article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
we use an external - cavity diode laser at @xmath39809 nm to seed a tapered amplifier ( moglabs , moa002 ) , which provides 1.8 w of output power .
the resulting beam is coupled into a single - mode fiber , and passed through three laser clean - up filters ( semrock ll01 - 808 ) .
this results in a 4 mm beam with @xmath39550 mw of power , which is split into an array of beams by an acousto - optic deflector ( aod ) ( aa opto - electronic model dtsx-400 - 800.850 ) .
these beams pass through a 1:1 telescope with 300 mm focal length and are then focused by a 0.5 na microscope objective ( mitutoyo g plan apo 50x ) into our vacuum chamber to form an array of optical tweezers ( fig .
1b of the main text ) .
these tweezers have a waist of @xmath39900 nm , and their centers are separated by 2.6@xmath40 m .
each beam has @xmath391 mw of optical power , corresponding to a trap depth of @xmath390.9 mk for @xmath41rb atoms .
the experimental sequence begins by laser cooling thermal @xmath41rb atoms in a magneto - optical trap ( mot ) around the traps for 100 ms ( fig .
we use a gradient field of 9.7 g / cm , and three intersecting retroreflected beams that are 17 mhz red detuned of the free space @xmath42 transition , overlapped with repumper beams resonant with the free space @xmath43 transition .
one of these beams is perpendicular to the optical table , and the other two are parallel to it ( intersecting at an angle of @xmath44 due to the geometric restriction imposed by the high resolution objectives ) .
all three beams have a diameter of @xmath391.5 cm , and carry @xmath391.5 mw of cooling light and @xmath390.4 mw of repumping light each .
to reduce the necessary time to load the mot , we shine uv light from a diode at 365 nm directly on the region of the glass cell within which the mot is loaded @xcite .
after 100 ms , the magnetic field gradient and the mot beams are turned off to allow the mot to dissipate over 28 ms . at the same time we turn on a set of probe beams which are 20 mhz red - detuned from the bare atom @xmath42 transition .
the probe beams have the same geometric configuration as the mot beams , but they have @xmath3950 times less power , and a diameter of @xmath391 mm , which largely reduces background light due to stray reflections during imaging .
the probe beams further cool the atoms through polarization - gradient cooling , and are left on for the remainder of the sequence .
the result of this process is the probabilistic loading of atoms into the traps .
for the data presented in fig . 3 of the main text we loaded on average 59@xmath455 atoms ( fig .
s1b ) .
* experimental sequence and atomic signal . * * a ) * pulse diagram of the experimental sequence . * b ) * distribution of number of initially loaded atoms .
* c ) * sample fluorescence count distribution for a single trap during 20 ms of exposure . ] the emccd camera ( andor ixon3 ) is triggered 128 ms after the beginning of the experimental sequence and acquires an image over the following 20 ms . cooling light from the probe beams
is scattered by the atoms and collected on the emccd , forming an image of the atoms in the array .
based on the collected photon statistics for each trap , we can set clear thresholds to determine the presence of an atom in a trap ( see fig .
furthermore , the bimodal nature of the photon statistics is an indication that the traps are occupied by either 0 or 1 atom .
once the emccd has finished acquiring the signal , the image file is transferred over the following 10 ms to a computer which determines trap occupations , using pre - calibrated regions of interest and threshold counts for each trap , in sub - ms time . using this information ,
the computer finds the correct pre - calculated waveform to displace each occupied trap during 3 ms from its initial position to its final position , and then adds up all of them into a multi - tone frequency - sweep waveform .
this computation takes @xmath390.2 ms for each trap to be displaced .
once the waveform is ready , it is sent to the aod to perform the trap displacement , and then it goes back to producing the original set of 100 traps .
it is important to note that the rearrangement waveform is only calculated for loaded traps , which means that all empty traps are turned off for the duration of the rearrangement but are restored immediately afterwards , so that the trap array returns after rearrangement to its original configuration .
after rearrangement , there is a @xmath397 ms buffer time before taking another 20 ms exposure image with the emccd . the entire process , consisting of image acquisition , transfer , analysis , waveform generation , rearrangement , and buffer , takes a total of 50 ms .
currently , the profile of the frequency sweeps is calculated to be piece - wise quadratic in time , over a duration of 3 ms . for shorter transport times
we observe an increase in the number of atoms lost during rearrangement .
for the experiments reported in the main text , atom losses are only slightly increased compared to the expected lifetime in static traps ( see fig .
these losses depend on the distance that the atoms move , and it is possible that they could be reduced by changing the length or the profile of the frequency sweep to minimize acceleration and jerk during the trajectory .
however , the fidelity of our rearrangement would not be significantly improved by minimizing these losses ( see fig .
3b of the main text )
. * rearrangement losses .
* losses as a function of total distance traveled by the atom during rearrangement for the dataset presented in fig . 3 of the main text .
the dashed line represents the expected loss for a stationary trap with 6.2 s lifetime . ]
when driving the aod with a single radio frequency ( rf ) tone , a portion of the input beam is deflected by an angle @xmath46 that depends on the frequency @xmath47 of the tone . by applying 100 different rf tones
@xmath48 , we generate 100 beams with output angles @xmath49 , where @xmath50 .
the waveform that we send into the aod is initially calculated by a computer that samples the desired waveform in the time - domain with a sampling rate of 100 mhz .
we stream these waveform samples to a software defined radio ( sdr ) ( ettus research , model usrp x310 , daughterboard ubx 160 ) which performs digital - analog conversion , low - pass filtering , and subsequent analog iq upconversion by a frequency of @xmath51 mhz , and outputs an analog waveform , which we then amplify and send to the aod .
the waveform that we calculate initially is given by : @xmath52 with @xmath53 and @xmath54 being the real amplitude and phase , respectively , of the rf tone with frequency @xmath55 .
since we generate all tones in the same waveform ( relative to the same local oscillator inside the sdr ) , the tones in our waveform have well - defined phases @xmath56 relative to one another . also , since all the frequencies we use are integer multiples of 1 khz , we calculate a 1 ms waveform which is streamed on a loop without needing to continuously generate new samples .
the finite power bandwidth , along with other imperfections of our system ( rf amplifier and aod ) generate a nonlinear response to the input signal . to the lowest order nonlinearity ,
the system acts as a mixer and generates new tones at the sum and difference of the input frequencies .
for example , for two tones , @xmath57 , @xmath58 , at the input , there will be a corresponding set of tones at the output : @xmath59 @xmath60 these terms are far removed from the main set of desired tones @xmath61 and can in principle be filtered by frequency .
however , they seed the next order of nonlinearity .
the next order of nonlinearity contains the mixing of these first order nonlinearities with the original tones .
for example , with two tones we would now see a mixing of the @xmath62 tone with the original @xmath63 tone to produce a sum tone at @xmath64 and a difference tone at @xmath65 . similarly , the @xmath62 tone would mix with the original @xmath65 tone to produce a tone at @xmath66 and @xmath63 . if the phases of each input tone @xmath56 are not carefully selected , these intermodulations interfere destructively with the original tones , as shown in fig .
s3a .
we address the issue of intermodulations by adjusting the phases of the different rf tones to almost completely cancel out the nonlinearities . as a first step ,
we generate a computer simulation of 100 tones evenly spaced in frequency and with random phases , and equal amplitudes . for each pair of traps
@xmath67 , we calculate the difference tone @xmath68 . by starting with random initial phases ,
the difference tones nearly completely destructively interfere with one another .
we then optimize each phase , one at a time , to further reduce the sum of all difference tones @xmath69 . after this
, we proceed to generate the waveform to be streamed onto the sdr .
the starting amplitudes for each tone are selected such that they individually produce a single deflection carrying the correct amount of optical power .
the frequencies span from 74.5 mhz to 123.01 mhz in steps of 0.49 mhz .
the next step in optimization consists of imaging the focused trap array on a ccd and performing 2d gaussian fits .
we use the amplitude of these fits to feed back on the amplitude of the individual tones .
once all the fitted amplitudes are approximately uniform , we continue to the last step of optimization .
since we are interested in the uniformity of the traps at the positions of the atoms , we measure the ac stark shifts induced by the traps , and use these values to feed back on the amplitude of each tone . in order to extract the ac stark shift ,
we shine a single laser beam onto the trapped atoms for @xmath70s , and measure the loss probability introduced by this
pushout " beam as a function of detuning from the bare @xmath42 resonance ( fig .
s4a inset ) . from the fits we extract the individual lightshift for each trap and use these values to perform feedback on the amplitudes .
we repeat the procedure until the shifted resonances are uniform to within @xmath71 across the trap array ( fig .
s4a ) . at this point
we have a set of optimal amplitudes @xmath72 and phases @xmath73 associated with the rf frequencies @xmath74 ( fig .
we interpolate between the values of @xmath72 to define the optimal amplitude as a function of frequency @xmath75 .
* characterization of trap and atom properties . *
( * a * ) the trapping light causes a lightshift on the atomic resonances that depends on the amplitude of the trap .
the measured lightshifts across the array are used to optimize the rf amplitudes @xmath76 .
this process results in homogeneous traps with lightshifts that are uniform to within @xmath392% .
inset shows the result of a `` pushout '' measurement ( see text ) on trap 26 that is used to determine the individual lightshift .
( * b * ) the trap frequencies are determined from a release and recapture measurement @xcite ( see inset for trap 26 ) .
the errors from the fits are smaller than the marker size for all the figures . ] to characterize the homogeneity of the final trap configuration , we perform experiments to determine the ac stark shift and trap frequency for each trap in the array ( fig .
s4 ) . as outlined in the previous section , the measurement of the ac stark shift is used to equalize the traps .
we find an average shift of 17.5 mhz with a standard deviation of 0.1 mhz across the whole array . the trap frequency ( fig .
s4b ) is found through a release and recapture technique @xcite .
we obtain the radial frequency from a fit to the probability of retaining an atom as a function of hold time ( fig .
s4b inset ) .
we find the average trap frequency to be 98.7 khz with a standard deviation of 1.7 khz across the array .
combining these measurements with an independent determination of the waist , we estimate a trap depth of @xmath390.9 mk . for most of the experimental sequence time ,
the traps are static .
however , during short bursts we stream new waveforms to the aod to rearrange them , and the atoms they hold .
we move our traps by sweeping the frequencies @xmath77 of the tones that correspond to the traps we wish to move , in a piecewise - quadratic fashion .
this way , the atoms experience a constant acceleration @xmath78 for the first half of the trajectory , and @xmath79 for the second half . during the sweep , we also continuously adjust the amplitude of the rf tone to match the optimized amplitude for its current frequency : @xmath80 .
further , by slightly adjusting the duration of each sweep , we enforce each trap to end with the optimal phase corresponding to its new position . using these parameters we pre - calculate all waveforms to sweep the frequency of a tone at
any given starting frequency in our array to any given final frequency , over 3 ms .
this amounts to @xmath81 precalculated trajectories .
the lifetime for each trap is found by an exponential fit to the probability of retaining an atom as a function of time . under optimal conditions , this results in an average lifetime for the traps in our array of 11.6 s with a standard deviation of 0.5 s across the array .
however , this value depends on the background pressure inside the vacuum chamber and therefore depends on the current with which we drive our dispenser atom source . for the measurements presented in fig . 3 of the main text ,
an average lifetime of 6.2 s was found from independent calibration measurements . in these measurements ,
we apply continuous laser cooling throughout the hold time .
we observe that without continuous cooling , the lifetime is reduced compared to these values . the retention as a function of time in this case
does not follow a simple exponential decay ( indicative of a heating process ) , and we define the time at which the retention probability crosses @xmath82 to be the lifetime . for the standard configuration in the main text ( 100 traps with 0.49 mhz spacing between neighboring frequencies ) ,
we find a lifetime of @xmath83 .
while we have not characterized the source of the lifetime reduction in detail , we have carried out a number of measurements in different configurations to distinguish various effects : * generating a single trap by driving the aod with a single frequency from a high - quality signal generator ( rhode & schwartz smc100a ) , we find a lifetime of @xmath84 .
this indicates that there are additional heating effects , such as photon scattering from trap light or intensity noise , that are independent of the use of the sdr or the fact that we drive the aod with a large number of frequencies .
possible improvements include using further detuned trap light and improving our intensity stabilization .
furthermore , in a separate experiment @xcite , we observed that using a titanium - sapphire laser instead of a ta significantly improved trap lifetimes even at the same trapping wavelength . * generating a single trap by driving the aod with a single frequency from the sdr , we find a lifetime of @xmath85 .
this indicates that the rf - waveforms from the sdr could have additional phase or intensity noise .
a possible source is the local oscillator used for iq upconversion in the sdr , which could be replaced with a more phase - stable version .
* we observe the same trap lifetime of @xmath85 when driving the aod with 70 frequencies at a spacing of @xmath86 mhz using the sdr .
this indicates that , in principle , there is no lifetime reduction associated with driving the aod with a large number of frequencies .
* we observe a reduction of lifetime for frequency spacings smaller than @xmath87 .
for example , we found a lifetime of @xmath83 , when setting the frequency spacing to @xmath88 .
( this lifetime is unchanged by increasing the number of traps from 70 to 100 . )
we interpret this effect to be the result of interference between neighboring traps .
due to the finite spatial overlap of the tweezer light , a time - dependent modulation occurs with a frequency given by the frequency spacing between neighboring traps . when bringing traps closer together , both the spatial overlap increases and the modulation frequency approaches the parametric heating resonance at @xmath89 given by twice the radial trapping frequency
. we would like to stress that these effects only play a role after the rearrangement procedure is completed . during rearrangement ,
continuous laser cooling is a valid and powerful method to reduce heating effects .
additionally , the flexibility of our system makes it possible to load and continuously cool atoms in a set of closely spaced traps , and to rearrange the filled traps to an array with larger separations , at which point cooling can be turned off .
this method of rearrangement takes advantage of the large number of atoms that can be initially loaded in our set of 100 traps separated by 0.49 mhz , while eliminating the effect of interference by setting a larger final frequency separation of , for example , 0.7 mhz .
furthermore , atoms could also be transferred into a fully static " trap array , such as an optical lattice , after rearrangement .
however , even with optimal cooling parameters , heating effects can not be always compensated .
for example , we observed a significant reduction of lifetime for frequency spacings below @xmath90 even with continuous laser cooling .
this sets a limit on the maximum number of traps that can be generated within the bandwitdh of our aod , and therefore limits the final sizes of the atomic arrays .
in this section we will discuss possible extensions of our method to form uniformly filled two - dimensional ( 2d ) arrays .
we will describe two different rearrangement strategies and compare their performance given realistic parameters for loading efficiencies and lifetimes . using a 2d aod
we could generate a 2d array of optical tweezers using two sets of rf tones , one set corresponding to rows and the other corresponding to columns .
after loading atoms probabilistically into the array , it would be possible to eliminate each defect by turning off the rf frequency that generates either the row or the column containing the defect , and then transport all remaining rows and columns to form a defect - free uniform array .
a different approach is to generate a static two - dimensional array of traps using techniques such as optical lattices or spatial - light modulators . in a first step ,
atoms would be probabilistically loaded into this static array .
after loading , we could use an independent aod to generate a linear array of traps , deeper than the ones forming the static array and overlapping precisely with one row of static traps . by rearranging the linear array
, we could transfer the atoms to their final locations in the static configuration , where they would remain after turning off the traps used for transport . doing this for
each row would make it possible to fill an entire region of the static 2d array .
the final size of the array will depend on the initial loading efficiency , while the probability to have a defect - free array after rearrangement will depend on the atom lifetimes in the traps , and the total feedback time .
this time consists of several blocks : image taking , image transfer , waveform calculation , and trap movement .
we can take an image in 20 ms . the time it takes to transfer the image from the camera to the computer takes a minimum of 9 ms and each row of atoms adds 0.8 ms to the transfer time . analyzing the image to determine
the location of the atoms and necessary frequency sweeps can be done in sub - ms time . generating the waveform takes @xmath390.2 ms for each sweep necessary .
this time scales as the final number of atoms ( @xmath91 ) for method 2 , and with the final number of rows and columns ( @xmath92 ) for method 1 .
finally , the rearrangement itself takes 3 ms for each set of frequency sweeps : for method 1 there are two sweeps , and for method 2 the number of sweeps equals the number of rows .
figure s5 shows the result of a monte carlo simulation for the expected number of atoms in a defect - free rectangular array using two sets of parameters , and both rearrangement methods described above .
given our current loading efficiency of @xmath93 and vacuum - limited lifetime of @xmath3910 s , we can expect more than 160 atoms in the final defect - free configuration ( fig .
s5a , c ) .
however , if we were to upgrade our vacuum setup to increase the lifetime from 10 s to 60 s , and we increased the loading efficiency to 0.9 using currently available techniques @xcite , we could expect more than 600 atoms in defect - free configurations ( fig .
s5b , d ) .
these numbers were calculated by simulating a sequence that performed repeated rearrangement until no more defects appeared , and every point on the plot is the average of 500 simulations .
_ ( see ancillary files ) _
vs1 : rearrangement procedure .
this video shows consecutive pairs of atom fluorescence images .
each pair consists of an initial image showing the random loading into the array of traps and a subsequent image after the loaded atoms have been rearranged to form a regular array .
the sequence of images demonstrates the effectiveness of the procedure to reduce the entropy associated with the random initial positions of the atoms by ordering them in long , uninterrupted chains . the 1 hz cycle rate in the video
is slowed down by a factor of five relative to the 5 hz experimental repetition rate .
vs2 : lifetime extension of 20 atom array .
this video shows consecutive fluorescence images of atoms being rearranged from an initial probabilistic loading into an ordered array of 20 atoms and a reservoir formed by surplus atoms .
the length of the target array is kept constant by transporting atoms from the reservoir to replace lost atoms in the target array .
this technique makes it possible to extend the average lifetime of a defect - free 20 atom array to @xmath398 s , at which point the reservoir typically has depleted .
the 10 hz frame rate of this video accurately reflects the rate at which these images were acquired .
vs3 : lifetime extension of 40 atom array .
this video shows consecutive fluorescence images of atoms being rearranged from an initial probabilistic loading into an ordered array of 40 atoms and a reservoir formed by surplus atoms .
the length of the target array is kept constant by transporting atoms from the reservoir to replace lost atoms in the target array .
this technique makes it possible to extend the average lifetime of a defect - free 40 atom array to @xmath392 s , at which point the reservoir typically has depleted .
the 10 hz frame rate of this video accurately reflects the rate at which these images were acquired . c. klempt et al .
a * 73 * , 013410 ( 2006 ) .
y. r. p. sortais et al . , phys .
a * 75 * , 013406 ( 2007 ) .
j. d. thompson et al . , science * 340 * , 1202 ( 2013 ) .
t. grnzweig , a. hilliard , m. mcgovern , and m. f. andersen , nat . phys .
* 6 * , 951 ( 2010 ) .
b. j. lester , n. luick , a. m. kaufman , c. m. reynolds , and c. a. regal , phys .
* 115 * , 073003 ( 2015 ) .
y. h. fung and m. f. andersen , new j. phys .
* 17 * , 073011 ( 2015 ) . | the realization of large - scale fully controllable quantum systems is an exciting frontier in modern physical science .
we use atom - by - atom assembly to implement a novel platform for the deterministic preparation of regular arrays of individually controlled cold atoms . in our approach ,
a measurement and feedback procedure eliminates the entropy associated with probabilistic trap occupation and results in defect - free arrays of over @xmath0 atoms in less than @xmath1 .
the technique is based on fast , real - time control of @xmath2 optical tweezers , which we use to arrange atoms in desired geometric patterns and to maintain these configurations by replacing lost atoms with surplus atoms from a reservoir .
this bottom - up approach enables controlled engineering of scalable many - body systems for quantum information processing , quantum simulations , and precision measurements .
the detection and manipulation of individual quantum particles , such as atoms or photons , is now routinely performed in many quantum physics experiments @xcite ; however , retaining the same control in large - scale systems remains an outstanding challenge .
for example , major efforts are currently aimed at scaling up ion - trap and superconducting platforms , where high - fidelity quantum computing operations have been demonstrated in registers consisting of several qubits @xcite .
in contrast , ultracold quantum gases composed of neutral atoms offer inherently large system sizes .
however , arbitrary single atom control is highly demanding and its realization is further limited by the slow evaporative cooling process necessary to reach quantum degeneracy . only in recent years has individual particle detection @xcite and basic single - spin control @xcite been demonstrated in low entropy optical lattice systems .
this report demonstrates a novel approach for rapidly creating scalable quantum matter with inherent single particle control via atom - by - atom assembly of large defect - free arrays of cold neutral atoms @xcite .
na microscope objective , which creates an array of tightly focused optical tweezers in a vacuum chamber .
an identical microscope objective is aligned to image the same focal plane .
a dichroic mirror allows us to view the trap light on a charge - coupled - device camera ( ccd ) while simultaneously detecting the atoms via fluorescence imaging on an electron - multiplied - ccd camera ( emccd ) .
the rearrangement protocol is realized through fast feedback onto the multi - tone radio - frequency ( rf ) field driving the aod.,scaledwidth=47.0% ] we use optical microtraps to directly extract individual atoms from a laser - cooled cloud @xcite and employ recently demonstrated trapping techniques @xcite and single - atom position control @xcite to create desired atomic configurations .
central to our approach is the use of single - atom detection and real - time feedback @xcite to eliminate the entropy associated with the probabilistic trap occupation @xcite ( currently limited to ninety percent even with advanced loading techniques @xcite ) .
related to the fundamental concept of
maxwell s demon " @xcite , this method allows us to rapidly create large defect - free atom arrays and to maintain them for long periods of time , providing an excellent platform for large - scale experiments based on techniques ranging from rydberg - mediated interactions @xcite to nanophotonic platforms @xcite and hubbard model physics @xcite .
the experimental protocol is illustrated in fig .
1a .
an array of @xmath2 tightly focused optical tweezers is loaded from a laser - cooled cloud .
the collisional blockade effect ensures that each individual tweezer is either empty or occupied by a single atom @xcite .
a first high - resolution image yields single - atom resolved information about the trap occupation , which we use to identify empty traps and to switch them off .
the remaining occupied traps are rearranged into a regular , defect - free array and we detect the final atom configuration with a second high - resolution image .
our implementation relies on fast , real - time control of the tweezer positions ( see fig .
1b ) , which we achieve by employing an acousto - optic deflector ( aod ) that we drive with a multi - tone radio - frequency ( rf ) signal .
this generates an array of deflected beams , each controlled by its own rf - tone @xcite .
the resulting beam array is then focused into our vacuum chamber and forms an array of optical tweezers , each with a gaussian waist of @xmath3 , a wavelength of @xmath4 , and a trap depth of @xmath5 ( @xmath6 , boltzmann constant ) that is homogeneous across the array within @xmath7 @xcite . the tweezer array is loaded from a laser - cooled cloud of rubidium-@xmath8 atoms in a magneto - optical trap ( mot ) . after the loading procedure , we let the mot cloud disperse and we detect the occupation of the tweezers with fluorescence imaging . fast , single - shot , single - atom resolved detection with @xmath9 exposure is enabled by a sub - doppler laser - cooling configuration that remains active during the remainder of the sequence @xcite ( see figs .
2a-2c ) .
our fluorescence count statistics show that individual traps are either empty or occupied by a single atom @xcite , and we find probabilistically filled arrays with an average single atom loading probability of @xmath10 ( see figs .
2a and 3a ) .
the central part of our scheme involves the rearrangement procedure for assembling defect - free arrays @xcite ( see fig .
1a ) . in the first step ,
unoccupied traps are switched off by setting the corresponding rf - amplitudes to zero . in a second step , all occupied
tweezers are moved to the left until they stack up with the original spacing of @xmath11 .
this movement is generated by sweeping the rf - tones to change the deflection angles of the aod and lasts @xmath12 @xcite .
finally , we detect the resulting atom configuration with a second high - resolution image . these steps implement a reduction of entropy via measurement and feedback .
the effect is immediately visible in the images shown in figs .
2a and 2b .
while the initial filling of our array is probabilistic , the rearranged configurations show highly ordered atom arrays .
our approach also allows us to construct flexible atomic patterns as shown in fig . 2c .
the rearrangement procedure creates defect - free arrays with high fidelity .
this can be quantified by considering the improvement of single atom occupation probabilities ( fig .
3a ) and the success probabilities , @xmath13 , for creating defect - free arrays of length @xmath14 ( fig .
3b ) .
the single atom occupation probability in the left - most forty traps increases from @xmath15 before rearrangement to @xmath16 after rearrangement , demonstrating our ability for high - fidelity single - atom preparation .
furthermore , the success probabilities for creating defect - free arrays show an exponential improvement .
prior to rearrangement , the probability of finding a defect - free array of length @xmath14 is exponentially suppressed with @xmath17 where @xmath18 ( blue circles , fig .
3b ) . after rearrangement ,
we find success probabilities as high as @xmath19 and @xmath20 ( red circles , fig .
3b ) .
the same exponential improvement is observed by considering the average wait time for producing defect - free arrays , given by @xmath21 , where @xmath22 is the cycling time of our experiment ( see fig .
3b ) .
for example , we are able to generate defect - free arrays of @xmath0 atoms with an average wait time of less than @xmath1 ( red circles , fig
. 3c ) .
in contrast , the likelihood of finding defect - free arrays of this length without rearrangement is so small that the corresponding wait time would be on the order of several hundred years .
the success probabilities can be further enhanced through multiple repetitions of the rearrangement protocol .
figure 4 illustrates the procedure in which we target an atomic array of fixed length and create a reservoir from surplus atoms in a separate zone . after the initial arrangement of atoms into the target and reservoir zones , we periodically take images to identify defects in the target array and pull atoms from the reservoir to fill in these defects .
this enhances our initial success probabilities at producing defect - free arrays within one mot - loading cycle to nearly the ideal limit ( see fig .
4c ) .
finally , a similar procedure can be used for correcting errors associated with atomic loss .
this becomes a significant limitation for large arrays since for a given lifetime of an individual atom in the trap @xmath23 , the corresponding lifetime of the @xmath14 atom array scales as @xmath24 . to counter this loss
, we repeatedly detect the array occupation after longer time intervals and replenish lost atoms from the reservoir .
as demonstrated in fig .
4d , this procedure leads to exponentially enhanced lifetimes of our arrays .
these results demonstrate the ability to generate and control large , defect - free arrays at a fast repetition rate .
the success probabilities are limited by two factors : the initial number of loaded atoms and losses during rearrangement .
for example , the average total atom number in our array is @xmath25
@xcite , which results in the cutoff in the success probability in fig .
3b starting from @xmath26 ( solid line ) . for shorter arrays ,
the fidelity is mostly limited by losses during rearrangement .
these losses are dominated by our finite vacuum - limited lifetime , which varies from @xmath27 to @xmath28 ( depending on the setting of our atomic dispenser source ) , and are only minimally increased by the movement of the atoms @xcite .
the single atom occupation probability is correspondingly reduced by a factor @xmath29 , where @xmath30 is the time for the whole rearrangement procedure @xcite .
this results in the success probabilities of creating length-@xmath14 arrays scaling as @xmath31 , which dominates the slope for @xmath32 in fig .
3b ( dashed line ) . currently , we reach vacuum limited lifetimes only with sub - doppler cooling applied throughout the sequence
. however , the lifetime without cooling could be improved , for example , by using a different trapping laser and trapping wavelength @xcite .
the size of the final arrays can be considerably increased by implementing a number of realistic experimental improvements .
for example , the initial loading probability could be enhanced to @xmath33 @xcite and the vacuum limited lifetime could be improved to @xmath34 in an upgraded vacuum chamber . increasing the number of traps in the current configuration is difficult due to the aod bandwidth of @xmath35 and strong parametric heating introduced when the frequency spacing of neighboring traps approaches @xmath36 @xcite . however , implementing two - dimensional ( 2d ) arrays could provide a path towards realizing thousands of traps , ultimately limited by the availability of laser power and the field of view of high - resolution objectives
. such 2d configurations could be realized by either directly using a 2d - aod or by creating a static 2d lattice of traps ( using spatial light modulators @xcite or optical lattices @xcite ) and sorting atoms with an independent aod @xcite .
realistic estimates for the rearrangement time @xmath37 in such 2d arrays indicate that the robust creation of defect - free arrays of hundreds of atoms is feasible @xcite . finally , the repetitive interrogation techniques , in combination with periodic reservoir reloading from a cold atom source ( such as a mot ) , could be used to maintain arrays indefinitely .
remarkable recent advances have been made in building , exploring , and applying engineered quantum matter in systems ranging from ion traps @xcite and ultracold atoms @xcite to solid - state qubits such as superconducting qubits @xcite and color - centers @xcite .
atom - by - atom assembly of defect - free arrays forms a novel scalable platform with unique possibilities .
it combines features that are typically associated with ion trapping experiments , such as single - qubit addressability @xcite and fast cycling times , with the flexible optical trapping of neutral atoms in a scalable fashion .
furthermore , in contrast to solid - state platforms , such atomic arrays are formed of indistinguishable particles and are mostly decoupled from their environment .
these features provide an excellent starting point for engineering interactions in multi - qubit experiments and for exploring novel applications .
in particular , our approach can be used for the realization of logical qubits based on quantum registers consisting of up to several dozens of atomic spin qubits .
the necessary entangling operations can be realized using atomic rydberg states @xcite .
the parallelism afforded by our flexible atom rearrangement enables efficient diagnostics of such rydberg - mediated entanglement .
for example , the use of atomic configurations demonstrated in fig . 2c , supplemented by the repetitive reloading mechanism in fig
. 4 , could allow fidelities to be optimized to their theoretical limit @xcite .
similar techniques can be used for novel approaches to quantum simulations using both coherent coupling and engineered dissipation @xcite and for creating large - scale maximally entangled quantum states for applications in precision measurements @xcite .
in addition , the realization of homogeneous arrays @xcite should allow for simultaneous sideband cooling of atoms in optical tweezers @xcite .
ground state fractions in excess of @xmath38 have already been demonstrated , and can likely be improved via additional optical trapping along the longitudinal tweezer axes .
this could enable bottom - up approaches to studying quantum many - body physics in hubbard models @xcite , where atomic mott insulators with fixed atom number and complex spin patterns could be directly assembled without the fluctuations and long cycling times associated with evaporative cooling .
moreover , these techniques could potentially be applied to create ultracold quantum matter composed of exotic atomic species or complex molecules @xcite that are difficult to cool evaporatively .
finally , our atom arrays are ideally suited for integration with hybrid atom - nanophotonic platforms @xcite .
these can be employed to couple the atoms within a local multi - qubit register or for efficient communication between the registers using a modular quantum network architecture @xcite .
+ * acknowledgments * + we thank k .- k .
ni , n. hutzler , a. mazurenko and a. kaufman for insightful discussion .
this work was supported by nsf , cua , nsseff and harvard quantum optics center .
hb acknowledges support by a rubicon grant of the netherlands organization for scientific research ( nwo ) .
+ + * supplementary materials * + materials and methods + figs .
s1 to s5 + movies s1 to s3 + |
the introduction of unitarity constraints in coupled channels in chiral perturbation theory has led to unitary extensions of the theory that starting from the same effective lagrangians allow one to make predictions at much higher energies .
one of the interesting consequences of these extensions is that they generate dynamically low lying resonances , both in the mesonic and baryonic sectors . by this
we mean that they are generated by the multiple scattering of the meson or baryon components , much the same as the deuteron is generated by the interaction of the nucleons through the action of a potential , and they are not preexistent states that remain in the large @xmath5 limit where the multiple scattering is suppressed . starting from the chiral lagrangians for the interaction of the octets of meson and baryon @xcite and using the n / d method to obtain a scattering matrix fulfilling exactly unitarity in coupled channels @xcite , the full set of transition matrix elements with the coupled channels is given in matrix form by @xmath6^{-1}\ , v~. \label{eq : bs1}\ ] ] here , the matrix @xmath7 , obtained from the lowest order meson baryon chiral lagrangian , contains the weinberg - tomozawa or seagull contribution , as employed e.g. in ref .
@xcite , @xmath8 for strangeness @xmath1 the @xmath9 coefficients are given in ref .
@xcite , and an averaged meson decay constant @xmath10 is used @xcite , with @xmath11mev the weak pion decay constant .
the diagonal matrix @xmath12 stands for the loop function of a meson and a baryon and is defined by a subtracted dispersion relation in terms of phase space with a cut starting at the corresponding threshold @xcite .
it corresponds to the loop function of a meson and a baryon once the logarithmic divergent constant is removed .
the analytical properties of @xmath12 are properly kept when evaluating the previous loop function in dimensional regularization and an explicit subtraction constant,@xmath13 , appears in the expression .
this meson baryon loop function was calculated in ref .
@xcite with a cut - off regularization .
the values of the @xmath14 constants are found to be around @xmath15 to agree with the results of the cut off method for cut off values of the order of the mass of the @xmath16 @xcite , which we call of natural size .
the study of ref .
@xcite showed the presence of poles in eq .
( [ eq : bs1 ] ) around the @xmath2 and the @xmath17 for isospin @xmath18 and around the @xmath19 in @xmath20 . the same approach in @xmath21 leads to the resonance @xmath22 @xcite and in @xmath23 to the @xmath24 @xcite .
one is thus tempted to consider the appearance of a singlet and an octet of meson baryon resonances . nevertheless , the situation is more complicated because indeed in the su(3 ) limit there are _ two _ octets and not just one , as we discuss below .
the presence of these multiplets was already discussed in ref .
@xcite after obtaining a pole with @xmath1 in the @xmath20 channel , with mass around 1430 mev , and two poles with @xmath18 , of masses around that of the @xmath2 .
the appearance of a multiplet of dynamically generated mesons and baryons seems most natural once a state of the multiplet appears .
indeed , one must recall that the chiral lagrangians are obtained from the combination of the octet of pseudoscalar mesons ( the pions and partners ) and the octet of stable baryons ( the nucleons and partners ) .
the su(3 ) decomposition of the combination of two octets tells us that @xmath25 thus , on pure su(3 ) grounds , should we have a su(3 ) symmetric lagrangian , one can expect e.g. one singlet and two octets of resonances , the symmetric and antisymmetric ones . the lowest order of the meson baryon chiral lagrangian is exactly su(3 ) invariant if all the masses of the mesons , or equivalently the quark masses , are set equal . as stated above [ see eq .
( [ eq : ampl2 ] ) ] , in ref . @xcite the baryon masses take their physical values , although strictly speaking at the leading order in the chiral expansion they should be equal to @xmath26 . for eq .
( [ eq : ampl2 ] ) being su(3 ) symmetric , all the baryons masses @xmath27 must be set equal as well . when all the meson and baryon masses are equal , and these common masses are employed in evaluating the @xmath28 functions , together with equal subtraction constants @xmath29 , the @xmath30matrix obtained from eq .
( [ eq : bs1 ] ) is also su(3 ) symmetric .
if we do such an su(3 ) symmetry approximation and look for poles of the scattering matrix , we find poles corresponding to the octets and singlet .
the surprising result is that the two octet poles are degenerate as a consequence of the dynamics contained in the chiral lagrangians . indeed ,
if we evaluate the matrix elements of the transition potential @xmath7 in a basis of su(3 ) states , we obtain something proportional to @xmath31 taking the following order for the irreducible representations : @xmath32 , @xmath33 , @xmath34 , @xmath35 , @xmath36 and @xmath37 , with positive sign meaning attraction .
hence we observe that the states belonging to different irreducible representations do not mix and the two octets appear degenerate .
the coefficients in @xmath38 clearly illustrate why there are no bound states in the @xmath35 , @xmath36 and @xmath37 representations . in practice ,
the same chiral lagrangians allow for su(3 ) breaking . in the case of refs .
@xcite the breaking appears because both in the @xmath39 transition potentials as in the @xmath28 loop functions one uses the physical masses of the particles as well as different subtraction constants in @xmath28 , corresponding to the use of a unique cut - off in all channels . in ref .
@xcite the physical masses are also used in the @xmath28 functions , although these functions are evaluated with a unique subtraction constant as corresponds to the su(3 ) limit . in both approaches ,
physical masses are used to evaluate the @xmath28 loop functions so that unitarity is fulfilled exactly and the physical thresholds for all channels are respected . by following the approach of ref .
@xcite and using the physical masses of the baryons and the mesons , the position of the poles change and the two octets split apart in four branches , two for @xmath18 and two for @xmath20 , as one can see in fig .
[ fig : tracepole ] . in the figure
we show the trajectories of the poles as a function of a parameter @xmath40 that breaks gradually the su(3 ) symmetry up to the physical values .
the dependence of masses and subtraction constants on the parameter @xmath40 is given by @xmath41 where @xmath42 . in the calculation of fig .
[ fig : tracepole ] , the values @xmath43 mev , @xmath44 mev and @xmath45 are used .
the complex poles , @xmath46 , appear in unphysical sheets . in the present search
we follow the strategy of changing the sign of the momentum @xmath47 in the @xmath48 loop function for the channels which are open at an energy equal to re(@xmath49 ) .
the splitting of the two @xmath18 octet states is very interesting .
one moves to higher energies to merge with the @xmath17 resonance and the other one moves to lower energies to create a pole , quite well identified below the @xmath50 threshold , with a narrow width . on the other hand
, the singlet also evolves to produce a pole at low energies with a quite large width .
we note that the singlet and the @xmath18 octet states appear nearby in energy and what experiments actually see is a combination of the effect of these two resonances .
similarly as for the @xmath18 octet states , we can see that one branch of the @xmath20 states moves to higher energies while another moves to lower energies .
the branch moving to higher energies finishes at what would correspond to the @xmath19 resonance when the physical masses are reached .
the branch moving to lower energies fades away after a while when getting close to the @xmath50 threshold . the model of ref .
@xcite reproduces qualitatively the same results , but the @xmath20 pole also stays for @xmath51 .
nevertheless , in both approaches there is an @xmath20 amplitude with an enhanced strength around the @xmath52 threshold .
this amplitude has non negligible consequences for reactions producing @xmath53 pairs in that region .
this has been illustrated for instance in ref .
@xcite , where the photoproduction of the @xmath2 via the reaction @xmath54 was studied .
it was shown there that the different sign in the @xmath20 component of the @xmath55 , @xmath56 states leads , through interference between the @xmath20 and the dominant @xmath18 amplitudes , to different cross sections in the various charge channels , a fact that has been confirmed experimentally very recently @xcite .
once the pole positions are found , one can also determine the couplings of these resonances to the physical states by studying the amplitudes close to the pole and identifying them with @xmath57 the couplings @xmath58 are in general complex valued numbers .
in the table we summarize the pole positions and the complex couplings @xmath58 obtained from the model of ref .
@xcite for isospin @xmath18 .
the results with the model of @xcite are qualitatively similar .
.pole positions and couplings to @xmath18 physical states from the model of ref .
@xcite [ cols="^,^,^,^,^,^,^ " , ] [ tab : jido0 ] we observe that the second resonance with @xmath18 couples strongly to @xmath52 channel , while the first resonance couples more strongly to @xmath53 .
in a given reaction the @xmath2 resonance is always seen in @xmath53 mass distribution .
however , the @xmath2 can be produced through any of the channels in the table .
hence , it is clear that , should there be a reaction which forces this initial channel to be @xmath50 , then this would give more weight to the second resonance , @xmath59 , and hence produce a distribution with a shape corresponding to an effective resonance narrower than the nominal one and at higher energy .
such a case indeed occurs in the reaction @xmath60 studied theoretically in ref .
it was shown there that since the @xmath61 system has a larger energy than the resonance , one has to lose energy emitting a photon prior to the creation of the resonance and this is effectively done by the bremsstrahlung from the original @xmath62 or the proton .
hence the resonance is initiated from the @xmath61 channel .
this is also the case in the reaction @xmath63 which has also the @xmath2 initiated by @xmath52 through the vertex @xmath64 @xcite .
given the success of the chiral unitary approach in generating dynamically low energy resonances from the interaction of the octets of stable baryons and the pseudoscalar mesons , in @xcite the interaction of the decuplet of @xmath65 with the octet of pseudoscalar mesons was studied and shown to lead to many states which were associated to experimentally well established @xmath66 resonances .
the lowest order chiral lagrangian for the interaction of the baryon decuplet with the octet of pseudoscalar mesons is given by @xcite @xmath67 where @xmath68 is the spin decuplet field and @xmath69 the covariant derivative given by in @xcite .
the identification of the physical decuplet states with the @xmath68 can be seen in @xcite , where a detailed study of this interaction and the resonances generated can be seen .
the study is done along the lines of the former sections , looking for poles in the second riemann sheet of the complex plane , the coupling of the resonances to the different channels and the stability of the results with respect to variations of the input parameter , which in our case is just the subtraction constant , a .
this allows the association of the resonances found to existing states of the particle data book , and the prediction of new ones .
a detail of the results obtained can be seen in fig .
2 . another interesting result is the generation of an exotic state of @xmath3 and @xmath20 which is generated by the interaction of the @xmath70 channels and stands as a @xmath4 resonance .
the pole appears in a riemann sheet below threshold when also the sign of the momentum is changed , but it leads to a @xmath70 amplitude which accumulates strength close to threshold and produces a broad peak in the cross section , see fig.3 , in contrast to the @xmath71 cross section which is much smaller and very smooth . in order to investigate this interaction
we propose the study of the following reactions : 1 ) @xmath72 , 2 ) @xmath73 , 3 ) @xmath74 . in the first case the @xmath75 state produced
has necessarily @xmath20 . in the second case
the @xmath76 state has @xmath71 . in the third case
the @xmath77 state has mostly an @xmath20 component .
the study of the @xmath4 invariant mass distribution in these reactions would provide experimental information to eventually prove the prediction made here , and thus giving support to this new exotic baryonic state , which , although in the quark model would require at least five quarks and could qualify as a pentaquark , finds a much simpler interpretation as a resonant @xmath4 state . for @xmath20,width=377 ]
this work is partly supported by dgicyt contract number bfm2003 - 00856 , cicyt ( spain ) grant nos .
fpa2002 - 03265 and fpa2004 - 03470 , the e.u .
euridice network contract no .
hprn - ct-2002 - 00311 and the research cooperation program of the jsps and the csic .
99 g. ecker , prog .
* 35 * ( 1995 ) 1 ; v. bernard , n. kaiser and u .- g .
meiner , int .
j. mod .
e * 4 * ( 1995 ) 193 .
j. a. oller and u .- g .
meiner , phys .
b * 500 * ( 2001 ) 263 .
e. oset , a. ramos and c. bennhold , phys .
b * 527 * ( 2002 ) 99 [ erratum - ibid .
b * 530 * ( 2002 ) 260 ] .
n. kaiser , p. b. siegel and w. weise , nucl .
phys . a * 594 * ( 1995 ) 325 t. inoue , e. oset and m. j. vicente vacas , phys .
c * 65 * ( 2002 ) 035204 .
d. jido , j.a .
oller , e. oset , a. ramos , u.g .
meissner , nucl . phys .
a * 725 * ( 2003 ) 181 .
j. c. nacher , e. oset , h. toki and a. ramos , phys .
b * 455 * , 55 ( 1999 ) .
j. k. ahn et al . , proceedigns of the xvi particles and nuclei international conference ( panic02 ) , eds .
k. imai , t. kishimodo and h. toki , nucl .
a721 ( 2003 ) 715 - 718 . j. c. nacher , e. oset , h. toki and a. ramos , phys .
b * 461 * ( 1999 ) 299 .
t. hyodo , a. hosaka , m.j .
vicente vacas and e. oset , phys .
b*593 * ( 2004)75 - 81 .
e. e. kolomeitsev and m. f. m. lutz , phys .
b * 585 * ( 2004 ) 243 .
e. jenkins and a. v. manohar , phys .
b * 259 * ( 1991 ) 353 .
s. sarkar , e. oset and m. j. vicente vacas , arxiv : nucl - th/0407025 . | in this talk we report on the use of a chiral unitary approach for the interaction of the octets of meson and baryon and the octet of mesons with the decuplet of baryons .
two octets of @xmath0 baryon states and a singlet are generated dynamically in the first case , resulting in the case of strangeness @xmath1 in two poles of the scattering matrix close to the nominal @xmath2 resonance . in the second case
many resonances are also generated , among them an exotic baryon with @xmath3 corresponding to a @xmath4 resonance .
we make suggestions of experiments which could show evidence for the existence of these states . |
cataclysmic variables ( cvs ) result when the cool , low mass secondary in a binary contacts its roche lobe and transfers material to the white dwarf primary .
to arrive at this stage , the standard evolutionary paradigm involves having the more massive component in the binary evolve off the main sequence and engulf its companion in a short - lived common envelope ( ce ) phase ( e.g. , iben & livio 1993 ) . during the ce phase , the outer atmosphere of the primary star
is removed by interaction with the secondary star , and this interaction leads to either a stellar merger , or a more compact binary ( politano & weiler 2007 ) .
angular momentum loss , such as through `` magnetic braking '' ( verbunt & zwaan 1981 ) , eventually leads to a semi - contact binary and a cv is born . in most of the early population synthesis calculations ( cf .
, howell et al . 2001 , and references therein ) the vast majority of donor stars in cvs were expected to be unevolved .
if true , the donor stars should generally appear to be similar to their main sequence counterparts , except for the effects introduced by the mass loss process .
thus , while they will have a slightly different mass - radius relationship due to the secondary being out of thermal equilibrium , their photospheres should not show any abundance peculiarities . the standard evolutionary paradigm though is slowly changing to reflect the fact that many long period cvs appear to have donors with unusual masses and/or radii .
this suggests that the secondaries in some subset of cvs have undergone significant evolution prior to becoming interacting binaries .
podsiadlowski et al .
( 2003 ) constructed a population synthesis for cvs that allowed for evolved secondaries with initial masses of up to 1.4 m@xmath7 .
inclusion of these objects allowed them to explain the wide dispersion in the observed temperatures of donor stars in cvs with p@xmath8 @xmath4 5 hr .
this result confirmed the conclusions of baraffe & kolb ( 2000 ) , that long period cvs are dominated by systems containing evolved secondaries .
recently , goliasch & nelson ( 2015 ) have produced a new population synthesis calculation that includes detailed nuclear evolution .
they confirm the results of podsiadlowski et al . , showing that there is a large range of masses and radii for donors in cvs with p@xmath8 @xmath4 5 hr when nuclear evolved secondary stars are considered .
two other interesting results from this study are that cvs with evolved secondaries do not have a period gap p@xmath8 @xmath9 3 hr ( see kolb et al .
1998 ) . ] , and they can evolve to much shorter periods than predicted by models where evolved donors are not considered .
podsiadlowski et al .
suggest that the secondaries that have undergone pre - cv evolution should show evidence for nuclear processing , in particular , products produced via the cno cycle .
two alternative paths that might allow the donor stars in cvs to acquire peculiar abundance patterns are for them to accrete such material during the original common envelope phase , or during the briefer common envelope phase that results from a classical nova eruption .
marks & sarna ( 1998 ) have investigated a large number of scenarios for the composition of the surface layers of cv donor stars that have accreted both types of material , and find that significant changes are possible .
a common theme amongst these models are large deficits of carbon . in a series of papers we have presented near - ir @xmath2-band spectra that demonstrated that the secondaries of many cvs appeared to have significant carbon deficits ( harrison et al .
2004a , 2005ab , 2009 ) .
recently , harrison & hamilton ( 2015 , `` h&h '' ) used synthetic spectra to quantify that ss cyg , ru peg , and gk per had carbon abundances that were 10 to 30% of the solar value .
in contrast , however , they found that the metallicities ( [ fe / h ] ) of the donor stars were solar , or only slightly subsolar .
the study by h&h was limited to the @xmath2-band and , as they describe , there are not many strong atomic absorption lines in that bandpass to derive more precise values of [ fe / h ] , nor enough spectral lines from any species other than carbon to allow derivation of their abundances . to remedy this issue ,
we have developed the capability to generate synthetic spectra that cover the @xmath10 bandpasses .
this allows us to investigate the abundances of other important elements for the subset of cvs we have observed using cross - dispersed spectrographs . from modeling these data
we find a wide range of values for [ fe / h ] , and for [ c / fe ] .
in addition , however , we find a number of cvs that appear to have unusual abundances of aluminum , magnesium , and/or sodium . due to the lack of an extensive line list ( and other uncertainties discussed below )
our abundance analysis has been limited to cvs whose secondaries have spectral types earlier than m0 .
recently , relationships have been developed to extract metallicities for m dwarfs . using those relationships it is possible to examine the cooler secondary stars found in short period cvs for abundance anomalies .
we describe the observations in the next section , the generation of synthetic spectra in section 3 , the results on cvs with k - dwarf secondaries in section 4 , the process and data needed to generate metallicities for m dwarfs in section 5 , discuss our results in section 6 , and present our conclusions in section 7 .
the majority of the new data presented below was obtained with spex on the irtf .
spex was used in cross - dispersed mode generating spectra spanning the wavelength range from 0.8 to 2.45 @xmath0 m . in the @xmath2-band ,
the dispersion is 5.4 / pix .
nearby a0 v stars were used for telluric correction , and these data were telluric - corrected and flux- and wavelength - calibrated using the spextool package ( cushing et al .
spextool attempts to fit , and remove the h i lines in the telluric a star . while this process can work quite well , many times there remain residuals at the positions of the strongest h i lines .
thus , the line strengths and profiles of strong h i absorption or emission features in these data should not be taken too seriously .
a log of all of our observations for previously unpublished data is presented in table [ obslog ] , and it lists the ut date , start and stop times , the length of the exposure times used for each object , the instrument used , the orbital period , the orbital phase at the mid - point of the observational sequence ( except for u gem where full orbits were covered ) , and the outburst state for each of the cvs .
we also present previously unpublished spectra obtained with nirspec on keck .
we used nirspec in low - resolution , single order mode with a 0.38 " slit .
the grating tilt was set so as to cover the wavelength region 2.04 @xmath0 m @xmath9 @xmath11 @xmath9 2.46 @xmath0 m , with a dispersion of 4.27 / pixel .
we employed the two - nod script , and used 4 minute exposure times for all of the program cvs . to correct for telluric absorption , we observed bright a0v stars located close to the program objects so as to minimize their relative differences in air mass .
these data were reduced using the idl routine redspec , specially developed for nirspec . finally , we used triplespec ( wilson et al .
2004 ) on the apache point observatory ( apo ) 3.5 m telescope to obtain phase - resolved near - ir spectra of u gem over two orbital periods , and for 50% orbital period coverage of qz ser .
triplespec is similar to spex in that it provides cross - dispersed spectra covering the near - ir , from 0.95 @xmath0 m to 2.46 @xmath0 m , at a resolution of r @xmath12 3500 .
the 1.1 " slit was used , and all exposures were four minutes in length .
these data were reduced using `` triplespectool , '' a version of spextool modified by michael cushing for use at apo .
nearby a0v stars were used to provide telluric correction . in the following we will also be analyzing the spectra of cvs
observed in just the @xmath2-band with the irtf using spex , at keck with nirspec , or at the vlt with isaac .
all of these data have been previously published in harrison et al .
( 2004a , 2005ab , 2007a , 2009 ) , and hamilton et al .
( 2011 ) . unless otherwise noted , all of the spectra have been doppler corrected using published radial velocity data and orbital ephemerides
in h&h we described the construction of a python - based program ( named `` @xmath13 '' ) that used moog ( sneden 1973 ) to generate a large grid of synthetic @xmath2-band spectra , and to autonomously search through this grid to derive the best fitting result for each program object .
as described there , the focus of that effort was in modeling the co features , and the capability to investigate the abundances of other elements was not implemented . with the desire to expand our coverage to shorter wavelengths , we decided to abandon further development of @xmath13 in favor of using `` spectrum '' by r. gray .
the main driver for this decision was the existence of a line list covering the visual and near - ir .
as discussed in h&h , the process of constructing and validating a @xmath2-band line list was quite onerous , and one we wished to avoid if possible .
as in h&h , we fixed the micro - turbulence to 2 km s@xmath14 for all synthetic spectra , a value that appears to best represent normal stars ( husser et al . 2013 , smith et al .
likewise , all of our synthetic spectra were run with log@xmath15 = 4.5 .
cesetti et al .
( 2013 ) have examined the use of moderate resolution data from the irtf spectral library to derive diagnostics for cool stars .
they found very few features that allow for accurate quantification of log@xmath15 for higher gravity stars . given the large rotation velocities of cv donors , the limited spectral resolution , and the presence of numerous emission features , it is not possible to constrain the surface gravity from the data discussed herein .
unlike @xmath13 , spectrum is not natively designed to use a multiple processor machine that greatly aids in speeding - up the process of generating large grids of models . to allow us to quickly construct large grids of synthetic spectra using spectrum
, we used the gnu shell tool `` _ parallel _ '' with our 32 core linux workstation .
we found that we could generate models as quickly with spectrum @xmath16 @xmath17 , as with @xmath13 .
the other main difference between the use of spectrum vs. @xmath13 is that the latter uses marcs atmosphere models when computing synthetic spectra , while spectrum was designed for the input of kurucz models .
thus , before we began our program we first had to insure that the results generated with spectrum matched those generated by @xmath13 ( see h&h for validation of @xmath13 ) .
we ran a small grid of models with identical parameters using both @xmath13 and spectrum .
in addition , we also translated the marcs models used for this test into the format used by kurucz to determine if the input atmosphere model resulted in significant differences .
the spectra produced by the three processes were identical as far as we could tell .
as in h&h , the next step was to validate the line list using template k dwarf spectra . for this process
we used the same set of dwarfs as in h&h ( the data are from the irtf spectral library , cushing et al
we found that in general , the `` luke.nir '' ( covering 6800 @xmath9 @xmath11 @xmath9 1 @xmath0 m ) and `` luke.ir '' ( covering 1 @xmath0 m @xmath9 @xmath11 @xmath9 4 @xmath0 m ) line lists were quite reliable .
we did adjust the oscillator strengths for some of the strongest mg i lines , but otherwise few changes were necessary . in fig .
1 we present the spectra in the @xmath10 bandpasses for the k dwarf templates with the best fitting models calculated using spectrum .
we identify the strongest atomic and molecular absorption features in each panel to show the redundancy of spectral features for each of the elements , and as a guide for when we discuss features in the spectra of the program cvs .
except for the cn bandhead near 1.1 @xmath0 m , the models reproduce all of the strongest spectral features .
the spectral regions near 1.1 @xmath0 m and 2.0 @xmath0 m are afflicted with strong telluric features that compromise the observations . as we did for the @xmath2-band in h&h , to determine the best fitting models we constructed heat maps from a @xmath18 analysis .
those for the @xmath19-band , covering the spectral types from k0v to k5v , are shown in fig .
we tabulate the best fitting value for t@xmath3 and [ fe / h ] for each of the template stars from a @xmath18 analysis of the synthetic spectral grid for each of the bandpasses in table 2 . in this table
we also list the averages from those fits , and their nominal values ( see table 2 in h&h ) . in h&h ,
the typical error bars in the determination of [ fe / h ] and t@xmath3 from the spex data using just the @xmath2-band were @xmath20 0.25 and @xmath20 250 k , respectively .
as shown in table 2 , the errors in both quantities , when averaging over the four bandpasses , are all smaller , and easily within the combined error bars of our models and the averages of the published values .
we must note that the temperature grid of the kurucz atmospheres steps in intervals 250 k , while the abundance grid at each temperature was [ fe / h ] = @xmath160.5 , @xmath160.25 , 0.0 , @xmath210.1 , @xmath210.2 , @xmath210.3 , @xmath210.5 , and @xmath211.00 ( we also interpolated between the @xmath210.5 and @xmath211.00 models to create atmospheres with [ fe / h ] = @xmath210.75 ) .
thus , there is an uneven quantization in the model values of [ fe / h ] .
the average of the differences in temperature between our models and the nominal values is only @xmath1622 k , suggesting our temperature derivations using these low resolution spectra are reliable : @xmath22 = @xmath20 110 k. for the values of [ fe / h ] , however , the average of the differences is @xmath210.11 , suggesting that we tend to derive slightly lower metallicities than the means of the published values .
thus , while the formal error on our derivation of [ fe / h ] for the templates is only @xmath22 = @xmath20 0.04 , the offset we find , and the actual error bars on the nominal values , suggests that the true standard deviation on our metallicities is at least a factor of two larger than this . in the following subsection
will use this same technique for all of the cvs with k - type secondaries .
@xmath18 minimization will be used to derive t@xmath3 , [ fe / h ] , and [ c / fe ] .
note that the @xmath18 minimization process was only run on select regions of the spectra in each of the photometric bands . as described in h&h
, we avoid regions with strong emission features intrinsic to the cv , and regions where strong telluric features compromise the calibration . in the @xmath2-band ,
the analysis was first conducted on the spectral region from 2.19 @xmath0 m to 2.29 @xmath0 m to arrive at values for t@xmath3 and [ fe / h ] .
then the region from 2.19 to 2.36 @xmath0 m was used to find the best fitting value of [ c / fe ] . if we were to include the co region into the initial @xmath18 analysis for the @xmath2-band , the results are greatly biased by the strength of the co features ( which might have a non - solar abundance ) . in the @xmath19-band the analysis region
was constrained to 1.16 @xmath9 @xmath11 @xmath9 1.26 @xmath0 m . in the @xmath23- and @xmath24-bands ,
the available spectral regions for analysis are compromised by h i and/or ca ii emission lines .
there were no cvs for which a @xmath18 analysis was possible in the @xmath23-band .
while there are gaps between the paschen series in the @xmath23-band , no strong absorption features are visible in those gaps . for most of the cvs , the @xmath18 minimization calculation for the @xmath24-band was run over the wavelength interval 1.55 @xmath9 @xmath11 @xmath9 1.67 @xmath0 m .
there were cvs , however , where the brackett series made this impossible ( e.g. , gk per ) . in all cases , we visually inspected the reasonableness of the fit of the final model to the data in the @xmath23- and @xmath24-bands that was derived from the @xmath19- and @xmath2-band analysis . for several of the cvs
it is clear that the abundances of magnesium , sodium , and/or aluminum appeared anomalous . in those cases we ran models with the best - fit values for t@xmath3 and [ fe / h ] , but with small changes in the abundances of the element that seemed peculiar .
we then derived the value for the deficit / enhancement of that element from visual inspection .
the final results for all of the objects are listed in table [ results ] . in this table , and in the remainder of the paper , cv subclasses , orbital periods and inclinations are from the online version of the ritter & kolb ( 2003 ) catalog ( `` rkcat '' ) , unless otherwise noted . in table [ results ] we list the object name , the cv subtype , p@xmath8 , t@xmath3 of the donor derived from the modeling , its metallicity ( [ fe / h ] ) , its carbon abundance ( [ c / fe ] ) , listing any other abundance anomalies in the final column . in several cases , we set the abundance to a prescribed value before modeling , when this occurs , the value in the table is followed by an `` = '' sign . using the methods described above , we now present the results from modeling the program objects .
we want to stress again that the presence of h i emission throughout the range of these spectra severely limits the @xmath18 analysis .
thus , the precision of the derived metallicities for the cvs are almost certainly poorer than those derived from the k - type templates .
the following subsections on the individual cvs are ordered by orbital period . as shown in knigge et al .
( 2011 ) , cvs with orbital periods above 5 hrs are expected to have masses of m@xmath25 @xmath12 0.6 m@xmath7 for their donor stars .
the spectral type of a main sequence star with such a mass is m0v .
thus , we will start with the assumption that all cvs with p@xmath8 @xmath4 5 hr will probably have k type donors ( though our analysis may lead to a different answer ) .
gk per is a long period ( p@xmath8 = 48.1 hr ) cv that had a classical nova eruption in 1901 ( see harrison et al .
the secondary star in this system is clearly a subgiant .
h&h modeled the @xmath2-band spectrum of gk per observed with both keck nirspec and irtf spex , finding t@xmath3 = 5100 k , [ fe / h ] = @xmath210.125 , and [ c / fe ] = @xmath211.0 .
the cross - dispersed spex data are shown in fig .
[ gkper ] . as detailed in table 3 , we find t@xmath3 = 5000 k , [ fe / h ] = @xmath210.3 , and [ c / fe ] = @xmath210.5 . as discussed in h&h , the lower s / n of the gk per data set meant more uncertain results than for their other two cvs . with the wider spectral coverage ,
we are better able to constrain the metallicity and carbon abundance , leading to slightly revised results ( but within the error bars of both results ) .
in addition to a strong deficit of carbon , we also find that all of the mg i lines ( see fig . 1 for identifications )
are weaker than they should be for this spectral type .
we conclude that the magnesium abundance in the subgiant donor is 50% of the solar value .
ey cyg is a low inclination dwarf nova of long orbital period , p@xmath8 = 11.02 hr ( echevarria et al .
echevarria et al . found a spectral type for the secondary of k0v , in agreement with that derived by kraft ( 1962 ) .
gnsicke et al .
( 2003 ) found an anomalous n v / c iv emission line ratio that suggested a deficit of carbon .
analysis of _ fuse _ and _ hst _ spectroscopy by sion et al .
( 2004 ) found a carbon abundance that was 20% solar , and a silicon abundance that was 10% of solar . from the spex data ( fig .
[ eycyg ] ) we derive t@xmath3 = 5250 k , consistent with a spectral type of k0v , and [ fe / h ] = 0.0 .
we confirm the reported carbon deficit , finding [ c / fe ] = @xmath210.5 . while the si i lines in the @xmath19-band do appear to be weaker than expected , the s / n of these data are too poor to determine an absolute silicon abundance .
ae aqr is an intermediate polar ( ip ) , a system with a highly magnetic white dwarf that is spinning much more rapidly than the orbital period : p@xmath8 = 9.88 hr , p@xmath26 = 0.55 min . harrison et al .
( 2007a ) have already published the @xmath10 spectra from spex , and noted the near - absence of the co features in the @xmath2-band .
those data , and the best fitting model , are plotted in fig .
[ aeaqr ] .
we derive a temperature of t@xmath3 = 4750 k ( @xmath12 k4 ) , [ fe / h ] = 0 , and [ c / fe ] = @xmath211.0 .
the derived temperature is in agreement with the phase - dependent spectral type range of k0 ( @xmath27 = 0.4 , 0.9 ) to k4 ( @xmath27 = 0.65 ) , found by echevarria et al .
some of this spectral type range might be due to starspots , as doppler tomography found that @xmath12 18% of the northern hemisphere of the secondary star is covered with spots ( watson et al .
except for the deficit of carbon , no other spectroscopic anomalies are evident .
ru peg , p@xmath8 = 8.99 hr , was modeled by h&h , and they found t@xmath3 = 4700 k , [ fe / h ] = 0 , and [ c / fe ] = @xmath210.75 .
the spex data are presented in fig .
[ rupeg ] .
the best fitting model has t@xmath3 = 5000 k , [ fe / h ] = @xmath210.3 , and [ c / fe ] = @xmath210.4 .
while the global metallicity in our new result is substantially lower , the absolute carbon deficit has not changed .
obviously , the values of [ fe / h ] and [ c / fe ] are intertwined , and both models suggest a large , total deficit of carbon .
it is not as clear why we now find a hotter star with a lower metallicity ( though again , within the combined error bars of both results ) . as discussed in h&h
( see their fig .
18 ) , with a limited data set , hotter models with super - solar abundances are nearly equivalent with cooler models with sub - solar abundances . here
we have the opposite trend .
the increased wavelength coverage makes the new determination more reliable . in the following
we will use the newly constrained properties of the secondary stars to estimate the distances to cvs using spectroscopic parallaxes . when possible , we compare these spectroscopic distances to those derived from astrometric parallaxes .
harrison et al . ( 2004b ) used the fine guidance sensors ( fgs ) on the @xmath28 to measured a parallax for ru peg that leads to a distance of 282 @xmath20 20 pc .
the 2mass data for ru peg , obtained during quiescence , is : ( @xmath29 ) = 0.44 , ( @xmath30 ) = 0.16 , and @xmath2 = 10.46 .
these colors are consistent with the k4 temperature we derive from the spectra .
the parallax , however , gives m@xmath31 = 3.2 , whereas a main sequence k4v has m@xmath31 = 4.48 .
the secondary star would have to be a g4v to explain the observed m@xmath31 !
the donor in ru peg is three times more luminous than a main sequence star of the same spectral type .
h&h modeled the @xmath2-band data for ss cyg and found t@xmath3 = 4700 k , [ fe / h ] = @xmath210.25 , and [ c / fe ] = @xmath210.50 .
they noted that the mg i line at 2.28 @xmath0 m was weaker than expected , but could not derive an abundance for magnesium .
modeling the @xmath10 spectra , fig .
[ sscyg ] , we find very similar results : t@xmath3 = 4750 k , [ fe / h ] = @xmath210.3 , and [ c / fe ] = @xmath210.40 . with the more expansive wavelength coverage , we conclusively find that there is a deficit of magnesium , [ mg / fe ] = @xmath210.3 .
we can use the temperature we found above , and published values that characterize the secondary star , to generate an updated spectroscopic parallax .
bitner et al . (
2007 ) conducted a radial velocity study of ss cyg and found that the secondary star was 10 to 50% larger than a main sequence dwarf , which implies an absolute magnitude that is brighter by 0.21 to 0.88 mag . in their paper
they assumed a k4.5v star . here , the temperature is closer to a k3.5v .
for such an object m@xmath31 @xmath32 4.40 , with their offsets in absolute magnitude , the limits become 3.52 @xmath9 m@xmath33 @xmath9 4.19 . with @xmath34 = 9.45 ( harrison et al .
2007b ) , the spectroscopic distance to ss cyg is 102 @xmath35 153 pc .
unfortunately , this range includes all of the published astrometric distances for this object .
the mean , 128 pc , is very similar to the distance that resulted from the revised fgs analysis = 7.41 @xmath20 0.20 mas .
note that this value for the parallax differs from that quoted in the abstract of that paper due to further analysis of the data set .
] of harrison & mcarthur ( 2016 ) .
as will be discussed below , evidence suggests that the spectral types derived from our spectroscopic observations are inclination angle - dependent . for an object with the binary inclination angle of ss cyg , @xmath36 @xmath37 45@xmath38 ( harrison et al .
2007b ) , the true spectral type of the secondary is likely to be one full subclass earlier , @xmath12 k2.5v .
therefore , with m@xmath31 @xmath32 4.2 , the resulting distance estimate , assuming an unbloated main sequence donor , is 112 pc ; in agreement with the radio parallax of miller - jones et al .
given the result for ru peg , combined with the fact that the secondary star in ss cyg has a subsolar metallicity , and deficits of both carbon and magnesium , suggests that the assumption of a main sequence luminosity for the donor is suspect .
rx and has an orbital period of 5.03 hr , and schreiber et al .
( 2002 ) suggest it has characteristics of both z cam dwarf novae ( dne ) , and vy scl stars .
the z cam subclass of dne have `` standstills , '' where the system brightens to a level well above minimum , but stays a magnitude or more below maximum .
vy scl stars , on the other hand , frequently show very low states , where the mass transfer drops to low levels , and the system fades by a magnitude or more below its typical minimum light level . vy scl stars do not show dwarf novae outbursts .
rx and clearly has dwarf novae outbursts , but also extended standstills and very low states .
the shorter orbital period of rx and suggests that its donor star may have a spectral type of m0 , or later .
we will discuss deriving the metallicities for m stars in 4 .
the line lists used for modeling the spectra of k dwarfs with spectrum are not expected to be reliable for m dwarfs . however , our main reluctance in modeling m dwarf data using spectrum ( or moog ) is the lack of the incorporation of water vapor absorption . at m0v , the water vapor absorption features are not very strong , and our continuum division process can easily remove them .
thus , we ran a grid of models that included t@xmath3 = 3750 k to attempt to constrain the nature of the secondary star in rx and . to demonstrate how well the line list works for this lower temperature , in fig .
7 we also plot our model spectra on top that of the irtf spectral template hd 19305 . there are no published values for the metallicity of hd 19305 , however , using the techniques described below ( 4 ) , we derived [ fe / h ] = 0.0 , and used this value in generating model spectra .
the resulting synthetic spectrum does a reasonable job of matching hd 19305 in the @xmath23 , @xmath19 , and @xmath2-bands , though nearly all of the lines from ca i are poorly fit by the model spectrum in these bandpasses .
the model @xmath24-band spectrum does a very poor job at reproducing the observations .
this is presumably due to ignoring molecules such as h@xmath25o .
we found that the best fit for rx and was achieved with the t@xmath3 = 3750 k model , and [ fe / h ] = @xmath210.3 , but the donor star has several abundance anomalies .
while the carbon deficit was mild , [ c / fe/ ] = @xmath210.1 , the abundances of both mg and al were significantly diminished ( e.g. , mg i at 1.173 and 1.771 @xmath0 m ; the al i doublets 1.314 and 2.11 @xmath0 m ) . the donor star in rx and could have a normal carbon abundance if the effective temperature was slightly higher . in agreement with the results of sepinsky
et al . ( 2002 ) who found a solar abundance for carbon .
sepinsky et al . also found a normal abundance for silicon , and the si i lines in the @xmath19-band for rx and appear to be at their proper strength .
the mg i and al i lines were reproduced for hd 19305 by the synthetic spectrum .
the strengths of both of these species decline with decreasing temperature , but this would require a much cooler donor star than m0 to explain their strength here . using the techniques described in 4 , the metallicity analysis procedure suggests that the donor in rx and has a spectral type of m2 , and [ fe / h ] = @xmath160.07 .
this is a later spectral type than would be expected at this orbital period .
as shown in table [ obslog ] , the secondary was close to superior conjunction , thus any irradiation would have made it appear hotter than it would on average . obviously , synthetic spectra using models with cooler effective temperatures are necessary to quantify the properties of the donor in rx and .
a cooler star would have weaker mg i and al i lines , so their deficiencies can thus be explained .
it is also important to note the strong ca ii triplet emission in the @xmath23-band .
these three lines are narrower than the nearby h i paschen series , suggesting they have an origin on the secondary star .
we will see further evidence of ca ii emission below .
such lines could be used to generate a radial velocity curve for the donor objects in systems where the absorption features can not be easily seen . over the years , we have collected @xmath2-band spectra for a large number of cvs .
most of these data were presented in harrison et al .
( 2004a , 2005ab ) .
the spectral types and carbon abundances were qualitatively examined in those papers .
here we have generated models for those objects .
the results for these objects are listed in table [ results ] ; we briefly discuss each object below . as noted earlier , having only a single bandpass of moderate resolution data leads to poorer constraints on both t@xmath3 and [ fe / h ] .
however , the results for some objects indicate very low values for the metallicities , and argue for deeper , multi - bandpass infrared spectroscopy .
some of the data analyzed below have low s / n . where the @xmath39 analysis is ambiguous ,
we have chosen the _
least _ unusual model as the final result .
sy cnc is a long period dwarf nova of the z cam subclass .
the @xmath2-band spectrum was presented in harrison et al .
( 2004a ) , where a spectral type of g2 was derived .
connon smith et al .
( 2005 ) obtained a radial velocity curve for this object and find a low mass for the secondary star ( m@xmath25 = 0.36 m@xmath7 ) , and conclude that its spectral type is near m0 . while the @xmath2-band data are of low quality , the strength of the blue ca i features , na i , ca i , and co are most consistent with t@xmath3 = 5500 k , [ fe / h ] = 0.0 , and [ c / fe ] = 0.0 . the temperature we derive corresponds to a spectral type of g8v , and is more consistent with the type of donor expected at this orbital period ( knigge 2006 ) . note that sy cnc was in outburst at the time of
this spectrum , and irradiation might result in finding a hotter temperature for the secondary star .
the visibility of the donor in @xmath2-band spectroscopy during outburst suggests that an excellent radial velocity curve and @xmath40sin@xmath36 rotation rate could be obtained for this bright object ( @xmath2 = 10.9 ) using nir spectroscopy .
such data are needed to better determine the masses of both components in this system .
harrison et al . ( 2004a ) presented the @xmath2-band spectrum of ch uma and derived a spectral type of k7 @xmath20 2 , consistent with the m0 type found by friend et al .
this is cooler than expected for this orbital period ( p@xmath8 = 8.232 hr ) .
uv spectroscopy reveals that ch uma has a large n v / c iv ratio , suggesting a deficit of carbon ( gnsicke et al .
sion et al .
( 2007 ) found that their best fit white dwarf model for the @xmath41 spectrum of ch uma indicated very low abundances for both carbon and silicon .
the @xmath39 analysis indicates that the best fit temperature is near t@xmath3 = 4750 k. such a value would be closer to that expected for this orbital period .
while the donor star appears to have a near - solar metallicity , the co features in the @xmath2-band spectrum are exceptionally weak , and our analysis finds [ c / fe ] = @xmath211.0 .
the si i lines in the @xmath2-band spectrum are not prominent , even at solar metallicities for these temperatures , but they do not appear to be unusual in the spectrum of ch uma . in harrison
( 2005a ) we presented the @xmath2-band spectra of five polars with m - type companions , and found that they all appeared to have normal carbon abundances ( see below ) .
since that time , we obtained a @xmath2-band spectrum of the very long period polar v1309 ori ( p@xmath8 = 7.98 hr ) using nirspec on keck ( howell et al .
2010 ) . we have generated a grid of synthetic model spectra and find that , assuming a solar metallicity
, the temperature is between 4750 k
@xmath9 t@xmath3 @xmath9 5000 k. hotter temperatures do not have the strong nai and ca i features seen here , while cooler temperatures would require reduced metallicities to reproduce the strength of these lines .
our result for t@xmath3 is similar to the late k - type derived by howell et al .
( 2010 ) , but is in contrast to garnavich et al .
( 1994 ) who suggested an m0v .
given the poor quality and limited spectral range of the data for v1309 ori , we can not fully exclude cooler donor stars .
co is clearly not in emission in this object , nor are their any signs of cyclotron emission ( see harrison & campbell 2015 ) .
we conclude that carbon is highly deficient no matter the spectral type .
for t@xmath3 = 4750 k , we derive [ c / fe ] = @xmath211.0 .
mg i at 2.28 @xmath0 m is also very weak .
the profile of the mg i line at 2.1 @xmath0 m is corrupted by he i emission . to reproduce
the strength of the 2.28 @xmath0 m line implies that [ mg / fe ] @xmath12 @xmath210.50 .
szkody & silber ( 1996 ) found that the n v / c iv ratio was large , suggesting a carbon deficit .
thus , there is at least one polar whose secondary star does not appear to be `` normal . ''
it is interesting that this deviation occurs in the longest period polar .
em cyg is an eclipsing z cam dwarf novae that has an orbital period of 6.982 hr .
thus the donor star should have a mid - k spectral type .
harrison et al . ( 2004a ) had a difficult time classifying its @xmath2-band spectrum , finding that some features suggested a k0v , while others were more consistent with a k5v .
this ambiguity can be understood as the best fitting family of models from our analysis have both sub - solar metallicities , and modest deficits of carbon .
godon et al .
( 2009 ) have modeled @xmath41 data for em cyg and found that nitrogen , sulphur and silicon are highly overabundant , while carbon _ appeared _ to have a solar abundance .
unfortunately , our spectrum of em cyg is too noisy to examine the strength of the si i lines , though absorption features are present at the locations of each of the strongest such lines .
v426 oph is a bright , z cam - type dwarf nova that was observed by harrison et al .
( 2004a ) .
they found an early k spectral type , consistent with the k3 type derived by hessman ( 1988 ) .
evidence is mounting that v426 oph is actually an ip ( ramsay et al .
2008 , and references therein ) .
the favored synthetic model spectrum has t@xmath3 = 4750 k , [ fe / h ] = 0.0 , and with [ c / fe ] = @xmath210.5 , producing an excellent fit to the spex data .
@xmath2-band spectra for tt crt have been presented in harrison et al .
( 2004a ) where a spectral type of k5v was found .
szkody et al .
( 1992 ) and thorstensen et al .
( 2004a ) found that the donor star in tt crt ( p@xmath8 = 6.44 hr ) is easily seen in visual spectroscopy , and appears to be a mid to late k - type star .
our best fitting model agrees with these three estimates , having t = 4500 k , and [ fe / h ] = 0.0 .
the co features are weaker than they should be at this spectral type , and we derive a carbon abundance of [ c / fe ] = @xmath210.50 . ah her is a dne with a period of 6.19 hr , and would be expected to have a mid / late - k type secondary . the @xmath2-band data for ah her was obtained with spex and , as described in harrison et al .
( 2004a ) , this dne was going into outburst at the time of these observations .
the data quality is poor , but the two best fitting families of models have either t@xmath3 = 5500 k , [ fe / h ] = 0 , and [ c / fe ] = 0 , or t@xmath3 = 4750 k , [ fe / h ] = @xmath210.75 , and [ c / fe ] = @xmath210.4 .
the hotter model matches most of the spectrum except for the ca i lines near 1.98 @xmath0 m .
given the orbital period , the lower temperature for the donor would be expected , and indicates a low metallicity secondary with a significant carbon deficit .
our result may have been compromised by irradiation and/or dilution , though the system was near inferior conjunction .
cz ori is a dne with a period of 5.25 hr .
ringwald et al .
( 1994 ) estimated that the secondary star has a spectral type of m2.5 @xmath20 1 from the presence of tio absorption . while the @xmath2-band spectrum of cz ori is poor ( see fig .
[ czori ] ) , it is clear that the co features have an unusual profile , suggesting co emission is present . using the technique described in h&h , we added co emission spectra to the models generated by spectrum and find that a donor star with t = 4250 k , [ fe / h ] = @xmath211.0 and [ c / fe ] = 0.0 , combined with a co emission spectrum broadened to a velocity of 800 km s@xmath14 , reproduces the observations .
this is only the second cv shown to have co emission features , the other being the ultra - short period system wz sge ( howell et al .
2004 , harrison 2016 ) ; several additional examples will be discussed below .
a temperature of 4250 k corresponds to a spectral type near k7 .
hotter stars fit nearly equally as well , and have larger metallicities ( e.g. , [ fe / h ] = @xmath210.75 for t@xmath3 = 4750 k ) .
cooler stars require much lower metallicities to fit these data , and also need to have carbon deficits .
if the metallicity was truly as low as indicated here , the tio features would probably have been too weak to have extracted a spectral type from optical data .
higher quality , multi - bandpass infrared spectroscopy of this object is needed .
ex dra has p@xmath8 = 5.04 hr , and baptista et al .
( 2000 ) found m@xmath25 = 0.54 m@xmath7 .
this mass for the secondary would suggest a mid - m type star , while the expected spectral type for this period would be near the k / m boundary .
harrison et al . ( 2004a ) derived a spectral type of k7 from the spex data .
they noted that the co features and the mg i lines were weaker than expected at this spectral type .
our analysis finds a best fitting model of t@xmath3 = 4000 k , [ fe / h ] = @xmath210.5 , [ c / fe ] = @xmath210.2 , and [ mg / fe ] = 0.0 . except for the carbon deficit
, there are no obvious abundance anomalies .
in our infrared spectroscopic surveys of shorter period cvs , we presented evidence that the carbon abundances in many of the secondaries with m spectral types appeared deficient .
the main issue with such analysis was that the template stars to which we were comparing the cv spectra had known spectral types , but unknown metallicities .
thus , if the spectral type of the cv donor was somewhat uncertain , comparing it to the spectra of several template stars might lead to erroneous conclusions due to the possibility that the template stars might have significantly non - solar values for [ fe / h ] .
a proper qualitative analysis requires us to be able to compare stars of known properties to those that we seek to characterize . while models for the structure of the atmospheres of m dwarfs have improved ( e.g. , baraffe et al .
2015 ) , it is still difficult to generate synthetic spectra that are adequate for abundance measures due to incomplete line lists , and the presence of water vapor features throughout the nir ( however , see nehag et al .
2012 , and bean et al .
thus , alternative methods have been sought that calibrate various spectral features for m dwarfs with known metallicities determined using their association with hotter stars , for which such derivations are simpler ( see rojas - ayala et al .
2010 , mann et al .
2013 , 2014 , and newton et al .
mann et al .
( 2013 , 2014 ) have performed a rigorous analysis and developed relationships to quantify the value of [ fe / h ] for dwarfs spanning the spectral type range k5 to m9 .
these relationships use the equivalent widths of the na i doublet at 2.20 @xmath0 m , and the ca i triplet at 2.26 @xmath0 m , along with an `` h2o index '' ( from covey et al .
2010 ) to derive [ fe / h ] from @xmath2-band spectra .
the h2o index gives the temperature dependence , and thus an estimate of the spectral type . in mann
et al . ( 2013 ) the relationship for determining [ fe / h ] using these three features for earlier spectral types ( @xmath42 m4 ) is given , while in mann et al .
( 2014 ) , a different formulation is found that covers the range m4 to m9 .
we can now derive [ fe / h ] for the various template m dwarfs that can used for comparison to the cv spectra , as well as attempt to derive [ fe / h ] directly for the donor stars . note that due to the weakness of the co features in some of the cvs , the value of the h2o index could be erroneous ( since its reddest continuum bandpass is at 2.37 @xmath43 m ) . in these cases , we can use the calibration by newton et al . for [ fe / h ]
that only uses the na i doublet .
that relationship is only valid for spectral types from m2 to m6 .
the 1@xmath22 error bars on [ fe / h ] for the newton et al .
( 2014 ) relation is @xmath20 0.12 dex , while mann et al .
( 2013 , 2014 ) claim an error of @xmath20 0.07 dex for their calibration .
there is one vitally important caveat that could dramatically alter our metallicity estimates using these techniques , and that is the dependence of the derived value of [ fe / h ] on the abundances of carbon and oxygen .
veyette et al .
( 2016 ) have used phoenix models of cool stars to attempt to probe how the derived values of [ fe / h ] using the above techniques are dependent on the c / o abundance ratio .
less carbon in the atmosphere means more h@xmath25o , leading to changes in the continuum of the cooler m dwarfs . as an example , for an object with [ c / fe ] = @xmath210.2 and [ o / fe ] = 0.0 , the derived abundance for a solar metallicity star , using the mann et al .
( 2013 ) relations with t@xmath3 = 3500 k , would be [ fe / h ] = @xmath210.15 . for t@xmath3 = 3000 k
, it would be [ fe / h ] = @xmath210.4 .
thus , if the carbon abundance is low , we could dramatically underestimate the true metallicity of the secondary star .
we will return to this subject below . with the ability to constrain [ fe / h ] for the templates and donor stars
, we can fully investigate whether the lines of any other elements are at unusual strengths in the cv secondaries . like the preceding section , we first begin with the analysis of cross - dispersed spectra for the individual program objects ordered by orbital period .
we then examine the objects with only @xmath2-band data .
unless otherwise indicated , we have used visual inspection to determine the best fitting spectral type and to note any abundance anomalies .
the results are listed in table [ results ] .
the value of t@xmath3 listed in this table uses the calibration in rajpurohit et al .
( 2013 ) to turn estimated spectra types into temperatures .
that calibration assumes solar metallicity .
ss aur is a typical dne with an orbital period of 4.39 hr , and harrison et al .
( 2005b ) found a spectral type of m4 for the donor star .
using the mann et al .
relationship gives [ fe / h ] = 0.03 , but the h2o index suggests a spectral type near m1 .
thus , the value for [ fe / h ] might be influenced by a noisy continuum at the red end of the @xmath2-band . using the newton et al .
calibration , we derive [ fe / h ] = @xmath210.05 . the donor star in ss aur appears to have a solar metallicity . in fig .
[ ssaur ] we present the spex data for ss aur , and compare it to gl 388 , for which the metallicity calculation gives [ fe / h ] = @xmath160.18 , and indicates a spectral type of m3.4 . except for the @xmath23-band , where the continuum of ss aur is poorly defined , the data presented in this figure for both objects have been continuum divided .
the co region of the spectrum for ss aur has low s / n , but suggests that the features are similar in strength to that of the m3v template .
the normality of the co features is confirmed with the @xmath2-band spectrum of ss aur obtained with nirspec ( fig .
[ ssaurkeck ] ) . except for the slightly bluer continuum of ss aur ,
the m3v template is an excellent match for ss aur .
the metallicity analysis using the nirspec data gives [ fe / h ] = @xmath210.13 , and a spectral type of m2.7 .
it is clear that ss aur was in a higher state of activity when observed with nirspec than with spex , as the he i line at 2.06 @xmath0 m is much stronger , and the donor star absorption features at 2.1 @xmath0 m are corrupted by he i emission in the nirspec data .
the aavso database indicates that ss aur went into outburst six days after the keck observations . for the spex observations , ss aur had just returned to quiescence from an outburst that had its maximum three weeks prior .
the ca i lines near 1.98 @xmath0 m appear to be very weak in the spex data for ss aur , but the telluric correction is almost certainly responsible for this deviation .
the mg i features at 1.5 and 1.7 @xmath0 m also appear to be weaker in ss aur than the template , but the depth of those lines is altered by emission lines of he i. potassium , aluminum , and manganese appear to have solar abundances .
note that the ca ii triplet in the @xmath23-band is in emission , and superposed on the tio bandhead at 8432 .
we conclude that the donor star in ss aur is completely consistent with a near - solar metallicity dwarf of spectral type m3 .
given the normality of the donor star in ss aur , we can estimate a distance to this system .
with @xmath2 = 12.00 , and m@xmath44 = 6.5 , @xmath45 = 126 pc .
the @xmath28 parallax gave @xmath45 = 167 @xmath20 9 pc .
the spectroscopic parallax is not in close agreement with the astrometric parallax . for that to occur
, requires that the secondary in ss aur to actually have the luminosity of an m2v ( the spectroscopic distance is then @xmath45 = 155 pc )
. we will encounter a similar issue for u gem .
u gem has an orbital period that is very similar to ss aur , but we will find that it has a very different secondary star .
u gem was observed using triplespec at apo over a time interval of 8.3 hrs , with a 30 minute break two thirds of the way through to obtain telluric standard data , as well as observe the m - type star gj 402 .
thus , our data set spanned almost two complete orbital periods .
the triplespectool routine uses night sky emission lines on the exposed frames for wavelength calibration .
we determined a wavelength solution for each pair of spectra , resulting in a very precise calibration of the data .
the triplespec data for u gem are presented in fig .
[ ugemspec ] . as first noted by harrison et al .
( 2000 ) , the co features are extremely weak .
the metallicity analysis leads to [ fe / h ] = @xmath160.05 , and a spectral type of m3.9 .
given the weak co features , we use the newton et al . relationship to find [ fe / h ] = @xmath160.20 . in fig .
[ ugemspec ] we also plot the data for gj 402 obtained on the same night , with the same instrument
. published spectral types for this star range from m3v to m5v .
the metallicity analysis gives a spectral type of m3.8 , and [ fe / h ] = @xmath160.27 .
thus , it should provide an excellent template for u gem .
as shown in fig .
[ ugemspec ] , the match between u gem and gj 402 in the wavelength interval 0.94 @xmath1 1.34 @xmath0 m is very good .
the only issue in this bandpass is the poor s / n of the spectrum of gj 402 through the telluric feature at 1.112 @xmath0 m .
the feh feature at 0.99 @xmath0 m shows that they both have similar metallicities .
the mg i , k i , and al i features all overlap . the two spectra do not match as well in the @xmath24-band , with the middle sections of the continuum of u gem well above that of gj 402 .
naively , one would assume this deviation is due to h i emission from the brackett series , since the doppler correction for the secondary s motion makes the h i lines appear very broad ( fwhm @xmath12 1400 km s@xmath14 ) ; but this explanation is inadequate .
first , the influence of the two strongest lines of the bracket series in the @xmath24-band , br10 ( @xmath111.74 @xmath0 m ) and br11 ( @xmath111.68 @xmath0 m ) , on the spectrum of u gem can just be discerned , as they fill - in some of the absorption features of the m dwarf .
they would have to be extremely broad ( and much stronger ) to create an excess in the continuum between these two wavelengths .
this would also require broader h i emission lines in the @xmath24-band than seen in either the @xmath19- or @xmath2-bands .
secondly , adding in a strong brackett continuum at @xmath46 1.6 @xmath0 m should raise the blue end of the @xmath24-band above that of an m4v , and this is not seen .
the excess in the @xmath24-band is not due to h i emission .
suspiciously , the excess in the @xmath24-band is centered near 1.65 @xmath0 m , the location of the peak of the h@xmath47 opacity minimum . as discussed in wing & jrgensen ( 2003 ) , and displayed in fig .
[ giantdwarf ] , lower gravity giants have larger h@xmath47 opacity `` bumps '' in the @xmath24-band , as compared to higher gravity dwarfs . in fig .
[ ugemhcomp ] , we plot the @xmath24-band spectra of u gem , gj 402 , and that of hd4408 , an m4iii . the excess due to h@xmath47 is obvious in the red giant spectrum . the strong and broad absorption due to water vapor seen in the dwarfs at the red end of the @xmath24-band , is weaker in the red giant .
the water vapor features in red giants are much weaker than those of dwarfs with the same effective temperature .
thus , the most obvious explanation for the peculiarity of the @xmath24-band spectrum is that the secondary of u gem has an _ apparent _ gravity that is much lower than that of gj 402 .
the na i doublet in the @xmath2-band was easily seen in the individual spectra , and we have used iraf to individually measure the positions of each of the lines of the doublet to allow us to construct the radial velocity curve shown in fig .
[ ugemrv ] .
@xmath39 analysis of sinewaves with various amplitudes fit to these data finds a semi - amplitude of k@xmath25 = 310 @xmath20 10 km s@xmath14 .
this result agrees with the value of 309 @xmath20 3 km s@xmath14 found by friend et al .
( 1990 ) , and k@xmath25 = 302 km s@xmath14 found by naylor et al .
( 2005 ) .
naylor et al . attempted to determine the @xmath40sin@xmath36 rotation rate of the secondary , and depending on the template they used , found values from 96 km s@xmath14 ( with an m5.5v ) , to 142 km s@xmath14 ( with an m3v ) . using the m4v template gj 402 for comparison , and working in the @xmath19-band ( 5@xmath48 order , a dispersion of @xmath49 = 1.7 )
, we find a best fit value for @xmath40sin@xmath36 = 155 @xmath20 20 km s@xmath14 .
this large value may result from the fact that gj 402 has a slightly higher metallicity than u gem .
the spectral features for a higher metallicity object are going to be slightly deeper , and thus would require a larger velocity convolution to best match the observed spectrum ( see shahbaz 1998 ) .
to test this , we used spectrum to generate synthetic @xmath19-band spectra using t@xmath3 = 3,200 k , log@xmath15 = 4.5 , and marcs models with [ fe / h ] = 0.0 and @xmath210.50
. given the uncertainties in modeling m dwarf atmospheres we limited our analysis to 1.24 @xmath50 1.26 @xmath0 m ( essentially the two k i lines ) .
a model generated using the parameters of gj 402 reproduced these two lines .
the best fit ( @xmath39 ) value for the rotation velocity of u gem using the solar metallicity model has @xmath40sin@xmath36 = 145 km s@xmath14 , while the sub - solar metallicity model gives @xmath40sin@xmath36 = 120 km s@xmath14 .
it is obviously critical to have the appropriate metallicity template when using this technique to estimate @xmath40sin@xmath36 .
thus we note that , except for gl 51 and gl 65a , the late - type templates used by naylor et al . for their @xmath40sin@xmath36 derivation have sub - solar metallicities .
gl 51 and gl 65a are slightly cooler ( t@xmath3 = 3030 k , @xmath32 m5v ) than the donor in u gem .
the closest temperature match is gj 213 , but it has the lowest metallicity of the sample ( [ fe / h ] = @xmath210.22 , mann et al . 2015 ) . convolving the spectrum of gj 463 gave their highest @xmath40sin@xmath36 estimate , but there are no measurements of [ fe / h ] for this star ( though it has a similar temperature to u gem , t@xmath3 = 3300 k , houdebine & mullan 2015 ) . given the uncertainties in this process , our result for @xmath40sin@xmath36 is certainly consistent with the conclusions of naylor et al .
this confirms their result that u gem has a massive white dwarf , m@xmath51 @xmath32 1.2 m@xmath7 .
the mass of the secondary that they derive , m@xmath25 = 0.44 m@xmath7 , is similar to those of main sequence m2/m3 dwarfs . given the parameters of the solution found by naylor et al .
( 2005 ) , the surface gravity of the secondary in u gem would be log@xmath15 = 4.8 . a value similar to those of normal m dwarfs ( cf . , fernandez et al . 2009 ) .
our larger values of k@xmath25 and @xmath40sin@xmath36 result in m@xmath25 = 0.42 m@xmath7 , and an equatorial radius of r@xmath25 = 0.58 r@xmath7 , leading to log@xmath15 = 4.5 .
this would imply a gravity at the equator that is one half that of a typical m dwarf . applying the relations in warner ( 1995 ) to u gem ,
the equatorial radius is 1.64 times larger than the volume radius `` r@xmath52 '' of the roche lobe ( the radius of the equivalent sphere ) .
thus , using our result , the surface gravity of the equivalent sphere is then log@xmath15 = 5.0 .
since u gem has a large inclination ( @xmath36 = 69@xmath38 ) , the equatorial regions of the secondary will dominate the spectrum . since these regions are further away from the center of the star , they will be cooler , leading to a cooler observed spectral type .
this is , of course , the effect of `` gravity darkening '' found for rotating stars , first quantified by von zeipel ( 1924 ) .
we explore whether we can reconcile the results for u gem with the expected levels of gravity darkening in 6.2 .
this inference would resolve the paradox from the @xmath28 parallax measurement , where the luminosity of the secondary star was equivalent to an m2v , but the photometry and spectroscopy indicated an m4v ( harrison et al .
it also explains the `` too cool for comfort '' finding of friend et al ( 1990 ) , where the observed spectral type was later than expected if the donor was to obey the main sequence mass - radius relationship . a more robust value of @xmath40sin@xmath36 for u gem is required to confirm this result .
to attempt to quantify the carbon deficit , we have continuum divided the @xmath2-band spectra of both u gem and gj 402 to remove the water vapor features . in the process , we also remove much of the broader absorption due to co , and thus these features will also be weakened by this process , resulting in an artificial deficit in carbon .
we have used _ kmoog _ to generate model spectra to determine a _ relative value _ for the carbon deficit between u gem and gj 402 . as noted above , the line lists and atmosphere models for m dwarfs
are poorly known , so the following process has considerable uncertainty .
kmoog _ was run with t@xmath3 = 3200 k , [ fe / h ] = 0 , and @xmath15 = 4.5 , using a marcs atmosphere model , over a wide range of carbon abundances ( 0.0 @xmath9 [ c / fe ] @xmath9 @xmath212.5 ) .
the best fit results are shown in fig .
[ ugemkcomp ] .
the na i doublet and
ca i triplet in the synthetic spectra do a reasonable job fitting the observations .
we find that we need to artificially have a carbon abundance of [ c / fe ] = @xmath210.5 to match the features in gj 402 , while u gem has to have [ c / fe ] = @xmath211.5 .
thus , we estimate that the c abundance for the donor star in u gem is @xmath12 10% of solar , in agreement with the value derived from uv spectroscopy of the white dwarf primary ( long & gilliland 1999 )
. one additional insight from this modeling process are the strengths of the @xmath53co features in u gem .
harrison et al . ( 2005b ) commented on the ease at which the @xmath53co@xmath54 ( @xmath112.345 @xmath0 m ) feature was seen in their spex data for u gem , given the weakness of the @xmath55co features . in fig .
[ ugemkcomp ] , we identify the locations of the @xmath55co bandheads with solid lines , and the @xmath53co bandheads with dashed lines . at the normal isotopic ratio of @xmath55co/@xmath53co = 89 used in the gj 402 model , only the @xmath53co@xmath54 bandhead at 2.345
@xmath0 m is visible . in u gem , this feature is of similar strength to the @xmath55co@xmath54 bandhead at 2.294 @xmath0 m ! the @xmath53co@xmath56 bandhead at 2.374 @xmath0 m falls at the position of a strong telluric feature that is always difficult to remove .
the next @xmath53co bandhead at 2.403 @xmath0 m ( @xmath53co@xmath57 ) , however , falls in a cleaner region , and is clearly present in u gem . in our synthetic spectrum for u gem
we used an isotopic ratio of @xmath55co/@xmath53co = 1 .
such a ratio is beyond the maximum ratio ( 4 ) that can be achieved by the cno cycle . with the uncertainties inherent to modeling m dwarf spectra , a more precise value will require better synthetic data .
we conclude that @xmath53c is highly overabundant in the donor star of u gem .
given the caveat noted earlier , assuming the results by veyette et al .
( 2016 ) are correct , the actual value for [ fe / h ] for u gem is much higher than derived from the metallicity relationships due to the lack of carbon in its spectrum .
if we assume the o abundance is normal in u gem ( leading to c / o = 0.055 ) , then we have underestimated [ fe / h ] by @xmath12 @xmath210.3 .
if this offset is true , u gem has a very high , super - solar metallicity of [ fe / h ] @xmath32 0.35 , bringing it into closer agreement with gj 402 .
the long - term aavso light curve of ww cet is unusual , showing frequent dne outbursts , periods of intermediate brightness that suggest z cam - like behavior , but also low states like vy scl .
ringwald et al .
( 1996 ) find that the behavior of ww cet is not consistent with a z cam classification , nor are its low states like the vy scl stars .
the orbital period of ww cet , p@xmath8 = 4.22 hr , is just a bit shorter than that of ss aur and u gem .
thus , the odd behavior of ww cet is puzzling .
the spex data for ww cet are presented in fig .
[ wwcet ] .
the metallicity relationship gives a spectral type of @xmath12m0v , and [ fe / h ] = @xmath210.17 , suggesting that noise at the red end of the @xmath2-band might be causing issues with the h@xmath25o index .
however , input of the same h@xmath25o index into the metallicity relationship as that for u gem leads to an identical value of [ fe / h ] . using only the na i equivalent width
we derive [ fe / h ] = @xmath210.34 .
it appears certain that the secondary in ww cet has a subsolar metallicity .
the irtf spectral library fortunately has spectra for a number of sub - solar metallicity m dwarfs in the spectral range m2v to m4v .
we compared the spectra of each of these to ww cet , and find that the best fit results from the m2.5v dwarf gl 381 .
the continuum of ww cet is slightly bluer than this star , with the excess increasing to the blue , suggesting the presence of emission from the white dwarf @xmath16 accretion disk .
the metallicity relationship gives [ fe / h ] = @xmath210.06 for gl 381 , while gaidos et al .
( 2014 ) find [ fe / h ] = @xmath210.07 .
all of the atomic absorption features in the template are stronger than their counterparts in ww cet .
this would be expected given the lower metallicity computed for ww cet .
even accounting for the difference in metallicity , the co features in ww cet are weaker than the template ( note the excess at the red end of the @xmath2-band ) .
sion et al .
( 2006 ) found that the carbon abundance of the white dwarf was 10% solar , and silicon was 50% solar .
clearly , the deficit of carbon is not as extreme in ww cet as it is in u gem .
building on the experience gained above , we estimate that the carbon abundance of the donor in ww cet is [ c / fe ] @xmath58 @xmath210.3 .
ls peg , p@xmath8 = 4.19 hr , had previously been classified as an sw sex star ( taylor et al .
1999 ) . recently , however , it has been re - classified as an ip by ramsay et al .
they believe that the lack of an x - ray beat period , a phenomenon common to ips , is due to the fact that the magnetic and spin axis of the white dwarf are aligned . the cross - dispersed spectrum of ls peg is shown in fig .
[ lspeg ] . except
, perhaps , for the na i doublet at 0.82 @xmath0 m , no absorption features from the secondary are evident .
the spectrum is dominated by h i and he i emission .
in addition , there are strong co emission features .
he ii at 2.18 @xmath0 m appears to be present , and a small peak on the red wing of h i br11 might be [ fe ii ] .
it is also clear that ca ii is in emission , as two of the members of the high order h i paschen series ( pa13 and pa15 ) are elevated above their expected levels . of the eight ips for which we have obtained nir spectra ,
none of them have had co in emission ( harrison et al .
2007a , 2010 ) .
it is now apparent , however , that ca ii emission was present in the spectrum of v603 aql ( fig . 4 in harrison et al .
the same set of unidentified emission lines in the spectrum of ls peg ( see fig . [ lspeg ] ) , were also present in the spectrum of v603 aql .
kt per is a z cam - type dwarf nova with a period of 3.90 hr ( thorstensen & ringwald 1997 ) .
they derived a spectral type of m3 for the secondary star .
the irtf data for kt per are presented in fig .
[ ktper ] .
the shorter orbital period , as compared to u gem and ss aur , would lead to the expectation of a mid - m spectral type for the donor star .
the metallicity analysis gives [ fe / h ] = @xmath210.53 , and a spectral type of m1 .
the na i relation gives [ fe / h ] = @xmath210.96 , though if the spectral type of the secondary was truly an m1v , this could possibly be erroneous ( see newton et al .
unfortunately , the irtf spectral library does not have a template with similar properties .
the closest match is gl 806 for which the metallicity relation gives [ fe / h ] = @xmath210.08 , and a spectral type of m2.2 .
mann et al .
( 2015 ) lists gl 806 as an m2.0 with [ fe / h ] = @xmath210.15 .
it is clear that the continuum of gl 806 closely matches that of kt per in the @xmath2-band . at shorter wavelengths
, there is an increasing contribution from the white dwarf @xmath16 accretion disk .
it is obvious that _ all _ of the spectral absorption features in the spectrum of kt per are much weaker than those in gl 806 .
this includes the water vapor features , confirming the very low metallicity estimate .
given this result , the co features then appear to be normal .
the ca ii triplet lines in the @xmath23-band are in absorption for the donor star . given the low level of contamination in the @xmath2-band , it is possible to estimate a distance to kt per if we assume a normal m2v .
the 2mass catalog has @xmath2 = 12.63 . to insure that this measurement reflects true quiescence , we consulted the @xmath59 single exposure source catalog .
there are 83 measurements for kt per in this database , and they show that the faintest it gets is w1(3.6 @xmath0 m ) = 12.25 . for a normal m2v , ( @xmath2 @xmath21 w1 )
= 0.15 ( pecaut & mamajek 2013 ) .
kt per is slightly redder than expected , but clearly the 2mass value must be close to quiescence .
assuming m@xmath60 = 6.05 , gives @xmath45 = 207 pc , just within the error bars on the distance derived from its parallax : @xmath45 = 180@xmath61 pc ( thorstensen et al . 2008 ) .
thorstensen et al .
note that there is a common proper motion companion to kt per which has a mid - k spectral type .
using the data in the 2mass catalog for this object , the nir colors are consistent with a k5v , and the resulting distance to this companion is 299 pc .
there is either significant contamination to the @xmath2-band luminosity of kt per , or else the secondary is 0.8 mag more luminous than a main sequence m2v , having the absolute magnitude of an m1v .
the 2mass colors suggest the contamination is low , and this is supported by the ease at which thorstensen & ringwald detected the donor in optical spectra .
unfortunately the orbital inclination for kt per is unconstrained . if we assume an offset of one spectral subtype due to inclination effects , the spectral type and distance can be reconciled if @xmath36
@xmath12 40@xmath38 .
our final short period cv with cross - dispersed spex data is tt ari , p@xmath8 = 3.30 hr .
the data are presented in fig .
[ ttari ] , and show no absorption features from a late - type secondary star . however , the ca ii triplet is in emission , and presumably originates on the donor .
if we assume that the secondary in tt ari has a spectral type of m4 , we can add a hot ( t@xmath3 = 10,000 k ) blackbody continuum source until the spectral slope in the @xmath2-band is reproduced . from this iterative process , we find that the blackbody source has 3.5@xmath62 the luminosity of the secondary in the @xmath2-band . given that tt ari has @xmath63 = 10.88 , and an m4v has m@xmath60 = 7.43 , the distance to tt ari is d @xmath12 104 pc .
tw vir has a period very similar to ss aur and u gem .
the spex data were presented in harrison et al .
( 2005a ) .
tw vir has a blue continuum , with unusually strong na i doublet and
ca i triplet features . running
the metallicity analysis for this spectrum finds a spectral type of m0v , and [ fe / h ] = @xmath210.3 .
clearly , this result is compromised by the significant contamination of the secondary by the white dwarf and accretion disk . subtracting a blackbody with t@xmath3 = 10,000 k that has 67% of the @xmath2-band flux results in a continuum that is consistent with an m1v , while subtraction of a hot blackbody that supplies 77% of the @xmath2-band flux results in a spectrum whose continuum matches that of an m4v .
the metallicity analysis on the subtracted spectra yield [ fe / h ] = 0.56 , and 1.2 , respectively .
the m1v dwarf hd 42581 in the irtf spectral library has [ fe / h ] = 0.2 , while the m4.5v dwarf gl 268ab from the same source has [ fe / h ] = 0.26 .
it is clear that both na i doublets , and the ca i triplet in the subtracted tw vir spectra are much stronger than those in either hd42581 or gl 268ab .
obviously , such high values of [ fe / h ] for tw vir are not reasonable , being as high , or higher , than seen anywhere in the galaxy . using the inferences of veyette et al .
( 2016 ) , to bring the metallicity of tw vir in line with normal stars would require an oxygen deficiency .
if [ o / fe ] = @xmath210.2 , the derived abundance for tw vir would be reduced by @xmath64[fe / h ] = @xmath210.6 .
it is not clear how to produce such a deficit without it being pre - existing . as shown by veyette et al .
, there are few such objects in the solar neighborhood . the result that we will derive for qz ser in the next section may provide the necessary insight for explaining this object . clearly , however , new higher s / n data spanning the entire nir will be essential for further progress . cw mon is a typical dne with an orbital period of 4.238 hr , and exhibits grazing eclipses ( kato et al .
cw mon has a significant x - ray luminosity ( verbunt et al .
1997 ) , suggesting that it might be an ip .
kato et al . found a possible white dwarf spin period that would aid in this classification , but such a periodicity only appears to show up during outbursts .
cw mon was observed with nirspec , and the spectrum is shown in fig .
[ cwmon ] .
the metallicity analysis indicates a spectral type of m2.3 , and [ fe / h ] = @xmath210.23 .
the irtf template gl 806 has a spectral type of m2.2v , and [ fe / h ] = @xmath210.15 , and proves to be an excellent match for cw mon .
there are no abundance anomalies , though the continuum of gl 806 appears to be slightly bluer than cw mon , indicating a slightly later spectral type for the donor . the irtf spectral template star gl 381 ( m2.9 , [ fe / h ] = @xmath210.06 ) provides a much better match to the continuum , but not to the depth of the na i and ca i absorption features .
verbunt et al .
list a distance for cw mon of 290 pc , though the origin of this `` adopted distance '' is not obvious . in the 2mass catalog ,
cw mon has ( @xmath29 ) = 0.53 , ( @xmath30 ) = 0.32 , and @xmath2 = 13.02 , roughly consistent with an m2/3 .
the aavso database shows that cw mon was at minimum light during the epoch of the 2mass observations .
if the secondary star is a normal m2v , the distance to cw mon is d = 247 pc . if the `` u gem effect '' is universal , the large inclination of cw mon , @xmath36 = 65@xmath38 ( szkody & mateo 1986 ) , would suggest a true spectral type closer to m1v , giving d = 311 pc , closer to the verbunt et al . estimate . as noted by kato et al .
, at that distance , the visual maxima of cw mon are nearly one magnitude less luminous than expected .
yy / do dra ( see patterson & eisenman 1987 for comments on its designation ) is an ip with an orbital period of 3.969 hr , and a spin period of 264.6 s ( patterson & szkody 1993 ) . like a small subset of its brethren
( see szkody et al .
2002 ) , it shows infrequent dwarf novae outbursts of short duration .
metallicity analysis on these data find a spectral type of m1.8 , and [ fe / h ] = @xmath210.20 .
the actual continuum of yy / do dra is better matched by the m2.5v template gl 381 ( m2.9 , [ fe / h ] = @xmath210.06 ) .
the nirspec data of yy / do dra are shown in fig .
[ yydra ] , along with the spectrum of gl 381 . clearly , the absorption lines in the template are stronger than those seen in yy / do dra , consistent with their difference in metallicity .
it is also obvious , however , that the co features in yy / do dra are weaker than those of the template .
we estimate that [ c / fe ] @xmath9 @xmath210.3 .
as noted in the discussion of u gem , weaker co features should lead to an underestimation of the metallicity if the conclusions of veyette et al .
( 2016 ) are correct ( and we assign the entire deficit to carbon ) . in the case of yy
/ do dra , it clearly has weak co , but the derived value of [ fe / h ] and the spectral comparison are consistent with it being lower metallicity than that of the template .
if [ c / fe ] = @xmath210.2 , the results of veyette et al . would lead to the derived metallicity being underestimated by @xmath64[fe / h ] = @xmath210.15 .
this offset would imply that both objects have the same metallicities , which is clearly not true .
the spectrum of uu aql was discussed in harrison et al .
( 2005a ) where a spectral type of m2 was found to match the continuum , but not the absorption features .
especially noteworthy was the absence of co absorption .
the metallicity analysis suggests a spectral type of k8 , and [ fe / h ] = @xmath210.34 .
the nai - only relation of newton et al . gives [ fe / h ] = @xmath210.98 .
the expectation at the orbital period of uu aql is for a mid - m type donor .
we compare the spectrum of uu aql to the m2v template gl 806 in fig .
[ uuaql ] , and it is clear that co is in emission in this system .
this explains the discrepant spectral type estimation from the h2o index .
the weakness of the blue ca i lines , and the na i doublet in uu aql , vis - - vis gl 806 , suggests that it has a much lower metallicity . if we use the h2o index of gl 806 for uu aql , the mann et al .
( 2013 ) relation gives [ fe / h ] = @xmath210.72 .
even with the uncertainty induced by co emission , the secondary star of uu aql appears to have an extremely low metallicity . in the 2mass catalog ,
uu aql has ( @xmath29 ) = 0.69 , ( @xmath30 ) = 0.24 , and @xmath2 = 13.33 , consistent with an m2v .
there are no aavso data for the epoch of these observations , but they occurred halfway between two maxima , suggesting that uu aql was near quiescence . if we assume a normal m2v donor , this suggests a distance of d = 286 pc .
sion et al .
( 2007 ) found that from modeling @xmath41 observations , they could not settle on a distance for uu aql , with both d = 150 pc , and d = 300 pc producing nearly identical results .
a model of theirs with the silicon and carbon abundances at 10% the solar value , however , gave a distance of 314 pc .
perhaps dramatically lowering the global abundances in their white dwarf / accretion disk models might lead to more robust fits .
it is important to note that woolf & west ( 2012 ) show that low metallicity m dwarfs are less luminous than more metal - rich stars . for [ fe / h ]
= @xmath210.5 , they found @xmath64m@xmath65 = @xmath160.53 .
if that offset is applied here , the distance to uu aql is reduced to 224 pc .
the orbital period of cn ori is about 30 s shorter than uu aql , so one would expect them to appear similar . the nirspec spectrum for cn ori
is shown in fig .
[ cnori ] , where we compare it to gl 806 .
the continua of the two sources match quite well .
the metallicity analysis gives the spectral type as m1.1 , and [ fe / h ] = @xmath210.49 . while more subtle than seen in uu aql , co emission is clearly present in the spectrum of cn ori , so the spectral type calculated from the h2o index will be corrupted .
if we again assume the h2o index of gl 806 , we find [ fe / h ] = @xmath210.71 for cn ori . essentially identical to uu aql .
h&h modeled several co emission scenarios and showed that when the velocity width of the emission is low , the underlying co absorption features of the secondary star appear much narrower , with small blue emission wings .
the red end of the model spectrum at the bottom of their fig .
21 looks nearly identical to that of cn ori ( see also fig .
[ czori ] ) .
as we found for uu aql , the main absorption features in the spectrum of cn ori are much , much weaker than their counterparts in gl 806 .
this supports the very low metallicity calculation .
verbunt et al .
( 1997 ) lists a distance of 295 pc .
the 2mass colors of cn ori are ( @xmath29 ) = 0.50 , ( @xmath30 ) = 0.20 , and @xmath2 = 13.10 .
the aavso light curve data shows that the 2mass observations fell in - between two outbursts .
the ( @xmath29 ) color is slightly bluer than an m2v by 0.1 mag , while ( @xmath30 ) matches such an object exactly .
the estimated distance is 257 pc .
again , due to the low metallicity of the secondary star , this may be an overestimate .
ip peg is an eclipsing cv with an orbital period of p@xmath8 = 3.797 hr .
ribeiro et al . ( 2007 ) modeled @xmath66 light curves of ip peg in quiescence and found @xmath36 = 84.5 , t@xmath25 = 3100 @xmath20 500 k , and a distance of 115 pc .
the nirspec data for ip peg is shown in fig .
[ ippeg ] , where it is compared to the m0v template hd 19305 .
metallicity analysis on the nirspec data finds a spectral type of m0.9 , and [ fe / h ] = 0.18 .
the same analysis on hd 19305 derives a spectral type of m0.7 , and [ fe / h ] = 0.02 .
thus , hd 19305 should provide a reasonable comparison , and this is confirmed in fig .
[ ippeg ] . except for the al i doublet at 2.1 @xmath0 m
, the main atomic absorption features in ip peg are slightly stronger than the template , consistent with the metallicity difference between the two objects .
the al i features might be compromised by he i emission , though the always much stronger he i emission line at 2.06 @xmath0 m is quite weak in ip peg .
the co features in ip peg appear to be marginally weaker than the template , though the telluric correction at the red end of the nirspec spectrum is not of the highest quality . as shown in table 1 , the secondary of ip peg was at inferior conjunction at the time of the nirspec observations , and the aavso light curve database shows that the system was quiescent ( though it would go into outburst 10 days later ) .
thus , it is somewhat surprising that we obtain a much earlier spectral type than found by ribeiro et al .
( @xmath12 m3.5 ) .
2mass data was obtained at @xmath27 = 0.35 , and with ( @xmath67 ) = 0.88 , suggests an m2v . with @xmath63
= 11.72 , and assuming the secondary is a normal m1v , the distance to ip peg is 172 pc . am herculis is the proto - type for the polars , and has an orbital period of 3.09 hr . as shown in campbell et al .
( 2008 ) , am her has strong cyclotron emission in the @xmath2-band during much of its orbit .
the spex data for am her presented in harrison et al .
( 2005a ) were obtained during a polar high state , and near inferior conjunction , when the cyclotron emission is at its strongest .
campbell et al .
observed am her over a full orbit during a low state using the low resolution `` prism '' mode of spex . during that time they found a spectral type of m5v
the metallicity analysis on the higher resolution spex data in harrison et al .
( 2005a ) finds a spectral type of m1.2 , and [ fe / h]= @xmath210.25 . to attempt to determine whether this metallicity is at all reliable ,
we have continuum - divided the moderate resolution spex data for am her to compare with the spectral templates .
the m4v template hip9291 ( h2o index gives m4.7 ) has [ fe / h ] = 0.15 , while gl 213 ( m4.6 ) has [ fe / h ] = @xmath210.31 .
the resulting match between the spectra of gl 213 and am her is excellent , except that the co features in am her are weaker than those in the template . running the metallicity routine on the low resolution spex data from campbell et al .
( 2008 ) at phase 0.5 , _ when the cyclotron emission is near its minimum _ , we derive a spectral type of m5.0 , and [ fe / h]= @xmath210.44 .
it is possible that spex prism mode spectroscopy is of insufficient resolution for these routines to work reliably , however , the results using such data are consistent with the those derived using the other techniques .
we conclude that secondary star in am her has a low metallicity , and probably a carbon deficit .
higher spectral resolution observations during a low state , and at phases of low cyclotron emission , are needed to confirm this result .
thorstensen ( 2003 ) obtained a parallax for am her that leads to a distance of 79@xmath68 pc .
an m5v at this distance would have @xmath2 = 13.00 , while an m4v would have @xmath2 = 11.9 .
campbell et al .
( 2008 ) present a @xmath66 light curves for am her , and from their ellipsoidal model , @xmath34 = 11.45 , and @xmath36 = 50@xmath38 .
the donor in am her has the luminosity of an m3.5v , 1.5 subtypes earlier than that derived from spectroscopy .
we lump together the results for the polars discussed in harrison et al .
( 2005b ) to examine their similarities .
there are four short period polars that were observed using nirspec on keck : ar uma , mr ser , st lmi , and vv pup .
they also observed the longer period low accretion rate polar mq dra (= sdss1553 + 5516 ) .
none of the donor stars in these systems appeared to be unusual in any way , with normal carbon abundances for their spectral types .
as shown in table [ results ] , the secondaries in all four short period polars have subsolar metallicities , while mq dra has a super - solar metallicity .
harrison & campbell ( 2015 ) discussed the field strengths for the four short period polars , and examined / modeled @xmath69 mid - ir light curve data for these systems .
ar uma has a large field strength , thus it has no cyclotron emission in the nir , and its @xmath2-band spectrum will be uncorrupted from non - thermal emission sources .
the spectral type from the h2o index , m4.5 , agrees with the classification in harrison et al .
( 2005b ) .
it is likely that there is weak cyclotron emission from the @xmath70 = 2 harmonic in the @xmath2-band for mr ser . in modeling its infrared light curve data ,
harrison & campbell ( 2015 ) used a donor with t = 2400 k ( @xmath12 m8 ) .
this contrasts with the m4.5 type derived from the metallicity analysis .
the nirspec and spex data for mr ser ( see fig . 2 in harrison et al .
2005b ) show no evidence for the short or long wavelength turnovers due to water vapor that would be expected from a late type m dwarf .
harrison & campbell ( 2015 ) present a phase - resolved set of spex prism spectra for mr ser , but their data did not cover a full orbit , and at no phase does the cyclotron emission in this system drop to very low levels like that in am her .
st lmi can have strong cyclotron emission in the @xmath2-band during part of its orbit ; fortunately the nirspec observations missed those phases .
the spectral type derived from the metallicity analysis , m6.1 , agrees with the m6v classification from campbell et al .
harrison & campbell find that the field strengths of the two poles in vv pup are b = 57 and 33 mg , thus there are no cyclotron harmonics in the @xmath2-band .
harrison et al . ( 2005b ) derived a spectral type of m7 , in agreement with that from the h2o index .
infrared light curves of mq dra are presented in szkody et al .
( 2008 ) and show that there is no evidence for cyclotron emission in the @xmath2-band .
thus , the metallicity analysis for this object should be robust . thorstensen et al .
( 2008 ) have published parallaxes for four of the five polars discussed here . for vv pup
, they found @xmath45 = 124@xmath71 pc , ar uma has @xmath45
= 86@xmath72 pc , mr ser has @xmath45 = 126@xmath73 pc , and mq dra has @xmath45 = 162@xmath74 . from harrison & campbell ( 2015 ) , @xmath75 = 13.2 , and @xmath76 = 14.0 . from szkody
et al . ( 2008 ) , @xmath77 = 13.7 , while szkody & capps ( 1980 ) found @xmath78 = 15.1 .
the absolute @xmath2 magnitudes of ar uma and mr ser ( m@xmath60 = 8.5 ) are identical to that of an m5v , while vv pup is a full magnitude less luminous , being more consistent with an m6v .
mq dra has the most luminous secondary of these four objects , m@xmath60 = 7.6 , suggesting the donor star has a luminosity slightly fainter than that of an m4v main sequence star .
the classifications derived from their luminosities are in good agreement with those found from the spectral analysis .
the donor stars in these five systems are very similar to normal main sequence stars .
the nirspec data for rz leo have been presented in howell et al .
( 2010 ) , and hamilton et al .
rz leo is a short period dne in the wz sge family .
this subclass of cvs have infrequent , but large amplitude outbursts .
both howell et al . and hamilton et al .
classify the secondary spectral type as m4 .
we have used measurements of the wavelength of the na i doublet to doppler correct the nirspec data ( see fig .
[ rzleo ] ) .
the metallicity analysis on the resulting spectrum gives a spectral type of m3.5 , and [ fe / h ] = @xmath210.44 .
however , measurement of the na i doublet and its local continuum is probably compromised by the broad h i br@xmath79 emission .
in addition , the ca i triplet has a much broader profile than it should .
comparison to the various late - type templates finds that gl 268ab ( m4.8 , [ fe / h ] = 0.26 ) provides an excellent match to the continuum and to both na i doublets of rz leo .
this suggests that the donor has a solar metallicity .
while the co features in rz leo appear to be only slightly weaker than the template , the first overtone of @xmath53co at 2.345 @xmath0 m is very strong .
the data are noisy , but suggest that the @xmath53co@xmath57 feature at 2.403 @xmath0 m is also present .
it is rare for such a short period system to have such a prominent secondary .
we suspect that there is contamination from a hot source that makes the continuum appear bluer than expected .
if so , the spectral type of the secondary is probably later than m5 , and that would imply a substantial carbon deficit .
for example , if we subtract - off a 10,000 k blackbody that supplies 30% of the @xmath2-band flux ( fig .
[ rzleo ] ) , the continuum of rz leo is very similar to that of gj 1111 ( m6.8 , [ fe / h ] = @xmath210.17 ) .
the donor in rz leo would then have a substantial subsolar abundance , and a significant carbon deficit ( note , however , that @xmath53c would not be strongly enhanced in this scenario ) .
new multi - bandpass near - infrared spectroscopy of this system would be extremely valuable .
v893 sco is an eclipsing su uma system .
hamilton et al .
( 2011 ) presented a @xmath2-band spectrum obtained using isaac on the vlt .
they derived a spectral type of m6v .
the spectrum has low s / n , but boxcar smoothing by 21 pixels , and then running the metallicity analysis , leads to a spectral type of m5.2 , and [ fe / h ] = @xmath210.82 .
there are no templates in the library with such a low metallicity .
the closest , gl 399 ( m5.0 , [ fe / h ] = @xmath210.58 ) , is not a very good match for the spectrum of v893 sco . while there is a significant co / h@xmath25o break in the spectrum of v893 sco , the absorption from na i and co are extremely weak .
hamilton et al .
( 2011 ) examined the luminosity of this cv , and found that it had an observed @xmath2-band magnitude that was 2.5 mags brighter than an m6v at the distance implied by the parallax .
the @xmath2-band spectrum of v893 sco suffers from substantial contamination .
multi - bandpass data will be needed to confirm the low metallicity of this system .
the @xmath2-band spectrum from isaac on the vlt was presented in hamilton et al .
vw hyi is an su uma dwarf novae for which hamilton et al . estimated a secondary star mass of 0.11 m@xmath7 , and derived a spectral type of m4 .
the metallicity analysis run for this spectrum finds a spectral type of m2.9 , and [ fe / h ] = @xmath210.32 .
the newton et al .
relation gives [ fe / h ] = @xmath210.34 .
there are a number of irtf templates with similar properties for comparison .
we find the best match occurs with hd95735 ( m3.0 , [ fe / h ] = @xmath210.33 ) , though gl 213 ( m4.6 , [ fe / h ] = @xmath210.31 ) provides a very similar fit .
thus , the low metallicity is confirmed , and the carbon abundance appears to be normal .
hamilton et al .
( 2011 ) presented @xmath2-band spectra for ex hya , which is a short period ip .
they derived a spectral type of m5v .
the metallicity analysis on the nirspec spectrum gives a spectral type of m2.0 , and [ fe / h ] = @xmath210.55 .
however , m2v templates , such as gl 806 , have bluer continua than ex hya . the spectral template gl 273 ( m4.2 , [ fe / h ] = @xmath210.13 ) provides a better match to the continuum , though all of the absorption features in the spectrum of ex hya are much weaker than in this object . the low metallicity template gl 299 ( m5.0
, [ fe / h ] = @xmath210.58 ) provides a good match to the strength of the na i doublet and ca i triplet absorption features , though its continuum is too red , and the co features are weaker in ex hya . assuming no contamination in the @xmath2-band
, we conclude that the donor in ex hya has spectral type near m4 , a low metallicity , and [ c / fe ] @xmath80 @xmath210.30 .
beuermann et al .
( 2003 ) obtained a parallax for ex hya using the fgs on the @xmath28 , giving a distance of 64 pc .
the 2mass catalog has ( @xmath29 ) = 0.32 , ( @xmath30 ) = 0.08 , and @xmath2 = 11.69 .
these colors are much bluer than a mid - m type star , confirming the conclusion in hamilton et al .
( 2011 ) , that there is some contamination of the @xmath2-band flux . with the astrometric distance , and the 2mass @xmath2-band magnitude , m@xmath60 = 7.76 .
this is 0.3 mag fainter than an m4v .
this suggests that the true spectral type of the donor in ex hya is closer to m5 , and contamination is responsible for the resulting continuum .
this would imply an even larger carbon deficit .
vy aqr is an su uma - type dwarf nova with an orbital period of 1.51 hr .
harrison et al . ( 2009 ) present the @xmath2-band spectrum of this source obtained with nirspec .
they showed that the continuum of the observed spectrum was most consistent with an m0v .
the metallicity analysis gives a similar answer : k9.6v ( and [ fe / h ] = @xmath160.37 ) .
harrison et al . found that analysis of archival vlt data for vy aqr suggested a spectral type closer to m6 , though the co features were much too weak for _ any _ m dwarf .
the problem with such a classification is that subtraction of a hot blackbody contaminating source does not lead to a realistic spectrum . to merely flatten the spectrum
so that it superficially resembles an m dwarf requires us to subtract off a hot source that supplies 95% of the @xmath2-band flux .
the result is a spectrum with ludicrously strong na i and ca i absorption features : [ fe / h ] = 13.9 ! _
though the data are noisy , it is clear that vy aqr does not have co in emission . _ in 5 , we will examine synthetic spectra for stars with hydrogen deficits .
those types of models appear to be required to explain unusual objects like vy aqr .
thorstensen ( 2003 ) derives an astrometric distance of 97@xmath81 pc to vy aqr .
the 2mass data for vy aqr are ( @xmath29 ) = 0.42 , ( @xmath30 ) = 0.27 , and @xmath2 = 14.59 .
there are no aavso measurements for this period , but using the _ neowise _ database , the faintest vy aqr gets is @xmath82 = 14.4 .
this suggests vy aqr was in quiescence at the time of the 2mass observations , giving m@xmath60 = 9.5 .
assigning 100% of the @xmath2-band luminosity to the donor star , implies a spectral type of m6 . while this agrees with the spectral type found by harrison et al .
( 2009 ) , the @xmath2-band spectra remain inconsistent with such a classification .
the su uma system v436 cen has a period that is nearly identical to vy aqr .
the @xmath2-band spectrum has been presented by hamilton et al .
( 2011 ) . while the s /
n of those data are not high , the secondary is quite prominent in the vlt spectrum .
hamilton et al .
estimated a spectral type of m8v for this object .
our analysis derives a spectral type of m6.9 , and [ fe / h ] = @xmath210.31 .
fortunately , the irtf spectral library has several late - type templates for comparison .
lp944 - 20 is listed as having a spectral type of m9 , but the metallicity program gives m7.7 , and [ fe / h ] = @xmath210.60 . the same analysis for gl 644c gives m7.4 , [ fe / h ] = @xmath210.32 .
the match of the latter to v436 cen is much better than the former , and suggests that the metallicity of the donor in v436 cen is low , while the c abundance is normal .
the orbital period of v2051 oph , an eclipsing su uma system , is 6.5 s shorter than v436 cen .
hamilton et al .
derive a spectral type of m7 for the donor from vlt data .
the metallicity analysis estimates a spectral type of m9.2 , and [ fe / h ] = @xmath210.53 .
there are no templates in the library with this late of spectral type , and this low of an abundance , to use for comparison .
analysis for gl 752b gives m8.6 , and [ fe / h ] = @xmath210.11 .
gl 752b provides an excellent match to the na i doublet , and to the co features .
we conclude that the noisiness of the continuum near na i is leading to both a very low value for the metallicity , and the very late spectral type . except for vy aqr , all of the donor stars for the su uma cvs with periods near 1.5 hr appear to have subsolar metallicities .
wx cet is an ultra - short period cv that appears to be an intermediary between the normal su uma dne , and the wz sge stars .
sterken et al . (
2007 ) show that wx cet has normal , short duration su uma type eruptions , but also rare superoutbursts of large amplitude .
they find that the time between short eruptions is about @xmath12200 d , while superoutbursts occur every @xmath12880 d. these recurrence times are four to five times longer than seen in the typical su uma system .
wx cet is extremely faint in the near - ir .
it does not appear in the 2mass point source catalog , though it is present in the @xmath19-band image .
aperture photometry gives @xmath83 = 17.2 .
the spectrum presented in fig .
[ wxcet ] , has been smoothed by 30 . besides the strong h i and he i emission lines , absorption from the first overtone of @xmath55co is clearly present .
the metallicity regimen assigns a spectral type of m4.9 , with [ fe / h ] = @xmath210.57 .
it is likely that the continuum is contaminated by white dwarf @xmath16 accretion disk emission . if 25% of the @xmath2-band flux is supplied by this source , the estimated spectral type is m6.5 , and the metallicity is [ fe / h ] = @xmath210.25 . if the donor provides only 50% of the @xmath2-band luminosity , the spectral type is m9.4 , and [ fe / h ] = 0.35 .
the irtf template gl 752 ( m8.6 , [ fe / h ] = @xmath210.11 ) is a reasonable match to the spectrum where the secondary supplies 75% of the @xmath2-band flux .
we conclude that the donor in wx cet is a late m - type star , and has a near - solar abundance with normal levels of carbon .
there are two objects for which we have obtained spectra for that appear to be extremely unusual cvs : qz ser and ei psc . both appear to have k - type secondary stars , while having short orbital periods . here
we analyze the spectra of these two objects .
thorstensen et al .
( 2002a ) recount the discovery of qz ser , and used time series spectroscopy to derive an orbital period of 1.996 hr .
unexpectedly , for this orbital period , the secondary appeared to have a spectral type of k4 @xmath20 2 .
they also believed that the sodium abundance was probably enhanced .
we observed qz ser with triplespec , and the data are presented in fig .
[ qzser ] . as found by thorstensen et al .
, the continuum of this spectrum is matched by that of a k5v . beyond that , the spectrum is highly unusual : the only emission lines are from he i , there is no evidence for co absorption , and the na i lines are extremely strong across all bandpasses . especially prominent is the na i doublet at 1.638 @xmath0 m , a feature that is usually lost amongst numerous absorption features found in this spectral region .
we generated synthetic spectra over a temperature range of 4000 @xmath9 t@xmath3 @xmath9 5000 k , log@xmath15 = 4.5 , and @xmath211.00 @xmath9 [ fe / h ] @xmath9 0.5 .
none of these models matched the depth of the na i , ca i , or the mg i lines . to fit these common absorption features required huge overabundances of these three elements .
after adjusting their abundances , the resulting models fit the strong lines better , but they did not reproduce the depth of the fe i lines , even in the metal - rich models . as an alternative , taking the suggestion of thorstensen et al . that the donor star might be deficient in hydrogen and have enhanced helium abundances , we ran models with _ solar abundance patterns _ , while changing the abundances of h and he .
we found the best fitting spectrum occurred when the hydrogen abundance was reduced to 30% of its normal value , and the helium abundance was increased by a factor of ten . with these changes , iron , magnesium and calcium have normal abundances .
this is not true for carbon , sodium , or aluminum .
the blue end of the @xmath24-band is sensitive to both the h and c abundances : reduce either one too much , and the continuum becomes too blue . the final model , plotted in fig .
[ qzser ] , has [ c / fe ] = @xmath211.7 ! the sodium abundance needed to explain the absorption features is 32@xmath62 the solar value .
the aluminum abundance also has to be higher to fit the data , and we found [ al / fe ] = @xmath160.5 .
none of our models fit the na i doublet at 1.638 @xmath0 m .
this doublet comes from transitions between the 2p@xmath844p@xmath6p@xmath85 and 2p@xmath846s@xmath6s@xmath86 ( we used the oscillator strengths listed in cunha et al .
these lines are more prominent in red giants ( ibid . ) , than in dwarfs , and may suggest a lower gravity for this peculiar star . obviously , we are in uncharted territory for abundance analysis . to enable us to produce realistic synthetic spectra for an object where the absolute abundances of h or he are non - solar , we need to start with a stellar atmosphere that has the correct abundance patterns .
while we believe the best solution to the difficulty in modeling the spectrum of qz ser is a hydrogen deficiency , in all of our models , the na overabundance is very large .
how could such an anomaly arise ? there are two proton capture cycles beyond the cno cycle that are likely responsible for the peculiar abundances of qz ser : the nena cycle , and the mgal cycle ( see wallerstein et al .
1997 , and references therein ) .
prantzos et al .
( 1991 ) model the production of sodium in 2 and 15 m@xmath7 stars .
they found that just before he ignition occurs , the sodium abundance in the innermost regions of the star are enhanced by a factor of five in both models .
for the lower mass star , the altered values of na were constrained to the inner 0.43 m@xmath7 of the h - exhausted core . inside this radius
, the carbon abundance would also be very low .
if the donor star of qz ser was once a red giant whose outer atmosphere has been stripped , simple stellar evolution models can probably explain the peculiar na abundances .
this would require an initial mass for the donor in qz ser that is higher than typically envisioned under the standard evolutionary paradigm for cvs .
a similar story holds for the mgal cycle .
proton capture onto mg can result in al . to operate
this cycle , however , appears to require more massive stars ( @xmath125 m@xmath7 , ventura et al .
smiljanic et al .
( 2016 ) have surveyed the na and al abundances in giants and dwarfs .
they only found large al enhancements for red giants with m @xmath4 3 m@xmath7 .
it appears that it will require an even more massive progenitor to explain the overabundance of al seen in qz ser , if it must arise from simple stellar evolution .
alternatively , the peculiar atmosphere of qz ser might be acquired during the common envelope phase of a classical novae ( cne ) eruption .
the rapid proton capture process ( the `` rp - process '' ) , during explosive hydrogen burning can produce a wide variety of nucleosynthetic products .
denissenkov et al .
( 2014 ) have shown that significant production of na or al in a cne eruption is possible , but almost certainly requires enrichment of material from an underlying one white dwarf .
these same models also produce significant amounts of carbon .
if the relative abundances of our model are even approximately correct , the accretion of cne processed material can not explain qz ser .
above , we encountered two systems , vy aqr and tw vir , that had anomalously strong metal absorption lines for their spectral types .
instead of having super - solar metallicities , it is just as likely that they suffer from hydrogen deficits .
even an object such as u gem has some parallels with qz ser .
unfortunately , hydrogen - deficient model atmospheres are not yet available for stars cooler than t@xmath3 @xmath12 10,000 k ( cf . , behara & jeffery 2006 ) ,
so further investigation of these cvs is not currently possible .
thorstensen et al .
( 2002b ) present spectra and time series photometry of ei psc and find an ultra - short period of 64 minutes , well below the period minimum for cvs with donor stars of normal composition ( see howell et al .
2001 , and references therein ) .
thorstensen et al .
believe that the he abundance is enhanced .
harrison et al .
( 2009 ) presented a @xmath2-band spectrum for ei psc , obtained with nirspec .
as discussed in harrison et al . , and thorstensen et al .
( 2002b ) , the spectral type appears to be close to a k5v .
the issue with this spectral classification , however , is that the ca i triplet is much weaker in ei psc than the na i doublet ( see fig .
[ eipsc ] ) . for k dwarfs ,
these two features have a similar strength , and it is not until the mid - m dwarfs ( @xmath12 m5v ) where the ratio of the equivalent widths for the na i and ca i features becomes as large as seen here ( 2.2 ) . like qz ser , there is no evidence for co absorption features in this spectrum . unlike qz ser , there is h i emission . if we assume the secondary has a temperature of 4500 k , the best fit model to the spectrum , has [ fe / h ] = @xmath210.2 , and [ c / fe ] = @xmath211.7 . like qz ser , na i appears to be enhanced , and we find [ na / fe ] = @xmath160.9 provides a good match to the two doublets in the @xmath2-band . while we did not need to invoke an h deficit to explain the nirspec data , it is likely that the short orbital period of this system requires it to be hydrogen deficient .
thorstensen et al .
show that if the donor had an initial mass of 1.2 m@xmath7 , and mass transfer began shortly after the end of h burning , the secondary star could evolve to short periods while maintaining a high temperature .
this model predicts a depletion of c , and enhanced levels of n in the photosphere . using our results for qz ser as a guide , if the donor in ei psc has an h deficiency of 30% like the thorstensen et al .
model , the na i enhancement would be much smaller , and the abundances of the other metals would have to be substantially subsolar .
we have used near - ir spectra to derive metallicities and effective temperatures for forty one cataclysmic variables spanning most of the major subclasses .
we have extended the program described in h&h to enable us to generate synthetic @xmath10 spectra , allowing us to model cross - dispersed spectroscopy of cvs we have obtained with spex on the irtf , or triplespec at apo .
we then constructed grids of models to compare to both cross - dispersed and @xmath2-band spectra for analysis of those cvs with k type secondary stars .
we have used the recent metallicity calibrations that employ @xmath2-band spectra to derive temperatures and values of [ fe / h ] for cvs with m - type donor stars .
these calibrations were applied to template spectra enabling the direct comparison of similar objects , bolstering the determinations of t@xmath3 , [ fe / h ] , and allowing for estimates of [ c / fe ] . as found in h&h , and implied in the data from harrison et al .
( 2004a , 2005a , b , 2009 ) , carbon deficits were found for numerous ( 21/36 ) objects .
the trend , however , is that these deficits were more likely to occur in long period cvs ( p@xmath8 @xmath4 4 hr ) , then in short period systems .
we discuss this , and the other main results in the subsections that follow .
h&h investigated whether it was possible for co emission to create apparent carbon deficits .
they found that the presence of co emission , presumably from the accretion disk , overlaying the absorption features of the secondary star creates a spectrum with unique co absorption feature profiles : blue emission wings , and dramatically narrower bandheads . using those model results , it is very simple to identify co emission even in noisy data .
we found four cvs with such emission : uu aql , ls peg , cn ori , and cz ori . with the inclusion of wz sge
, this brings the number of cvs that exhibit co emission to five . _ none _ of the other cvs discussed above show evidence for co emission .
thus , any derived carbon deficits are real , though the quality of the spectra often limited our ability to derive precise values of [ c / fe ] .
the ability to compare templates with known metallicities to cvs with m - type secondaries makes the identification of carbon deficits in the shorter period cvs more robust than previous efforts that lacked this information .
one relevant question is why do these five cvs show co emission ?
as described in h&h , co emission is believed to originate in the outer , cooler parts of the accretion disk .
we believe that there are three possible explanations .
the first is that co emission is a transient phenomenon .
the second is that the mass transfer rate in these five systems is just right to create the environment amenable to co emission . or third
, the structure of their accretion disks is somehow altered by composition , leading to conditions that fosters co emission .
only a larger spectroscopic survey of cvs can answer the first possibility , though the co emission in wz sge was present for at least a full year ( see harrison 2016 ) .
the low s / n , low resolution @xmath2-band spectrum of wz sge presented by dhillon et al .
( 2000 ) , is highly suggestive of co emission being present in wz sge as far back as 1997 .
this would rule it out as an artifact of the 2001 eruption .
given its long inter - outburst intervals and low accretion rate , one would expect the disk structure in wz sge to be relatively stable .
thus its co emission might be persistent .
we have no second epoch observations for the other sources with co emission .
it is interesting to note that the observations of cn ori and uu aql were obtained just a few days after they had returned to minimum light following outbursts .
we have a handful of other objects that have observed more than once . in h&h ,
gk per , ss cyg , and ru peg were observed twice in one week with no change in the co features .
ru peg and ss cyg were also previously observed by harrison et al .
( 2004a ) , and dhillon et al .
( 2002 ) observed ss cyg .
there is no evidence for co emission in those data .
we have three epochs of observation for ss aur , including harrison et al .
( 2005a ) , and it has never shown the signature of co emission .
it will take additional epochs of observation to ascertain whether co emission is transitory .
if we assume co emission is tied to accretion rate , the second explanation does not seem especially viable , as ls peg clearly has a very large accretion rate , while wz sge has one of the lowest .
the third explanation might be a possibility , given that the donors in cz ori , uu aql and cn ori all have large subsolar values of [ fe / h ] .
wehrse & shaviv ( 1995 ) explore the role that metal opacities have on the structure of accretion disks .
the group of objects with the most extreme values for their carbon deficits ( [ c / fe ] @xmath9 @xmath211.0 ) are a heterogeneous set : ae aqr , ch uma , v1309 ori , u gem , qz ser , vy aqr , and ei psc .
essentially one object from each of the subclasses in the survey .
ae aqr , u gem , v1309 ori and ch uma were previously shown to have deficits of carbon from uv studies .
hamilton et al .
( 2011 ) have discussed the direct correlation between weak co features in @xmath2-band spectra , and uv - derived carbon deficits .
it is tempting to suggest that like qz ser , the objects with the most extreme carbon deficits are hydrogen deficient .
in contrast to harrison et al .
( 2005b ) , we now have evidence that polars are not always `` normal , '' as both v1309 ori and am her have carbon deficits .
knigge et al .
( 2011 ) published a compilation of the distribution of cv donor spectral types vs. p@xmath8 .
they show that a reasonable fit to the distribution results from both standard evolutionary models , as well as models with different angular momentum loss prescriptions .
though the distribution of spectral types in their fig
. 15 could reasonably be summarized by saying that the spectral type of a cv donor star is m4 @xmath20 2 for p@xmath8 @xmath42 6 hr .
the origin of the spectral types plotted in their figure come from a diverse set of methodologies ( see knigge 2006 for references ) .
given that the majority of the donor stars in the current survey have subsolar metallicities , most of the spectral types in that compilation are , at best , suspect .
some of them have been shown to be wrong ( such as the l0 spectral type for vy aqr ) .
our version of this diagram ( less gk per ) is presented as fig .
[ teffvsporb ] .
we code the various subclasses using different symbols .
we use t@xmath3 in this diagram as that is quantity we derive from our synthetic spectra modeling .
we use the effective temperature scale from rajpurohit et al .
( 2013 ) to convert m spectral types found above to t@xmath3 .
while there is a general trend of shorter period cvs to have cooler donor stars , the spread at any p@xmath8 or t@xmath3 can be significant .
for example , there are six cvs with t@xmath3 = 4750 k that have orbital periods spanning the range six to ten hours .
in contrast , there are ten cvs with p@xmath8 @xmath12 4 hr , that span the range 2800 k @xmath9
t@xmath3 @xmath9 4600 k. a similar grouping is found below p@xmath8 @xmath42 2 hr , where 2500 k @xmath9 t@xmath3 @xmath9 3050 k. orbital period is a poor predictor for the _ observed _ donor star temperature . because main sequence stars have a well defined mass - radius relationship , their mean density is therefore defined .
mann et al .
( 2015 ) present the relationship for the mass range 0.1 m@xmath7 @xmath9 m @xmath9 0.7 m@xmath7 .
fortunately , metallicity plays no significant role . for late - type dwarfs , mann et al . also found that there is a fairly tight correlation between radius and temperature .
thus , for hydrogen burning main sequence stars , there is an easily defined relationship between mean density and temperature .
given that the mean density of a roche - lobe filling star is closely tied to orbital period ( @xmath87 = 107p@xmath88 gm @xmath89 , equation 2.3b in warner 1995 ) , and the mean density of the roche lobe filling star is @xmath87 = 3m/4@xmath90r@xmath91 , there are only a small number of variables involved that can act to create a range in observed temperature at any particular orbital period .
the equivalent spherical radius of the roche - lobe filling donor star , r@xmath52 , is a function of the mass ratio , and dependent on the size of the semi - major axis of the orbit . for p@xmath8 = 4.3 hr and m@xmath51 = 1 m@xmath7 , r@xmath25 = 0.5 and 0.4 r@xmath7 , at @xmath92 = 0.5 and 0.25 , respectively .
for main sequence stars , these two donor stars would differ by @xmath12 350 k. thus , a large difference in the mass ratio can have a significant influence on the observed temperature of the donor star at a fixed orbital period .
returning to the objects with periods near 4.3 hr , the mean density of the donor at this period is 6.1 gm @xmath89 . as shown in fig .
[ teffvsporb ] , a main sequence star with this density has t@xmath3 @xmath12 3900 k , hotter than that observed for any of the cv donors at this period . that the secondary stars of cvs are
cooler than the equivalent main sequence stars has been discussed repeatedly ( e.g. , echevarria 1983 , friend et al .
1990 , beuermann et al .
1998 , baraffe & kolb 2000 ) . for u gem ,
the secondary has a mass of 0.42 m@xmath7 , and a mean density of 13.3 gm @xmath89 .
a main sequence star of this mass and density should have a temperature of 3500 k ( @xmath12 m2v ) . yet
the best fitting spectral template suggests t@xmath3 = 3050 k. as discussed earlier , the @xmath28 parallax indicated that the donor in u gem has the luminosity of an m2v .
thus , it is only the _ observed _ spectral type that appears incongruent . as noted above , the large inclination of u gem ( @xmath36 = 69@xmath38 ) means that the observed luminosity is dominated by the equatorial regions of the donor , and the enlargement in this direction from the roche geometry means a lower gravity , and a cooler appearance . in the case of u gem ,
the observed spectral type is two subtypes later than expected .
ss aur , with an orbital period very similar to u gem , assuming all things being equal , should have approximately the same type donor star .
the observed spectral type was found to be identical to an m3 . as in the case of u gem ,
the @xmath28 parallax suggested that the luminosity of ss aur was equivalent to that of an m2v .
the observed spectral type would then be one subtype too late .
the rkcat lists the inclination of ss aur as @xmath36 = 38@xmath38 .
it appears that inclination is playing a significant role in what we observe ( we address contamination issues below ) .
the other non - magnetic systems near this orbital period are rx and ( p@xmath8 = 5.03 hr , t@xmath3 = 3500 k = m1.5v , @xmath36 = 51@xmath38 ) , tw vir ( p@xmath8 = 4.38 hr , t@xmath3 = 3600 k = m1v , @xmath36 = 43@xmath38 ) , cw mon ( p@xmath8 = 4.24 hr , t@xmath3 = 3300 = m2.5v , @xmath36 = 65@xmath38 ) , and ww cet ( p@xmath8 = 4.22 hr , t@xmath3 = 3300 = m2.5v , @xmath36 = 54@xmath38 ) .
if we blindly apply the offsets found from u gem and ss aur to these other systems , we would predict that rx and should have a secondary with a spectral type that is slightly earlier than m0v , tw vir should have an m0v donor , cw mon should have an m0.5v , and ww cet should have an m1v .
secondary stars with these spectral types would have temperatures between @xmath12 3600 k and 3900 k , _ in good agreement with the predictions from the relationship for main sequence stars_. kt per , which has an orbital period of 3.9 hr , was found to have an inferred secondary star spectral type that agrees with the main sequence mass - radius relationship at its orbital period ( though it has a very low metallicity ) .
the donor in am her was found to have a spectral type of m5 , while the parallax suggests the luminosity of an m3.5v .
the inclination of am her is @xmath36 = 50@xmath38 .
if inclination is the source of the later - than - expected spectral types in cvs , than we are postulating that the gravity darkening of the secondary stars is responsible for this result . the gravity darkening can be expressed as t@xmath3 @xmath93 @xmath94 , where @xmath95 is the gravity darkening exponent . for radiative stars
@xmath95 = 0.25 , while for convective stars @xmath95 = 0.08 .
this assumes that the stars are in radiative equilibrium .
if the donor in u gem is truly an m2v ( t@xmath3 = 3400 k ) , yet it appears to be an m4v ( t@xmath3 = 3050 k ) , and the derived values for the polar and equatorial gravities noted above are assumed , we derive that @xmath95 = 0.09 .
a highly unusual limb darkening law does not appear to be necessary to explain the observations .
however , the inclination of u gem is not 90@xmath38 , but 69@xmath38 .
this would imply that the actual equatorial temperature is cooler than we have found , and the limb darkening exponent would have to be larger than we have derived .
fortunately , espinosa lara & rieutord ( 2012 ) have theoretically investigated the gravity darkening for the roche - lobe filling component in semi - contact binaries . while not calculating a model with the exact properties of u gem , we can interpolate the values of the gravity darkening exponent from their fig .
4 , giving @xmath96 0.22 . for u gem
, this would imply an equatorial temperature of t@xmath3 = 2640 k. in a similar context , shahbaz ( 2003 , see also shahbaz 1998 , and bitner & robinson 2006 ) have constructed the mean spectrum of the roche - lobe filling secondary in the black hole x - ray transient system gro j1655 - 40 ( v1033 sco ) in an attempt to measure its mass ratio . for modeling this system ,
the roche surface is broken into 2048 surface elements .
for each element the surface gravity is determined , and an effective temperature calculated using the appropriate gravity darkening formulation .
the most appropriate nextgen model spectrum is then assigned to that element .
the spectra for all of the elements are then summed together to create the effective spectrum . for gro j1655 - 40 ,
@xmath36 = 70@xmath38 , and the secondary star has an apparent spectral type of mid - f .
for their models they used both of the two standard limb darkening coefficients ( @xmath95 = 0.08 , and @xmath95 = 0.25 ) .
for the smaller exponent , the temperature across the stellar surface , assuming @xmath97 t@xmath3 @xmath98 = 6336 k ( f6.5v ) , varied from 6490 k ( f5.5v ) at the poles , to 5120 k ( k1v ) at the equator .
as expected , for @xmath95 = 0.25 , the temperature range was larger : 3230 k ( m3v ) to 6780 k ( f3v ) . in the end ,
their best fitting model had @xmath97 t@xmath3 @xmath98 = 6600 k with @xmath95 = 0.08 .
we conclude that for the shorter period systems , inclination appears to be the main source of the `` cooler than expected '' spectral types .
any remaining scatter can then be attributed to chemical peculiarities in the individual objects , or the various systems could have dramatically different binary mass ratios . with analogies to qz ser , the odd abundance patterns in u gem and tw vir suggest hydrogen deficiencies , and thus one would not expect them to exactly mimic the properties of normal main sequence stars . the other main peculiarity in the p@xmath8 @xmath21 t@xmath3 diagram is the existence of a set of six cvs ( ae aqr , ch uma , v1309 ori , v426 oph , ss cyg , and ah her ) , with t@xmath3 = 4750 k , but with a large spread in orbital period . in every case , we find that these objects have significant carbon deficiencies .
the three longer period cvs of these six appear to have solar values of [ fe / h ] ( ignoring v1309 ori where we set [ fe / h ] = @xmath160.0 ) .
ss cyg and ah her have subsolar metallicities .
the most reasonable explanation for this result is that the donor star underwent nuclear evolution prior to the system formally becoming a cv , and these objects have unusual radii and/or masses .
ru peg is an example of a donor in a long period cv that is more than a magnitude more luminous than it would be if it had the radius of a main sequence star . as first shown in podsiadlowski et al .
( 2003 ) , and re - addressed by goliasch & nelson ( 2015 ) , the inclusion of evolved donors in cv population studies leads to a considerable spread in t@xmath3 at orbital periods @xmath4 5 hr .
this effect arises due to some donors being zero age main sequence stars at the time they become semi - contact binaries , while others have undergone significant nuclear processing .
our results , combined with the growing numbers of donors in long period cvs that have unusual physical properties ( e.g. , connon smith et al .
2005 , thoroughgood et al .
2005 , bitner et al .
2007 , echevarria et al .
2007 , neustroev & zharikov 2008 , rodrguez - gil et al . 2009 ) , confirm the conclusions of baraffe & kolb ( 2000 ) that cvs with evolved secondaries dominate the population .
many cv donor stars must have had initial masses that were larger than 1 m@xmath7 , and our measurements of magnesium deficiencies for several long period cvs strengthens this argument . in the preceding
we have mostly ignored the possibility of contamination influencing the determination of donor star spectral types . here
we wish to briefly explore how a hot source , such as the white dwarf , or hot spot on the accretion disk , might alter the spectral type derivation .
qualitatively , the addition of a hot blackbody source to the spectrum of a late - type star should result in the derivation of an earlier spectral type than reality .
this argument can be mostly blunted by noting that the depth of various atomic and molecular absorption features are dependent on the temperature of the underlying donor star .
thus , the match of the continua _ and _ absorption features of two objects with similar metallicities confirms they have similar spectral types . here
, however , we have the possibility of non - solar metallicities as well as non - standard abundance patterns , and thus can not blindly assume that a match of the continuum and absorption features in two objects signals concordance of temperature . to test this ,
we simply have added blackbodies with various fractions of the @xmath2-band flux to the spectrum of a m - type template star ( note that for the analysis of cvs with k - type secondaries , we have continuum divided them , removing any subtle contamination effects ) .
if the contamination is at a low level , @xmath12 25% , we find that a donor with a true spectral type of m5v could be confused with an m3v .
larger levels of contamination , however , will not result in further confusion , as the depression of the continuum due to water vapor at both ends of the @xmath2-band spectrum of an m5v remains imprinted on the continuum of the contaminated spectrum .
thus , the spectrum of a highly contaminated m5v can not be confused with templates that have spectral a type of m2v or earlier .
it is possible for those objects with very low values of [ fe / h ] to have weak water vapor features .
their continua would be easier to reproduce with larger contamination fractions .
the weakening / dilution of the spectral features that comes with a larger contamination would imply an even lower metallicity .
this suggests that several of the shortest period systems with the lowest metallicities in table 3 , such ex hya and v893 sco , are affected by this process .
but we have already identified that these two objects suffer significant contamination .
thus , the process of deriving t@xmath3 from nir spectra is surprisingly robust . the other argument that we are not finding earlier spectral types than reality comes from analysis of those objects with astrometric distances . in nearly every case
, the luminosity of the cv system in the @xmath2-band implies an earlier spectral type for the donor than found here .
there is no doubt that the @xmath2-band spectra of u gem and ss aur suffer little contamination : harrison et al .
( 2000 ) found that the contamination of the @xmath2-band luminosities of both objects was less than 5% .
yet we find that in both cases , the observed spectral type is too late to explain their observed @xmath2-band luminosities . in ss aur
, we need an m2v to explain the @xmath2-band luminosity , while we derive a donor spectral type of m3v .
an m2v is a full magnitude more luminous than an m3v , and the contaminating source would have to supply more than 70% of the observed flux . in u gem the situation is even worse , with the contaminating source having to supply over 90% of the observed luminosity . while contamination is present at some level in every cv , it certainly is not seriously affecting our results .
one of the most consistent findings from our study is the sub - solar metallicities of the ips and polars .
the only real outlier is mq dra , which appears to have a slightly super - solar value for [ fe / h ] .
mq dra (= sdss1553 + 5516 ) has been classified as a low - accretion rate polar ( larp , schwope et al .
2002 ) . from x - ray observations and
the width of the cyclotron features seen in optical spectroscopy , the derived plasma temperatures at the accretion shock region in larps are much cooler than typically found in polars ( @xmath12 1 kev , vs. @xmath99 5 kev ) .
the derived accretion rates in larps are also very low compared with normal polars ( 10@xmath100 m@xmath7 yr@xmath14 , vs. @xmath4 10@xmath101 m@xmath7 yr@xmath14 ) . in addition , no `` high states , '' periods of increased mass transfer rate that frequently occur in polars , have yet been observed for a larp .
schmidt et al .
( 2005 ) suggest that these objects are `` pre - polars , '' in that the orbital period has not yet evolved to short enough values for the secondaries in these systems to contact their roche lobes .
the cyclotron emission then results from mass transfer through a stellar wind from the donor star .
the position of mq dra in the t@xmath3 vs. p@xmath8 diagram is very unusual , having a donor spectral type that is much cooler than the non - magnetic cvs at its orbital period .
the derived spectral type is nearly as late as those seen in the shortest period polars .
the derived inclination from light curve modeling in szkody et al .
( 2008 ) indicates an inclination near @xmath36 = 35@xmath38 .
if we assume it is a normal main sequence star , and its spectral type is 1 subtype earlier than observed , it would have a radius of r@xmath25 = 0.16 m@xmath7 . to fill its roche lobe
, it would need to have a radius that was twice this size .
schmidt et al .
( 2005 ) and schwartz et al .
( 2001 ) found similar results for other larps .
the other short period polars would be expected to have temperatures near t@xmath3 = 3100 k ( m4v ) , consistent with our results ( not accounting for inclination effects ) .
thus , the suggestion that larps are pre - polars appears to be valid .
the super - solar metallicity of mq dra could also be interpreted as a sign of youth .
it would be extremely useful to obtain @xmath2-band spectra of all of the other larps to explore their metallicities , and to confirm / refine the spectral types of their donor stars .
we have ascertained the metallicity for the donor stars in a large number of cvs using nir spectroscopy . for cvs with k - type donor stars ,
temperatures and abundances have been derived through the comparison of synthetic spectra to observations . for cvs with m - type stars , the values for [ fe / h ]
were derived from recently developed relationships for field m dwarfs that employ @xmath2-band spectroscopy .
we find that the donor stars in cvs generally have subsolar metallicities .
this is consistent with results for field stars in the solar neighborhood , where the local stellar population is shown to be a mix of thin and thick disk stars ( nordstrm , et al .
an analysis of the kinematics of cvs by ak et al .
( 2015 ) derives a similar result .
the donors in longer period cvs appear to have more dramatic abundances anomalies than do the short period systems . in agreement with predictions from population synthesis modeling .
the presence of mg deficits in several long period cvs argues that the donor stars in these systems had large initial masses . a caveat for the determination of [ fe / h ] exists for cvs with m - type donors , as variations in the c / o ratio could lead to the derivation of artificially low values for [ fe / h ] ( veyette et al .
this concern is especially relevant given that many cvs have carbon deficits .
our comparison of the high s / n spectrum of u gem to that of gj 402 showed that the predicted metallicity offset implied by the observed carbon deficit produced a better agreement between the two objects .
however , application of a similar offset to the derived [ fe / h ] value for carbon - deficient yy / do dra made it _
less _ consistent with the best - fitting template .
due to the extreme carbon deficits seen in some donors , cvs could provide the data to better establish and/or calibrate the effect of the c / o ratio on the metallicity estimates for normal field stars .
this could be easily accomplished with new multi - bandpass nir spectroscopy of those cvs in table 3 with proven carbon deficits .
it is also now abundantly clear that while the donor stars in cvs may superficially resemble their main sequence cousins , many of them have undergone significant nuclear evolution prior to becoming a cv .
the inclusion of such objects into recent population studies is a welcome change , and demonstrates that stellar evolution processes alone are able to explain many of our results .
it would be extremely useful to have model atmospheres for cool stars that had both hydrogen deficiencies and enhanced levels of helium , so as to enable the generation of more realistic synthetic spectra .
new models for the outbursts of dwarf novae that employ non - solar abundance patterns and/or hydrogen deficiencies should also be considered , so as to investigate whether composition has any affect on outburst properties .
the same is true for models of classical novae eruptions , as the composition of the accreted material appears to alter the nature of these events ( shen & bildsten 2009 ) .
we close by noting that the total mass of the u gem system is larger than the chandrasekhar limit , and if the mass lost to cne eruptions is not too great , the system could become a type ia supernova . note that if the donor star in u gem is even modestly hydrogen deficient , one of the main arguments against single - degenerate cvs as snia progenitors is removed : the lack of h i emission lines in their post - deflagration spectra .
pakmor et al .
( 2008 ) have found that models for the amount of matter stripped from a main sequence donor star in an snia explosion is roughly consistent with the limits implied by observations .
hydrogen deficits in cv donors would only aid this result .
teh was partially supported by a grant from the nsf ( ast-1209451 ) .
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 .
we acknowledge with thanks the variable star observations from the aavso international database contributed by observers worldwide and used in this research .
this publication makes use of data products from neowise , which is a project of the jet propulsion laboratory / california institute of technology .
neowise is funded by the national aeronautics and space administration .
the author wishes to acknowledge s. howell and p. szkody who aided in obtaining observations of several of the targets included here .
ak , t. , bilir , s. , zd"[o]nmez , a. , soydugan , f. , soydugan , e. , et al .
2015 , apss , 357 , 72 + baptista , r. , cataln , m. s. , & costa , l. 2000 , mnras , 316 , 529 + baraffe , i. , homeier , d. , allard , f. , & chabrier , g. 2015 , a&a , 577 , 42 + baraffe , i. , & kolb , u. 2000 , mnras , 318 , 354 + bean , j. , sneden , c. , hauschildt , p. h. , & johns - krull , c. m. , 2006 , apj , 652 , 1604 + behara , n. t. , & jeffery , c. s. 2006 , baltast , 15 , 115 + beuermann , k. , harrison , t. e. , mcarthur , b. e. , benedict , g. f. , & gnsicke , b. t. 2003 , a&a , 412 , 821 + beuermann , k. , baraffe , i. , kolb , u. , & wiechhold , m. 1998 , a&a , 339 , 518 + bitner , m. a. , robinson , e. l. , & behr , b. b. 2007 , apj , 662 , 564 + bitner , m. a. , & robinson , e. l. 2006 , aj , 131 , 1712 + campbell , r. k. , harrison , t. e. , & kafka , s. 2008 , apj , 684 , 409 + cesetti , m. , pizzella , a. , ivanov , v. d. , morelli , l. , corsini , e. m. , et al .
2013 , a&a , 549 , 129 + connon smith , r. , mehes , o. , vande putte , d. , & hawkins , n. a. 2005 , mnras , 360 , 364 + covey , k. r. , lada , c. j. , roman - zuniga , c. , meunch , a. a. , forbrich , j. , & ascenso , j. 2010 , apj , 722 , 971 + cushing , m.c .
rayner , j.t .
, & vacca , w.d . , apj , 2005 , 623 , 1115 + cunha , k. , smith , v. v. , johnson , j. a. , bergemann , m. , mszros , s. , et al .
2015 , apj , 798 , l41 + cushing , m. c. , vacca , w. d. , & rayner , j. t. 2004 , pasp , 116 , 362 + denissenkov , p. a. , truran , j. w. , pignatari , m. , trappitsch , ritter , c. et al .
mnras , 442 , 2058 + dhillon , v. s. , littlefair , s. p. , marsh , t. r. , sarna , m. j. , & boakes , e. h. 2002 , a&a , 393 , 611 + dhillon , v. s. , littlefair , s. p. , howell , s. b. , ciardi , d. r. , et al .
2000 , mnras , 314 , 826 + echevarria , j. , connon smith , r. , costero , r. , zharikov , s. , & michel , r. 2008 , mnras , 387 , 1564 + echevarria , j. , michel , r. , costero , r. , & zharikov , s. 2007 , a&a , 462 , 1069 + echevarria , j. , 1983 rmxaa , 8 , 109 + espinosa lara , f. , & rieutord , m. 2012 , a&a , 547 , 32 + fernandez , j. m. , latham , d. w. , torres , g. , everett , m. e. , mandushev , g. , et al .
2009 , apj , 701 , 764 + friend , m. t. , martin , j. s. , smith , r. c. , & jones , d. h. p. 1990 , mnras , 246 , 637 + gaidos , e. , mann , a. w. , lpine , s. , buccino , a. , james , d. , et al .
2014 , mnras , 443 .
2561 + gnsicke , b. t. , szkody , p. , de martino , d. , beuermann , k. , long , k. s. , et al .
( 2003 ) , apj , 594 , 443 + garnavich , p. m. , szkody , p. , robb , r. m. , zurek , d. r. , & hoard , d. w. 1994 , apj , 435 , 141 + godon , p. , sion , e. m. , barrett , p. e. , & linnell , a. p. 2009 , apj , 699 , 1229 + goliasch , j. , & nelson , l. 2015 , apj , 809 , 80 + gonzalez , g. , g. , & wallerstein , g. 2000 , aj , 119 , 1839 + hamilton , r. t. , harrison , t. e. , tappert , c. , & howell , s. b. 2011 , apj , 728 , 16 + harrison , t. e. 2016 , apj , 816 , 4 + harrison , t. e. , & mcarthur , b. e. 2016 , aas 227 , 239.07 + harrison , t. e. , & hamilton , r. t. 2015 ( h&h ) , aj , 150 , 142 + harrison , t. e. , & campbell , r. k. 2015 , apjsupp , 219 , 32 + harrison , t. e. , bornak , j. , mcarthur , b. e. , & g. f. benedict 2013 , apj , 767 , 7 + harrison , t. e. , bornak , j. , rupen , m. p. , & howell , s. b. 2010 , apj , 710 , 325 + harrison , t. e. , bornak , j. , howell , s. b. , mason , e. , szkody , p. , et al .
2009 , aj , 137 , 4061 + harrison , t. e. , campbell , r. k. , howell , s. b. , cordova .
f. a. , & schwope , a. d. 2007a , apj , 656 , 444 + harrison , t. e. , howell , s. b. , szkdoy , p. , & cordova , f. a. 2007b , aj , 133 , 162 + harrison , t. e. , howell , s. b. , szkody , p. , & cordova , f. a. 2005a , apj , 632 , l123 + harrison , t. e. , osborne , h. l. , & howell , s. b. 2005b , aj , 129 , 2400 + harrison , t. e. , osborne , h. l. , & howell , s. b. 2004a , aj , 127 , 3493 + harrison , t. e. , johnson , j. j. , mcarthur , b. e. , benedict , g. f. , szkody , p. 2004b
, aj , 127 , 460 + harrison , t. e. , mcnamara , b. j. , szkody , p. , & gilliland , r. l. 2000 , aj , 120 , 2649 + hessman , f. v. , 1988 , a&as , 72 , 515 + houdebine , e. r. , & mullan , d. j. 2015 , apj , 801 , 106 + howell , s. b. , harrison , t. e. , szkody , p. , & silvestri , n. m. 2010 , aj , 139 , 1771 + howell , s. b. , harrison , t. e. , & szkody , p. 2004
, apj , 602 , l49 + howell , s. b. , nelson , l. a. , & rappaport , s. 2001 , apj , 550 , 897 + husser , t. -0 .
, wende - von berg , s. , dreizler , s. , homeier , d. , reiners , a. , et al .
2013 , a&a , 553 , 6 + iben , i. , & livio , m. 1993 , pasp , 150 , 1373 + kato , t. , uemura , m. , kiyota , s. , tanabe , k. , koizumi , m. et al .
2003 , pasj , 55 , 489 + kolb , u. , king , a. r. , & ritter , h. 1998 , mnras , 298 , 29 + knigge , c. , baraffe , i. , & patterson , j. 2011 , apjsupp , 194 , 28 + knigge , c. 2006 , mnras , 373 , 484 + kraft , r. p. 1962 ,
apj , 135 , 408 + long , k. s. , & gilliland , r. l. 1999 , apj , 511 , 916 + mann , a. w. , feiden , g. a. , gaidos , e. , boyajian , t. , & von braun , k. 2015 , apj , 804 , 64 + mann , a. w. , deacon , n. r. , gaidos , e. , ansdell , m. , brewer , j. m. , et al .
2014 , aj , 147 , 160 + mann , a. w. , brewer , j. m. , gaidos , e. , lepine , s. , & hilton , e. j. 2013 , aj , 145 , 52 + marks , p. b. , & sarna , m. j. 1998 , mnras , 301 , 699 + miller - jones , j. c. a. , sivakoff , g. r. , knigge , c. , krding , e. g. , templeton , m. , et al . 2013 , sci , 340 , 950 + naylor , t. , allan , a. , & long , k. s. 2005 , mnras , 361 , 1091 + newton , e. r. , charbonneau , d. , irwin , j. , berta - thompson , z. k. , rojas - ayala , b. , et al .
2014 , aj , 147 , 20 + nehag , a. , heiter , u. , gustafsson , b. , piskunov , n. , plez , b. , & reiners , a. 2012 , a&a , 542 , 33 + neustroev , v. v. , & zharikov , s. 2008 , mnras , 386 , 1366 + nordstm , b. , mayor , m. , andersen , j. , holmberg , j . , pont , f. , et al .
2004 , a&a , 418 , 989 + pakmor , r. , rpke , f. k. , weiss , a. , & hilllebrandt , w. 2008 , a&a , 489 , 943 + parsons , s. g. , marsh , t. r. , copperwheat , c. m. , dhillon , v. s. , littlefair , s. p. , et al .
2010 , mnras , 402 , 2591 + patterson , j. 1998 , pasp , 110 , 1132 + patterson , j. , & szkody , p. 1993 , pasp , 105 , 1116 + patterson , j. , & eisenman , n. 1987 , ibvs 3079 + pecaut , m. j. , & mamajek , e. e. 2013 , apjs , 208 , 9 + podsiadlowski , ph .
, han , z. , & rappaport , s. 2003 , mnras , 340 , 1214 + politano , m. , & weiler , k. p. 2007 , apj , 665 , 663 + prantzos , n. , coc , a. , & thibaud , j. p. 1991
, apj , 379 , 729 + rajpurohit , a. s. , rehl , allard , f. , homeier , d. , schultheis , m. , et al .
2013 , a&a , 556 , 15 + ramsay , g. , wheatley , p. j. , norton , a. j. , hakala , p. , & baskill , d. 2008 , mnras , 387 , 1157 + ribeiro , t. , baptista , r. , harlaftis , e. t. , dhillon , v. s. , & rutten , r. g. m. 2007 , a&a , 474 , 213 + ringwald , f. a. , thorstensen , j. r. , honeycutt , r. k. , & smith , r. c. 1996 , aj , 111 , 2077 + ringwald , f. a. , thorstensen , j. r. , & hamwey , r. m. 1994 , mnras , 271 , 323 + ritter , h. , & kolb , u. 2003 , a&a , 404 , 301 + rodrguez - gil , p. , torres , m. a. p. , gnsicke , b. t. , muoz - darias , t. , steeghs , d. , et al .
2009 , a&a , 496 , 805 + rojas - ayala , b. , covey , k. r. , muirhead , p. s. , & lloyd , j. p. 2010
, apjl , 720 , l113 + schmidt , g. d. , szkody , p. , vanlandingham , k. m. , anderson , s. f. , barentine , j. c. , et al .
2005 , apj , 630 , 1037 + schwartz , r. , schwope , a. d. , & staude , a. 2001 , a&a , 374 , 189 + schwope , a. d. , brunner , h. , hambaryan , v. , schwarz , r. , staude , a. , et al .
2002 , in asp conf .
261 , the physics of cataclysmic variables and related objects , ed .
gnsicke , k. beuermann , & k. reinsch ( san francisco : asp ) , 102 + schreiber , m. r. , gnsicke , b. t. , & mattei , j. a. 2002 , a&a , 384 , 6 + sepinski , j. f. , sion , e. m. , szkody , p. , & gnsicke , b. t. 2002 , apj , 574 , 937 + shahbaz , t. 2003 , mnras , 339 , 1031 + shahbaz , t. 1998 , mnras , 298 , 153shen , k. j. , & bildsten , l. 2009 , apj , 692 , 324 + sion , e. m. , godon , p. , cheng , f. , & szkody , p. 2007 , aj , 134 , 886 + sion , e. m. , cheng , f. h. , godon , p. , & szkody ,
p. 2004 , revmex , 20 , 194 + sion , e. m. , winter , l. , urban , j. a. , tovmassian , g. h. , zharikov , s. , et al . 2004 , aj , 128 , 1795 + smiljanic , r. , romano , d. , bragaglia , a. , donati , p. , magrini , l. , et al .
2016 , a&a , 589 , 115 + smith , v. v. , cunha , k. , shetrone , m. d. , meszaros , s.,allende prieto , c. , et al .
2013 , 765 , 16 + sneden , c. 1973 , phd thesis , univ .
texas + sterken , c. , vogt , n. , schreiber , m. r. , uemura , m. , & tuvikene , t. 2007 , a&a , 463 , 1053 + szkody , p. , linnell , a. p. , campbell , r. k. , plotkin , r. m. , harrison , t. e. , et al .
2008 , apj , 683 , 967 + szkody , p. , nishikida , k. , erb , d. , mukai , k. , hellier , c. , et al .
2002 , aj , 123 , 413 + szkody , p. , & silber , a. 1996 , aj , 112 , 289 + szkody , p. , williams , r. e. , margon , b. , howell , s. b. , & mateo , m. 1992 , apj , 387 , 357 + szkody , p. , & mateo , m. 1986 , apj , 301 , 286 + szkody , p. , & capps , r. w. 1980 , aj , 85 , 882 + taylor , c. j. , thorstensen , j. r. , & patterson , j. 1999 , pasp , 111 , 184 + thoroughgood , t. d. , dhillon , v. s. , steeghs , d. , watson , c. a. , buckley , d. a. h. , et al .
2005 , mnras , 357 , 881 + thorstensen , j. r. , lepine , s. , & shara , m. 2008 , aj , 136 , 2107 + thorstensen , j. r. , fenton , w. h. , & taylor , c. j. 2004a , pasp , 116 , 300 + thorstensen , j. r. 2003 , aj , 126 , 3017 + thorstensen , j. r. , fenton , w. h. , patterson , j. , kemp , j. , halpern , j. , et al . 2002a , pasp , 114 , 1117 + thorstensen , j. r. , fenton , w. h. , patterson , j. o. , kemp , j. , krajci , t. et al . 2002b , apj , 567 , l49 + thorstensen , j. r. , & ringwald , f. a. 1997 , pasp , 109 , 483 + ventura , p. , di criscienzo , m. , carini , r. , & dantona , f. 2013 , mnras , 431 , 3642 + verbunt , f. , bunk , w. h. , ritter , h. , & pfeffermann , e. 1997 , a&a , 327 , 602 + verbunt , f. , & zwaan , c. 1981 , a&a , 100 , 7 + veyette , m. j. , muirhead , p. s. , mann , a. w. , & allard , f. 2016 , arxiv160504904 + von zeipel , h. 1924 , mnras , 84 , 665 + wallerstein , g. , iben , i. , parker , p. , boesgaard , a. m. , hale , g. m. 1997 , rvmp , 69 , 995 + warner b. , 1995 , cataclysmic variable stars .
cambridge univ . press ,
cambridge , p 33 + watson , c. a. , dhillon , v. s. , shahbaz , t. 2006 , mnras , 368 , 637 + wehrse , r. & shaviv , g. 1995 , in asp conf .
78 , astrophysical applications of powerful new databases , ed .
s. j. adelman & w.l .
wiese ( san francisco , ca : asp ) , p301 + wilson , j. c. , henderson , c. p. , herter , t. l. , matthews , k. , skrutskie , m. f. , et al .
2004 , spie , 5492 , 1295 + wing , r. f. , & jrgensen , u. g. 2003 , jaavso , 31 , 110 + woolf , v. m. , & west , a. w. 2012 , mnras , 422 , 1489 + lccccccc gk per & 2004 - 08 - 15 & 13:53 to 14:19&240s&spex & 48.1&0.94 & min .
+ ey cyg & 2004 - 08 - 16 & 11:05 to 11:57&180s&spex & 11.02&0.97 & min .
+ ae aqr & 2004 - 08 - 15 & 08:05 to 09:48 & 90s&spex & 9.88&0.61 & min .
+ ru peg & 2004 - 08 - 15 & 10:25 to 10:59&240s&spex & 8.99&0.85 & min
. + ss cyg & 2004 - 08 - 15 & 08:55 to 09:20&120s&spex & 6.60&0.40 & min .
+ cz ori & 2005 - 02 - 17 & 05:00 to 05:20&180s&nirspec & 5.25&0.66 & nearing min .
+ rx and & 2004 - 08 - 15 & 13:07 to 13:41&240s&spex & 5.03&0.59 & min . + ss aur & 2005 - 09 - 01 & 15:02 to 15:26&240s&spex & 4.39&0.12 & min . + ss aur & 2007 - 03 - 05 & 06:00 to 07:18&120s&nirspec & 4.39&0.99 & min . + u gem & 2012 - 02 - 03 & 02:22 to 07:51&240s&tspec & 4.25&&min . + u gem & 2012 - 02 - 03 & 08:40 to 10:34&240s&tspec & 4.25&&min .
+ cw mon & 2005 - 02 - 17 & 05:29 to 05:40&180s&nirspec & 4.24&0.28&min .
+ ww cet & 2004 - 08 - 15 & 12:20 to 12:54&240s&spex & 4.22&0.60&min .
+ ls peg & 2004 - 08 - 15 & 11:34 to 12:08&240s&spex & 4.19&0.68&min .
+ ip peg & 2003 - 09 - 06 & 06:39 to 07:00&120s&nirspec & 3.98&0.03&min .
+ do dra & 2005 - 02 - 17 & 09:50 to 09:59&240s&nirspec & 3.97&0.86&min .
+ cn ori & 2005 - 02 - 17 & 05:44 to 05:49&180s&nirspec & 3.92&0.33&min .
+ kt per & 2004 - 08 - 16 & 14:31 to 14:56&240s&spex & 3.90&0.16&intermed .
+ tt ari & 2004 - 08 - 15 & 14:56 to 15:22&240s&spex & 3.30&0.08&min . + qz ser & 2012 - 02 - 03 & 11:08 to 12:45&240s&tspec & 2.00&0.67&@xmath102 = 15.6 + [ obslog ] lccclll gk per & ip & 47.92 & 5000 & @xmath210.3 & @xmath210.5 & [ mg / fe ] = @xmath210.3 + ey cyg & ug & 11.02 & 5250 & @xmath160.0 & @xmath210.5 & + ae aqr & ip & 9.88 & 4750 & @xmath160.0 & @xmath211.0 & + sy cnc & zc & 9.18 & 5500 & @xmath160.0 & @xmath160.0 & + ru peg & ss & 8.99 & 5000 & @xmath210.3 & @xmath210.4 & [ mg / fe ] = @xmath210.14 + ch uma & ug & 8.24 & 4750 & @xmath160.0 & @xmath211.0 & + v1309 ori&am & 7.98 & 4750 & @xmath160.0=&@xmath211.0 & [ mg / fe ] = @xmath210.5 + em cyg & zc & 6.98 & 4500 & @xmath210.5 & @xmath210.2 & + v426 oph & zc & 6.85 & 4750 & @xmath160.0 & @xmath210.5 & + ss cyg & ug & 6.60 & 4750 & @xmath210.3 & @xmath210.4 & [ mg / fe ] = @xmath210.3 + tt crt & ug & 6.44 & 4500 & @xmath160.0 & @xmath210.5 & + ah her & ug / zc & 6.19 & 4750 & @xmath210.7 & @xmath210.4 & + cz ori & ug & 5.25 & 4250 & @xmath211.0 & @xmath160.0= & + ex dra & ug & 5.04 & 4000 & @xmath210.5 & @xmath210.2 & [ mg / fe ] = @xmath160.0 + rx and & zc ? & 5.03 & 3500 & @xmath160.1 & @xmath160.0 & + mq dra & am & 4.39 & 2800 & @xmath160.2 & @xmath160.0 & + ss aur & ug & 4.39 & 3100 & @xmath210.1 & @xmath160.0 & + tw vir & ug & 4.38 & 3600 & @xmath160.6 & & + u gem & ug & 4.25 & 3050 & @xmath160.2 & @xmath211.0 & + cw mon & ug & 4.24 & 3300 & @xmath210.2 & @xmath160.0 & + ww cet & ug & 4.22 & 3300 & @xmath210.3 & @xmath1030.3 & + yy / do dra&ip & 3.97 & 3200 & @xmath210.2 & @xmath1040.3 & + uu aql & ug & 3.92 & 3375 & @xmath210.7 & & + cn ori & ug & 3.92 & 3375 & @xmath210.7 & & + kt per & zc & 3.90 & 3580 & @xmath210.5 & @xmath160.0 & + ip peg & ug & 3.80 & 3625 & @xmath160.2 & @xmath90.0 & + am her & am & 3.09 & 2900 & @xmath210.4 & @xmath90.0 & + qz ser & su & 2.00 & 4530 & @xmath160.0=&@xmath211.7 & [ na / fe ] = @xmath161.5 , [ al / fe ] = @xmath160.5 + ar uma & am & 1.93 & 2925 & @xmath210.5 & @xmath160.0 & +
st lmi & am & 1.90 & 2775 & @xmath210.4 & @xmath160.0 & + mr ser & am & 1.89 & 3000 & @xmath210.2 & @xmath160.0 & + rz leo & wz & 1.83 & 2925 & @xmath210.3 & @xmath90.0 & + v893 sco & su & 1.82 & 2860 & @xmath210.8 & & + vw hyi & su & 1.78 & 3050 & @xmath210.3 & @xmath160.0 & + vv pup & am & 1.67 & 2690 & @xmath210.3 & @xmath160.0 & + ex hya & ip & 1.64 & 3050 & @xmath210.6 & @xmath1040.3 & + vy aqr & su & 1.51 & 2800 & @xmath160.4 & @xmath80 @xmath211.0 & + v436 cen & su & 1.50 & 2670 & @xmath210.3 & @xmath160.0 & + v2051 oph&su & 1.50 & 2500 & @xmath210.1 & @xmath160.0 & + wx cet & su & 1.40 & 2600 & @xmath160.0 : & & + ei psc & su & 1.07 & 4530 & @xmath210.2 & @xmath211.7 & [ na / fe ] = @xmath160.9 + [ results ] analysis of a grid of synthetic spectra compared to those of the k dwarf templates ( k0v top left , k5v bottom right , k7v not shown ) for the @xmath19-band .
the red `` x '' in each panel marks the location of the nominal value for t@xmath3 and [ fe / h ] .
the line segments that pass through the vertex of the each of the xs are not error bars , but display the full range of published measurements for each star ( see table 2 for their formal error bars ) . in this heat map ,
blue denotes @xmath105 @xmath9 0.7 , green is 0.7 @xmath9 @xmath105 @xmath42 0.9 , yellow is 0.9 @xmath9 @xmath105 @xmath42 1.5 , magenta is 1.5 @xmath9 @xmath105 @xmath42 2.5 , and red is @xmath105 @xmath58 2.5.,title="fig:",width=453 ] -band ( top panel ) and in the @xmath19-band ( bottom panel ) . for hd 19305 a model spectrum with t@xmath3 = 3750 k and [ fe / h ] = 0.0 is plotted in red . for rx and , the synthetic spectrum also has t@xmath3 = 3750 k , but [ fe / h ] = @xmath210.3.,title="fig:",width=453 ] -band spectrum of cz ori ( black ) with a co emission spectrum with ( @xmath40sin@xmath36 = 800 km s@xmath14 ) that has been added to a synthetic spectrum ( red ) with t@xmath3 = 4250 k , and [ fe / h ] = @xmath211.0,title="fig:",width=453 ] -band spectrum for u gem ( black , top panel ) , and gj 402 ( black , bottom panel ) .
we plot a synthetic spectra ( red ) with t@xmath3 = 3200 k and [ fe / h ] = 0.0 , in black in both panels .
the vertical solid lines are the location of the @xmath55co bandheads , while the vertical dashed lines are for the @xmath53co features.,title="fig:",width=453 ] co bandheads are indicated by vertical magenta lines . [ fe ii ] is in red , the ca ii triplet in cyan , and the na i doublet at 0.82 @xmath0 m in blue .
unidentified emission lines are indicated with a `` ? '' . , title="fig:",width=453 ] vs. p@xmath8 relationship for cvs using the results in table [ results ] .
u gem - type dwarf novae are plotted as filled circles , while su uma type dwarf - novae are plotted as open circles .
z cam - type dwarf novae are plotted with star symbols .
polars are plotted as filled squares , and ips as open squares .
the solid line was constructed using the mass - radius relationship for late - type main sequence stars from mann et al .
( 2013).,title="fig:",width=453 ] | we derive metallicities for 41 cataclysmic variables ( cvs ) from near - infrared spectroscopy .
we use synthetic spectra that cover the 0.8 @xmath0 m @xmath1 2.5 @xmath0 m bandpass to ascertain the value of [ fe / h ] for cvs with k - type donors , while also deriving abundances for other elements . using calibrations for determining [ fe / h ] from the @xmath2-band spectra of m - dwarfs , we derive more precise values for t@xmath3 for the secondaries in the shortest period cvs , and examine whether they have carbon deficits . in general , the donor stars in cvs have sub - solar metallicities .
we confirm carbon deficits for a large number of systems .
cvs with orbital periods @xmath4 5 hr are most likely to have unusual abundances .
we identify four cvs with co emission .
we use phase - resolved spectra to ascertain the mass and radius of the donor in u gem .
the secondary star in u gem appears to have a lower _ apparent _ gravity than a main sequence star of its spectral type . applying this result to other cvs , we find that the later - than - expected spectral types observed for many cv donors is mostly an affect of inclination .
all of the magnetic cvs , except the low accretion rate polar mq dra , have donors with subsolar metallicities .
we find that two systems with unusual spectra , ei psc and qz ser , that have large excesses of sodium , and extreme deficits of carbon .
synthetic spectra that have a reduced abundance of hydrogen are best able to explain the spectra of these two objects . _
key words : _ infrared : stars stars : cataclysmic variables stars : abundances stars : individual ( rx and , ae aqr , vy aqr , uu aql , tt ari , ss aur , sy cnc , v436 cen , ww cet , wx cet , tt crt , em cyg , ey cyg , ss cyg , ex dra , mq dra , yy / do dra , u gem , ah her , am her , ex hya , vw hyi , rz leo , st lmi , cw mon , v426 oph , v2051 oph , cn ori , cz ori , v1309 ori , ip peg , ls peg , ru peg , gk per , kt per , ei psc , vv pup , v893 sco , mr ser , qz ser , ar uma , ch uma , tw vir ) @xmath5 visiting observer , w. m. keck observatory , which is operated as a scientific partnership among the california institute of technology , the university of california , and the national aeronautics and space administration .
+ @xmath6 visiting astronomer at the infrared telescope facility , which is operated by the university of hawaii under contract from nasa . |
this article describes a combinatorial way to reconstruct a tiling space and its groupoid , using bratteli diagrams . in a forthcoming paper , this description will be used for tilings given by multi - substitutions @xcite .
given a repetitive , aperiodic tiling @xmath0 with finite local complexity ( flc ) in @xmath1 , its tiling space @xmath2 , or hull , is a compact space obtained by taking a suitable closure of the family of tilings obtained by translating @xmath0 . by construction , the translation group @xmath1 acts on @xmath2 by homeomorphisms making the pair @xmath10 a topological dynamical system .
it is well known that repetitivity ( or _ uniform repetitivity _ , as it is called in the symbolic one - dimensional case ) is equivalent to the minimality of this action , see @xcite .
equivalently , this dynamical system can be described through a groupoid @xcite , denoted by @xmath11 , called the _ crossed product _ of the tiling space by the action .
if the tiles of @xmath0 are punctured , the subset of @xmath2 made of tilings with one puncture at the origin of @xmath1 is a compact subset @xmath3 called the _
canonical transversal _ @xcite . for quasi - crystals
this identifies with the so - called _ atomic surface _
it has been shown that flc implies that the transversal is completely disconnected @xcite , while aperiodicity and minimality eliminate isolated points , making it a cantor set .
similarly , there is an _ tale _
groupoid @xmath12 associated with it @xcite , which plays a role similar to the poincar first return map in usual dynamical systems .
this groupoid is defined by the equivalence relation @xmath4 identifying two tilings of the transversal differing by a space translation .
while the existence of the hull does not require much information , it can be quite involved to have an effective description which allows computations .
the first step in this direction came from the work of anderson and putnam for substitution tilings @xcite .
a substitution is a rule describing how each tiles , after suitable rescaling , is covered by other tiles touching along their faces .
the most publicized example is the penrose tiling in its various versions , like kites and darts @xcite .
anderson and putnam built a @xmath6-complex @xmath13 of dimension @xmath14 , with @xmath14-cells given by suitably decorated prototiles and showed that the substitution induces a canonical map @xmath15 .
then the hull can be recovered as the inverse limit @xmath16 .
this construction has several generalizations for repetitive , aperiodic , and flc tilings without a substitution rule , see @xcite .
it is proved that there is a sequence of finite @xmath6-complexes @xmath17 of dimension @xmath14 and maps @xmath18 so that the hull is given by the inverse limit @xmath19 .
it is even possible to choose these @xmath6-complexes to be smooth branched manifolds @xcite .
the first goal of the present paper is to describe the construction of a bratteli diagram from these data in section [ bratteli09.ssect-brat ] .
a bratteli diagram is a graph with a marked vertex @xmath20 called the _
root_. the set of vertices @xmath21 is graded by a natural integer , called the _ generation _ , so that @xmath22 with @xmath23 and @xmath24 if @xmath25
. then edges exist only between @xmath26 and @xmath27 . in the present construction , each vertex in @xmath26
is given by a @xmath14-cell of @xmath28 .
then there is an edge between a vertex @xmath29 and @xmath30 if and only if the @xmath14-cell @xmath31 can be found inside @xmath32 .
each edge will be labeled by the translation vector between the puncture of @xmath32 and the one of @xmath31 inside @xmath32 .
namely at each generation @xmath33 , the bratteli diagram encodes how a @xmath14-cell of @xmath34 is filled by @xmath14-cells of @xmath28 .
more precisely it encodes not only which cells in @xmath28 occur but also their position relative to the @xmath14-cell of @xmath34 . as in @xcite , since the family @xmath17 is not unique , there are several choices to build a bratteli diagram associated with a tiling space .
conversely , starting from such a bratteli diagram , a tiling can be described as an infinite path starting from the root .
such a path describes how a given patch at generation @xmath33 , is embedded in a larger patch at generation @xmath35 . by induction on the generations ,
an entire tiling is reconstructed .
in addition , the origin of the tiling is defined uniquely at each step , thanks to the label of the edges involved . if each patch involved in the construction is decorated by its _
, namely provided that the tiling _ forces its border _ in the sense of @xcite , the tiling obtained eventually covers @xmath1 , even if the origin is at a fixed distance from the boundary of the patch , because the collar , beyond this boundary , increases in size as well . as a result the transversal @xmath3 is recovered , as a compact topological space , from the space of infinite rooted paths in the diagram ( theorem [ bratteli09.thm-isotrans ] ) . in a bratteli diagram ,
the _ tail equivalence _ identifies two infinite paths differing only on a finite number of edges ( definition [ bratteli09.def-afer ] ) .
it gives rise to a groupoid called the _ af - groupoid _ of the diagram . in the language of tiling spaces ,
this groupoid describes the translation structure inside the tiling , up to an important obstacle .
namely , a tiling built from an infinite path with the origin at a fixed distance from the boundary of the corresponding patches can not be identified , modulo the tail equivalence , with a tiling built from a path with origin at a fixed distance from the same boundary but located on the other side of it . as a result
, a tiling obtained in this way will be subdivided into regions , that will be called _ af - regions _ here ( definition [ bratteli09.def-afregion ] ) , separated by boundaries .
note that matui has found similar features for a class of @xmath36-dimensional substitution tilings in @xcite .
by contrast , the groupoid of the transversal @xmath12 ( definition [ bratteli09.def-gpd ] ) allows to identify two such regions through translation . so that recovering @xmath12 from the bratteli diagram requires to change the definition of the tail equivalence .
the present paper offers a solution to this problem by adding a _ horizontal structure _ to the diagram making it a _ collared bratteli diagram _ ( definition [ bratteli09.def-cbrat ] ) .
its aim is to describe in a combinatorial way , from local data , how to locate a tile in a patch , relative to its boundary .
practically it consists in adding edges between two vertices of the same generation , describing pairs of tiles , in a pair of patches , each in the collar of one another ( see figure [ bratteli09.fig-horizontal ] ) .
these edges are labeled by the translation between these tiles . in other words ,
the horizontal edges describe how to glue together two patches in a tiling across an af - boundary .
then it becomes possible to extend the tail equivalence into an tale equivalence relation on the collared bratteli diagram ( definition [ bratteli09.def-eqrel ] and theorem [ bratteli09.def-etale ] ) in order to recover the groupoid of the transversal ( theorem [ bratteli09.thm-eqrel ] and corollary [ bratteli09.cor-groupoid ] ) , with the af - groupoid as a sub - groupoid ( remark [ bratteli09.rem-afcomp ] ) . in the particular case of @xmath37-dimensional tilings ,
this larger equivalence relation is generated by the af - relation and finitely many pairs of minimal and maximal paths in the diagram , which are derived from the collared diagram in a natural and explicit way ( proposition [ bratteli09.prop-1d ] ) .
the examples of the fibonacci and thue - morse tilings are illustrated in details in section [ bratteli09.sect-examples ] .
this study of @xmath37-dimensional tilings also allows to view the paths whose associated tilings have their punctures at a minimum distance to an af - boundary ( corollary [ bratteli09.cor-fsigma ] ) as generalizations of extremal paths for @xmath37-dimensional systems .
this work gives one more way of describing tilings or their sets of punctures , and their groupoids , liable to help describing various properties of tiling spaces , such as their topology or their geometry .
it benefited from almost thirty years of works with original motivation to describe more precisely the properties of aperiodic solids .
in particular the notion of hull , or tiling space , was described very early as a fundamental concept @xcite encoding their macroscopic _ translation invariance _ , called homogeneity .
see for instance @xcite for reviews and updates . during the eighties , the discovery of quasicrystals @xcite was a landmark in this area and stimulated a lot of mathematical works to describes this new class of materials @xcite .
in particular , it was very convenient to represent the structure of these materials by various examples of tilings , such as the penrose tiling or its @xmath38-dimensional analogs @xcite .
it was shown subsequently @xcite that such structures are liable to describe precisely the icosahedral phase of quasicrystalline alloys such as . in the end of the eighties , connes attracted the attention of mathematicians to the subject by showing that the penrose tiling was a typical example of a noncommutative space @xcite .
it is remarkable that already then , connes used a bratteli diagram to encode the combinatorics of patches between generations .
its is important to realize , though , that the construction given by connes was based on the _ substitution _ proposed originally by penrose @xcite but ignored entirely the additional structure given by space _
translations_. in the context of the present paper , including the translations is the key reason leading to the extension of the @xmath5-relation .
this leads to a non-@xmath5 @xmath39-algebra instead , a difficulty at the source of so many works during the last twenty years .
bratteli created his diagrams in @xcite to classify @xmath5-algebras , namely @xmath39-algebras obtained as the unions of finite dimensional @xmath39-algebras .
it took two decades before it was realized that , through the notion of _ vershik map _
@xcite such diagrams could encode any minimal homeomorphism of the cantor set @xcite .
the corresponding crossed product @xmath39-algebra was shown to characterize the homeomorphism up to orbit equivalence .
this classification was a natural extension of a similar problem for ergodic actions of @xmath40 , a problem solved by krieger and connes within the framework of von neumann algebras ( see @xcite for instance ) .
this program was continued until recently and lead to the proof of a similar result for minimal actions of @xmath41 on the cantor set @xcite .
in particular , as a consequence of this construction , giordano matui putnam skau proved that any minimal @xmath41-action on the cantor set is orbit equivalent to a @xmath40-action @xcite .
moroever , the groupoid of the transversal of every aperiodic repetitive flc tiling space is orbit equivalent to a minimal @xmath40-action on the cantor set @xcite .
as it turns out the formalism described in the present work is similar to and , to a certain extend inspired by , the construction of _ refined tessellations _ made by giordano _
_ in @xcite .
this paper is organized as follows : the section [ bratteli09.sect-tilgpd ] contains the basic definitions about tilings , groupoids , and the constructions existing in the literature required in the present work .
section [ bratteli09.sect-gpdbrat ] is dedicated to the definition and properties of the bratteli diagram associated with a tiling .
the main reconstruction theorems are reproduced there .
the last section [ bratteli09.sect-examples ] illustrates the construction for @xmath37-dimensional substitution tilings , and treats the two examples of the fibonacci and thue morse tilings in details .
* acknowledgements : * the three authors are indebted to johannes kellendonk for his invaluable input while starting this work .
j.s . would like to thank john hunton for discussions and for communicating information about his research using the notion of bratteli diagrams .
this work was supported by the nsf grants no .
dms-0300398 and no .
dms-0600956 , by the school of mathematics at georgia tech in the spring 2009 and by the sfb 701 ( universitt bielefeld , germany ) in the group led by michael baake .
this section is a reminder .
the basic notions about tilings and their groupoids are defined and described .
the finite volume approximations of tiling spaces by branched manifolds are also summarized @xcite .
all tilings in this work will be subsets of the @xmath14-dimensional euclidean space @xmath1 .
let @xmath42 denote the open ball of radius @xmath43 centered at @xmath44 .
[ bratteli09.def-tiling ] a. a _ tile _ is a compact subset of @xmath1 which is the closure of its interior .
b. a _ punctured tile _
@xmath45 is an ordered pair consisting of a tile @xmath46 and a point @xmath47 . c. a _ partial tiling _ is a collection @xmath48 of tiles with pairwise disjoint interiors .
the set @xmath49 is called its _
support_. d. a _ patch _ is a finite partial tiling .
a patch is punctured by the puncture of one of the tiles that it contains .
e. a _ tiling _ is a countable partial tiling with support @xmath1 .
a tiling is said to be punctured if its tiles are punctured .
f. the _ inner radius _ of a tile or a patch is the radius of the largest ball ( centered at its puncture ) that is contained in its support .
the _ outer radius _ is the radius of the smallest ball ( centered at its puncture ) that contains its support .
[ tlap09.hypo-punccw ] from now all tiles and tilings are punctured .
in addition , each tile is assumed to have a finite @xmath6-complex structure .
the @xmath6-complex structures of the tiles in a tiling are compatible : the intersection of two tiles is a subcomplex of both . in particular , the support of a tiling gives a @xmath6-complex decomposition of @xmath1 .
[ bratteli09.rem- ] _ _ a. definition [ bratteli09.def-tiling ] allows tiles and patches to be _
disconnected_. b. tiles or patch are considered as subsets of @xmath1 .
if @xmath46 is a tile and @xmath50 a patch , the notations _ @xmath51 and @xmath52 _ mean _ @xmath46 is a tile and @xmath50 is a patch of the tiling @xmath0 , at the positions they have as subsets of @xmath1_. the results in this paper are valid for the class of tilings that are _ aperiodic , repetitive _ , and have _ finite local complexity_. [ bratteli09.def-repflc ] let @xmath0 be a tiling of @xmath1 .
a. @xmath0 has _ finite local complexity _
( flc ) if for any @xmath53 there are up to translation only finitely many patches of outer radius less than @xmath54 .
b. if @xmath0 has flc then it is _ repetitive _ if given any patch @xmath50 , there exists @xmath55 such that for every @xmath56 , there exists some @xmath57 such that @xmath58 .
c. for @xmath59 let @xmath60 denote the translate of @xmath0 by @xmath61 .
then @xmath0 is aperiodic if @xmath62 , for all vector @xmath63 . from now on
, we will only consider tilings which satisfy the conditions above .
note that flc implies that there are only finitely many tiles up to translation .
one of the authors defined in @xcite a topology that applies to a large class of tilings ( even without flc ) . in the present setting , this topology can be adapted as follows .
let @xmath64 be a family of tilings .
given an open set @xmath65 in @xmath1 with compact closure and an @xmath66 , a neighborhood of a tiling @xmath0 in @xmath64 is given by @xmath67 where @xmath68 is the notation for the set of all cells of @xmath0 which intersect @xmath65 .
[ bratteli09.def-hull ] a. the _ hull _ or _ tiling space _ of @xmath0 , denoted @xmath2 , is the closure of @xmath69 for the topology defined above .
b. the _ canonical transversal _
, denoted @xmath3 , is the subset of @xmath2 consisting of tilings having one tile with puncture at the origin @xmath70 . by flc condition
, the infimum of the inner radii of the tiles of @xmath0 is @xmath71 .
so the canonical transversal is actually an abstract transversal for the @xmath1 action in the sense that it intersects every orbit , and @xmath72 for all @xmath57 small enough ( @xmath73 ) .
the hull of a tiling is a dynamical system @xmath74 which , for the class of tilings considered here , has the following well - known properties ( see for example @xcite section 2.3 ) .
[ bratteli09.thm-hull ] let @xmath0 be a tiling of @xmath1 .
a. @xmath2 is compact .
b. @xmath0 is repetitive if and only if the dynamical system @xmath75 is minimal .
c. if @xmath0 has flc , then its canonical transversal @xmath3 is totally disconnected @xcite . d. if @xmath0 is repetitive and aperiodic , then @xmath2 is _ strongly aperiodic _ ,
_ i.e. _ contains no periodic points .
e. if @xmath0 is aperiodic , repetitive , and has flc , then @xmath3 is a cantor set . by the minimality property
, if @xmath2 is the hull of a tiling @xmath0 , any patch @xmath50 of any tiling @xmath76 appears in @xmath0 in some position . in case ( iv ) , the sets @xmath77 for @xmath78 a patch of @xmath0 , form a base of clopen sets for the topology of @xmath3 .
[ bratteli09.rem-topohull ] _ a metric topology for tiling spaces has been used in the literature .
let @xmath0 be a repetitive tiling or @xmath1 with flc .
the orbit space of @xmath0 under translation by vectors of @xmath1 , @xmath79 , is endowed with a metric as follows ( see @xcite section 2.3 ) . for @xmath80 and @xmath81 in @xmath79 , let @xmath82 denote the set of @xmath83 in @xmath84 such that there exists @xmath85 for which @xmath86 and @xmath87 agree on @xmath88 , _ i.e. _ their tiles whose punctures lie in the ball are matching , then @xmath89 hence the diameter of @xmath79 is bounded by @xmath37 . with this distance ,
the action of @xmath1 is continuous . _ for the class of repetitive tilings with flc , the topology of the hull given in definition [ bratteli09.def-hull ] is equivalent to this @xmath90-metric topology @xcite .
let @xmath2 be the tiling space of an aperiodic , repetitive , and flc tiling of @xmath1 , and let @xmath3 be its canonical transversal . [ bratteli09.def-erhull ]
the equivalence relation @xmath91 of the tiling space is the set @xmath92 with the following topology : a sequence @xmath93 converges to @xmath94 if @xmath95 in @xmath2 and @xmath96 in @xmath1 .
the equivalence relation of the transversal is the restriction of @xmath91 to @xmath3 : @xmath97 note that the equivalence relations are _ not _ endowed with the relative topology of @xmath98 and @xmath99 . for example , by repetitivity , for @xmath61 large , @xmath0 and @xmath100 might be close to each other in @xmath2 , so that @xmath101 is close to @xmath102 for the relative topology , but not for that from @xmath103 .
the map @xmath104 from @xmath103 to @xmath105 has a dense image , coinciding with @xmath91 and is one - to - one because @xmath2 is strongly aperiodic ( contains no periodic points ) .
the topology of @xmath91 is the topology induced by this map .
[ bratteli09.def-etale ] an equivalence relation @xmath106 on a compact metrizable space @xmath13 is called _ tale _ when the following holds .
a. the set @xmath107 is closed in @xmath108 and the maps sending @xmath109 in @xmath108 to @xmath110 in @xmath106 , @xmath111 in @xmath106 , and @xmath112 in @xmath106 are continuous .
b. the diagonal @xmath113 is open in @xmath106 . c. the range and source maps @xmath114 given by @xmath115 , are open and are local homeomorphisms .
a set @xmath116 is called an _ @xmath106-set _
, if @xmath65 is open in @xmath106 , and @xmath117 and @xmath118 are homeomorphisms .
it is proved in @xcite that @xmath4 is an tale equivalence relation . a _ groupoid _
@xcite is a small category ( the collections of objects and morphisms are sets ) with invertible morphisms .
a topological groupoid , is a groupoid @xmath119 whose sets of objects @xmath120 and morphisms @xmath119 are topological spaces , and such that the composition of morphisms @xmath121 , the inverse of morphisms @xmath122 , and the source and range maps @xmath123 are all continuous maps . given an equivalence relation @xmath106 on a topological space @xmath13 , there is a natural topological groupoid @xmath119 associated with @xmath106 , with objects @xmath124 , and morphisms @xmath125 .
the topology of @xmath119 is then inherited by that of @xmath106 .
[ bratteli09.def-gpd ] the _ groupoid of the tiling space _ is the groupoid of @xmath4 , with set of objects @xmath126 and morphisms @xmath127 there is also a notion of tale groupoids @xcite .
essentially , this means that the range and source maps are local homeomorphisms .
it can be shown that @xmath12 is an tale groupoid @xcite .
let @xmath0 be an aperiodic , repetitive and flc ( punctured ) tiling of @xmath1 .
since @xmath0 has finitely many tiles up to translation , there exists @xmath128 , such that the minimum of the inner radii of its tiles equals @xmath43 , and the maximum of the outer radii equals @xmath106 .
the set @xmath129 of punctures of tiles of @xmath0 is a _
delone set _
( or @xmath130-delone set ) . that is @xmath129 is _ uniformly discrete _ ( or @xmath43-uniformly discrete ) : for any @xmath131 , @xmath132 contains at most a point , and _ relatively dense _ ( or @xmath106-relatively dense ) : for any @xmath131 , @xmath133 contains at least a point .
[ bratteli09.rem-delone ] _ properties of aperiodicity , repetitivity , and flc can be defined for delone sets as well .
see @xcite for instance .
the study of delone sets is equivalent to that of tilings with the same properties .
any tiling @xmath0 defines the delone set @xmath129 .
conversely , a delone set @xmath134 defines its _ voronoi tiling _ as follows .
let @xmath135 be the _ voronoi tile _
at @xmath136 ( it is a closed and convex polytope ) .
the voronoi tiling @xmath137 is the tiling with tiles @xmath138 . _ [ bratteli09.def-ctile ]
let @xmath0 be a tiling such that @xmath129 is @xmath43-uniformly discrete for some @xmath71 .
a. the _ collar _ of a tile @xmath139 is the patch @xmath140 b. a _ prototile _ is an equivalence class of tiles under translation .
collared prototile _ is the subclass of a prototile whose representatives have the same collar up to translation
. then @xmath141 and @xmath142 will denote the set of prototiles and collared prototiles of @xmath0 respectively . c. the _ support _
@xmath143 $ ] of a prototile @xmath144 $ ] is the set @xmath145 with @xmath146 a representant in the class of @xmath144 $ ] , and @xmath147 is the vector joining the origin to the puncture of @xmath146 . as can be checked easily , the set @xmath145 does not depend upon which tile @xmath146 is chosen in @xmath144 $ ] .
hence @xmath143 $ ] is a set obtained from any tile in @xmath144 $ ] by a translation with puncture at the origin .
the support of a collared prototile is the support of the tiles it contains .
a collared prototile is a prototile where a local configuration of its representatives has been specified : each representative has the same neighboring tiles up to translation . in general
, @xmath148 may contain more tiles than just the ones intersecting @xmath46 , as illustrated in figure [ bratteli09.fig-collar ] .
since @xmath0 is flc , it follows that its has only a finite number of collared prototiles .
moreover , any tiling in its hull have the same set of collared prototiles .
the previous description allows to represent a tiling @xmath0 in a more combinatorial way . given a prototile @xmath149 ( collared or not ) ,
let @xmath150 be the subset of @xmath129 made of punctures corresponding to tiles in @xmath149 ( it should be noted that it is also an aperiodic repetitive flc delone set ) .
[ bratteli09.def-combtil ] the set @xmath151 of pairs @xmath152 , where @xmath153 is a prototile and @xmath154 , will be called the combinatorial representation of @xmath0 .
similarly , the set @xmath155 of pairs @xmath152 , where @xmath156 is a collared prototile and @xmath154 , will be called the combinatorial collared representation of @xmath0 .
there is a one - to - one correspondence between the family of tiles of @xmath0 and @xmath157 , namely , with each tile @xmath51 is associated the pair @xmath158)$ ] where @xmath159 is the puncture of @xmath46 . by construction
this map is one - to - one .
conversely @xmath160 will denote the tile corresponding to @xmath152 .
thanks to the hypothesis [ tlap09.hypo-punccw ] , the tiles of @xmath0 are finite and compatible @xmath6-complexes .
the next definition is a reformulation of the anderson
putnam space @xcite that some of the authors gave in @xcite . [ bratteli09.def-ap ]
let @xmath161 be the prototiles of @xmath0 .
let @xmath160 be the support of @xmath162 .
the _ prototile space _ of @xmath0 is the quotient @xmath6-complex @xmath163 where two @xmath164-cells @xmath165 and @xmath166 are identified if there exists @xmath167 for which @xmath168 , with @xmath169 .
the _ collared prototile space _
@xmath170 is built similarly out of the collared prototiles of @xmath0 .
[ bratteli09.prop-projhull ] there is a continuous map @xmath171 .
let @xmath172 be the quotient map . and let @xmath173 be defined as follows .
if @xmath61 belongs to the intersection of @xmath164 tiles @xmath174 , in a tiling @xmath175 , with @xmath176 , then the point @xmath177 belongs to @xmath178 .
moreover , all these points are identified after taking the quotient , namely @xmath179 if @xmath180 .
therefore @xmath181 is well defined .
this allows to set @xmath182 .
this map sends the origin of @xmath1 , that lies in some tile of @xmath183 , to the corresponding tile @xmath184 s at the corresponding position . in @xmath185
, points on the boundaries of two tiles @xmath160 and @xmath184 are identified if there are neighboring copies of the tiles @xmath186 somewhere in @xmath0 such that the two associated points match .
this ensures that the map @xmath187 is well defined , for if in @xmath1 tiled by @xmath183 , the origin belongs to the boundaries of some tiles , then the corresponding points in @xmath188 given by @xmath189 are identified by @xmath190 .
let @xmath61 be a point in @xmath185 , and @xmath191 and open neighborhood of @xmath61 .
say @xmath61 belongs to the intersection of some tiles @xmath192 .
let @xmath183 be a preimage of @xmath61 : @xmath193 .
the preimage of @xmath191 is the set of tilings for which the origin lies in some neighborhood of tiles that are translates of @xmath194 , and this defines a neighborhood of @xmath183 in the @xmath2 .
therefore @xmath187 is continuous . a _ nested sequence of tilings _
@xmath195 , is a countable infinite sequence of tilings such that @xmath196 and for all @xmath197 the tiles of @xmath198 are ( supports of ) patches of @xmath199 . without loss of generality , it will be assumed that @xmath200 , _
i.e. _ has a puncture at the origin .
to built such a sequence , the following procedure will be followed : assume that the tiling @xmath199 has been constructed and is aperiodic , repetitive and flc with one puncture at the origin ; then a. let @xmath201 be a finite patch with a puncture at the origin , and let @xmath202 be the delone set of the punctures of the translated copies of @xmath203 within @xmath199 ; b. let @xmath204 be the voronoi tiling of @xmath205 ( remark [ bratteli09.rem-delone ] ) ; to each tile of @xmath199 a tile @xmath206 will be assigned ( see the precise definition below ) ; c. for each tile @xmath206 , let @xmath207 be the union of the tiles of @xmath199 that have been assigned to it ; then @xmath198 is defined as the tiling @xmath208 .
it is worth remarking that @xmath205 is an aperiodic , repetitive , flc , delone set as well . thus the voronoi tiling inherits these properties .
note however that in point ( iii ) , two tiles @xmath209 may have the same shape .
however , if they correspond to different patches of @xmath199 , they should be labeled as different .
the second step of the construction above needs clarification since the tiles of @xmath204 are not patches of @xmath199 , but convex polytopes built out of @xmath205 ( see remark [ bratteli09.rem-delone ] ) .
first the set of patches of @xmath0 is countable , and second for any patch @xmath52 the voronoi tiling of @xmath210 has flc , thus has finitely many prototiles .
hence there is a vector @xmath57 which is not in the span of any of the subspaces generated by the faces of the tiles @xmath211 for all @xmath50 .
a point @xmath131 is called @xmath212-interior to a closed set @xmath213 , and write @xmath214 , if @xmath215 .
given a voronoi tile @xmath206 the patch associated with it is defined by ( here @xmath159 denotes the puncture of @xmath46 ) @xmath216 this gives an unambiguous assignment of tiles of @xmath199 to tiles of @xmath198 .
there is therefore a natural subdivision of each tile of @xmath198 into tiles of @xmath199 .
let @xmath217 be the combinatorial representation of @xmath198 ( see definition [ bratteli09.def-combtil ] ) .
there is a map @xmath218 , describing how to assign a tile in @xmath199 to the tile of @xmath198 , namely to a patch of @xmath199 it belongs to .
the `` inverse map '' defines a _ substitution _ , denoted by @xmath219 , namely a map from the set of tiles of @xmath198 to the set of patches of @xmath199 defined by @xmath220 such a substitution defines a map , also denoted by @xmath219 , from the @xmath6-complex @xmath198 onto the @xmath6-complex @xmath199 , if the tiles @xmath221 are given the @xmath6-complex structure inherited by the ones of the unions of the @xmath222 for @xmath223 . in much the same way , this gives a map on the prototile space , and thus on the anderson
putnam complex as well .
a similar construction holds if tiles and prototiles are replaced by collared tiles and collared prototiles .
it leads to a canonical map @xmath224 , also denoted by @xmath219 .
let us denote by @xmath225 be the minimum of the inner radii of the tiles of @xmath226 , and @xmath227 the maximum of the outer radii .
[ bratteli09.def-nested ] a nested sequence of tilings @xmath228 with substitution maps @xmath229 is called a _ proper nested sequence _ , if for all @xmath33 the tiling @xmath198 is aperiodic , repetitive , and has flc , and the following holds for all @xmath230 : a. for each tile @xmath231 , @xmath232 is a patch of @xmath199 , b. for each tile @xmath231 , @xmath232 contains a tile of @xmath199 in the interior of its support , c. there exists @xmath53 ( independent of @xmath33 ) , such that @xmath233 and @xmath234 .
the nested sequence constructed above can be made proper by choosing , for each @xmath33 , the patch @xmath203 to be large enough .
these conditions are sufficient to ensure that @xmath235 is _ zoomed out _ of @xmath236 for all @xmath230 , in the terminology of @xcite ( definition 2.43 ) .
the map @xmath237 also _ forces the border _ ( see @xcite , definition 15 ) because it is defined on the _ collared _ prototile spaces .
this suffices to recover the tiling space from the sequence @xmath238 .
[ bratteli09.thm-isohull ] let @xmath195 be a proper nested sequence of tilings .
then the tiling space @xmath2 of @xmath80 is homeomorphic to the inverse limit of the complexes @xmath235 : @xmath239 this theorem was first proved in @xcite and it was also proved that the @xmath1-action could also be recovered from this construction ( see also @xcite ( theorem 5 ) ) .
the reader can also easily adapt the proof of theorem [ bratteli09.thm-isotrans ] given in section [ bratteli09.ssect-brat ] to deduce theorem [ bratteli09.thm-isohull ] . for each @xmath33
, we can see the tiling @xmath198 as a `` subtiling '' of @xmath80 .
the map @xmath240 of proposition [ bratteli09.prop-projhull ] can be immediately adapted to a map onto @xmath241 for each @xmath33 .
[ bratteli09.prop-projhulln ] there is a continuous map @xmath242 .
let @xmath243 .
the origin of @xmath1 lies in some patch @xmath244 which is a translate of @xmath245 for some @xmath246 .
we set @xmath247 to be the corresponding point in image of @xmath221 in @xmath241 .
bratteli diagrams were introduced in the seventies for the classification of @xmath5-algebras @xcite .
they were then adapted , in the topological setting , to encode @xmath40-actions on the cantor set @xcite .
a specific case , close from our present concern , is the action of @xmath40 by the shift on some closed , stable , and minimal subset of @xmath248 ( this is the one - dimensional symbolic analog of a tiling ) .
then , bratteli diagrams were used to represent @xmath249 @xcite , and recently @xmath41 @xcite cantor dynamical systems , or to represent the transversals of substitution tiling spaces @xcite .
[ bratteli09.def-bdiag ] a _ bratteli diagram _ is an infinite directed graph @xmath250 with sets of vertices @xmath251 and edges @xmath252 given by @xmath253 where the @xmath254 and @xmath255 are finite sets .
the set @xmath256 consists of a single vertex , called the _
root _ of the diagram , and noted @xmath20 .
the integer @xmath257 is called the _ generation index_. and there are maps @xmath258 called the _ source _ and _ range _ maps respectively .
we assume that for all @xmath257 and @xmath259 one has @xmath260 and @xmath261 ( _ regularity _ ) .
we call @xmath7 _ stationary _ , if for all @xmath262 , the sets @xmath254 are pairwise isomorphic and the sets @xmath255 are pairwise isomorphic .
regularity means that there are there are no `` sinks '' in the diagram , and no `` sources '' apart from the root .
a bratteli diagram can be endowed with a _ label _ , that is a map @xmath263 to some set @xmath264 . [ bratteli09.def-paths ] let @xmath7 be a bratteli diagram .
a. a _ path _ in @xmath7 is a sequence of composable edges @xmath265 , with @xmath266 and @xmath267 .
b. if @xmath268 is finite , @xmath269 , is called a _ finite path _ , and @xmath268 the length of @xmath270 .
we denote by @xmath271 the set of paths of length @xmath268 .
c. if @xmath268 is infinite , @xmath272 , is called an _
infinite path_. we denote by @xmath8 the set of infinite paths .
d. we extend the range map to finite paths : if @xmath273 we set @xmath274 .
in addition to definition [ bratteli09.def-bdiag ] , we ask that a bratteli diagram satisfies the following condition .
[ bratteli09.hypo-split ] for all @xmath275 , there are at least two distinct infinite paths through @xmath31 .
the set @xmath8 is called the _ boundary _ of @xmath7 .
it has a natural topology inherited from the product topology on @xmath276 , which makes it a compact and totally disconnected set .
a base of neighborhoods is given by the following sets : @xmath277 = \ { x \in \partial { { \mathcal b}}\ ; \ \gamma \text { is a prefix of } x\}.\ ] ] hypothesis [ bratteli09.hypo-split ] is the required condition to make sure that there are no isolated points .
this implies the following . with this topology , @xmath8 is a cantor set .
we now build a bratteli diagram associated with a proper nested sequence of tilings .
[ bratteli09.def-tilbrat ] let @xmath228 be a proper nested sequence of tilings .
let @xmath278 , be the collared prototiles of @xmath198 , and @xmath279 the representative of @xmath280 that has its puncture at the origin .
the bratteli diagram @xmath281 associated with @xmath228 is given by the following : a. @xmath282 , and @xmath283 , b. @xmath284 : @xmath285 if and only if @xmath286 and @xmath287 , c. @xmath288 with @xmath289 and @xmath290 , if and only if there exists @xmath291 such that @xmath292 is a tile of the patch @xmath293 , d. a label @xmath294 , with @xmath295 for @xmath296 ( and @xmath297 on @xmath298 ) .
figure [ bratteli09.fig-edge ] illustrates condition ( iii ) .
we extend the label as a map on finite paths @xmath299 : for @xmath300 we set @xmath301 we can associate to each finite path @xmath302 in @xmath7 , with @xmath303 , the patches of @xmath80 @xmath304 where @xmath305 is a `` collared patch '' , in the sense that it is the set of tiles of @xmath80 which make up the collar of a tile of @xmath198 .
[ bratteli09.thm-isotrans ] let @xmath3 be the transversal of the tiling space of @xmath80 .
there is a canonical homeomorphism @xmath306 .
let @xmath307 , and set @xmath308 .
define @xmath309 where @xmath310 is the clopen set of tilings that have the patch @xmath311 at the origin ( see equations and ) . since @xmath3 is compact , and @xmath312
is closed and contains @xmath313 for all @xmath33 , by the finite intersection property @xmath314 is a closed and non - empty subset of @xmath3 .
let us show it consists of a single tiling .
let then @xmath315 . for all @xmath33 , @xmath316 .
and since @xmath311 is a collared patch , it contains a ball of radius @xmath317 ( see definition [ bratteli09.def-nested ] ) centered at its puncture . using the metric of remark [ bratteli09.rem-topohull ]
this implies that @xmath318 for all @xmath33 . by condition ( iii ) in definition [ bratteli09.def-nested ] , @xmath319 as @xmath320 , therefore @xmath321 and @xmath322 . if @xmath323 then @xmath324 for some @xmath33 , thus @xmath325 and thus @xmath326 .
this proves that @xmath327 is injective . to prove that it is onto we exhibit an inverse
let @xmath328 .
for each @xmath33 , @xmath329 ( proposition [ bratteli09.prop-projhulln ] ) lies in the interior of some tile @xmath221 and therefore we can associate with @xmath0 a sequence of edges through those vertices .
this defines an inverse for @xmath327 . to prove that @xmath327 and @xmath330 are continuous it suffices to show that the preimages of base open sets are open .
clearly @xmath331 ) = \xi(p^c_\gamma)$ ] , hence @xmath330 is continuous .
conversely , since @xmath3 is compact and @xmath8 is hausdorff , the continuity of @xmath327 is automatic .
we now endow @xmath7 with a _ horizontal structure _ to take into account the adjacency of prototiles in the tilings @xmath198 , @xmath257 .
[ bratteli09.def-cbrat ] a _ collared bratteli diagram _ associated with a proper nested sequence @xmath228 , is a graph @xmath332 , with @xmath333 the bratteli diagram associated with @xmath228 as in definition [ bratteli09.def-tilbrat ] , and where @xmath334 is the set of _ horizontal edges _ : @xmath335 given by @xmath336 with @xmath337 , if and only if there exists @xmath338 , such that @xmath339 , with @xmath340 and we extend the label @xmath212 to @xmath334 , and set @xmath341 .
there is a horizontal arrow in @xmath342 between two tiles @xmath343 , if one can find `` neighbor copies '' in @xmath198 where each copy belongs to the collar of the other .
in other words there exists a patch @xmath344 ( _ i.e. _ its tiles have pairwise disjoint interiors ) with @xmath345 such that @xmath346 see figure [ bratteli09.fig-horizontal ] for an illustration . for @xmath347 , we define its _ opposite edge _
@xmath348 by @xmath349 clearly , for all @xmath350 in @xmath334 , @xmath348 also belongs to @xmath334 , and @xmath351 .
also , the definition allows _ trivial edges _ , that is edges @xmath350 for which @xmath352 and @xmath353 .
[ bratteli09.def-afer ] let @xmath7 be a bratteli diagram and let @xmath354 with the product topology ( discrete topology on @xmath355 ) .
the _ @xmath5-equivalence relation _ is the direct limit of the @xmath356 given by @xmath357 with the direct limit topology .
for @xmath358 we write @xmath359 and say that the paths are _ tail equivalent_. it is well known that @xmath360 is an @xmath5-equivalence relation , as the direct limit of the compact tale relations @xmath361 , see @xcite .
assume now that @xmath7 is a bratteli diagram associated with a proper sequence of nested tilings @xmath228 .
we show now what this @xmath5-equivalence relation represents for @xmath80 and its transversal @xmath3 . for a finite path @xmath362 ,
let us write @xmath363 which is the support of the patch @xmath364 as defined in equation : @xmath365 .
for @xmath366 , we can see @xmath367 as a subset of @xmath368 for all @xmath230 .
we characterize the subset of @xmath8 for which @xmath369 stays close to the boundary of @xmath370 for all @xmath33 , and its complement .
recall that a @xmath371 is a countable intersection of open sets , and an @xmath372 a countable union of closed sets .
[ bratteli09.lem-minimal ] for any @xmath257 , there exists a @xmath373 such that for any @xmath259 and any @xmath374 , there is a path in @xmath8 from @xmath31 to @xmath375 . in particular , any @xmath307 ,
the @xmath5-orbit of @xmath44 is dense .
the definition of the @xmath5 topology and the repetitivity of the underlying tilings are the two key elements of this proof .
first , we prove that for any @xmath259 , there exists @xmath376 such that for any @xmath374 , there is a path from @xmath31 to @xmath375 .
this is repetitivity : a vertex @xmath31 corresponds to a tile in @xmath198 . by repetitivity
, any tile of @xmath198 appears within a prescribed range , say @xmath106 .
now , pick @xmath164 such that the inner radius of the tiles of @xmath377 is greater than @xmath106 .
it means that any tile of @xmath198 appears in ( the substitute ) of any tile of @xmath377 .
this is exactly equivalent to the existence of a path from any @xmath259 to any @xmath374 .
now , consider @xmath378 .
let us show that @xmath379 can be approximated by elements @xmath380 in @xmath7 which are all @xmath5-equivalent to @xmath44 .
let @xmath380 be defined as follows : @xmath381 .
we just proved that there is a @xmath164 such that there is a path from @xmath382 to @xmath383 .
continue @xmath380 with this path , and define its tail to be the tail of @xmath44 .
then the sequence @xmath384 is the approximation we were looking for .
[ bratteli09.prop-gdelta ] the subset @xmath385 is a _
dense @xmath371 in @xmath8_. for @xmath386 , let @xmath387 .
then @xmath388 .
show that every @xmath389 is a dense open set in @xmath8 . remark that if for some @xmath390 , @xmath391 , then this property holds for all @xmath392 .
let us first prove that @xmath389 is dense .
let @xmath257 .
then there is a @xmath164 such that there is a path from any @xmath393 to any @xmath374 , by lemma [ bratteli09.lem-minimal ] .
let @xmath394 be such that @xmath395 , where @xmath361 is the outer radius of the tiles of @xmath198 .
then , let @xmath270 be a path from @xmath396 to @xmath397 corresponding to the inclusion of a tile of @xmath377 in the middle of a path of @xmath398 . now , for any path @xmath399 of length @xmath33
, it is possible to join @xmath399 to @xmath270 .
extend then this path containing @xmath399 and @xmath270 arbitrarily to an infinite path @xmath44 in @xmath8 .
then @xmath44 satisfies @xmath400 , so @xmath401 .
it proves that @xmath389 is non - empty . since we could do this construction for all @xmath257 and all @xmath399 of length @xmath33
, it proves that @xmath389 is dense .
finally , @xmath389 is open because if @xmath401 and satisfies @xmath402 , then the tail of @xmath44 after generation @xmath33 can be changed without changing this property .
it proves that @xmath389 contains a neighborhood around all of its points , and so it is open .
it proves that @xmath119 is a @xmath371 as intersections of dense open sets .
since @xmath8 is compact , it satisfies the baire property and so @xmath119 is dense .
[ bratteli09.cor-fsigma ] the subset @xmath403 is a _ dense @xmath372 in @xmath8_. with the notation of proposition [ bratteli09.prop-gdelta ] consider the closed set @xmath404 .
we have @xmath405 and @xmath406 , thus @xmath407 is an @xmath372 .
the proof of the density of @xmath407 in @xmath8 is similar to that for @xmath119 in proposition [ bratteli09.prop-gdelta ] .
let @xmath307 .
fix @xmath408 and consider the patch @xmath409 as in equation . by repetitivity of @xmath80
there exists @xmath410 such that @xmath80 has a copy of @xmath411 in each ball of radius @xmath410 .
hence there exists @xmath412 such that for all @xmath413 the tiles of @xmath198 ( viewed as patches of @xmath80 under the map @xmath414 ) contain a copy of @xmath411 that lies within a distance @xmath410 to their boundaries : @xmath415 .
we can thus extend the finite path @xmath416 to @xmath417 such that @xmath418 for all @xmath413 .
set @xmath419 .
we clearly have @xmath420 .
the sequence @xmath421 built in this way converges to @xmath44 in @xmath8 .
this proves that @xmath407 is dense in @xmath8
. * notation .
* we will use now the following notation : @xmath422 where @xmath327 is the homeomorphism of theorem [ bratteli09.thm-isotrans ] .
for each @xmath423 the equivalence relation @xmath424 induces an equivalence relation on @xmath129 : @xmath425 [ bratteli09.def-afregion ] an @xmath5-_region _ in a tiling @xmath200 is the union of tiles whose punctures are @xmath426-equivalent in @xmath129 .
[ bratteli09.prop-afregion ] a tiling @xmath423 has a single @xmath5-region if and only if @xmath427 .
assume @xmath428 has a single @xmath5-region .
fix @xmath53 .
pick @xmath429 with @xmath430 . since @xmath431
, there exists @xmath390 such that @xmath329 and @xmath432 belong to the same tile in @xmath235 for all @xmath433 ( see proposition [ bratteli09.prop-projhulln ] ) .
therefore @xmath434 for all @xmath433 .
since @xmath54 was arbitrary this proves that @xmath435 , _
i.e. _ that @xmath436 .
assume that @xmath437 .
choose @xmath438 .
since @xmath439 , there exists @xmath390 such that for all @xmath433 the patch @xmath440 contains a ball of radius @xmath441 around the origin ( where @xmath442 are the parameters of the delone set @xmath443 ) .
therefore @xmath444 agree with @xmath445 on a ball of radius @xmath446 .
hence @xmath447 and @xmath448 belong to the same tile in @xmath235 for all @xmath433 .
hence @xmath449 for all @xmath433 .
so we have @xmath450 , _
i.e. _ @xmath451 .
since @xmath61 was arbitrary , this shows that @xmath452 has a single @xmath5-region .
[ bratteli09.rem-aforbit ] _ let @xmath453 be the equivalence relation on @xmath3 that is the image of the equivalence relation @xmath360 induced by the homeomorphism of theorem [ bratteli09.thm-isotrans ] .
proposition [ bratteli09.prop-afregion ] shows that the @xmath454-orbit of a tiling @xmath428 is the set of all translates of @xmath0 by vectors linking to punctures that are in the @xmath5-region of the origin : @xmath455_{af } = \bigl\ { t -a \ : \ a \in t^{\text{\rm punc}}\ , , \ ; a \,\dot{\sim}\ , 0 \bigr\}\,,\ ] ] so if @xmath0 has more than one @xmath5-region , _
i.e. _ if @xmath456 , then its @xmath454-orbit is only a proper subset of its @xmath4-orbit ( definition [ bratteli09.eq-ertrans ] ) .
so we have : @xmath455_{af } = [ t]_{r_\xi } \iff x_t \in g\,.\ ] ] and for all @xmath0 in the dense subset @xmath457 we have @xmath458_{af } \subsetneq [ t]_{r_\xi}$ ] . _ as noted in remark [ bratteli09.rem-aforbit ] , the images in @xmath3 of the @xmath360-orbits do not always match those of @xmath4 . in this section
, we build a new equivalence relation on @xmath8 that `` enlarges '' @xmath360 , and from which we recover the full equivalence relation @xmath4 on @xmath3 .
we consider a collared bratteli diagram @xmath459 associated with a nested sequence @xmath228 ( definition [ bratteli09.def-tilbrat ] and [ bratteli09.def-cbrat ] ) , and we denote by @xmath3 the canonical transversal of the tiling space of @xmath80 . [ bratteli09.def-comdiag ] a _ commutative diagram _ in @xmath460 is a closed subgraph @xmath461^{e } \ar[rr]^{h } & & \ar[dd]^{e ' } \\ & & \\
\ar[rr]^{h ' } & & } \ ] ] where @xmath462 , @xmath463 , and @xmath464 , for some @xmath33 , and such that @xmath465 figure [ bratteli09.fig-comdiag ] illustrates geometrically the conditions of adjacency required for tiles to fit into a commutative diagram . with the notion of commutative diagram
we can now define an equivalence relation on @xmath8 that contains @xmath360 .
[ bratteli09.def-equiv ] we say that two infinite paths @xmath272 and @xmath466 in @xmath8 are equivalent , and write @xmath467 , if there exists @xmath468 , and @xmath469 for all @xmath433 , such that for each @xmath470 the subgraph @xmath461^{e_n } \ar[rr]^{h_{n-1 } } & & \ar[dd]^{e'_n } \\ & & \\ \ar[rr]^{h_n } & & } \ ] ] is a commutative diagram . [ bratteli09.lem-taileq ] if @xmath471 in @xmath8 , then there exists @xmath472 such that @xmath473 .
let @xmath474 be as in definition [ bratteli09.def-equiv ] , and for @xmath470 set @xmath475 .
for all @xmath476 we have @xmath477 where the last equality occurs by commutativity of the diagram between generations @xmath478 and @xmath33 .
hence we have @xmath479 for all @xmath470 .
now for all @xmath470 , the patches @xmath480 and @xmath481 belong to @xmath482 ( and similarly @xmath483 ) . hence @xmath484 .
set @xmath485 to complete the proof .
[ bratteli09.def-eqrel ] we define the equivalence relation on @xmath8 @xmath486 with the following topology : @xmath487 converges to @xmath110 in @xmath9 , if @xmath488 converges to @xmath44 in @xmath489 , and @xmath490 in @xmath1 . the flc and repetitivity properties of @xmath80 imply that for all @xmath428 , the set of vectors linking its punctures , @xmath491 , equals @xmath492 , and is _ discrete _ and _ closed _
( see @xcite for instance ) .
the convergence @xmath493 in @xmath9 implies then that there exists @xmath474 such that @xmath494 for @xmath433 .
[ bratteli09.rem-afcomp ] _ if @xmath495 in @xmath9 are such that the horizontal edges @xmath496 of the commutative diagrams are all trivial , then for all @xmath433 we have @xmath497 , _ i.e. _ @xmath44 and @xmath379 are tail equivalent in @xmath7 : @xmath359 .
thus we have the inclusion @xmath498 in view of remark [ bratteli09.rem-aforbit ] , the two equivalence relation coincide on @xmath119 , but differ on @xmath407 ( see proposition [ bratteli09.prop-gdelta ] and corollary [ bratteli09.cor-fsigma ] ) hence the inclusion is not an equality . _ we now give a technical lemma to exhibit a convenient base for the topology of @xmath9 .
given two paths @xmath499 , such that there exists @xmath500 with @xmath501 and @xmath502 we define @xmath503 = \varphi^{-1 } \bigl ( \varphi([\gamma ] ) \cap \bigl ( \varphi([\gamma ' ] ) - a_{\gamma\gamma ' } \bigr ) \bigr)\,,\ ] ] where @xmath327 is the homeomorphism of theorem [ bratteli09.thm-isotrans ] .
so @xmath504 $ ] is the clopen set of tilings in @xmath3 which have the patch @xmath364 at the origin _ and _ a copy of the patch @xmath505 at position @xmath506 .
recall from definition [ bratteli09.def-etale ] that an @xmath9-set is an open set in @xmath9 on which the source and range maps are homeomorphisms .
[ bratteli09.lem-basersets ] for @xmath499 , @xmath507 , the sets @xmath508\ , , \
a(x , y)=a_{\gamma\gamma ' } \bigr\}\,,\ ] ] form a base of @xmath9-sets for the topology of @xmath9 .
we first prove that the @xmath509 form a base for the topology of @xmath9 .
a base open set in @xmath9 reads @xmath510 for a clopen @xmath511 and an open set @xmath512 .
as noted after definition [ bratteli09.def-eqrel ] , the set of vectors @xmath513 is a subset of the countable , discrete , and closed set @xmath514 .
hence we can write @xmath515 as a countable union of open sets of the form @xmath516 for some @xmath517 . since the sets @xmath518,\gamma \in\pi$ ] , form a base for the topology of @xmath3 we can write @xmath519 as a ( finite ) union of open sets of the form @xmath520,\ , a(x , y)= a\}$ ] for some @xmath521 .
and we can choose those @xmath522 for @xmath33 large enough such that @xmath61 belongs to a puncture of a tile in @xmath364 , that is @xmath523 for some @xmath524 , and thus @xmath525 .
note that this proves that the sets @xmath504 $ ] also form a base for the topology of @xmath489 .
hence any open set in @xmath9 is a union of @xmath509 , and therefore the sets @xmath509 form a base for the topology of @xmath9 .
we now prove that @xmath509 is an @xmath9-set . by definition @xmath509
is open , so it suffices to show that the maps @xmath526 and @xmath527 are homeomorphisms .
first note that @xmath528\ , , \quad \text { and } \quad r(o_{\gamma\gamma'})=[\gamma'\gamma ] \,.\ ] ] given @xmath529 , by lemma [ bratteli09.lem-taileq ] , we have @xmath530 . hence given @xmath531 $ ] , there is a unique @xmath532 $ ] such that @xmath533 . and
similarly , given @xmath532 $ ] , there is a unique @xmath531 $ ] such that @xmath533 .
therefore the maps @xmath526 and @xmath527 are one - to - one .
but equation shows that they map base open sets in @xmath9 to base open set in @xmath8 .
hence those maps are homeomorphisms .
we now state the main theorems , which characterize the equivalence relation @xmath9 , and compare it with @xmath4 .
[ bratteli09.thm-etale ] the equivalence relation @xmath9 is _
tale_. we check conditions ( i ) , ( ii ) , and ( iii ) of definition [ bratteli09.thm-etale ] .
\(iii ) let us show first that the maps @xmath43 and @xmath534 are continuous .
it suffices to show that @xmath535=\{(x , y)\,:\,x\in[\gamma],\ , x\sim y\}$ ] and @xmath536)=\{(x , y)\,:\,y\in[\gamma],\ , x\sim y\}$ ] are open in @xmath9 .
pick @xmath537 $ ] ( respectively @xmath538 $ ] ) .
since @xmath539 , there exists @xmath540 such that @xmath541\subset[\gamma]$ ] ( respectively @xmath542\subset[\gamma]$ ] ) . since @xmath543
we have @xmath544)$ ] ( respectively @xmath545)$ ] ) .
thus @xmath546)$ ] ( respectively @xmath536)$ ] ) is open .
we have showed in lemma [ bratteli09.lem-basersets ] that sets @xmath509 are base @xmath9-sets .
hence the maps @xmath43 and @xmath534 are local homeomorphisms . from equation
, we see that they are also open .
\(i ) pick @xmath547 , with @xmath548 , in @xmath549 .
let @xmath550 be such that for all @xmath551\times [ \gamma'_3]$ ] we have @xmath552 .
choose @xmath553 , @xmath554 , with @xmath555 , such that @xmath556 . and choose similarly @xmath557 , @xmath558 , with @xmath559 , such that @xmath560 .
the set @xmath561 is open in @xmath549 and contains @xmath32 .
hence @xmath549 is open , and therefore @xmath562 is closed in @xmath563 .
call @xmath564 the map that sends @xmath109 to @xmath111 , @xmath565 that which sends it to @xmath112 , and @xmath566 that which sends it to @xmath110 .
we have @xmath567 ) \times o_{\gamma\gamma'}\\ p_2^{-1}(o_{\gamma\gamma ' } ) & = & r^{-1}([\gamma\gamma ' ] ) \times s^{-1}([\gamma'\gamma])\\ p_3^{-1}(o_{\gamma\gamma ' } ) & = & o_{\gamma\gamma ' } \times r^{-1}([\gamma'\gamma ] ) \end{array } \right . \,,\ ] ] and the sets on the right hand sides are all open sets in @xmath562 .
hence the maps @xmath568 , and @xmath566 are continuous .
\(ii ) let @xmath569 , so we have @xmath552
. for each @xmath33 pick @xmath570 $ ] , and define @xmath571 to coincide with @xmath379 on @xmath355 , and with @xmath380 on its tail .
since @xmath552 we have @xmath572 , hence @xmath573 , for all @xmath433 for some @xmath390 . since @xmath574 , by lemma [ bratteli09.lem-taileq ] , there exists @xmath575 such that for @xmath576 we have @xmath577 .
set @xmath578 .
we have proved that the sequence @xmath579 has all its elements in @xmath580 , and is such that : @xmath581 in @xmath8 , and @xmath494 .
therefore it converges to @xmath110 in @xmath580 .
this proves that @xmath580 is closed in @xmath9 , hence that @xmath582 is open in @xmath9 .
[ bratteli09.thm-eqrel ] the two equivalence relations @xmath9 on @xmath8 , and @xmath4 on @xmath3 , are homeomorphic : @xmath583 the homeomorphism is induced by @xmath584 from theorem [ bratteli09.thm-isotrans ] .
consider the map @xmath585 , given by @xmath586 since @xmath327 is a homeomorphism , @xmath587 is injective . to prove that is surjective , consider @xmath588 and
let us show that @xmath589 belongs to @xmath9 , _
i.e. _ that @xmath590 . for each @xmath507 , call @xmath591 , the tiles in @xmath254 such that @xmath592 and @xmath593 ( see proposition [ bratteli09.prop-projhulln ] )
. the nested sequence of tilings @xmath228 induces a nested sequence @xmath594 with @xmath595 . in @xmath596
the origin lies in a translate @xmath597 of @xmath370 , and the point @xmath61 in a translate @xmath598 of @xmath599 . since @xmath319 ( condition ( iii ) in definition [ bratteli09.def-nested ] )
, there exists @xmath390 such that for all @xmath470 those two tiles @xmath597 and @xmath598 are within a distance @xmath317 to one another : @xmath600 .
therefore we have @xmath601 and @xmath602 ( definition [ bratteli09.def-ctile ] ) .
this means that for all @xmath470 there exists a horizontal edge @xmath496 with source @xmath370 and range @xmath599 ( definition [ bratteli09.def-cbrat ] ) .
as @xmath603 and @xmath604 are the infinite paths in @xmath489 through the vertices @xmath370 and @xmath599 respectively , we have @xmath605 .
this proves that @xmath587 is a bijection .
now a sequence @xmath487 converges to @xmath110 in @xmath9 if and only if @xmath606 in @xmath8 , and @xmath607 in @xmath1 .
this is the case if and only if @xmath608 in @xmath3 and @xmath607 in @xmath1 , since for all @xmath33 one has @xmath609 by lemma [ bratteli09.lem-taileq ] . hence the map @xmath587 and its inverse are continuous .
[ bratteli09.cor-groupoid ] the groupoid of the equivalence relation @xmath9 is homeomorphic to @xmath12 .
we illustrate here our construction for dimension @xmath37 substitutions . in this case , if one chooses the proper sequence of tilings according to the substitution , then one recovers the usual formalism of bratteli diagrams associated with the ( abelianization matrix of the ) substitution .
we show here that the horizontal structure introduced in definition [ bratteli09.def-cbrat ] allows to recover natural minimal and maximal infinite paths , and that the equivalence relation @xmath9 is then exactly generated by @xmath360 and the set of minimal and maximal paths @xmath610 such that @xmath611 , where @xmath612 is the vershik map on @xmath7 ( corresponding to an associated ordering of the edges ) .
this will be shown carefully in examples , but this fact is general : the following result holds .
[ bratteli09.prop-1d ] let @xmath7 be a collared bratteli diagram associated with a @xmath37-dimensional tiling , with labelled edges .
then there is a partial order on edges , which induces a partial order on infinite paths .
furthermore , there is a one - to - one map @xmath613 from the set of maximal paths to the set of minimal paths , such that : @xmath614 the fact that our labels on vertical edges give a partial ordering on edges is immediate : given @xmath275 , the set @xmath615 is a set of edges encoding the inclusions of tiles in the substitution of @xmath616 . in dimension
@xmath37 , it makes sense to define the edge of @xmath617 which corresponds to the _ leftmost _ tile included in the substitution of @xmath616 ( it is the edge with the _ bigger _ label in @xmath618 ) .
a minimal ( respectively maximal ) path is then a path made uniquely of minimal ( respectively maximal ) edges .
it is then an exercise to show that given two paths in @xmath619 then they are tail - equivalent to a maximal path , respectively a minimal path .
this gives a pairing of minimal with maximal paths , and thanks to the tiles decorations , this pairing is one - to - one .
the fact that the groupoid relation can be recovered from the @xmath5 relation and a finite number of pairs is already known , and our formalism recovers this here .
furthermore , the pairing @xmath613 corresponds actually to the translation of the associated tilings ( more precisely to the action of the fist return map on @xmath3 ) .
this map , the vershik map , can be read from the bratteli diagram ( from the partial ordering of vertices ) , see @xcite for example for the definition of this map .
we treat in details the cases of the fibonacci and the thue - morse tilings .
those tilings have been extensively studied , and we refer the reader to @xcite for a short presentation , and to @xcite for further material .
both tiling spaces are strongly aperiodic , repetitive , and flc @xcite .
this implies that the substitution induces a homeomorphism on the tiling space .
let @xmath620 denote either the fibonacci or thue - morse tiling space , @xmath3 its canonical transversal , and @xmath190 the inflation constant of the substitution .
fix @xmath428 .
we can build a nested sequence of tilings associated with @xmath0 .
for @xmath164 large enough , the nested sequence @xmath621 is proper , because the @xmath164-th substitute of any tile contains one in its interior .
one can easily see however that condition ( ii ) in definition [ bratteli09.def-nested ] can be replaced by the weaker assumption : for example , it is straightforward to see that the proof of theorem [ bratteli09.thm-isotrans ] goes through with this weaker condition . as noted earlier , we can take here @xmath626 independently of the tile , for some fixed @xmath164 .
we will thus consider the sequence @xmath228 , with @xmath627 .
the bratteli diagram built in definition [ bratteli09.def-tilbrat ] , without the label @xmath212 , is therefore exactly the usual bratteli diagram associated with a substitution ( its abelianization matrix ) .
for example , the encoding of words by paths , corresponds also exactly to the encoding of patches given in equation .
let @xmath2 be the fibonacci tiling space , and @xmath3 its canonical transversal .
each tiling in @xmath2 has two types of tiles up to translation , denoted @xmath628 and @xmath37 .
the prototile @xmath628 is identified with the closed interval @xmath629 $ ] with puncture at the origin , and the prototile @xmath37 is identified with @xmath630 $ ] with puncture at the origin , where @xmath631 is the golden mean .
the substitution is given by @xmath632 , @xmath633 , and its inflation constant is @xmath634 .
we write @xmath635 for the collared prototiles , where @xmath636 and where the dot indicates the tile that holds the puncture .
so @xmath637 , and @xmath638 correspond to the tile @xmath628 but with different labels , while @xmath14 corresponds to @xmath37 .
the substitution on collared tiles reads then @xmath639 let @xmath428 , and consider the sequence @xmath228 , with @xmath640 .
the bratteli diagram @xmath7 associated with this proper sequence has then the following form between two generations ( excluding the root ) : @xmath641^(.9){2 } \ar@{-}[ddrrrr]^(.9){2 } & & b_{n-1 } \ar@{-}[ddrrrr]^(.9){1 } & & c_{n-1 } \ar@{-}[ddllll]_(.9){2 } & & d_{n-1 } \ar@{-}[ddllllll]^(.9){1 } \ar@{-}[ddllll]_(.9){1 } \ar@{-}[ddll]_(.9){1}\\ & & & & & & \\ a_n & & b_n & & c_n & & d_n \\ } \ ] ] where @xmath370 is the support of @xmath642 for @xmath643 .
we write an arrow as @xmath644 with @xmath645 and @xmath646 .
we have for @xmath230 @xmath647 note that given a path @xmath648 , the patch @xmath364 of equation corresponds exactly to the word associated with that path in the usual formalism of the bratteli diagram of a substitution .
for example , for the path @xmath649 , we associate the patch or word @xmath650 . where the indices show the patches corresponding to the edges ( with left - right orientation ) .
we have not shown trivial edges , and have identified an edge and its opposite in the drawing .
if @xmath652 is one of the patches ( with this orientation ) in the above graph , we write an horizontal edge @xmath653 with @xmath654 and @xmath655 . and
we will simply use @xmath656 to avoid confusions ( for example between @xmath657 and @xmath658 ) .
for @xmath659 we have @xmath660 there are only two commutative diagrams that one can write between two generations , namely @xmath661^{e^n_{bd } } & & \ar[ll]_{h^{n-1}_{ba } } \ar[dd]_{e^n_{ac } } a_{n-1 } & & & d_{n-1 } \ar[dd]^{e^n_{db } } & & \ar[ll]_{h^{n-1}_{dc } } \ar[dd]_{e^n_{ca } } c_{n-1 } \\ & & & & & \\ d_n & & \ar[ll]^{h^n_{cd } } c_n & & & b_n & & \ar[ll]^{h^n_{ab } } a_n } \ ] ] the translation do match in those diagrams , we have : @xmath662 , and @xmath663 .
let us write @xmath664 and @xmath665 for the left and right above diagrams respectively .
the top horizontal edge of @xmath666 matches the bottom horizontal edge of @xmath667 , and the top horizontal edge of @xmath668 matches the bottom one of @xmath664 .
we can thus `` compose '' those diagrams , and consider the two infinite sequences @xmath669 and @xmath670 .
each of those sequences contains exactly two infinite paths , namely : @xmath671 ( where we have added an edge to the root ) .
if we order the edges in @xmath7 as shown in equation , those are the two minimal and maximal infinite paths . and if we let @xmath612 denote the vershik map on @xmath8 , we have @xmath672 , for @xmath673 .
all those paths are pairwise non - equivalent in @xmath360 , but by definition they are equivalent in @xmath9 .
now given two infinite paths @xmath674 such that @xmath675 and @xmath676 , for @xmath677 or @xmath36 , we have @xmath539 in @xmath9 .
we have thus shown that @xmath9 is generated by @xmath360 and the two pairs @xmath678 and @xmath679 : @xmath680 let @xmath2 be the thue - morse tiling space , and @xmath3 its canonical transversal .
each tiling has two types of tiles up to translation , denoted @xmath628 and @xmath37 .
each prototile is identified with the closed interval @xmath629 $ ] with puncture at the origin .
the substitution is given by @xmath632 , @xmath681 , and its inflation constant is @xmath36 .
we write @xmath682 for the collared prototiles , where @xmath683 and where the dot indicates the tile that holds the puncture .
so @xmath637 , and @xmath684 correspond to the tile @xmath628 but with different labels , while @xmath685 and @xmath686 correspond to @xmath37 .
the substitution on collared tiles reads then @xmath687 let @xmath428 , and consider the nested sequence @xmath228 , with @xmath688 .
the bratteli diagram @xmath7 associated with that sequence has then the following form between two generations ( excluding the root ) : @xmath689^(.9){2 } \ar@{-}[ddrrrrrrrrrr]_(.92){2 } & & b_{n-1 } \ar@{-}[ddll]_(.9){1 } \ar@{-}[ddrrrrrr]_(.9){1 } & & c_{n-1 } \ar@{-}[ddll]_(.94){2 } \ar@{-}[ddrrrr]^(.9){2 } & & d_{n-1 } \ar@{-}[ddll]_(.9){1 } \ar@{-}[ddrrrr]^(.9){1 } & & e_{n-1 } \ar@{-}[ddllllll]^(.9){1 } \ar@{-}[ddllll]^(.9){2 } & & f_{n-1 } \ar@{-}[ddllllllllll]^(.92){2 } \ar@{-}[ddllll]_(.9){1 } \\ & & & & & & & & & & \\ a_n & & b_n & & c_n & & d_n & & e_n & & f_n } \ ] ] where @xmath370 is the support of @xmath642 for @xmath690 .
we write an arrow as @xmath644 with @xmath645 and @xmath646 .
we have for @xmath230 @xmath691 note that given a path @xmath648 , the patch @xmath364 of equation corresponds exactly to the word associated with that path in the usual formalism of the bratteli diagram of a substitution .
for example , for the path @xmath692 , we associate the patch or word @xmath693 . the horizontal graph @xmath342 has the form @xmath694^{ab } \ar@{-}[ddrrrr]^{da } \ar@{-}[dd]_{fa } & & b_{n } \ar@{-}[rr]^{bc } \ar@{-}[ddll]^{bf } & & c_{n } \ar@{-}[dd]^{cd } \ar@{-}[ddll]^{ec } \\ & & & & \\
f_{n } \ar@(ur , ul)@{-}[rr]_{fe } \ar@(dr , dl)@{-}[rr]^{ef } & & e_n \ar@{-}[rr]_{de } & & d_n } \ ] ] where the indices show the patches corresponding to the edges ( with left - right orientation ) .
we have not shown trivial edges , and have identified an edge and its opposite in the drawing . if @xmath652 is one of the patches ( with this orientation ) in the above graph , we write an horizontal edge @xmath653 with @xmath654 and @xmath655 . and we will simply use @xmath656 to avoid confusions ( for example between @xmath695 and @xmath696 ) . for @xmath659 , we have @xmath697 there are only four commutative diagrams that one can write between two generations , namely @xmath698^{e^n_{af } } & & \ar[ll]_{h^{n-1}_{ab } } \ar[dd]_{e^n_{be } } b_{n-1 } & & & f_{n-1 } \ar[dd]^{e^n_{fa } } & & \ar[ll]_{h^{n-1}_{fe } } \ar[dd]_{e^n_{eb } } e_{n-1 } \\ & & & & & \\ f_n & & \ar[ll]^{h^n_{fe } } e_n & & & a_n & & \ar[ll]^{h^n_{ab } } b_n } \ ] ] and @xmath699^{e^n_{ce } } & & \ar[ll]_{h^{n-1}_{cd } } \ar[dd]_{e^n_{df } } d_{n-1 } & & & e_{n-1 } \ar[dd]^{e^n_{ec } } & & \ar[ll]_{h^{n-1}_{ef } } \ar[dd]_{e^n_{fd } } f_{n-1 } \\ & & & & & \\ e_n & & \ar[ll]^{h^n_{ef } } f_n & & & c_n & & \ar[ll]^{h^n_{cd } } d_n } \ ] ] the translation do match in those diagrams , for example we have : @xmath700 , and @xmath701 .
let us write @xmath664 and @xmath665 for the left and right diagrams in equation respectively , and @xmath702 and @xmath703 for those on the left and right of equation respectively .
the diagrams @xmath664 and @xmath665 , and @xmath702 and @xmath703 are `` composable '' : the top horizontal edge of one at generation @xmath35 matches the bottom horizontal edge of the other at generation @xmath33 .
we consider the four infinite sequences @xmath669 and @xmath670 , as well as @xmath704 and @xmath705 .
each of those sequences contains exactly two infinite paths , namely : @xmath706 ( where we have added edges to the root ) .
if we order the edges in @xmath7 as shown in equation , those are the four minimal and four maximal infinite paths . and
if we let @xmath612 denote the vershik map on @xmath8 , we have @xmath672 , for @xmath707 .
all those paths are pairwise non - equivalent in @xmath360 , but by definition they are equivalent in @xmath9 . now given two infinite paths @xmath674 such that @xmath675 and @xmath676 , for @xmath708 or @xmath36 , we have @xmath539 in @xmath9 .
we have thus shown that @xmath9 is generated by @xmath360 and the two pairs @xmath709 for @xmath707 : @xmath710 jean bellissard , georgia institute of technology , school of mathematics and school of physics + postal address : georgia tech , school of math .
, 686 cherry street , atlanta ga , 30332 - 0160 , usa + antoine julien , universit de lyon , universit lyon 1 , institut camille jordan , umr 5208 du cnrs , 43 boulevard du 11 novembre 1918 ,
f-69622 villeurbanne cedex , france + jean savinien , universit de lyon , universit lyon 1 , institut camille jordan , umr 5208 du cnrs , 43 boulevard du 11 novembre 1918 , f-69622 villeurbanne cedex , france bellissard j. , _
@xmath712-theory of @xmath39-algebras in solid state physics _ , in _ statistical mechanics and field theory , mathematical aspects _ ,
dorlas , m. n. hugenholtz & m. winnink eds . , lecture notes in physics , * 257 * , 99 - 156 , ( 1986 ) . bellissard j. , `` gap labelling theorems for schrdinger s operators '' , in _ from number theory to physics _ , pp
. 538 - 630 , les houches march 89 , springer , j.m . luck , p. moussa & m. waldschmidt eds . , ( 1993 ) .
bellissard j. , hermmann d. , zarrouati m. , `` hull of aperiodic solids and gap labeling theorems '' , in _ directions in mathematical quasicrystals _
, crm monograph series , bf 13 , 207 - 259 , m.b .
baake & r.v .
moody eds . , ams providence , ( 2000 ) .
bellissard j. , `` noncommutative geometry of aperiodic solids '' , in _ geometric and topological methods for quantum field theory _ , ( villa de leyva , 2001 ) , pp . 86 - 156 , world sci .
publishing , river edge , nj , ( 2003 ) .
skau c. , `` minimal dynamical systems , ordered bratteli diagrams and associated @xmath714-crossed products '' , _ current topics in operator algebras ( nara , 1990 ) _ , 264280 , world sci .
publ . , river edge , nj , 1991 . | let @xmath0 be an aperiodic and repetitive tiling of @xmath1 with finite local complexity .
let @xmath2 be its tiling space with canonical transversal @xmath3 .
the tiling equivalence relation @xmath4 is the set of pairs of tilings in @xmath3 which are translates of each others , with a certain ( tale ) topology . in this paper @xmath4
is reconstructed as a generalized `` tail equivalence '' on a bratteli diagram , with its standard @xmath5-relation as a subequivalence relation . using a generalization of the anderson
putnam complex @xcite @xmath2 is identified with the inverse limit of a sequence of finite @xmath6-complexes .
a bratteli diagram @xmath7 is built from this sequence , and its set of infinite paths @xmath8 is homeomorphic to @xmath3 . the diagram @xmath7 is endowed with a horizontal structure : additional edges that encode the adjacencies of patches in @xmath0 .
this allows to define an tale equivalence relation @xmath9 on @xmath8 which is homeomorphic to @xmath4 , and contains the @xmath5-relation of `` tail equivalence '' . |
thequest for the origin of matter dominated universe " is a long standing puzzle of our physical world ( see e.g. @xcite ) .
answers to this question might come from further exploration of * the lepton sector cp violation by precise testing of the neutrino oscillation processes * * measure precisely the @xmath2 and the mixing angle @xmath1 ; * * examine matter effect in neutrino oscillation process and possibly conclude the mass hierarchy of neutrinos .
* proton decay : * * search for p @xmath8 @xmath9 k@xmath10 and p @xmath8 e @xmath11 in the range @xmath12 years with assuming non - equilibrium environment in the evolution of universe .
the primary motivation of the t2k experiment @xcite is to improve the sensitivity to the @xmath13 conversion phenomenon in the atmospheric regime by about an order of magnitude in the mixing angle @xmath14 compared to the chooz experimental limit @xcite .
the discovery of a non - vanishing @xmath1 angle would ascertain the @xmath15 nature of the lepton flavor mixing matrix .
the measurements in t2k might indicate that the @xmath1 angle is in a region where the simultaneous determination of neutrino mass hierarchy and the cp violating phase becomes possible .
the j - parc accelerator complex which includes the 180 mev linac , the 3 gev rapid cycling synchrotron ( rcs ) and the 30 - 50 gev main ring synchrotron ( mr ) is planned to be commissioned in 2008 . the j - parc neutrino beam facility , under construction for the t2k experiment ,
is foreseen to begin operation in 2009 .
the final goal for the t2k experiment is to accumulate an integrated proton power on target of @xmath16 mw@xmath17 seconds . within a few years of run , critical information , which will guide the future direction of the neutrino physics , will be obtained based on the data corresponding to about @xmath18 mw @xmath19 10@xmath20 seconds integrated proton power on target ( roughly corresponding to a 3@xmath21 discovery at sin@xmath62@xmath22@xmath23@xmath24 0.05 and 0.03 , respectively ) @xcite .
it is well documented in the literature ( see e.g. refs .
@xcite ) that the most challenging task for next generation long baseline experiments is to unfold the unknown oscillation parameters @xmath25 , @xmath2 and mass hierarchy , @xmath26 , and from the measurement of the energy dependence of the oscillation signal to resolve the so - called problems of `` correlations '' and `` degeneracies '' .
the most important experimental aspects here are the beam profile ( e.g. the ability to cover with sufficient statistics the @xmath27 maximum of the oscillation , the @xmath27 minimum , and the @xmath28 maximum ) , the visible energy resolution of the detector , with which the neutrino energy can be reconstructed , and the spectrum of the misidentified background ( e.g. @xmath11 spectrum , etc . ) .
one possible approach , advocated in the literature @xcite , is to locate a very massive coarse water cerenkov detector at very long baselines ( @xmath29 km ) from a new multi - gev neutrino source and to detect more than one oscillation maximum . in this kind of configuration , although the oscillation peaks are well separated in energy , an important issue is the ability to subtract the backgrounds , in particular that coming from misidentified neutral current events with leading @xmath11 s which introduce new sources of systematic errors . in order to alleviate this problem , a different approach assuming two similar far water cerenkov detectors , one located at kamioka ( @xmath3 km ) and the other in korea ( @xmath30 km ) at the same oa2.5@xmath31 from the j - parc neutrino beam ,
was discussed in ref .
the two detectors would see the same sub - gev neutrino beam ( in fact the same as in t2k ) but the different baselines would allow to study @xmath32 regions corresponding to the 1st and 2nd oscillation maxima .
the approach of our paper is different .
we consider the physics performance of a single 100 kton next generation liquid argon time projection chamber ( lar tpc ) located in the region of okinoshima islands corresponding to a ( relatively modest ) baseline @xmath0 km .
a detector in this location will automatically see the j - parc neutrino beam at an off - axis angle of @xmath33 . as we show below
, this configuration allows to study both 1st and 2nd oscillation maximum peaks with good statistics .
we rely on the realistic opportunity offered by the increase of intensity of the j - parc mr and on the higher precision of the lar tpc than other large detectors to separate the two peaks .
in addition , the @xmath11 background is expected to be highly suppressed thanks to the fine granularity of the readout , hence the main irreducible background will be the intrinsic @xmath34 component of the beam . at this stage of our investigations ,
the emphasis is placed on the measurement of the @xmath1 and @xmath2 parameters , possibly following indications for a non - vanishing @xmath1 in t2k .
we also compare the possibility to determine the @xmath25-@xmath2 parameters with a single detector configuration at other possible baselines like for example the t2k ( tokai - kamioka , oa2.5@xmath31 , @xmath3 km ) or the tokai - eastern korean coast ( @xmath4 km ) with an off - axis oa1.0@xmath31
@xcite . more complete physics performance studies will be presented elsewhere .
if a significant @xmath34 signal were to be observed at t2k , an immediate step forward to a next generation experiment aimed at the discovery of cp - violation in the lepton sector would be recommended with high priority .
we have conducted a case study for the discovery of lepton sector cp violation based on the j - parc mr power improvement scenario .
naturally , next generation far neutrino detectors for lepton sector cp violation discovery will be very massive and large . as a consequence
, the same detector will give us the rare and important opportunity to discover proton decay .
thus , we also discuss the proton decay discovery potential with a huge underground detector .
compared with the t2k experimental conditions , lepton sector cp violation discovery requires * an improved neutrino beam condition ( intensity increase , broader energy spectrum , possibly re - optimization of the focusing optics , ... ) ; * an improved far neutrino detector ( by improvements we primarily mean increased signal reconstruction efficiency , better background separation , better energy resolution , ... , and not only its volume ) .
detector improvements include * detector technology ; * its volume ; * its baseline and off - axis angle with respect to the neutrino source
. as for the neutrino beam intensity improvement , a realistic first step power improvement scenario at j - parc accelerator complex has been recently analyzed and proposed ( linac energy is recovered to be 400 mev , h=1 operation at rcs and 1.92 seconds repetition cycle operation at mr ) . as for the detector technology , we assume a 100 kton liquid argon time projection chamber for our case study .
the effects of the cp phase @xmath2 appear either * as a difference between @xmath9 and @xmath35 behaviors ( this method is sensitive to the @xmath36-odd term which vanishes for @xmath37 or 180@xmath31 ) ; * in the energy spectrum shape of the appearance oscillated @xmath34 charged current ( cc ) events ( sensitive to all the non - vanishing @xmath2 values including 180@xmath31 ) . since antineutrino beam conditions are known to be more difficult than those for neutrinos ( lower beam flux due to leading charge effect in proton collisions on target , small antineutrinos cross - section at low energy , etc . ) , we concentrate on the possibility to precisely measure the @xmath34 cc appearance energy spectrum shape with high resolution during a neutrino - only run .
the liquid argon time projection chamber ( lar tpc ) ( see ref .
@xcite and references therein ) is a powerful detector for uniform and high accuracy imaging of massive active volumes .
it is based on the fact that in highly pure argon , ionization tracks can be drifted over distances of the order of meters .
imaging is provided by position - segmented electrodes at the end of the drift path , continuously recording the signals induced .
@xmath38 is provided by the prompt scintillation light . in this paper , we assume a lar tpc detector with a mass of the order of 100 kton , for example of a kind based on the glacier concept @xcite . unlike other liquid argon tpcs operated or planned which rely on immersed wire chambers to readout the ionization signals , a double phase operation with charge extraction and amplification in the vapor phase is considered here in order to allow for very long drift paths and for improved signal - to - noise ratio .
we assume that successful application of such novel methods will be an important milestone in the r&d for very large lar tpc detectors in the range of 100 kton . at this stage ,
a ton - scale prototype based on this scheme has been developed and is under test @xcite .
challenges to realize liquid argon tpc s with a scale relevant to this paper have been reviewed e.g. in ref .
@xcite .
we refer to previous physics performance studies assuming such a detector configuration @xcite and recall that since liquid argon tpc has advantages on * good energy resolution / reconstruction , * good background suppression , * good signal efficiency it is suitable for a precision measurement of the neutrino energy spectrum to extract cp information .
thus we concentrate on the @xmath34 appearance energy spectrum shape measurement to extract leptonic cp phase information .
the j - parc neutrino beam line was designed and constructed in order to allow an off - axis angle with respect to super - kamiokande ( sk ) between 2.5@xmath31 and 3@xmath31 . a beam setup yielding an oa2.5@xmath31 at sk
was chosen for the t2k experiment . in this configuration
, the center of the t2k neutrino beam will go through underground beneath sk , and will automatically reach the okinoshima island region with an off - axis angle @xmath33 and eventually the sea level east of the korean shore with an off - axis angle @xmath39 .
larger off - axis angles are obtained moving inland korea ( either north , south or west ) .
figure [ fig : blcand ] shows these baseline options using the same beam configuration as the t2k experiment .
parameters for different baselines and beam axes are summarized in figure [ fig : blcand ] and table [ tab : blcand ] .
lccc + & kamioka & okinoshima & korea eastern shore + baseline & 295 km & @xmath40658 km & @xmath401000 km + off - axis angle & 2.5@xmath41 & @xmath400.8@xmath41 & @xmath401.0@xmath41 + the neutrino flavor oscillation probability including atmospheric , solar and interference terms , as well as matter effects , can expressed using the following equation @xcite @xmath42 where , @xmath43}{(1-a)^{2 } } \nonumber\\ t_{2 } & = & \sin \delta_{cp } \cdot \sin 2\theta_{12 } \cdot \sin 2\theta_{23 } \cdot \sin \delta \frac{\sin(a\delta)}{a } \cdot \frac{\sin[(1-a)\delta]}{(1-a ) } \nonumber\\ t_{3 } & = & \cos \delta_{cp } \cdot \sin 2\theta_{12 } \cdot \sin 2\theta_{23 } \cdot \cos \delta \frac{\sin(a\delta)}{a } \cdot \frac{\sin[(1-a)\delta]}{(1-a ) } \nonumber\\ t_{4 } & = & \cos^2 \theta_{23 } \cdot \sin^2 2\theta_{12 } \frac{\sin^2(a\delta)}{a^2}.\end{aligned}\ ] ] where @xmath44 , @xmath45 , @xmath46 .
@xmath47 , @xmath48 , @xmath1 is the mixing angle of the 1st and 3rd generations , while @xmath49 is that for 1st and 2nd , and @xmath50 is that for 2nd and 3rd generations
. oscillations as a function of the e(gev)/l(km ) for @xmath51 ( for other oscillation parameters see text ) .
black curve shows the case for the @xmath2=0 , red shows that for @xmath2=90@xmath31 , blue shows that for @xmath2=270@xmath31 , respectively . , scaledwidth=90.0% ] the term which includes @xmath52 is the `` atmospheric term '' , those including @xmath53 and @xmath54 are `` interference terms '' , and the one that includes @xmath55 is the `` solar term '' .
the `` interference terms '' are sensitive to the @xmath2 phase and play therefore an important role to extract the cp phase . for definite calculations , we use the following parameters ( we assume that most of these parameters will be precisely measured within the timescale of the one discussed here ) : * @xmath50 is @xmath56/4 .
* @xmath49 is 0.572904 rad .
* @xmath57 is 8.2@xmath1910@xmath58 ev@xmath59 .
* @xmath60 is 2.5@xmath1910@xmath61 ev@xmath59 .
* earth density for matter effects are 2.8 g/@xmath62 . * normal hierarchy is assumed unless mentioned otherwise .
figure [ fig : oploe ] shows the oscillation probability as a function of the e(gev)/ l(km ) ( neglecting for the moment matter effects ) . in the plot , the mixing angle @xmath14 was assumed to be 0.1 .
if the distance between source and detector is fixed , the curves can be easily translated to that for the expected neutrino energy spectrum of the oscillated events . as can be seen ,
if the neutrino energy spectrum of the oscillated events could be reconstructed with sufficiently good resolution in order to distinguish first and second maximum , useful information to extract the cp phase would be available even only with a neutrino run .
we note that the following observables depend on leptonic cp phase : * position / height of the first oscillation maximum peak ; * position / height of the second oscillation maximum peak .
the position of the first oscillation minimum is unaffected by it ( in this point all terms of the oscillation probability vanish except the solar term ) .
to effectively experimentally study these observables , we point out that : * the second oscillation maximum peak should end up at sufficiently high energy in order to be measurable ; * the beam should have a sufficiently wide energy range to cover the 1st and 2nd maximum ; * the neutrino flux should be maximized in order to increase as much as possible the statistical significance of the first and second maxima . in order to cover a wider energy range
, we accordingly favor a detector location which is near on - axis . if one assumes that the second oscillation maximum has to be located at an energy larger than about 400 mev , the oscillation baseline should be longer than about 600 km .
in addition , in order to collect enough statistics , we choose a baseline which is not too much longer than above stated . taking into account all of the above mentioned considerations , we privilege the okinoshima region : placing a detector in an appropriate location on the island will probe neutrino oscillations at a baseline of @xmath63 km away from the source at an off - axis angle of @xmath33 .
we assume realistic parameters of the j - parc beam after all accelerator complex upgrades described in the kek road - map are accomplished , as follows @xcite : * the average beam power is reaching 1.66 mw ; * a total of 3.45 @xmath19 10@xmath64 pot is delivered on target per year ; * the optimal kinetic energy of the incident protons is 30 gev . to remain conservative
, we focus on an analysis which uses a neutrino run only during five years under the best j - parc beam assumption .
an antineutrino beam ( opposite horn polarity ) might be considered in a second stage in order to cross - check the results obtained with the neutrino run ( in particular see section [ sec : masshier ] on the mass hierarchy problem ) .
the parameters of the okinoshima location and the assumed beam are therefore as follows : * distance from j - parc is 658 km ; * the axis of the beam is off by 0.76@xmath31 ; * five years operation with horn setting to neutrino run .
the expected neutrino flux calculated under these assumptions is shown in figure [ fig : flux ] where the curves correspond to one year run ( 3.45@xmath65 pot ) .
the black , red , green , blue lines show @xmath66 , @xmath67 , @xmath34 , @xmath68 fluxes , respectively .
the interacting neutrino cross section on argon was computed using the nuance programme @xcite .
we use the followings parameters for the cross section calculation : * number of protons is 18 , and that of neutrons is 22 .
* medium density is 1.4 g/@xmath62 . *
nucleus fermi gas model with binding energy of 30 mev and fermi momentum of 242 mev / c .
taking these into account , we obtain the neutrino cross section shown in figure [ fig : xs ] .
table [ tab : nurate ] shows that total number of charged current ( cc ) events at okinoshima for the null oscillation case and the number of cc @xmath34 events from @xmath69 oscillations for three different mixing angles , and in various @xmath2 scenarios , normalized to five years neutrino run at 1.66 mw and 100 kton fiducial mass .
lcccc + & @xmath66 & @xmath34 & @xmath70 & @xmath71 + beam components ( null oscillation ) & 82000 & 750 & 1460 & 35 + + @xmath72 & @xmath73 & 90@xmath41 & 180@xmath41 & 270@xmath41 + @xmath14 = 0.1 & 2867 & 2062 & 2659 & 3464 + @xmath14 = 0.05 & 1489 & 1119 & 1342 & 1908 + @xmath14 = 0.03 & 942 & 506 & 829 & 1266 + the number of expected events is already a good indicator of the signal , but for this analysis we use binned likelihood fitting for the energy spectrum , therefore all information including total number of events is taken into account , as described in the next section .
images taken with a liquid argon tpc are comparable with pictures from bubble chambers . as it is the case in bubble chambers , events can be analyzed by reconstructing 3d - tracks and particle types for each track in the event image , with a lower energy threshold of few mev for electrons and few tens of mev for protons .
the particle type can be determined from measuring the energy loss along the track ( @xmath75 ) or from topology ( i.e. observing the decay products ) .
additionally , the electronic readout allows to consider the volume as a calorimeter adding up all the collected ionization charge .
the calorimetric performance can be excellent , as we will show in section [ sec : eneres ] , depending on event energy and topology . in order to understand the effect of resolution on physics performance , we show in this section the @xmath34 charged current ( cc ) event energy spectra using different simple models .
one is the perfect resolution case as a reference , and others are 100 mev/200 mev gaussian resolution cases for more realistic cases . this `` resolution '' should include neutrino interaction effects as well as detector resolution .
possible smearing or backgrounds affecting the measurement of the neutrino energy spectrum are listed below subdivided into 4 classes : * neutrino interaction : * * fermi motion and nuclear binding energy , * * nuclear interactions of final state particles within the hit nucleus ( fsi ) , * * vertex nuclear remnant effects ( e.g. nuclear break - up signal ) , * * neutral current ( nc ) @xmath76 event shape including vertex activity . * detector medium : * * ionization processes , * * scintillation processes , * * correlation of between amount of charge and light , * * charge and light quenching , * * hadron transport in argon and secondary interactions , * * charge diffusion and attenuation due to impurity attachment . * readout system including electronics system : * * signal amplification or lack thereof , * * signal - to - noise ratio , * * signal shaping and feature extraction . * reconstruction : * * pattern recognition * * background processes ( nc @xmath11 , @xmath77 cc , ... ) and their event shape * * particle identification efficiency and purity very few real events in liquid argon tpcs have so far been accumulated to allow a full understanding of these complex effects and their interplay .
the only sample comes from a small 50 lt chamber developed by the icarus collab . and
exposed to the cern wanf high energy neutrino beam which collected less than 100 quasi - elastic events @xcite .
it is clear that the energy resolution will depend on several detector parameters , including the readout pitch , the readout method chosen and on the resulting signal - over - noise ratio ultimately affecting the reconstruction of the events .
we therefore stress that significantly improved experimental studies with prototypes exposed to neutrino beams of the relevant energies and sufficient statistics are mandatory to assess and understand these effects . in the meantime , we give in section [ sec : eneres ] preliminary estimates for potentially achievable energy resolutions in a liquid argon medium .
the results are based on full geant3 @xcite simulation of the energy deposited by final state particles in the detector volume , however do not include all possible contributing effects .
figure [ fig : es0mev ] shows the energy spectra of electron neutrino at the cases of @xmath2 equal 0@xmath41 , 90@xmath41 , 180@xmath41 , 270@xmath41 , respectively .
shaded region is common for all plots and it shows the background from beam @xmath34 . here
perfect resolution is assumed for reference to later cases .
= 0@xmath41 ( top - left ) , 90@xmath41 ( right - top ) , 180@xmath41 ( left - bottom ) , 270@xmath41 ( right - bottom ) cases .
, scaledwidth=105.0% ] = 0@xmath41 ( top - left ) , 90@xmath41 ( right - top ) , 180@xmath41 ( left - bottom ) , 270@xmath41 ( right - bottom ) cases . ]
if we smear the energy spectra shown in figure [ fig : es0mev ] with gaussian of sigma equals to 100 or 200 mev independent from original neutrino energy , we obtain spectra shown in figures [ fig : es100mev ] and [ fig : es200mev ] . as seen easily , an energy resolution below 100 mev is crucial since the robustness of the neutrino oscillation is directly determined by the visible second oscillation peak around 400 mev in the energy spectrum . in 200 mev resolution case , the second peak is hidden by the smearing of the 1st oscillation maximum peak .
= 0@xmath41 ( top - left ) , 90@xmath41 ( right - top ) , 180@xmath41 ( left - bottom ) , 270@xmath41 ( right - bottom ) cases .
, width=340 ]
assuming all others were measured very precisely , there are only two free parameters @xmath14 and @xmath2 to be fitted using the energy spectra .
as shown in figure [ fig : oploe ] , the value of @xmath2 varies the energy spectrum , especially the first and the second oscillation peaks ( heights and positions ) , therefore comparison of the peaks determine the value @xmath2 , while the value of @xmath14 changes number of events predominantly . to fit the free parameters , a binned likelihood method is used .
the poisson bin - by - bin probability of the observed data from the expected events ( for an assumed pair of values ( @xmath14 , @xmath2 ) ) is calculated .
the fit procedure is validated by testing the result on a pseudo - experiment .
figure [ fig : ape ] shows one typical pseudo - experiment for the perfect resolution case ( with @xmath2 equals to 0 , and @xmath14 equals to 0.1 ) .
best fit gives reasonable result . at this stage , we only take statistical uncertainty into account for the fitting , thus other systematic uncertainties like far / near ratio , beam @xmath74 shape , energy scale , and so forth are not considered .
also , oscillated signal and beam @xmath74 are only accounted in the fit , i.e. other background like neutral current @xmath76 or beam @xmath68 contamination is assumed to be negligible compared to beam @xmath74 contamination .
as a reference , we first extract allowed regions in the perfect resolution case ( see figure [ fig : allowed_0mev ] ) .
twelve allowed regions are overlaid for twelve true values , @xmath14=0.1 , 0.05 , 0.02 , and @xmath2=0@xmath41 , 90@xmath41 , 180@xmath41 , 270@xmath41 , respectively .
the @xmath2 sensitivity is 20@xmath4030@xmath31 depending on the true @xmath2 value .
= 0.1 , 0.05 , 0.02 , and @xmath2=0@xmath41 , 90@xmath41 , 180@xmath41 , 270@xmath41 , respectively . , width=340 ] = 0.1 , 0.05 , 0.02 , and @xmath2=0@xmath41 , 90@xmath41 , 180@xmath41 , 270@xmath41 , respectively . , width=340 ] = 0.1 , 0.05 , 0.02 , and @xmath2=0@xmath41 , 90@xmath41 , 180@xmath41 , 270@xmath41 , respectively . ,
width=340 ] allowed regions are then extracted for 100 and 200 mev resolution cases .
the used energy spectra are same as figures [ fig : es100mev ] and [ fig : es200mev ] .
results are shown as figures [ fig : allowed_100mev ] and [ fig : allowed_200mev ] .
one obvious but important issue to be pointed out is the robustness of the fitting.the fit procedure shows that results could also be extracted with the 200 mev resolution : this result is as expected statistically ; however , we stress that in this case there is no second oscillation maximum peak visible in figure [ fig : es200mev ] .
hence , we think it is mandatory to keep an energy resolution less than 100 mev as goal for the credibility of this experiment .
the influence of matter on neutrino oscillations was first considered by wolfenstein @xcite . as is well - known ( although never directly experimentally verified ) , oscillation probabilities get modified under these conditions .
matter effects are sensitive to the neutrino mass ordering and different for neutrinos and antineutrinos .
as mentioned earlier , we consider in this paper the possibility of a neutrino - only run .
hence , we briefly address in this section the question of normal hierarchy ( nh ) versus inverted hierarchy ( ih ) .
since we are focusing on the potential discovery of cp - violation in the leptonic sector , our discussion is geared towards possible ambiguities that would arise if the mass hierarchy was unknown .
indeed , figure [ fig : dmresults ] illustrates the results of a fit of a pseudo - data with nh by both nh ( black ) and ih ( red ) hypotheses , assuming only the neutrino run . the best fit likelihood value with one assumed hierarchy is used to calculate the likelihood variation @xmath78 for both hierarchies .
one could claim that cp - violation is discovered if the experimental results of a given experiment exclude the @xmath2 phase to be either 0 or 180@xmath31 .
hence , the danger is that the lack of knowledge of the mass hierarchy ( or rather the `` wrong '' choice in hypothesis when selecting the hierarchy in the fit of the data ) , gives a result for @xmath2 consistent with either 0 or 180@xmath31 . on the other hand ,
if fits of a given experiment with both assumed hierarchies provide neither 0 nor 180@xmath31 , one can declare `` discovery '' although the mass hierarchy could not be determined . alternatively ,
if it turned out in the actual experiment that one of results of the fit is consistent with @xmath37 or 180@xmath31 , one would still have the possibility to consider an anti - neutrino run in order to solve this ambiguity .
, one can declare discovery of cp violation in the leptonic sector . if any of the fits results in a @xmath2 of 0 or 180@xmath31 , then an anti - neutrino run could be envisaged .
, scaledwidth=90.0% ] for 5 years of neutrino run comparing 100 kton lar at okinoshima with sk at kamioka and 20 kton lar at kamioka . , scaledwidth=90.0% ] our results indicate that cp - violation would be unambiguously discovered under our assumption for true values of @xmath2 s in the region of 90@xmath31 and 270@xmath31 .
alternatively , in case of true values of @xmath2 s near 0@xmath31 or 180@xmath31 an antineutrino flux would help untangle the two solutions if the neutrino mass hierarchy was unknown .
although this paper has focused on the possibility to measure the @xmath25 and @xmath2 oscillation parameters , it is instructive to estimate the sensitivity of the potential setup in the case of negative result from t2k .
the corresponding sensitivity to discover @xmath1 in the true ( @xmath79 plane at 90% c.l . and
@xmath80 is shown in figure [ fig : t13sensi ] , where we assumed 5 years of neutrino run comparing the 100 kton lar at okinoshima with increased statistics in sk at kamioka and a potential 20 kton lar at kamioka . in order to discover a non - vanishing @xmath25 ,
the hypothesis @xmath81 must be excluded at the given c.l . as input ,
a true non - vanishing value of @xmath25 is chosen in the simulation and a fit with @xmath82 is performed , yielding the `` discovery '' potential .
this procedure is repeated for every point in the ( @xmath83 ) plane . at the @xmath80 c.l .
the sensitivity of the t2k experiment is 0.02 @xcite . continuing to collect data at sk with an improved j - parc neutrino beam for another 5 years
would improve the t2k sensitivity by a factor @xmath84 . in comparison ,
our simulations indicate that a 20 kton lar tpc is expected to perform better than this although the masses are comparable , since the signal efficiency is higher than sk and the nc background is assumed to be negligible contrary to sk .
even better , a 100 kton lar tpc at okinoshima would further improve the sensitivity by about a factor six compared to sk at kamioka ( or a factor 10 compared to t2k ) , thanks to its bigger mass , increased cross - section at higher energies and reduced off - axis angle , although the neutrino fluxes at the same off - axis angle would be reduced by a factor @xmath85 because of the longer baseline .
as mentioned previously , the neutrino energy resolution expected in lar medium depends on several parameters and their interplay that ultimately need to be measured experimentally . in principle ,
energy and momentum conservation allow to estimate the incoming neutrino energy in the detector via a precise measurement of decay products , with the exception of the smearing introduced by fermi motion and other nuclear effects ( nuclear potential , re - scattering , absorption , etc . ) for interactions on bound nucleons . in this section
, we try to estimate the smearing introduced by these effects with the help of exclusive neutrino event final states distributed with the okinoshima flux and generated with the genie mc @xcite subsequently fully simulated with geant3 .
nuclear effects in neutrino interactions can be roughly divided into those of the nuclear potential and those due to reinteractions of decay products .
bound nucleons and other hadrons in nuclei are subject to a nuclear potential . the fermi energy ( or momentum ) must be calculated from the bottom of this nuclear potential well , and the removal of a nucleon from any stable nucleus is always an endothermic reaction .
when hadrons are produced in the nucleus , some energy is spent to take it out of this well : for a nucleon , the minimum energy is given by the nucleon separation energy ( around 8 mev ) , and corresponds to a nucleon at the fermi surface .
in this case , the daughter nucleus is left on its ground state .
more deeply bound nucleons , leaving a hole in the fermi sea , correspond to an excitation energy of the daughter nucleus , and an additional loss of energy of the final state products .
this energy is then spent in evaporation and/or gamma deexcitation .
thus , the energy of the final state products is expected to be always slightly smaller than for interactions on free nucleons , and spread over a range of about 40 mev .
correspondingly , the fermi momentum is transferred to the decay products and compensated by the recoil of the daughter nucleus .
additional momentum distortions come from the curvature of particle trajectories in the nuclear potential .
reinteractions in the nuclear medium also play an important role .
interaction products can lose part of their energy in collisions , or even be absorbed in the same nucleus where they have been created .
this is particularly true for pions , that have an important absorption cross section on nucleon pairs , while kaons have smaller interaction probability .
once final state products have exited the nucleus , they will propagate in the medium with the possibility that further interactions occur . in the case of the liquid argon tpc ,
the medium is fully homogeneous and the detector is fully active .
all deposited energy in the medium ( above a certain threshold ) will be eventually collected .
several methods can be adopted to reconstruct the neutrino energy and we list in the following three : = 0@xmath41 : perfect reconstruction ( top - left ) , lar using deposited energy(right - top ) , ibid but with charged pion mass correction ( left - bottom ) using final state lepton only ( like in wc ) ( right - bottom ) cases . , scaledwidth=97.5% ] 1 .
* final state lepton : * this method , traditionally used in large water cerenkov detectors like kamiokande , imb or sk , relies on the precise measurement of the energy and direction of the outgoing lepton and kinematically constrains the incoming neutrino energy by assuming a quasi - elastic configuration : @xmath86 where @xmath87 , @xmath88 and @xmath89 are the energy , momentum and scattering angle of the outgoing lepton .
this method is sensitive to the final state configuration and to fermi motion ( up to @xmath90 mev / c ) which randomizes the direction of the outgoing lepton .
* momentum conservation : * neglecting the incoming neutrino mass , one obtains @xmath91 where @xmath92 is the momentum of the outgoing lepton , @xmath93 is the momentum of the outgoing hadron @xmath94 and @xmath95 is the fermi motion of the hit nucleon ( up to @xmath90 mev / c ) .
this method is also sensitive to fermi motion since the recoiling remnant hit nucleus ( with momentum @xmath96 ) is not measured .
* energy conservation : * using a calorimetric approach , one can sum the deposited energies of all outgoing particles . in a tracking - calorimeter
this is obtained by summing the @xmath75 measurements along each ionizing track to obtain the associated kinetic energies @xmath97 .
one should identify final state particles in order to take into account their rest masses .
this method is sensitive to the nuclear binding energy of the decay products and is intrinsically more precise than the above method relying on momenta for the energies considered here .
mathematically one can write this result as : @xmath98 all methods are sensitive to potential re - interactions of the outgoing hadrons within the nucleus . in order to estimate the potential of a fully homogeneous and sensitive medium like the liquid argon tpc , we performed full simulations of neutrino interactions : the event 4-vector are generated with the genie mc and final state particles ( after nuclear reinteraction ) were propagated through the liquid argon medium with geant3 .
the geometry of the setup is an infinite lar box and at this stage we have neglected charge quenching ( i.e. we assume a linear response to @xmath75 ) .
we also assume a 100% efficiency to identify final state charged pions . the achievable incoming neutrino energy resolution in the liquid argon medium has been estimated using @xmath34 cc events distributed with the @xmath77 flux expected at the okinoshima location .
the results , using the energy conservation method , are shown in figure [ fig : esgeantene ] .
although okinoshima looks a very good candidate as mentioned so far , the favorable geography would allow , in principle , for a few baseline candidates for the detector locations ; kamioka for a relatively short baseline , okinoshima for a medium baseline and korea for the longest baseline .
it is worthwhile to compare the physics potential at okinoshima with other scenarios with the same strategy and same assumed beam conditions , i.e. five years neutrino run at 1.66 mw .
figure [ fig : blresults ] illustrates the analyzed allowed regions using each option .
a perfect resolution was assumed here to directly compare the physics potential of the various sites .
as seen , sensitivity of @xmath2 is similar between okinoshima and korea , but that in kamioka is much worse than others .
this indicates the analysis strategy using only neutrino run is not suitable for kamioka case , as was expected since the oa2.5@xmath31 available at kamioka does not allow to cover 1st and 2nd oscillation maxima peaks . as for the sensitivity of the @xmath14
, okinoshima is better than korea , due to the increased statistics at the shorter baseline .
grand unification of the strong , weak and electromagnetic interactions into a single unified gauge group is an extremely appealing idea @xcite which has been vigorously pursued theoretically and experimentally for many years .
the detection of proton or bound - neutron decays would represent its most direct experimental evidence .
the physics potentialities of very large underground liquid argon tpc was recently carried out with detailed simulation of signal efficiency and background sources , including atmospheric neutrinos and cosmogenic backgrounds @xcite .
it was found that a liquid argon tpc , offering good granularity and energy resolution , low particle detection threshold , and excellent background discrimination , should yield very good signal over background ratios in many possible decay modes , allowing to reach partial lifetime sensitivities in the range of @xmath99 years with exposures up to 1000 kton@xmath19year , often in quasi - background - free conditions optimal for discoveries at the few events level , corresponding to atmospheric neutrino background rejections of the order of @xmath100 .
multi - prong decay modes like e.g. @xmath101 or @xmath102 and channels involving kaons like e.g. @xmath103 , @xmath104 and @xmath105 are particularly suitable , since liquid argon imaging provides typically an order of magnitude improvement in efficiencies for similar or better background conditions compared to water cerenkov detectors .
up to a factor 2 improvement in efficiency is expected for modes like @xmath106 and @xmath107 thanks to the clean photon identification and separation from @xmath11 .
channels like @xmath108 or @xmath109 , dominated by intrinsic nuclear effects , yield similar efficiencies and backgrounds as in water cerenkov detectors .
thanks to the self - shielding and 3d - imaging properties of the liquid argon tpc , the result remains valid even at shallow depths where cosmogenic background sources are important . in conclusion , a lar tpc would not necessarily require very deep underground laboratories even for high sensitivity proton decay searches .
once again , we stress the importance of an experimental verification of the physics potentialities to detect , reconstruct and classify events in the relevant gev energy range
. this experimental verification will require the collection of neutrino event samples with high statistics , accessible e.g. with a detector located at near sites of long baseline artificial neutrino beams .
in this paper , we have reported our case study for a new future long baseline neutrino and nucleon decay experiment with a 100 kton liquid argon tpc located in the region of okinoshima using the j - parc neutrino facility .
our study assumes a realistic upgrade of the j - parc main ring operation yielding an average 1.66 mw beam power for neutrinos coupled to the new far detector at okinoshima , which would then see the beam at an off - axis angle of @xmath110 and at a baseline of @xmath111 km . in this first discussion , we focused on the possibility to measure the @xmath1 and @xmath2 parameters .
our strategy is based on the precise measurement of the energy spectrum of the oscillated events , and in particular on the comparison of the features of 1st maximum oscillation peak with those of the 2nd maximum oscillation peak . in order to efficiently determine and study those features
, we rely on the excellent neutrino event reconstruction and incoming neutrino energy resolution , as is expected in the case of the fully - sensitive and fine charge imaging capabilities of the liquid argon tpc . beyond this case study
, we intend to perform more detailed simulations of the performance of the experiment including for example different sources of systematic errors and sources of backgrounds which have been up - to - now neglected .
the construction and operation of a 100 kton liquid argon tpc certainly represents a technological challenge at the present state of knowledge of the technique . for it to become a realistic option
, we stress the importance and necessity of further r&d and of dedicated experimental measurement campaigns . at this stage , we intend to pursue our investigations on a ton - scale prototype based on the novel double - phase readout imaging method . options to operate small devices in the j - parc neutrino beam are being assessed in parallel . in addition , we have started to address the possibility of an `` intermediate '' prototype , presumably in the range of 1 kton of mass , based on a similar but scaled down design of the potential 100 kton detector .
part of this work was supported by the swiss national foundation and eth zurich .
m. apollonio _ et al . _
[ chooz collaboration ] , `` limits on neutrino oscillations from the chooz experiment , '' phys .
b * 466 * , 415 ( 1999 ) t. hasegawa , talk given at the 4th international workshop on nuclear and particle physics at j - parc ( np08 ) .
slides available at ` http://j-parc.jp/np08/ ` j. burguet - castell , m. b. gavela , j. j. gomez - cadenas , p. hernandez and o. mena , `` on the measurement of leptonic cp violation , '' nucl .
b * 608 * , 301 ( 2001 ) [ arxiv : hep - ph/0103258 ] .
h. minakata and h. nunokawa , `` exploring neutrino mixing with low energy superbeams , '' jhep * 0110 * , 001 ( 2001 ) [ arxiv : hep - ph/0108085 ] . v. barger , d. marfatia and k. whisnant , `` breaking eight - fold degeneracies in neutrino cp violation , mixing , and mass hierarchy , '' phys . rev .
d * 65 * , 073023 ( 2002 ) [ arxiv : hep - ph/0112119 ] .
w. j. marciano , `` extra long baseline neutrino oscillations and cp violation , '' arxiv : hep - ph/0108181 .
m. v. diwan _ et al .
_ , `` very long baseline neutrino oscillation experiments for precise measurements of mixing parameters and cp violating effects , '' phys . rev .
d * 68 * ( 2003 ) 012002 [ arxiv : hep - ph/0303081 ] .
m. ishitsuka , t. kajita , h. minakata and h. nunokawa , `` resolving neutrino mass hierarchy and cp degeneracy by two identical detectors with different baselines , '' phys .
d * 72 * , 033003 ( 2005 ) [ arxiv : hep - ph/0504026 ] .
k. hagiwara , n. okamura and k. i. senda , `` physics potential of t2kk : an extension of the t2k neutrino oscillation experiment with a far detector in korea , '' phys .
d * 76 * , 093002 ( 2007 ) [ arxiv : hep - ph/0607255 ] .
a. meregaglia and a. rubbia , `` neutrino oscillations with a next generation liquid argon tpc detector in kamioka or korea along the j - parc neutrino beam , '' arxiv:0801.4035 [ hep - ph ]
. s. amerio _ et al .
_ , `` design , construction and tests of the icarus t600 detector '' , nucl .
meth . a 527 ( 2004 ) 329 and references therein .
a. rubbia , `` ardm : a ton - scale liquid argon experiment for direct detection of dark matter in the universe , '' j. phys .
* 39 * ( 2006 ) 129 [ arxiv : hep - ph/0510320 ] .
a. marchionni , talk given at the 4th international workshop on nuclear and particle physics at j - parc ( np08 ) .
slides available at ` http://j-parc.jp/np08/ ` a. meregaglia and a. rubbia , `` neutrino oscillation physics at an upgraded cngs with large next generation liquid argon tpc detectors , '' jhep * 0611 * , 032 ( 2006 ) [ arxiv : hep - ph/0609106 ] .
m. freund , `` analytic approximations for three neutrino oscillation parameters and probabilities in matter , '' phys . rev .
d * 64 * , 053003 ( 2001 ) [ arxiv : hep - ph/0103300 ] .
a. cervera , a. donini , m. b. gavela , j. j. gomez cadenas , p. hernandez , o. mena and s. rigolin , `` golden measurements at a neutrino factory , '' nucl .
b * 579 * , 17 ( 2000 ) [ erratum - ibid .
b * 593 * , 731 ( 2001 ) ] [ arxiv : hep - ph/0002108 ] . ` http://nuint.ps.uci.edu/nuance/ ` f. arneodo _ et al .
_ [ icarus - milano collaboration ] , `` performance of a liquid argon time projection chamber exposed to the wanf neutrino beam , '' phys .
d * 74 * , 112001 ( 2006 ) [ arxiv : physics/0609205 ] .
r. brun _ et al .
_ , geant3 reference manual , .
cern program library long writeup .
w5013 , dd / ee/84 - 1 , 1987 . for hadronic secondary interactions ,
we select gfluka .
j. c. pati and a. salam , `` unified lepton - hadron symmetry and a gauge theory of the basic interactions , '' phys .
d * 8 * , 1240 ( 1973 ) . h. georgi and s. l. glashow , `` unity of all elementary particle forces , '' phys .
lett . * 32 * , 438 ( 1974 ) .
a. bueno _ et al .
_ , `` nucleon decay searches with large liquid argon tpc detectors at shallow depths : atmospheric neutrinos and cosmogenic backgrounds , '' jhep * 0704 * , 041 ( 2007 ) [ arxiv : hep - ph/0701101 ] . | in this paper , we consider the physics performance of a single far detector composed of a 100 kton next generation liquid argon time projection chamber ( lar tpc ) possibly located at shallow depth , coupled to the j - parc neutrino beam facility with a realistic 1.66 mw operation of the main ring .
the new far detector could be located in the region of okinoshima islands ( baseline @xmath0 km ) .
our emphasis is based on the measurement of the @xmath1 and @xmath2 parameters , possibly following indications for a non - vanishing @xmath1 in t2k , and relies on the opportunity offered by the lar tpc to reconstruct the incoming neutrino energy with high precision compared to other large detector technologies .
we mention other possible baselines like for example j - parc - kamioka ( baseline @xmath3 km ) , or j - parc - eastern korean coast ( baseline @xmath4 km ) .
such a detector would also further explore the existence of proton decays . + * a possible future long baseline neutrino and nucleon decay experiment with a 100 kton liquid argon tpc at okinoshima using the j - parc neutrino facility * a.badertscher@xmath5 , t.hasegawa@xmath6 , t.kobayashi@xmath6 , a.marchionni@xmath5 , a.meregaglia@xmath5 , t.maruyama@xmath7 , k.nishikawa@xmath6 , and a.rubbia@xmath5 + _ ( 1 ) eth zrich , ( 2 ) kek ipns , ( 3 ) university of tsukuba _ + |
leptogenesis @xcite is one of the most attractive mechanisms for explaining the observed baryon asymmetry of the universe , @xmath0 @xcite . in the type
i see - saw scenario @xcite , where the asymmetry is generated via the out - of - equilibrium decay of the same heavy right - handed neutrinos which are involved in generating neutrino masses , it has been studied intensively . in models with a left - right symmetric particle content like minimal left - right symmetric models , pati - salam models or grand unified theories ( guts ) based on @xmath1 ,
the type i see - saw mechanism is typically generalized to a type ii see - saw ( see e.g. @xcite ) , where an additional direct mass term @xmath2 for the light neutrinos is present .
the effective mass matrix of the light neutrinos is then given by @xmath3 is the type i see - saw mass matrix .
one motivation for considering the type ii see - saw is that it allows to construct models for partially degenerate neutrinos in a natural way , e.g. via a type ii upgrade @xcite , which is otherwise difficult to achieve in type i models . from a rather model independent viewpoint , the type ii mass term can be considered as an additional contribution to the lowest dimensional effective neutrino mass operator . in the literature ,
the most discussed case is where the type ii contribution is realized via su(2)@xmath4-triplets .
there are in general two possibilities to generate the baryon asymmetry : via the decay of the lightest right - handed neutrino @xmath5 or via the decay of one or more su(2)@xmath4-triplets @xcite . in the first case
, there are additional one - loop diagrams where virtual triplets are running in the loop @xcite .
referring to the contributions to the decay asymmetries for @xmath5 proportional to @xmath6 as @xmath7 and to the ones proportional to @xmath2 as @xmath8 , either or both contributions can be important for generating the baryon asymmetry .
in many studies of leptogenesis in left - right symmetric models , it has been assumed that @xmath7 dominates even if neutrino masses stem dominantly from @xmath2 ( see e.g. @xcite ) .
the case where @xmath8 dominates over @xmath7 has recently been studied in @xcite .
it has the interesting feature that unlike in the type i see - saw scenario , there is in general no upper bound on the absolute neutrino mass scale @xcite from type ii leptogenesis , as has been pointed out in @xcite . in this work ,
we analyze the consequences of the neutrino mass scale for baryogenesis via the out - of - equilibrium decay of the lightest right - handed ( s)neutrinos in type ii see - saw models .
first , we calculate the type ii contributions to the decay asymmetries for minimal scenarios based on the standard model ( sm ) and on the minimal supersymmetric standard model ( mssm ) , where the additional direct mass term for the neutrinos stems from the induced vev of a triplet higgs .
the result for the supersymmetric case is new and we correct the previous result in the scenario based on the standard model .
we then develop an effective approach to type ii leptogenesis , assuming a gap between the mass @xmath9 of the lightest ( s)neutrino and the masses of the heavier particles involved in generating neutrino masses .
leptogenesis in this framework is approximately independent of the specific realization of the neutrino mass operator .
the calculation of the decay asymmetries using the effective approach confirms our results for the triplet scenarios in the limit of heavy triplets .
we subsequently derive a general upper bound on the decay asymmetry and find that it increases with the neutrino mass scale .
it leads to a lower bound on the mass of the lightest right - handed neutrino , which is significantly below the type i bound for partially degenerate neutrinos .
it is worth emphasizing that these results are in sharp contrast to the type i see - saw mechanism where an upper bound on the neutrino mass scale is predicted .
here we find no upper limit on the neutrino mass scale which may be increased arbitrarily .
indeed we find that the lower bound on the mass of the lightest right - handed neutrino decreases as the physical neutrino mass scale increases .
this allows a lower reheat temperature , making thermal leptogenesis more consistent with the gravitino constraints in supersymmetric models @xcite ) .
[ sec : tripletdecayasymmetries ] we now consider minimal type ii models based on the sm and the mssm , where the type ii see - saw is realized via an additional heavy su(2)@xmath4-triplet .
we focus on the case where the asymmetry is generated via the decay of the lightest right - handed ( s)neutrinos and assume hierarchical masses of the right - handed ( s)neutrinos and @xmath10 . in the mssm extended by chiral superfields @xmath11 @xmath12 , which contain the right - handed neutrinos @xmath13 and @xmath14-triplet higgs superfields @xmath15 and @xmath16 with weak hypercharge @xmath17 and @xmath18 , respectively , the relevant parts of the superpotential are @xmath19 the dot indicates the @xmath14-invariant product , @xmath20 , with @xmath21 @xmath22 being the pauli matrices .
superfields are marked by hats and we have written the @xmath23-triplets as traceless @xmath24-matrices @xmath25 in the sm , we only consider one triplet scalar field @xmath26 , using an analogous notation as in the extended mssm .
the corresponding terms of the lagrangian are @xmath27 at low energy in the sm and in the mssm , the type i contribution to the neutrino mass matrix of the light neutrinos is approximately given by the see - saw formula of equation ( [ eq : seesawformula ] ) , @xmath28 @xmath29 is the vacuum expectation value ( vev ) of the neutral component of the higgs doublet which couples to the right - handed neutrinos and the lepton doublets , i.e. @xmath30 in the mssm and @xmath31 in the sm .
an induced see - saw suppressed vev @xmath32 of the neutral component of the scalar field contained in @xmath15 gives a naturally small direct mass for the left - handed neutrinos .
it can also be viewed as resulting from realizing the effective neutrino mass operator by integrating out the triplet below its mass threshold at @xmath33 .
the type ii contribution to the effective neutrino mass matrix is given by @xmath34 the complete neutrino mass matrix in the minimal type ii scenarios based on the sm and on the mssm is thus given from the above equations as @xmath35 in this subsection we calculate the relevant decay asymmetries diagrammatically .
the asymmetry from the decay of the lightest right - handed neutrino into a lepton doublet and a higgs is defined as @xmath36 with the decay rate @xmath37 .
in addition , in the supersymmetric case , we need the decay asymmetries @xmath38 for the decay of @xmath5 into slepton and higgsino and for the decays of the sneutrino @xmath39 . at tree level ,
the decay rates are @xmath40 the contributions to the decay asymmetries arise from the interference of the diagrams for the tree - level decays with the loop diagrams .
the one - loop diagrams for the decay @xmath41 are shown in figure [ fig : leptogenesis_triplet_susy ] .
compared to the supersymmetric type i see - saw case , there are additional diagrams contributing to @xmath42 , [ fig : leptogenesis_triplet_susy](c ) and [ fig : leptogenesis_triplet_susy](f ) , which involve the triplet higgs and its superpartner .
the additional diagrams corresponding to the decay of @xmath5 into slepton and higgsino and to the decays of the sneutrino @xmath39 are not shown explicitly , but are included in the analysis .
@xmath43{typeiiloops.eps}}}$ ] using feyncalc @xcite , the calculation of the decay asymmetries corresponding to the diagrams [ fig : leptogenesis_triplet_susy](c ) and [ fig : leptogenesis_triplet_susy](f ) of figure [ fig : leptogenesis_triplet_susy ] yields [ eq : decayastriplet]@xmath44}{(y_\nu^\dagger y_\nu)_{11 } } \ , y \,\left [ - 1 + y \ln \left ( \frac{y + 1}{y}\right ) \!\right ] \!,\\ \varepsilon^{(f)}_{1 } & \!\!=\!\ ! & -\frac{3}{8 \pi } \frac{m_{\mathrm{r}1}}{v^2_\mathrm{u } } \frac{\sum_{fg}\mbox{im}\ , [ ( y^*_\nu)_{f1 } ( y^*_\nu)_{g1 } ( m_{\mathrm{ll}}^{\mathrm{ii}})_{fg}]}{(y_\nu^\dagger y_\nu)_{11 } } \ , y \,\left[1 - ( 1+y ) \ln \left ( \frac{y + 1}{y}\right ) \!\right ] \!.\end{aligned}\ ] ] we have defined @xmath45 . for dealing with lepton number violating interactions
, we use the methods derived in @xcite .
the results for the contributions to the decay asymmetries from the triplet in the sm and from the triplet superfields in the mssm are @xmath46 the mssm results are new . in the sm , we correct the previous result of @xcite by a factor of @xmath47 .
as we will see below , our results in the limit @xmath48 agree with the calculation in the effective approach , where the particles much heavier than @xmath9 are integrated out . in the mssm
, we furthermore obtain @xmath49 in addition , we reproduce the known results @xcite for the decay asymmetries corresponding to the diagrams ( a ) ,
( b ) , ( d ) and ( e ) which contribute to @xmath50 in the sm and in the mssm : [ eq : effdecayass_mintypeiiscenarios]@xmath51 } { \sum_f \ , |(y_\nu)_{f1}|^2 } \ , \sqrt{x_j } \,\left[1 - ( 1+x_j ) \ln \left ( \frac{x_j + 1}{x_j}\right ) \right ] , \\
\varepsilon^{(b)}_{1 } & = & \frac{1}{8 \pi } \frac{\sum_{j\not=1 } \mbox{im}\ , [ ( y^\dagger_\nu y_\nu)^2_{1j } ] } { \sum_f \ , |(y_\nu)_{f1}|^2 } \ , \sqrt{x_j } \,\left [ \frac{1}{1-x_j } \right ] , \\
\varepsilon^{(d)}_{1 } & = & \frac{1}{8 \pi } \frac{\sum_{j\not=1}\mbox{im}\ , [ ( y^\dagger_\nu y_\nu)^2_{1j } ] } { \sum_f \ , |(y_\nu)_{f1}|^2 } \ , \sqrt{x_j } \,\left [ - 1 + x_j \ln \left ( \frac{x_j + 1}{x_j}\right ) \right ] , \\ \varepsilon^{(e)}_{1 } & = & \frac{1}{8 \pi } \frac{\sum_{j\not=1 } \mbox{im}\ , [ ( y^\dagger_\nu y_\nu)^2_{1j } ] } { \sum_f \ , |(y_\nu)_{f1}|^2 } \ , \sqrt{x_j } \,\left [ \frac{1}{1-x_j } \right],\end{aligned}\ ] ] with @xmath52 for @xmath53 .
the results for the type i contribution to the decay asymmetries in the sm and in the mssm are @xmath54 in the mssm , the remaining decay asymmetries are equal to @xmath55 @xcite , @xmath56 note that the type i results can be brought to a form which contains the neutrino mass matrix using @xmath57 } { \sum_f \ , |(y_\nu)_{f1}|^2 } \ , \frac{1}{\sqrt{x_j } } = -\frac{1}{8 \pi } \frac{m_{\mathrm{r}1}}{v^2_\mathrm{u } } \frac{\sum_{fg}\mbox{im}\ , [ ( y^*_\nu)_{f1 } ( y^*_\nu)_{g1 } ( m_{\mathrm{ll}}^{\mathrm{i}})_{fg}]}{(y_\nu^\dagger y_\nu)_{11}}\,.\end{aligned}\ ] ] in the limit @xmath48 and @xmath58 for all @xmath53 , which corresponds to a large gap between the mass @xmath9 and the masses @xmath59 , @xmath60 and @xmath33 , using @xmath61 & \stackrel{z \gg 1}{\rightarrow } & - \frac{1}{2 } \ ; , \\ z \,\left [ \frac{1}{1-z } \right ] & \stackrel{z \gg 1}{\rightarrow } & - 1 \ ; , \\ z \,\left [ - 1 + z\ln \left ( \frac{z + 1}{z}\right ) \right ] & \stackrel{z \gg 1}{\rightarrow } & - \frac{1}{2 } \end{aligned}\ ] ] for @xmath62 , we obtain the simple results for the decay asymmetries @xmath63 and @xmath64 , @xmath65}{\sum_h \ , |(y_\nu)_{h1}|^2}\ ; , \\ \varepsilon^{\mathrm{mssm}}_{1 } & = & \frac{3}{8 \pi } \frac{m_{\mathrm{r}1}}{v_\mathrm{u}^2 } \frac{\sum_{fg}\mbox{im}\ , [ ( y^*_\nu)_{f1 } ( y^*_\nu)_{g1 } ( m^{\mathrm{i}}_{\mathrm{ll}}+ m^{\mathrm{ii}}_{\mathrm{ll}})_{fg}]}{\sum_h \ , in the presence of such a mass gap , the calculation can also be performed in an effective approach after integrating out the two heavy right - handed neutrinos and the heavy triplet , generating contributions to the effective neutrino mass operator , as we now discuss .
in the sm and the mssm , viewed as effective theories , neutrino masses can be introduced via the lowest dimensional effective neutrino mass operator [ eq : the3formsofnumassop ] @xmath66 after electroweak symmetry breaking , the effective operator yields majorana masses for the light neutrinos , @xmath67 let us assume that the lepton asymmetry is generated via the decay of the lightest right - handed neutrino and that all other additional particles , in particular the ones which generate the type ii contribution , are much heavier than @xmath9 .
we furthermore assume that we can neglect their population in the early universe , e.g. that their masses are much larger than the reheating temperature @xmath68 and that they are not produced non - thermally in a large amount .
we also assume that they approximately do not contribute to washout processes .
this scenario is motivated by supersymmetric guts , where additional charged particles like e.g. su(2)@xmath4-triplets with intermediate scale masses could spoil gauge coupling unification at @xmath69 gev . for a minimal effective approach ,
it is convenient to isolate the type i contribution from the lightest right - handed neutrino as follows : @xmath70 .\ ] ] @xmath71 includes type i contributions from the heavier right - handed neutrinos , plus any additional ( type ii ) contributions from heavier particles .
examples for realizations of the neutrino mass operator can be found e.g. in @xcite . at @xmath9 , the most minimal extension of the sm or the mssm
would then be to introduce the effective neutrino mass operator @xmath71 plus one right - handed neutrino @xmath72 with mass @xmath9 and yukawa couplings @xmath73 to the lepton doublets @xmath74 , defined as @xmath75 in the superpotential of the mssm and @xmath76 in lagrangian of the sm .
the situation is illustrated in figure [ fig : gap ] .
graphical illustration of the effective description of neutrino masses at the leptogenesis scale . ] the contributions to the decay asymmetries in the effective approach stem from the interference of the diagram for the tree - level decay with the loop diagrams containing the effective operator . in the sm
, the interference with diagram ( a ) of figure [ fig : effleptogenesis_susy ] gives the simple result @xmath77}{(y_\nu^\dagger y_\nu)_{11 } } = : - \frac{3}{16 \pi } \frac{m_{\mathrm{r}1}}{v_\mathrm{u}^2 } \<m^{\mathrm{bau}}\ > \;,\end{aligned}\ ] ] where we have introduced the effective mass for leptogenesis @xmath78 , @xmath79}{(y_\nu^\dagger y_\nu)_{11 } } \ ; .\end{aligned}\ ] ] for the supersymmetric case , diagram ( a ) and diagram ( b ) contribute to @xmath80 and we obtain : @xmath81}{(y_\nu^\dagger y_\nu)_{11 } } = : - \frac{3}{8 \pi } \frac{m_{\mathrm{r}1}}{v_\mathrm{u}^2 } \<m^{\mathrm{bau}}\ > .\end{aligned}\ ] ] explicit calculation furthermore yields @xmath82 the results are independent of the details of the realization of the neutrino mass operator @xmath71 .
note that , since the diagrams where the lightest right - handed neutrino runs in the loop do not contribute to leptogenesis , we have written @xmath83 instead of @xmath84 in the formulae ( [ eq : effdecayass_sm ] ) - ( [ eq : effdecayass_mssm ] ) .
having done this , the decay asymmetries are then seen to be directly related to the neutrino mass matrix @xmath85 . for neutrino masses via the type i see - saw mechanism , they are in agreement with the known results @xcite ( equation ( [ eq : effdecayass_mintypeiiscenarios ] ) ) in the limit @xmath86 . in the limit @xmath10 , the results obtained in the effective approach are also in agreement with our full theory calculation in the minimal type ii scenarios with su(2)@xmath4-triplets in equation ( [ eq : decayastriplet ] ) .
in particular , we confirm the correction by the factor @xmath47 compared to the previous result of @xcite in the sm .
the generated b - l asymmetry , i.e. the ratio of the number density over the entropy density @xmath87 , can be written as [ eq : leptas]@xmath88 .
\end{aligned}\ ] ] @xmath80 ( and @xmath89 ) are the decay asymmetries of the lightest right - handed neutrino into ( s)lepton and higgs(ino ) and @xmath90 ( and @xmath91 ) are the decay asymmetries of the lightest right - handed sneutrino .
ignoring supersymmetry breaking , the right - handed neutrinos and sneutrinos have equal mass @xmath9 . @xmath92 and @xmath93 are the number densities of the neutrino and sneutrino at @xmath94 if they were in thermal equilibrium , normalized with respect to the entropy density .
they are given by @xmath95 where @xmath96 is the effective number of degrees of freedom , which amounts @xmath97 in the sm and @xmath98 in the mssm , and @xmath99 is the boltzmann constant .
equation ( [ eq : leptas ] ) also provides the definition for the efficiency factor @xmath100 for leptogenesis .
it can be computed from a set of coupled boltzmann equations ( see e.g. @xcite ) and it is subject to e.g. thermal correction @xcite and corrections from spectator processes @xcite , @xmath101 processes involving gauge bosons @xcite and from renormalization group running @xcite . in the effective approach and for thermal leptogenesis with a reheating temperature @xmath102 , which is most independent of the cosmological model and of the model for neutrino masses
, we assume that to a good approximation the efficiency factor depends only on the quantity @xmath103 @xcite , defined by @xmath104 and on the initial population of right - handed ( s)neutrinos .
this means , we neglect e.g. the contribution to washout processes from diagrams involving the additional particles which are involved in realizing the effective operator @xmath71 . for example , in type i see - saw scenarios the effects from the heavier right - handed neutrinos and their yukawa couplings can be neglected if @xmath9 is much smaller than @xmath105 gev . under this assumption , we can use the results for @xmath100 from type i see - saw models . see e.g. @xcite for figures showing @xmath106 for various initial populations of right - handed ( s)neutrinos .
a population of right - handed ( s)neutrinos required for leptogenesis can also be produced non - thermally , e.g. via the decay of the inflaton .
such scenarios depend on the specific cosmological model
. they could be very efficient , since @xmath5 and @xmath39 would be almost completely out - of - equilibrium when they decay . from the decay of the right - handed ( s)neutrinos ,
a lepton asymmetry is produced which is then transformed into a baryon asymmetry via sphaleron transitions . since the latter conserve b - l , we write the negative of the lepton asymmetry as b - l asymmetry in equation ( [ eq : leptas ] ) . the baryon asymmetry is then related to the b - l asymmetry @xmath107 via @xmath108 and with @xmath109 being the number of higgs doublets . with @xmath110 from equation ( [ eq : susydecayasymmetries ] ) in the mssm and using @xmath111 , the produced baryon asymmetry in terms of the baryon to photon ratio in the sm and in the mssm is approximately given by [ eq : bau_epsilon_eta]@xmath112 note that the sign of the produced asymmetry is a relevant quantity here .
@xmath113 has to be positive since we calculate in the convention that we consist of matter and not of anti - matter . in terms of the effective mass for leptogenesis , defined in equation ( [ eq : effmassforleptogenesis ] ) ,
@xmath114 is required for obtaining @xmath115 .
in the effective approach , we can calculate a model - independent upper bound for the decay asymmetries @xmath117 and @xmath118 from the requirement of successful thermal leptogenesis . for obtaining this bound , it is useful to choose a basis where @xmath119 ( and @xmath120 ) are diagonal .
note that the decay asymmetry is independent of the basis for @xmath121 . in this basis , we can write @xmath122 with real and positive @xmath123 and @xmath124 . for the effective mass for leptogenesis @xmath125 , defined in equation ( [ eq : effmassforleptogenesis ] ) ,
we obtain @xmath126}{(y_\nu^\dagger y_\nu)_{11 } } \le \frac{\widetilde y^2_{11 } m_1 + \widetilde y^2_{21 } m_2 + \widetilde y^2_{31 } m_3 } { \widetilde y^2_{11 } + \widetilde y^2_{21 } + \widetilde y^2_{31 } } \nonumber \\ & \le & m^\nu_\mathrm{max } \ ; , \end{aligned}\ ] ] with @xmath127 being the largest neutrino mass at the energy scale @xmath9 . using equation ( [ eq : effdecayass_sm ] ) and equation ( [ eq : effdecayass_mssm ] ) ,
this leads to the upper bounds [ eq : boundseffdecayass]@xmath128 for the decay asymmetries .
thus , the upper bound increases with increasing mass scale of the light neutrinos . note that compared to the low energy value , the neutrino masses at the scale @xmath9 are enlarged by renormalization group ( rg ) effects by @xmath129 in the mssm and @xmath130 in the sm , which raises the bounds on the decay asymmetries by the same values .
more accurate results can be found e.g. in figure 4 of @xcite . using equation ( [ eq : bau_epsilon_eta ] ) , for a given efficiency factor @xmath100 and using an upper bound for @xmath131
, it can be transformed into a lower type ii bound for the mass of the lightest right - handed neutrino : [ eq : boundsmr1]@xmath132 the bound on @xmath9 is lower for a larger neutrino mass scale .
it is shown in figure [ fig : lowerboundsmr1 ] as a function of the neutrino mass scale , i.e. of the mass of the lightest neutrino .
@xmath133{lowerboundsmr1_typeii.eps}}}$ ] the situation in the type ii framework is very different to the type i see - saw case : e.g. , for a normal mass ordering , the type ii bound on the decay asymmetry is proportional to @xmath134 , whereas the type i bound is proportional to @xmath135 .
in addition , thermal type i leptogenesis gets less efficient for a larger neutrino mass scale since @xmath136 , with @xmath137 .
together with an improved bound @xcite on the type i decay asymmetry , this strongly increases the type i bound on @xmath9 @xcite for increasing @xmath138 and finally leads to an upper bound for the absolute mass scale of the light neutrinos of @xmath139 ev @xcite . in the type ii scenario where @xmath103 is in general independent of @xmath131 ,
there is no bound on the neutrino mass scale from the requirement of successful leptogenesis . a neutrino mass scale @xmath140 ev on the contrary allows for a mass of the lightest right - handed neutrino of about an order of magnitude below the bound in type i models , which might help thermal leptogenesis with respect to the gravitino problem in supersymmetric models .
note that in non - thermal leptogenesis scenarios , a lower bound on @xmath9 does in general not lead to a conflict with respect to the gravitino problem .
in this work , we have investigated type ii leptogenesis via the decay of the lightest ( s)neutrinos . in the mssm with
the type ii contribution realized via an additional su(2)@xmath141-triplet superfield , we have calculated the decay asymmetries for the lightest right - handed neutrino @xmath5 and its superpartner @xmath39 . in the sm
, we have recalculated the decay asymmetry @xmath8 and corrected the previous result . we have developed an effective approach , assuming a gap between the mass @xmath9 of the lightest ( s)neutrino and the masses of the remaining particles involved in generating the neutrino masses .
we have calculated the effective decay asymmetries in the sm and in the mssm .
leptogenesis in this framework is independent of the specific realization of the neutrino mass operator .
the total decay asymmetry @xmath142 is proportional to the complete neutrino mass matrix @xmath143 .
we have derived a general upper bound ( equation ( [ eq : boundseffdecayass ] ) ) on the total decay asymmetry and found that it increases with the neutrino mass scale , in sharp contrast to the type i case which leads to an upper bound of about 0.1 ev on the neutrino mass scale .
it leads to a lower bound ( equation ( [ eq : boundsmr1 ] ) and figure [ fig : lowerboundsmr1 ] ) on the mass of the lightest right - handed neutrino significantly below the type i bound for partially degenerate neutrinos .
increasing the neutrino mass scale allows a lower reheat temperature , making thermal type ii leptogenesis more consistent with the gravitino constraints in supersymmetric models .
we would like to thank michael ratz for useful discussions . we acknowledge support from the pparc grant ppa / g / o/2002/00468 .
t. yanagida in _ proc . of the workshop on unified theory and baryon number of the universe _ ,
kek , japan , 1979 ; s.l .
glashow , cargese lectures ( 1979 ) ; m. gell - mann , p. ramond and r. slansky in sanibel talk , calt-68 - 709 , feb 1979 , and in _ supergravity _ ( north holland , amsterdam 1979 ) ; r. n. mohapatra and g. senjanovic , phys . rev . lett . * 44 * ( 1980 ) 912 .
m. magg and c. wetterich , phys .
b * 94 * ( 1980 ) 61 ; g. lazarides , q. shafi and c. wetterich , nucl .
b * 181 * ( 1981 ) 287 ; r. n. mohapatra and g. senjanovi , phys . rev .
* d23 * ( 1981 ) , 165 ; c. wetterich , nucl . phys .
* b187 * ( 1981 ) , 343 .
s. antusch , j. kersten , m. lindner , and m. ratz , _ running neutrino masses , mixings and cp phases : analytical results and phenomenological consequences _ ,
b674 * ( 2003 ) , 401433 , ` hep - ph/0305273 ` . | we discuss the effect of the neutrino mass scale on baryogenesis via the out - of - equilibrium decay of the lightest right - handed ( s)neutrinos in type ii see - saw models .
we calculate the type ii contributions to the decay asymmetries for minimal scenarios based on the standard model and on the minimal supersymmetric standard model , where the additional direct mass term for the neutrinos arises from a higgs triplet vacuum expectation value .
the result in the supersymmetric case is new and we correct the previous result in the scenario based on the standard model .
we confirm and generalize our results by calculating the decay asymmetries in an effective approach , which is independent of the realization of the type ii contribution .
we derive a general upper bound on the decay asymmetry in type ii see - saw models and find that it increases with the neutrino mass scale , in sharp contrast to the type i case which leads to an upper bound of about 0.1 ev on the neutrino mass scale .
we find a lower bound on the mass of the lightest right - handed neutrino , significantly below the corresponding type i bound for partially degenerate neutrinos .
this lower bound decreases with increasing neutrino mass scale , making leptogenesis more consistent with the gravitino constraints in supersymmetric models .
shep/0415 type ii leptogenesis and the neutrino mass scale + s. antusch , s. f. king + _ department of physics and astronomy , university of southampton , + southampton , so17 1bj , u.k . |
active galactic nuclei ( agn ) are sources of luminous x - ray emission , and at cosmological distances agn are routinely selected from deep x - ray ( @xmath4kev ) exposures ( @xcite ) .
highly obscured ( @xmath5 ) agn are thought to be a major contributor to the hard x - ray background ( @xcite ) .
however , the majority of them might not be detected in these x - ray surveys because a large fraction of their soft x - ray , uv , and optical emission is absorbed , and presumably reradiated in the infrared ( ir ) .
numerous attempts have been made to detect this population of heavily obscured agn , many of which have focused on the mir emission where the obscured radiation is expected to be reemitted ( e.g. , @xcite ) , or on combinations of mir and multiwavelength data ( e.g. , @xcite ) . in the mid - ir , agn can often be distinguished by their characteristic power - law emission ( e.g.,@xcite ) .
this emission is not necessarily due to a single source , but can arise from the combination of non - thermal nuclear emission and thermal emission from various nuclear dust components ( @xcite ) .
we summarize here the properties of galaxies showing the characteristic power - law behavior expected for agn in the _ spitzer _
@xmath0 m bands detected in the _ chandra _ deep field north and south ( cdf - n and cdf - n ) studied in @xcite and @xcite , respectively .
we also discuss results from other ir - based methods to detect high-@xmath6 obscured agn .
we use @xmath7 , @xmath8 , and @xmath9 .
we chose two cosmological fields with deep x - ray coverage ( cdf - n : 2ms and cdf - s : 1ms , see @xcite for details ) to look for obscured agn .
we selected as power - law galaxies sources that were detected in each of the four irac ( @xmath10 and @xmath11 m ) bands and whose irac spectra could be fitted as @xmath12 , where @xmath13 is the spectral index .
we used a minimum @xmath14 criterion to select galaxies whose irac seds followed a power law with spectral index @xmath15 .
the choice for the spectral index was based on the empirical spectral energy distributions ( seds ) of bright qso selected in the optical , x - rays and near - ir ( e.g. , @xcite ) and seyfert galaxies ( e.g. , @xcite ) .
there are two slight differences in the two catalogs of ir power - law galaxies . in
the cdf - s @xcite started the selection of the power - law candidates from the @xmath16 m catalog of @xcite , without any further requirements on the flux limits of the irac catalogs .
@xcite in the cdf - n instead imposed a strict s / n@xmath17 flux density cut in each of the irac bands but did not require a @xmath16 m detection , although virtually all of the power - law galaxies in the cdf - n were also detected at @xmath16 m down to @xmath18jy ( equivalent to the 80% completeness limit of the cdf - s , see @xcite ) . to minimize the chances of selecting non - active galaxies we constructed optical - mir seds ( see 4 ) , and compared them with theoretical and observational templates of star - forming galaxies .
we rejected any source selected via the power - law criteria whose sed resembled a star - forming galaxy .
the final samples included 92 and 62 galaxies in the cdf - s and cdf - n , respectively ( see @xcite and @xcite for details ) .
a small ( @xmath19 , depending on the field ) fraction of the power - law galaxies , typically the optically bright x - ray sources ( see next section ) , have spectroscopic redshifts ( e.g. , @xcite in cdf - s ) .
we supplemented the available spectroscopic redshifts with photometric ones estimated with an improved version of the method described by @xcite .
we find that the ir power - law galaxies tend to lie at significantly higher redshifts ( @xmath20 ) than the x - ray sources ( median @xmath21 , see @xcite for a review ) in both fields .
@xcite and @xcite defined agn selection criteria based on _
spitzer_/irac color - color diagrams .
the lacy et al criterion is based on sdss qsos , and therefore excludes agn in which the host galaxy dominates the mir ( see e.g. , @xcite ) , as well as agn obscured in the mid - ir .
our power - law galaxies fall along a straight line well within the lacy et al diagram , although they do not cover completely the available color space ( see fig . 1 ) .
the stern et al criteria are based on the observed properties of spectroscopically classified agn , and provide a closer match to our power - law technique . while the color - color selected samples comprise a higher fraction of the low x - ray luminosity agn than does the power - law selected sample , the color criteria select more sources not detected in x - rays , due at least in part to a higher degree of contamination from ulirgs dominated by star formation ( fig . 1 and @xcite ) .
recently @xcite obtained follow - up optical spectroscopy of objects selected according to @xcite in the _ spitzer _ first look survey and swire xmm - lss fields .
their sample is flux - limited at @xmath16 m , although their objects are much brighter ( @xmath22mjy , median of 5mjy , and median @xmath23-band magnitudes @xmath24mag ) than our power - law galaxies ( @xmath25mjy , and @xmath26mag for those detected by combo-17 see next section ) , and on average their agn are closer @xmath27 compared with @xmath28 for our power - law galaxies .
the location of this sample on the @xcite irac color - color diagram ( figure 7 in @xcite ) is almost identical to the positions of the power - law galaxies shown in fig 1 ( right panel ) .
their selection technique has proven to be very effective at selecting agn as their follow - up spectroscopy shows that approximately 90% have agn signatures with one - third of them showing broad - line regions , thus an obscured - to - unobscured ratio of 2:1 ( see 5 for the power - law galaxies ) .
all these properties seem to indicate that these color - color selected galaxies represent the brightest end of the power - law galaxies .
in both cdf - n and cdf - s we found that approximately 50% of the ir power - law galaxies were detected in at least one _ chandra _ band using the x - ray catalogs of @xcite . in the cdf - s we stacked the x - ray data of a few individually undetected galaxies ( at off - axis angles of @xmath29 ) and found a significant detection in the hard - band ( @xmath30 ) and a tentative detection in the soft band ( @xmath31 ) .
for @xmath32 these would correspond to observed soft and hard luminosities of @xmath33 and @xmath34 , respectively .
this is consistent with obscured agn . since the x - ray exposure of the cdf - n is twice that of cdf - s we searched for faint x - ray emission at the positions of the power - law galaxies not in the @xcite catalog .
we found that the x - ray detection rate increases to 77% at the @xmath1 level .
the power - law galaxies make up a significant fraction of the high x - ray luminosity sample , as our selection criteria require the agn to be energetically dominant .
the lower luminosity x - ray sources not identified as power - law galaxies tend to be dominated by the @xmath35 m stellar bump in the optical to near - ir bands ( see also @xcite ) .
a large fraction of ir luminous high-@xmath6 galaxies have been found to host agn ( e.g. , scuba galaxies @xcite ) , and ir luminous galaxies at @xmath36 ( e.g. , @xcite ) .
we measured the total ir ( @xmath2 m ) luminosity of the cdf - s power - law galaxies from the rest - frame @xmath37 m luminosity .
although our procedure to compute ir luminosities is similar to that of @xcite , we took special care to use @xmath37 m to ir luminosity ratios specific to the class of galaxies in study .
in particular , galaxies whose seds resemble those of optical qsos ( see 4 and fig . 3 ) show @xmath37 m to ir luminosity ratios significantly lower than the typical values of cool ulirgs and some warm ulirgs ( e.g. , mrk 231 ) .
all the ir power - law galaxies are highly luminous .
about 30% are in the hyperluminous class ( @xmath38 ) , 41% are ulirgs ( @xmath39 ) , and all but one of the rest are lirgs ( @xmath40 ) . at the lower ir luminosity end ( @xmath41 )
a large fraction are detected in x - rays and tend to have seds similar to those of optical qsos ( see next section ) .
about one - quarter of the cdf - s ir power - law galaxies are optically faint i.e. , were not detected by combo-17 ( @xcite ) down to a limit of @xmath42mag .
moreover , the fraction of power - law galaxies not detected in x - rays increases toward fainter @xmath23-band magnitudes ( see fig .
2 ) , an indication of their obscured nature as also revealed by their seds ( see also 5 ) . a number of works ( e.g. , @xcite and references therein ) have demonstrated that x - ray to optical flux ratios can be useful for distinguishing between agn and star - forming galaxies for sources detected in deep x - ray exposures . fig .
2 ( left panel ) shows that the majority of the galaxies ( or their upper limits ) in the cdf - s are consistent with being agn or transition objects based on the x - ray vs. @xmath23-band diagram ( similar results are found for the cdf - n galaxies ) .
the location of the ir power - law galaxies on this diagram ( see @xcite ) indicates x - ray luminosities ( or upper limits ) above @xmath43 , as also shown by the right panel of fig . 2 .
the rest - frame hard x - ray/@xmath37 m ratios of the ir power - law galaxies are similar to those of local warm ulirgs ( i.e. , those containing an agn ) and qso . included in the sample of predominantly star - forming galaxies selected at @xmath16 m by @xcite .
, width=445 ]
using available multiwavelength datasets ( see @xcite for references ) of the two cosmological fields we constructed seds for our sample .
the spectroscopic classifications ( broad vs. narrow lines ) tend to agree with two distinct types of seds . about 40% of the cdf - s ir
power - law galaxies detected in x - rays are classified as blagn and have seds ( fig .
3 ) similar to the average radio - quiet qso sed of @xcite , that is , with an optical to mid - ir continuum almost flat in @xmath44 with a uv bump .
the remaining x - ray sources with narrow lines ( nlagn ) have seds similar to the blagn but their uv and optical continua are much steeper ( obscured ) , and some of them resemble local warm ulirgs .
the majority of the power - law galaxies not detected in x - rays ( fig .
4 ) have steep seds similar to the nlagn or ulirg class as they tend to be optically fainter ( see 3 ) and possibly more obscured ( see 5 ) than the x - ray sources .
in contrast with our power - law galaxies , massive galaxies from the virmos vlt deep survey ( vvds , @xcite ) at @xmath45 with @xmath16 m detections from the sample of @xcite are predominantly star - forming galaxies with a prominent stellar bump at @xmath35 m due to an evolved ( red giants and supergiants ) stellar population . moreover , the seds of power - law galaxies are also significantly different from those of the majority of optically - dull agn in the cdf - s which show seds dominated by stellar light originating in the host galaxy ( see @xcite ) .
4 shows the great observatories origins deep survey ( goods ) acs ( @xcite ) observed optical morphologies of optically - bright power - law galaxies in the cdf - s .
the power - law galaxies spectroscopically classified as blagn ( and sed type ) display bright nuclear point sources suggesting that the optical light is dominated by the agn , which is consistent with the fact that our power - law criteria selects the most x - ray luminous agn ( see also @xcite ) .
the power - law galaxies without broad lines ( those classified as hex or lex by @xcite ) have disky or irregular morphologies .
a morphological characterization of the ir power - law galaxies not detected in x - rays is difficult as only about half of them are detected in the goods / acs images and they are faint . as can be seen for a few examples in fig .
4 they have irregular , knotty , and/or interacting morphologies , and do not appear to contain bright point sources .
in the distant universe the x - ray background and luminosity synthesis models predict global obscured ( @xmath46 ) to unobscured ratios of 3:1 to 4:1 ( e.g. , @xcite ) , significantly higher than the observed ratios of spectroscopically identified x - ray sources in deep fields ( e.g. , @xcite ) including those detected in the mid - ir ( e.g. , @xcite ) .
@xcite estimated the intrinsic column densities of each of the x - ray well detected and weakly x - ray emitting power - law galaxies in the cdf - n .
we also included a few cdf - s ir power - law galaxies ( @xcite ) for which @xcite estimated the x - ray column densities .
the column density distribution of the power - law galaxies ( fig .
7 ) is significantly different from that of optically bright x - ray agn with spectroscopic classifications ( see also @xcite ) , including the optically - dull agn which are believed to suffer strong obscuration ( @xcite ) . from fig .
7 it is clear that the weakly detected ir power - law galaxies are consistent with being obscured ( @xmath47 ) but not compton - thick ( @xmath48 ) . if all the x - ray non - detected power - law galaxies
are obscured , the maximum obscured ratio is 4:1 ( for the cdf - n power - law galaxies ) .
this @xmath49-based obscured fraction of power - law galaxies agrees well with the ratio of blagn ( unobscured ) sed vs. nlagn ( obscured ) seds found in the cdf - s .
we can finally estimate the ratio of obscured to unobscured mid - ir detected agn in the cdf - s .
the unobscured agn are all those x - ray sources ( detected in the hard band to make sure they are agn ) with @xmath50 , whereas in the obscured category we include all obscured x - ray sources with @xmath51 , and all the obscured ir power - law galaxies .
we find an observed ratio of obscured to unobscured agn of 2:1 in the cdf - s .
comparing with the predictions of @xcite for the @xmath16 m detected agn number density we find that our sample of power - law galaxies only accounts for approximately 20% of all the mid - ir emitting obscured agn in the cdf - n .
this fraction can be as high as @xmath52 in the cdf - s as a result of larger sample of power - law galaxies there ( we did not impose irac high s / n detections , see 2 and @xcite for details ) . the remainder should have seds dominated by or strongly affected by the host galaxy or red power - law seds that fall below the irac detection limit .
this is not surprising as our power - law criteria require the agn to be energetically dominant in the near to mid - ir .
the selection of ir - bright optically - faint galaxies has been suggested as another method for identifying obscured agn ( @xcite ) , although the selection criteria ( @xmath53 and @xmath54mjy ) were set so that irs follow - up spectroscopy could be obtained .
these criteria select mostly high-@xmath6 ( median @xmath55 ) galaxies , with only a small fraction showing the characteristic aromatic feature emission of star formating galaxies .
the majority have irs spectra similar to local agn - dominated ulirgs with either a featureless power - law rest - frame mid - ir continuum or deep silicate features at @xmath56 m .
in addition , their seds lack a strong @xmath35 m stellar bump , and are similar to those of ir power - law galaxies .
their properties are consistent with being optically obscured agn - powered ulirgs with @xmath57 ( see @xcite for details ) .
only a few galaxies in the cdf - n and cdf - s fall within the @xcite flux density cuts .
we can instead compare with the @xmath16m/@xmath11 m vs. @xmath16m/@xmath23-band diagram criterion proposed by @xcite to select obscured agn .
the location of our cdf - n power - law galaxies on this diagram can be seen in fig .
we also show the comparison x - ray sources ( from @xcite ) and other irac sources in the field that do not meet the power - law criteria .
we find that all these samples cover a large range in colors , but the power - law galaxies comprise a significant fraction ( @xmath58 ) of the highly optically reddened members of the comparison x - ray sample .
this suggests that the power - law selection is capable of detecting both optically obscured and unobscured agn , and that a large fraction of the ir - bright / optically - faint sources in the comparison sample have power - law seds in the near and mid - ir ( fig .
radio emission is another good way to select agn as it is unaffected by dust absorption .
@xcite in the cdf - n used a radio to mid - ir ratio to select galaxies that are too bright in radio to be star - forming galaxies .
they found that @xmath59% of their radio - loud agn are not detected in x - rays suggesting strong obscuration .
@xcite looked for a population of radio intermediate and radio quiet agn by selecting @xmath16 m sources ( @xmath60mjy ) with radio emission and imposed a flux density cut at @xmath61 m to filter out type-1 and radio - loud qsos .
they found a population of qsos at @xmath62 ( median @xmath32 , the epoch of qso maximum activity ) with a ratio of obscured to unobscured of ( @xmath63):1 , and postulated that this population of obscured agn may be responsible for most of the black hole growth in the young universe .
m over @xmath23-band flux ratio.,width=453 ]
deep x - ray cosmological surveys are efficient at detecting agn at high-@xmath6 but they only account for a @xmath64 1:1 ratio of obscured - to - unobscured agn , whereas synthesis models of the x - ray background require ratios of between 3:1 and 4:1 .
thus there is a significant population of obscured ( @xmath65 ) agn being missed by current deep x - ray ( @xmath4kev ) observations .
we describe searches for this population .
our selection criterion is based on the characteristic ir power - law emission shown by local qsos . by selecting galaxies with power - law emission in the _ spitzer_/irac bands ( @xmath66 m )
we avoid high-@xmath6 galaxies whose seds are dominated by stellar emission and star formation peaking at @xmath35 m .
only @xmath67% of ir power - law galaxies are detected in deep ( @xmath68ms ) x - ray exposures .
this fraction increases to 75% if we include weakly detected x - ray sources at the @xmath1 level in the field with the deepest x - ray exposure ( cdf - n ) .
the optical ( faint ) , ir ( mostly ulirgs and hyper luminous ir galaxies ) , and x - ray properties , the x - ray column densities @xmath49 ( moderately obscured , but not compton - thick ) and redshift ( @xmath20 ) distributions , and sed shapes of a large fraction ( up to @xmath69 ) of ir power - law galaxies are significantly different from bright x - ray selected agn .
this may indicate that a large fraction of ir power - law galaxies are good candidates to host obscured agn , and could account for a ratio of 2:1 of obscured - to - unobscured agn at high-@xmath6 .
other mid - ir based criteria ( e.g. , @xcite ) are also finding populations of obscured agn
. there might be a significant overlap between populations of bright mid - ir agn selected with all these methods , and thus a complete census of the _ entire _ obscured agn population is needed to determine whether we can account for the obscured fraction of the x - ray background .
a. a. h. acknowledges support from the spanish plan nacional del espacio under grant esp2005 - 01480 and p. g. p .-
from the spanish programa nacional de astronoma y astrofsica under grant aya 2004 - 01676 and the comunidad de madrid astrid i+d project .
support for this work was also provided by nasa through contract no .
960785 and 1256790 issued by jpl / caltech .
alexander et al : aj * 126 * , 539 ( 2003 ) d.m .
alexander et al : nature * 434 * , 738 ( 2005 ) a. alonso - herrero et al : apjs * 154 * , 155 ( 2004 ) a. alonso - herrero et al : apj * 640 * , 167 ( 2006 ) a. barger et al : aj * 126 * , 632 ( 2003 ) p. barmby et al : apj * 642 * , 126 ( 2006 ) f.e .
bauer et al : aj * 128 * , 2048 ( 2004 ) w.n .
brandt & g. hasinger : ara&a * 43 * , 827 ( 2005 ) a. comastri et al : a&a * 296 * , 1 ( 1995 ) j.l .
donley et al : apj * 634 * , 169 ( 2005 ) j.l .
donley et al : apj submitted ( 2006 ) r.a .
edelson et al : apj * 321 * , 233 ( 1987 ) m. elvis et al : apjs * 95 * , 1 ( 1994 ) a. franceschini et al : aj * 129 * , 2074 ( 2005 ) m. giavalisco et al : apj * 600 * , l93 ( 2004 ) r. gilli : advances in space research * 34 * , 2470 ( 2004 ) j. houck et al : apj * 622 * , l105 ( 2005 ) z. ivezi et al : aj * 124 * , 2364 ( 2002 ) j.k .
kuraszkiewicz et al : apj * 590 * , 128 ( 2003 ) m. lacy et al : apjs * 154 * , 166 ( 2004 ) m. lacy et al : apj in press , astro - ph/0609594 ( 2006 ) o. le fvre et al : a&a * 428 * , 1043 ( 2004 ) a. martnez - sansigre et al : nature * 436 * , 666 ( 2005 ) r.f .
mushotzky et al : nature * 404 * , 459 ( 2000 ) g. neugebauer et al : apj * 230 * , 79 ( 1979 ) c. papovich et al : apjs * 154 * , 70 ( 2004 ) p.g .
prez - gonzlez et al : apj * 630 * , 82 ( 2005 ) m.c .
polletta et al : apj * 642 * , 673 ( 2006 ) g.h .
rieke & m.j .
lebofsky : apj * 250 * , 87 ( 1981 ) j.r .
rigby et al : apjs * 154 * , 160 ( 2004 ) j.r .
rigby et al : apj * 645 * , 115 ( 2006 ) d. stern et al : apj * 631 * , 136 ( 2005 ) g.d .
szokoly et al : apjs * 155 * , 271 ( 2004 ) p. tozzi et al : a&a * 451 * , 457 ( 2006 ) e. treister et al : apj * 616*,123 ( 2004 ) m.j .
ward et al : apj * 315 * , 74 ( 1987 ) d. weedman et al : apj * 651 * , 101 ( 2006 ) c. wolf et al : a&a * 421 * , 193 ( 2004 ) m.a .
worsley et al : mnras * 352 * , l28 ( 2004 ) m.a .
worsley et al : mnras * 357 * , 1281 ( 2005 ) l. yan et al : apjs * 154 * , 60 ( 2004 ) l. yan et al : apj * 628 * , 604 ( 2005 ) | we summarize multiwavelength properties of a sample of galaxies in the _ chandra _ deep field north ( cdf - n ) and south ( cdf - s ) whose spectral energy distributions ( seds ) exhibit the characteristic power - law behavior expected for agn in the _ spitzer_/irac
@xmath0 m bands . agn selected this way tend to comprise the majority of high x - ray luminosity agn , whereas agn selected via other irac color - color criteria might contain more star - formation dominated galaxies .
approximately half of these ir power - law galaxies in the cdf - s are detected in deep ( 1ms ) _ chandra _ x - ray imaging , although in the cdf - n ( 2ms ) about 77% are detected at the @xmath1 level .
the seds and x - ray upper limits of the sources not detected in x - rays are consistent with those of obscured agn , and are significantly different from those of massive star - forming galaxies .
about 40% of ir power - law galaxies detected in x - rays have seds resembling that of an optical qso and morphologies dominated by bright point source emission .
the remaining 60% have seds whose uv and optical continuum are much steeper ( obscured ) and more extended morphologies than those detected in x - rays .
most of the ir power - law galaxies not detected in x - rays have ir ( @xmath2 m ) above @xmath3 , and x - ray ( upper limits ) to mid - ir ratios similar to those of local warm ( ie , hosting an agn ) ulirgs .
the sed shapes of power - law galaxies are consistent with the obscured fraction ( 4:1 ) as derived from the x - ray column densities , if we assume that all the sources not detected in x - rays are heavily absorbed .
ir power - law galaxies may account for between 20% and 50% of the predicted number density of mid - ir detected obscured agn .
the remaining obscured agn probably have rest - frame seds dominated by stellar emission . |
the abba collaboration is proposing to perform a measurement of a `` complete set '' of correlations in the neutron @xmath0-decay using the same apparatus , and improve the precision of the correlation coefficients @xmath1 , @xmath2 , @xmath3 , and @xmath4 by up to an order of magnitude .
geant4 is a general - purpose software package for simulation of the passage of particles through matter that provides a complete set of tools for all domains of detector simulation @xcite . in particular
, the geant4 toolkit currently provides particle tracking in non - uniform magnetic and electric fields and handles combined electromagnetic fields transparently @xcite .
the geant4 low energy electromagnetic physics group validates the low energy electromagnetic processes for electrons down to 250ev @xcite . in this report
we describe a geant4 simulation of the abba spectrometer and outline the algorithm for the extraction of the physics decay parameters .
[ ! tpb ] in the tentative design of the abba spectrometer the decay particles ( electrons and protons ) are guided by the electric and magnetic fields and interact only in sensitive detectors thus avoiding the energy losses and scatterings in apertures , grids , or windows @xcite .
the simplified geometry of the detector defines two sensitive planar silicon detectors with a @xmath5mm@xmath6 area and a 2 mm thickness .
the two si detectors are separated by 4 meters .
the coordinate system is defined with the si detectors at @xmath7 meters ( fig .
[ det ] ) .
a passive solenoid magnet is placed around the decay region , with its axis of symmetry perpendicular to the incident neutron beam , along the @xmath8 coordinate .
the 3 m long magnet with a 0.8 m radius can produce a 4 t central magnetic field that decreases to 1 t at the detector positions , thus guiding charged particles from the decay region to the si detectors .
a tubular electrode held at @xmath930kv accelerates the protons so they can be detected in a silicon detector .
the magnetic field along the @xmath8 axis of the detector solenoid is given by : @xmath10 where @xmath11 is the length of the solenoid , @xmath12 its radius , and @xmath8 the axial coordinate .
meanwhile , @xmath13 denotes the number of turns per unit length , and @xmath14 is the electrical current in the closely wound cylindrical coil .
thanks to axial symmetry , the magnetic field off - axis , outside its sources , can be represented in terms of the magnetic field strength @xmath15 along the axis : @xmath16 and @xmath17 where @xmath18 is the magnetic scalar potential and @xmath19 is the axial radius coordinate .
these fields have been programmed into the geant4 user routine .
the electrons from the neutron @xmath0-decay are generated from @xmath20mm@xmath21 central volume with the relativistic differential decay rate given by @xcite : @xmath22 } \vert m\vert ^2,\ ] ] where @xmath23 and @xmath24 ( @xmath25 and @xmath26 ) are the electron ( neutrino ) energy and momentum , @xmath27 is the neutron mass , @xmath28 is the fermi constant , @xmath29 is the cabbibo - kobayashi - maskawa matrix element , @xmath30 , and @xmath31 is the fermi function that describes the interaction of the electron and the recoil proton . -2.5
cm ) , while the mc column depends both on generated variables ( subscript @xmath32 ) and simulated detector response ( subscript @xmath33).,width=529 ] -1.5 cm the transition matrix element squared @xmath34 is given by + & & c_0(e_e ) ( 1 + 3 ^2 ) \ { 1 + ( 1 + _ ^(2 ) ) c_1(e_e ) _ + b ( ) .
+ & & . + ( 1 + _
^(2 ) ) [ c_2(e_e)+c_3(e_e ) _ ] . + & & .
+ [ c_4(e_e)+c_5(e_e ) _ ] _ } , where @xmath35 is the proton mass , @xmath36 is the fine structure constant , @xmath37 is a low energy constant , @xmath38 s are model - independent radiative corrections , @xmath39 is the axial coupling constant , and the correlation coefficients @xmath1 , @xmath3 , @xmath4 are incorporated into the recoil corrections @xmath40 @xcite .
a geant4 simulation of abba detector energy and timing response was performed for @xmath41 neutron @xmath0-decays .
we used the values of the correlation coefficients from ref .
@xcite ( @xmath42 , @xmath43 , @xmath44 , @xmath45 ) . for each event we recorded the neutron polarization ,
generated momenta of the final state particles and measured energy depositions and timing hits in the silicon detectors .
the separate geant4 run which included systematic effects ( energy and timing resolutions of si detectors , detector calibration uncertainties , detector response nonlinearities , magnetic field inhomogeneities , neutron polarization uncertainty , etc . )
was used to simulate the experimental data .
( `` mc data '' ) .
the flow chart of the physics analysis is summarized in fig .
the correlation coefficients and their fitted uncertainties are extracted using the standard minuit code @xcite .
we present two example results : ( i ) the extraction of the parameter @xmath1 with unpolarized neutron beam in fig .
[ sa ] , and ( ii ) the asymmetry coefficient @xmath3 for 80% polarized neutron beam in fig .
[ asy ] . at the current stage of development ,
a geant4 simulation limited to @xmath41 neutron decay events and @xmath41 mc data events , runs 24 cpu hours on a 1ghz linux computer . given the limited event statistics , analysis of mc data results in the coefficient @xmath47 and the coefficient @xmath48 , where statistical and systematic uncertainties are combined .
the code will be made faster by using the adiabatic invariants for charged particle tracking in the electromagnetic field , which will markedly improve the uncertainties of our method . | the abba collaboration proposes to conduct a comprehensive program of precise measurements of neutron @xmath0-decay coefficients @xmath1 ( the correlation between the neutrino momentum and the decay electron momentum ) , @xmath2 ( the electron energy spectral distortion term ) , @xmath3 ( the correlation between the neutron spin and the decay electron momentum ) , and @xmath4 ( the correlation between the neutron spin and the decay neutrino momentum ) at a cold neutron beam facility .
we have used a geant4-based code to simulate the propagation of decay electrons and protons in the electromagnetic spectrometer and study the energy and timing response of a pair of silicon detectors .
we used these results to examine systematic effects and find the uncertainties with which the physics parameters @xmath1 , @xmath2 , @xmath3 , and @xmath4 can be extracted from an over - determined experimental data set .
key words : neutron beta decay asymmetry parameters , detector monte carlo simulation , geant4 .
* monte carlo study of the abba experiment : + detector response and physics analysis * + e. frle , for the abba collaboration + department of physics , university of virginia , + charlottesville , va 22904 - 4714 , usa + |
the analysis reported here was done in the years 20023 in collaboration with two senior members of the bes collaboration .
the work was funded by the royal society and queen mary college via an official agreement with the chinese academy of sciences and the bes collaboration .
this agreement guaranteed access to bes 2 data for the purpose of publications .
since then , the bes management has refused to appoint internal referees , hence blocking a collaboration publication and comments from the collaboration .
the data illuminate the spectroscopy of @xmath19 mesons and should be in the public domain .
there is a responsibility to publish work which has been supported from public funds with approximately @xmath24 .
the wording of the article is close to a document submitted late in 2008 to the bes management , except for some clarification of wording .
there have been many earlier studies of @xmath11 in @xmath2 radiative decays , @xmath25 annihilation and @xmath26 interactions .
extensive references are to be found in the listings of the particle data group ( pdg ) @xcite .
their review concludes that two nearby @xmath27 resonances are required in the mass range 14001500 mev .
the lower one is at 1405 mev , decaying to @xmath20 and @xmath9 , where @xmath28 stands for the @xmath29 s - wave .
the upper state is at 1475 mev , decaying dominantly to @xmath30 .
however , almost all of these analyses are based on fitting the strong @xmath30 channel with a breit - wigner resonance of constant width .
the threshold for this channel is at 1390 mev , although smeared out to some extent by the 50 mev full - width of the @xmath31 .
the @xmath32 channel with @xmath3 is produced in a p wave . as a result ,
the intensity of the @xmath32 channel increases with kaon momentum @xmath33 as @xmath34 .
this rapid @xmath5-dependence enhances the upper side of the @xmath32 peak strongly .
it also gives rise to a strong dispersive correction in the breit - wigner denominator , as described in subsection 4.3 .
only one previous analysis includes the first feature and none contains the second .
they are therefore prejudiced towards two separate resonances . as well as the possible @xmath12 and @xmath13
, there is evidence from four experiments on @xmath35 for @xmath36 @xcite , @xcite , @xcite and @xcite . in these data ,
the possibile exchange of both natural and unnatural parity mesons requires a rank 2 analysis , with partial interference between the two alternatives , hence incomplete interferences between @xmath17 and @xmath36 .
the possibility of no interference between @xmath17 and @xmath36 leaves some room for uncertainty in conclusions .
there is no definite evidence for @xmath36 in @xmath37 decays and only weak evidence in @xmath25 annihilation @xcite .
there are therefore three controversial candidates for two radial excitations of @xmath38 and @xmath39 .
some people have argued that @xmath36 has been confused with @xmath17 @xcite .
others argue from the strong production of @xmath12 and @xmath13 in @xmath2 radiative decays that one of them is the @xmath40 glueball @xcite .
however , lattice gauge calculations predict the glueball above 2 gev @xcite
. one must remember too that the @xmath38 and @xmath39 are well established to have compositions @xmath41 where @xmath42 and @xmath43 if there is no glueball component @xcite . in @xmath2
radiative decays , the @xmath39 is produced with an intensity a factor 5 larger than the @xmath44 .
similar mixing for radial excitations could explain surprisingly strong production of @xmath12 and/or @xmath13 in @xmath2 decays and @xmath25 annihilation .
there are two likely scenarios .
the first is that the @xmath36 does not exist and @xmath12 may be assigned dominantly as @xmath45 and @xmath13 as @xmath46 .
the second scenario is that @xmath36 does exist and @xmath47 and @xmath13 are really one resonance .
these two alternatives are fitted to data presented here .
the @xmath5-dependence of the breit - wigner amplitudes is treated fully for the first time .
the higher mass range for @xmath20 is also of considerable interest .
bes ii data provide evidence for a 7 standard deviation signal in @xmath48 for a narrow @xmath49 with @xmath50 mev @xcite .
it is important to know if it appears in @xmath20 and if so with what branching fraction .
section 2 presents technical details of the selection of data for the channels @xmath23 , @xmath51 and @xmath52 and discusses backgrounds .
the @xmath51 channel has not been examined before .
it is useful in establishing the magnitude of possible @xmath53 contributions to @xmath54 .
sections 3 and 4 present features of the data and the formulae needed to treat the @xmath5-dependence of breit - wigner amplitudes .
section 5 gives details of the fit to the mass range up to 1600 mev .
this is the centre - piece of the paper .
section 6 considers the high mass range above 1600 mev and estimates the magnitude possible for @xmath55 ; the low value observed is inconsistent with the @xmath49 being dominantly @xmath45 .
section 7 returns to the interpretation of @xmath11 and @xmath18 .
section 8 summarises conclusions .
the data fitted here were taken with the besii detector .
full details of the detector and its upgrade are reported by bai et al .
@xcite , @xcite .
it has cylindrical symmetry around the intersecting @xmath56 beams .
its essential features are ( i ) a main drift chamber ( mdc ) for the measurement of charged particles , ( ii ) time of flight detectors with a @xmath57 of 180 ps , and ( iii ) a 12 radiation length barrel shower counter comprised of gas proportional tubes interleaved with lead sheets . the mdc measures @xmath58 .
together with the time of flight detectors , it separates @xmath59 and @xmath60 up to @xmath61 mev / c .
outside the @xmath62 detectors is a magnet providing a field of 0.4 t .
this magnet is instrumented with muon detectors , but for present work they serve simply to reject @xmath63 pairs .
the main drift chamber provides full coverage of charged particles for lab angles @xmath64 with @xmath65 .
these events have been selected by demanding exactly three photons and two charged tracks with balancing charges .
both charged particles must be positively identified by time - of - flight and/or @xmath58 as pions ( for @xmath66 ) or as kaons ( for @xmath67 ) .
the vertex must lie within a cylinder of 1.5 cm radius , 30 cm long , centred on the intersecting @xmath56 beams .
two photons must have an invariant mass within 70 mev of the @xmath44 .
events are rejected if any two photons produce a mass within 40 mev of the @xmath68 .
photons may also originate from interaction of charged tracks in the detector .
any photons lying within a @xmath69 cone around a charged track are discarded .
candidate events have been subjected to 4-constraint kinematic fits to @xmath70 and @xmath71 fits to @xmath72 ( or correspondingly @xmath73 and @xmath74 ) ; they are accepted only if the @xmath75 probability of each fit is @xmath76 .
any events where two photons are consistent with the final state @xmath77 or @xmath78 are rejected if @xmath79 .
they are also rejected if the measured particles are consistent with the loss of one photon and a fit to @xmath80 or @xmath81 with @xmath79 .
[ htb ] -14 mm -6 mm the main background arises from the loss of one photon from the final state @xmath82 , which is dominated by @xmath83 .
this background has been studied using the standard simbes monte carlo simulation of the detector .
after the loss of one soft photon , it leads to a strong peak at high @xmath20 masses 2300 - 2800 mev .
this background may be reconstructed from data by assuming one missing photon in the kinematic fit . for events fitted this way , fig .
1(a ) displays the @xmath84 mass distribution .
there is a clear peak due to @xmath85 .
if @xmath86 lies in the mass range 670870 mev , one finds the @xmath87 mass distribution of fig .
it shows clear @xmath88 and @xmath89 peaks . in order to suppress these backgrounds further
, the missing momentum is calculated from the measured @xmath66 of each candidate event .
assuming the missing particle to be a @xmath68 , any event with @xmath90 in the mass range 670 - 870 mev is rejected .
however , a minor problem is that events lying in the charge zero peak of the @xmath91 can not be eliminated without rejecting required @xmath92 events , where @xmath57 is a shortand for the @xmath93 s - wave .
1(c ) shows the mass distribution for the @xmath94 pair closest to the @xmath44 for all combinations fitting @xmath70 .
1(e ) shows the corresponding plot for candidate @xmath67 events .
1(d ) shows the @xmath20 mass distribution from the final data sample .
peaks are visible at 1285 and 1415 mev , followed by a sharp rise at 1490 mev to a shoulder at @xmath95 mev .
there is a further peak at @xmath96 mev , then a rapid fall to 2 gev . above 2 gev
, there is a large contamination from the background explained above .
the monte carlo simulation of the @xmath97 background is fitted to events in the mass region 2300 - 2800 mev , hence establishing its magnitude .
it agrees with the pdg branching fraction of @xmath98 within its quoted error of @xmath99 .
the physics analysis will be restricted to the mass range below 2 gev .
since @xmath83 is the dominant background by a factor 10 , the background at low @xmath66 masses is taken from the monte carlo simulation of this channel .
integrated up to an @xmath20 mass of 2 gev , this background is 18.2% . its distribution with @xmath20 mass is illustrated below in fig .
7(d ) . in the mass range below 1600 mev
it is small and featureless .
although there is a background of misidentified events , direct feedthrough from @xmath97 produces at most 33 events in @xmath100 , mostly above 1600 mev .
this small @xmath89 background surviving in data will be pointed out later . at the end of the data selection , there are 18,751 events up to an @xmath20 mass of 2000 mev of which an estimated 3350 are background .
the average efficiency for data reconstruction is @xmath101 .
the branching fraction is @xmath102 = ( 7.5 \pm 0.8(syst ) ) \times 10^{-3}\ ] ] up to a mass of 2 gev , after correction for all charge combinations and background .
the error is mostly systematic .
it covers the estimated background and the uncertainty in partial wave amplitudes due to loss of events through the @xmath103 holes in the detector around the beam entrance and exit .
the pdg quotes an average branching fraction @xmath104
. however , this refers to a crystal ball measurement @xcite claiming a peak in @xmath20 at 1700 mev , substantially different to what is observed here . for this final state ,
the main background arises from @xmath106 after the loss of one photon .
the contamination is estimated to be @xmath107 from a study of side - bins either side of the @xmath44 peak , together with a monte carlo estimate of the shape of the background .
the reconstruction efficiency is 7.6% and varies little with @xmath108 mass .
1(f ) shows the mass distribution of events fitting as @xmath109 .
the number of events is too small to allow a meaningful physics analysis .
however , the data are useful in establishing an upper limit on the number of events due to the final state @xmath10 , @xmath110 .
the essential point is that the number of signal events is a factor 100 smaller than for @xmath72 .
after background subtraction , there are 22 events up to an @xmath108 mass of 1.8 gev and 125 up to 2 gev .
the latter corresponds to a branching fraction @xmath111 = ( 1.4 \pm 0.4 ) \times 10^{-4}\ ] ] up to 2 gev .
errors are again systematic , from the acceptance .
these events have been selected initially with at least one photon and with 4 charged tracks having balancing charges .
one and only one charged kaon must be identified by time - of - flight or @xmath58 ; at least two charged tracks must be positively identified as pions .
events are then subjected to a @xmath112 kinematic fit , requiring a @xmath113 probability @xmath76 .
the mass of one @xmath114 pair must lie within 25 mev of the @xmath115 .
this pair must be consistent with a vertex within 6 cm of the second vertex defined by the other two charged particles .
2 shows the mass distribution for the reconstructed @xmath115 before the kinematic fit .
there is a clean peak with only a @xmath116 background , which can be evaluated accurately .
events are rejected if the spectator pion and either of the other two particles ( fitted as pions ) have an invariant mass within 25 mev of the @xmath115 .
[ htb ] -16 mm -6 mm events are rejected if there is a better fit to final states @xmath117 or @xmath118 without an additional photon . in order to reject the second process more strongly ,
events are discarded if the missing mass attributed to neutral particles is within 50 mev of the @xmath68 mass .
the kinematic fit is made assuming that the @xmath115 decays at the primary vertex ; the effect of this assumption on the selection of events is simulated by a monte carlo study .
there is a small loss of events , but no significant bias to kinematics .
alternative selections have been made fitting two separate vertices .
these lead to a larger loss of events because of the difficulties of reconciling two vertices with the kinematics ; within statistics , there is no change to physics conclusions . at the end of the data selection , there are 5638 events including an estimated background of 1068 .
the background is substantially smaller over the reduced mass range up to 1600 mev .
the average efficiency for data reconstruction is @xmath119 .
the resulting branching ratio , corrected for charge combinations , is @xmath120 = [ 2.2 \pm 0.3(syst)]\times 10^{-3},\ ] ] up to a mass of 1800 mev .
the error is almost entirely systematic .
a @xmath121 error arises from extrapolating the partial wave analysis over the holes of the detector , @xmath122 uncertainty in the effects of cuts , and @xmath123 in the background ; these errors are added in quadrature .
the pdg quotes an average of @xmath124 , though this is derived from four values spanning the range @xmath125 .
[ htb ] -16 mm -6 mm the monte carlo used here to simulate the bes detector was the standard simbes package released in february 2002 , in its version as of march 2003 . during discussions with the bes management
, they have claimed that alterations to the calibration of the time - of - flight system after march 2003 could affect the selection of events .
this point is discussed in detail in the accompanying paper on @xmath126 data @xcite . in that analysis , no systematic discrepancy is observed between @xmath127 and time - of - flight , and the same is true for present data ; no contamination of @xmath128 events by @xmath129 is observed above the @xmath130 level . in the present data ,
there is again no observed confusion between @xmath114 and @xmath131 . for @xmath72 data , contamination from @xmath74 is obviously @xmath132 because of the small observed branching fraction for that channel .
there is no evidence for contamination from @xmath106 , which is kinematically well separated from @xmath133 ; such contamination is estimated to be @xmath132 . for @xmath134 data ,
the selection of the @xmath115 limits possible background to @xmath116 .
the observed background under the @xmath115 is consistent with that originating from @xmath135 ; it rises with @xmath136 mass as expected for that source .
3 shows the @xmath136 mass distribution in @xmath137 .
a small @xmath138 peak is visible at 1285 mev . then there is a strong but asymmetric peak at @xmath139 mev on a slowly rising background .
there may be a small structure causing a drop at 1530 mev , though this could be a large statistical fluctuation .
a similar shoulder is visible at 1510 mev in mark iii data of bai et al .
several alternative prescriptions for fitting this shoulder will be described . the raw mass distribution for @xmath66 is shown in fig .
1(d ) and fits to it will be shown in detail below in figs . 7 and 11 .
there is a narrow peak at @xmath140 mev with a full width of 60 mev .
there is also a sharp rise at 1500 mev .
it is vital to realise from the outset that these structures are superimposed on a large and broad @xmath40 signal extending over the whole mass range from 1200 to 2000 mev and interfering coherently , see fig .
7(b ) below . without the inclusion of this broad component , fits fail completely to describe the @xmath20 mass projection . in particular
, the broad component plays a major role in fitting the shoulder at 1550 mev .
the @xmath141 data contain a similar broad peak at 1600 mev in @xmath142 with @xmath3 @xcite .
it is natural that some @xmath142 pairs will de - excite to @xmath143 or @xmath6 with @xmath144 in the decay .
it is also possible that the @xmath142 peak at 1600 mev is resonant .
[ htb ] -8 mm -6 mm dalitz plots for @xmath145 are shown for three mass ranges in fig .
the upper 3 panels are for data and the second three from the fit to @xmath11 reported below .
the lower set does not contain the statistical fluctuations of the upper set and the plotting procedure makes it appear that there are some differences between data and fit .
however , a check on all bins shows no evidence for systematic discrepancies .
the main feature in all three cases is a peak due to two crossing @xmath88 bands .
[ htb ] -6 mm -6 mm dalitz plots for @xmath136 are shown in fig 5 for three mass ranges . here
, the main feature is due to two crossing @xmath146 bands .
there is a strong enhancement near the intersection of these bands in figs .
5(a ) and ( b ) .
it is much reduced in fig .
this is due to the zero in the @xmath29 amplitude near the middle of the @xmath31 band , arising from the angular dependence of the partial wave amplitude in the decay .
the aim of the analysis is to fit the data in terms of conventional resonances , plus the broad interfering @xmath3 component observed both here and in the @xmath147 data
. it would be good if the spin composition of @xmath20 and @xmath136 could be determined in narrow slices of mass , and this has been tried .
the main conclusion from that attempt is that @xmath40 is strongly dominant for both sets of data . in @xmath136 data , contributions from @xmath148
are identified clearly in the narrow mass ranges of @xmath17 and @xmath16 .
there is also definite evidence for a small @xmath149 contribution continuing above @xmath16 , up to @xmath95 mev .
however , it is sufficiently small that its line - shape can not be determined with any precision from slice fits .
the contribution from @xmath150 is small .
there are however two difficulties with trying to analyse the data in slices .
the first is that the only available information is a differential cross - section .
this does not allow independent determinations of amplitude and phase .
a related difficulty is that there are complex interferences between @xmath6 , @xmath7 and @xmath10 in @xmath20 data and likewise in @xmath136 data between @xmath32 , @xmath9 and @xmath6 , @xmath151 . in the absence of any strong interferometer in the data with known quantum numbers ,
the slice analysis can not sort out these interferences .
this leaves no choice but to relate magnitudes and phases through breit - wigner amplitudes fitted directly to the data .
partial waves which are tried are @xmath3 , ( orbital angular momentum @xmath152 in the production process @xmath153 ) , @xmath149 ( @xmath154 ) , @xmath155 ( @xmath152 ) and @xmath156 @xmath157 ) . of these
, @xmath158 is negligible .
decay channels which are considered are @xmath6 , @xmath30 , @xmath9 , @xmath159 , @xmath160 , @xmath10 and @xmath7 .
there is an immediate question how @xmath57 and @xmath28 should be parametrised . in @xmath161 and @xmath29 elastic scattering , there are adler zeros which suppress the amplitudes strongly near threshold .
this is well understood in terms of chiral symmetry breaking , which creates the adler zeros . however , in many sets of data , it is established that @xmath57 and @xmath28 appear as low mass peaks .
the origin of this effect is straightforward .
if the elastic amplitude is written as @xmath162 , the denominator must be universal .
however , the numerator can be quite different in elastic scattering and production reactions .
bes ii data for @xmath163 @xcite may be fitted accurately with @xmath164 a constant .
this is also true for bes data on @xmath165 , where there is again a low mass peak in the @xmath29 s - wave due to the @xmath28 pole @xcite .
the present work uses the @xmath28 amplitude fitted in ref .
@xcite to lass data for elastic scattering @xcite , bes ii , and e791 data .
a problem is that it is not yet fully understood why or how the adler zero disappears from the numerator of the amplitude in some production processes . in principle
, there could be contributions with a numerator @xmath166 where @xmath167 and @xmath168 are complex constants
. however , this unfortunately leads to large destructive interferences between the two terms .
this is a well known symptom of over - fitting the data , so final fits must choose between @xmath57 and @xmath28 poles or alternatively amplitudes the same as for elastic scattering .
empirically , the present @xmath169 data give a considerably better fit ( by 46 in log likelihood ) with the @xmath28 pole amplitude , i.e. a constant numerator @xmath164 .
[ log likelihood is defined here so that it changes by 0.5 for a one standard deviation change in the fit ] .
the @xmath23 data show less discrimination , but still prefer a fit with the @xmath57 pole , i.e. no adler zero in @xmath164 .
this fit is better than one using the @xmath161 elastic scattering amplitude by 16 in log likelihood .
fits to present data are almost identical using the formula for the @xmath57 of zou and bugg @xcite or a recent formula fitted to the @xmath57 pole @xcite .
the first formula is simpler and is used here .
both @xmath57 and @xmath28 are fitted using a constant for @xmath164 .
the fit to data is made using 200k monte carlo events for @xmath170 and @xmath171k for @xmath1 to simulate detector acceptance .
these monte carlo events are generated with the standard simbes package .
all figures show experimental results uncorrected for acceptance .
these corrections are included automatically into the log likelihood fit .
the acceptance varies only slowly across figures .
a technical detail is that partial wave amplitudes are written in terms of relativisic tensors ; explicit formulae are given in @xcite .
in @xmath37 radiative decays , an important point is that the dominant production process is expected to be via intermediate @xmath172 . because of the high mass of these quarks and the interaction of the photon at a point ,
the production process is expected to be almost pointlike .
data on @xmath173 in the accompanying paper verify this @xcite .
they contain a dominant @xmath3 signal in @xmath142 with a dependence on @xmath174 mass close to the expected @xmath175 for @xmath40 production with @xmath152 ; here @xmath176 is the photon lab momentum .
the consequence for present data is that production should proceed only through the lowest available orbital angular momentum @xmath177 .
there are well identified @xmath17 and @xmath16 signals in @xmath169 final states . in both cases , any amplitudes with @xmath178
are negligible , producing improvements in log likelihood @xmath179 . for production of @xmath40 states
, there is likewise no evidence for @xmath180 . with this simplification ,
there are distinctive differences between final states with @xmath3 , @xmath149 , @xmath155 and @xmath156
. fits to the 1450 mev @xmath136 peak immediately reveal the need for both @xmath148 and @xmath40 amplitudes .
the @xmath16 amplitude is conspicuous from its @xmath181 dependence , where @xmath182 is the decay angle in the rest frame of the @xmath31 to @xmath60 and @xmath59 .
a free fit gives @xmath183 mev , compared with the value @xmath184 mev of the pdg .
if the @xmath16 is removed from the final fit , log likelihood is worse by @xmath185 for two less fitting parameters .
statistically this is nearly a 14 standard deviation effect .
there is also a small but definite @xmath17 contribution , decaying to @xmath9 .
if the @xmath36 is included in the fit , its contribution is small and will be discussed in section 7 .
earlier analyses have freqently fitted the @xmath88 freely to @xmath15 data .
the @xmath88 contribution lies along the upper right - hand edge of dalitz plots of fig .
5 . for low @xmath136 masses
, it interferes strongly with the two @xmath30 components , see figs .
5(a ) and ( b ) .
it can easily be confused with the intersection between the two @xmath146 bands .
the @xmath6 contribution in @xmath20 data does not suffer from this problem .
fits have therefore been made constraining the decay branching ratio of @xmath88 between @xmath186 and @xmath187 to a value consistent with the flatt ' e form for @xmath88 @xcite .
integrated over the @xmath88 peak up to 1.3 gev , the branching fraction to @xmath187 is @xmath188 of that in @xmath186 .
this constraint results in small but well defined contributions of @xmath88 to @xmath136 data .
production phases are fitted freely .
it is essential to include into the breit - wigner amplitude the strong @xmath5-dependence of the phase space for p - wave production of @xmath32 .
this phase spaces rises as @xmath34 , where @xmath33 is the kaon momentum in the @xmath14 rest frame . as a reminder , the intensity for a two - channel resonance is @xmath189 where @xmath190 is the total width .
the factor @xmath191 in the numerator arises from the phase space of the final @xmath192 state .
the same @xmath5-dependence _ must _ be included into the denominator , otherwise an inconsistency with unitarity arises between numerator and denominator .
the correct form for the breit - wigner amplitude for decay to channel @xmath193 ( for example @xmath32 ) is @xmath194 where @xmath195 is the coupling constant for each channel @xmath196 and @xmath197 is its phase space .
a form factor @xmath198 is included , and ( where required ) a centrifugal barrier factor @xmath168 .
details will be given shortly .
the evaluation of phase space is done numerically , using eq .
( 40 ) of ref .
@xcite which will be repeated here for completeness . as an example , the 3-body phase space for @xmath32 is given by latexmath:[\[\rho _
{ k^*\bar k}(s ) = \int ^{(\sqrt { s } - m_k)^2 } _ { 4m^2_k } \frac { ds_1}{\pi } \frac { 4|k||k_1|}{\sqrt { s s_1 } } amplitude for the @xmath31 .
here @xmath5 refers to the resonance @xmath200 and @xmath33 to the momentum of the @xmath31 or @xmath60 in the rest frame of @xmath200 ; @xmath201 and @xmath202 refer to the @xmath31 and the momenta of decay @xmath60 and @xmath59 in its rest frame . for cases where there is angular momentum in the decay ,
the centrifugal barrier factor @xmath203 appears in the numerator .
[ htb ] -6 mm -6 mm the term @xmath204 is a dispersive correction to the real part of the amplitude , required to make the amplitude fully analytic : @xmath205 it includes a subtraction on resonance making the integral strongly convergent .
the resulting @xmath206 for @xmath11 is illustrated in fig .
6 . in the 1440 mev mass region ,
the term @xmath204 actually varies more strongly with @xmath5 than the term @xmath207 in the breit - wigner denominator , leading to major deviations of phases from those of a breit - wigner resonance of constant width .
this is one of the essential new features of the present analysis . in fitting @xmath20 data , it is unavoidable to introduce a broad @xmath19 signal
. it needs to accomodate peaks observed in @xmath208 , @xmath209 , @xmath210 and @xmath211 near their thresholds .
a natural explanation is that it arises from @xmath212 , where @xmath213 are gluons , followed by coupling of gluons to vector mesons by colour neutralisation .
an essential feature of the @xmath214 data is the requirement for a peak at @xmath215 mev , naturally associated with @xmath216_{l=1 } \to [ \eta \sigma]_{l=1}$ ] and @xmath217_{l=1}$ ] .
the strategy adopted here is to examine how far one can get by fitting this broad component by a single broad resonance @xmath200 in the production process ; exactly the same formula as ( 7 ) is used to fit @xmath218 data in the accompanying paper @xcite .
the mass @xmath219 optimises empirically at @xmath220 gev .
this is far removed from @xmath11 so the only significant phase variation arises from ( important ) dispersive effects associated with the opening of each threshold .
this broad background will be described as ` the broad @xmath40 ' . although eq .
( 7 ) may look complicated , it is straightforward to programme and gives an excellent fit to present data and simultaneously to data on @xmath173 .
each of the channels @xmath7 , @xmath6 , @xmath221 , @xmath32 , @xmath9 , @xmath142 , @xmath209 , @xmath222 and @xmath211 requires just one parameter @xmath223 constrained ( iteratively ) to reproduce observed branching fractions of these channels .
this constrains the @xmath5-dependence of the broad @xmath40 amplitude closely .
the first five channels are essential to fit @xmath224 and @xmath136 data .
the inclusion of the @xmath210 threshold examines whether or not the drop in the @xmath20 data above 1850 mev can be explained by the @xmath210 threshold .
the channels @xmath209 and @xmath211 have little effect but are included for completeness .
further details concern the form factor @xmath198 and centrifugal barriers @xmath168 in eq .
the form factor for @xmath225 , @xmath226 is the product of a weak form factor @xmath227 for the first step and a factor @xmath228 for decays of @xmath200 through channel @xmath193 . both are taken as gaussians : @xmath229 where @xmath60 is the production momentum of @xmath200 in the @xmath37 rest frame and @xmath33 is the decay momentum in the rest frame of x ( for example , the momentum of @xmath44 in the channel @xmath7 )
. values of @xmath230 and @xmath231 are taken from the work on @xmath173 : @xmath232 gev@xmath233 and @xmath234 gev@xmath233 , corresponding to rms radii @xmath235 fm , @xmath236 fm .
the production form factor is very weak and has little effect here , but is included for consistency with @xmath218 data .
the factor @xmath168 is likewise the product of the @xmath152 centrifugal barrier for @xmath237 and ( where necessary ) a centrifugal barrier for the decay , for example an @xmath238 barrier for the decay of @xmath40 to @xmath32 . as a reminder , for @xmath238 , @xmath239 the same value of @xmath240
is used in the centrifugal barrier as in the form factor , though this may be an approximation .
two types of fit will be discussed .
the first ( fit a in tables ) tries to fit both @xmath20 and @xmath241 data with a single @xmath11 resonance .
the second ( fit b ) uses two separate resonances like @xmath12 and @xmath13 , though their masses and widths need re - optimising .
[ htb ] -6 mm -6 mm fig .
7 shows details of the fit to @xmath242 data with a single resonance which optimises at a mass of @xmath243 mev .
panel ( a ) shows the overall fit to the mass distribution ; the lower histogram shows the coherent sum of @xmath11 and the broad @xmath40 .
panel ( b ) shows the full @xmath40 intensity , also the broad component from eqs .
( 7)(9 ) and thirdly @xmath244 .
panel ( c ) shows the small peak due to @xmath17 .
panel ( d ) shows background as the upper histogram and the small @xmath155 contribution required above 1600 mev from @xmath245 and @xmath246 .
a detail is that it is necessary to check that the @xmath10 contribution to @xmath20 is consistent with data on @xmath247 . the dashed curve on fig .
1(f ) shows the signal predicted for @xmath10 with @xmath248 .
it is consistent with data near 1600 mev .
the sharp rise of the @xmath20 mass distribution at 1495 mev is not fitted perfectly in fig .
7(a ) : the data would prefer a rise 10 mev lower than the fit .
the rapid rise requires a strong contribution from @xmath10 .
the fit which is shown uses the bes parametrisation of @xmath53 @xcite .
a reparametrisation is desirable including the threshold cusp in the real part of the amplitude at the @xmath187 threshold .
this has been tried , but does not cure the problem : it simply sharpens the @xmath53 peak near the @xmath187 threshold .
[ htb ] -10 mm -6 mm fig .
8 shows the fit to @xmath169 data .
the peak sits on a small experimental background shown in ( a ) and a slowly rising physics background , clearly visible above experimental background .
this physics background has @xmath3 and is parametrised by the broad @xmath40 contribution decaying to @xmath32 , @xmath9 and @xmath249 , @xmath250 .
the coupling constants of the first two are fitted freely ; the last component is derived from the @xmath88 fitted to @xmath20 data using the branching ratio between @xmath186 and @xmath187 in ref .
@xcite .
the @xmath136 peak of fig . 8 is decidedly asymmetric .
the main source of the asymmetry can immediately be traced to the @xmath5-dependence of the @xmath32 decay channel whose contribution is shown in fig .
the lower histogram in fig .
8(a ) shows the contribution from @xmath148 .
all fits require some @xmath149 contribution above @xmath16 ; this extra contribution is parametrised in the present fit by @xmath18 with mass and width quoted by the pdg . changing the parameters to values of aston et al .
@xcite changes log likelihood only by 3.6 .
attempts have been made to fit the small shoulder in fig .
8 at 1530 mev in several ways .
this mass is close to the opening of the @xmath251 channel at 1539 mev .
it raises the possibility of a drop in the line - shape associated with the opening of this channel ; this is a unitarity effect - the opening of the @xmath108 channel robs @xmath136 of some signal .
however , this is immediately excluded by the small branching ratio of @xmath108 compared to @xmath20 .
a more subtle possibility is that it may be associated with the @xmath252 channel , which also peaks at 1539 mev , since @xmath53 itself peaks at the @xmath187 threshold .
the fit to @xmath20 data requires a strong contribution from @xmath10 .
however , it is spread out quite significantly according to phase space for the @xmath10 final state , as shown below in fig .
13(a ) ; that is too broad to explain a narrow shoulder at 1530 mev .
note that there is independent evidence for this @xmath10 contribution to @xmath3 @xcite .
the @xmath253_{l=1}$ ] does not account for the 1550 mev structure in @xmath20 or the 1530 mev shoulder in @xmath136 ; there is no visible @xmath89 peak at this mass in @xmath186 or @xmath187 and the angular dependence is wrong .
furthermore , the magnitude which can be fitted to @xmath254 in @xmath141 rules out the possibility of explaining the 1530 mev shoulder with decays to @xmath255_{l=1}$ ] .
[ htb ] -10 mm -6 mm one further possible explanation remains . in the accompanying paper on @xmath173
, there is evidence that the @xmath142 peak at @xmath256 mev is actually resonant at @xmath257 mev .
if this possible resonance has a weak decay to @xmath32 , the shoulder at 1530 mev can be fitted as an interference effect .
the resulting fit is shown in fig . 9 in identical format to fig .
8 ; the addition of @xmath258 with a width of 280 mev improves log likelihood by 62 .
this explanation is clearly speculative .
it is to be expected that two neighbouring resonances at 1440 and 1560 mev will provide considerably more flexibility in the fit than @xmath11 alone .
however , the magnitude of the contribution from @xmath258 is fairly small , as would be expected if its mixing with @xmath259 is small ; its intensity is @xmath260 of that from @xmath11 .
this additional resonance does not provide the 10 mev mass shift near 1500 mev needed for a perfect fit to the @xmath20 mass projection of fig .
7(a ) . a possible remedy for this detail will be discussed in section 7 .
[ htb ] -14 mm -6 mm let us now turn to the @xmath261 contribution .
10(a ) shows the poor fit without @xmath18 . the remaining component due to @xmath17 and @xmath16
is shown as the lower histogram .
the @xmath16 is parametrised including the @xmath5-dependence of the @xmath32 channel and the associated dispersive term .
the result is to extend the tail of @xmath16 to slightly higher masses , but does not substitute for @xmath18 .
the mass of @xmath16 is held fixed at the pdg value and its full width at half maximum is fixed to the width quoted by the pdg .
the magnitude fitted to @xmath16 increases in fig .
10 to 743 events compared with 488 in fig .
the reason for this is the requirement for some significant @xmath149 signal in the mass range around 1500 mev .
an alternative is to increase the width of @xmath16 quite significantly .
this possibility will be discussed in section 7 .
the full @xmath5-dependence is fitted to both states including the dispersive terms .
this requires a determination of the channels which contribute to each .
the @xmath6 and @xmath7 channels are naturally attributed to @xmath12 , as in earlier work .
attempts to include these channels into @xmath13 lead to negligible contributions .
the @xmath32 channel is naturally associated with @xmath13 , as in earlier work .
if it is added to @xmath12 , log likelihood improves only by 3.9 , which is insignificant . here ,
it is necessary to comment on earlier work of adams et al . @xcite . their fig .
4(d ) shows a low mass @xmath32 component in addition to that of @xmath13 .
however , from the present analysis it is clear that this is an artefact arising from their free fit to @xmath6 , @xmath250 .
the resulting large @xmath88 intensity in @xmath262 is inconsistent with the @xmath20 data by a factor 10 , so a constraint on the magnitude of @xmath151 is essential .
this eliminates any @xmath32 contribution from @xmath12 .
if @xmath9 is fitted freely to both @xmath12 and @xmath263 , it goes almost entirely into @xmath12 .
removing the small component from @xmath13 makes log likelihood worse by only 6.2 , which is barely significant .
furthermore , it removes a destructive interference with @xmath32 , symptomatic of instability .
the final fit includes @xmath9 only in @xmath12 .
the remaining component is @xmath10 .
it improves log likelihood of the fit by 162 , a decisive amount .
it plays a strong role in creating the sharp rise in the @xmath20 mass projection of fig .
7(a ) at 1500 mev . one might therefore naturally associate it with @xmath13 .
however , the fit contradicts this possibility
. a free fit to both @xmath12 and @xmath13 leads to negligible contribution to @xmath13 from @xmath10 .
it is easy to locate the reason .
associated with the @xmath10 channel is a large dispersive term @xmath204 in the denominator of any resonance to which is couples .
the phase variation created by this dispersive term is not compatible with @xmath264 .
the final outcome is that @xmath13 has only a @xmath32 contribution and all other channels contribute via @xmath12 .
the peak in @xmath20 data is at @xmath265 mev . however , because of the interference with the broad @xmath40 contribution , the mass which can be fitted is quite flexible , with an uncertainty of @xmath266 mev .
the fit actually optimises at 1429 mev because the phase variation associated with @xmath204 in the breit - wigner denominator favours the higher mass . for @xmath13
, there is a strong correlation between the fitted width and the optimum mass .
the strict optimum is at 1505 mev .
however , log likelihood is worse by only 4 if the pdg mass of 1476 mev is used . in order to allow comparison with published results ,
the mass is fixed at the pdg average , 1476 mev and the width is then optimised .
the fitted width at half - maximum is 78 mev , compared with the pdg average of @xmath267 mev .
the fit to the @xmath20 mass distribution using these values is shown in fig .
11 . near the 1415 mev peak
, it is somewhat different in shape to that of fig .
the fit to the rapid rise at 1500 mev is no better than in fig .
[ htb ] -6 mm -6 mm the fit to @xmath136 data is shown in fig .
the mass distribution of the peak is fitted rather better than in fig .
the reason is obvious : two resonances allow much greater flexibility .
in particular , the phase variation with @xmath5 allowed by 2 resonances has much greater freedom than the fixed relation between magnitude and phase for @xmath11 alone . with two resonances , it is possible to arrange that the mass distribution in fig .
12 rises rapidly and falls rapidly , but with something approaching a flat top .
however , there is very little difference in log likelihood between fits with one resonance or two .
the fit to @xmath20 data with two states is better by 6.5 in log likelihood and the fit to @xmath136 data is better by only 2.6 , making a total of 9.1 for the two sets of data .
this hardly warrants confidence in the need for two resonances .
[ htb ] -12 mm -6 mm table 1 shows changes in log likelihood for both types of fit when individual components are removed and magnitudes and phases of all other components are re - optimised .
the third column shows results for fit a with one resonance , @xmath11 , and the fourth shows results for fit b with two resonances .
this table also includes branching fractions for channels or combinations of channels ( where interferences are large ) .
a warning is that for @xmath11 there are strong interferences between @xmath9 and @xmath268 and these are somewhat different for fits a and b. there are likewise large interferences between @xmath9 and @xmath32 for the broad @xmath40 signal .
errors given for branching fractions cover uncertainties in interferences . because of large interferences between the broad @xmath40 and @xmath244 , branching fractions of table 1 do not add up to the total branching fraction given in eq .
the one feature which might favour a fit with two resonances rather than one is the @xmath18 contribution .
its parameters are fixed at pdg averages : @xmath269 mev , @xmath270 mev .
it improves the fit with a single @xmath11 and an @xmath18 , which is regarded by some authors as questionable , by 54.0 .
the improvement for the fit with 2 resonances is 38.9 .
so the fit without any @xmath18 is better for fit b. both fits require some @xmath18 contribution , or something resembling it .
a test has been made of the requirement for @xmath18 , as distinct from a @xmath40 contribution .
a further @xmath40 state is added to fit b using @xmath12 and @xmath13 .
it is given identical mass and width to @xmath18 .
the result is that the @xmath18 contribution dominates over that from @xmath271 , confirming a definite requirement for a @xmath149 component in this mass range . in section 7 , a possible explanation in terms of the line - shape of @xmath16 will be presented .
.changes in log likelihood when individual components are removed and all other components are re - optimised in magnitude and phase .
column 3 shows results fitting @xmath11 alone and column 4 results fitting both @xmath12 and @xmath13 .
columns 5 and 6 show branching fractions for @xmath153 , multiplied by @xmath272 .
[ cols="^,^,^,^,^,^",options="header " , ] from table 2 , the ratio @xmath274 the errors have been obtained by adding in quadrature the errors for @xmath32 and @xmath275 to get overall branching fractions for @xmath136 in the range @xmath276 , then dividing by the corresponding @xmath20 fraction . in their review , masoni et al .
@xcite present in their table 8 their estimate of the same branching ratio , or more exactly its inverse . converting their entries to the definition used here , they estimate values in the range @xmath277 from mark iii and dm2 data , @xmath278 using crystal barrel and bubble chamber data of baillon et al .
@xcite , or @xmath279 from crystal barrel and obelix data .
they use the disagreement between these sources as an argument in favour of separate @xmath12 and @xmath13 .
however , this analysis is misleading .
when one analyses data for @xmath280 annihilation , it is vital to allow for the suppression of the @xmath32 channel by the available phase space .
this was ignored by masoni et al .
the factor @xmath281 multiplying the line - shape of @xmath11 is : latexmath:[\[h(m ) = \int ^{(2m_p - \sqrt { s_1})^2}_{4m^2_{\pi } } ds_1 \ , \frac { |p_{prodn}|}{2m_p } \frac { |p_1(\sigma \to\pi \pi ) |}{\sqrt { s_1 } } |t_\sigma ( s_1)|^2 spectator @xmath161 pair , and @xmath283 is the momentum of @xmath11 or the spectator @xmath57 in the @xmath25 rest frame .
next , @xmath284 and @xmath285 are the masses of the proton and pion .
also @xmath286 is the scattering amplitude for the spectator @xmath161 s - wave pair and @xmath198 is the form factor for the production process , taken here as @xmath287 with @xmath283 in gev / c .
-6 mm -6 mm the variation of @xmath281 with the mass of @xmath11 is shown in fig . 15 .
it is not clear whether the @xmath161 spectator pair will be produced via the @xmath57 pole or via the @xmath161 elastic scattering amplitude .
the former is shown as the full curve and the latter as the dashed curve .
they are sufficiently close that one can work with the average .
the high mass tail of @xmath11 is cut off strongly in @xmath25 annihilation .
it goes to zero at 1.60 gev : the highest mass which can be produced in @xmath25 annihilation with two spectator pions .
there are two consequences .
firstly , the factor @xmath281 amplifies the lower side of @xmath11 with respect to the upper side . for the narrow @xmath7 and @xmath88 channels ,
the effect is to shift the apparent peak of @xmath11 downwards by 4.7 mev ; i.e. the breit - wigner mass is actually 4.7 mev higher than is fitted with a breit - wigner amplitude of constant width . for the @xmath288 channel ,
the effect is larger because the phase space for this channel , shown on fig .
13(b ) , increases significantly with @xmath289 .
the result is that the mass fitted to this channel is moved downwards by 9.7 mev , so the true mass fitted as a breit - wigner resonance of constant width should be increased by this amount .
the second effect is that a considerable part of the @xmath136 signal is lost in data on @xmath25 annihilation ; all of it is lost above 1.60 gev
. a check will be presented here on the ratio @xmath290 from data on @xmath25 annihilation .
attention will be drawn to many sources of uncertainty .
the check is done using data of anisovich et al .
@xcite for @xmath291 and bubble chamber data of baillon et al @xcite .
table 2 of anisovich et al .
gives branching fractions for all amplitudes fitted to their data , in particular @xmath292 and @xmath6 .
this table reveals an important point .
there is large destructive interference in those data between these two amplitudes .
this destructive interference has a strong effect on the branching fraction quoted for @xmath11 .
however , in reality there are major uncertainties in determining the interference .
there are actually four charge cominations in those data allowing considerable freedom in interferences .
the individual contribution from @xmath7 is quoted as @xmath293 , that from @xmath6 is @xmath294 and the interference between them contributes @xmath295 .
it clearly makes a large difference how this interference is taken into account .
relative magnitudes of all decay channels of a resonance should be the same in all production processes .
the procedure adopted by anisovich et al . , and used here , is to derive coupling constants from the fit to data , then evaluate the intensity of amplitudes by taking the modulus squared of individual amplitudes .
interference between decay channels of one resonance are kept , since these are a property of the resonance .
interferences with other resonances in the data are ignored because such interferences are a feature of the particular environment in which the @xmath11 is produced .
the resulting branching fraction for production of @xmath11 in @xmath20 via @xmath25 annihilation is @xmath296 in the observed charge states .
this needs to be multiplied by a factor @xmath297 to allow for other charge states , giving a result of @xmath298 for @xmath299 , @xmath300 .
the result for all charges of @xmath136 quoted by baillon et al .
is @xmath301 .
this gives a ratio @xmath302 .
masoni et al .
use the smaller branching ratio @xmath303 found by amsler et al .
@xcite in an earlier analysis of @xmath25 data .
however , that publication does not specify if any allowance is made for the effect of destructive interference between @xmath304 and @xmath6 .
the effect of @xmath281 on the integrated intensities for final states @xmath7 and @xmath6 is small because the phase space @xmath197 for these channels varies slowly with @xmath289 and there is an approximate cancellation between upper and lower sides of the resonance . for the @xmath9 channel
, there is a noticeable effect because of its fairly rapidly increasing phase space , see fig .
13(b ) . using fig
15 , the @xmath9 component in @xmath25 annihilation needs to be multiplied by a factor 1.38 and the @xmath32 contribution needs to be multiplied by @xmath305 . if we take the relative contributions of @xmath268 and @xmath9 to be in the ratio 40:35 of table 2 , the result is to increase the @xmath136 branching fraction of baillon et al by a factor 2.3 , resulting in a value @xmath306 of 1.88 .
this is still a little smaller than the value derived from table 2 , namely @xmath307 .
however , the essential conclusion is that the branching fraction between @xmath136 and @xmath20 is subject to a large systematic correction when evaluated from @xmath25 data .
the result from @xmath37 data does not suffer from these systematic problems .
note that the line - shapes fitted to present data are perturbed by the slowly varying factor @xmath175 for production with @xmath308 ; this factor and the associated form factor have been included in the fit and the arithmetic of branching ratios .
one should also note that the signal assigned by obelix to the @xmath136 channel is not visible as a peak in the data , but is obtained from the difference in the line - shape of the @xmath241 signal from a breit - wigner of constant width fitted to @xmath47 .
a further detail is that the phases of individual decay channels are affected by multiple scattering with other components of the fit .
they may differ from one production process to another . as a result , it is difficult to give a definitive result for the branching ratio between @xmath136 and @xmath20 .
the reality of the situation is that there is major flexibility in the magnitudes of contribution which can be fitted to @xmath11 in @xmath20 channels . in the data presented here ,
there is flexibility in what can be fitted to the broad interfering component .
this flexibility is apparent from the fact that fits with a single @xmath11 or separate @xmath309 and @xmath13 differ very little in log likelihood .
there is yet further flexibility if the @xmath7 channel receives additional contributions proportional to the @xmath161 elastic amplitude .
-16 mm in @xmath20 data shown in fig . 7 , the broad @xmath40 signal in fig .
7(b ) obviously fails to explain fully the @xmath20 signal above 1600 mev .
the @xmath210 threshold may contribute something to the drop above 1870 mev , but alone does not fit the high mass spectrum .
the remedy adopted here is to fit a resonance form as an expedient , although there is no phase information , hence no definite evidence for a resonance .
the optimum fitted mass and width are @xmath310 mev , @xmath311 mev .
the strongest decays are to @xmath6 , but there is also a rather strong decay to @xmath10 .
16 show mass projections from @xmath20 dalitz plots on to @xmath161 and @xmath186 mass .
the @xmath53 is clearly visible in fig .
16(g ) at 1 gev . in figs .
16(h ) , there is a a small but definite @xmath89 peak .
it does not fit naturally to decays of @xmath312 and/or @xmath313 .
it is of roughly the right magnitude to originate from @xmath83 background .
the @xmath314 signal from this background is also just visible in fig .
. on the upper side of the 1850 mev peak is a small @xmath155 component , which may be fitted with @xmath312 . when this contribution is omitted from the fit , log likelihood gets worse by 84 .
the relative intensities in @xmath6 , @xmath159 , @xmath160 and @xmath315 decays are fixed to accurate values from crystal barrel data @xcite .
production phases , attached to each amplitude in the isobar model , are fitted freely .
so there are five fitted parameters .
statistically , the significance of the @xmath155 component is 10 standard deviations , despite the signal being small .
the reason for this significance is that amplitudes have a distinctive angular dependence which is readily identified . for completeness ,
a contribution is also included for @xmath316 .
it is much smaller and barely significant . in ref .
@xcite , it is suggested that @xmath316 is a standard @xmath317 state , the isospin partner to @xmath318 ; @xmath312 is interpreted as a hybrid .
this interpretation is consistent with the observations here .
the production of a hybrid is naturally explained in terms of @xmath319 , followed by conversion of one gluon @xmath320 to @xmath317 .
the bes collaboration has presented evidence for an @xmath49 in @xmath48 with a production branching fraction @xmath321 \times 10 ^{-4}$ ] and a width of 68 mev @xcite .
there is no obvious signal in @xmath66 data of fig .
7 , although the fit to data is slightly ragged in this mass range . a free fit including it with published mass and width gives 189 @xmath221 events , 94 @xmath6 , 164 @xmath7 and 86 @xmath159 .
log likelihood improves by 58 for 8 additional fitting parameters .
there is flexibility in the fit , and the total number of events can be increased from 533 to 1182 if log likelihood is allowed to get worse by 2 for every channel , corresponding to an overall 4 standard deviation change .
this is a cautious limit taking account of the fact that the components fitting this mass range are uncertain .
the corresponding branching fraction is @xmath322 .
no statement can be made about its possible coupling to @xmath136 , since the mass range of the present data extends only to 1800 mev .
huang and zhu @xcite point out that its mass , total width , production rate and decay pattern favour its assignment as the second radial excitation of @xmath39 .
more precisely , the small branching fraction in @xmath20 favours a mixing angle between @xmath323 and @xmath324 close to that of the @xmath39 .
the @xmath325 , @xmath11 and @xmath49 would then form a regge trajectory with a spacing in @xmath326 of @xmath327 gev@xmath328 , close to that of other well establshed trajectories @xcite .
the peak observed by bes2 at 2240 mev in @xmath211 would fit naturally as the fourth member of this sequence @xcite .
however , resonant phase variation needs to be established for @xmath329 and the @xmath211 peak .
ma @xcite tries to interpret @xmath330 as a baryonium state associated with the threshold peak observed in @xmath25 in bes data @xcite .
this interpretation leads to a decay stronger to @xmath20 than @xmath48 , and is clearly ruled out by the present data .
if there is just one @xmath11 in the mass range 14001500 mev , it either has a large su(3 ) singlet @xmath317 component or a glueball component .
either would explain its strong production in @xmath2 radiative decays , where @xmath331 appear in a singlet combination .
it could likewise be produced strongly in @xmath25 annihilation .
however the @xmath14 channel is su3 octet , so the @xmath11 ( or @xmath13 ) can not be purely singlet .
it is well known that isolated @xmath259 states such as @xmath332 are produced weakly in @xmath25 annihilation .
if the @xmath13 is to be produced strongly in @xmath25 annihilation , it should have a substantial @xmath323 component , but that is inconsistent with decays purely to @xmath32 .
the @xmath12 decays much more strongly to @xmath288 and @xmath10 than to @xmath7 + @xmath6 .
this would not arise naturally if it is a pure @xmath323 state .
so the conclusion is that a substantial @xmath259 ( or gluball ) component must be present in @xmath11 or @xmath333 .
the existence of @xmath36 depends primarily on data in three papers on @xmath334 and one on @xmath335 @xcite . in this process
, there should be a clear distinction between @xmath36 which decays via the s - wave and @xmath17 which decays via the p - wave .
if production is limited to a single exchange process , these are immediately distinguishable . in @xmath25 annihilation
the situation is much more complex , since there are four combinations of @xmath20 produced in every event @xcite , making the separation between @xmath36 and @xmath17 much more difficult .
those reviews which argue against the existence of @xmath36 base their arguments on the fact that there could be more than one exchange mechanism in @xmath336 production data , leading to possible confusion between the flat angular distribution from @xmath36 and rank 2 processes involving @xmath17 .
it is unfortunate that the papers of primary importance published on @xmath36 do not give quantitative estimates of its significance .
if the data still survive , it would be valuable if the groups concerned would publish values of log likelihood or @xmath113 showing quantitatively how their fit changes when the @xmath36 is omitted .
this information would sharpen the arguments for or against @xmath36 . in the data presented here ,
most of the peak near 1285 mev certainly is due to @xmath17 ; a free fit including @xmath36 improves log likelihood only by 3.7 for four parameters fitted to magnitudes and phases of decay amplitudes to @xmath6 and @xmath7 .
seven events are fitted to @xmath36 compared with 230 to @xmath17 .
so the evidence for @xmath36 in present data is not significant .
is the @xmath11 really split into two components ?
although fit a optimises @xmath12 at 1429 mev , a persistent feature of data fitted to @xmath12 in other analyses is a mass 1405 mev in @xmath20 and 1414 mev in @xmath9 .
as shown in section 5.3 , data from @xmath25 annihilation require corrections to these masses for the phase space in the production process .
this shifts the mass fitted to @xmath20 data up by 4.7 mev and that in @xmath136 data by 9.7 mev .
weighting these with the intensities fitted to @xmath20 and @xmath9 in table 2 , the mean mass of @xmath12 becomes 1421 mev .
there is then a discrepancy in mass of @xmath337 mev with the mass in fit a with a single resonance .
this is not convincing evidence for two separate resonances near 1440 mev .
however , it must be said that one can not move the @xmath11 much below 1439 mev without a poor fit to the peak in @xmath136 . how could the difference of 18 mev be explained between @xmath20 data and @xmath187 ?
klempt @xcite has pointed out a possible explanation .
a radial excitation has a node in its wave function causing a zero in decays to @xmath249 in this general mass range .
firstly this attenuates the @xmath20 signal and may explain why it is weaker than @xmath9 and @xmath32 . secondly , such a node introduces a further @xmath5-dependence into @xmath20 .
suppose the amplitude for the @xmath249 channel is multiplied by a factor @xmath338 , where @xmath339 is the location of the zero in the amplitude .
a resonance at 1439 mev can have a peak at 1410 mev if there is a zero in the amplitude at 1.47 gev , just where there is a dip in the @xmath224 mass spectrum , fig .
however , the peak is attenuated more on its upper side than its lower side and the peak is somewhat asymmetric .
tests on the present data show that this explanation does cure the 10 mev discrepancy between @xmath20 data and the fit near 1500 mev . however , a factor 10 higher statistics are needed to test the idea precisely . it can not be tested in @xmath25 annihilation because of the four different charge combinations .
zeros would appear at different masses in @xmath7 and @xmath9 , depending on matrix elements , leading to a potentially complicated line - shape with peaks at different masses in different channels .
there is one further possible complication .
the dominant decay of @xmath11 is through @xmath136 .
there is inevitably some rescattering of @xmath187 through @xmath88 to the @xmath20 channel via what are called triangle graphs .
this was discussed in some detail in ref .
@xcite . it will have the effect of increasing the @xmath136 decay branching fraction of @xmath11 .
let us turn to @xmath244 . the general pattern of the broad components in the @xmath20 mass projection of fig .
7 is that there are two components peaking at 1890 and 1600 mev .
it is possible that the latter is resonant . in the @xmath141 data there is evidence for a resonance at 1970 mev with a width of 210 mev .
if the parameters of @xmath244 are changed to those values , log likelihood for @xmath20 data is worse by 42 .
this is a modest amount , in view of the systematic uncertainties about how to fit the @xmath20 mass range above 1600 mev and the strong @xmath210 threshold nearby .
it is therefore possible that these two signals arise from the same resonance , though this can not be proved at the moment .
consider next @xmath16 and @xmath18 . in fig .
8(a ) , one sees that the effect of @xmath18 is to produce a slight bump at 1490 mev and extend the tail of @xmath16 .
longacre @xcite has suggested that @xmath16 may be an abnormal state ( or ` molecule ' in current terminology ) with an @xmath238 pion orbiting a @xmath187 s - wave combination . in this picture , there are final state interactions from two combinations of @xmath31 and a further final - state interaction between the two s - wave kaons .
my suggestion is that these final state interactions enhance this state strongly at the lowest @xmath187 masses , but the rapid rise of the @xmath32 phase space with momentum produces a long tail extending into the mass region of what is called @xmath18 .
it must be said in favour of the data of aston et al . that they identified the @xmath340 @xmath341 state and it was subsequently confirmed by crystal barrel data @xcite .
furthermore , there is a @xmath342 peak in their data which they did not claim as a resonance , but was subsequently identified as a resonance by crystal barrel @xcite and confirmed by wa102 @xcite .
the conclusion is that the data of aston et al .
appear reliable .
the primary question is one of interpretation of their @xmath343 signal .
they did not include @xmath16 in their fit .
so it is possible that the @xmath16 has a more extended line - shape than has previously been fitted .
the main experimental results are the observation of three peaks in the @xmath20 mass spectrum : a narrow peak at 1415 mev , a shoulder at 1550 mev and a broad peak at @xmath344 mev . in @xmath136
, there is an asymmetric peak at @xmath139 mev .
the asymmetry is due to the @xmath5-dependence of the @xmath32 channel ; its dalitz plot immediately identifies a dominant @xmath32 decay .
the intensity for that decay increases rapidly from its threshold at @xmath345 mev as @xmath34 , where @xmath33 is the momentum of decay @xmath60 and @xmath31 in the @xmath14 rest frame . when analysing @xmath25 data , it is necessary to allow for the variation of phase space for production of @xmath20 and @xmath9 with mass .
this corrects the mass of @xmath12 upwards by 4.7 mev in @xmath20 and by 9.7 mev in @xmath9 . as a result ,
the mean mass difference between the possible @xmath12 and @xmath346 is no larger than @xmath347 mev .
there is a dominant broad @xmath40 amplitude in @xmath20 .
when this is included into the partial wave fit , interference with a single @xmath11 can account naturally for the observed peaks at 1415 mev in @xmath20 and at 1450 mev in @xmath136 .
there is no evidence for two separate @xmath12 and @xmath13 from the present data .
the @xmath329 is not observed in the present @xmath20 data , and an upper limit has been derived for the branching fraction for its production in @xmath37 decays .
this limit is a factor @xmath348 smaller than the @xmath329 observed in the @xmath48 channel .
the natural explanation is that @xmath329 is a radial excitation of @xmath39 and @xmath11 . as an aid to further work
, the fully annotated fortran code for the broad @xmath40 component and @xmath11 is available from the author .
this code includes numerical values of @xmath223 from the present work , but they would need to be re - optimised if further data become available .
i wish to acknowledge financial support from the royal society and major help from members of the bes collaboration in processing data and running the monte carlo simulation of acceptance and backgrounds . 99 particle data group ( pdg ) , physics letters * b667 * 1 ( 2008 ) s. fukui _
et al . _ ,
b * 267 * 293 ( 1991 ) d. alde _ et al . _
( gams collaboration ) , yad .
fis . * 60 * 458 ( 1997 ) , translated in part . at .
nuclei * 60 * 386 ( 1997 ) j.j .
( e852 collaboration ) , phys . rev . *
d62 * 012003 ( 2000 ) g.s .
( e852 collaboration ) , phys .
b516 * 264 ( 2001 ) a.v .
et al . _ , nucl .
a * 690 * 567 ( 2001 ) e. klempt and a. zaitsev , phys . rep . * 454 * 1 ( 2007 ) a. masoni , c. cicalo and g.l .
usai , j. phys .
* 32 * r293 ( 2006 ) c.j .
morningstar and m. peardon , phys .
* d60 * 034509 ( 1999 ) r. escribano , arxiv : hep - ph/0802.3909 m. ablikim _
( bes collaboration ) , phys .
lett . * 95 * 262001 ( 2005 ) j.z
( bes collaboration ) , nucl .
methods * a344 * 319 ( 1994 ) j.z .
( bes collaboration ) , nucl .
methods * a458 * 627 ( 2001 ) d. edwards _ et al . _
( crystal ball collaboration ) , phys .
51 * 859 ( 1983 ) d.v .
bugg , accompanying paper arxiv:0907.3021 z. bai _ et al . _
( mark iii collaboration ) , phys .
* 65 * 2507 ( 1990 ) m. abklikim _
( bes collaboration ) , phys .
b * 598 * 149 ( 2004 ) m. ablikim _
( bes collaboration ) , phys .
* b633 * 681 ( 2006 ) d.v .
bugg , phys .
lett . b632
* 471 * ( 2006 ) d. aston _
et al . _ ( lass collaboration ) , nucl .
b * 296 * 493 ( 1988 ) b.s .
zou and d.v .
bugg , phys .
* d48 * r3948 ( 1993 ) d.v .
bugg , j. phys .
* g34 * 151 ( 2007 ) b.s . zou and d.v .
bugg , eur .
j * a 16 * 537 ( 2003 ) d.v .
bugg , phys .
d * 78 * 074023 ( 2008 ) d.v .
bugg , a. v. sarantsev and b.s .
zou , nucl .
b * 471 * 59 ( 1996 ) m. ablikim _
( bes collaboration ) phys .
* b607 * 243 ( 2005 ) d. aston _
et al . _ ( lass collaboration ) phys
* b201 * 573 ( 1988 ) a.v .
* b472 * 168 ( 2000 ) p. baillon _
_ , nu . cim .
* 50a * 383 ( 1967 ) c. amsler _ et al . _
( crystal barrel collaboration ) , phys .
b358 * 389 ( 1995 ) a.v .
( crystal barrel collaboration ) , phys .
b * 477 * 19 ( 2000 ) t. huang and s - l .
zhu , phys . rev . *
d73 * 014023 ( 2006 ) d. v. bugg , phys . rep . * 297 * 257 ( 2004 ) m. ablikim _
et al . _ ,
( bes collaboration ) , phys .
* b662 * 330 ( 2008 ) y - l .
ma , arxiv : 0803.3764 m. ablikim _
( bes collaboration ) , phys .
lett . * 95 * 262001 ( 2005 ) e. klempt , hep - ph/0409148 and beijing 2004 , ichep 2004 , vol .
2 , p1082 r.s .
longacre , phys . rev . *
d42 * 874 ( 1990 ) a. abele _ et al .
_ ( crystal barrel collaboration ) , phys . lett . * b415 * 280 ( 1997 ) j. adomeit _
( crystal barrel collaboration ) , z. phys .
c * 71 * 227 ( 1996 ) d. barberis _ et al . _
( wa102 collaboration ) , phys .
b413 * 217 ( 1997 ) | data on @xmath0 and @xmath1 from 58 m @xmath2 hadronic interactions in the bes ii detector are analysed .
they throw new light on @xmath3 mesons throughout the mass range up to 2 gev , notably @xmath4 . for the first time
, this state is fitted including the full @xmath5-dependence of decays to @xmath6 , @xmath7 , @xmath8 , @xmath9 and @xmath10 , including the dispersive term associated with this @xmath5-dependence .
two types of fit are reported .
the first uses a single @xmath11 .
the second uses separate @xmath12 and @xmath13 .
the strong @xmath5-dependence of p - wave production of @xmath14 shifts the fitted mass of @xmath13 down to 1440 mev . as a result ,
the two types of fit give almost identical results and the conclusion from these data is that there is no significant evidence for two separate resonances . in @xmath15 , there are definite though small contributions from @xmath16 and @xmath17 .
the optimum fit requires either a small additional @xmath18 or alternatively a small perturbation to the high mass tail of @xmath16 . at higher masses
, there is a broad @xmath19 peak in @xmath20 at 1850 mev .
there is also a weak but definite signal consistent with @xmath21 . _
pacs : _
13.25.-k , 14.40.cs -15 mm 6.75 in -0.75 in 6.75 in data on @xmath22 and @xmath23 d.v .
bugg , + queen mary , university of london , london e14ns , uk |
recent results of the superkamiokande ( sk ) experiment confirmed the solar neutrino deficit at the level less than data / ssm @xmath8 @xcite
. there are well - known msw , vo , rsfp solutions to the solar neutrino problem ( snp ) that have different signatures which could be observed through day / night ( d / n ) , seasonal and 11 year periodicities of the solar neutrino flux correspondingly .
we study here new aperiodic spin - flavor conversion ( asfc ) scenario@xcite that , on the one hand , is similar with the rsfp scenario because same neutrino transition magnetic moment is assumed while , on the other hand , differs significantly from the rsfp case since the presence of random magnetic fields leads to a large amount of electron antineutrinos produced efficiently _ by a non - resonant way _ within the convective zone of the sun even for small mixing in the cascade conversions @xmath4 . for the large mixing , @xmath9 , antineutrinos
are produced through last step of such conversions for both scenarios on the way to the earth in the solar wind as well as through usual vacuum neutrino oscillations , @xmath10 , without helicity change .
non - observation of antineutrinos in the sk experiment ( @xmath11 for @xmath12 , @xcite ) puts the bound on neutrino mixing angles in the rsfp like @xmath13 .
we checked this statement from our numerical code using different regular magnetic field profiles @xcite and this bound almost coincides with the limit @xmath14 obtained earlier in @xcite .
moreover , we confirm another result in @xcite that the moderate mixing angle region @xmath15 for which @xmath5 maybe detectable is still acceptable for the rsfp solution to the snp for the lowest allowed strengths of the magnetic field , @xmath16 and for the tipical mass parameter @xmath17 @xcite .
we make here accent on much smaller mixing angles , @xmath18 , when the rsfp mechanism fails to produce @xmath5 while our asfc scenario remains very efficient .
moreover , the allowed @xmath1-region occurs quite different for the case of random magnetic fields . in all four solar neutrino experiments one
measures the integral spectrum , i.e. the number of neutrino events per day for the sk experiment , @xmath19 or the number of neutrino events in snu ( 1 snu = @xmath20 captures per atom / per sec ) in the case of gallex ( sage ) and homestake experiments , @xmath21 where @xmath22 is the integral flux of neutrinos of kind `` i '' ( @xmath23 ) assumed to be constant and uniform at a given distance @xmath24 from the center of the sun , @xmath25 is the normalized differential flux , @xmath26 ; @xmath27 are corresponding cross sections ; the thresholds @xmath28 for gallex ( sage ) and homestake are @xmath29 and @xmath30 correspondingly while for sk at present @xmath31 .
the probabilities @xmath32 , where the subscript @xmath33 equals to @xmath34 for @xmath35 , @xmath36 for @xmath5 , @xmath37 for @xmath38 and @xmath39 for @xmath40 correspondingly , satisfy the unitarity condition @xmath41 and @xmath42 corresponds to averaging over the solar interior if necessary .
there exists , however , a problem with large regular solar magnetic fields , essential for the rsfp scenario .
it is commonly accepted that magnetic fields measured at the surface of the sun are weaker than within interior of the convective zone where this field is supposed to be generated .
the mean field value over the solar disc is about of order @xmath43 and in the solar spots magnetic field strength reaches @xmath44 .
because sunspots are considered to be produced from magnetic tubes transported to the solar surface due to the boyancy , this figure can be considered as a reasonable order - of - magnitude observational estimate for the mean magnetic field strength in the region of magnetic field generation . in the solar magnetohydrodynamics
@xcite one can explain such fields in a self - consistent way if these fields are generated by dynamo mechanism at the bottom of the convective zone ( or , more specific , in the overshoot layer ) .
but its value seems to be too low for effective neutrino conversions .
the mean magnetic field is however followed by a _ small scale _ , random magnetic field .
this random magnetic field is not directly traced by sunspots or other tracers of solar activity .
this field propagates through convective zone and photosphere drastically decreasing in the strength value with an increase of the scale . according to the available understanding of solar dynamo ,
the strength of the random magnetic field inside the convective zone is larger than the mean field strength .
a direct observational estimation of the ratio between this strengthes is not available , however the ratio of order 50 100 does not seem impossible . at least , the ratio between the mean magnetic field strength and the fluctuation at the solar surface is estimated as 50 ( see e.g. @xcite ) .
this is the main reason why we consider here an analogous to the rsfp scenario , an aperiodic spin - flavour conversion ( asfc ) , based on the presence of random magnetic fields in the solar convective zone .
it turns out that the asfc is an additional probable way to describe the solar neutrino deficit in different energy regions , especially if current and future experiments will detect electron antineutrinos from the sun leading to conclusion that neutrinos are majorana particles .
the termin `` aperiodic '' simply reflects the exponential behaviour of conversion probabilities in noisy media ( cf .
@xcite , @xcite ) . as well as for the rsfp mechanism
all arguments for and against the asfc mechanism with random magnetic fields remain the same ones that have been summarized and commented by akhmedov ( see @xcite and references therein ) .
the random magnetic field is considered to be maximal somewhere at the bottom of convective zone and decaying to the solar surface . to take into account a possibility , that the solar dynamo action is possible also just below the bottom of the convective zone ( see @xcite )
, we accept , rather arbitrary , that it is distributed at the radial range @xmath45 , i.e. it has the same thickness as the convective zone , while its correlation length is @xmath46 that is close to the mesogranule size .
we suppose also that all the volume of the convective zone is covered by a net of rectangular domains where the random magnetic field strength vector is constant .
the magnetic field strength changes smoothly at the boundaries between the neighbour domains obeying the maxwell equations . since one can not expect the strong influence of small details in the random magnetic field within and near thin boundary layers between the domains this oversimplified model looks applicable . in agreement with the ssm the neutrino source is supposed to be located inside = -0.32 cm -0.58 cm the solar core with the radius of the order of @xmath47 . for the sake of simplicity
we do not consider here the spatial distribution of neutrino emissivity from a unit of a solid angle of the core image and the differences in @xmath48 for different neutrino kinds `` i '' .
different parallel trajectories directed to the earth cross different magnetic domains because the domain size @xmath49 ( in the plane which is perpendicular to neutrino trajectories ) is much less than the transversal size @xmath50 of the full set of parallel rays , @xmath51 , see fig .
the whole number of trajectories ( rays ) with statisticaly independent magnetic fields is about @xmath52 . at the stage of numerical simulation of the random magnetic field
we generate the set of random numbers with a given r.m.s .
value and for each realization of random magnetic fields along each of @xmath53 rays solve the cauchi problem for the @xmath54 master equation @xcite .
then we calculate the dynamical probabilities @xmath55 , and average them in transversal plane , i.e. obtain the mean arithmetic probabilities as functions of mixing parameters @xmath2 , @xmath56 .
we argue that being additive functions of the area of the convective zone layer ( or , the same , of the number of rays ) these probabilities with increasing area become self - averaged .
to demonstrate the properties of our model we briefly discuss how random magnetic field influences the small - mixing msw solution to snp .
it turns out that for the ssm exponential density profile , typical borone neutrino energies @xmath57 , and @xmath58 , the msw resonance occurs well below the bottom of the convective zone .
thus we can divide the neutrino propagation problem and consider two successive stages .
first , after generation in the middle of the sun , neutrinos propagate in the absence of any magnetic fields , undergo the non - adiabatic ( non - complete ) msw conversion @xmath59 and acquire certain nonzero values @xmath60 end @xmath61 , which can be treated as initial conditions at the bottom of the convective zone . for small neutrino mixing @xmath62 the @xmath63-master equation @xcite
then splits into two pairs of independent equations describing correspondingly the spin - flavor dynamics @xmath64 and @xmath65 in noisy magnetic fields .
in addition , once the msw resonance point is far away from the convective zone , one can also omit @xmath66 and @xmath67 in comparison with @xmath68 . for @xmath69 conversion
this results into a two - component schr@xmath70dinger equation @xmath71 with initial conditions @xmath72 as normalized probabilities @xmath73 and @xmath74 ( satisfying the conservation law @xmath75 ) are the only observables , it is convenient to recast eq.([(01 ) ] ) into an equivalent integral form @xmath76\times \nonumber \\ & & \times \cos2\delta(t_1 - t_2 ) -[b_x(t_1)b_y(t_2 ) - \nonumber\\ & & - b_x(t_2)b_y(t_1)]\sin2\delta(t_1 - t_2)\ } , \label{(03)}\end{aligned}\ ] ] where @xmath77 is the third component of the polarization vector @xmath78 .
let us trace the line of further derivation in a spirit of our numerical method .
dividing the interval of integration into a set of equal intervals of correlation length @xmath79 and assuming that possible correlations between @xmath80 and @xmath81 under the integral are small , @xmath82 itself varies very slowly within one correlation cell , and making use of statistical properties of random fields , ( @xmath83 ) different transversal components within one cell are independent random variables , and ( @xmath84 ) magnetic fields in different cells do not correlate , we can average eq.([(03 ) ] ) thus obtaining a finite difference analogue .
returning back to continuous version we get @xmath85~ , \label{finite}\end{aligned}\ ] ] where we retained possible slow space dependence of the r.m.s . magnetic field value . for @xmath86 and constant r.m.s .
@xmath87 we obtain the simple @xmath88-correlation result @xcite , @xcite , otherwise there remains an additional stabilizing factor @xmath89 , demonstrating to what extent the vacuum / medium neutrino oscillations within one correlation cell can suppress the spin - flavor dynamics due to the magnetic field only .
another important issue is the problem of temporal dependence of higher statistical moments of @xmath90 as @xmath91 enter the eq.([integ ] ) for the number of events one should be certain that the averaging procedure does not input large statistical errors , otherwise there will be no room for the solar neutrino puzzle itself . for @xmath92 basing on eq .
( [ ( 03 ) ] ) we evaluate the dispersion @xmath93 . here
we present only the final result , the details of derivation will be published elsewhere : @xmath94 @xmath95 where @xmath96 .
we see that relative mean square deviation of @xmath60 from its mean value tends with @xmath97 to its maximum asymptotic value @xmath98 irrespectively of the initial value @xmath99 this estimate is true , evidently , only for one neutrino ray .
averaging over @xmath100 independent rays lowers the value ( [ ( c10 ) ] ) in @xmath101 times . that is for our case of @xmath102 rays we get that maximum relative error should not exceed approximately @xmath103 thus justifying the validity of our approach . for smaller magnetic fields
the situation is always better . to conclude this section , it is neccesary to repeat that the above estimate eq .
( [ ( c10 ) ] ) indicates possible danger when treating numerically the neutrino propagation in noisy media .
indeed , usually adopted one - dimensional ( i.e. along one ray only ) approximation for the @xmath54 master equation @xcite or @xmath104 eq . (
[ ( 01 ) ] ) can suffer from large dispersion errors and one should make certain precautions when averaging these equations over the random noise _ before _ numerical simulations .
otherwise , the resulting error might be even unpredictable .
in order to find the regions in the @xmath1 , @xmath2 plane excluded from nonobservation of @xmath5 in sk it is necessary to compute and plot the isolines of the ratio @xmath105 . indeed ,
as antineutrinos were not detected in the sk experiment one should claim that antineutrino flux is smaller than the sk background @xmath106 @xcite .
deviding this inequality by the ssm borone neutrino flux with energies @xmath107 , @xmath108 we find the bound on the averaged over cross - section and spectrum , cascade transition probability @xmath109 , @xmath110 or @xmath111% . here
^ 2 $ ] is the cross - section of the capture reaction @xmath113 .
notice that the above bound is not valid for low energy region below the threshold of the antineutrino capture by protons , @xmath114 . for low magnetic fields , @xmath115 , both for regular and random ones we do not find any violation of the bound eq .
( [ percent ] ) .
however , for strong magnetic fields such forbidden parameter regions appear in different areas over @xmath56 and @xmath2 for different kind of magnetic fields ( regular and random ) . in general , the influence of random magnetic fields is more pronounced as compare to the regular ones of the same strength .
it follows that the more intensive the r.m.s .
field @xmath116 in the convective zone the more effective spin - flavour conversions lead to the production of the right - handed @xmath5 , @xmath40-antineutrinos . in the case of small mixing this
is also seen from eq .
( [ finite ] ) .
our results , however , show that there exists strong dependence not only on the r.m.s . magnetic field strength but on its correlation length also .
in particular , when for the same @xmath117 we substitute @xmath118 ( a granule size ) instead of @xmath119 ( a mesogranule size treated in numerical simulation ) the small mixing msw region is fully excluded via the sk bound eq .
( [ percent ] ) ( see fig . 14 in @xcite ) .
this is easily explained due to the maximum value of asfc seen from eq .
( [ finite ] ) ( @xmath120 ) .
vice versa , if the small - mixing msw solution is valid and a large neutrino magnetic moment exists we can extract from neutrino data important constraints on the structure of random magnetic fields deep under surface of the sun . extrapolating the tendency above either with a decrease of the correlation length @xmath121 or with an increase of the magnetic field values @xmath122 ( or for both changes )
we could also exclude lma msw region or both standard msw regions would be forbidden .
however , for such case one finds low msw with @xmath123 , @xmath124 becomes to be allowed being forbidden for larger correlation lengths ( @xmath125 ) and lower r.m.s .
values ( @xmath126 ) ( see fig . 14 in @xcite ) .
thus , we develop a model of neutrino spin - flavour conversions in the random magnetic fields of the solar convective zone supposing in consistence with modern mhd models of solar magnetic fields that random fields are naturally much higher than large - scale magnetic fields created and supported continuously from the small - scale random ones @xcite .
it follows that if neutrinos have a large transition magnetic moment their dynamics in the sun is governed by random magnetic fields that , first , lead to _ aperiodic _ and rather _ non - resonant _ neutrino spin - flavor conversions , and second , inevitably lead to production of electron antineutrinos for _ low energy _ or _ large mass difference _ region .
analogous results were obtained in the work @xcite where the averaging of the redfield equation for the @xmath127 density matrix over @xmath88-correlated random magnetic fields allowed to get simple asymptotical solutions for strong random fields and led to the same aperiodic form of the probabilities as in eq.([(c9 ) ] ) . however , for the realistic case of solar random magnetic fields with a finite correlation length the method of direct numerical simulation of the schroedinger equation given in @xcite is more appropriate .
then the averaging of numerical solutions ( not schroedinger equation ! ) over an ensemble of random field configurations ( see fig . 1 ) gives correct results for this case .
if antineutrinos @xmath5 would be found with the positive signal in the borexino experiment@xcite or , in other words , a small - mixing msw solution to snp fails , this will be a strong argument in favour of magnetic field scenario with asfc in the presence of a large neutrino transition moment , @xmath128 for the same small mixing angle .
we would like to emphasize the importance of future low - energy neutrino experiments ( borexino , hellaz ) which will be sensitive both to check the msw scenario and the @xmath5-production through asfp . as it was shown in a recent work@xcite a different slope of energy spectrum profiles for different scenarios would be a crucial test in favour of the very mechanism providing the solution to snp .
akhmedov , _ the neutrino magnetic moment and time variations of the solar neutrino flux , preprint ic/97/49_. invited talk given at the 4-th international solar neutrino conference ( heidelberg , germany , april 8 - 11 , 1997 ) .
j. b. benziger et al .
, status report of borexino project : _ the counting test facility and its results . a proposal for participation in the borexino solar neutrino experiment _
( princeton , october 1996 ) .
talk presented by t.e .
hagner at x baksan international school _ particles and cosmology _ ( april 1999 ) . | the magnetic field in the solar convective zone has a random small - scale component with the r.m.s . value substancially exceeding the strength of a regular large - scale field . for two majorana neutrino flavors @xmath0 two helicities in the presence of a neutrino transition magnetic moment and
nonzero neutrino mixing we analize the displacement of the allowed ( @xmath1 - @xmath2)-parameter region reconciled for all underground experiments with solar neutrinos in dependence on the r.m.s . magnetic field value @xmath3 . in contrary with the rsfp scenario with a regular large - scale magnetic field
, we find an effective production of electron antineutrinos in the sun even for small neutrino mixing through the cascade conversions like @xmath4 .
it was found that usual sma and lma msw parameter regions maybe forbidden while opening low msw as the allowed one from the non - observation of @xmath5 in the sk experiment if random magnetic fields have strengths @xmath6 and correlation lengths shorter than @xmath7 . |
besii @xcite running at bepc and cleoc @xcite running at cesr are the two detectors operating in the @xmath7-charm energy region .
both experiments have collected large data samples of charmonium decays including 58 m @xmath0 events , 14 m @xmath1 events , and 33 pb@xmath8 data around @xmath2 peak at besii , and 4 m @xmath1 events , and 281 pb@xmath8 @xmath2 events at cleoc . to study the continuum background in
the charmonium decays , special data samples at the center of mass energy lower than the @xmath1 mass were taken both at besii ( @xmath9 gev ) and at cleoc ( @xmath10 gev ) , the luminosities are 6.4 pb@xmath8 and 21 pb@xmath8 respectively .
these data samples are used for the study of the hadron spectroscopy , the @xmath11 decay properties and the ckm matrix , as well as the charmonium decay dynamics . in this paper , we focus on the the extensive study of the `` @xmath12 puzzle '' in @xmath0 and @xmath1 decays , the relative phase between strong and electromagnetic amplitudes of @xmath1 decays , and the non-@xmath5 decays of @xmath2 . the study of @xmath6 decays are also reviewed . it should be noted that performance of the cleoc detector is much better than the besii detector , especially in the photon detection , this makes the 4 m @xmath1 events data sample at cleoc produces results with similar precision as from 14 m @xmath1 events from besii .
it has been determined that for many two - body exclusive @xmath0 decays , like vector pseudoscalar ( vp ) , vector vector ( vv ) , and pseudoscalar pseudoscalar ( pp ) meson decays and nucleon anti - nucleon ( n@xmath13 ) decays , the relative phases between the three - gluon and the one - photon annihilation amplitudes are near @xmath14 @xcite . for @xmath1 decays , the available information about the phase is much more limited because there are fewer experimental measurements .
it has been argued that the relative phases in @xmath1 decays should be similar to those in @xmath0 decays @xcite , but the analysis of @xmath1 to vp decays in ref .
@xcite indicates this phase is likely to be around @xmath15 .
another analysis of this mode though shows the relative phase observed in @xmath0 decays could also fit these decays @xcite , but it could not rule out the @xmath15 possibility due to the big uncertainties in the experimental data .
bes measured the branching fraction of @xmath16 @xcite to be @xmath17 , together with the known branching fractions of @xmath18 and @xmath19 , two possible solutions of the phase are found , which is either @xmath20 or @xmath21 , following the procedure developed in ref .
benefits from the good detector performance , cleo measures all three pseudoscalar meson pair decay modes @xcite .
the branching fraction of @xmath16 is @xmath22 , that of @xmath19 is @xmath23 ; the signal of @xmath18 is not significant , and the upper limit of the branching fraction is determined to be @xmath24 at 90% c.l .
cleo measures the relative phase to be @xmath25 , assuming there is no interference between continuum amplitude and the resonance decay amplitudes , which is questionable since the two components of the resonance decays has a relative phase around @xmath14 , as measured by both besii and cleoc experiments .
following the same procedure developed in ref .
@xcite , cleo measurements also result in two solutions for the phase , one at around @xmath26 degrees , while another at around @xmath27 degrees , in good agreement with the bes result .
the error is large due to the small statistics of the data samples , and there is no way to determine which solution is the physical one .
from perturbative qcd ( pqcd ) , it is expected that both @xmath0 and @xmath1 decaying into light hadrons are dominated by the annihilation of @xmath28 into three gluons or one virtual photon , with a width proportional to the square of the wave function at the origin @xcite .
this yields the pqcd `` 12% rule '' , @xmath29 a large violation of this rule was first observed in decays to @xmath12 and @xmath30 by mark ii @xcite , known as _
the @xmath12 puzzle _ , where only the upper limits on the branching fractions were reported in @xmath1 decays .
since then , many two - body decay modes of the @xmath1 have been measured by the bes collaboration and recently by the cleo collaboration ; some decays obey the rule while others violate it @xcite .
the @xmath4 mode of the vector charmonia decays is essential for this study , since this is the first puzzling channel found in @xmath0 and @xmath1 decays .
the new measurements , together with the old information , show us a new picture of the charmonium decay dynamics .
besii measured the @xmath32 branching fraction using its @xmath0 and @xmath33 data samples @xcite , and babar measured the same branching fraction using @xmath0 events produced by initial state radiative ( isr ) events at @xmath34 gev @xcite . together with the measurement from mark - ii @xcite
, we get a weighted average of @xmath35 . to extract the @xmath36 branching fraction , partial wave analysis ( pwa ) is needed to consider the possible contribution from the excited @xmath37 states , the only available information on the fraction of @xmath12 in @xmath38 was got from mark - iii . using the information given in ref .
@xcite , we estimate @xmath39 , with the error from an educated guess based on the information in the paper since we have no access to the covariant matrix from the fit showed in the paper . from this number and @xmath40 got above , we estimate @xmath41 .
this is substantially larger than the world average listed by pdg @xcite , which is @xmath42 .
@xmath44 was studied both at besii @xcite and cleoc @xcite . after selecting two charged pions and two photons , clear @xmath45 signals
are observed in the two photon invariant mass spectra , the numbers of signals are found to be 229 and 196 from besii and cleoc respectively , and the branching fraction of @xmath46 is measured to be @xmath47 and @xmath48 at besii and cleoc respectively .
the two experiments give results in good agreement with each other .
to extract the branching fraction of @xmath44 , however , besii uses a pwa including the high mass @xmath37 states and the interference between them , while cleoc counts the number of events by applying a @xmath37 mass cut .
the branching fraction from besii is @xmath49 , while that from cleoc is @xmath50 , the difference is large .
although a big difference exists between the besii and cleoc results , it does mean that the @xmath44 signal exists , rather than the signal is completely missing as has been guessed before .
if we take a weighted average neglecting the difference between the two measurements , we get @xmath51 . comparing @xmath52 with @xmath53 , one gets @xmath54 the suppression compared to the 12% rule is obvious .
there are many more new measurements on @xmath1 decays for the extensive study of the 12% rule @xcite , among which the vp modes are measured as first priority .
the ratios of the branching fractions are suppressed for almost all these vp modes compared with the 12% rule ; while the pp modes , @xmath55 and @xmath56 , are enhanced compared with the 12% rule .
the multi - hadron modes and the baryon - antibaryon modes are either suppressed , or enhanced , or normal , which are very hard to be categorized simply .
the various models , developed for interpreting specific mode can hardly find solution for all these newly observed modes .
one observation is that many of the attempts to interpret the @xmath4 puzzle are based on the potential models for the charmonium which were developed more than 20 years ago , as the b - factories discovered many new charmonium states @xcite which are hard to be explained in the potential models , it may indicate even the properties of @xmath0 and @xmath1 are not as expected from the potential models .
the further understanding of the other high mass charmonium states may shed light on the understanding of the low lying ones .
since @xmath2 is above the @xmath5 threshold , it decays predominantly into charmed mesons , however , since the old measurements may indicate big contribution of @xmath2 charmless decays @xcite , and large fraction of charmless decays of @xmath2 is expected if @xmath2 is a mixture of @xmath57- and @xmath11-wave charmonium states and the mixing is responsible for the 12% rule violation in @xmath0 and @xmath1 decays @xcite , both besii and cleoc experiments tried to search for the non-@xmath5 decays of @xmath2 in both exclusive modes and inclusive measurement .
it has been pointed out that the continuum amplitude plays an important role in measuring @xmath2 decays into light hadrons @xcite .
in fact , there are two issues which need to be clarified in @xmath2 decays , that is whether @xmath2 decays into light hadrons really exist , and if it exists , how large is it .
the searches for the non-@xmath5 decays are performed by comparing the cross sections on and off the @xmath2 resonance peak . by comparing the cross sections of @xmath59 at the @xmath2 resonance peak ( @xmath60 gev ) and at a continuum energy point ( @xmath9 gev at besii and @xmath61 gev at cleoc ) below the @xmath1 peak
, both besii and cleoc found that @xmath62 at continuum is larger than that at @xmath2 resonance peak .
the average of the two experiments @xcite are @xmath63 @xmath64 the difference , after considering the form factor variation between 3.650 and 3.773 gev , is still significant , and it indicates that there is an amplitude from @xmath2 decays which interferes destructively with the continuum amplitude , and makes the cross section at the @xmath2 peak smaller than the pure contribution of continuum process . for the @xmath12 mode , besii can only give upper limit of its cross section due to the limited statistics of the data sample , the upper limit at 90% c. l. is found to be 6.0 pb @xcite at the @xmath2 peak , which is in consistent with the measurement from cleoc using a much larger data sample : @xmath65 @xcite ; while the cross section at the continuum is @xmath66 pb measured by cleoc . to extract the information on the @xmath67 branching fraction , besii developed a method based on the measured cross sections at @xmath2 resonance peak and at the continuum @xcite . by neglecting the electromagnetic decay amplitude of @xmath2 ,
there are two amplitudes which contribute to the cross section at the @xmath2 peak , the strong decay amplitude of @xmath2 and the continuum amplitude .
taking the continuum amplitude as a real number , the @xmath2 strong decay amplitude is described as one real number for the magnitude , and one phase between @xmath2 strong and electromagnetic decays to describe the relative phase between the two amplitudes . since
only two measurements are available ( at @xmath2 peak and at continuum ) , one can only extract @xmath2 decay branching fraction as a function of the relative phase .
besii measurement on the upper limit of the @xmath68 cross section at @xmath2 peak , together with the cleoc measurement of the continuum cross section restrict the physics region of the branching fraction and the relative phase as shown in fig .
[ besii_pspp ] . from the figure
, we see that the branching fraction of @xmath67 is restricted within @xmath69 and @xmath70 , and the phase is between @xmath71 and @xmath72 , at 90% c.l .
the observation of the @xmath68 signal at @xmath2 peak and the measurement of the cross section @xcite at cleoc further make the physical region in the branching ratio and relative phase plane smaller : the cleoc measurement gives a similar outer bound of the physical region as bes gives , while also indicates the central part of the physical region in fig .
[ besii_pspp ] is not allowed by physics . by using a toy monte carlo to simulate the cleoc selection criteria and the interference between the resonance and continuum amplitudes
, we found that the @xmath67 branching fraction could be either @xmath73 or @xmath74 from the cleoc measurements , if the relative phase between @xmath2 strong and electromagnetic decay amplitudes is @xmath75 as observed in @xmath0 and @xmath1 decays @xcite .
the @xmath2 decays into light hadrons were searched for in various @xmath2 decay modes , including two - bady and multi - hadron modes @xcite .
however , only the comparison between the cross sections at continuum and those at @xmath2 resonance peak are given , instead of giving the @xmath2 decay branching fractions . in current circumstances
, it is still not clear whether the @xmath2 decays into light hadrons with large branching fractions , since , as has been shown in the @xmath12 case , there could be two solutions for the branching fraction , and the two values could be very different .
as the luminosity at the @xmath2 peak is large enough , current study is limited by the low statistics at the continuum : at cleoc , the luminosity at continuum is more than an order of magnitude smaller than that at peak , this prevents from a high precision comparison between the cross sections at the two energy points .
one conclusion we can draw from the existing data is that the measurements do not contradict with the assumption that the relative phase between @xmath2 strong and electromagnetic decay amplitudes is around @xmath75 , and the @xmath2 decays into light hadrons could be large .
the inclusive non-@xmath5 decays of @xmath2 is searched for by measuring the @xmath5 cross section and the total hadronic cross section above the @xmath76 continuum contribution at the @xmath2 resonance peak .
bes and cleo measure the @xmath5 cross section @xcite using both single tag and double tag methods , the results are shown in fig .
good agreement between bes and cleo on the @xmath77 and @xmath78 , as well as the total @xmath5 cross sections are found .
the weighted average of the two experiments is @xmath79 nb for the total @xmath5 cross section .
the total hadronic cross section of @xmath2 decay is obtained by comparing the cross sections at the @xmath2 peak and the data points below the resonance peak .
the contributions of the radiative tails of @xmath0 and @xmath1 are also subtracted . by comparing this cross section and the @xmath5 cross section , bes found a non-@xmath5 decay branching fraction of @xmath80 @xcite , while cleo measured the non-@xmath5 cross section of @xmath81 nb @xcite , which corresponds to an upper limit of the non-@xmath5 decay branching fraction of 11% at 90% c. l. although the bes and cleo measurements are not in contradiction considering the large uncertainties , the big difference between the two central values calls for a better understanding of the procedure from which the total inclusive cross section is obtained .
an obvious difference between the two measurements is how to consider the interference between the resonance and the continuum amplitudes .
since each @xmath6 ( @xmath82 ) is produced about 10% of the @xmath1 decays , they are studied at bes to understand the p - wave charmonium decay dynamics , as well as the light hadron spectroscopy .
two modes , @xmath84 @xcite and @xmath85 , were measured recently , while the former is the first observation , the latter improves the precision .
the results are summarized in table [ vv ] , together with the measurement of @xmath86 @xcite .
these results are used to predict the decay branching fractions of @xmath6 to other vector meson pairs , like @xmath87 and @xmath88 @xcite , large double ozi suppressed amplitude is expected .
.branching fractions of @xmath89 vv ( in @xmath90 ) , the results on @xmath85 are bes preliminary . [
cols="^,^,^",options="header " , ] the above results supply important information on the understanding of the natures of the scalar states @xcite , as well as the decay dynamics of @xmath6 decays into pair of scalar particles .
there are many new results in charmonium decays from besii and cleoc experiments .
the decay properties of the vector charmonium states have been studied for more than three decades , but they are still far from being understood , one extreme example is the `` @xmath12 puzzle '' in @xmath0 and @xmath1 decays .
further studies of all these are expected from the besiii at bepcii which will start its data taking in late 2007 .
i would like to thank the organizer for the invitation to give the talk .
i thank my colleagues in bes collaboration for many helpful discussions .
this work was supported in part by national natural science foundation of china ( 10491303 ) and `` 100-talent program '' of cas under contract no .
u-25 .
many results may be found in ref .
@xcite ; more recent results may be found in bes collaboration , j. z. bai _ et al .
d * 69 * , 072001 ( 2004 ) ; phys .
* 92 * , 052001 ( 2004 ) ; bes collaboration , m. ablikim _ et al .
d * 70 * , 112007 ( 2004 ) ; phys .
d * 70 * , 112003 ( 2004 ) ; and phys .
b * 614 * , 37 ( 2005 ) .
bes collaboration , m. ablikim _ et al .
d * 71 * , 072006 ( 2005 ) ; phys .
d * 73 * , 052004 ( 2006 ) ; hep - ex/0605031 ; cleo collaboration , r. a. briere _ et al .
lett . * 95 * , 062001 ( 2005 ) ; cleo collaboration , t. k. pedlar _ et al .
d * 72 * , 051108(r ) ( 2005 ) .
# 1#2#3 phys .
d * # 1 * , # 2 ( # 3 ) g. d. bodwin , e. braaten , and g. p. lepage , .
b * 630 * , 21 ( 2005 ) ; .
, ; ; ; phys .
lett . * 81 * , 3091 ( 1998 ) .
s. j. brodsky , g. p. lepage , phys .
d * 24 * , 2848 ( 1981 ) .
d * 73 * , 052006 ( 2006 ) .
s. m. h. wong , eur .
j c * 14 * , 643 ( 2000 ) .
r. g. ping , b. s. zou , and h. c. chiang , eur .
j. a * 23 * , 129 ( 2005 ) . | recent results from bes and cleoc experiments on charmonium decays using @xmath0 , @xmath1 and @xmath2 data samples collected in @xmath3 annihilation are reviewed , including the measurement of the relative phase between strong and electromagnetic decays of @xmath1 , the study of the `` @xmath4 puzzle '' in @xmath0 and @xmath1 decays , and the search for the non-@xmath5 decays of @xmath2 .
the decays of @xmath6 produced in @xmath1 radiative transition are also reviewed .
these new results shed light on the understanding of qcd . |
the electronic properties of ultrathin films are significantly different from those of their bulk states due to their reduced dimensionality and the influence of substrates . in the case of oxides on metallic substrates
, it has been predicted that characteristic features such as the on - site coulomb interaction energy @xmath0 and the charge transfer energy @xmath4 from a ligand to a neighboring cation would be substantially altered for atomically thin films.@xcite altieri _
et al._@xcite observed that for an ultrathin mgo film on ag(001 ) , both @xmath0 and @xmath4 decreased monotonically as the film became thinner .
they attributed the decrements from the bulk values of @xmath0 and @xmath4 in the film , @xmath2 and @xmath5 , to the extra - atomic relaxation energies @xmath6 that developed in response to the altered charge states of the ions .
as the major sources of @xmath6 , they considered both the image charge potential energy @xmath7 between an extra charge and its image induced in the metal substrate and the polarization energy @xmath8 of the oxide caused by the extra charge also known as the madelung potential energy .
the magnitude of the image potential energy should be larger for thinner films due to the smaller mean distance between an extra charge in the film and its image in the substrate .
the polarization energy of the film should be different from that of the bulk oxide because the volume of the oxide in the film is reduced , whereas the polarizability is enhanced at the surface of the oxide .
the resulting variation between @xmath6 in the film and in the bulk state , @xmath9 , even quantitatively reproduced the experimental @xmath2 for a 1-monolayer ( ml ) mgo film on ag(001).@xcite nevertheless , the suspicion was raised that the successful reproduction of the experimental @xmath2 by @xmath9 could have been fortuitous , as there were many other effects ( such as dipole - dipole interactions ) that were not taken into account as well as unjustified assumptions ( such as @xmath10 dependence of the image potential on the atomic distance ) .
moreover , chambers and droubay@xcite reported that both fe@xmath11o@xmath12 and cr@xmath11o@xmath12 films on pt(111 ) exhibited negligible @xmath2 and @xmath5 .
this contrasting observation was attributed to effective intrinsic screening of charge transfer , which reduced the extra - atomic relaxation to an undetectable level .
thus , no comprehensive elucidation of the electronic properties of ultrathin oxide films on highly polarizable substrates seems to exist , and the number of experimental studies are very limited to properly assess the existing hypotheses . the objective of the present work is to assess the existing hypotheses by comparing their predictions with experimental results for a different system : ultrathin nio films on ag(001 ) .
bulk nio is prototypical as a charge - transfer insulator , @xcite and its charge fluctuation energies have already been studied.@xcite furthermore , the lattice mismatch between nio(001 ) and ag(001 ) is only @xmath13 2 @xmath14 , and the pseudomorphic growth of an nio film is well established.@xcite in other words , nio films grown on ag(001 ) are well suited for studying the thickness dependency of charge fluctuation energies such as @xmath0 and @xmath4 .
however , it is difficult to obtain the coulomb interaction energy @xmath0(ni @xmath15 ) between ni @xmath15 electrons via the method of altieri _
et al._,@xcite because the ni @xmath15 spectrum is difficult to isolate due to its overlap with the ag @xmath16 band of the substrate . instead , we study the interaction energy between ni @xmath1 holes @xmath0(ni @xmath1 ) .
as the film becomes atomically thin , @xmath0(ni @xmath1 ) exhibits a substantial reduction from its bulk value .
moreover , the extra - atomic relaxation energies represented by both @xmath7 and @xmath8 well reproduce the change in @xmath0(ni @xmath1 ) from bulk to thin film , @xmath2(ni @xmath17 , for a 1 ml nio film on ag(001 ) . using the observed values of @xmath2 and @xmath5 , we estimate the nel temperature @xmath18 in the mean field approximation that is found to be compatible with the experimental value of @xmath18 for a 3-ml nio film.@xcite these results reinforce the idea that the extra - atomic relaxation represented mainly by @xmath7 and @xmath8 determines @xmath2 and @xmath5 for ultrathin oxide films of nio , as well as mgo , on highly polarizable substrates .
we performed _ in situ _ scanning tunneling microscopy ( stm ) , photoelectron spectroscopy ( pes ) , and auger electron spectroscopy ( aes ) on ultrathin nio films grown on ag(001 ) .
the stm work was carried out with a variable - temperature stm ( omicron ) .
the nio films were grown in an attached preparation chamber , where the preliminary characterization of both ag substrate and nio film was accomplished by x - ray photoelectron spectroscopy ( xps ) and low - energy electron diffraction ( leed ) .
the pes and aes work were carried out with a soft x - ray beamline ( 7b1 ) at pohang light source in korea .
the end station of the beamline is composed of both an analysis chamber and a preparation chamber .
the analysis chamber is equipped with a hemispherical electron energy analyzer with a multichannel detector . for the pes ,
the photoelectrons are collected at a take - off angle of 45@xmath19 with respect to the surface normal of the sample .
the pes resolving power is @xmath13 4000.@xcite the zero point of the binding energy is determined in reference to the binding energy of the ag @xmath15 ( 368.3 ev ) of the clean ag substrate .
all spectra presented in this work were recorded with the sample maintained at room temperature . for both stm and pes ,
no charging effects were observed .
the nio films were grown in preparation chambers for both stm and pes .
the base pressures were @xmath20 @xmath21 torr for both chambers .
wedge - shaped nio films were grown by e - beam evaporation of high purity ( 5n ) ni rod onto clean ag(001 ) at room temperature at an ambient o@xmath11 pressure ( @xmath22 ) of @xmath23 torr .
the films were then thermally annealed at @xmath24 k at @xmath22 @xmath25 torr . in the present work ,
we are especially interested in films within the monolayer limit , as this enables a definite comparison of experimental @xmath2 with theoretical values obtained by considering extra - atomic relaxation energies . however , for films less than 2 ml , the growth mode is somewhat complicated due to the ( @xmath26 ) reconstruction and the bilayer growth of the nio film.@xcite under the aforementioned growth conditions , we were able to grow 1 ml ( @xmath27 ) nickel oxide films , as assessed by a combination of techniques , including stm , leed , and xps .
( further details are given in the following section . ) according to our previous extensive pes of nio films , such growth conditions also minimize the chemical defects.@xcite the thickness of each film was mainly determined by the ratio of the peak intensity of the ag @xmath15 in the nio - covered region to that of the clean ag substrate , assuming layer - by - layer ( lbl ) growth of the film . because the growth of a nio film does not follow lbl in an ideal fashion , the thicknesses described in the present work are nominal .
for coverage @xmath13 0.5 ml , the film is mainly composed of monolayer - high islands and can be taken as a model for a 1 ml film ( fig .
( further discussion is presented below . ) moreover , up to 0.5 ml , the coverage recorded by a quartz microbalance is in reasonable agreement with the nominal coverage estimated by the reduction of ag @xmath15 intensity , assuming layer - by - layer growth of the nio film .
based on these estimated film thicknesses , the growth rate is adjusted to @xmath13 0.25 ml / min throughout the experiments .
of @xmath28 torr .
the nio film coverage is ( a ) 0.5 ml , ( b ) 0.67 ml , and ( c ) 1 ml .
the scanning voltage and tunneling current were -2.7 v ( sample ) and 0.1 na , respectively . figs .
( b ) and ( c ) show leed images of ( @xmath27 ) patterns at an electron energy of 137 ev.,scaledwidth=75.0% ] figure 1 ( a)-1 shows a typical image of a nickel oxide film with @xmath13 0.5 ml coverage , consisting of nickel oxide patches . the line profile ( fig . 1 ( a)-3 ) across a typical patch in fig . 1 ( a)-2 displays a plateau of apparent height @xmath29 nm , which corresponded to 1 ml in our previous stm study of nio film on ag(001 ) under similar tunneling conditions.@xcite as the film ( with its nominal coverage of @xmath13 0.5 ml ) is mostly composed of islands of thickness 1 ml , we regard its ni and o spectra asrepresentative of the electronic properties of a 1 ml nio film . after further deposition to the nominal coverage
is around 1 ml , the second layer is preferably occupied ( fig . 1 ( b ) ) , and the film is almost bilayered .
( fig . 1 ( c ) )
all films exhibit the ( @xmath27 ) leed pattern ( fig . 1 ( b ) and ( c ) )
, whereas the well - known ( @xmath26 ) reconstruction is observed only sporadically in the stm images ( figure not shown ) .
however , the ( @xmath26 ) reconstruction becomes abundant if @xmath22 is lowered beneath 10@xmath30 torr .
and ( b ) ni @xmath1 spectra shift to the lower binding energy side , while ( c ) the centroid of the ni@xmath31 auger transition moves to the higher kinetic energy side .
( d ) the binding energy of the main peaks of the o @xmath32 spectra becomes smaller as the film becomes thinner .
the film thickness ranges from 0.5 to 15.00 ml .
all the spectra are normalized by the incident photon intensity.,scaledwidth=100.0% ] @xmath0(ni @xmath1 ) can be obtained by comparing the energy of a two - hole state , ni @xmath33 , to that of two one - hole state , ni @xmath34 , in accordance with the relationship @xmath35 the variation of @xmath0(ni @xmath1 ) from that of bulk ( actually bulk - like thick film ) @xmath2(ni @xmath1 ) can then be determined by the following relation , @xmath36 which is obtained via the approach of altieri _ et al._@xcite to estimate @xmath2(ni @xmath1 ) , we measured the xps spectra of ni @xmath37 , ni @xmath1 , and the ni @xmath38 auger transition as functions of the thickness of nio film .
we also utilized the o 1s spectra to estimate @xmath5(o @xmath37 @xmath39 ni @xmath15 ) , as described below .
figure 2 ( a ) , ( b ) , ( c ) , and ( d ) , respectively , show the ni @xmath37 , ni @xmath1 , ni@xmath31 auger transition , and o @xmath32 spectra of the nio films .
the thicknesses of the films range from submonolayer ( 0.5 ml ) to 15.00 ml .
even visual inspection reveals monotonic shifts of the major peaks of those spectra with variation of the film thickness .
however , the core - level spectra of ni comprise many peaks of various origins , such as final state effects and non - local screening,@xcite making identification of the main peak uncertain .
this complication is also transferred to the ni@xmath31 auger transition .
thus , we estimate shifts of the peak positions of ni @xmath37 , ni @xmath1 , and the ni@xmath31 auger transition with variation in the film thickness in terms of shifts of the centroids of their spectra , anticipating that if shifts of the peak positions are mainly caused by extra - atomic relaxation , then all of the component peaks should shift by the same amount . each spectrum
is fitted with a minimal number ( three or four ) of gaussian - convoluted lorentzian peaks with shirley backgrounds , which are used to obtain the centroid position .
the red dotted lines overlapping the experimental spectra in fig .
2 are best - fit curves .
to determine the peak positions of the o @xmath32 spectra , we fit each spectrum with gaussian - convoluted lorentzian peaks .
the full widths at half - maxima of major peaks are @xmath13 2.0 ev . the tick marks in the spectra indicate the resulting peak positions of the o @xmath32 spectra .
the curve - fitting results suggest the existence of some chemical defects , possibly ni@xmath11o@xmath12 , ni(oh)@xmath11 and/or nio(oh ) , which appear as small shoulders in the spectra.@xcite , ( b ) ni @xmath1 , ( c ) ni@xmath31 , and ( d ) the main peak position for o @xmath32 relative to those of the 15 ml film are shown as functions of the thickness of the nio film on ag(001 ) .
error limits were set by the scatter of the centroid positions , depending on the fitting parameters.,scaledwidth=50.0% ] figure 3 shows the centroid energies of ni @xmath37 , @xmath1 , the auger ni@xmath31 spectra , and the energy of the o @xmath32 main peak as functions of the film thickness relative to the corresponding energies of the 15 ml film that is considered as a bulk - like film .
even though the data points exhibit some scatter , we may readily observe that as the film becomes thinner , the peak positions of all the photoelectron spectra tend to shift toward the lower binding energy side , whereas the auger transition energy of ni@xmath31 increases monotonically .
for an ultrathin mgo film on ag(001 ) , a similar reduction of the binding energies of the photoelectrons and increase of the auger electron energy for relevant transitions are also observed as the thickness of the film decreases.@xcite in fig .
4 ( a ) , the values of @xmath2(ni @xmath1 ) obtained from eq . ( 2 ) are plotted relative to the value for the 15 ml film .
@xmath2(ni @xmath1 ) decreases monotonically as the film becomes thin , as is the case for @xmath2(mg @xmath37 ) of ultrathin mgo films on ag(001 ) .
however , @xmath2(ni @xmath1 ) changes very rapidly with increasing film thickness and is already negligible for films thicker than 5 ml .
this behavior is in contrast to that of mgo films on ag(001)@xcite , which exhibit substantial @xmath2 even for 10 ml ( although both mgo and nio of films have similar @xmath2 values for 1 ml coverage , as shown in fig .
4 ( a ) ) .
this is attributed to the larger polarizabilities @xmath40(o@xmath41 ) and @xmath40(ni@xmath42 ) of nio compared with mgo . in the bulk state ,
@xmath40(o@xmath41 ) of nio ( 1.98 @xmath43 ) is larger than that of mgo ( 1.65 @xmath43 ) . furthermore , @xmath40(ni@xmath42 ) @xmath13 0.68 @xmath43 , which is much larger than @xmath40(mg@xmath42 ) @xmath13 0.09 @xmath43 for mgo , possibly due to its closed - shell nature
. the larger polarizabilities of nio should make the screening of extra charges in the cation more effective , so extra - atomic relaxation should be more localized in nio films than in mgo films . hence , in response to charge fluctuation , nio films exhibit bulk - like behavior at smaller thicknesses than mgo films .
note that for fe@xmath44o@xmath45 , @xmath40(o@xmath41)@xmath46 is 2 @xmath13 2.91 @xmath43 ( ref .
@xcite ) , which is even larger than that of nio . thus , for the oxide film
, the coverage at which nonzero @xmath2 is observed would be further limited according to the above argument , possibly below the experimental limit , as chambers and droubay@xcite did not observe any @xmath2 for ultrathin fe@xmath44o@xmath45 films on pt(111 ) .
these authors also attributed the absence of @xmath2 to the large polarizabilities of the oxide .
( ni 3p ) and ( b ) @xmath47 ( defined in eq .
4 ) on the nominal thickness of the nio film on ag(001 ) .
the error limits are set by the fitting uncertainty .
the red line in each figure is the best - fit line for the data .
the blue lines show the theoretical values of @xmath2 and @xmath48 using the respective polarizabilities of bulk and surface.,scaledwidth=50.0% ] the shifts of peak positions summarized in fig .
3 can be suspected to originate from band bending due to charge transfer at the interface between the nio film and ag substrate .
however , the amount of peak shift varies for different transitions in the same film , as fig .
3 indicates .
hence , the peak shifts can not be attributed to band - bending effects .
furthermore , hybridization between the nio film and the ag substrate at the interface is shown to be very weak by photoelectron spectroscopy of the valence bands of the films@xcite , and this is also predicted by first principle calculations.@xcite
.both experimental and theoretical values of @xmath2(ni @xmath1 ) , @xmath47 , and e@xmath49 of a 1 ml nio film are summarized .
e@xmath50 for bulk ( 1 ml film ) is calculated using the polarizability of bulk ( surface ) nio .
the definition of @xmath47(ni @xmath1 ) is given in the text . [ cols="^,^,^,^ " , ] we investigate whether the extra - atomic relaxation represented by both @xmath7 and @xmath8 can account for the reduction of @xmath0 for the nio films , as well as for the 1 ml mgo film on ag(001).@xcite because it is not easy to acquire layer - resolved @xmath2 values experimentally for films thicker than 2 ml , we calculated @xmath7 , @xmath8 , and thus @xmath2 only for a 1 ml film .
@xmath2 is obtained from the following relation@xcite , @xmath51 where @xmath52 and @xmath53 are both chosen to be positive , following the convention of altieri _
et al._@xcite the contribution of @xmath7 to @xmath0(ni @xmath1 ) is obtained by comparing a two - hole state , ni @xmath33 , with two one - hole states , ni @xmath34 .
hence , @xmath7 is the difference between ( 2e)@xmath54 ) for @xmath55 and two one - hole states 2@xmath56e@xmath54 ) , corresponding to @xmath57 . here
, @xmath58 is the distance between a real charge and its image in the ag substrate . by the analysis of image potential surface states on clean ag(001 ) , the image plane is located 1.26 above the ag atoms in the surface layer.@xcite as a result , the ni atoms are separated from the image plane by 1.11 .@xcite thus , according to eq .
( 2 ) , @xmath7 contributes @xmath59 ev to @xmath2 for a 1 ml film , assuming that e@xmath49 is null for bulk nio .
@xmath8 is determined by the difference between the polarization energies of oxide for a two - hole state and two singly charged holes : @xmath60 - 2 @xmath61 . for the calculation of bulk @xmath8
, we employ the polarizabilities @xmath40(o@xmath41 , ni@xmath42 ) of bulk nio , @xmath40(o , ni)@xmath46 , while the polarizabilities of both o and ni at the surface of bulk nio , @xmath40(o , ni)@xmath62 , are used to calculate @xmath8 of the 1 ml film . for @xmath40(o@xmath63 , three values
have been reported : 1.49 , ( ref .
@xcite ) , 1.98 ( ref .
@xcite ) , and 2.64 ( ref .
@xcite ) @xmath43 . among these ,
1.98 @xmath43 is widely accepted .
a value of 2.43 has been reported for @xmath40(o@xmath64 ( ref .
@xcite ) , which best fits the leed i / v ( spot intensity versus electron energy ) of a bulk - terminated nio(001 ) surface.@xcite @xmath40(ni@xmath65 is obtained from the empirical relationship @xmath40(o@xmath41 ) + @xmath40(ni@xmath65 = 2.66 @xmath43 .
@xcite ) this relationship was obtained for bulk nio , but we tentatively assume that it holds down to a 1 ml film . the resulting @xmath8 s for both @xmath40(o@xmath63 and @xmath40(o@xmath64 are summarized in table i , along with @xmath7 for 1 ml nio film . using the @xmath8 and @xmath7 values in table i , we obtain the theoretical value of @xmath2 for a 1 ml nio film from eq .
the experimental value of @xmath2 ( -2.2 ev ) for the nominal 0.5 ml film , which is a model system for a 1 ml film , is well reproduced by the theoretical value , -2.48 ev ( see table i and fig .
this observation suggests that @xmath8 and @xmath7 are the major origins of @xmath2 for nio films , as well as for mgo films@xcite , and reinforces the model of duffy _
et al._@xcite and altieri _
et al._@xcite altieri _ et al .
_ suggested that manipulation of charge fluctuation energies of ultrathin oxide films can be used to control their physical properties , such as nel temperature @xmath18.@xcite reduced charge fluctuation energies affect the superexchange interaction in nio films . according to anderson s expression for the superexchange ,
the coupling constant @xmath66 depends on both @xmath0(ni @xmath15 ) and @xmath4(ni @xmath15 ) as follows : @xmath67 .
therefore , one can expect that the reduced values of @xmath0(ni @xmath15 ) and @xmath4(ni @xmath15 ) for an nio film would lead to an increase in the superexchange interaction for the film . in line with this conjecture , altieri _ et al .
_ found that for a 3 ml nio film on ag(001 ) , @xmath18 did not decay as much as for an nio film on an mgo substrate ( in which no image charge screening is expected , and which therefore exhibits less reduction of @xmath0 and @xmath4).@xcite to evaluate @xmath68 from eq . ( 4 ) , we use @xmath2 ( ni 3p ) in place of @xmath2 ( ni 3d ) ( which is not available ) .
the variation in @xmath0 has an extra - atomic origin , and thus we can expect that @xmath2(ni @xmath15 ) will differ little from @xmath2(ni @xmath1 ) .
we may then estimate @xmath5 along the lines of altieri _
et al._@xcite : @xmath69 here , we use @xmath2(ni @xmath1 ) in place of @xmath2(ni @xmath15 ) and denote the resulting @xmath5 by @xmath47 . from the intrapolation of @xmath2(ni @xmath1 ) and @xmath47 in fig .
4 , the values of @xmath2(ni @xmath1 ) and @xmath47 for a 3 ml film are estimated to be -0.62 and -0.47 , respectively . for @xmath0 and @xmath4 of bulk nio , we use 6.5 and 4.0 ev , respectively , referring to the report of taguchi _
et al._@xcite combining the above input , @xmath66 is found to be @xmath70 @xmath56 0.0364 for a 3 ml nio film on ag , while @xmath66 for bulk nio is @xmath70 @xmath56 0.0252 . here
, we assume the transfer integral @xmath71 between the anion o @xmath37 and the cation ni @xmath15 is the same for both bulk and the 3 ml film , as @xmath71 is a very local property and is assumed to be little influenced by extra - atomic effects .
we can now estimate @xmath18 for a 3 ml film in the mean field approximation . in the mean field approximation , @xmath72 , where s is the spin moment of an ni ion , and n the mean number of nearest neighbors of ni ions .
then , @xmath73 for a bulk - like thick nio film on ag(001 ) , @xmath18 was experimentally determined to be 535 k.@xcite 1.90 ( ref .
@xcite ) and 2.2 ( ref .
@xcite ) @xmath74 have been reported for the total magnetic moment @xmath75 of bulk nio .
first principle calculations predict that @xmath76 of a 3 ml nio film on ag(001 ) reduces to @xmath13 1.67 @xmath77 ( the average of the moments of the 1st , 2nd , and 3rd layers).@xcite if the ratio @xmath78 of the orbital moment to the spin moment is assumed to be the same ( 0.34 ) ( ref .
@xcite ) for both bulk and film , @xmath79 is 0.81 ( ref .
@xcite ) or 0.94 ( ref .
@xcite ) , and @xmath80 is 0.71 .
the mean number of nearest neighbors of a 3 ml film is @xmath13 9.33 .
if all this input is taken into account , then according to eq .
( 6 ) , @xmath18 is between 400 and 498 k for a 3 ml nio film on ag(001 ) .
the wide variation in @xmath18 originates mainly from the large uncertainty in the spin moment of bulk nio .
the experimentally determined @xmath18 of a 3 ml nio film is 390 k@xcite , which is close to the range of the present estimate . despite the many simplifications and assumptions , @xmath2(ni @xmath1 ) and @xmath48(ni @xmath1 )
seem to provide a reasonable estimate of the range of @xmath18 for an nio film under the mean field approximation .
most importantly , the lower limit of the present estimate ( 400.00 k ) is still much higher than the @xmath81 40 k observed for a 3 ml nio film on an mgo substrate.@xcite at the very least , this supports the argument that the reduction of charge fluctuation energies on a polarizable substrate gives rise to high values of @xmath18 ( as observed for the 3 ml nio film on ag(001 ) in comparison with the results for mgo(001 ) ) .
using both photoelectron spectroscopy and auger electron spectroscopy , we found that the on - site coulomb interaction energy @xmath0 of ultrathin nio films on ag(001 ) decreases monotonically in a manner analogous to the case of ultrathin mgo films on ag(001 ) .
the observed value of @xmath2 ( ni @xmath1 ) for a 1 ml film was well reproduced by considering extra - atomic relaxations represented by image charge screening by the ag substrate and modified polarization energy of the film , thus affirming the pictures of duffy _
et al._@xcite and altieri _
et al._@xcite furthermore , using @xmath2 ( ni @xmath1 ) , we estimated the value of t@xmath82 for a 3 ml nio film on ag(001 ) , and the estimate was comparable to experimental observation .
hence , the model proposed by the aforementioned authors seems to offer a unified picture of the variation in charge fluctuation energies of ultrathin mgo and nio films , even though further refinement is necessary in view of the many assumptions / estimations employed without precise quantitative justification .
this work was supported by the kosef grant no .
r01 - 2007 - 000 - 20249 - 0 and the nrf grant no . | we investigate the on - site coulomb interaction energy @xmath0(ni @xmath1 ) between two @xmath1 holes in an ultrathin nio film on ag(001 ) using both x - ray photoelectron spectroscopy and auger electron spectroscopy .
as the film becomes thinner , @xmath0(ni @xmath1 ) decreases monotonically , and the difference between the values of @xmath0(ni @xmath1 ) for a 1-monolayer ( ml ) film and a bulk - like thick film , @xmath2(ni @xmath1 ) , reaches @xmath3 ev . the observed value of @xmath2(ni @xmath1 ) for a 1 ml film is well reproduced by the differences between both the image potentials and polarization energies of the 1 ml film and the bulk - like thick film .
hence , the present results provide evidence in favor of the reduction of charge fluctuation energies in ultrathin films on highly polarizable substrates , as originally predicted by duffy _
et al . _ [ j. phys .
c : solid state phys . , * 16 * , 4087 ( 1983 ) ] and altieri _
et al._[phys . rev .
b * 59 * , r2517 ( 1999 ) ] . _ _ |
let @xmath8 \times { \mathbf r}\to { \mathbf r}$ ] be continuous and consider the differential equation @xmath9 , \end{aligned}\ ] ] where @xmath10 denotes the derivative of @xmath0 .
how complex can the solution @xmath0 be , assuming that @xmath3 is polynomial - time computable ? here , polynomial - time computability and other notions of complexity are from the field of _ computable analysis _ @xcite and measure how hard it is to approximate real functions with specified precision ( section [ section : preliminaries ] ) . if we make no assumption on @xmath3 other than being polynomial - time computable , the solution @xmath0 ( which is not unique in general ) can be non - computable .
table [ table : related ] summarizes known results about the complexity of @xmath0 under various assumptions ( that get stronger as we go down the table ) . in particular , as the third row says , if @xmath3 is ( globally ) lipschitz continuous , then the ( unique ) solution @xmath0 is known to be polynomial - space computable but still can be @xmath4-hard @xcite . in this paper , we study the complexity of @xmath0 when we put stronger assumptions about the smoothness of @xmath3 .
.complexity of the solution @xmath0 of assuming @xmath3 is polynomial - time computable [ cols= " < , < , < " , ] in numerical analysis , knowledge about smoothness of the input function ( such as being differentiable enough times ) often helps to apply certain algorithms or simplify their analysis
. however , to our knowledge , this casual understanding that smoothness is good has not been rigorously substantiated in terms of computational complexity theory .
this motivates us to ask whether , for our differential equation , smoothness really reduces the complexity of the solution . at one extreme
is the case where @xmath3 is analytic : @xmath0 is then polynomial - time computable ( the last row of the table ) by an argument based on taylor series , since we can efficiently compute the taylor sequence of @xmath0 from that of @xmath3 . ] ( this does not necessarily mean that computing the values of @xmath0 from those of @xmath3 is easy ; see the last paragraph of section [ section : constructive ] ) .
thus our interest is in the cases between lipschitz and analytic ( the fourth and fifth rows ) .
we say that @xmath3 is of class @xmath11 if the partial derivative @xmath12 ( often also denoted @xmath13 ) exists and is continuous for all @xmath14 and @xmath15 ; is of class @xmath7 if it is of class @xmath16 for all @xmath17 , @xmath18 with @xmath19 .
] it is said to be of class @xmath20 if it is of class @xmath11 for all @xmath21 .
we will show that adding continuous differentiability does not break the @xmath4-completeness that we knew from @xcite for the lipschitz continuous case : [ differentiableispspace ] there exists a polynomial - time computable function @xmath8 \times [ -1,1 ] \to { \mathbf r}$ ] of class @xmath22 such that the equation has a @xmath4-hard solution @xmath23 \to { \mathbf r}$ ] . the complexity notions ( computability and hardness ) in this and the following theorems will be explained in section [ section : preliminaries ] .
when @xmath3 is more than once differentiable , we did not quite succeed in proving that @xmath0 is @xmath4-hard in the same sense , but we will prove it @xmath24-hard , where @xmath25 is the counting hierarchy ( see section [ section : counting hierarchy ] ) : [ ktimesisch ] let @xmath26 be a positive integer .
there is a polynomial - time computable function @xmath8 \times [ -1,1 ] \to { \mathbf r}$ ] of class @xmath27 such that the equation has a @xmath24-hard solution @xmath23 \to { \mathbf r}$ ] . in theorems [ differentiableispspace ] and [ ktimesisch ] ,
we said @xmath8 \times [ -1 , 1 ] \to { \mathbf r}$ ] instead of @xmath8 \times { \mathbf r}\to { \mathbf r}$ ] , because the notion of polynomial - time computability of real functions in this paper is defined only when the domain is a bounded closed region
. do not hold as they are , because polynomial - time compubable functions @xmath3 , such as @xmath28 , could yield functions @xmath0 , such as @xmath29 , that grow too fast to be polynomial - time ( or even polynomial - space ) computable .
bournez , graa and pouly ( * ? ? ?
* theorem 2 ) report that the statement about the analytic case holds true if we restrict the growth of @xmath0 appropriately .
] this makes the equation ill - defined in case @xmath0 ever takes a value outside @xmath30 $ ] . by saying that @xmath0 is a solution in theorem [ differentiableispspace ]
, we are also claiming that @xmath31 $ ] for all @xmath32 $ ] .
this is no essential restriction , because any pair of functions @xmath8 \times { \mathbf r}\to { \mathbf r}$ ] and @xmath23 \to { \mathbf r}$ ] satisfying the equation could be scaled down in an appropriate way ( without affecting the computational complexity ) to make @xmath0 stay in @xmath30 $ ] . in any case , since we are making stronger assumptions on @xmath3 than lipschitz continuity , the solution @xmath0 , if it exists , is unique . whether smoothness of the input function reduces
the complexity of the output has been studied for operators other than solving differential equations , and the following negative results are known .
the integral of a polynomial - time computable real function can be @xmath33-hard , and this does not change by restricting the input to @xmath34 ( infinitely differentiable ) functions ( * ? ? ?
* theorem 5.33 ) .
similarly , the function obtained by maximization from a polynomial - time computable real function can be @xmath35-hard , and this is still so even if the input function is restricted to @xmath34 ( * ? ? ?
* theorem 3.7 ) needs to be replaced by , e.g. , @xmath36 ] .
( restricting to analytic inputs renders the output polynomial - time computable , again by the argument based on taylor series . ) in contrast , for the differential equation we only have theorem [ ktimesisch ] for each @xmath26 , and do not have any hardness result when @xmath3 is assumed to be infinitely differentiable .
theorems [ differentiableispspace ] and [ ktimesisch ] are about the complexity of each solution @xmath0 .
we will also discuss the complexity of the final value @xmath37 and the complexity of the operator that maps @xmath3 to @xmath0 ; see sections [ section : final value ] and [ section : constructive ] .
[ [ notation ] ] notation + + + + + + + + let @xmath38 , @xmath39 , @xmath40 , @xmath41 denote the set of natural numbers , integers , rational numbers and real numbers , respectively .
we assume that any polynomial is increasing , since it does not change the meaning of polynomial - time computable or polynomial - space computable .
let @xmath42 and @xmath43 be bounded closed intervals in @xmath41 .
we write @xmath44 for @xmath45 .
a function @xmath45 is of _ class @xmath46 _ ( or _ @xmath17-times continuously differentiable _ ) if all the derivatives @xmath47 exist and are continuous . for a function @xmath3 of two variables , we write @xmath48 and @xmath49 for the derivatives of @xmath3 with respect to the first and the second variable , respectively , when they exist .
a function @xmath50 is of _ class @xmath11 _ if for each @xmath51 and @xmath52 , the derivative @xmath53 exists and is continuous .
this derivative @xmath53 is then written @xmath54 ( and is known to equal @xmath55 for any sequence @xmath56 of @xmath57 @xmath58s and @xmath59 @xmath60s ) .
a function @xmath3 is of _ class @xmath20 _ if it is of class @xmath11 for all @xmath21 .
we assume that the reader is familiar with the basics of complexity theory , such as the classes @xmath61 and @xmath4 and the notions of reduction and hardness @xcite .
we briefly review the class @xmath24 .
the counting hierarchy @xmath24 is defined analogously to the polynomial - time hierarchy @xmath62 ( * ? ? ?
* kapitel 10.4 ) using oracle machines corresponding to the class @xmath63 ( * ? ? ?
* kapitel 3.3 ) instead of @xmath35 : thus , @xmath64 , where each level @xmath65 is defined inductively by @xmath66 and @xmath67 .
we leave the details of this definition to the original papers @xcite .
all we need for our purpose is the fact ( lemma [ lemma : cnp - complete ] below ) that @xmath65 is characterized by the following complete problem @xmath68 : given @xmath59 lists @xmath69 , , @xmath70 of propositional variables , a propositional formula @xmath71 with all free variables listed , and numbers @xmath72 , , @xmath73 in binary , determine whether @xmath74 is true . here
, @xmath75 is the _ counting quantifier _ : for every formula @xmath76 with the list @xmath77 of @xmath78 free variables , we write @xmath79 where formula @xmath80 is identified with the function @xmath81 such that @xmath82 when @xmath76 is true . [ lemma : cnp - complete ] for every @xmath83 , the problem @xmath68 is @xmath84-complete . by this
, we mean that any problem @xmath42 in @xmath65 reduces to @xmath68 via some polynomial - time function @xmath85 in the sense that @xmath86 if and only if @xmath87 . note
that this is analogous to the polynomial hierarchy @xmath62 , whose @xmath59th level @xmath88 has the complete problem that asks for the value of a formula of the form @xmath89 with @xmath59 alternations of universal and existential quentifiers . note also that the counting quantifier generalizes the universal and existential quentifiers , and hence @xmath90 in fact , it is known @xcite that @xmath62 is contained already in @xmath91 and thus in the second level of the counting hierarchy .
we review some complexity notions from computable analysis @xcite .
we start by fixing an encoding of real numbers by string functions .
a function @xmath92 is a _ name _ of a real number @xmath93 if for all @xmath94 , @xmath95 is the binary representation of @xmath96 or @xmath97 , where @xmath98 and @xmath99 mean rounding down and up to the nearest integer . in effect , a name of a real number @xmath93 takes each string @xmath100 to an approximation of @xmath93 with precision @xmath101 .
we use _ oracle turing machines _
( henceforth just _ machines _ ) to work on these names ( fig .
[ fig : model - of - function ] ) .
let @xmath102 be a machine and @xmath80 be a function from strings to strings .
we write @xmath103 for the output string when @xmath102 is given @xmath80 as oracle and string @xmath104 as input .
thus we also regard @xmath105 as a function from strings to strings . computing a real function @xmath106
let @xmath42 be a bounded closed interval of @xmath41 .
a machine @xmath102 _ computes _ a real function @xmath45 if for any @xmath107 and any name @xmath108 of it , @xmath109 is a name of @xmath110 .
computation of a function @xmath45 on a two - dimensional bounded closed region @xmath111 is defined similarly using machines with two oracles .
a real function is ( _ polynomial - time _ ) _ computable _ if there exists some machine that computes it ( in polynomial time ) .
polynomial - time computability of a real function @xmath112 means that for any @xmath94 , an approximation of @xmath110 with error bound @xmath113 is computable in time polynomial in @xmath59 independent of the real number @xmath93 . by the time the machine outputs the approximation of @xmath114 of precision @xmath101 , it knows @xmath93 only with some precision @xmath115
this implies that all computable real functions are continuous .
if the machine runs in polynomial time , this @xmath57 is bounded polynomially in @xmath59 .
this implies in the following lemma , which characterizes polynomial - time real functions by the usual polynomial - time computability of string functions without using oracle machines .
[ lem : type1representation ] a real function is polynomial - time computable if and only if there exist a polynomial - time computable function @xmath116 ) \times \{0 , 1\ } ^ * \to { \mathbf q}$ ] and polynomial @xmath117 such that for all @xmath118 $ ] and @xmath94 , @xmath119 and for all @xmath120 $ ] , @xmath94 , @xmath121 where each rational number is written as a fraction whose numerator and denominator are integers in binary . to speak about hardness of a real function ,
we define the notion of a language to it .
a language @xmath122 is identified with the function @xmath123 such that @xmath124 when @xmath125 .
[ definition : reduction ] a language @xmath126 _ reduces _ to a function @xmath127 \to { \mathbf r}$ ] if there exist a polynomial - time function @xmath128 and a polynomial - time oracle turing machine @xmath102 ( fig . [ fig : reduction ] ) such that for any string @xmath129 , 1 .
@xmath130 is a name of a real number @xmath131 , and 2 . for any name @xmath80 of @xmath132
, we have that @xmath133 accepts if and only if @xmath125 .
this reduction may superficially look stronger ( and hence the reducibility weaker ) than the one in kawamura @xcite in that @xmath102 can make multiple queries adaptively , but it is not hard to see that this makes no difference . for a complexity class @xmath134 ,
a function @xmath112 is _
@xmath134-hard _ if all languages in @xmath134 reduce to @xmath112 .
to a function @xmath135 \to { \mathbf r}$ ] ]
the proofs of theorems [ differentiableispspace ] and [ ktimesisch ] proceed as follows . in section [ section : divp ]
, we define _ difference equations _ , a discrete version of the differential equations .
we show that these equations with certain restrictions are @xmath4- and @xmath24-hard . in section [
subsection : ode family ] , we show that these classes of difference equations can be simulated by families of differential equations satisfying certain uniform bounds on higher - order derivatives . in section [
subsection : proof of theorems ] , we prove theorems [ differentiableispspace ] and [ ktimesisch ] by putting these families of functions together to obtain one differential equation with the desired smoothness ( @xmath22 and @xmath27 ) .
the idea of simulating a discrete system with limited feedback by differential equations was essentially already present in the proof of the lipschitz version of these results @xcite .
we look more closely at this limitation on feedback , and express it as a restriction on the _ height _ of the difference equation .
we show that a stronger height restriction allows the difference equation to be simulated by smoother differential equations ( see the proof of lemma [ ktimesfamily ] and the discussion after it ) , leading to the @xmath24-hardness for @xmath27 functions . in this section ,
we define difference equations , which are a discrete version of differential equations .
then we show the @xmath4- and @xmath24-hardness of families of difference equations with different height restrictions .
let @xmath136 $ ] denote @xmath137 .
let @xmath138 \times [ q ] \times [ r ] \to \{-1 , 0 , 1\}$ ] and @xmath139 \times [ q+1 ] \to [ r]$ ] .
we say that @xmath140 is the solution of the _ difference equation _ given by @xmath141 if for all @xmath142 $ ] and @xmath143 $ ] ( fig .
[ fig : divp ] ) , @xmath144 we call @xmath145 , @xmath146 and @xmath147 the _ height _ , _ width _ and _ cell size _ of the difference equation . the equations and are similar to the initial condition @xmath148 and the equation @xmath149 in , respectively . in section [
subsection : ode family ] , we will use this similarity to simulate difference equations by differential equations . of the difference equation given by @xmath141 ]
we view a family of difference equations as a computing system by regarding the value of the bottom right cell ( the gray cell in fig .
[ fig : divp ] ) as the output .
a family @xmath151 of functions @xmath152 \times [ q_u ] \times [ r_u ] \to \{-1 , 0 , 1\}$ ] _ recognizes _ a language @xmath126 if for each @xmath129 , the difference equation given by @xmath153 has a solution @xmath154 and @xmath155 .
a family @xmath151 is _ uniform _ if the height , width and cell size of @xmath153 are polynomial - time computable from @xmath129 ( in particular , they must be bounded by @xmath156 , for some polynomial @xmath157 ) and @xmath158 is polynomial - time computable from @xmath159 .
a family @xmath151 has _ polynomial height _ ( and _ logarithmic height _ , respectively ) if the height @xmath160 is bounded by @xmath161 ( and by @xmath162 , respectively ) . with this terminology , the key lemma in ( * ?
* lemma 4.7 ) can be written as follows : [ divppolyispspacehard ] there exists a @xmath4-hard language @xmath126 that is recognized by some uniform family of functions with polynomial height .
is in @xmath4 , because a uniform family with polynomial height can be simulated in polynomial space . ]
kawamura @xcite obtained the hardness result in the third row of table [ table : related ] by simulating the difference equations of lemma [ divppolyispspacehard ] by lipschitz continuous differential equations .
likewise , theorem [ differentiableispspace ] follows from lemma [ divppolyispspacehard ] by a modified construction that keeps the function in class @xmath22 ( sections [ subsection : ode family ] and [ subsection : proof of theorems ] ) .
we show further that difference equations restricted to logarithmic height can be simulated by @xmath27 functions for each @xmath26 ( sections [ subsection : ode family ] and [ subsection : proof of theorems ] ) .
theorem [ ktimesisch ] follows from this simulation and the following lemma .
[ divplogischhard ] there exists a @xmath24-hard language @xmath126 that is recognized by some uniform family of functions with logarithmic height .
we define the desired language @xmath126 by @xmath163 using the @xmath65-hard language @xmath68 from lemma [ lemma : cnp - complete ] .
then @xmath126 is obviously @xmath24-hard .
we construct a logarithmic - height uniform function family @xmath151 recognizing @xmath126 .
we will describe how to construct @xmath164 for a string @xmath129 of the form @xmath165 , where @xmath80 is a formula , and @xmath69 ,
, @xmath70 are lists of variables with lengths @xmath72 , , @xmath73 , respectively .
we write @xmath166 and @xmath167 . for each @xmath168 and @xmath169 , we write @xmath170 for the truth value ( @xmath58 or @xmath171 ) of @xmath172 .
note that @xmath173 , @xmath174 , and the relation between @xmath175 and @xmath176 is given by @xmath177 where we define @xmath178 by @xmath179 for @xmath180 , we write @xmath181 for the @xmath17th digit of @xmath182 written in binary ( @xmath183 being the least significant digit ) , and @xmath184}$ ] for the string @xmath185 . for each @xmath186 \times [ 2^{s_n}+1 ] \times [ 2^{|u|}]$ ] , we define @xmath187 as follows .
the first row is given by @xmath188 } , t_{[s_1 + 1,s_2 ] } , \dots , t_{[s_{n-1}+1,s_n]}),\ ] ] and for @xmath189 , we define @xmath190 } = 10 \cdots 0 , \\ 0 & \text{otherwise}. \end{cases } \label{eq : def - gu : case - nonzero}\ ] ] this family @xmath151 is clearly uniform , and its height @xmath191 is logarithmic in @xmath192 .
let @xmath154 be the solution ( as defined in and ) of the difference equation given by @xmath153 .
we prove by induction on @xmath17 that @xmath193 $ ] for all @xmath182 , and that @xmath194 } , \dots , v_{[s_{n-1}+1 , s_n]})\ ] ] if @xmath195 } = 10 \cdots 0 $ ] ( otherwise it is immediate from the definition that @xmath196 ) . for @xmath197 , the claims follow from .
for @xmath198 , suppose that the induction hypothesis @xmath199 } , \dots , v _ { [ s_{n-1}+1 , s_n]})\ ] ] holds . since @xmath200 is the solution of the difference equation given by @xmath164 , we have @xmath201 since the assumption implies that flipping the bit @xmath202 of any @xmath203 reverses the sign of @xmath204 , most of the summands in cancel out .
the only nonzero terms that can survive are the ones corresponding to those @xmath203 that satisfy @xmath205 } = 10 \cdots 0 $ ] and lie between the numbers whose binary representations are @xmath206 and @xmath207
. there are at most @xmath208 such terms , and each of them is @xmath171 or @xmath58 , so @xmath209 $ ] . in particular ,
if @xmath210 } = 10 \cdots 0 $ ] , we have @xmath211 } , \dots , v_{[s_{n-1}+1 , s_n]}).\ ] ] by this and , we have @xmath212 } , \dots , v_{[s_{n-1}+1 , s_n]}).\ ] ] by this and , we have , completing the induction step . by substituting @xmath59 for @xmath17 and @xmath213 for @xmath203 in , we get @xmath214 .
hence @xmath215 .
note that for @xmath24-hardness , we could have defined @xmath126 using a faster growing function than @xmath216 in .
this would allow the difference equation in lemma [ divplogischhard ] to have height smaller than logarithmic .
we stated lemma [ divplogischhard ] with logarithmic height , simply because this was the highest possible difference equations that we were able to simulate by @xmath27 functions ( in the proof of lemma [ ktimesfamily ] below ) .
we show that certain families of smooth differential equations can simulate @xmath4- and @xmath24-hard difference equations from the previous section . before stating lemmas [ ktimesfamily ] and [ differentiablefamily ]
, we extend the definition of polynomial - time computability of real functions to that of _ families _ of real functions .
a machine @xmath102 _ computes _ a family @xmath217 of functions @xmath218 indexed by strings @xmath129 if for any @xmath107 and any name @xmath108 of @xmath93 , the function taking @xmath219 to @xmath220 is a name of @xmath221 .
we say a family of real functions @xmath217 is polynomial - time if there is a polynomial - time machine computing @xmath217 .
[ ktimesfamily ] there exist a @xmath24-hard language @xmath126 and a polynomial @xmath222 , such that for any @xmath223 and polynomial @xmath224 , there are a polynomial @xmath225 and families @xmath226 , @xmath227 of real functions such that @xmath226 is polynomial - time computable and for any string @xmath129 : 1 .
[ enum : kf : start ] @xmath228 \times [ -1,1]\to { \mathbf r}$ ] , @xmath229 \to [ -1,1]$ ] ; 2 .
[ enum : equation ] @xmath230 and @xmath231 for all @xmath232 $ ] ; 3 .
[ enum : differentiability ] @xmath233 is of class @xmath27 ; 4 .
[ enum : boundary ] @xmath234 for all @xmath21 and @xmath235 $ ] ; 5 .
[ enum : smooth ] @xmath236 for all @xmath21 and @xmath237 ; 6 .
[ enum : kf : end ] @xmath238 .
[ differentiablefamily ] there exist a @xmath4-hard language @xmath126 and a polynomial @xmath222 , such that for any polynomial @xmath224 , there are a polynomial @xmath225 and families @xmath226 ,
@xmath227 of real functions such that @xmath226 is polynomial - time computable and for any string @xmath129 satisfying ( [ enum : kf : start])([enum : kf : end ] ) of lemma [ ktimesfamily ] with @xmath239 . in lemmas [ ktimesfamily ] and [ differentiablefamily ] , we have the new conditions ( [ enum : differentiability])([enum : smooth ] ) about the smoothness and the derivatives of @xmath233 that were not present in ( * ? ? ?
* lemma 4.1 ) . to satisfy these conditions , we construct @xmath233 using the smooth function @xmath112 in following lemma .
[ smoothfunction ] there exist a polynomial - time function @xmath127 \to { \mathbf r}$ ] of class @xmath34 and a polynomial @xmath240 such that * @xmath241 and @xmath242 ; * @xmath243 for all @xmath83 ; * @xmath112 is strictly increasing ; * @xmath244 is polynomial - time computable for all @xmath83 ; * @xmath245 for all @xmath83 .
although the existence of the polynomial @xmath240 was not explicitly stated in ( * ? ? ?
* lemma 3.6 ) , it can be proved easily .
we will prove lemma [ ktimesfamily ] using lemma [ divplogischhard ] as follows .
let @xmath151 be a family of functions obtained by lemma [ divplogischhard ] , and let @xmath246 be the family of the solutions of the difference equations given by @xmath151 .
we construct @xmath247 and @xmath233 from @xmath154 and @xmath153 so that @xmath248 for each @xmath249 , , @xmath250 and @xmath231 .
the polynomial - time computability of @xmath226 follows from that of @xmath151 .
we omit the analogous and easier proof of lemma [ differentiablefamily ] .
let @xmath126 and @xmath151 be as in lemma [ divplogischhard ] , and let @xmath154 be the solutions of the difference equations given by @xmath153 .
let @xmath112 and @xmath240 be as in lemma [ smoothfunction ] .
we may assume that @xmath251 \times [ 2^{q(|u| ) } ] \times [ 2^{r(|u| ) } ] \to \{-1 , 0 , 1\}$ ] for some @xmath157 , @xmath252 , @xmath253 , where @xmath157 has logarithmic growth and @xmath252 and @xmath254 are polynomials .
we may also assume , similarly to the beginning of the proof of ( * ? ? ?
* lemma 4.1 ) , that @xmath255 for some @xmath256 \to [ p ( \lvert u \rvert ) ] $ ] .
we construct families of real functions @xmath226 and @xmath227 that simulate @xmath164 and @xmath200 in the sense that @xmath257 , where the constant @xmath43 and the ( increasing ) function @xmath258 \to { \mathbf n}$ ] are defined by @xmath259 where @xmath260 is a polynomial satisfying @xmath261 ( which exists because @xmath157 is logarithmic ) .
thus , the value @xmath262 $ ] will be stored in the @xmath263th digit of the base-@xmath43 expansion of the real number @xmath264 .
this exponential spacing described by @xmath265 will be needed when we bound the derivative @xmath266 in below . for each @xmath267 \times [ -1 , 1]$ ] , there exist unique @xmath268 , @xmath269 , @xmath270 and @xmath271 such that @xmath272 and @xmath273 . using @xmath112 and a polynomial @xmath240 of lemma [ smoothfunction ]
, we define @xmath274 \to { \mathbf r}$ ] , @xmath275 \times [ -1 , 1 ] \to { \mathbf r}$ ] and @xmath276 \to [ -1 , 1 ] $ ] by @xmath277 we will verify that @xmath226 and @xmath227 defined above satisfy all the conditions stated in lemma [ ktimesfamily ] .
polynomial - time computability of @xmath226 can be verified using lemma [ lem : type1representation ] .
the condition ( [ enum : kf : start ] ) is immediate , and ( [ enum : equation ] ) follows from the relation between @xmath164 and @xmath200 ( by a similar argument to ( * ? ? ?
* lemma 4.1 ) ) .
since @xmath278 is constructed by interpolating between the ( finitely many ) values of @xmath164 using a @xmath279 function @xmath112 , it is of class @xmath279 and thus satisfies ( [ enum : differentiability ] ) . calculating from
, we have @xmath280 for each @xmath21 . by this and
, we have @xmath281 for each @xmath21 and @xmath282 for each @xmath21 and @xmath283 . substituting @xmath284 ( @xmath285 ) into ,
we get @xmath286 , as required in ( [ enum : boundary ] ) .
we show that ( [ enum : smooth ] ) holds with @xmath287 .
note that @xmath222 is a polynomial independent of @xmath26 and @xmath224 , and that @xmath288 by .
hence , @xmath289 by , and our choice of @xmath43 .
we have ( [ enum : kf : end ] ) with @xmath290 , because @xmath291 we used the exponential positioning function @xmath265 defined at , so that we have @xmath292 for the second inequality of for @xmath6 . because of this , we had to restrict @xmath157 to be logarithmic , because otherwise the function @xmath260 in would have to be superpolynomial and so would @xmath225 in .
the proof of lemma [ differentiablefamily ] is analogous , starting with the @xmath126 and @xmath151 of lemma [ divppolyispspacehard ] .
the only difference is that @xmath157 is now a polynomial and therefore we use @xmath293 in ( and @xmath294 in ) . using the function families @xmath226 and @xmath227 obtained from lemmas [ ktimesfamily ] or [ differentiablefamily ]
, we construct the functions @xmath3 and @xmath0 in theorems [ differentiableispspace ] and [ ktimesisch ] as follows .
divide @xmath295 into infinitely many subintervals @xmath296 $ ] , with midpoints @xmath297 .
we construct @xmath0 by putting a scaled copy of @xmath247 onto @xmath298 $ ] and putting a horizontally reversed scaled copy of @xmath247 onto @xmath299 $ ] so that @xmath300 , @xmath301 and @xmath302 where @xmath303 is a polynomial . in the same way
, @xmath3 is constructed from @xmath226 so that @xmath3 and @xmath0 satisfy .
we give the details of the proof of theorem [ ktimesisch ] from lemma [ ktimesfamily ] , and omit the analogous proof of theorem [ differentiableispspace ] from lemma [ differentiablefamily ] .
let @xmath126 and @xmath222 be as lemma [ ktimesfamily ] .
define @xmath304 , @xmath305 and for each @xmath129 let @xmath306 , @xmath307 , @xmath308 , where @xmath309 is the number represented by @xmath129 in binary notation .
let @xmath225 , @xmath226 , @xmath227 be as in lemma [ ktimesfamily ] corresponding to the above @xmath224 .
we define @xmath310 for each string @xmath129 and @xmath311 , @xmath312 $ ] .
let @xmath313 and @xmath314 for any @xmath235 $ ] .
it can be shown similarly to the lipschitz version ( * ? ? ?
* theorem 3.2 ) that @xmath3 and @xmath0 satisfy and @xmath3 is polynomial - time computable .
here we only prove that @xmath3 is of class @xmath27 .
we claim that for each @xmath21 and @xmath237 , the derivative @xmath315 is given by @xmath316 for each @xmath317 and @xmath318 $ ] , and by @xmath319 .
this is verified by induction on @xmath320 .
the equation follows from calculation ( note that this means verifying that follows from the definition of @xmath3 when @xmath321 ; from the induction hypothesis about @xmath322 when @xmath197 and @xmath323 ; and from the induction hypothesis about @xmath324 when @xmath198 ) . that @xmath319 is immediate from the induction hypothesis if @xmath197 .
if @xmath198 , the derivative @xmath325 is by definition the limit @xmath326 this can be shown to exist and equal @xmath171 , by observing that the first term in the numerator is @xmath171 and the second term is bounded , when @xmath327 $ ] , by @xmath328 where the second inequality is from lemma [ ktimesfamily ] ( [ enum : smooth ] ) and the fourth inequality holds for sufficiently large @xmath329 by our choice of @xmath224 .
the continuity of @xmath315 on @xmath330 $ ] follows from and lemma [ ktimesfamily ] ( [ enum : boundary ] ) . the continuity on @xmath331 $ ] is verified by estimating @xmath332 similarly to .
in the previous section , we considered the complexity of the solution @xmath0 of the ode as a real function . here
we discuss the complexity of the final value @xmath333 and relate it to tally languages ( subsets of @xmath334 ) , as did ko @xcite and kawamura ( * ? ? ?
* theorem 5.1 ) for the lipschitz continuous case .
we say that a language @xmath126 _ reduces to _ a real number @xmath93 if there is a polynomial - time oracle turing machine @xmath102 such that @xmath105 accepts @xmath126 for any name @xmath80 of @xmath93 .
note that this is the same as the reduction in definition [ definition : reduction ] to a constant function with value @xmath93 .
[ theorem : final value of once ] for any tally language @xmath335 , there are a polynomial - time computable function @xmath8 \times [ -1,1 ] \to { \mathbf r}$ ] of class @xmath22 and a function @xmath336 \to { \mathbf r}$ ] satisfying such that @xmath126 reduces to @xmath37 . [ theorem : final value of fixed ] let @xmath26 be a positive integer . for any tally language @xmath337 , there are a polynomial - time computable function @xmath8 \times [ -1,1 ] \to { \mathbf r}$ ] of class @xmath27 and a function @xmath336 \to { \mathbf r}$ ] satisfying such that @xmath126 reduces to @xmath37
. we can prove theorem [ theorem : final value of fixed ] from lemma [ ktimesfamily ] in the same way as the proof of ( * ? ? ? * theorem 5.1 ) .
we skip the proof of theorem [ theorem : final value of once ] since it is similar .
let @xmath182 be any tally language in @xmath24 and @xmath26 be any positive integer , and let @xmath126 and @xmath222 be as lemma [ ktimesfamily ] .
define @xmath338 , @xmath339 and let @xmath225 , @xmath226 , @xmath227 be as in lemma [ ktimesfamily ] corresponding to the @xmath224 . since @xmath126 is @xmath24-hard , there are a polynomial - time function @xmath85 such that @xmath340 for all @xmath17 .
let @xmath341 and @xmath342 .
define @xmath3 and @xmath0 as follows : when the first variable is in @xmath295 , let @xmath343 for each @xmath94 , @xmath232 $ ] , @xmath312 $ ] and @xmath344 ; when the first variable is @xmath58 , let @xmath345 it can be proved similarly to the proof of theorem [ ktimesisch ] that @xmath3 is polynomial - time computable and of class @xmath27 and that @xmath3 and @xmath0 satisfy .
the equation implies that @xmath182 reduces to @xmath37 .
both theorems [ differentiableispspace ] and [ ktimesisch ] state the complexity of the solution @xmath0 under the assumption that @xmath3 is polynomial - time computable . but
how hard is it to `` solve '' differential equations , i.e. , how complex is the operator that takes @xmath3 to @xmath0 ? to make this question precise , we need to define the complexity of operators taking real functions to real functions .
recall that , in order to discuss complexity of real functions , we used string functions as names of elements in @xmath41 .
such an encoding is called a _ representation _ of @xmath41 .
likewise , we now want to encode real functions by string functions to discuss complexity of real operators . in other words , we need to define representations of the class @xmath346}$ ] of continuous functions @xmath336 \to { \mathbf r}$ ] and class @xmath347 \times [ -1 , 1]}$ ] of lipschitz continuous functions @xmath348 \times [ -1 , 1 ] \to { \mathbf r}$ ] .
the notions of computability and complexity depend on these representations .
following @xcite , we use @xmath349 and @xmath350 as the representations of @xmath351}$ ] and @xmath347 \times [ -1 , 1]}$ ] , respectively .
it is known that @xmath349 is the canonical representation of @xmath346}$ ] in a certain sense @xcite , and @xmath350 is the representation defined by adding to @xmath349 the information on the lipschitz constant .
since these representations use string functions whose values have variable lengths , we use _ second order polynomials _ to bound the amount of resources ( time and space ) of machines @xcite , and this leads to the definitions of second - order complexity classes ( e.g. @xmath352 , polynomial - space computable ) , reductions ( e.g. @xmath353 , polynomial - time weihrauch reduction ) , and hardness . combining them with the representations of real functions mentioned above
, we can restate our theorems in the constructive form as follows .
let @xmath354 be the operator mapping a real function @xmath355 \times [ -1 , 1]}$ ] to the solution @xmath356}$ ] of .
the operator @xmath354 is a partial function from @xmath347 \times [ -1 , 1]}$ ] to @xmath346}$ ] ( it is partial because the trajectory may fall out of the interval @xmath30 $ ] , see the paragraph following theorem [ ktimesisch ] ) . in (
* theorem 4.9 ) , the @xmath357-@xmath352-@xmath353-completeness of @xmath354 was proved by modifying the proof of the results in the third row of table [ table : related ] .
theorem [ differentiableispspace ] can be rewritten in a similar way .
that is , let @xmath358 be the operator @xmath354 restricted to class @xmath22 .
then we have : to show this theorem , we need to verify that the information used to construct functions in the proof of theorem [ differentiableispspace ] can be obtained easily from the inputs .
we omit the proof since it does not require any new technique .
theorem [ theorem : c1 constructive ] automatically implies theorem [ differentiableispspace ] because of ( * ? ? ?
* lemmas 4.7 and 4.8 ) and the @xmath353 versions of ( * ? ? ?
* lemmas 3.11 and 3.12 ) .
in contrast , the polynomial - time computability in the analytic case ( the last row of table [ table : related ] ) is _ not _ a consequence of a statement based on @xmath349 .
it is based on the calculation of the taylor coefficients , and hence we only know how to convert the taylor sequence of @xmath3 at a point to that of @xmath0 . in other words ,
the operator @xmath354 restricted to the analytic functions is not @xmath357-@xmath359-computable , but @xmath360-@xmath359-computable , where @xmath361 is the representation that encodes an analytic function using its taylor coefficients at a point in a suitable way .
more discussion on representations of analytic functions and the complexity of operators on them can be found in @xcite .
o. bournez , d. graa , and a. pouly . solving analytic differential equations in polynomial time over unbounded domains . in _ proceedings of the 36th international symposium on mathematical foundations of computer science _
( mfcs ) , lncs 6907 , 170181 , 2011 .
a. kawamura .
on function spaces and polynomial - time computability .
dagstuhl seminar 11411 : computing with infinite data , 2011 .
a. kawamura and s. cook . complexity theory for operators in analysis . _
acm transactions on computation theory _ , 4(2 ) , article 5 , 2012 .
a. kawamura , n. mller , c. rsnick , and m. ziegler .
parameterized uniform complexity in numerics : from smooth to analytic , from np - hard to polytime .
a. kawamura , h. ota , c. rsnick , and m. ziegler .
computational complexity of smooth differential equations . in
_ proceedings of the 37th international symposium on mathematical foundations of computer science _ ( mfcs ) , lncs 7464 , 578589 , 2012 .
i. wegener .
_ komplexittstheorie : grenzen der effizienz von algorithmen_. springer , 2003 . in german .
english translation by r. pruim : _ complexity theory : exploring the limits of efficient algorithms _ , springer , 2005 .
( h. ota , a. kawamura , m. ziegler , and c. rsnick .
complexity of smooth ordinary differential equations .
presented at the tenth eatcs / la workshop on theoretical computer science , kyoto , 2012 . ) in japanese . | the computational complexity of the solution @xmath0 to the ordinary differential equation @xmath1 , @xmath2 under various assumptions on the function @xmath3 has been investigated .
kawamura showed in 2010 that the solution @xmath0 can be @xmath4-hard even if @xmath3 is assumed to be lipschitz continuous and polynomial - time computable .
we place further requirements on the smoothness of @xmath3 and obtain the following results : the solution @xmath0 can still be @xmath4-hard if @xmath3 is assumed to be of class @xmath5 ; for each @xmath6 , the solution @xmath0 can be hard for the counting hierarchy even if @xmath3 is of class @xmath7 . |
the mid - infrared ( mid - ir ) spectra of galaxies either with active nucleus ( agn ) and/or starburst ( sb ) show emission features attributed to polycyclic aromatic hydrocarbons ( pah ) , which can be considered to be originated in very small amorphous carbon dust grains or very large carbon - rich ring molecules ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the most prominent , well - known pah emissions , are the 6.2@xmath1 m , 7.7@xmath1 m , 8.6@xmath1 m , 11.2@xmath1 m and 12.7@xmath1 m bands ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
besides , the mid - ir spectral region also presents prominent forbidden ionic emission lines , such as [ neii]12.8@xmath1 m , [ nev]14.3@xmath1 m , [ neiii]15.5@xmath1 m , [ siii]18.7@xmath1 m and 33.48@xmath1 m , [ oiv]25.89@xmath1 m and [ siii]34.8@xmath1 m ( e.g. * ? ? ? * ; * ? ? ?
the nature of the ionizing source can be assessed by computing the intensity ratios between forbidden lines such as [ neii]12.8@xmath1 m , [ neiii]15.5@xmath1 m and [ nev]14.3@xmath1 m ( e.g. * ? ? ?
in addition , the detection of [ nev ] at 14.3@xmath1 m and 24.2@xmath1 m or [ nevi ] at 7.6@xmath1 m , due to ions of high ionization potential , is an evidence of a hard radiation field associated with an agn @xcite .
several diagnostic diagrams based on the line intensity ratios between the brightest ionic lines and the pah features have been proposed to classify the mid - ir spectra of galaxies according to the degree of activity ( agn , low - ionization nuclear emission - line region - liner , and sb galaxies ) .
for instance , @xcite , using data from the short wavelength spectrometer ( sws ) on board of the infrared space observatory ( iso ) , propose the diagnostic diagram , [ oiv]25.9@xmath1m/[neii]12.8@xmath1 m versus the strength of the 7.7@xmath1 m pah band to separate star - forming galaxies from agns . from the study of the mid - ir spectra of quiescent spirals , galactic hii regions , planetary nebulae , and photodissociation regions ( pdrs ) , @xcite found that the ratios between the 6.2@xmath1m/8.6@xmath1 m and 7.7@xmath1m/8.6@xmath1 m pah bands do not vary significantly ,
while the ratios 6.2@xmath1m/11.3@xmath1 m and 7.7@xmath1m/11.3@xmath1 m varies over one order of magnitude .
they concluded that the properties of the pahs for different types of galaxies are very similar and that variations of the pah band ratios are due to the fraction of ionized to neutral pah molecules .
thus , this ratio could be used to analyze the physical conditions of the molecular gas in the regions where the emission is originated .
in addition , @xcite found that the equivalent width ( ew ) of pah features , observed in hii regions of the m101 galaxy are correlated with the ionization index , and does not depend on the hii regions metallicity . according to @xcite the ionized to neutral fraction of pahs
can be estimated using pah 7.7@xmath1m/11.3@xmath1 m ratio , which is sensitive to the type of nuclear activity , decreasing from sb to seyfert .
these authors , suggest that this behavior is due to the selective destruction of the 7.7@xmath1 m pah molecule with the increasing hardness of the radiation field ( similar results are found by * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
such results are supported by laboratory experiments , which show that certain pahs are effectively destroyed by individual uv and x - ray photons , and can not survive within a few kiloparsecs from an active nucleus , unless they are shielded from the agn s x - ray emission by absorbing material @xcite .
it is worth mentioning that the above results , concerning the ionization fraction of pah bands in agns are carried out with small samples of this type of object
. therefore , we use the largest sample of agns and sb galaxies available up to date in spitzer public archive , to study in detail the effect of the hardness of the radiation field on the pah bands emission , in these kind of galaxies .
this paper is structured as follows : in sect .
[ sample ] we describe the sample selection , data reduction process and emission line measurements . in sect .
[ results ] we present results and discussions on the shape of the continuum spectra , emission line frequency and diagnostic diagrams of pah band ratios and emission lines ratios . sect .
[ correlacaopahne ] we discuss the behavior of the ionization fraction and ew of pah bands with the hardness of radiation field . summary and conclucion are presented in sect .
[ conclusions ] .
we have analyzed a well defined sample of 186 galaxies which have spitzer spectra available from public archive .
being 83 agns taken from @xcite , 22 sb from @xcite , 59 hii and liner from @xcite and 7 hii regions of m101 galaxy from @xcite , plus 15 agns spitzer spectra published here for the first time .
details on the reduction process of the 15 new objects are given below .
the 15 new spectra of agn galaxies ( 4 sy 1 and 11 sy 2 ) were obtained from the spitzer public archive .
observations were performed with the infrared spectrograph ( irs ; * ? ? ? * ) using the short - low ( sl ) and long - low ( ll ) modules , covering the interval between 5@xmath1 m and 36@xmath1 m with a resolving power of 64 - 128 .
the raw data were processed using the spitzer pipeline version 17.2 .
the nuclear spectra were extracted using the spitzer irs custom extractor - spice .
default point - source extractions were selected for all objects .
only a few sources show a noticeable step in flux between the overlapping ll and ls module spectra , which results from the extended circumnuclear emission contribution to the size sl module slit of [email protected] x 57@xmath5 and ll module slit of [email protected] x 168@xmath5 .
although the observed regions of almost all of the galaxies are large in both modules the agn light dominates the spectra . fig . [ profile ]
shows the emission spatial profile of ngc3786 .
note that the luminosity falls off to almost zero for @xmath6 3 pixels from the center of the nucleus .
for this galaxy , 3 pixels correspond to @xmath6 1 kpc , adopting a radial velocity of 2678 km sec@xmath7 and a hubble constant of 74 km sec@xmath7mpc@xmath7 .
the general parameters for the objects listed in table [ tabgeral ] were taken from nasa / ipac extragalactic database ( ned ) . all the spectra were corrected for redshift .
the rest - frame spectra were grouped according to their activity , and are shown in fig .
[ allnorm ] . for the 15 new agns listed in the table [ tabgeral ] and
sb taken from @xcite , low resolution irs spectra have been decomposed using the pahfit code , described in detail by @xcite .
in short , they assume that the irs spectra are composed by dust continuum , starlight , prominent emission lines , individual and blended pah emission bands , and that the light is attenuated by extinction due to silicate grains
. the default input parameters of pahfit are : ( i ) the infrared emission from a blackbody with t= 5000k ; ( ii ) different weights for thermal dust continuum components represented by the temperatures 35 , 40 , 50 , 90 , 135 , 200 , 300k ; ( iii ) spectral line features , including the pure rotational lines of the molecular hydrogen ; ( iv ) dust features are represented by individual and blended drude profiles , which are theoretical frequency profiles for a classical damped harmonic oscillator and are the best choice to model pah emission . for more details see @xcite . with the above constraints we were able to isolate the pah features and measure their fluxes , as well as those of the ionized gas emission lines .
the fluxes and the ew are listed in tables [ fluxpah ] to [ fluxionic ] .
we checked the consistency of the results with the liner code @xcite that integrates the flux of a gaussian function fitted to the line , obtaining similar results in both procedures .
the observed irs spectra of mrk334 , a sy 1 galaxy , is shown in fig .
[ pahfit ] , where we also plot the detailed components of the pahfit spectral decomposition model .
clearly , the model reproduces well the observed spectrum .
the main goal of this section is to characterize the continuum shapes and spectral features observed in the 15 new objects listed in table [ tabgeral ] , with respect to the type of nuclear activity ( sy 1 and sy 2 ) .
therefore , we normalize to unity the continuum of all spectra at @xmath0 = 23 @xmath1 m and grouped them according to nuclear activity . for each agn class , data were sorted according to spectral shape , from bluest to reddest , top to bottom of fig .
[ allnorm ] , respectively .
one can clearly see that the majority of spectra have a continuum shape increasing for longer wavelengths , in agreement with results found by @xcite .
this continuum range can be fitted with a warm blackbody distribution of t @xmath2 150 300k .
interestingly , a similar trend in the continuum of agns is also observed in the near - infrared region @xcite .
we compare the spectroscopic properties of the 15 new objects with those of the @xcite groups .
we found that the spectra of the sy 2 galaxies : ngc3786 , ngc5728 , ngc7682 , mrk471 , mrk609 , mrk622 , mrk883 , mrk1066 , mrk883 and the sy 1 mrk334 and ngc4748 , have the same characteristic as their group 1 ( strong pah emission and red continuum , see fig [ allnorm ] ) . on the other hand the sy 2 , ngc 1275 , ngc2622 ,
mrk3 and sy 1 mrk 478 can be classified as group 3 ( power - law continuum and weak pah features ) , while only the sy 1 mrk279 galaxy show a broken continuum and silicate emission at 9.8@xmath1 m ( group 2 ) . in summary , for our and @xcite samples
the mid - ir spectra for most of sy 2 galaxies have same spectroscopic characteristics as those of group 1 ( @xmath8 50@xmath9 ) .
[ hist ] shows a histogram of the commonest emission features present in each type of activity ( note that we have used the 186 galaxies spectra ) .
it is clear from this figure that the pah features are present in almost all the sy 1 , sy 2 and sb galaxies , as well as the h@xmath10 molecules at 9.6@xmath1 m and 17@xmath1 m .
in addition , the most frequent ionic lines are : [ siii ] 34.8@xmath1 m , [ arii ] 6.9@xmath1 m , [ siii ] 18.7 and 33.4@xmath1 m , [ siv ] 10.5@xmath1 m , [ neiii ] 15.53@xmath1 m and [ nev ] 14.3@xmath1 m . all
studied sb galaxies show pah bands and the following forbidden emission lines : [ siii]34.8@xmath1 m , [ arii]6.9@xmath1 m , [ siii]18.7 and 33.4@xmath1 m .
in addition , nearly 80@xmath9 show molecular h@xmath10 lines .
similarly , @xmath480@xmath9 of the sy 2 show pah features , and the same ionic lines detected in sb galaxies .
however , we have found that pah lines in sy 1 galaxies behave differently from the other types of nuclear activity .
the pah bands at 7.7@xmath1 m , 11.3@xmath1 m and 12.7@xmath1 m are present in 80% of the galaxies , but only 50% of them show the pah lines at 6.2@xmath1 m and 8.6@xmath1 m . nevertheless , the high ionization lines , such as [ nev]14.3@xmath1 m and [ oiv]25.8@xmath1 m are more common in sy 1 .
numerical studies of the vibrational energy distribution of pah molecules show that the relative strengths of the 3.3@xmath1 m , 6.2@xmath1 m , 7.7@xmath1 m , 8.6@xmath1 m and 11.3@xmath1 m pah features depend on grain size and on the charging conditions @xcite .
small pahs radiate strongly at 6.2@xmath1 m and 7.7@xmath1 m , while large pahs emit mostly at longer wavelengths @xcite .
another important conclusion of @xcite is that neutral pahs have higher values of 11.3@xmath1m/7.7@xmath1 m ratio , while ionized pahs have lower ratios . on the other hand ,
the 6.2@xmath1m/7.7@xmath1 m ratio decreases with the increasing number of carbon atoms of the pah molecules , fig . 16 from @xcite
shows how the relative strengths of these emission vary depending on the size and charge state of pah molecules .
@xcite used this diagram to analyze these pah ratios for a sample formed by galaxies dominated by star formation , galaxies with significant stellar and agn component and agn - dominated galaxies .
they found a weak trend in the direction of constant ionization fraction with changing grain size .
in addition , they suggest that the presence of an agn component is correlated with the reduction in the 7.7@xmath1m/11.3@xmath1 m ratio indicating that smaller pah grains ( number of carbon atoms @xmath11180 ) are destroyed in agns .
we present in fig .
[ pahs ] the theoretical intensity ratios , for neutral and ionized pahs ( 6.2@xmath1m/7.7@xmath1 m @xmath3 11.3@xmath1m/7.7@xmath1 m ) , from @xcite .
the authors calculated these quantities assuming grains illuminated either by the @xcite spectrum or a blackbody with t= 3 x 10@xmath12k varying the number of carbon atoms in each molecule .
we overplot in this diagram the observed pah emission line ratios of our sample .
clearly , the observed ratios are located between neutral and ionized pah theoretical lines , in agreement with previous studies @xcite .
the above results suggest that the observed pah bands in both , agns and low ionization objects ( sb and liners ) , are formed by an appropriate mixture of pah molecules with different sizes and adequate neutral to ionized fraction .
high ionization objects ( agns ) present a tendency for molecules @xmath8 180 carbon atoms ( 6.2@xmath1m/7.7@xmath1 m @xmath11 0.2 ) , on the other hand most of low ionization objects are biased towards molecules with less than 180 carbon atoms .
the latter are typically located in the region 0.2 @xmath116.2@xmath1m/7.7@xmath1 m @xmath11 0.4 ( fig .
[ pahs ] ) . furthermore
, seyfert galaxies appear to be located near to region populated by ionized pah molecules , while the majority of sb , hii , and liner are near to the line representing the neutral pahs
. moreover , these results are consistent with a picture that agn have higher ionization fraction and larger pah molecules than sb galaxies .
another way to assess the ionization fraction of the pah molecules is using the following diagnostic diagram 6.2@xmath1m/11.3@xmath1 m @xmath3 7.7@xmath1m/11.3@xmath1 m .
such diagram was used by @xcite , @xcite and @xcite , they demonstrate that there is a good correlation between both ratios .
however , we would like to emphasize the fact that both samples were mostly composed by sb galaxies .
in addition , these ratios are controlled by the fraction of ionized to neutral pahs @xcite .
in fig [ pahspahs ] ( top ) we show these diagram for our sample . clearly , if we only consider the sb galaxies our results are consistent with the correlation described above . by including the agns ( sy 1 and sy 2 ) ,
an unexpected large dispersion is observed for both type of activity .
alternatively , we also test the diagram 6.2@xmath1m/8.6@xmath1 m @xmath3 7.7@xmath1m/8.6@xmath1 m involving ionized ( 7.7@xmath1 m and 6.2@xmath1 m ) and neutral ( 8.6@xmath1 m ) pah molecules .
this diagram is shown in fig.[pahspahs ] ( bottom ) , where a similar dispersion is observed .
interestingly , in both diagnostic diagrams the ratios in the vertical axes tend to separate the agns ( sy 1 and sy 2 ) from sb galaxies .
this is even clearer in the diagram involving the 7.7@xmath1m/8.6@xmath1 m ratio , leading us to the conclusion that 7.7@xmath1m/8.6@xmath1 m @xmath4 6 for agns
( dotted line ) and smaller for low ionization objects .
synthetic spectra in the 6.2@xmath1 m 9.0@xmath1 m region , for large symmetric pahs cations , anions and neutral , have been computed by @xcite .
the trends , in the band position and intensite , is a function of the molecule size , charge and geometry .
large pah cations and anions ( @xmath8 110 carbon atmos ) produce prominente bands at 7.7@xmath1 m and 8.5@xmath1 m , the intensite ratio 7.7@xmath1m/8.6@xmath1 m increase with the pah sizes .
thus , we interpret the separation in activity types ( sb , sy 1 and sy 2 ) , along the vertical axis in both diagrams of fig .
[ pahspahs ] , as due to the fact that in average the emitting pah molecules in seyferts are larger than those of sb .
in addition , we suggest that the high values for the ratios 7.7@xmath1m/8.6@xmath1 m and 7.7@xmath1m/11.3@xmath1 m is due to the fact that almost all the molecules are ionized . in summary , we point out that both diagrams ( figs . [ pahs ] and [ pahspahs ] )
clearly shows that the ionization fraction and molecules size increase from sb to agns .
however , it is worth mentioning that , no separation between seyfert type is observed .
although , the four ratios depends on the pahs ionization fractions , ratios involving 6.2@xmath1 m band have the same range of variation for both types of nuclear activity .
this can be associated with the fact that the 6.2@xmath1 m pah emission is due to small molecules @xcite , which probably are destroyed by the hard radiation field of seyfert galaxies . as shown in sect .
[ freqemlin ] ( fig .
[ hist ] ) , forbidden ionic emission lines are observed in sb and active galaxies . in the first case , the gas is ionized by massive hot stars , while in the latter the ionization is due to a non - thermal continuum .
moreover , the emission cloud in agns are located in the narrow line regions ( nlr ) @xcite .
the mid - ir emission from dusty nlr of agns has been explored by @xcite .
their models give a direct connection between the dust emission and the line emission from the photoionized gas .
these authors also show that the emission line ratios [ nev]/[neii ] @xmath3 [ neiii]/[neii ] can be used to separate agns from sbs .
in addition , @xcite using starburst evolutionary models , which incorporate new stellar atmosphere models for massive stars , found that [ neiii]/[neii ] is sensitive to hardness of the radiation field . in fig .
[ nene ] we show the diagnostic diagram [ nev]14.3@xmath1m/[neii]12.8@xmath1 m @xmath3 [ neiii]15.5@xmath1m/ [ neii]12.8@xmath1 m for our sample .
we also overplot the nlr models for different ionization and pressure parameters taken from @xcite .
clearly , all the observations have values lower than those predicted by the models . in this diagram , agns are closely correlated , while the sb galaxies show a large scattering .
the [ nev]14.3@xmath1 m has a high ionization potential of 97.1 ev and is often detected in spectra of agns ( e.g. * ? ? ?
* ) however is rarely detected in sb galaxies @xcite .
the lack of correlation for sb galaxies in this diagram , thus , may be due to a strong contamination of [ nev ] with pah 14.19@xmath1 m emission or to the absence of the former line in the sb galaxies spectra .
in addition , we call the attention to the fact that the deblending of these lines is very difficult in the low resolution spitzer spectra .
it is also clear in this figure that sy 1 galaxies have [ neiii]/[neii ] @xmath4 0 , while sb and sy 2 have this ratio @xmath13 0 ( solid line in fig . [ nene ] ) .
this suggests that both line ratios are sensitive to the hardness of the radiation field , confirming the result of @xcite . the fact that sy 2 galaxies are spread over the lower left of the fig .
[ nene ] can be associated to @xcite predictions , which states that starburst contribution would move the points in that direction .
it is worth mentioning that there are 3 sb galaxies located in the top - right side of fig .
[ nene ] ( ngc1097 , ngc4676 and ngc520 ) .
the former , was reclassified as sy 1 by @xcite , while the two latter are strongly interacting sources and may have the these emission lines enhanced @xcite .
thus we suggest that further , more accurate studies ( i.e. with higher spectral resolution spectra ) , are necessary to better explain the position of the sb galaxies in the diagram .
@xcite have studied the correlation between the intensity ratio of the strongest pah bands 7.7@xmath1m/11.3@xmath1 m with the hardness of the radiation field indicator , [ neiii]/[neii ] . they have found that galaxies with hii regions or starburst optical spectra exhibit a nearly constant 7.7@xmath1m/11.3@xmath1 m across the whole range of radiation hardness , including low - metallicity galaxies , which have [ neiii]/[neii ] @xmath11 2 .
in contrast , the agns are located below the hii - type galaxies and their pah ratio falls rapidly with increasing radiation field .
this result was confirmed by @xcite through the analysis of pah emission bands of a sample of 92 typical star forming galaxies .
they interpreted the decrease in the pahs band ratio for agns as due to a selective destruction of the ionized 7.7@xmath1 m pah molecules .
the variation of 7.7@xmath1m/11.3@xmath1 m pah bands , observed in our sample ( 186 objects ) , with respect to the hardness of the radiation field , is shown in fig .
[ pahne ] .
it is clear from this figure , that the fraction of ionized to neutral pah bands does not depend on the hardness of the radiation field in contrast with the previous results of @xcite and @xcite .
we interpret this behavior as due to the fact that the strengths of both pah bands would have the same dependence on the radiation field .
besides , this diagram show a tendency in the sense that sb galaxies have [ neiii]/[neii ] @xmath11 0.4 , while for sy 1 and sy 2 [ neiii]/[neii ] @xmath8 0.4 . in addition ,
sy 1 are biased towards location on the upper right corner .
some of the sy 2 galaxies are located in the sb region , which suggest that these galaxies may have a circumnuclear starburst .
no clear separation between sy 1 and sy 2 is observed for [ neiii]/[neii ] @xmath8 0.4 . in order to test the dependence of the pah bands strength with the radiation field , we compare in fig .
[ ewpahs ] the ew of the 7.7@xmath1 m and 11.3@xmath1 m pah bands with [ neiii]/[neii ] hardness indicator .
it can be seen that ew of both pah bands are constant with the radiation field indicator for [ neiii]/[neii ] @xmath13 0.8 , and fall rapidly with the increasing radiation field .
galaxies , powered by star formation , have a nearly constant ew across the full range of [ neiii]/[neii ] .
while for sy 1 the ews of the pah bands decrease rapidly , for the sy 2 the ew values are distributed either in the sb or sy 1 region .
their location in the diagram may reflect the stellar or agns dominant component contribution to the spectra ( i.e. in some case the sy 2 may be dominated by a non - thermal component , and in other by circumnuclear massive star formation ) . in order to have a more quantitative analysis of fig .
[ ewpahs ] , we have computed an exponential regression taking into account all objects ( agns+sbs , dotted line ) and another for a sub - sample formed only by agns ( solid line ) . for the latter we have found a correlation coefficient of @xmath2 -0.7 while for the former we have not found any correlation ( @xmath2 -0.2 ) .
we conclude this section by arguing that both pah molecules , either ionized ( 7.7@xmath1 m ) or neutral ( 11.3@xmath1 m ) , may be destroyed with increasing hardness of the radiation field .
we emphasize that similar results were found by @xcite , which compares the pah strength with the high to low excitation line ratios for a sample of ulirs ( ultra luminous infrared galaxies ) and agns .
in addition , @xcite has shown that the diagram ew pah 6.2@xmath1 m @xmath3 [ ne v ] 14.3@xmath1m/[ne ii ] 12.8@xmath1 m separate agns from sb galaxies .
we have analyzed the emission line properties of a sample of 186 spitzer galaxy spectra .
171 objects were taken from the literature and we have added 15 unpublished agns spitzer spectra , taken from the public archive .
the continuum properties of the 15 new spectra were also studied .
our main results can be summarized as follows . *
the normalized mid - ir continuum spectra
( at @xmath0=23@xmath1 m ) , of the 15 new agns , steeply rise for redder wavelengths and can be fitted with a warm blackbody distribution of t @xmath2 150 - 300k . * the brightest pah spectral bands ( 6.2 , 7.7 , 8.6 , 11.3 , and 12.7@xmath1 m ) and the forbidden emission lines of [ siii ] 34.8@xmath1 m , [ arii ] 6.9@xmath1 m , [ siii ] 18.7 and 33.4@xmath1 m were detected in all the sb and in @xmath280% of the sy 2 .
taking under consideration only the pah bands at 7.7@xmath1 m , 11.3@xmath1 m , and 12.7@xmath1 m we find they are present in @xmath280% of the seyfert 1 , while only half of this type of activity show the 6.2@xmath1 m and 8.6 @xmath1 m pah bands .
* comparison between observed line intensity ratios of neutral and ionized pahs ( 6.2@xmath1m/7.7@xmath1 m @xmath3 11.3@xmath1m/7.7@xmath1 m ) with theoretical models suggests that these ratios , in sy 1 , sy2 and sb galaxies , are produced by an appropriate mixture of pah molecules , with different sizes and adequate neutral to ionized fraction . * the pah emission lines observed in sy 1 and sy 2 may be due to pah molecules with @xmath8 180 carbon atoms , while for most of the low ionization objects ( sb and liners ) it is biased towards molecules with @xmath11 180 carbon atoms .
in addition , seyfert galaxies appear to be located near to region populated by ionized pah molecules , while the majority of sb , hii , and liner populate the region of neutral pahs
. these results are consistent with a picture where sy 1 and sy 2 have higher ionization fraction and larger pah molecules than sb galaxies .
* we have investigated the pah line intensity ratio diagrams 6.2@xmath1m/11.3@xmath1 m @xmath3 7.7@xmath1m/ 11.3@xmath1 m and 6.2@xmath1m/8.6@xmath1 m @xmath3 7.7@xmath1m/8.6@xmath1 m .
the separation in activity types ( sb , sy 1 and sy2 ) , along the vertical axis in both diagrams , is interpreted as due to the fact that in average the emitting pah molecules in seyferts are larger than those of sb , with ionization fraction and emitting molecules sizes increasing from sb to agns .
no separation between seyfert type is observed .
* diagnostic diagram [ nev]14.3@xmath1m/[neii]12.8@xmath1 m @xmath3 [ neiii]15.5@xmath1m/[neii]12.8@xmath1 m shows that agns are correlated , while sb galaxies show a large scattering .
moreover , for most of the sy 1 [ neiii]/[neii ] @xmath4 0 , while for sb and sy 2s [ neiii]/[neii ] @xmath13 0 , indicating that both line ratios can be used as an indicator of radiation field hardness . *
the ratio between the 7.7@xmath1 m and 11.3@xmath1 m pah bands is nearly constant with the increase of [ neiii]15.5@xmath1m/[neii ] 12.8@xmath1 m , indicating that the fraction of ionized to neutral pah bands does not depend on the hardness of the radiation field . * the equivalent width of both pah features ( 7.7@xmath1 m and 11.3@xmath1 m )
show the same dependence with the [ neiii]/[neii ] ratio . in the case of sy 1 , they are nearly constant for [ neiii]/[neii ] @xmath13 0.8 , and fall sharply with increasing radiation field . for sb galaxies the pah ew is constant with the increase of [ neiii]/[neii ] .
sy 2s are distributed either in the sb or sy 1 regions .
these results suggest that the pah molecules emitting either ionized ( 7.7@xmath1 m ) or neutral ( 11.3@xmath1 m ) bands , may be destroyed with increasing hardness of the radiation field .
we thank an anonymous referee for useful comments , as well as charles bonatto , daniel ruschel - dutra and rogemar andr riffel for useful discussions .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
this work is based on observations made with the spitzer space telescope , which is operated by the jet propulsion laboratory , california institute of technology under a contract with nasa .
allamandola , l. j. , hudgins , d. m. , & sandford , s. a. 1999 , , 511 , l155 baum , s. a. , gallimore , j. f. , odea , c. p. , buchanan , c. l. , noel - storr , j. , axon , d. j. , robinson , a. , elitzur , m. , dorn , m. , & staudaher , s. 2010 , , 710 , 289 bauschlicher , jr . , c. w. , peeters , e. , & allamandola , l. j. 2008 , , 678 , 316 brandl , b. r. , bernard - salas , j. , spoon , h. w. w. , devost , d. , sloan , g. c. , guilles , s. , wu , y. , houck , j. r. , weedman , d. w. , armus , l. , appleton , p. n. , soifer , b. t. , charmandaris , v. , hao , l. , higdon , j. a. , marshall , s. j. , herter , t. l. 2006 , , 653 , 1129 buchanan , c. l. , gallimore , j. f. , odea , c. p. , baum , s. a. , axon , d. j. , robinson , a. , elitzur , m. , & elvis , m. 2006 , , 132 , 401 bernard - salas , j. , spoon , h. w. w. , charmandaris , v. , lebouteiller , v. , farrah , d. , devost , d. , brandl , b. r. , wu , y. , armus , l. , hao , l. , sloan , g. c. , weedman , d. , & houck , j. r. 2009 , , 184 , 230 burtscher , l. , jaffe , w. , raban , d. , meisenheimer , k. , tristram , k. r. w. , & rttgering , h. 2009 , , 705 , l53 deo , r. p. , richards , g. t. , crenshaw , d. m. , & kraemer , s. b. 2009 , , 705 , 14 draine , b. t. , & lee , h. m. 1984 , , 285 , 89 draine , b. t. , & li , a. 2001 , , 551 , 807 draine , b. t. , & li , aigen 2007 , , 657 , 810 galliano , f. 2006 , arxiv : astro - ph/0610852 galliano , f. , madden , s. c. , tielens , a. g. g. m. , peeters , e. , jones , a. p. 2008 , 679 , 310 gallimore , j. f. , yzaguirre , a. , jakoboski , j. , stevenosky , m. j. , axon , d. j. , baum , s. a. , buchanan , c. l. , elitzur , m. , elvis , m. , odea , c. p. , & robinson , a. 2010 , , 187 , 172 genzel , r. , et al .
1998 , , 498 , 579 gordon , k. d. , engelbracht , c. w. , rieke , g. h. , misselt , k. a. , smith , j .- d .
t. , & kennicutt , r. c. , jr .
2008 , , 682 , 336 groves , b. , dopita , m. , & sutherland , r. 2006 , , 458 , 405 houck , j. r. , et al .
2004 , , 154 , 18 hunt , l. k. , thuan , t. x. , izotov , y. i. , & sauvage , m. 2010 , , 712 , 164 kaneda , h. , onaka , t. , sakon , i. , kitayama , t. , okada , y. , & suzuki , t. 2008 , 684 , 270 mathis , j. s. , mezger , p. g. , & panagia , n. 1983 , , 128 , 212 odowd , m. j. , schiminovich , d. , johnson , b. d. , treyer , m. a. , martin , c. d. , wyder , t. k. , charlot , s. , heckman , t. m. , martins , l. p. , seibert , m. , & van der hulst j. m. 2009 , , 705 , 885 osterbrock , donald e. & gary j. ferland , astrophysics of gaseous nebulae and active galactic nuclei , 2nd ed .
( university science books , sausalito , ca , 2006 ) pogge , r. w. , & owen , j. m. 1993 , osu internal report 93 - 01 puget , j. l. , & leger , a. 1989 , , 27 , 161 read , a. m. , & ponman , t. j. 1998 , , 297 , 143 riffel , r. , rodrguez - ardila , a. , & pastoriza , m. g. 2006 , , 457 , 61 riffel , r. , pastoriza , m. g. , rodrguez - ardila , a. , & bonatto , c. 2009 , , 400 , 273 roche , p. f. , aitken , d. k. , smith , c. h. , & ward , m. j. 1991 , , 248 , 606 schutte , w. a. , tielens , a. g. g. m. , & allamandola , l. j. , 1993 , , 415 , 397 smith , j. d. t. , draine b. t. , dale d. a. , et al .
2007 , , 656 , 770 storchi - bergmann , t. , eracleous , m. , ruiz , m. t. , livio , m. , wilson , a. s. , & filippenko , a. v. 1997 , , 489 , 87 sturm , e. , lutz , d. , tran , d. , feuchtgruber , h. , genzel , r. , kunze , d. , moorwood , a. f. m. , & thornley , m. d. 2000 , , 358 , 481 sturm , e. , lutz , d. , verma , a. , netzer , h. , sternberg , a. , moorwood , a. f. m. , oliva , e. , & genzel , r. 2002 , , 393 , 821 thornley , m. d. , schreiber , n. m. f. , lutz , d. , genzel , r. , spoon , h. w. w. , kunze , d. , & sternberg , a. 2000 , , 539 , 641 tielens , a. g. g. m. 2008 , , 46 , 289 voit , g. m. 1992 , , 399 , 495a voit , g. m. 1992 , , 258 , 841b weedman , d. w. , et al .
2005 , , 633 , 706 wu , y. , charmandaris , v. , huang , j. , spinoglio , l. , tommasin , s. 2009 , , 701 , 658 lccc name & ra & dec & _ z _ + & & & + sey 1 & & & + mrk279 & 13h53m03.4s & + 69d18m30s & 0.030451 + mrk334 & 00h03m09.6s & + 21d57m37s & 0.021945 + mrk478 & 14h42m07.4s & + 35d26m23s & 0.079055 + ngc4748 & 12h52m12.4s & -13d24m53s & 0.014630 + sey 2 & & & + mrk3 & 06h15m36.3s & + 71d02m15s & 0.013509 + mrk471 & 14h22m55.4s & + 32d51m03s & 0.034234 + mrk609 & 03h25m25.3s & -06d08m38s & 0.034488 + mrk622 & 08h07m41.0s & + 39d00m15s & 0.023229 + mrk883 & 16h29m52.9s & + 24d26m38s & 0.037496 + mrk1066 & 02h59m58.6s & + 36d49m14s & 0.012025 + ngc1275 & 03h19m48.1s & + 41d30m42s & 0.017559 + ngc2622 & 08h38m10.9s & + 24d53m43s & 0.028624 + ngc3786 & 11h39m42.5s & + 31d54m33s & 0.008933 + ngc5728 & 14h42m23.9s & -17d15m11s & 0.009353 + ngc7682 & 23h29m03.9s & + 03d32m00s & 0.017140 + lcccccccccccccccccccc name & 6.2@xmath1 m & ew & 6.7@xmath1 m & ew & 7.7@xmath1 m & ew & 8.3@xmath1 m & ew & 8.6@xmath1 m & ew & 10.7@xmath1 m & ew & 11.3@xmath1 m & ew & 12.0@xmath1 m & ew & 12.7@xmath1 m & ew & 17.0@xmath1 m & ew + + mrk279 & [email protected] & 0.035 & [email protected] & 0.095 & 15.6@xmath14 1.82 & 0.220 & - & - & [email protected] & 0.001 & 0.84@xmath14 0.12 & 0.013 & 2.94@xmath14 0.36 & 0.049 & - & - & [email protected] & 0.002 & [email protected] & 0.154 + mrk334 & [email protected] & 0.548 & [email protected] & 0.183 & [email protected] & 1.881 & [email protected] & 0.138 & [email protected] & 0.351 & 1.12@xmath14 0.09 & 0.019 & 28.3@xmath14 0.35 & 0.479 & 8.05@xmath14 0.19 & 0.131 & 16.7@xmath14 0.39 & 0.263 & 18.5@xmath14 0.83 & 0.229 + + mrk3 & [email protected] & 0.006 & [email protected] & 0.037 & 38.2 @xmath14 2.98 & 0.315 & - & - & [email protected] & 0.009 & 1.22@xmath14 0.17 & 0.009 & 3.28@xmath14 0.55 & 0.022 & 1.61@xmath14 0.37 & 0.009 & 1.45@xmath14 0.16 & 0.008 & [email protected] & 0.143 + mrk471 & [email protected] & 0.820 & [email protected] & 1.030 & 6.78 @xmath141.15 & 1.415 & - & - & [email protected] & 0.382 & 0.05@xmath14 0.04 & 0.007 & [email protected] & 0.474 & 0.56@xmath14 0.19 & 0.078 & [email protected] & 0.226 & [email protected] & 0.359 + mrk609 & [email protected] & 1.020 & 8.57 @xmath140.49 & 0.485 & 58.0 @xmath141.74 & 3.115 & 4.62@xmath14 0.20 & 0.245 & [email protected] & 0.617 & 0.53@xmath14 0.07 & 0.027 & 17.7@xmath14 0.32 & 0.903 & 5.76@xmath14 0.23 & 0.282 & 10.4@xmath14 0.48 & 0.474 & 16.5@xmath14 0.66 & 0.469 + mrk622 & 5.01 @xmath140.19 & 0.516 & [email protected] & 0.194 & 14.6 @xmath14 1.65 & 0.986 & - & - & [email protected] & 0.093 & [email protected] & 0.015 & 6.45@xmath14 0.28 & 0.253 & 2.40@xmath14 0.18 & 0.089 & 4.41@xmath14 0.41 & 0.154 & [email protected] & 0.044 + mrk883 & [email protected] & 1.100 & [email protected] & 0.920 & 10.7 @xmath14 1.18 & 1.630 & - & - & [email protected] & 0.161 & [email protected] & 0.001 & 3.76@xmath14 0.22 & 0.324 & 0.83@xmath14 0.18 & 0.068 & 1.37@xmath14 0.34 & 0.107 & 2.53@xmath14 0.30 & 0.111 + mrk1066 & [email protected] & 0.757 & [email protected] & 0.263 & 199 @xmath142.76 & 3.567 & [email protected] & 0.340 & [email protected] & 0.592 & 1.29@xmath14 0.11 & 0.020 & 52.1@xmath14 0.43 & 0.713 & 15.3@xmath14 0.23 & 0.181 & [email protected] & 0.328 & 43.0@xmath14 1.41 & 0.325 + ngc1275 & [email protected] & 0.097 & [email protected] & 0.139 & - & - & - & - & - & - & 10.6@xmath14 0.10 & 0.043 & 17.4@xmath14 0.33 & 0.070 & 2.90@xmath14 0.22 & 0.011 & - & - & [email protected] & 0.199 + ngc2622 & [email protected] & 0.233 & [email protected] & 0.339 & - & - & - & - & - & - & 0.32@xmath14 0.05 & 0.015 & [email protected] & 0.135 & 0.78@xmath14 0.14 & 0.034 & 2.03@xmath14 0.13 & 0.086 & 2.42@xmath14 0.46 & 0.0472 + ngc5728 & [email protected] & 0.379 & [email protected] & 0.167 & 65.1 @xmath14 3.00 & 3.050 & [email protected] & 0.245 & [email protected] & 0.424 & 0.04@xmath14 0.21 & 0.001 & 21.4@xmath14 0.65 & 0.783 & 3.97@xmath14 0.35 & 0.117 & 9.89@xmath14 0.78 & 0.258 & 32.6@xmath14 1.14 & 0.471 + ngc7682 & [email protected] & 1.840 & [email protected] & 0.456 & 160 @xmath141.48 & 5.200 & [email protected] & 0.328 & [email protected] & 0.828 & 1.07@xmath14 0.05 & 0.024 & 40.9@xmath14 0.19 & 0.913 & 10.1@xmath14 0.12 & 0.221 & [email protected] & 0.522 & [email protected] & 0.641 + lcccccccccccccccccccc name & [ arii]6.9@xmath1 m & ew & [ ariii]8.9@xmath1 m & ew & [ siv]10.5@xmath1 m & ew & [ neii]12.8@xmath1 m & ew & [ neiii]15.5@xmath1 m & ew & [ siii]18.7@xmath1 m & ew & [ oiv]25.9@xmath1 m & ew & [ feii]25.9@xmath1 m & ew & [ siii]33.4@xmath1 m & ew & [ siii]34.8@xmath1 m & ew + + mrk279 & [email protected] & 0.014 & 0.05@xmath14 0.10 & 0.001 & [email protected] & 0.014 & 0.82@xmath14 0.05 & 0.022 & 1.26 @xmath14 0.07 & 0.039 & 0.64 @xmath140.10 & 0.024 & [email protected] & 0.067 & - & - & - & - & [email protected] & 0.088 + mrk334 & [email protected] & 0.036 & [email protected] & 0.001 & [email protected] & 0.016 & 5.32@xmath14 0.03 & 0.122 & [email protected] & 0.050 & [email protected] & 0.050 & - & - & [email protected] & 0.008 & [email protected] & 0.068 & [email protected] & 0.129 + mrk478 & 0.47 @xmath140.03 & 0.012 & 0.06 @xmath14 0.06 & 0.002 & [email protected] & 0.016 & 0.59 @xmath14 0.04 & 0.032 & 0.52 @xmath140.01 & 0.032 & [email protected] & 0.027 & 0.75 @xmath140.02 & 0.066 & - & - & - @xmath14- & - & 0.01 @xmath140.06 & 0.001 + ngc4748 & 0.45 @xmath140.21 & 0.030 & 0.28 @xmath14 0.12 & 0.018 & [email protected] & 0.144 & 1.14 @xmath14 0.06 & 0.065 & 3.78 @xmath140.12 & 0.123 & [email protected] & 0.058 & 6.81 @xmath140.13 & 0.269 & - & - & 2.69 @xmath14 0.24 & 0.131 & 3.42 @xmath140.29 & 0.167 + + mrk3 & [email protected] & 0.040 & 2.63@xmath14 0.12 & 0.034 & [email protected] & 0.072 & 8.75@xmath14 0.08 & 0.073 & 20.9 @xmath14 0.12 & 0.111 & 6.70 @xmath14 0.11 & 0.037 & [email protected] & 0.159 & - & - & [email protected] & 0.054 & [email protected] & 0.156 + mrk471 & [email protected] & 0.153 & - & - & [email protected] & 0.028 & 0.32@xmath14 0.02 & 0.064 & [email protected] & 0.020 & [email protected] & 0.013 & [email protected] & 0.049 & - & - & [email protected] & 0.285 & [email protected] & 0.331 + mrk609 & [email protected] & 0.083 & - & - & [email protected] & 0.022 & 2.22@xmath14 0.03 & 0.144 & [email protected] & 0.047 & [email protected] & 0.048 & [email protected] & 0.034 & - & - & [email protected] & 0.070 & [email protected] & 0.159 + mrk622 & [email protected] & 0.046 & - & - & [email protected] & 0.021 & 0.70@xmath14 0.03 & 0.036 & [email protected] & 0.019 & [email protected] & 0.009 & [email protected] & 0.020 & - & - & [email protected] & 0.028 & [email protected] & 0.128 + mrk883 & [email protected] & 0.147 & [email protected] & 0.019 & [email protected] & 0.036 & [email protected] & 0.150 & [email protected] & 0.069 & [email protected] & 0.043 & [email protected] & 0.058 & - & - & [email protected] & 0.052 & [email protected] & 0.161 + mrk1066 & [email protected] & 0.096&[email protected] & 0.023 & [email protected] & 0.038 & 9.60@xmath14 0.03 & 0.146 & 4.67 @xmath14 0.09 & 0.055 & [email protected] & 0.050 & [email protected] & 0.031 & 0.25@xmath14 0.29 & 0.002 & [email protected] & 0.035 & [email protected] & 0.108 + ngc1275 & [email protected] & 0.019 & - & - & [email protected] & 0.017 & 3.37@xmath14 0.05 & 0.019 & 4.25 @xmath14 0.14 & 0.023 & 1.77 @xmath14 0.11 & 0.009 & - & - & - & - & [email protected] & 0.001 & [email protected] & 0.096 + ngc3786 & [email protected] & 0.031 & - & - & [email protected] & 0.031 & 1.49@xmath14 0.02 & 0.087 & [email protected] & 0.042 & [email protected] & 0.024 & [email protected] & 0.066 & 0.04 @xmath140.01 & 0.020 & [email protected] & 0.076 & [email protected] & 0.096 + ngc2622 & [email protected] & 0.069 & [email protected] & 0.009 & [email protected] & 0.031 & 0.50@xmath14 0.02 & 0.031 & [email protected] & 0.025 & [email protected] & 0.012 & [email protected] & 0.036 & [email protected] & 0.009 & [email protected] & 0.036 & [email protected] & 0.165 + ngc5728 & [email protected] & 0.061 & [email protected] & 0.035 & [email protected] & 0.197 & 2.26@xmath14 0.06 & 0.084 & [email protected] & 0.162 & [email protected] & 0.081 & [email protected] & 0.244 & - & - & [email protected] & 0.094 & [email protected] & 0.145 + ngc7682 & [email protected] & 0.073 & - & - & [email protected] & 0.029 & 5.10@xmath14 0.02 & 0.164 & [email protected] & 0.109 & [email protected] & 0.093 & [email protected] & 0.077 & - & - & [email protected] & 0.102 & [email protected] & 0.203 + | we study the polycyclic aromatic hydrocarbons ( pah ) bands , ionic emission lines , and mid - infrared continuum properties , in a sample of 171 emission line galaxies taken from the literature plus 15 new active galactic nucleus ( agn ) _ spitzer _ spectra .
we normalize the spectra at @xmath0=23@xmath1 m and grouped them according to the type of nuclear activity .
the continuum shape steeply rises for longer wavelengths and can be fitted with a warm blackbody distribution of _ t _
@xmath2 150300k .
the brightest pah spectral bands ( 6.2 , 7.7 , 8.6 , 11.3 , and 12.7@xmath1 m ) and the forbidden emission lines of [ siii ] 34.8@xmath1 m , [ arii ] 6.9@xmath1 m , [ siii ] 18.7 and 33.4@xmath1 m were detected in all the starbursts and in @xmath280% of the seyfert 2 .
taking under consideration only the pah bands at 7.7@xmath1 m , 11.3@xmath1 m , and 12.7@xmath1 m we find that they are present in @xmath280% of the seyfert 1 , while only half of this type of activity show the 6.2@xmath1 m and 8.6 @xmath1 m pah bands .
the observed intensity ratios for neutral and ionized pahs ( 6.2@xmath1m/7.7@xmath1 m @xmath3 11.3@xmath1m/7.7@xmath1 m ) were compared to theoretical intensity ratios , showing that agns have higher ionization fraction and larger pah molecules ( @xmath4 180 carbon atoms ) than sb galaxies .
the ratio between the ionized ( 7.7@xmath1 m ) and the neutral pah bands ( 8.6@xmath1 m and 11.3@xmath1 m ) are distributed over different ranges for agns and sb galaxies , suggesting that these ratios could depend on the ionization fraction , as well as on the hardness of the radiation field .
the ratio between the 7.7@xmath1 m and 11.3@xmath1 m bands is nearly constant with the increase of [ neiii]15.5@xmath1m/[neii ] 12.8@xmath1 m , indicating that the fraction of ionized to neutral pah bands does not depend on the hardness of the radiation field .
the equivalent width of both pah features show the same dependence ( strongly decreasing ) with [ neiii]/[neii ] , suggesting that the pah molecules , emitting either ionized ( 7.7@xmath1 m ) or neutral ( 11.3@xmath1 m ) bands , may be destroyed with the increase of the hardness of the radiation field . |
core - excited states of na - like ions have two valence electrons outside and one hole inside a closed @xmath3=2 core and , therefore , present an excellent model for studying strong correlations in systems with closely - spaced levels .
we carry out second - order relativistic many - body perturbation theory ( rmbpt ) calculations for na - like ions starting from a @xmath4 dirac - fock ( df ) potential .
all possible @xmath5 holes and @xmath6 particles leading to 121 odd - parity and 116 even - parity @xmath72l^{-1}(j)$ ] states are considered .
we calculate energies of the 237 core - excited states together with the @xmath8 singly - excited states in na - like ions with nuclear charges @xmath2 ranging from 14 to 100 .
many experimental values for the energy levels and fine - structure intervals are now available up to very high nuclear charge ( @xmath2=57 ) for @xmath72l^{-1}(j)$ ] levels ; additionally , experimental rates for some transitions between these levels are available .
the objective of this paper is to present a comprehensive set of calculations for @xmath72l^{-1}(j)$ ] energies to compare with previous calculations and experiments for the entire na isoelectronic sequence .
most earlier measurements and calculations focused on the @xmath92p^{-1}\ ^2p_j$ ] states and low - lying @xmath102p^{-1}\ ^4d_j$ ] and @xmath112p^{-1}\ ^4f_j$ ] levels .
very few results exist for other @xmath12 , @xmath13 , and @xmath14 levels .
the large number of possible transitions have made experimental identification difficult .
experimental verification should become simpler and more reliable using this more accurate and complete set of calculations .
we start with a brief overview of previous theoretical and experimental studies of properties of core - excited states in sodium - like ions .
energies , collision strengths , and oscillator strengths for @xmath15 and @xmath16 transitions in na - like ions with nuclear charges @xmath17 were calculated by @xcite , where energies and mixing coefficients were obtained using the cowan code @xcite with a scaling factor @xmath18 . in the paper by @xcite , auger and radiative transition energies and rates
were calculated for 18 ions with nuclear charges in the range @xmath19 using the multiconfiguration dirac - fock ( mcdf ) method .
wavelengths , radiative transition rates , auger rates , satellite intensity factors , and fluorescence yields were presented by @xcite for dielectronic satellites of 14 neon - like ions ( ar@xmath20 - w@xmath21 ) including @xmath22@xmath23 and @xmath22 -@xmath24 transitions .
the calculations in @xcite were made using the yoda code which is based on multiconfiguration relativistic functions for the bound states and distorted - wave dirac wave functions for continuum electrons .
new comprehensive theoretical analyses of the @xmath25 and @xmath26 ( @xmath3=3 - 8 ) doubly - excited states in fe@xmath27 and cu@xmath28 were presented by @xcite . in these papers ,
energy levels , radiative transitions probabilities , and autoionization rates were calculated by using the cowan and yoda codes .
recently , r - matrix calculations of electron - impact collision strengths for excitation from the inner @xmath29-shell into doubly excited states of fe@xmath27 were presented by @xcite .
energy positions of the @xmath22 and @xmath30 states were calculated in that work using the superstructure code of @xcite .
it was shown in @xcite that disagreement between calculations and data recommended by @xcite and @xcite ranges from 0.5% to 5% .
studies of the @xmath31 configurations in na - like ions are of continuing interest from both theoretical and experimental points of view .
experimentally , these configurations are studied by photon and electron emission spectroscopy .
to our knowledge , the first measurements of @xmath32 and @xmath33 transitions in na - like fe ( which included the @xmath34 configurations ) using photon emission spectroscopy were done by @xcite in @xcite .
the soft x - ray spectra produced by a focused laser source and a vacuum spark device were obtained in @xcite using a grazing incidence spectrograph .
identification of the spectra was based on theoretical calculations using the cowan code .
the @xmath35 @xmath36 - @xmath37 @xmath38 transition in fexvi was observed by the beam - foil technique by @xcite .
the identification of this transition was based on a study of the difference in wave numbers between experimental values and theoretical predictions obtained from the cowan code in na - like ions from s vi to cu xix .
auger electron spectra of na - like fe ions excited in collisions of 170-kev fe xviii on he and ne were studied by @xcite .
the dominant spectral structures were due to auger decay of states with @xmath39 configurations in fexvi .
special attention was paid to the metastable level @xmath35 @xmath36 which was studied in detail ; the absolute energy of this level was measured .
three of the 17 lines observed in this experiment were identified with states from @xmath40 configurations .
this analysis was based on theoretical predictions using the multiconfigurational dirac - fock method including the transverse breit interaction .
a few years later this analysis was extended to include an isoelectronic study of double electron capture in slow ion - atom collisions of f - like ions ( si@xmath41 , ar@xmath42 , sc@xmath43 , ti@xmath44 , fe@xmath45 , and cu@xmath46 ) with he atoms @xcite .
it was reported that the metastable level @xmath35 @xmath36 was observed in all auger spectra except si@xmath47 . recently in @xcite , a study of the double electron capture in low - energy fe@xmath45 + he collisions was conducted .
except for the confirmation of the identification of the 12 peaks in the 225 - 340 ev interval , predictions were made for identification of four additional peaks in the 585 - 670 ev interval .
a comparison of solar and tokamak fexvii x - ray spectra with synthetic spectra including dielectronic satellites with @xmath48 and @xmath49 transitions was presented by @xcite .
atomic data were calculated in @xcite using the superstructure code .
recently , the three fe@xmath27 inner - shell satellite lines were thoroughly studied at the lawrence livermore national laboratory electron beam ion trap ebit - ii by @xcite . in the present paper
, we use rmbpt to determine energies of @xmath30 and @xmath50 states for na - like ions with nuclear charges in the range of @xmath2 = 14100 .
our calculations are carried out to second order in perturbation theory and include second - order coulomb and breit interactions .
corrections for the frequency - dependent breit interaction are taken into account in the lowest order .
screened self - energy and vacuum polarization data given by blundell @xcite are used to determine the qed correction .
our perturbation theory calculations are carried out using single - particle orbitals calculated in the df potential of the ne - like core . as a first step , we determine and store the single - particle contributions to the energies for three @xmath3=2 hole states ( @xmath51 , @xmath52 , and @xmath53 ) and the five @xmath3=3 valence states ( @xmath54 , @xmath55 , @xmath56 , @xmath57 , and @xmath58 ) in lowest , first , and second orders .
next , we evaluate and store the 155 two - particle @xmath59 matrix elements and the 178 hole - particle @xmath60 matrix elements of the effective hamiltonian in the first and second orders .
it should be noted that these one - particle , two - particle , and hole - particle matrix elements were used previously to evaluate energies of the @xmath61 levels in mg - like ions @xcite and energies of the @xmath62 levels in neonlike ions @xcite .
finally , second - order particle - particle - hole matrix elements are evaluated . combining these data using the method described below , we calculate one- , two- , and three - body contributions to the energies of na - like ions .
the present calculations are compared with experimental and predicted results from refs .
the rmbpt formalism developed previously @xcite for b - like and al - like ions is used here to describe the perturbed wave functions and to obtain the second - order energies @xcite of the core - excited states of na - like ions .
the particle - particle - hole contribution to the energies of those states is , however , different from the three - particle contribution to the energies of b - like and al - like ions , and is derived in the appendix .
including hole - particle matrix elements in na - like ions leads to 237 @xmath63 states instead of 148 @xmath64 states of al - like ions or 15 @xmath65 of b - like ions and , consequently , to more laborious numerical calculations .
the model space for core - excited @xmath63 states of na - like ions includes 121 odd - parity states consisting of 25 @xmath66=1/2 states , 37 @xmath66=3/2 states , 31 @xmath66=5/2 states , 19 @xmath66=7/2 states , 7 @xmath66=9/2 states , and two @xmath66=11/2 states . additionally , there are 116 even - parity states consisting of 25 @xmath66=1/2 states , 35 @xmath66=3/2 states , 31 @xmath66=5/2 states , 17 @xmath66=7/2 states , 7 @xmath66=9/2 states , and one @xmath66=11/2 state .
the evaluation of the second - order energies for the @xmath672l^{-1 } ( j)$ ] states in na - like ions follows the pattern of the corresponding calculation for mg - like and ne - like ions given in refs .
in particular , we use the second - order one - particle and two - particle matrix elements for mg - like ions calculated in @xcite and hole - particle matrix elements for ne - like ions calculated in @xcite .
these matrix elements are recoupled as described below to obtain the one - body and two - body contributions for na - like ions .
we will discuss how these matrix elements are combined to obtain the one - body and two - body contributions to energies of na - like ions .
we refer the reader to refs .
@xcite for a discussion of the evaluation of the one - body and two - body matrix elements .
an intrinsic particle - particle - hole diagram also contributes to the second - order energy for na - like ions .
the expression for particle - particle - hole diagram differs from the expression for the three - particle diagram given in ref .
@xcite owing to different angular parts . the expression for the intrinsic particle - particle - hole diagram
is derived in appendix . in fig .
[ e2-d]a , we present the one - body , @xmath68 , two - body , @xmath69 , and three - body , @xmath70 coulomb contributions to the diagonal matrix element with @xmath712p_{1/2}\ ( 1/2)$ ] as functions of @xmath2 . here and in the sequel
, we drop the superscript @xmath72 on the @xmath73 hole state for simplicity .
we see from fig .
[ e2-d]a that the value of @xmath70 is almost constant for all @xmath2 .
the one - body contribution is the largest one .
the sum of @xmath69 and @xmath70 values is almost equal to zero for small-@xmath2 ions ; however , the three - body contribution becomes more important with increasing @xmath2 . as a result ,
the total second - order coulomb contribution @xmath74=@xmath68+@xmath69+@xmath70 differs from the one - body contribution and is almost constant for all @xmath2 . in fig .
[ e2-d]b , we present the one - body , @xmath75 , two - body , @xmath76 , and three - body , @xmath77 breit - coulomb contributions to the diagonal matrix element with @xmath712p_{1/2}\ ( 1/2)$ ] as function of @xmath2 .
the one - body contribution is the largest among other contributions and the total breit second - order contribution @xmath78=@xmath75+@xmath76+@xmath77 is almost equal to the one - body contribution .
the one - body contribution vanishes for non - diagonal matrix elements . in fig .
[ e2-nd]a , we illustrate the contributions @xmath79 and @xmath80 for two non - diagonal elements @xmath81 and @xmath82 with @xmath83=@xmath842p_{1/2}(1/2)$ ] and @xmath85=@xmath862p_{1/2}(1/2)$ ] .
the corresponding breit contributions are plotted in fig .
[ e2-nd]b .
we see from fig .
[ e2-nd]a that the ratio of @xmath80 to @xmath79 is about 3050% for high-@xmath2 ions . fig .
[ e2-nd]a illustrates the asymmetry of the nondiagonal energy matrix elements for low-@xmath2 ions .
the asymmetry of the two - particle second - order matrix elements in rmbpt calculation was discussed in ref .
values of @xmath79 and @xmath80 have different signs and the total second - order contribution @xmath87 is about half of @xmath79 . the behavior of non - diagonal second - order breit contribution is shown in fig . [ e2-nd]b .
the @xmath88 and @xmath89 increase with @xmath2 ; however , they have opposite signs .
the total second - order contribution @xmath90 can be about half of @xmath88 .
finally , we observe some small deviations from smooth @xmath2-dependencies of @xmath91 and @xmath92 in figs .
[ e2-nd]a and [ e2-nd]b . as we can see from eq .
( [ eq10 ] ) in the appendix , the denominator @xmath93 in the expression for the three - body diagram contains two valence energies @xmath94 , @xmath95 , one hole energy @xmath96 , and one virtual state energy @xmath97 .
the sum @xmath98 is positive ; therefore , the denominator can be close to zero when @xmath99 is positive .
in such a case , a singularity arises from the contribution of the continuous part of spectra to the sum over states for the @xmath80 and @xmath89 matrix elements , resulting in sharp features in the corresponding @xmath2-dependencies . after evaluating the energy matrices , we calculate eigenvalues and eigenvectors for states with given values of @xmath66 and parity
. there are two possible methods to carry out the diagonalization : ( a ) diagonalize the sum of zeroth- and first - order matrices , then calculate the second - order contributions using the resulting eigenvectors or ( b ) diagonalize the sum of the zeroth- , first- , and second - order matrices together . following ref .
@xcite , we choose the second method here .
the importance of second - order contributions to the energies is illustrated in fig .
[ contr ] . in this figure , the second - order energy @xmath87 ,
the first- and second - order breit contributions @xmath100 and @xmath90 , and the qed contribution @xmath101 are plotted as functions of nuclear charge @xmath2 for the @xmath102 ( a ) and @xmath103 ( b ) states of na - like ions .
we can see from the figure that @xmath87 dominates up to @xmath2=34 and @xmath2=37 for the @xmath104 and @xmath103 states , respectively . the qed contribution
@xmath101 is smaller than the @xmath87 energy up to @xmath2=50 for the energy level shown in fig .
[ contr]a and @xmath2=60 for the energy level shown in fig . [ contr]b . the second - order breit energy @xmath90 is smaller than the second - order energy @xmath87 up to @xmath2=97 ( fig .
[ contr]a ) and @xmath2=83 ( fig .
[ contr]b ) , respectively . the sharp features in the curves shown in the fig .
[ contr]a are explained by strong mixing of states inside of the odd - parity complex with @xmath66=1/2 .
the main contribution to the @xmath102 level comes from the @xmath842p_{1/2}$ ] state for ions with small @xmath2 up to @xmath2=48 .
starting from @xmath2=49 , the main contribution to this level comes from @xmath1052p_{3/2}$ ] and @xmath1062p_{3/2}$ ] states instead of @xmath842p_{1/2}$ ] state .
the particle - particle - hole levels studied in this work may be divided to two groups by their energies : those with energies in the range of 1523 a.u and those with energies greater than 25 a.u .
the first group of levels includes @xmath842p_j$ ] , @xmath1072p_{j'}$ ] , @xmath1082p_{j''}$ ] , and @xmath1092p_{j'}$ ] levels ( 64 levels total ) .
the second group includes the remaining 173 levels .
the first group of levels has been studied experimentally , while there are no experimental data for the second group . when starting calculations from relativistic df wave functions , it is natural to use @xmath110 designations for uncoupled energy matrix elements ; however , neither @xmath110 nor @xmath111 coupling describes _ physical _ states properly , except for the single - configuration state @xmath1122p_{3/2 } \equiv 3p3d[^3f]2p\ ^3\ !
g_{11/2}$ ] .
both designations are given in our subsequent tables .
the @xmath2-dependence of the eigenvalues of the @xmath1132l''j''(j)$ ] odd - parity states with @xmath66=1/2 is given in figs .
[ f - et]a , b , c , and d , where the energies are divided by @xmath114 . scaled energies @xmath115 decrease with @xmath2 for the first group of levels ( figs .
[ f - et]a , b ) and have parabolic @xmath2 dependence for the second group ( figs .
[ f - et]c , d ) .
it is evident from figs .
[ f - et]a , b that there is a strong mixing between @xmath842p_{1/2}$ ] , @xmath1072p_{j'}$ ] , and @xmath1082p_{j''}$ ] levels .
we already mentioned that strong mixing happens between @xmath842p_{3/2}$ ] , @xmath1052p_{3/2}$ ] and @xmath1062p_{3/2}$ ] states for @xmath2=4849 .
it should be noted that the @xmath111 designations are chosen by comparing results with available experimental data for low-@xmath2 ions .
we used the labeling of levels from refs .
@xcite , where the identification of levels in fe@xmath27 and cu@xmath28 was based on the cowan code .
we already mentioned that there are no measurements for the second group of levels , therefore , we do not give @xmath111 designations for the levels shown in figs .
[ f - et]c , d. the change of relative magnitude of mixing coefficients is illustrated by the behavior of the six levels in fig .
[ f - et]a and the five levels in fig .
[ f - et]b .
the crossing of levels inside of a complex ( set of levels with the same @xmath66 and parity ) is forbidden by wigner - neumann theorem @xcite .
we have calculated energies of the 121 odd - parity states and the 116 even - parity excited states for na - like ions with nuclear charges ranging from @xmath2=14 to 100 . [ cols= " < , > , > , > , > , > , > , > , > , < " , ] in table [ tab - com1 ] , our theoretical results for @xmath116 , @xmath117 , @xmath118 , and @xmath119 levels of na - like fe are compared with predicted auger energies given by @xcite .
it should be noted that the predicted auger energies in ref .
@xcite differ by 1 ev from experimental measurements presented more than 15 years ago by @xcite , where the method of zero - degree auger spectroscopy was used to study the excitation and decay of auger states formed in 170-kev fe@xmath45 ions with he and ne target atoms .
our rmbpt energies are in much better agreement with recent results from @xcite than with experimental measurements from @xcite .
llllll & & + & & & & & + + @xmath842p_{3/2}$ ] & 1.5 & 4.0555 & 4.0547 & 2.6294 & 2.6294 + @xmath842p_{1/2}$ ] & 0.5 & 3.8312 & 3.8309 & 2.4692 & + @xmath1202p_{3/2}$ ] & 1.5 & 3.7423 & 3.7427 & 2.4249 & + @xmath1212p_{1/2}$ ] & 2.5 & 3.5707 & 3.571 & 2.2788 & + @xmath1222p_{1/2}$ ] & 0.5 & 3.5701 & 3.571 & 2.2798 & + @xmath1202p_{1/2}$ ] & 1.5 & 3.5700 & 3.571 & 2.2766 & + @xmath1232s_{1/2}$ ] & 1.5 & 3.4261 & 3.426 & 2.1805 & + @xmath1232s_{1/2}$ ] & 0.5 & 3.4121 & 3.412 & 2.1739 & + in table [ tab - com2 ] , our rmbpt wavelengths for na - like ag and la are compared with experimental results presented by @xcite .
spectra of the @xmath3=3 to @xmath3=2 transitions in neonlike silver obtained from the princeton large torus were recorded with high - resolution bragg - crystal spectrometer .
the measurements covered the wavelength interval 3.34.1 .
eight lines of na - like ag were observed .
as one can see from table [ tab - com2 ] , our theoretical calculations are in excellent agreement with results from ref .
measurement of a single line in na - like la was carried out with high - resolution crystal spectrometer on the plt tokamak by @xcite .
our rmbpt result agrees with experimental wavelengths from @xcite to five significant figures .
we list our theoretical data for other lines in la@xmath124 in table [ tab - com2 ] .
llll & & & + + 15 & 729.74 & & 728.3 + 16 & 613.99 & [email protected] & 614.20 + 17 & 534.12 & [email protected] & 532.67 + 18 & 472.79 & [email protected] & 471.22 + 19 & 424.23 & & 424.05 + 20 & 384.80 & & 384.15 + 21 & 350.86 & & 351.98 + 22 & 325.76 & [email protected] & 324.89 + 23 & 302.61 & & 301.69 + 24 & 281.75 & [email protected] & 281.60 + 25 & 264.32 & & 263.98 + 26 & 248.73 & [email protected] & 248.39 + 27 & 234.76 & & 234.48 + 28 & 222.26 & [email protected] & 221.98 + 29 & 210.93 & [email protected] & 210.68 + 30 & 200.63 & & 200.39 + the @xmath126@xmath127 transitions in p@xmath128-zn@xmath129 ions were investigated by @xcite .
identification of spectral lines originating from the transitions between core - excited configurations in na isoelectronic sequence was made from analyses of beam - foil spectra in @xcite .
wavelengths @xmath130 from @xcite are shown in table [ tab - com3 ] together with our rmbpt data @xmath131 .
we also present the predicted data @xmath132 from ref .
@xcite in the same table .
the predicted wavelengths were obtained in @xcite from the fit of the difference between experimental data and cowan code calculations .
our _ ab initio _ calculations are in good agreement with those predictions .
no measurements of fine - structure intervals were made by observing the wavelength differences between transitions within the doublet or quartet states .
the intervals are obtained by subtracting the excitation energies of the corresponding levels ( see , for example , auger data in table [ tab - com1 ] ) .
these fine structures are quite regular throughout the isoelectronic sequence , as seen from fig .
[ split]a . in this figure
, we present the fine - structure splitting divided by @xmath114 for the doublet term @xmath135 .
small deviation from smooth line of fig .
[ split]a in the region of @xmath2=48 - 49 is due to the strong mixing in the odd - parity complex with @xmath66=1/2 as illustrated in fig .
[ contr]a .
the @xmath136 level has smaller energy than the @xmath102 level in contradiction with land interval rules .
the same situation occurs for other fine - structure intervals in figs .
[ split]b , c , and d. the @xmath137 level has the smallest energy among @xmath138 levels , the next level is @xmath139 , and the @xmath140 level has the largest energy .
the value of @xmath141 decreases with @xmath2 while the value of @xmath142 increases with @xmath2 . as a result
, energies of @xmath140 and @xmath139 become almost equal for high @xmath2 .
the energy difference between the @xmath143 and @xmath144 levels in the @xmath145 quartet term is very small , as shown in fig .
[ split]c .
we find that the @xmath146 quartet terms are ordered as follows : @xmath147 , @xmath36 , @xmath143 , and @xmath144 .
energy differences between the @xmath148 , @xmath149 , and @xmath150 levels in the @xmath151 quartet term are very small , as shown in fig .
[ split]d .
the unusual splitting is caused by mixing of states within the complexes with different @xmath66 .
further experimental confirmation would be very helpful in verifying the correctness of these occasionally sensitive mixing parameters .
a systematic second - order rmbpt study of the energies of the core - excited states of na - like ions has been presented .
rmbpt gives results in good agreement with experimental and predicted data .
it would be beneficial if experimental data for other highly - charged na - like ions were available . at the present time , there are almost no experimental data between @xmath2=47 and @xmath2=100 for the sodium isoelectronic sequence .
availability of such data could lead to an improved understanding of the relative importance of different contributions to the energies of highly - charged ions .
these calculations are presented as a theoretical benchmark for comparison with experiment and theory .
the present calculations could be further improved by including third - order correlation corrections .
the work of w.r.j . and m.s.s .
was supported in part by national science foundation grant no .
phy-01 - 39928 .
u.i.s . acknowledges partial support by grant no .
b516165 from lawrence livermore national laboratory .
the work of j.r.a was performed under the auspices of the u. s. department of energy by the university of california , lawrence livermore national laboratory under contract no . w-7405-eng-48 .
the model space state vector for an ion with two valence electrons ( @xmath152 ) and one hole ( @xmath83 ) can be represented as @xcite : @xmath153 where @xmath154 is the state vector for the core ( @xmath4 in our case ) , @xmath155 describes a particle - particle - hole state with quantum numbers @xmath156 , and intermediate momentum @xmath157 .
we use the notation @xmath158 and @xmath159 . the sum in eq.([eq1 ] ) is over magnetic quantum numbers @xmath160 , @xmath161 , @xmath162 , and @xmath163 .
the quantity @xmath164 is a clebsch - gordan coefficient : @xmath165 the above representation of the state vector is somewhat inconvenient ; for example , it leads to an expression containing 36 terms for the second - order diagram with six free ends .
it was shown in ref .
@xcite that it is more efficient to express the state vector in a manifestly symmetric form in the case of three - electron system . here
, we can only use symmetry of two particles @xmath166 and @xmath167 . as a result
, we obtain the expression for the particle - particle - hole contribution @xmath91 in second order consisting of a sum of @xmath168 , @xmath169 , @xmath170 , and @xmath171 terms : @xmath172 @xmath173aj , x^{0}y^{0}[j_{13}]bj)=-\sum_{vw}\sum_{v^{% \prime } w^{\prime } } p_{j_{12}}(v^{0}v , w^{0}w)p_{j_{13}}(x^{0}v^{\prime } , y^{0}w^{\prime } ) \\ \times ( vw|k_{12})(k_{12}a|k)(v^{\prime } w^{\prime } |k_{13})(k_{13}b|k)\sum_{n}\frac{g(vwan)g(nbv^{\prime } w^{\prime } ) } { % \varepsilon ( v)+\varepsilon ( w)-\varepsilon ( a)-\varepsilon ( n)}\ , , \end{gathered}\ ] ] @xmath174aj , x^{0}y^{0}[j_{13}]bj)=\sum_{vw}\sum_{v^{% \prime } w^{\prime } } p_{j_{12}}(v^{0}v , w^{0}w)p_{j_{13}}(x^{0}v^{\prime } , y^{0}w^{\prime } ) \\ \times ( vw|k_{12})(k_{12}a|k)(v^{\prime } w^{\prime } |k_{13})(k_{13}b|k)\sum_{n}\left [ \frac{g(vww^{\prime } n)g(nbv^{\prime } a)-g(vww^{\prime } n)g(nbav^{\prime } ) } { \varepsilon ( v)+\varepsilon ( w)-\varepsilon ( w^{\prime } ) -\varepsilon ( n)}\right]\,,\end{gathered}\ ] ] @xmath175aj , x^{0}y^{0}[j_{13}]bj)=\sum_{vw}\sum_{v^{% \prime } w^{\prime } } p_{j_{12}}(v^{0}v , w^{0}w)p_{j_{13}}(x^{0}v^{\prime } , y^{0}w^{\prime } ) \\ \times ( vw|k_{12})(k_{12}a|k)(v^{\prime } w^{\prime } |k_{13})(k_{13}b|k)\sum_{n}\left [ \frac{g(vban)g(nwv^{\prime } w^{\prime } ) -g(bvan)g(nwv^{\prime } w^{\prime } ) } { \varepsilon ( v)+\varepsilon ( b)-\varepsilon ( a)-\varepsilon ( n)}\right]\,,\end{gathered}\ ] ] @xmath176aj , x^{0}y^{0}[j_{13}]bj)=\sum_{vw}\sum_{v^{% \prime } w^{\prime } } p_{j_{12}}(v^{0}v , w^{0}w)p_{j_{13}}(x^{0}v^{\prime } , y^{0}w^{\prime } )
\\ = \sum_{vw}\sum_{v^{\prime } w^{\prime } } ( vw|k_{12})(k_{12}a|k)(v^{\prime } w^{\prime } |k_{13})(k_{13}b|k)\sum_{n}\frac{1}{\varepsilon ( w)+\varepsilon ( b)-\varepsilon ( v^{\prime } ) -\varepsilon ( n ) } \\ \times \left [ -g(wbv^{\prime } n)g(nvw^{\prime } a)+g(bwv^{\prime } n)g(nvw^{\prime } a)+g(wbv^{\prime } n)g(nvaw^{\prime } ) -g(bwv^{\prime } n)g(nvaw^{\prime } ) \right]\,.\end{gathered}\ ] ] the quantity @xmath177 is a symmetry coefficient defined by @xmath178 \,,\ ] ] where @xmath179 is a normalization factor given by @xmath180 and @xmath181 is the one - body df energy of the state @xmath182 .
the quantities @xmath183 are clebsch - gordan coefficients and @xmath184 differs from the definition ( [ cg ] ) by the @xmath185 factor and @xmath186 in the @xmath187 coefficient .
the indices @xmath166 , @xmath167 , @xmath188 , and @xmath189 distinguish the valence states , indices @xmath83 and @xmath85 refer to the core states and index @xmath3 refers to excited states .
two - particle matrix element @xmath190 is the sum of the two - particle coulomb and breit matrix elements and is presented as a product of radial and angular parts : @xmath191 radial part @xmath192 of the coulomb matrix element is given by @xmath193 the quantities @xmath194 are normalized spherical harmonics and @xmath195 are slater integrals .
the expression for the breit matrix elements @xmath196 is given in ref .
@xcite . carrying out angular reduction
, we obtain the final expression for the four contributions @xmath197 , @xmath198 , @xmath199 , and @xmath200 to second - order diagram of particle - particle - hole interaction : @xmath201aj , x^{0}y^{0}[j_{13}]bj)=\sum_{vw}\sum_{v^{% \prime } w^{\prime } } p_{j_{12}}(v^{0}v , w^{0}w)p_{j_{13}}(x^{0}v^{\prime } , y^{0}w^{\prime } ) \frac{\sqrt{[j_{12}][j_{13}]}}{[j]}(-1)^{j_{12}+j_{13 } } \\ \times \sum_{n}\sum_{k}\sum_{k^{\prime } } \frac{x_{k}(vwan)x_{k^{\prime } } ( v^{\prime } w^{\prime } bn))}{\varepsilon ( v)+\varepsilon ( w)-\varepsilon ( a)-\varepsilon ( n)}\delta ( j , j_{n})\left\ { \begin{array}{lll } j_{12 } & j_{a } & j_{n } \\ k & j_{w } & j_{v } \end{array } \right\ } \left\ { \begin{array}{lll } j_{13 } & j_{b } & j_{n } \\ k^{\prime } & j_{w^{\prime } } & j_{v^{\prime } } \end{array } \right\ } ( -1)^{j_{a}+j_{b}+j_{w^{\prime } } + j_{w}+k+k^{\prime } } \,,\end{gathered}\ ] ] @xmath174aj , x^{0}y^{0}[j_{13}]bj)=\sum_{vw}\sum_{v^{% \prime } w^{\prime } } p_{j_{12}}(v^{0}v , w^{0}w)p_{j_{13}}(x^{0}v^{\prime } , y^{0}w^{\prime } ) \sqrt{[j_{12}][j_{13 } ] } \\ \times \sum_{n}\sum_{k}\sum_{k^{\prime } } \frac{x_{k^{\prime } } ( vww^{\prime } n)}{\varepsilon ( v)+\varepsilon ( w)-\varepsilon ( w^{\prime } ) -\varepsilon ( n)}\left\ { \begin{array}{lll } j_{w^{\prime } } & j_{n } & j_{12 } \\
j_{w } & j_{v } & k^{\prime } \end{array } \right\ } ( -1)^{1+k+k^{\prime } }
\\ \times \left [ + x_{k}(nbav^{\prime } ) \left\ { \begin{array}{lll } j_{b } & j_{v^{\prime } } & k \\ j_{w^{\prime } } & j & j_{13 } \end{array } \right\ } \left\ { \begin{array}{lll } j_{a } & j_{n } & k \\ j_{w^{\prime } } & j & j_{12 } \end{array } \right\ } ( -1)^{j_{12}+j_{13}+j_{a}+j_{v^{\prime } } + j_{n}+j_{w}}\right . \\
+ x_{k}(nbv^{\prime } a)\left\ { \begin{array}{lll } j_{b } & j_{a } & k \\
j_{12 } & j_{13 } & j \end{array } \right\ } \left\ { \begin{array}{lll } j_{n } & j_{v^{\prime } } & k \\
j_{13 } & j_{12 } & j_{w^{\prime } } \end{array } \right\ } ( -1)^{j_{b}+j_{w^{\prime } } + j_{w}-j}\right]\,,\end{gathered}\ ] ] @xmath175aj , x^{0}y^{0}[j_{13}]bj)=\sum_{vw}\sum_{v^{% \prime } w^{\prime } } p_{j_{12}}(v^{0}v , w^{0}w)p_{j_{13}}(x^{0}v^{\prime } , y^{0}w^{\prime } ) \sqrt{[j_{12}][j_{13 } ] } \\ \times \sum_{n}\sum_{k}\sum_{k^{\prime } } \frac{x_{k^{\prime } } ( v^{\prime } w^{\prime } vn)}{\varepsilon ( v^{\prime } ) + \varepsilon ( w^{\prime } ) -\varepsilon ( v)-\varepsilon ( n)}\left\ { \begin{array}{lll } j_{v } & j_{n } & j_{13 } \\ j_{w^{\prime } } & j_{v^{\prime } } & k^{\prime } \end{array } \right\ } ( -1)^{1+k+k^{\prime } } \\ \times \left [ x_{k}(wban)\left\ { \begin{array}{lll } j_{a } & j_{w } & k \\
j_{v } & j & j_{12 } \end{array } \right\ } \left\ { \begin{array}{lll } j_{b } & j_{n } & k \\
j_{v } & j & j_{13 } \end{array } \right\ } ( -1)^{j_{13}+j_{a}+j_{w^{\prime } } + j_{w}+j_{v}}\right . \\ \left .
+ x_{k}(bwan)\left\ { \begin{array}{lll } j_{b } & j_{a } & k \\
j_{12 } & j_{13 } & j \end{array } \right\ } \left\ { \begin{array}{lll } j_{n } & j_{w } & k \\
j_{12 } & j_{13 } & j_{v } \end{array } \right\ } ( -1)^{j_{b}+j_{w^{\prime } } + j_{n}+j+j_{12}}\right]\ , , \end{gathered}\ ] ] @xmath176aj , x^{0}y^{0}[j_{13}]bj)=\sum_{vw}\sum_{v^{% \prime } w^{\prime } } p_{j_{12}}(v^{0}v , w^{0}w)p_{j_{13}}(x^{0}v^{\prime } , y^{0}w^{\prime } ) \sqrt{[j_{12}][j_{13 } ] } \\ \times \sum_{n}\sum_{kk^{\prime } } \frac{1}{\varepsilon ( w)+\varepsilon ( b)-\varepsilon ( v^{\prime } ) -\varepsilon ( n)}% ( -1)^{1+j_{n}+j_{w}+j_{a}+j_{v^{\prime } } + j_{12}+j_{13 } }
\\ \times \left [ x_{k}(nvw^{\prime } a)x_{k^{\prime } } ( bwv^{\prime } n)\left\ { \begin{array}{lll } j_{a } & j_{v } & k \\
j_{w } & j & j_{12 } \end{array } \right\ } \left\ { \begin{array}{lll } j_{b } & j_{v^{\prime } } & k^{\prime } \\ j_{w^{\prime } } & j & j_{13 } \end{array } \right\ } \left\ { \begin{array}{lll } j_{w } & j_{n } & k^{\prime } \\ j_{w^{\prime } } & j & k \end{array } \right\ } \right . \\ \left .
+ x_{k}(nvw^{\prime } a)x_{k^{\prime } } ( wbv^{\prime } n)\left\ { \begin{array}{lll } j_{a } & j_{v } & k \\
j_{w } & j & j_{12 } \end{array } \right\ } \sum_{\kappa } [ \kappa ] \left\ { \begin{array}{lll } j_{b } & j_{v^{\prime } } & \kappa \\ j_{w^{\prime } } & j & j_{13 } \end{array } \right\ } \left\ { \begin{array}{lll } j_{w } & j_{n } & \kappa \\ j_{w^{\prime } } & j & k \end{array } \right\ } \left\ { \begin{array}{lll } j_{b } & j_{v^{\prime } } & \kappa \\ j_{w } & j_{n } & k^{\prime } \end{array } \right\ } \right . \\
+ x_{k}(nvaw^{\prime } ) x_{k^{\prime } } ( bwv^{\prime } n)\left\ { \begin{array}{lll } j_{b } & j_{v^{\prime } } & k^{\prime } \\ j_{w^{\prime } } & j & j_{13 } \end{array } \right\ } \sum_{\kappa } [ \kappa ] \left\ { \begin{array}{lll } j_{a } & j_{v } & \kappa \\ j_{w } & j & j_{12 } \end{array } \right\ } \left\ { \begin{array}{lll } j_{w^{\prime } } & j_{n } & \kappa \\ j_{w } & j & k^{\prime } \end{array } \right\ } \left\ { \begin{array}{lll } j_{a } & j_{v } & \kappa \\ j_{w^{\prime } } & j_{n } & k \end{array } \right\ } \right . \\ \left .
+ x_{k}(nvaw^{\prime } ) x_{k^{\prime } } ( wbv^{\prime } n)\left\ { \begin{array}{lll } j_{a } & j_{v } & k \\
j_{w } & j & j_{12 } \end{array } \right\ }
\left\ { \begin{array}{lll } j_{b } & j_{v^{\prime } } & k^{\prime } \\ j_{w^{\prime } } & j & j_{13 } \end{array } \right\ } \sum_{\kappa \kappa ^{\prime } } [ \kappa ] [ \kappa ^{\prime } ] \left\ { \begin{array}{lll } j_{a } & j_{v } & \kappa \\ j_{w^{\prime } } & j_{n } & k \end{array } \right\ } \left\ { \begin{array}{lll } j_{b } & j_{v^{\prime } } & \kappa ^{\prime } \\
j_{w } & j_{n } & k^{\prime } \end{array } \right\ } \left\ { \begin{array}{lll } j_{w } & j_{n } & \kappa ^{\prime } \\ j_{w^{\prime } } & j & \kappa \end{array } \right\ } \right]\,.\end{gathered}\ ] ] | energies of @xmath0 and @xmath1 states for sodiumlike ions with @xmath2 = 14100 are evaluated to second order in relativistic many - body perturbation theory starting from a neonlike dirac - fock potential .
second - order coulomb and breit interactions are included .
correction for the frequency - dependence of the breit interaction is taken into account in lowest order . the lamb shift correction to energies
is also included in lowest order .
intrinsic particle - particle - hole contributions to energies are found to be 20 - 30% of the sum of one- and two - body contributions .
comparisons are made with available experimental data .
these calculations are presented as a theoretical benchmark for comparison with experiment and theory . |
collapsed structures identified within the context of dissipationless cosmological simulations are invariably found to exhibit some invariant , or universal , form for their density profiles irrespective of their mass or the epoch at which they are identified ( see , e.g. , and the references therein ) which , moreover , seems to reflect an invariant underlying phase space density distribution ( ) .
these results are potentially of fundamental importance ; they may signal the existence of a generic tendency in initially cold and nearly homogeneous gravitational configurations to evolve towards definite final states , with a consistency reminiscent of that characterising the approach to a gibbs distributions in laboratory systems , irrespective of the details of the initial conditions or the intermediate dynamics .
the existence of invariant final states in systems where the microscopic dynamics is known to be time reversible ( newtonian equations in our case ) inevitably echoes the protracted debate pertaining to irreversibility in laboratory statistical mechanics , the main conundrum being that such ` attractor ' like behaviour is characteristic of dissipative systems . nevertheless ,
as is well known , dissipative dynamics can be effectively introduced in otherwise reversible systems by incorporating a degree of randomness .
this can either be achieved by the presence of a fluctuating force , representing otherwise intractable interactions , or because the intrinsic dynamics is so complicated that tracking its detailed evolution from any given initial condition becomes impossible the accompanying loss of information has consequences similar to the incorporation of random forces . systems where such processes are in action may exhibit the required macroscopic irreversibility that leads to invariant final states .
even if , in the cosmological context , the ` universality ' may be only approximate , the relatively quaint state of affairs that transpires , rare in astrophysical research , has generated intense interest .
discussions dealing with the origin of the identified density distributions can be broadly divided into two main theses along basic nature ( the profiles result from the initial collapse : ) or nurture ( they arise from cumulative effect repeated mergers or interactions : ) lines .
these situations encompass the two contexts alluded to above : irreversible evolution can either arise from the ` violent ' internal evolution involving complicated dynamics of the mean field during the initial collapse or merger events ; or phenomena arising from random encounters between , and dynamical friction forces on , component clumps .
it is the latter category that concerns us here .
the fact that haloes are _ always _ found to have basically the same forms for their profiles suggests that _ both arguments may be relevant _ : for , while numerical calculations simulating the cold collapse of isolated gravitational systems do indeed show that the process can account for halo profiles ( e.g. , hiotelis 2002 ) , a halo s tenure in the context of a cosmological environment will be punctuated by repeated mergers ; it will contain substructure , which will be constantly replenished by continual infall .
the density distribution needs to be invariant under the action of such interactions if the profile is to survive .
evidently , this will be the case if the underlying phase space structure is unaffected by the interactions . in this paper
we endevour to show that this is in fact true ; that the presence of substructure leaves the total phase space mass distribution of haloes with density distributions similar to those of cosmological - simulation - haloes ( at least approximately ) invariant .
what needs to be shown is that when dynamical friction acts on clumps ( representing dark matter substructure ) inside a larger halo , causing them to sink towards the centre , the increased density that results from the mass deposition is precisely balanced by an outflow in the parent halo material ; and that mass stripping from clumps does not change that basic conclusion .
since the evolution necessarily involves macroscopically irreversible behaviour , due to dynamical friction and random clump - clump encounters , the formulation presented in this paper may also shed light on the process of attainment of the universal profile _ via _ substructure interaction .
though this issue is only touched upon briefly here . as in laboratory statistical mechanics , effective randomness , born of the presence of granularity in a system , can be introduced by considering a _ macroscopic _ transport equation that contains a collision term where this randomness is incorporated . in the context of gravitational systems
the usual approximation involves the fokker - planck equation , extensively used in the study of globular clusters for example ( spitzer 1987 ) .
it has also been invoked in the context of cosmological haloes ( ) .
the basic assumption in deriving it from more basic kinetic equations is that the encounters are weak .
this implies that it can not describe the dynamics leading to major mergers between a few large components of roughly equal mass .
the assumption however is well founded when what is involved is the gravitational interaction of a large number of clumps ( representing substructure ) and their motion though a smooth background medium ( representing a parent halo ) .
this is the picture we will have in mind in this paper . in section [ sec : pasim ]
we discuss simple numerical experiments illustrating the phase space dynamics of such a clump - background system . following this , in section [ sec : physical ]
, we tackle the question of correspondence between the almost exact phase space invariance present in these calculations and the persistence of density profiles in simulations of the significantly more sophisticated cosmological context .
we then outline the fokker - planck formulation used in this paper ( section [ sec : fp ] ) and apply it to pure power law systems ( section [ sec : scalefree ] ) .
section [ sec : times ] discusses the evolution timescales on which the total mass distribution changes , relative to the dynamical friction timescale it takes for the clumps to sink in .
these are then applied to the power law solutions in section [ sec : timescales ] . in section [ sec : instab ] , the evolution of the _ individual _ components is considered in more detail .
the consequences of the solutions obtained are outlined in several sections that follow . in particular
, section [ sec : core ] discusses the formation of a constant density core .
this occurs in the background material if no stripping is invoked .
the effect of stripping is dealt with in section [ sec : strip ] .
both configurations correspond to approximate equilibria , with the ones involving stripping longer lived .
technical background material is outlined in several appendices .
its inclusion was found necessary because of the variety of conventions , terminology and notation used in the literature of the fokker - planck equation as applied to gravitational systems .
moreover , to the author s knowledge , derivations of some of the results required for this paper have only been been published in the ( french language ) paper of henon ( 1961 ) .
alternative derivations are presented .
[ fig : sim ] shows the density and velocity dispersion distributions of systems of massive clumps and smooth background .
the simulations are identical to one of the runs presented in el - zant et al .
the clumps are represented by softened point particles whose ` size ' is about @xmath4 of the scale length of the navarro , frenk & white ( 1997 ) initial density distribution .
there are 900 000 background ( ` light ' ) particles through which moves a system of 900 clumps with combined mass representing @xmath5 the system s total .
they are all started from the same initial density and velocity distributions .
the system is spherically symmetric with isotropic velocities , and is evolved using a polar particle mesh code .
as already noted by el - zant et al ( 2004 ) , the total density distribution of the clump - background system remains virtually unchanged for many dynamical times ( given in terms of the average density inside the initial halo scale length by @xmath6 ) ; it is identical , except perhaps at the very centre and up to statistical noise , to the passively evolved system containing no clumps .
this statement is also true at all times smaller than the timescale shown in the figure ( the invariance breaks down completely on timescales almost an order of magnitude larger , for reasons we discuss in the next subsection ) .
we have tried different values of the nfw concentration parameter @xmath7 .
the results presented in this figure refer to a system with @xmath8 .
we ran , in addition , simulations with of @xmath9 and @xmath10 , with almost identical results .
different heavy particle masses and fractions were also investigated . as long as the problem is rescaled in terms of the relevant dynamical and dynamical - friction ( see next subsection below ) timescales , the situation described by fig .
[ fig : sim ] remains for time intervals of the order of those shown . in el - zant et al ( 2004 )
we had interpreted the clump distribution to be composed of baryonic material . in that case , the conclusions outlined in that paper still stand : the dark matter , made of lighter background particles , is heated and forms a constant density core ; but the total distribution ( interpreted to be composed of baryons and dark matter ) remains constant . however
, if the clumps are also made of dark matter , this invariance implies that the total distribution of dark matter remains constant . a rather remarkable result , transpiring from fig . [ fig : sim ] ,
is that it is not only the physical density that remains unchanged , but also the velocity dispersion .
the invariance of the velocity distribution is all the more remarkable because the initial ` temperature inversion ' the fact that the velocity dispersion decreases towards the centre is a thermodynamically unstable state of affairs , that should be washed out under the action of energy transfer ; the system then evolving towards an isothermal distribution , as heat flows from hotter to colder regions .
it so happens , nevertheless , that the ` heating ' of the lighter particles is precisely balanced by a corresponding ` cooling ' in the clump distribution , so as to keep the total ` temperature ' constant .
the constancy of the mass and velocity dispersion distributions in a spherical isotropic system , where the distribution function is dependent only on energy , must be associated with an invariant phase space structure . in section [ sec : physical ] we argue that _ this state of affairs , so stark in this idealised setting , may remain relevant , resulting in the observed persistence of universal profiles , in the highly more complex cosmological context . _
the remainder of the paper is an attempt to explain and derive the simulations presented here , and to quantitatively justify this basic contention . the state of affairs described by fig .
[ fig : sim ] is metastable ; in the sense it is long - lived , but not infinitely so .
this should be obvious , since once the massive particles form a self gravitating component , relaxation resulting from their mutual interactions will result in evolution . for timescales of order @xmath11
, we have verified that the center expands , as energy is transferred from the outer hotter regions .
this is in accordance to what was found , for example , by hayashi et al .
although we have not performed integrations for longer intervals , it is expected that it would then proceed towards core collapse .
the notion of solid clumps can not possibly be useful however in either of these regimes ; the effect of stripping , which will transform clump material into background , ensures that the former can never form an autonomous dynamical system ( cf . , section [ sec : physical ] ) .
we will see later in this paper ( section [ sec : times ] and section [ sec : instab ] ) , that the characteristic timescale for the evolution that significantly modifies the distribution of the ( light and heavy ) components , but keeps the total virtually invariant , is the local dynamical friction time
. it will be useful therefore to parametrise the timescale of the simulation just analysed in terms of that quantity . to get a rough estimate of the relation between the time the system is evolved in fig .
[ fig : sim ] and the average dynamical friction time within a given radius , we write the latter as ( e.g. , el - zant , shlosman & hoffman 2001 ) .
it also assumes that the velocity of the clumps is comparable to the dispersion of the background , which has a maxwellian velocity distribution . ]
@xmath12 where @xmath13 is of the total system mass inside radius @xmath14 divided by the mass of a clump and @xmath15 is the coulomb logarithm .
we denote this timescale with the label ` sim ' to distinguish it from an analogous timescale @xmath16 that we derive later , in a more rigorous manner , from diffusion coefficients ( section [ sec : times ] ) . for a system with significant softening ,
as is the case here , @xmath17 is determined by the softening length .
the maximum impact parameter can be taken as the radius inside which we are calculating the average dynamical friction timescale .
we plot in fig .
[ fig : dfsim ] the dynamical friction time , relative to the simulated timescale , for an nfw system with @xmath18 .
note that the dynamical friction timescale is smaller than the simulated one only in the very inner region , and then only by a factor of a few .
( it increases again , because the softening dominates .
probably by then however , the formula above becomes entirely inadequate , the dynamics being dominated by a non - newtonian contributions to the softened potential ) .
thus , in terms of dynamical friction time , the system is fairly young , despite it being of significant dynamical age ( parametrised in terms of @xmath19 ) . to explain the results of the simulation discussed in section [ sec : sim ]
therefore , it suffices to show that the evolution timescales characterising the change in total phase space mass density of a system with central density cusp are much longer than the dynamical friction time . noting also that most of the evolution in the lighter component in fig .
[ fig : sim ] takes place deep inside the @xmath20 cusp , it should be possible , for the purpose of explaining the aforementioned results , to approximate the system as a scale free configuration .
there are two glaring idealisations pertaining to the calculations presented in the previous section : the heavier bodies are solid softened point particles ; they also all have the same mass .
the advantage , of course , is that the simplified models serve as to highlight and isolate the interesting phenomenon involving the invariance of the phase space distribution and identify its origin in the dynamical friction interaction between a distinguishable background and a system of massive objects , the number of which is large enough so that statistical reasoning may be relevant . but how do these toy models relate to the actual situation found in realistic simulations of halo evolution in the context of cdm cosmologies ? in other words , would the discovered invariance survive , at least in some approximate sense , the removal of the idealisations ? in the following we attempt to make this connection the arguments are only suggestive ; though most are broadly supported by the analysis based on the fokker - planck formulation presented in subsequent sections of the paper , future detailed numerical investigation should also be undertaken .
subhaloes found in cosmological simulations do not all have the same mass .
indeed , their differential number distribution follows the rather steep mass function @xmath21 ( e.g. , gao et al 2004c ) .
the mass is also a deciding factor as to how strongly a body will interact with a background the rate of energy lost by that object varying as @xmath22 .
it would thus appear that in the assumption of equal masses lies a drastic approximation , compromising the description of the evolution .
nevertheless , it may precisely be the steep mass spectrum , and the sensitive dependence of the dynamical friction on the mass , that ensure that the situation described by the single ( clump ) mass model survives . for , by virtue of these properties , a separation of timescales will ensue ; the heavier mass clumps sinking in first , while the lower mass ones are hardly affected by dynamical friction .
if clump - clump interaction is not of major importance , this initial evolution will mimic that of the single mass model ( clump - clump interaction will cause the dynamical evolution of small mass clumps to follow the background material ) . the total distribution of the system just described is also an equilibrium .
formally , we show in section [ sec : fp ] , that the fundamental equation deciding the presence of an equilibrium is linear with respect to the distribution of massive objects : if a set of distributions , possibly corresponding to different mass species , is a zero of that equation , so is their sum ( or any linear combination thereof ) .
thus , one can superpose an equilibrium distribution made of the evolved system , composed of the heaviest clumps plus background , and the distribution of the remaining ( non - participating ) clumps , which will remain unchanged if clump - clump interactions can be neglected ( or , if such interactions are included , to various distributions of the remaining clumps ) .
there is a caveat .
detailed invariance will require that the clumps representing substructure be initially distributed in a similar manner to the background .
but the viability of a statistical description will be severely affected by the relatively small number of massive objects in each mass range .
in particular , the most massive bodies , which initially contribute most , will be represented only by a few specimen .
the invariance therefore would only hold in an ensemble averaged sense .
ma & boylan - kolchin ( 2004 ) found that substructure may , in individual runs increase or decrease the steepness of density profiles ; and cosmological simulations do indeed show large scatter in the the profiles of identified haloes ( e.g. , klypin et al .
. more work , involving a series of simulations representing a statistical sample of initial sustructure distribution and concentrations , in the controlled context of isolated halo calculations , is in order . if the clumps come to completely dominate the mass distribution at small radii ( as happens to the evolved system in section [ sec : sim ] ) , then the effect of dynamical friction will be limited to the most massive specimen by the time , at any given energy , the dynamical friction coupling is strong enough to affect the next group of clumps , most of the background particles that these clumps should couple with would have been removed .
one is then left with a region that can be best described as a system of self gravitating clumps , with fluctuating and dissipative forces roughly balancing each other , but where energy transfer ( ` relaxation ' ) leads to eventual evolution ( cf .
sections [ sec : theinvariance ] and [ sec : core ] ) .
it is quite likely however that stripping would be effective in transforming much of the substructure into background , before such a situation transpires .
this represents another form of cutoff for the dynamical friction - mediated coupling .
two question are clearly pertinent .
they are concerned with the timescale for stripping and its effect on the total phase space mass distribution function .
satellites can no longer be regarded as solid clumps if any of the following processes are efficient 1 .
particles are removed from the satellites due to pruning by the background ( this includes the possibility of complete dissolution ) .
particles are removed due to weak encounters with other satellites .
satellites participate in strong mutual encounters or merge .
the first of these mechanisms does not , to a very good approximation , change the total phase space mass distribution function : particles are gently removed from a satellite ; they leave with approximately the same velocity and position as the satellite .
thus , material with virtually the same location in phase space is transferred from clump to background , the total phase space mass density remaining nearly constant .
this will also be true of the second mechanism .
weak encounters will lead to the gentle removal of particles , they may also facilitate stripping by the main halo ( cf . pennerubia & benson 2004 ) , both processes through which the total phase space mass distribution is closely conserved .
the situation is different however if violent encounters and mergers are involved .
it is possible to estimate the timescale associated with such events by averaging the clump distribution inside a given radius and invoking an effective collision crossection .
the envisioned ` collisions ' will not necessarily lead to mergers ( this will depend on their orbital parameters etc ) , but we will assume that when two satellites so interact , particles in their outer regions can be removed with significantly modified phase space coordinates .. the collision mean free time for a randomly moving system of @xmath23 spheres of radius @xmath24 , confined within a spherical volume of radius @xmath14 , can be written as ( e.g. , saslaw 1985 ) @xmath25 where @xmath26 , is the time for an object to make one full crossing of the system .
clumps of different masses will have different diameters , but one can define an effective crossection @xmath27 let the differential mass function @xmath28 , and the characteristic radius of a clump be @xmath29 , where the parameter @xmath30 relates the virial radius and mass ( it depends on the redshift and the cosmology ; cf .
a1 of nfw ) and @xmath31 is the typical fraction of the virial radius of a clump that survives stripping by means of mechanisms i ) and ii ) in the list above . in this context we can write @xmath32 where we have assumed that the maximal mass in the clump distribution is at least a few orders of magnitude larger than the minimal one . confider a radius @xmath33 , inside a host halo of virial mass @xmath34 , and containing @xmath35 satellites .
we can rewrite eq ( [ eq : freet ] ) in these terms , replacing @xmath36 by the effective crossection @xmath37 , @xmath38 if @xmath39 , @xmath40 ( e.g. , if @xmath41 is of order @xmath42 and @xmath43 ) .
this represents an average collisional timescale inside the virial radius .
it is necessarily a lower limit , because it effectively assumes that stripping by the host halo , or _ via _
weak encounters , is negligible ( hence @xmath44 ) . as one averages over smaller radii ( that is @xmath45 ) ,
our estimate of @xmath46 will depend on how the ratio @xmath47 changes . a simple single power law that can be taken as a reasonable representation of the density distribution of host and satellites is @xmath48 ( we discuss this further in section [ sec : scalefree ] ) .
in this case , it is easy to show ( e.g. , el - zant , kim & kamionkowski 2004 ) that the tidal radius of a satellite is proportional to its radial position inside the host halo that is @xmath49 .
since @xmath50 necessarily decreases with @xmath14 , this suggests that the above estimate at the virial radius is indeed a lower limit not only there , but at all radii . during a time interval @xmath46
, substructure continues to be radically modified due to stripping by the host halo ; as can be seen from fig .
11 of taylor & babul ( 2004 ) for example , where for almost all haloes , well over @xmath51 of satellite mass is lost on such a timescale ( note that the radial period used there @xmath52 ; and that the authors do not consider clump - clump interactions at all .
not even weak encounters ) .
it therefore appears unlikely that strong encounters or mergers will be important at any stage during the evolution haloes being affected by gentle stripping on a significantly shorter timescale .
finally , note that only a substructure satellite for which @xmath53 will be affected by dynamical friction before most of its mass is lost _ via _ stripping . from eq .
( [ eq : dfsim ] ) it should have @xmath54 . thus stripping eventually puts an effective end to the dynamical friction coupling .
the discussion of the present section suggests that a situation whereas sinking clumps would come to dominate the mass distribution within a certain region ( as in fig .
[ fig : sim ] ) probably would not , in practice , arise satellites having been significantly stripped long before such a configuration materialises but that , nevertheless , the invariance found there will remain if 1 . stripping is due to gentle removal of particles from clumps due to their interaction with background or soft mutual encounters .
2 . substructure distributions , properly modified to take into account the effect of stripping , also constitute ( at least approximate ) equilibria .
the estimates of the mean free time presented here suggest that the first condition is satisfied . in section [ sec : strip ] we show that distributions satisfying condition 2 are indeed excellent approximate equilibria .
in fact , as may already be expected from the discussion above , such configurations are much longer lived , as they do not suffer from the problem of relaxation , characteristic of a system composed of self gravitating clumps .
the fokker - planck equation has been discussed in the context of gravitational systems in several textbooks ( saslaw 1985 ; binney & tremaine 1987 ; sptizer 1987 ) .
the basic assumption is that the perturbations due to the granularity in a system are weak , local and random .
the situation is then analogous to that of brownian motion in a laboratory setting .
the dynamical variables therefore undergo a random walk , the amplitude of which is limited by a deterministic force .
chandrasekhar ( 1943 ) reviewed these phenomena , both in the case of laboratory systems , where the approximations involved are much more easily justifiable , and the gravitational case ; for which he derived an explicit formula for the deterministic force , that he dubbed dynamical friction .
it proved remarkably successful in the face of numerical tests , despite the problems associated with justifying the assumptions of locality and randomness ( e.g. zaritsky & white 1988 ) .
the ratio of the amplitude of the fluctuating force causing the random walk to that of dynamical friction is determined by the mass of the object in question relative to that of the perturbers in the background .
this implies that more massive particles will typically have smaller velocities ; in the gravitational context , they will sink to the centre of the system .
the timescale for this to happen we call the dynamical friction time .
( [ eq : dfsim ] ) represents an estimate derived on the basis of the chandrasekhar formula alluded to above .
other evaluations can be directly deduced from the fokker - planck formulation ( sec .
[ sec : times ] ) .
the energy deposited among the population of lighter particles causes their velocities to increase and their distribution to expand .
the principal object of this paper is to show that these processes leave the total phase space mass density invariant for timescales significantly longer than the dynamical friction time ( so as to be negligible on timescales of that order ) . in physical space
, that would correspond to the expansion of the lighter particle system being almost exactly compensated by the inflow of massive clumps .
provided that the dynamical friction time is significantly longer than the orbital timescale ( @xmath55 ) , the fokker - planck equation can be averaged over particle trajectories , and expressed in terms of the integrals of motion in the smoothed out potential .
the orbit averaged fokker - planck equation and its properties have been elucidated in detail by henon ( 1961 ) , and discussed in binney & tremaine ( 1987 ) and spitzer ( 1987 ) .
other works include that of kohn ( 1979 , 1980 ) and merritt ( 1983 ) , the latter being particularly relevant to the situation at hand here .
in the case when the configuration is spherical and has isotopic velocity distribution , as we will assume here , the six dimensional phase space is also spherically symmetric .
the relevant integral of motion is the energy _ per unit mass _ e , which defines a radial coordinate in that space , in terms of which the distribution function and the fokker - planck equation may be written . in this paper
refers to the energy , and not to the binding energy , which has opposite sign , favoured in some of the aforementioned work . ] we will not be using the fokker planck equation directly here , but instead relations for the change in mass and energy , within a given surface @xmath56 , that can be readily derived from it ( e.g. , henon 1961 and appendices [ app : mass ] and [ app : energy ] of the present paper ) .
the advantage of this approach is twofolds .
first , the fokker - planck equation itself allows for solutions with non - vanishing constant mass flux . these are steady state solutions , even though they are unphysical , unless there is something to produce or absorb particles at the center of the system , a possibility we do not consider here .
the second is that steady state solutions of the fokker planck equation do not guarantee thermodynamic stability , we will need to estimate the effect of energy transfer on the steady state ( cf .
, section [ sec : energychange ] ) . for the mass evolution we have ( appendix [ app : mass ] ) @xmath57 where @xmath58 is the phase space mass density distribution function that is , the amount of mass contained in a volume element @xmath59 is @xmath60 .
this equation simply says that the rate of mass change inside the energy surface @xmath61 is determined by the mass flux @xmath62 that is , the amount of mass , per unit time , that enters ( negative sign ) or exists the phase space volume bounded by the energy surface @xmath61 and the mass acquired or lost at the edge of this volume because of its variation with time .
that phase space volume is related to the potential ; since , in a spherical phase space , @xmath63 and @xmath64 ( here @xmath65 is the maximum radius out to which a particle with specific energy @xmath61 can travel ) .
the ` density of states ' @xmath66 , the area _ in phase space _ of the surface with energy @xmath61 , relates to the volume inside that surface by @xmath67 .
accordingly , the mass enclosed within an energy interval @xmath68 $ ] is @xmath69 ; and the total mass within the surface at @xmath61 : @xmath70 note that , even when the fokker plank flux @xmath62 is zero , the mass distribution may be time dependent .
this state of affairs is characteristic , for example , of smooth ( collisionless ) systems in configurations out of dynamical equilibrium . in this paper
we assume that we are dealing not with the merger of a few clumps of comparable mass , but with a large number of them and a smooth background , with the combined system having reached a slowly evolving quasi - equilibrium state .
any evolution must therefore arise from the flux @xmath62 describing the collisional evolution .
we will therefore ignore evolution driven solely by the second term on the right hand side of ( [ eq : fok ] ) .
comparison of the isotropic fokker planck equation in flux conservation form ( eq .
[ eq : fpfc ] ) and its standard format ( e.g. , binney & tremaine 1987 ; spitzer 1987 ) yields @xmath71 from which follows the equation for the flux @xmath72 where @xmath73 and @xmath74 and where @xmath75 and @xmath76 are given by eq .
( [ eq : d1 ] ) and ( [ eq : d2 ] ) .
the fokker - planck flux is due to the deterministic and random forces , acting on the components of the system due to their mutual interactions ( as discussed in the last subsection ) .
the changes in energies that result from these forces are represented by the diffusion coefficients @xmath77 and @xmath78 respectively .
it is possible to heuristically describe how their forms , and the accompanying expression for the flux , arise . when ( [ eq : de ] ) and ( [ eq : dee ] ) are inserted into ( [ eq : fluxp ] ) , this expression is seen to consist of three components .
the first term , due to the component @xmath79 of ( [ eq : fluxp ] ) , i.e. @xmath80 , refers to a shell in phase space of volume @xmath81 . on average , due to systematic change in energy @xmath82 , particles will traverse such a shell during a unit time interval .
that is , particles crossing a surface at @xmath56 in a unit time interval , b will come from this shell .
the mass inside that shell is the volume multiplied by the phase space mass density .
hence the expression @xmath80 .
the random fluctuations characterised by @xmath83 , as opposed to their weighed gradients discussed below , can not produce a particle flux unless the distribution function has a gradient because particles will be as likely to enter or leave energy surface @xmath61 . the rms energy space
` length ' traversed in unit time by a typical particle due to random fluctuation is @xmath84 .
consider a constant energy surface centered on the interval @xmath85 $ ] .
the volume of each of the two shells bounded by @xmath61 and surfaces at the extremities of this interval is @xmath86 ; the difference between their average phase space mass density is @xmath87 ; and so the difference in mass they contain is @xmath88 .
since the walk in energy space described by @xmath83 is random , from each of those shells , on average , half the mass would transit in unit time through @xmath61 ( the other half crossing to still higher energies for the outer shell , or smaller energies in the case of the inner one ) .
the magnitude of the resulting mass flux through @xmath61 is then @xmath89 .
it is directed outwards ( in our convention positive ) when the phase space mass density is a decreasing function of energy hence the term @xmath90 in ( [ eq : fluxp ] ) .
systematic changes in energy can arise due to the deterministic dynamical friction forces ; or because the random fluctuations , weighed by the phase space available at energy @xmath61 , @xmath91 , are energy dependent .
is energy independent ; simply because , while particles will be as likely to cross energy surface @xmath61 from above or below , if @xmath66 has a gradient , there will be more particles at energy intervals @xmath92 $ ] corresponding to larger @xmath66 . ]
when this latter component is subtracted from total systematic energy change @xmath93 , the result represents flux whose sole source is dynamical friction that is , the term @xmath79 in ( [ eq : fluxp ] ) . in such circumstances ,
the average rates of change of the energy and its square are given by ( [ eq : deltemu ] ) , ( [ eq : delte ] ) , and ( [ eq : deltee ] ) .
it is then straightforward to show that @xmath95 for the light particles . for the heavy particles , it consists of the second term of eq .
( [ eq : delte ] ) .
contains an extra mass factor as compared with that adopted by merritt 1983 , who prefers to multiply the term @xmath77 in eq .
( [ eq : fluxp ] ) by the mass . ]
one can also see that @xmath78 is the same for both species , and depends only on the distribution of the heavy masses .
this is to be expected .
dynamical friction on the lighter particles is negligible ; they can also be considered non - interacting , their evolution in energy space being determined by the random energy they , on average , gain from the massive particles . for , if these heavier particles are removed , the remaining system , composed of light particles , is collisionless .
the flux in light ( background ) particles is then @xmath96 for the massive particles , on the other hand , we have @xmath97 adding the two equations we get @xmath98 by inserting the values of @xmath77 and @xmath78 into ( [ eq : flux ] ) we obtain what we will call the fundamental equation for the flux @xmath99 where we have assumed that the zero point of the potential is so chosen so that particles have positive energies ( @xmath100 ) . the flux is a radial vector in an energy - parametrised spherical phase space . in the context of the conventions adopted here , positive flux , pointing outwards
, corresponds to particle transport towards higher energies .
another equation that will be useful is the flux of background material only .
we have ( by using [ eq : flight ] ) @xmath101 which does not have a dynamical friction term , and thus invariably corresponds to mass outflow into higher energies levels . now note that _ eq . ( [ eq : fun ] ) is linear in @xmath102_. that is , if a set of distribution functions @xmath103 are solutions , any function @xmath104 is also a solution
. therefore one can construct solutions for the various components of a multimass system and add them up , assigning arbitrary weights , possibly derived from assumptions concerning the effects of mass segregation and stripping ( cf .
section [ sec : strip ] ) . thus our focus on two component systems
does not lead to any fundamental limitation from the point of view of finding equilibrium solutions .
this remarkable property is a derivative of the fact that the source term for the dynamical friction depends only on the total mass distribution ; while the heating , though originating only from the massive clumps , affects all particles .
note , nevertheless , that the evolution of components of different mass will not be the same , since the dynamical friction flux for _ each _ individual heavy species @xmath105 would be @xmath106 while the flux due to the fluctuating force on members of that species is @xmath107 their ratio is @xmath108 .
heavier clumps will therefore lose energy to the lighter ones . from this
it follows that the linearity of the fundamental equation does not imply that the evolution of the system can be decomposed into the sum of the evolution of each species plus background separately !
what it does imply is that , if each distribution corresponding to ( e.g ) distinct mass species happens to be an equilibrium solution for a given total mass distribution , the combined system will also be in equilibrium .
the amount of energy entering the energy surface @xmath61 per unit time is @xmath109 this equation is derived in appendix [ app : energy ] ( another derivation can be found in henon 1961 ; cf .
his eq 2.27 , 2.28 and 4.27 ) refers to our @xmath62 and @xmath62 refers to the distribution function .
he also uses the term ` fundamental equation ' to describe the fokker - planck equation instead of the flux equation , as the term is used here . ] .
the first term of ( [ eq : foken ] ) refers to the energy carried by the mass flux , while the second represents nonlocal contributions .
the last two terms correspond to the variation in both these components due to the change in phase space volume , which results from change in the potential .
they are analogous to the second term on the right hand side of eq .
( [ eq : fok ] ) .
this form takes into account the fact that particles entering into energy surface @xmath61 from higher energy levels , at any given time , do not end up at the same energy @xmath110 .
neither do they all originate from the same initial energy level @xmath111 .
a distribution in @xmath112 and @xmath113 is always present ( even if excessively large jumps are unlikely , because of the small deflection approximation assumed in deriving the diffusion coefficients ) .
the same goes for particles exiting energy surface @xmath61 to higher energies .
thus , even in the absence of a net mass flux through a given surface @xmath61 , energy can still be carried through it .
this can lead to evolution . because of the state of affairs just outlined
, the evolution driven by the energy flux will take the form of ` heat ' transfer from ` hotter ' regions of the system to those that are ` colder ' .
when a temperature gradient exists , statistically , particles crossing @xmath61 from above will end up at energies which , though smaller than @xmath61 , are larger than the compensating particles that crossed from below @xmath61 to exit towards higher energies again , even if the mass flux vanished across @xmath61 .
this is why thermodynamic equilibrium , for a system of solid single mass objects , can only be isothermal ; and all open gravitational systems in virial equilibrium , so composed , are thermodynamically unstable ( e.g. , saslaw , 1985 , 2000 ; padmanabhan 1990 ; el - zant 1998 ) .
nevertheless , we will show that in the case of the two component systems considered here , the evolution timescales associated with the energy flux can be quite long ; and that , under certain circumstances which may be associated with stripping , even exact equilibria that are not isothermal can exist .
in this section we consider the case in which both the clump and the background distribution functions are power laws in the energy .
the corresponding physical densities are also pure power laws
. perfect power law densities @xmath0 with @xmath114 are relevant representations in the inner regions of haloes , where the dynamical friction coupling is most efficient .
densities with indices @xmath115 are useful approximate forms at intermediate radii , up to the virial radius in haloes of small concentrations ( as discussed in the penultimate and final paragraphs of section [ sec : masssol ] ) .
we generally have in mind a range @xmath116 , although the case of @xmath117 , corresponding to the outer regions of haloes is briefly considered in section [ sec : steep ] . strictly speaking the case of @xmath118
requires separate treatment , because the potential is logarithmic in the radius .
nevertheless , the flux associated with profiles having @xmath119 continuously tends to the value at @xmath118 ( namely equilibrium ) .
so will skip such separate treatment for the sake of brevity .
[ sec : scalefree ] if the physical density can approximated by a power law such that @xmath120 the poisson equation implies a potential is expressed as @xmath121 with @xmath122 ) .
the distribution function that is an equilibrium of the unperturbed collisionless system takes the form ( cf . ,
evans 1994 ; evans & collett 1997 ) @xmath123 any distribution of a subset of particles having a phase space mass density function @xmath124 has corresponding physical space mass density given by @xmath125 or @xmath126 where the hypergeometric function @xmath127 is a constant ( that is with third argument equal to unity ) when both @xmath58 and @xmath102 are perfect power laws in the energy .
the density of that component is then still a power law , with index @xmath128 finally , systems with scale free form for their total mass distribution function have density of states of the form @xmath129 a steady state implies that all time derivatives must vanish . in terms of eq .
( [ eq : fok ] ) this in turn requires that @xmath130 . substituting ( [ eq : g ] ) and
( [ eq : gm ] ) into ( [ eq : fun ] ) one obtains @xmath131 where @xmath132 when @xmath133 eq .
( [ eq : sfsol ] ) with @xmath134 reduces to the form studied by evans & collett ( 1997 ) , thus their solution with @xmath135 and @xmath136 is relevant to clump - background systems considered here . as noted by these authors , the only other physical solution , that is a single power law , is the singular isothermal sphere ( @xmath119 ) . when the assumption @xmath137 is relaxed , there is another solutions for each value of @xmath138 .
it however corresponds to @xmath139 . in the central regions ( i.e , where @xmath140 ) , where the dynamical friction coupling is strongest , and which must therefore be the focus of our analysis
, the physical clump density would be larger than the total ; which is unphysical .
as an example , we plot in fig .
[ fig : scalefree ] the flux obtained from eq .
( [ eq : sfsol ] ) , normalised by the dynamical friction flux ( i.e. , by the first term of on the right hand side of that equation with @xmath141 substituted in explicitly ) , for the case of @xmath142 with @xmath135 an exact solution at @xmath133 .
in addition to this solution , there is another with @xmath143 , that is @xmath144 . as @xmath140
, this inevitably implies that the clump mass density is larger than the total ( clump and background combined ) .
this is clearly impossible ( unless a low energy cutoff is introduced ; see below ) . as one moves towards flatter clump density distributions relative to the total ,
i.e. @xmath145 , there is a net positive flux that increases sharply beyond @xmath146 .
because this flux is positive , and increases outwards , there is a decrease in phase space density ; an expansion of the system .
moreover , this decrease in density is largest as one moves to smaller energy suggesting that initial evolution proceeds in the direction of flatter density distribution . by inserting @xmath147 , in eq .
( [ eq : fun ] ) , with @xmath58 still given by ( [ eq : g ] ) and @xmath148 , one can arrive at a complementary result concerning the background component .
this time , because of the negative sign , one concludes that if the background component has a power law distribution that is less steep than the total , there is compression of the total distribution .
these deductions are only suggestive ; neither rigorous nor comprehensive .
they are nevertheless supported by the monte carlo experiments of el - zant , shlosman & hoffman ( 2001 ) , which show that systems with initially homogeneous clump distributions expand .
predicting the initial direction of evolution in the case when @xmath149 is even less straightforward than the case when @xmath136 and @xmath150 discussed above , because the flux diverges at small energies .
it would appear that the particle distribution is rapidly depleted at these energies , destroying the cusp and forming a core ; while the distribution steepens outside that ( expanding ) region ( because @xmath151 ) . in practice , a small core in the spatial clump distribution
must be always assumed to exist , because subhaloes have finite size .
it corresponds to a low energy cutoff in the phase space distribution function ( section [ sec : core ] ) , which would ensure well behaved forms for the flux at @xmath152 .
such a small energy cutoff in the clump distribution does not lead to appreciable deterioration in the accuracy of the solutions as we show in section [ sec : strip ] in connection with the issue of stripping .
an important feature of fig .
[ fig : scalefree ] is that _
although for @xmath133 the only exact solution ( other than the singular isothermal sphere ) is that with exponent @xmath136 , for @xmath153 the normalised flux is quite modest , suggesting that these configurations may be long lived .
_ in order to make such statements more precise , we consider , in section [ sec : times ] , the question of evolution timescales .
they determine how small the flux needs to be in order for a given approximate solution to be considered relevant for the timescales of interest .
power law densities with @xmath154 are steeper than inner cusps seen in numerical simulations .
they can nevertheless be retained as good represntations to the density profiles of cosmological haloes over a large radial range .
for example , as can be seen from fig .
1 of el - zant & shlosman ( 2002 ) , fits to cosmological halo density profiles correspond to softened @xmath155 distributions over more than three orders of magnitudes in the softening length .
these profiles do not incorporate faithful reproductions of the phase space structure of the inner cusp , since they lack the required temperature inversion in velocity space , but they should represent excellent approximations at intermediate scales ( on which velocity profiles are approximately isothermal ) , up to the virial radius for haloes of small concentrations . thus , the existence of approximate solutions in the range @xmath156 may be interpreted to mean that gravitational systems with density profiles similar to those of cosmological haloes are approximately invariant under substructure - induced interaction since dynamical friction coupling at @xmath157 should be negligible , except for the most massive satellites . nevertheless ,
as we note below , if @xmath157 , and power laws with @xmath158 no longer represent reasonable approximations to the density distribution , the normalised flux can be large , even divergent .
this , in principle , may have an effect if dynamical friction coupling on some satellites is not entirely negligible , and if a significant fraction of the host halo mass is contained in these regions .
although power laws with indices @xmath159 may be useful representations of the density up to the virial radii for cosmological haloes of small concentration , beyond a few scale length @xmath160 , @xmath161 . for @xmath117 , it is appropriate to replace the lower bounds in the first and third integral in eq .
( [ eq : fun ] ) by @xmath162 , and the upper bound in the second integral by @xmath163 .
also @xmath164 in equations ( [ eq : g ] ) , ( [ eq : gm ] ) and ( [ eq : p ] ) . the normalised flux is shown in fig .
[ fig : steep ] in the range @xmath165 . only in the immediate neighbourhood of the singular isothermal sphere is the flux comparable to the case of @xmath166 .
it diverges for @xmath167 .
the source of the blow up is the third integral in ( [ eq : fun ] ) . when the total mass distribution is a power law , so that @xmath168 , it can be seen to correspond ( in absolute value ) to a mass weighed average of the energy of the clump distribution ( recall that @xmath169 ) .
the excess of low energy clumps causes their average energy to diverge for @xmath170 .
the result is a divergent positive flux .
for @xmath171 the first integral also diverges ( because the mass does ) . in practice ,
a cutoff is introduced , because @xmath172 increases as @xmath173 .
the severity of the net flux will then depend on the mass inside the region where @xmath172 falls significantly below @xmath174 .
cosmological haloes do not generally contain a large fraction of their mass in regions beyond a few scale length @xmath160 .
the dynamical friction coupling is also exceedingly small in these low density regions .
nevertheless , the possibility that evolution on a fraction of a dynamical friction time occurs in the outer regions of highly concentrated haloes can not be completely ruled out within the context of the present analysis ; neither can the direction of any prospective evolution be determined .
a thermodynamically stable steady state requires that the time derivatives in eq .
( [ eq : foken ] ) vanish .
this in turn implies that @xmath175 .
for power law systems this requirement translates to @xmath176 ( where @xmath141 is given by [ eq : c ] ) . for @xmath133
, this is impossible , except for @xmath177 , corresponding to ( an unphysical ) completely homogeneous system , or @xmath178 ( the isothermal sphere ) .
all other solutions are thermodynamically unstable . to determine whether these are nevertheless long lived for the timescales of interest we must derive associated evolution times .
this is done in the next section .
but , for @xmath142 , there are exact solutions that are thermodynamically stable for all values of @xmath138 , provided that @xmath179 that is , the massive particle distribution is slightly less steep than that of the total .
this can not correspond to any evolution with solid clumps since , in this case , the clumps sink in , replacing the background to dominate the mass distribution near the centre of the system ( @xmath180 ) .
it may be relevant however in the presence of stripping ( section [ sec : strip ] ) , a process whereby clump material is transformed into background .
note that , because these configurations have @xmath181 , the corresponding mass flux is also quite small .
a standard procedure in physics is to derive a characteristic timescale by freezing the physical parameters of a system at a given moment and estimating the evolution time from there . in our case , using eq .
( [ eq : fok ] ) , a timescale for the evolution of the total mass distribution because of the existence of a non - vanishing mass flux can then be defined as @xmath182 which is the time interval the flux at energy @xmath61 takes to significantly modify the total mass distribution inside that energy surface .
suppose that , within that energy surface @xmath61 , the mass distribution is initially dominated by the smooth background component . then the fluctuations in the potential resulting in random heating ( described by the coefficient @xmath78 ) have negligible effect on the clumps .
they therefore sink in under the action of dynamical friction on a timescale similar to that given by ( [ eq : dfsim ] ) .
a closely related quantity , the time on which the mass in sinking clumps significantly modify the original mass in clumps inside energy level @xmath61 is given by @xmath183 where @xmath184 as usual corresponds to the first term or the right hand side of eq .
( [ eq : fun ] ) . the dynamical friction time itself
can be arrived at by using eq .
( [ eq : de ] ) , again neglecting self interaction terms arising from clump - clump heating due to their mutual encounters ( i.e. , the term involving @xmath83 ) .
one gets @xmath185 but @xmath186 ; so we have @xmath187 since @xmath188 is generally smaller than @xmath189 , this timescale is larger than that obtained from ( [ eq : taufdf ] ) ; it involves the effect of the flux on the local mass distribution of clumps , instead of its influence on the distribution inside the phase space surface bounded by @xmath61 .
one can also define an analogous timescale for the evolution of the total mass distribution .
the ratio of these timescales would be @xmath190 it determines how rapidly the total mass evolution takes place relative to the dynamical friction timescale ; which describes how fast the distribution of one of the components , namely the clumps , changes .
suppose now that the initial system consists of single power law configurations ( i.e , in terms of the terminology of section [ sec : scalefree ] , @xmath191 ) .
the dynamical friction flux for this initial system then is @xmath192 ( obtained by inserting [ eq :
g ] and [ eq : gm ] into [ eq : fun ] ) ; in terms of which one gets an explicit form of the dynamical friction timescale : @xmath193 we omit the explicit reference to the initial time in what follows , even thought this is implied .
for such a system to keep its total phase space mass density @xmath58 nearly constant ( eq .
[ eq : g ] remaining a good approximation ) , for timescales of the order of the initial dynamical friction time , it is necessary that the absolute value of the dimensionless flux @xmath194 be significantly smaller than unity .
note that this same result can be inferred from ( [ eq : tauf ] ) and ( [ eq : taufdf ] ) .
note also that , although this equation assumes a power law form for the _ total _ distribution , the flux @xmath62 can correspond to arbitrary distribution for any of the _ individual _
( i.e , background and clump ) components , which are necessarily time dependent .
the quantity @xmath195 _ defines a relative error _ , quantifying the validity of approximate solutions .
more precisely , a solution is a valid approximation , that is _ the system s total phase space distribution stays approximately invariant for a timescale @xmath196 if @xmath197 for all relevant energies .
_ at any given moment one can define a dimensionless characteristic evolution timescale @xmath198 it expresses the number of dynamical friction times the total phase space mass distribution can remain approximately invariant . a system can be considered unstable under the action of the coupling induced by the presence of substructure clumps if @xmath199 since , in that case , the total mass distribution changes if the configuration is evolved long enough for dynamical friction , the principal manifestation of such interaction , to be effective . as an example , the system studied in section [ sec : sim ] must have @xmath200 for energies corresponding to radii that are evolved over a dynamical friction time or longer .
[ fig : sim ] also suggests that the timescale for each of the components to be significantly modified is comparable to the dynamical friction time .
this is consistent with the results in this section .
note however that the replacement of background by clumps does not proceed linearly in time in which case it would take a time @xmath201 for the clumps to dominate inside some energy surface @xmath61 .
the somewhat more detailed analysis of section [ sec : instab ] shows that the evolution is actually exponential in @xmath16 ( at least initially ) .
this explains why the population exchange is virtually complete on timescales of the order of a dynamical friction time , even if the mass fraction in clumps is quite small .
finally , note that if the mass fraction in heavy clumps is not negligible , then the _ effective _ dynamical friction flux decreases by a factor of @xmath202 , because part of that flux is compensated by heating due to clump - clump interaction ( c.f . , eq .
[ eq : fun ] ) .
the corresponding dynamical friction time increases by an inverse of this factor . the effective evolution flux can then be defined as @xmath203 unless otherwise stated , we will have in mind a system where the background dominates , so that @xmath204 , and will be using eq .
( [ eq : fe ] ) ; since it corresponds to a definition of the dynamical friction time that is closer to conventional usage based on the chandrasekhar formula , which considers only one massive particle in an infinite background medium ( eq . [ eq : dfsim ] is based on that formulation ) ; obviously , conversion between @xmath195 and @xmath205 is trivial .
the same attitude adopted in the subsection above allows one to define timescales associated with the energy flux .
for example , the expression analogous to ( [ eq : tauf ] ) is @xmath206 one can also derive a corresponding ` evolution flux ' @xmath207
we now apply the results just derived to power law density distributions with @xmath191 ( considered in section [ sec : scalefree ] ) . the values of @xmath208 in the case of these scale free solutions
are shown in fig .
[ fig : scalefree ] . for @xmath209
the absolute value of this ratio is @xmath210 . for models with single power laws
@xmath211 determines the mass density of clumps relative to the total mass density .
( [ eq : fe ] ) then implies that evolution timescale associated with the mass flux ( given by eq .
[ eq : taue ] ) is necessarily large compared to the dynamical friction time .
for example , for the mass ratio of the simulations of sec .
[ sec : sim ] , @xmath212 , and so @xmath213 in the central region , where the evolution takes place and where @xmath214 .
this means that when both mass components are distributed in power law density configurations with @xmath209 , the total phase space mass distribution would evolve over @xmath215 dynamical friction timescales , very large compared to timescales of interest ( at @xmath160 , this corresponds to @xmath216 ; which is far larger than the age of cosmological haloes ) .
when @xmath217 the energy evolution flux can be written as @xmath218 this reaches a maximum of @xmath219 for @xmath220 .
it is smaller for other values of @xmath221 .
scale free systems in this range of @xmath172 are therefore also long lived from the energy transfer viewpoint that is , the time for energy transfer to affect the total distribution is much longer than the dynamical friction time .
( note that , for @xmath220 , @xmath222 , although there appears to be no reason to suppose this not to be entirely coincidental ) .
the mass evolution flux is not negligible when @xmath223 that is , for flat initial density distribution especially if the mass fraction in clumps is not small . in this case
one must use eq .
( [ eq : effflux ] ) , with @xmath224 ; which , for @xmath223 , implies that @xmath225 . therefore , a system which starts with a weak central cusp , and a significant fraction of its mass in clumps , is unstable if the coupling due to dynamical friction is non - negligible .
the same conclusion applies to the evolution driven by the energy flux when @xmath214 .
initial scale free solutions can only be adopted as zeroth order approximations .
as the system evolves , the coupling between the two components will cause the light background particles to move out of lower energy levels and be replaced by the massive clumps .
in addition , the latter can lose mass to the background under the action of stripping .
the goal of the remainder of this paper is to show that the _ total _ phase space mass density can nevertheless remain nearly constant , approximately keeping the power law form , even when the individual distributions are modified by the aforementioned processes .
( [ eq : fok ] ) , for the background particles alone , can be written as @xmath226 where @xmath227 is given by eq .
( [ eq : flun ] ) .
suppose now that the total phase space distribution is largely unchanged over timescales of interest ; that is , @xmath228 , and so @xmath58 continues to be expressed in terms of its initial value form , given by eq .
( [ eq : g ] ) .
the second term on the right hand side of ( [ eq : evol ] ) is then also negligible .
we assume a generalised power law solution for the background @xmath229 .
this , in itself , is not an approximation , since the function @xmath230 is arbitrary .
however , in order to obtain a solution , we assume , in addition , that for the purpose of evaluating the integrals in ( [ eq : flun ] ) , @xmath102 keeps its original scale free form , given by ( [ eq : gm ] ) . for the first integral in eq .
( [ eq : flun ] ) this is an excellent approximation at any given time @xmath196 and energy @xmath61 such that @xmath231 ; since , in that case , the effect of the dynamical friction coupling is just becoming important at energy @xmath61 , and is therefore unimportant at larger energies ( recall that @xmath232 ) . for the second integral , one notes that when the _ total _ mass distribution remains in scale free form @xmath233 .
this integral then represents a mass weighed average of the energy in massive clumps ( recall that @xmath234 ) .
it will decrease as the the latter lose energy because of dynamical friction .
nevertheless , if this occurs principally at low energies , where @xmath16 is smaller , the change is not drastic .
more formally , integration by parts transforms it into @xmath235 where the scale free form for ( eq . [ eq : p ] ) remains relevant , in accordance with our assumption that the total distribution is practically time independent . using eq .
( [ eq : mass ] ) , we get @xmath236 mass conservation implies that the first term in this expression is constant if the effect of dynamical friction is small inside @xmath61 .
it is also always larger than the second @xmath237 where the equality corresponds to the case when all the mass is concentrated near @xmath238 rather unlikely given the assumption that , at energy @xmath61 , the effect of dynamical friction is just being felt at some energy @xmath239 , ( it would imply a sharp discontinuity in the @xmath240 profile ) .
moreover , @xmath241 is always smaller than the first integral in eq .
( [ eq : flun ] ) , which as we have seen is virtually constant . in the initial scale
free system its contribution to the total outward flux is a fraction @xmath242 , which is less than @xmath243 for @xmath244 ( i.e. , for @xmath245 ) .
this ratio can only become smaller on timescales for which the inflow of massive particles across @xmath61 does not significantly change their total mass inside @xmath61 , but where redistribution inside this surface renders their phase space density profile steeper ( again this follows from eq . [
eq : flin ] ) .
the use of the initial scale free form for @xmath102 in evaluating the integrals on the right hand side of ( [ eq : flun ] ) is thus justified when @xmath231 .
since the total mass inside energy surface @xmath61 is still unaffected on that timescale , @xmath246 can also be be computed using the initial scale free distribution . in this context , eq . ( [ eq : evol ] ) transforms into @xmath247 where @xmath248 with @xmath249 a parameter of order unity .
( [ eq : aproel ] ) can be integrated to give @xmath250 here we have assumed , as usual , @xmath251 ( and so @xmath252 ) .
the function @xmath253 is in principle arbitrary , but is in fact independent of energy when the initial ratio of light to heavy particles has this characteristic .
and so we have @xmath254 the timescale @xmath255 determines the interval on which the exchange of populations , associated with the massive clump inflow accompanied by expulsion of background , takes place .
this segregation timescale is of the order of the dynamical friction time .
more explicitly , using eq .
( [ eq : taudfsf ] ) , @xmath256
the results of the previous section suggest that dynamical friction coupling causes depletion in the distribution of background particles ; first at the smallest energies , and then at progressively higher energy .
suppose that there is some cutoff energy @xmath257 below which no background particles exist . at energies @xmath258 , and still assuming that the clumps are indestructible ( no merging or striping ) , there are again equilibrium solutions which are single power laws this time for a one component system composed of the clumps , since one can now rewrite the fundamental equation only in terms of the clumps ( by putting @xmath259 in eq .
[ eq : fun ] ) .
the dimensionless evolution fluxes ( eq . [ eq : fe ] and [ eq : fee ] ) are now significantly larger , since @xmath260 ; the associated timescales correspondingly smaller
. nevertheless , this state will still evolve on a timescale that is @xmath261 , provided @xmath153 .
only when @xmath172 is significantly larger , corresponding to flat density profiles , will the evolution timescale be comparable to the dynamical friction time .
it is important to note here that the ultimate effect of dynamical friction by a smooth background on solid clumps is to produce a self gravitating core made of these clumps , with the total density distribution largely unchanged _ and not the sinking of the clumps to the centre _ as may be expected had the background been of infinite mass .
no matter how small the fraction of mass in clumps initially , the ratio will increase exponentially on a dynamical friction timescale ( as we have shown in the previous section ) , with the clumps eventually coming to dominate inside some ( time dependent ) energy level @xmath262 .
this puts an effective end to effects induced by dynamical friction from the background , which has been expelled from that region , and to the sinking of clumps to the physical centre of the system . for @xmath263 the system would still retain its original distributions of clump and background particles , the dynamical friction time being too long for there to be any effect .
the question therefore is what transpires at the boundary ; the transition region between the two equilibria .
suppose that @xmath264 for @xmath265 , but remains unmodified for larger energies . ) , so that the migration of low-@xmath61 particles has little effect on the distribution there ( as in fig .
[ fig : sim ] , where drastic changes in the inner distributions of light particles has little effect on the outer regions where they moved out ) . in
what follows we will be assuming this approximation holds .
] by writing @xmath147 in eq .
( [ eq : fun ] ) one can separate it into two components : one involving only functions of @xmath58 and another involving both @xmath58 and @xmath266 .
if it is assumed that the total phase space mass distribution function @xmath58 remains a perfect power law , then the flux due to the one species term ( involving only @xmath58 ) , can be made arbitrarily small by choosing an appropriate value for @xmath138 ( i.e , @xmath267 or @xmath178 ) . the two species component ( involving @xmath58 and @xmath266 ) has no such solutions .
in fact , for @xmath265 there is the residual flux @xmath268 where @xmath269 and @xmath270 which is always nonzero .
dividing by the dynamical friction flux , we obtain the associated evolution flux ( cf . , eq . [ eq : fe ] ) @xmath271 as may be expected from the discussion of the preceding subsection , for @xmath258 this residual flux becomes small ; the system of clumps , thus represented , is near equilibrium . for @xmath272 ( and @xmath273 ) however , it is smaller than unity but not negligible . for @xmath274
there is another non - zero residual term resulting from the abrupt cutoff .
it is given by @xmath275 with @xmath276 the associated evolution flux this time is @xmath277 it has a sign opposite to that of @xmath278 and decreases away from @xmath262 significantly faster .
it is also smaller at @xmath272 , but still is not completely negligible .
both residual terms can be made far smaller if one replaces the abrupt cutoff with a gradual one .
we do this by introducing a variable cutoff energy @xmath279 .
this conceptually amounts to dividing the background material into a continuum of populations , each with its own cutoff energy , such that @xmath280 changes in an interval @xmath281 by an amount @xmath282 .
we will suppose that at some energy @xmath283 , @xmath284 ; and at another , @xmath285 , it takes its initial value .
the evolution fluxes can then be written ( by summing over all the populations ) as @xmath286 and @xmath287 we can also rewrite eq . (
[ eq : sresidin ] ) and ( [ eq : sresidout ] ) in terms of the dimensionless variable @xmath288 . for any phase space point with energy
@xmath61 the ` inner flux ' from points with @xmath289 is then given by @xmath290 where @xmath291 and @xmath292 .
there will also be an ` outer flux ' from points with @xmath293 .
it is @xmath294 note that , both these integrals are minimised when @xmath230 is flat , that is @xmath295 , when @xmath230 is largest . by choosing functions with decreasing logarithmic derivatives
, one can make the integrals as small as one likes .
the fact that the two terms are always nonzero also means that they tend as to cancel each other . as we will see below , it is easy to find functional forms that render the total evolution fluxes quite small . in this case
the relevant quantity is @xmath296 , which replaces @xmath62 in calculations analogous to those just described .
it then follows ( using [ eq : fee ] and [ eq : rawresidin ] ) that @xmath297 and , in an analogous manner ( this time using [ eq : rawresidout ] ) @xmath298 thus , for @xmath299 , the inner residual energy flux is smaller , and the outer larger , compared to the corresponding mass fluxes by only a modest factor . since we will be interested in showing that the fluxes are actually smaller by an order of magnitude , or more , than those that would lead to significant change of the total phase space mass distributions , over timescales of interest , these factors are of secondary importance . in the illustrations that follow we will be focussing on the mass flux
equation ( [ eq : lin ] ) was arrived at under the assumption that the total phase space mass density remains constant during that evolution ; that is @xmath300 .
we now check this .
self consistency requires that when we insert back the solution into ( [ eq : fun ] ) , the evolution flux is indeed small .
we suppose that each element of the system can be evolved according to eq .
( [ eq : lin ] ) . according to the discussion of section [ sec : instab ] , this is a good approximation at points where the effect of dynamical friction is just becoming significant .
nevertheless , the distribution at smaller energies , where the dynamical friction time is smaller , need not follow this form .
this affects the calculation of the third integral in ( [ eq : fun ] ) and therefore involves a further approximation , even at points where @xmath301 ( note that it does not affect first integral which depends only on the total distribution ) .
but since , as also discussed in the aforementioned section , this integral ( denoted by @xmath241 ) is insensitive to the precise distribution inside energy surfaces @xmath61 for which @xmath302 , and in any case does not dominate the flux , this approximation should be valid when calculating that flux for points where the dynamical friction is just becoming important .
( [ eq : lin ] ) can be solved for @xmath303 at any given energy @xmath61 and time @xmath196 to give @xmath304 substituting this form for @xmath230 , using eq .
( [ eq : taus ] ) and ( [ eq : a ] ) , in ( [ eq : normresidin ] ) and ( [ eq : normresidout ] ) and ( since this functional form does not admit sharp cutoffs ) letting @xmath305 and @xmath306 , one obtains the residual fluxes .
these are shown in fig .
[ fig : expos ] in terms of @xmath307 and for @xmath308 .
since the exponential evolution was deduced under the assumption that material beyond our reference point at @xmath309 was largely unaffected , they are expected to be highly accurate only for @xmath310 .
for this is the value of @xmath311 that corresponds to one dynamical friction time for @xmath312 ( from eq .
[ eq : dftos ] ) .
the results of fig .
[ fig : expos ] do indeed show that for @xmath310 , the average value of the evolution timescale ( the inverse of the evolution flux shown ) correspond to @xmath313 which means that , up to one dynamical friction time , the timescale for the total mass to evolve away from its initial phase space distribution is twenty times longer .
the approximation breaks down completely when @xmath314 , where the angular brackets refer to averages of @xmath195 over times smaller than @xmath196 . from fig .
[ fig : expos ] , it may be estimated that the exponential solutions are relevant for the whole period of evolution that is plotted .
they are a good approximation ( @xmath315 ) up to @xmath316 , corresponding to about two dynamical friction times .
note that , when @xmath311 is very large , at @xmath317 the system tends to the one - component equilibrium state described in section [ sec : behaviour ] .
therefore , the accuracy of the exponential solution does not continue to deteriorate as @xmath311 increases .
however , because the gradient of @xmath230 is quite large , the decay in the inner flux approximately follows the sharp cutoff form given by eq .
( [ eq : insharp ] ) , and the tendency towards the one component solution is slow .
but the form given by ( [ eq : lingm0 ] ) has been derived under the assumption that @xmath318 , and is therefore no longer relevant . in the following , we present approximate solutions that describe a smooth transition between the two equilibria discussed in section [ sec : behaviour ] , with corresponding evolution fluxes that are very small at all points .
since , for power law forms of the total mass distribution function @xmath58 , there only corresponds two solutions for the individual components that can be also expressed in terms of powers of the energy ( section [ sec : scalefree ] ) , no solutions can be written as an infinite power series in the energy , and be exact . nevertheless , as we show here
, some such solutions can be exceptionally excellent approximations .
for example , we have tried solutions of the form ) and smooth exponential forms ( e.g. , @xmath320 , for @xmath321 ) with similar results . ]
@xmath322 where @xmath323 is some scaling parameter and @xmath324 corresponds to the unperturbed value ( i.e. to @xmath325 ) .
they represent systems where the background material is depleted at small energies due to interaction with solid massive clumps .
we insert @xmath147 into eq .
( [ eq : fun ] ) , with @xmath58 , the total phase space mass density assumed to remain constant .
consistency requires that the evolution flux ( eq . [ eq : fe ] ) is small .
this substitution breaks the equation into two additive terms . in the case when @xmath135 , the @xmath326 terms have exactly vanishing total flux . for @xmath327
there is an evolution flux of @xmath328 .
it will add to any evolution flux resulting from the @xmath329 term .
the total evolution fluxes are shown in fig .
[ fig : smooth ] ( again for the case where @xmath330 ) .
as may be expected from the discussion in section [ sec : behaviour ] , when @xmath135 , they tend to zero as @xmath331 ( corresponding to the one component equilibrium state ) ; and as @xmath332 ( where the initial , two component equilibrium , still holds ) . as may also be expected ( this time from section [ sec : theinvariance ] ) ,
the absolute values of the fluxes peak at the transition between these two regimes , and are larger when the transition is most rapid . note nevertheless that @xmath323 evolves on the local dynamical friction timescale , as the instability front moves out to higher energies on the segregation time scale @xmath333 of section [ sec : instab ] .
therefore no given region in energy space would remain , during the whole period of evolution , in the regime where the peaks lie .
like in the previous section where we discussed the initial exponential evolution , it is the average value of the flux that is relevant .
when @xmath327 , and @xmath334 , the residual flux connected to the @xmath326 ( one component ) term causes the total flux to tend to @xmath328 .
the associated evolution timescale is of the order of ten dynamical friction times .
the energy flux also becomes important on such timescales .
significant evolution in the total mass distribution is therefore expected .
we have verified by means of simulations ( like those of section [ sec : sim ] ) that evolution in the total distribution does indeed occur on these timescales .
it is initially in the direction of expanding core ( as found by hayashi et al .
we have not integrated our systems for longer times , but core collapse , driven by the evolution of the ( by then completely ) self gravitating system of clumps , probably occurs .
both these regimes however are well beyond any situation where the representation of substructure in terms of solid clumps can be expected to be of any use . for stripping would have radically modified the distribution and internal structure of satellite haloes .
the existence of a low energy cutoff in the phase space distribution will entail the removal of some particles of the affected species from the central region of the system in physical space ; it does not however imply the removal of all particles .
instead , the radial mass density distribution of the background will develop a nearly constant density core . for , as can be seen by replacing the lower integration limit by @xmath257 in eq .
( [ eq : dens ] ) , there will always be a radius where @xmath335 . for power law potentials of the form @xmath336
, the transition radius is @xmath337 .
deep inside this region the density tends to the constant value @xmath338 .
this is of course what is found for the background distribution in fig .
[ fig : sim ] ( left panel ) , where the density of background particles does indeed flatten off at small radii . in el - zant , shlosman & hoffman ( 2001 ) and el - zant et al .
( 2004 ) we had interpreted the clumpy component to be made of baryons ; and assumed that stripping was negligible and that this state of affairs actually materialised in practice ( the rationale being that baryonic material has formed by dissipation and is therefore dense enough to survive ) . in contrast , if the clumps are made of dark matter , there will be continuous transformation between the two components , as stripping transforms the clump material into background .
cosmological simulations actually suggest that this process may be so effective that _ the cutoff in the energy distribution in practice probably occurs in the clump distribution _ , and not the smooth background . for the density distribution of the clumps
consistently shows evolution towards a flattened core ( e.g. , ghigna et al .
2000 , fig .
as discussed in section [ sec : striptease ] , stripping does not significantly change the phase space mass density because stripped particles are likely to escape from the clumps with very small kinetic energy and near zero binding energy ( relative to the clump ) , the amount of mass in the phase space energy range @xmath339 $ ] remaining nearly constant ( the mass in the clumps in that energy range decreases , but is compensated by an equal increase in background ) . therefore ,
if the interaction due to dynamical friction leaves the system in equilibrium , it will remain so in the presence of tidal stripping .
one nevertheless still needs to show that satellite distributions that are affected by stripping also correspond to long lived equilibrium states for the total mass distribution .
suppose now the stripping turns most of the mass in clumps into background materials for energies @xmath340 .
then the role of the two components are reversed from what was described in section [ sec : approxal ] .
now , as appears to be the situation in evolved cosmological systems , it is the satellite distribution which would have a lower energy cutoff and develop a core ( in physical space ) , while the background would have a cusp . in that case , instead of substituting @xmath147 into eq .
( [ eq : fun ] ) , as we have done above , we can directly put @xmath341 since this simply represents a direct exchange of the role of the two components compared to what is described by eq .
( [ eq : smooth ] ) , we expect the situation to be similar except that the evolution flux is now expected to be even smaller , because the principal source for evolution , the massive clump distribution , vanishes at small energies , where the coupling is strongest . as can be seen from fig .
[ fig : stripsmooth ] , this is indeed the case . since there is now virtually no clump - clump interaction at small energy ,
there is no longer a relatively large flux as @xmath152 when @xmath342 .
the energy transfer flux is also always small .
physically this simply means that , if stripping is efficient over timescales smaller than the segregation time @xmath333 ( eq . [ eq : taus ] ) , the evolved regions tends towards a non - evolving collisionless state as @xmath323 increases , instead of the self gravitating system of satellites expected to transpire if the stripping time is arbitrarily long ( i.e. , when the clumps can be considered solid ) . of course
, in a realistic system , the distribution of clumps would be more complicated .
there will be a mass spectrum , and each species may have a different distribution . nonetheless , because of the linearity of the fundamental equation ( [ eq : fun ] ) with respect to @xmath102 ( cf .
section [ sec : linfun ] ) , one can add any number of approximate solutions of the form ( [ eq : strip ] ) ( or , for that matter , [ eq : smooth ] ) and still end up with a system that is close equilibrium the errors , that is the evolution fluxes , will simply add linearly .
the principal conclusion of this paper is that the phase space mass distribution associated with density distributions akin to the ` universal ' halo profiles are approximately invariant under the action of the interaction induced by the presence of substructure satellites ; which is shown to lead to negligible evolution over timescales comparable to halo lifetimes .
this conclusion applies to the central region ( @xmath343 ) , where the dynamical friction coupling is strongest , and up to radii where the profile can be approximated as @xmath48 . since haloes in cosmological simulations build up from smaller components ; are continually incorporating infalling material ; and retain significant substructure throughout their evolution , this is a necessary condition for the observed universal forms , apparently present at all redshifts and mass ranges even if the initial collapse naturally leads to the ` right ' distribution , evolution driven by the presence of substructure will modify it , unless the total mass distribution is shown to be invariant over relevant timescales .
the physical mechanism behind our claim is simple .
clumps and background are coupled via dynamical friction interaction .
the clumps move in , the lighter background particles move out , the total remains the same .
this equation , which we have shown in terms of distribution functions in energy space , naturally leads to invariance in physical space .
it is unmodified by the effect of inter - substructure interaction in the form of weak encounters .
we presented simple simulations which showed that for dozens of dynamical times ( inside the initial scale length @xmath160 ) , the physical as well as velocity space total distributions remain invariant under the dynamical friction mediated interaction between a ( solid ) clumpy component and a background of far lighter particles despite the fact that , during that same timescale , the distribution of each component is radically modified ( section [ sec : sim ] ) .
the rest of the paper had the two principal goals : 1 ) to explain the results of these simple simulations and 2 ) to discuss , both from the qualitative and quantitative viewpoints , the relevance of such idealised models to the more complex context of cosmological simulations .
the conceptual and computational framework employed is based on a fokker - planck formulation , dividing self gravitating structures into two components : a background made of light particles and a system of clumps representing substructure satellites .
the latter are assumed to interact with the background _ via _ dynamical friction , and among themselves through weak gravitational encounters .
although we usually have in mind a system where the background dominates , this is not an _ a priori _ assumption of the model .
it shows ( section [ sec : scalefree ] ) that there are exact scale free solutions for cusps with both components having scale free density distributions @xmath344 , and @xmath136 .
this solution , already found for single mass systems by evans & collett ( 1997 ) , is generalised for the case of clump - background systems considered here .
as long as the massive clumps are much heavier than the background particles , any mass distribution of the former is allowed ; by virtue of the linearity of the fundamental equation for the mass flux ( sec [ sec : linfun ] ) .
power law distributions , in the whole range of @xmath345 , are in approximate equilibrium , both in terms of the mass and energy transfer ; the total mass distribution in phase space remaining constant for many dynamical friction times ( sec .
[ sec : timescales ] ) .
solutions with @xmath346 do not correspond to any power law found in the central regions of cosmological haloes .
but they are faithful representations of the phase space structure at intermediate radii ( up to the virial radius for haloes of small concentration ; section [ sec : scalefree ] ) .
nevertheless , when @xmath157 and @xmath158 power laws can no longer be taken as faithful represntations of the density distribution , there can , in principle , be a large mass flux towards higher energies , if dynamical friction is not completely negligible at these radii , and if they contain a significant fraction of the halo mass ( section [ sec : steep ] ) .
even when the total mass distribution function is constant , there is continuous evolution in the distribution of each of the components .
two cases may be distinguished , depending on how efficient the stripping of the massive clumps is . in the case
when the clumps are considered as solid objects , we show that the evolution of each of the components away from the initial power law function takes the form of an exponential instability , which we termed the ` segregation instability ' , and which takes place on a characteristic timescale comparable to the local dynamical friction time .
it involves the replacement of lighter particles , which gain energy and exit lower energy levels , by the massive clumps ( section [ sec : instab ] ) .
this results in a situation whereby the distribution of light background particles develops a low energy cutoff in energy space with a corresponding constant density core in physical space ( section [ sec : core ] ) , in line with the evolution of the idealised @xmath23-body models discussed in section [ sec : sim ] .
long lived solutions for such models are given in section [ sec : approxal ] .
situations whereas substructure is formed by dissipationless collapse are unlikely to be represented by such models for any significant period of time .
nevertheless , these may be relevant if the clumps are made of significantly denser baryonic material ; which is more resistant to dissolution _ via stripping _ , because it dissipated during collapse ( e.g. gao et al .
one then recovers the situation described by el - zant et al .
( 2004 ) , where a core developed in the light dark matter particles but the total density distribution remained largely unchanged .
similar conclusions were reached by gao et al ( 2004a ) in a much more sophisticated cosmological setting .
if the timescale for mass loss from clumps stripping is significantly smaller than the segregation time , a cutoff in the _ satellite _ distribution function may result instead , with an accompanying constant density core in the physical space distribution of that component .
this appears to be what happens in the context of dissipationless cosmological simulations . in section [ sec : striptease ] we had argued that stripping does not modify the total mass distribution function .
therefore , if the dynamical friction interaction between clumps and background , and weak encounters between clumps , do not modify that distribution function , it suffices to show that solutions that take into account stripping that is ones incorporating a low energy cutoff in the clump distribution can be long lived .
such solutions , valid up to a thousand dynamical friction times ( much longer than the age of cosmological haloes ) , are given in section [ sec : strip ] . finally , it is interesting to note that , whereas in the case of a single mass system , or a multimass systems without stripping , the only solutions with no energy transfer ( i.e. , thermodynamically stable configuration ) , are isothermal spheres , the case may be different when there is a mass spectrum and stripping is effective ( section [ sec : energysol ] ) .
several issues raised in this paper would seem to require numerical work , either by means of the fokker - planck equation or direct simulations , to be settled in a satisfactory manner .
the two approaches are complementary : the full simulations including fewer physical assumptions and the fokker - planck models corresponding to more controlled calculations that are computationally far less intensive .
the question concerning statistical effects arising from the finite number of subhaloes , and the contention that stripping does not modify the total mass distribution function ( both discussed in section [ sec : physical ] ) , need to be empirically verified by means of full simulations , involving live @xmath23-body satellites .
other issues are best treated by a combination of the two approaches .
the effects of a mass spectrum and stripping may be explicitly included in fokker - planck models ( merritt 1983 ) .
it is also possible to incorporate a prescription for the evolution of the halo mass distribution as was done for example by nusser & sheth ( 1999 ) .
these authors used a stable clustering formulation , supplemented by an algorithm for the effect of dynamical friction on substracture and associated back reaction on the main halo .
because they have assumed that energy lost by sinking satellites is redistributed homogeneously among parent halo particles , their conclusion was that sinking substructure causes a steepening of the cusp , invalidating the ` universality ' assumption .
their basic model nevertheless remains relevant and can be coupled to the fokker - planck formulation , which provides a much better representation of the effect of energy deposition by substructure ( which in fact is better approximated as a local phenomenon ) . of prime interest
is the issue of stability , touched upon in the introduction whether the effectively dissipational interaction due to the presence of substructure may lead to an ` attractor ' like behaviour , with systems tending to preferred configurations similar to those observed in the large simulations .
there are several related points that are raised by the material presented here . the fundamental conclusion of this paper , the invariance of the total phase space mass density distribution under the action of the dynamical friction coupling , strictly speaking holds only when _ both components _ initially have the same phase space distributions .
simple considerations involving the direction of phase space fluxes and their gradients ( section [ sec : scalefree ] ) suggest that power law systems may expand if the the clumps distribution rises less steeply than the background , and contract if the situation is reversed .
what is available in terms of numerical data tends to confirm this .
for example , el - zant , hoffman & shlosman ( 2001 ) ran monte carlo simulations where solid substructure was given homogeneous spatial initial conditions inside an nfw halo ; the combined system was started from virial equilibrium and left to relax towards detailed dynamical equilibrium ; the dynamical friction , modelled on the chandrasekhar formula , was then turned on
. the total density distribution decreased .
heuristically , one may explain this from the fact that the binding energy available , per unit mass , is larger the more diffuse distribution . nevertheless , this issue is of sufficient importance so as to merit further work , involving a series of simulations whereby the spatial distribution of substructure within the halo , as well as the internal structure of individual clumps is varied .
the material in sections [ sec : scalefree ] and [ sec : timescales ] suggest that single power law systems with flat cusp ( @xmath223 ) are only marginally unstable they would evolve on a dynamical friction time only when @xmath347 ( which may be taken to represent the nearly homogeneous centre of a system ) and when the mass fraction in clumps is comparable to that in the background . whether any evolution actually occurs will clearly depend on if stripping is effective enough in order to quench the evolution , by modifying the mass ratio and cutting the source of the flux before any actual evolution takes place .
it is of interest to determine under what ( if any ) condition evolution does occur , and whether this happens in the direction of the cusps characterisitc of cosmological haloes .
finally , there is the issue of whether the large evolution flux associated with the very outer parts of haloes ( @xmath117 ; see section [ sec : steep ] ) has any significance in determining the structure of the outer profiles . dealing with these regions
may also require the incorporation of velocity anisotropies .
the author is grateful to peter goldreich for suggestions , inspiration and encouragement .
it is a pleasure to thank adi nusser and sergei shandarin for helpful discussions and critique .
comments by julio navarro were also useful .
this work originated from numerous discussions with peter goldreich , milos milosavljevic and isaac shlosman ; it was supported by nsf grant ast 00 - 98301 at caltech .
binney j. , tremaine s. 1987 , galactic dynamics .
princeton university press chandrasekhar s. , 1943 , rev .
15 , 1 cohn h. , 1979 , apj 234 , 1036 cohn h. , 1980 , apj 242 , 765 dekel a. , arad i. , devor j. , birnboim y. , 2003 , apj 588 , 680 dekel a. , devor j. , hertzoni g. , 2003 , mnras 341 , 326 el - zant a. a. , 1998 , phys .
e 58 , 4152 el - zant a. a. , shlosman i. , hoffman y. , 2001 , apj 560 , 636 el - zant a. a. , shlosman i. , 2002 , apj 577 , 626 el - zant a. a. , hoffman y. , primack j. p. , combes f. , shlosman i. , 2004 , apj 607 , l41 evans , n. w. 1994 , mnras , 267 , 333 evans n. w. , collett j. l. , 1997 , apj 480 , l103 gao l. , loeb a. , peebles p. j. e. , white s. d. m. , jenkins a. , 2004a , apj 614 , 17 gao l. , de lucia g. , white s. d. m. , jenkins a. , 2004b , mnras 352 , l1 gao l. , de lucia g. , white s. d. m. , jenkins a. , 2004c , mnras 355 , 819 ghigna s. , moore b. , governato f. , lake g. , quinn t. , stadel j. , 2000 , apj 544 , 616 hayashi e. , navarro j. f. , taylor j. e. , stadel j. , quinn t. , 2003 , apj 584 , 541 henon m. , 1961 , ann .
24 , 369 hiotelis n. , 2002 , a & a 382 , 84 klypin a. , kravtsov a. v. , bullock j. s. , primack j. r. , 2001 , apj 554 ,
903 lokas e. l. , hofmann y. , 2001 , apj 542 , l139 ma c .- p .
, bertschinger e. , 2004 , 612 , 28 ma c .- p . , boylan - kolchin m. , 2004 , , phys
93 , 021301 merritt d. , 1983 , apj 264 , 24 moore b. , quinn t. , governato f. , stadel t. , lake g. , 1999 , mnras 310 , 1147 navarro j. f. , frenk , c. s. , white s. d. m. 1997 , apj 490 , 493 ( nfw ) navarro j. f. et al . , 2004 , mnras 293 , 337 nusser a , sheth r. k. , 1999 , mnras 303 , 685 nusser a. , 2001 , mnras 325 , 1397 padmanabhan , t. , 1990 , phys .
188 , 285 penarrubia j. , benson , a. j. , 2004 ( astro - ph 0412370 ) saslaw w. c. , 1985 , gravitational physics of galaxies and clusters , cambridge university press .
saslaw w. c. , 2000 , the distribution of the galaxies : gravitational clustering in cosmology .
cambridge university press spitzer l. , 1987 , dynamical evolution of globular clusters .
princeton university press syer d. , white s. d. m. , 1998 , mnras 293 , 337 taylor j. e. , navarro j. f. , 2001 , apj 563 , 483 theuns t. , , 1996 , mnras 279 , 827 taylor j. e. , babul a. , 2004 , mnras 348 , 811 weinberg m. d. , 2001 , mnras 328 , 321 zaritsky d. , white s. d. m. , mnras 235 , 289
in general , a kinetic equation with a collisional term can be written as @xmath348 where @xmath58 is mass density distribution function the amount of mass inside a unit volume element in a six dimensional phase space . when the collisional term on the right hand side is zero
, one recovers the collisionless bolzmann equation : the dynamics conseerves the phase space mass density along the motion .
the collisional term can be written as @xmath349 where @xmath350 is the mass flux through phase space , defined as the amount of mass crossing unit area per unit time , and @xmath351 is the phase space gradient .
this is simply a statment of conservation of mass in phase space the change in mass density inside a phase space volume due to the effect of encounters is the a difference in the amount of mass entering and leaving as a result of these encounters . for a system where the distribution function depends only on energy ,
that is one whose phase space distribution is spherically symmetric , one can rewrite ( [ eq : collp ] ) as @xmath352 here @xmath62 is the mass flux travesing the energy surface @xmath61 .
since this does not have unit area , one divides by the phase space area of this surface , hence the factor @xmath353 .
the term on the left hand side of ( [ eq : kinetic ] ) can be written as @xmath354 which is the collisionless boltzman equation written in terms of the _ phase space _
velocity @xmath355 of a particle .
when the phase space is spherical that velocity is @xmath356 ; so that we have @xmath357 inserting this into ( [ eq : kinetic ] ) , taking into account ( [ eq : coll ] ) , one gets @xmath358 this can be rewritten , in terms of @xmath359 , as @xmath360 which , upon integration , gives @xmath361 for an isolated system in quasi - equilibrium the arbitrary flux @xmath362
. if there are no source or sinks of mass at the centre of the system then also @xmath363 .
this is the assumption we adopt here .
we nevertheless note that , strictly speaking , this can not not be true of the approximate scale free solutions presented in section [ sec : scalefree ] ; since , for these cases , the flux diverges as @xmath152 .
one therefore has to assume a small core that breaks the scale free solutions at the centre . in practice
this is naturally realised when the finite size of the clumps is taken into account .
under the zero central flux assumption we have @xmath364
the rate of energy change of a population of particles in an interval @xmath92 $ ] can be related to the rate of mass change inside that interval by @xmath365 or , in terms of the total mass and energy inside surface @xmath61 , @xmath366 integration ( by parts on the right hand side and setting arbitrary time dependent functions and central fluxes to zero ) gives @xmath367 which , when used in conjunction with ( [ eq : dmass ] ) yields the required energy change .
note that , in moving from the local energy @xmath368 to the integrated one @xmath369 , we have neglected the interaction potential of the particles , that is we have simply added their energy .
this is consistent with the context of the fokker - planck formulation , since the assumption of locality in relation to the encounters producing the energy changes ( c.f .
section [ sec : fp ] ) already incorporates the idea that the evolution is cuased by encounters which proceed independenly of the form of the mean field , which is responsible for this interaction .
this is no longer true if singificant energy changes are caused by an evolving mean field .
however , for our purposes , this point is not of central importance ; the rationale being that we assume that the system is in quasiquilibrium and evolves only due to the encounters .
the source of any evolution is then the fokker - planck flux .
in general , diffusion coefficients represent the averge rates that a certain quantity changes over time . in case of the isotropic orbit averaged fokker - planck equation
they represent the average rate of change of powers of the specific energy @xmath61 , averaged all orbits with energies in the interval @xmath370 $ ] .
( note that , as always with this equation , this involves the assumption that the actual rate is such that the change during an orbital time is small , thus permiting the averaging of quantities over orbits ) . in the weak encounter ( fokker - planck ) approximation ,
two diffusion coeficients are relavant : @xmath371 and @xmath372 . for a test particle of mass @xmath373 affected by encounters with field particles of mass @xmath374 and distribution function @xmath375 ,
they are given by ( spitzer 1987 ; theuns 1996 ) @xmath376 and @xmath377 where @xmath100 , @xmath378 being the coulomb logarithm ; @xmath379 is given by ( [ eq : q ] ) and @xmath380 ( note that in this paper we are assuming that the zero of the potential is taken so that energies are always positive ) . in a system consisting of several mass species
, field particles will include all other particles of the same species , plus those of the other species present .
the form of the diffusion coeficients is additive in the distribution of these different species , it is thus easy to generalise relations obtained for a two component system .
we therefore consider such a system ; with constituents having masses @xmath381 and @xmath382 . the phase space _ mass _ distribution functions are @xmath383 and @xmath384 .
we will assume that the phase space mass densities of the two components are comparable ; that is @xmath102 and @xmath266 are roughly of the same order , while at the same time @xmath94 .
for the average energy change of the light particles one therefore has @xmath385 while for the massive species it is @xmath386 where @xmath387 . | simple simulations suggest that the phase space structure of haloes identified in cosmological calculations is invariant under the dynamics induced by sinking substructure satellites the background expands so as to leave the total distribution unchanged .
we use a fokker - planck formulation to show that there are long lived solutions for densities @xmath0 and @xmath1 ; indices between @xmath2 and @xmath3 corresponding to the inner cusps of cosmological haloes , where the coupling is strongest ; steeper ones to intermediate radii .
we recover the exact solutions found by evans & collett ( 1997 ) ; reinterpret them in terms of well defined background - satellite interaction ; and show that these , and all other solutions , are valid for any mass spectrum of substructure , because the governing equation is linear in their mass weighed phase space distribution .
if the spatial distribution of substructure has a milder cusp than the total , the system expands ; when the background has a milder cusp there is compression .
it is not possible for the individual distributions to retain their original form : light particles are driven out of low energy states , being replaced by the sinking massive ones . if the clumps are considered solid , this takes the form of an exponential instability , with characteristic timescale of the order of the dynamical friction time , leading to a low energy cutoff in the distribution function of the background and a constant density core .
we show that there are long lived solutions with such a cutoff .
they would correspond to a situation whereas the clumps are made of dense baryonic material .
when stripping is important , as in the case of dissipationless substructure , it is likely that this situation is reversed the cutoff is now in the clump distribution function .
a simple description suggests that this renders equilibria even more long lived . in all cases it is possible to find solutions that are long lived from the thermodynamical ( energy transfer ) perspective . in systems without stripping the only truly stable solutions however are isothermal spheres , but there are double power law solutions that may be relevant if stripping is involved .
the results in this paper suggest that halo profiles similar to those found in dissipationless cosmological simulations are approximately invariant under the interaction induced by the presence of substructure satellites a necessary condition for the observed ` universality ' .
in addition , the total profile , including baryons , should also be invariant ; provided the latter are initially in the form of dense clumps , whose distribution follows that of the dark matter .
3mark iii # 1#2#3#4=.24 = .24 = .24 = .24 c dark matter galaxies : haloes diffusion galaxies : general galaxies : formation galaxies : structure |
it is an important subject in nuclear physics to understand the nuclear structure from the viewpoint of the nucleon - nucleon ( @xmath9 ) interaction .
the @xmath9 interaction has strong tensor forces at long and intermediate distances caused by the pion exchange and strong central repulsion at short distance caused by the quark dynamics .
it is important to investigate the nuclear structure by treating these characteristics of the @xmath9 interaction @xcite .
recently , it became possible to calculate nuclei up to mass around @xmath10 using a @xmath9 interaction with the green s function monte carlo method ( gfmc ) @xcite . at present , this method requires the extreme computational time to be applied to heavier nuclei .
it is desired to develop a new method to calculate nuclear structure with large nucleon numbers by taking care of the characteristic features of the @xmath9 interaction .
the presence of the tensor force in the @xmath9 interaction induces @xmath11-wave components in a nucleus , in particular , for proton - neutron ( @xmath12 ) pairs as the deuteron .
the @xmath11-wave component in the deuteron makes the system bound via the large @xmath13-@xmath11 coupling of the tensor force .
this @xmath11-wave component is also found to be spatially compact as compared with the @xmath13-wave component due to large momentum characters brought by the tensor force @xcite .
this effect originates from the pseudo - scalar nature of the one - pion exchange .
it is reported experimentally that a large fraction of @xmath12 pairs is observed as compared with @xmath14 or @xmath15 pairs in light nuclei @xcite .
this enhancement of @xmath12 pairs is hard to reproduce theoretically in a simple shell model ( mean - field picture ) @xcite except for the rigorous method as the one of gfmc @xcite , which treats the tensor force explicitly .
it is important to make efforts to study high momentum components caused by the tensor force in finite nuclei @xcite .
there are two important developments to perform nuclear structure calculations in heavy nuclei by including the necessary dynamics induced by the bare @xmath9 interaction .
one development is to find out that the strong tensor force is of intermediate range and we are able to express the tensor correlation in a reasonable shell model space @xcite .
we name this method as tensor optimized shell model ( tosm ) .
the other is the unitary correlation operator method ( ucom ) to treat the short - range correlation caused by the short - range repulsion @xcite
. we shall combine these two methods , tosm and ucom , to describe nuclei using a bare @xmath9 interaction . in the tosm part
, the wave function is constructed in terms of the shell model basis states with full optimization of two particle - two hole ( @xmath16 ) states .
there is no truncation of the particle states within the @xmath16 space in tosm , where the spatial shrinkage of the particle states is essential to obtain convergence of the tensor contribution involving high momentum components @xcite .
this treatment of the bare tensor force in tosm corresponds to the one - pair approximation correlated by the tensor force @xcite . in a few - body framework ,
the validity of tosm was confirmed by taking only the @xmath17-wave component connected with the @xmath18-wave state directly by the tensor force for the @xmath13-shell nuclei @xcite .
their calculation reproduces more than 90% of the tensor contribution and this result justifies the description of the tensor correlation in tosm .
the explicit inclusion of @xmath16 states in the extended mean field model for heavy nuclei has been formulated by ogawa _
_ as an extended brueckner hartree - fock theory @xcite .
so far , we obtained various successful results of tosm for the investigation of the tensor correlations in he isotopes . in @xmath19he
, we showed the selectivity of @xmath20 configuration in the @xmath16 space with the @xmath12 pair induced by the tensor force and this correlation was recognized as the deuteron - like state @xcite .
this specific @xmath16 excitation plays a decisive role to reproduce the spectra of neutron - rich he isotopes @xcite .
it is found that the @xmath2 occupation of extra neutrons increases the tensor correlation of nuclei from @xmath3he to @xmath21he , while the @xmath1 occupation of extra neutrons decrease the tensor correlation of those nuclei due to the pauli blocking between the specific @xmath16 excitations by the tensor force in @xmath19he and the motions of the extra @xmath1-neutrons .
this configuration dependence of the tensor correlation produces the right amount of splitting energy between the @xmath2- and @xmath1-dominant states in the he isotopes . in this paper
, we use the tosm+ucom for the study of the li isotopes from @xmath3li to @xmath22li , and discuss their structures focusing on the role of the tensor force on the excitation energies and configurations of those nuclei . before the present analysis , for neutron - rich @xmath23li and @xmath24li ,
we have performed the coupled two- and three - body cluster model analyses assuming the @xmath22li core described using the simplified tosm wave function , respectively @xcite .
the cluster model for two nuclei precisely describes the neutron wave functions in a loosely bound state around @xmath22li , such as a neutron halo structure in @xmath24li .
similar to the case of the he isotopes , we have shown that @xmath22li has the amount of the selected excitation from the @xmath8 orbit to the @xmath25 orbit by the tensor force , and the @xmath1 occupation of last one and two neutrons in @xmath23li and @xmath24li are blocked by this tensor correlation in @xmath22li . as a result , the @xmath26 occupation of last neutrons is relatively enhanced in two nuclei , and the virtual @xmath13-state in @xmath23li , and the large @xmath13-wave mixing and a neutron halo formation in @xmath24li are naturally explained .
on the basis of these successful results for two nuclei , in the present study , we perform a full microscopic analysis by using the tosm+ucom approach using the @xmath9 interaction for the li isotopes .
we investigate the systematic role of the tensor force in @xmath27li .
it is interesting to see the effect of tensor force on the structures of @xmath28li @xcite . in @xmath4li ,
the isospin @xmath29 and @xmath30 states coexist in the low excitation energy region .
the influence of the tensor force on two kinds of states in @xmath4li are important to understand the isospin dependence of @xmath4li in relation with the tensor correlation . in heavier li isotopes ,
the contribution of tensor force is investigated in the ground and excited states .
we also see the high momentum behavior of the kinetic energy in each state in relation with the tensor correlation .
the present subject of li isotopes becomes a foundation of the nuclear structure study using tosm+ucom and also of the previous study of @xmath24li and @xmath23li structures @xcite . in sec .
[ sec : model ] , we explain the method of the tosm+ucom approach . in sec . [
sec : result ] , we show the level structures of li isotopes and discuss their characteristics in relation with the tensor force .
a summary is given in sec .
[ sec : summary ] .
we explain the tensor optimized shell model ( tosm ) for open shell nuclei .
we begin with writing a many - body hamiltonian for a @xmath31 body system as , @xmath32 here , @xmath33 is the kinetic energy of each nucleon with @xmath34 being the center of mass kinetic energy .
we take a bare interaction @xmath35 such as av8@xmath0 @xcite consisting of central @xmath36 , tensor @xmath37 and spin - orbit @xmath38 terms and @xmath39 the coulomb term .
we obtain the many - body wave function @xmath40 by solving the schrdinger equation @xmath41 . in our previous works for @xmath42he and a few body systems
, we found that the tensor force can be described by taking @xmath16 excitations with large momentum components in the shell - model framework @xcite .
we prepare first a standard shell - model state for an open shell li nucleus with @xmath31 nucleons in order to introduce the tosm for open shell nuclei .
the standard shell - model state @xmath43 is written as _
s=_k_s a_k_s|(0s)^4(0p)^a-4;k_s . here , the @xmath44 shell is the valence shell and the index @xmath45 is used to distinguish various shell - model components .
the spirit of the tosm is that the tensor force works strongly for two nucleons in the standard shell - model state and excite two nucleons to various two particle states with high momentum components .
hence , we limit configurations up to two particle - two hole excitations from the standard shell - model state .
we extend therefore first the standard shell - model state by allowing two particles in the @xmath13 shell to excite into the @xmath44 shell .
we name these extended shell - model states as @xmath46 states , since no particle is excited into shell - model orbits outside of the @xmath47 shells .
we write the extended shell - model states as = |(0s)^n_s(0p)^n_p;k_0with the constraints @xmath48 and @xmath49 .
@xmath50 distinguishes various @xmath47 shell configurations .
we take now all the necessary high momentum components brought by the strong tensor force .
these components are included by exciting two nucleons from the @xmath47 shells to higher shells and therefore limit these configurations to @xmath16 states from the standard shell - model state @xmath43 .
hence , we have = |(0s)^n_s(0p)^n_p(higher)^2;k_2with the constraint @xmath51 and @xmath49 .
we introduce quantum numbers @xmath52 to distinguish various @xmath16 states , which amount to a large number of configurations . here , `` _ higher _ '' indicates higher shells outside of the @xmath47 shells , which are treated as particle states .
hence , we use the notation of @xmath16 states to specify that two particles are in higher shells outside of the @xmath47 shells .
we allow also @xmath53 excitations for shell model consistency .
= |(0s)^n_s(0p)^n_p(higher)^1;k_1with the constraint @xmath54 and @xmath49 .
these @xmath53 states are able to include high momentum components and improve the @xmath55 wave functions in the presence of the strong tensor correlation .
altogether , we write the tosm wave function @xmath40 for open shell nuclei as @xmath56 here , all the amplitudes @xmath57 are variational coefficients to be fixed by the energy minimization .
we write now the details of the radial wave functions .
the @xmath55 states are shell model states and expressed in terms of harmonic oscillator wave functions .
hence , the @xmath8 and @xmath58 shell - model wave functions are used to express the radial wave functions , whose length parameters are taken independently as variational parameters . the @xmath53 and @xmath16 states involve particle states with high momentum components .
the hole states correspond to the shell - model states in the @xmath47 shells .
the particle wave functions have to contain high momentum components to express the specific characters of the tensor force with all the possible angular momenta until the total energy converges . for particle states , we employ the gaussian wave functions to express single - particle states in higher orbits in order to describe high momentum properties due to the tensor force @xcite .
we prefer the gaussian wave functions over the shell - model states in order to effectively include the necessary high momentum components @xcite .
when we superpose a sufficient number of gaussian wave functions , the radial components of the particle states can be fully expressed .
gaussian basis states should be orthogonalized to the hole states and among themselves .
this condition is imposed by using the gram - schmidt orthonormalization . in order to use the non - orthogonal gaussian basis functions in the shell model framework , we construct the following orthonormalized single - particle basis function @xmath59 using a linear combination of gaussian bases @xmath60 with length parameter @xmath61 .
@xmath62 where @xmath63 is a number of basis functions for the orbit @xmath64 , and @xmath65 is an index to distinguish the bases with gaussian length of @xmath61 .
the explicit form of the gaussian basis function is written as @xmath66_j , \label{eq : gauss2 } \\
n_l(b_{\alpha,\nu } ) & = & \left [ \frac{2\ b_{\alpha,\nu}^{-(2l+3 ) } } { \gamma(l+3/2)}\right]^{\frac12},\end{aligned}\ ] ] where @xmath67 and @xmath68 are the orbital and total angular momenta of the basis states , respectively .
the weight coefficients @xmath69 are determined to satisfy the overlap condition in eq .
( [ eq : gauss3 ] ) .
this is done by using the gram - schmidt orthonormalization . following this procedure , we obtain the new single - particle basis states @xmath70 in eq .
( [ eq : gauss1 ] ) used in tosm . the particle states in @xmath53 and @xmath16 states are prepared to specify the basis wave functions , whose amplitudes are determined by the variational principle .
we note that we may use another method to obtain @xmath70 by solving the eigenvalue problem of the norm matrix of the gaussian basis set in eq .
( [ eq : gauss2 ] ) with the dimension @xmath63 .
this method gives different coefficients @xmath69 from those of the gram - schmidt method .
however , these two methods of making the orthonormalized single - particle basis states provide equivalent variational solutions for the total tosm wave function @xmath40 in eq .
( [ eq : config ] ) , because we start from the same gaussian basis functions in eq .
( [ eq : gauss2 ] ) with the same number @xmath63 of gaussian basis states .
therefore , the tosm solution does not depend on the construction of the orthonormal basis states when we start from the gaussian basis functions .
we construct gaussian basis functions of particle states to be orthogonal to the occupied states . for the @xmath26 orbit in the particle states , we prepare an extended @xmath26 basis function @xcite , which is orthogonal to the @xmath8 state and possesses a length parameter @xmath71 that can differ from @xmath72 of the @xmath8 state . in the extended @xmath26 basis functions , the polynomial part is changed from the usual @xmath26 basis states to satisfy the conditions of the normalization and the orthogonality to the @xmath8 state @xcite . for the @xmath73 orbits ,
we take the same method as used for the @xmath26 case . in the numerical calculation
, we prepare 10 gaussian basis functions at most with various range parameters to get a convergence of the energy and hamiltonian components .
we note here that when we write probabilities and occupation numbers of each orbit in various states in the numerical sections , those numbers are given by the summation of all the orthogonal orbits with the same spin having different radial behaviors due to the fact these wave functions are constructed by the orthonormalization . for hole states
, we do not have to sum up their numbers because the hole states are described by the single harmonic oscillator basis states .
we take care of the center - of - mass excitations by using the well - tested method of introducing a hamiltonian of center - of - mass motion in the many - body hamiltonian known as the lawson method @xcite . in the present study
, we take the value of @xmath74 for the center of mass motion as the averaged one used in the @xmath8 and @xmath58 orbits in the @xmath55 states with the weight of the occupation numbers in each orbit @xcite . adding this center of mass hamiltonian as the lagrange multiplier to the original hamiltonian in eq .
( [ eq : ham ] ) , we can effectively project out only the lowest harmonic oscilattor state for the center - of - mass motion . the variation of the energy expectation value with respect to the total wave function @xmath40 in eq . ( [ eq : config ] ) is given by @xmath75 which leads to the following equations : @xmath76 the total energy is represented by @xmath77 .
we solve two variational equations in eqs .
( [ eq : vari1 ] ) and ( [ eq : vari2 ] ) in the following steps . first
, fixing the length parameters @xmath61 and the partial waves of the basis states up to @xmath78 , we solve the linear equation for @xmath79 as an eigenvalue problem for @xmath80 .
we thereby obtain the eigenvalue @xmath77 , which is functions of @xmath81 and @xmath78 .
we try to adopt various sets of the length parameters @xmath81 and increase @xmath78 in order to find a better solution which minimizes the total energy @xmath77 . in tosm , we can describe the spatial shrinkage of particle states with an appropriate radial form in the individual configuration , which is important to describe the tensor correlation @xcite , as seen in the deuteron .
we briefly explain ucom for the short - range central correlation @xcite , in which the following unitary operator @xmath82 is introduced @xmath83 we express the correlated wave function @xmath40 in terms of a simple wave function @xmath84 as @xmath85 .
the transformed schrdinger equation becomes @xmath86 where the transformed hamiltonian is given as @xmath87 .
the operator @xmath82 is in principle a many - body operator . in case of the short - range correlation , we are able to truncate the modified operators at the level of two - body operators @xcite .
two - body hermite operator @xmath88 in eq . (
[ eq : ucom ] ) is defined as @xmath89 where the operator @xmath90 is the radial component of the relative momentum and is conjugate to the relative coordinate @xmath91 .
the function @xmath92 expresses the amount of the shift of the relative wave function at the relative coordinate @xmath91 for every nucleon pair in nuclei .
we use the tosm basis states to describe @xmath84 . in ucom , the function @xmath92 is determined variationally to minimize the total energy of the system .
we parametrize @xmath92 in the same manner as proposed by feldmeier and neff @xcite for four channels of spin - isospin pair .
the detailed forms of @xmath92 and their parametrization are explained in the previous paper @xcite . in the present analysis
, we use these @xmath92 functions commonly for every states of the li isotopes . to simplify the numerical calculation
, we adopt the ordinary ucom for the central correlation part instead of the @xmath18-ucom in this analysis .
we explain first the results of @xmath19he calculated by using tosm+ucom .
we then discuss how the structures of @xmath27li behave by comparing with the properties of @xmath19he .
we show results of @xmath19he using the av8@xmath0 interaction , which consists of central , @xmath5 and tensor terms .
the av8@xmath93 interaction is used in the calculation given by kamada _
et al _ , where the coulomb term is ignored @xcite .
the total energy of @xmath19he is obtained as @xmath94 mev in tosm+ucom .
the convergence of the solutions of @xmath19he has already been confirmed in the previous works @xcite .
the energy components of @xmath19he are discussed in the previous paper @xcite c|r @xmath95 & 84.14 + @xmath96 & 2.32 + @xmath97_{10}$ ] & 2.20 + @xmath98_{10}$ ] & 1.82 + @xmath99_{10}$ ] & 1.21 + & 8.31 + [ conf4 ] the dominant configurations of @xmath19he in tosm are listed in table [ conf4 ] with their probabilities .
the specific @xmath16 states such as @xmath96 show large probabilities and are essential to produce the tensor correlation in @xmath19he .
the selectivity of those @xmath16 configurations can be understood from the coupling by the tensor operator which changes both the orbital angular momentum and spin by two with the opposite direction of their @xmath100-components @xcite .
in addition , the @xmath16 state with @xmath97_{10}$ ] is also favored by the tensor interaction .
it is found that the dominant @xmath16 states commonly correspond to the excitations of a @xmath12 pair .
those features of the @xmath16 excitations play an important role to determine the structures of @xmath27li as will be discussed later . for comparison , we show the results with the minnesota ( mn ) @xmath9 interaction , which does not have the tensor force .
the mn interaction consists of the central and @xmath5 parts with @xmath101 parameter being 0.95 @xcite .
we use set iii of the mn interaction for the @xmath5 force .
tosm has been confirmed to reproduce the binding energy and the radius of @xmath19he obtained by few - body stocastic variational method@xcite .
it is obtained that the @xmath102 configuration is by far dominant in the @xmath19he wave function by 96.6% .
other @xmath16 configurations have the probability less than a percent .
this comparison makes the role of the tensor force clear on the energy spectra of the @xmath44-shell nuclei for the case of the av8@xmath0 interaction .
we also list the occupation numbers of nucleon in @xmath19he using av8@xmath0 in table [ occ4 ] .
it is shown that the @xmath1 orbit has the largest contribution among the particle states according to the large @xmath16 mixing including the @xmath1 component . in the mn case
, it is found that the component of the @xmath8 orbit is larger than the av8@xmath0 case and the enhancement of the @xmath1 orbit is not obtained .
these results mean that the tensor force brings the specific excitations from the @xmath13-shell to the @xmath44- and @xmath103-shells in @xmath19he .
c|cccccccc @xmath19he@xmath104 & @xmath105 & @xmath1 & @xmath2 & @xmath106 & @xmath107 & @xmath108 + av8@xmath0 & 3.77 & 0.06 & 0.04 & 0.03 & 0.04 & 0.01 + mn & 3.94 & 0.01 & 0.03 & 0.01 & 0.004 & 0.006 + [ occ4 ] .,width=321 ] with tosm+ucom.,width=302 ] .matter radii of li isotopes in comparison with the experiments ; a@xcite , b@xcite .
units are in fm . [ cols=">,<,<",options="header " , ] seeing the whole structures from @xmath4li to @xmath22li , it is found that their ground states possesses the largest tensor energies , in which the configurations of @xmath58-shell are constructed to activate the two - kinds of excitations by the tensor force ; one is from @xmath8-orbit to the @xmath1 orbit and the other is from @xmath58-shell to higher shell such as @xmath103-shell . in their ground states ,
the @xmath25 component in their most dominant configurations is included only in @xmath4li .
other heavier systems contain the @xmath109 configuration largely , which is the @xmath6-like one , to increase the tensor correlation from the @xmath8 orbit to the @xmath25 orbit . from this result ,
the @xmath4li nucleus is a specific one and this is considered to be related to the @xmath64+@xmath11 component in this nucleus , as was mentioned .
we have developed a method to describe nuclei with bare @xmath9 interaction on the basis of the tensor optimized shell model with the unitary correlation operator method , tosm+ucom .
we have treated the tensor force in terms of tosm , in which @xmath16 states are fully optimized to describe the deuteron - like tensor correlation .
the short - range repulsion in the @xmath9 interaction is treated by using the central correlation part of ucom .
we have shown the reliability of tosm+ucom using the av8@xmath0 interaction to investigate the structures of the li isotopes .
it is found that the excitation energy spectra are found quite consistent with the experimental spectra .
when we employ the effective minnesota interaction consisting of only the central and @xmath5 parts , the excitation energy spectra show quite a large deviation from experiment . it has been found that @xmath19he contains a relatively large amount of the @xmath12 pair in the @xmath1 orbit due to the tensor force .
this characteristics of the tensor force produces the state - dependence in the li isotopes . in @xmath3li , the @xmath110 state gains more tensor energy than the @xmath111 case .
the enhancement of the kinetic energy is also observed in the @xmath110 state of @xmath3li , which is brought by the tensor force . as a result
, tensor force dynamically produces the state dependence in @xmath3li and contributes to the @xmath5 splitting energy in @xmath3li . in @xmath4li
, the tensor force also makes the large mixing of the spin - triplet component in the configuration of the last @xmath12 pair in the ground state , which shows the @xmath5 coupling scheme .
this component increases the tensor energy and is related to the @xmath64+@xmath11 clustering configuration in @xmath4li .
this characteristics can not be seen in the minnesota interaction without the tensor force , which shows rather the conventional @xmath6-coupling scheme . in @xmath7li ,
the @xmath6-like configurations , which include more @xmath112 states than the @xmath25 states , can gain the tensor energy , because of the allowance of the @xmath12-pair excitation from the @xmath8-orbit to @xmath25 orbit in the @xmath19he core part . as a conclusion of the roles of the tensor force in the li isotopes , the tensor energy depends on the configurations of nucleons occupied mainly in @xmath58-shell in each state .
this property is related with the amount of the excitation of @xmath12 pair from @xmath8-shell by the tensor force .
the amount of the high momentum component also depends on the tensor energy in each state .
observation of the high momentum component experimentally in finite nuclei is desired in order to confirm the existence of the strong tensor correlation @xcite .
the tensor force would play important roles on various physical quantities such as charge radii , quadrupole and magnetic moments@xcite in addition to the energy spectra . in our previous study of @xmath24li ,
we discuss those quantities of @xmath24li in detail @xcite .
it is interesting to discuss the effect of the tensor force on these quantities in other li isotopes using the tosm+ucom method .
we thank professor hisashi horiuchi for fruitful discussions and continuous encouragement . this work was supported by a grant - in - aid for young scientists from the japan society for the promotion of science ( no .
21740194 ) and also jsps ( no .
21540267 ) .
numerical calculations were performed on a computer system at rcnp , osaka university .
00 y. akaishi , _ cluster models and other topics _ , international review of nuclear physics 4 ( world scientific , singapore , 1986 ) pp.259 - 394 . h. kamada et al . , , and the references therein .
k. arai , s. aoyama , y. suzuki , p. descouvemont , and d. baye , .
s.c . pieper and r. b. wiringa , .
b. s. pudliner , v. r. pandharipande , j. carlson , s. c. pieper and r. b. wiringa , . k. ikeda , t. myo , k. kat and h. toki , _ clusters in nuclei _
, lecture notes in physics 818 vol.1 , ( springer , berlin , 2010 ) , pp.165 - 221 .
r. subedi et al . .
e. c. simpson and j. a. tostevin , .
r. schiavilla , r. b. wiringa , s. c. pieper and j. carlson , .
i. tanihata , cern proceedings cern - proceedings-2010 - 001 , 31 ( 2010 ) .
t. myo , s. sugimoto , k. kat , h. toki and k. ikeda , .
t. myo , h. toki and k. ikeda , .
t. myo , a. umeya , h. toki and k. ikeda , .
h. feldmeier , t. neff , r. roth and j. schnack , .
t. neff and h. feldmeier , .
r. roth , t. neff and h. feldmeier .
h. toki , s. sugimoto and k. ikeda , .
s. sugimoto , k. ikeda and h. toki , .
y. ogawa , h. toki , s. tamenaga , s. sugimoto and k. ikeda , .
t. togashi and k. kat , .
k. horii , h. toki , t. myo and k. ikeda , .
y. ogawa and h. toki , t. myo , k. kat , k. ikeda , .
t. myo , k. kat , h. toki and k. ikeda , .
t. myo , k. kat , h. toki and k. ikeda , .
t. myo , y. kikuchi , k. kat , h. toki and k. ikeda , .
k. arai , y. suzuki , and k. varga , .
y. kikuchi , n. kurihara , a. wano , k. kato , t. myo , and m. takashina , .
e. hiyama , y. kino and m. kamimura , . s. aoyama , t. myo , k. kat and k. ikeda , .
gloeckner , r.d .
y. c. tang , m. lemere and d. r. thompson , .
i. reichstein and y.c .
k. varga and y. suzuki , .
y. suzuki , w. horiuchi , m. orabi and k. arai , .
i. tanihata _ et al . _ , .
g. d. alkhazov _ et al .
_ , . o. a. kiselev _
et al . _ ,
j. a*25 * , suppl . 1 ( 2005 ) 215 .
k. m. nollett , s. c. pieper , r. b. wiringa , j. carlson , and g. m. hale , . t. terasawa , and .
a. arima and t. terasawa , .
s. nagata , t. sasakawa , t. sawada and r. tamagaki , .
et al . _ , .
w. nrtershuser , t. neff , r. snchez , and i. sick , . | we study the li isotopes systematically in terms of the tensor - optimized shell model ( tosm ) by using a bare nucleon - nucleon interaction as the av8@xmath0 interaction .
the short - range correlation is treated in the unitary correlation operator method ( ucom ) . using the tosm+ucom approach
, we investigate the role of the tensor force on each spectrum of the li isotopes .
it is found that the tensor force produces quite a characteristic effect on various states in each spectrum and those spectra are affected considerably by the tensor force . the energy difference between the spin - orbit partner , the @xmath1 and @xmath2 orbits of the last neutron , in @xmath3li
is caused by opposite roles of the tensor correlation . in @xmath4li , the spin - triplet state in the @xmath5 coupling configuration
is favored energetically by the tensor force in comparison with @xmath6 coupling shell model states . in @xmath7li ,
the low - lying states containing extra neutrons in the @xmath2 orbit are favored energetically due to the large tensor contribution to allow the excitation from the @xmath8-orbit to the @xmath1 orbit by the tensor force .
those three nuclei show the @xmath6 coupling character in their ground states which is different from @xmath4li . |
in recent years , the kolmogorov postulate on local isotropy ( pli ) has been examined by means of laboratory and numerical experiments ( @xcite@xcite ) .
the current phenomenological interpretation of real turbulent flows is based on this postulate , which maintains that turbulence in fluids can achieve a universal state at small scales where fluctuations are supposed to become statistically isotropic , and as a consequence universal .
the contravention of this postulate implies that small scales in a turbulent system interact directly with larger scales .
thus , a sort of long - term interaction occurs and is able to induce anisotropy on the small scales , according to the symmetry properties of the system .
direct communication between large and small scales at the same time undermines the concept of small - scale universality and that of energy cascade .
the problem remains on how to examine the effects of large - scale anisotropy on small scales .
the answer is to examine statistics that are sensitive to flow anisotropy .
the main way of verifying the pli is based on turbulence studies at very high reynolds numbers ( @xmath0 and greater ) .
for example , a popular way of measuring small - scales anisotropies is to obtain the @xmath1 dependence of the isotropic and anisotropic statistical observables on the basis of velocity gradients . at different instants .
the second interaction layer is not visible in these images .
( c ) initial three - dimensional normalized spectra @xmath2 in homogeneous regions 1 and 2 .
dimensional reference data for the experiment : velocity fluctuation @xmath3 , kinetic energy @xmath4 j / kg , molecular viscosity @xmath5 m@xmath6/s , size of the flow domain @xmath7 m ( dimension along x ) , @xmath8 m ( dimension along @xmath9 and @xmath10 ) , correlation lengths : @xmath11 m , field 1 , @xmath12 m , @xmath13 m ; @xmath14 m ; field 2 , time scales : @xmath15 s , field 1 , @xmath16 s , @xmath17 s , @xmath18 s ( for field 2 yielding , the scale ratio is : 2.8 , 2.1,1.5 , respectively ) . ,
title="fig:",scaledwidth=56.0% ] + at different instants .
the second interaction layer is not visible in these images .
( c ) initial three - dimensional normalized spectra @xmath2 in homogeneous regions 1 and 2 .
dimensional reference data for the experiment : velocity fluctuation @xmath3 , kinetic energy @xmath4 j / kg , molecular viscosity @xmath5 m@xmath6/s , size of the flow domain @xmath7 m ( dimension along x ) , @xmath8 m ( dimension along @xmath9 and @xmath10 ) , correlation lengths : @xmath11 m , field 1 , @xmath12 m , @xmath13 m ; @xmath14 m ; field 2 , time scales : @xmath15 s , field 1 , @xmath16 s , @xmath17 s , @xmath18 s ( for field 2 yielding , the scale ratio is : 2.8 , 2.1,1.5 , respectively ) . , title="fig:",scaledwidth=56.0% ] -2 mm -5 mm , where @xmath19 and @xmath20 are two homogeneous and isotropic fields with integral scales @xmath21 and @xmath22 , @xmath23/2 $ ] is a smoothing function ; @xmath24 is a correction introduced to obtain a uniform energy distribution : @xmath25 is a spatial modulation in the smoothing region and @xmath26 is a random field with zero mean and uniform variance @xmath27 .
the value of @xmath27 is chosen a posteriori in order to generate a uniform energy level @xmath28 .
the coefficient _ a _ in function p(x ) is equal to 11 if the matching variable is normalized on the correlation length @xmath21 ( @xmath29 ) and is equal to 400 if the matching variable is normalized on @xmath30 ( @xmath31 ) .
( b ) initial distribution of the average turbulent kinetic energy . , title="fig:",scaledwidth=49.0% ] , where @xmath19 and @xmath20 are two homogeneous and isotropic fields with integral scales @xmath21 and @xmath22
, @xmath23/2 $ ] is a smoothing function ; @xmath24 is a correction introduced to obtain a uniform energy distribution : @xmath25 is a spatial modulation in the smoothing region and @xmath26 is a random field with zero mean and uniform variance @xmath27 . the value of @xmath27 is chosen a posteriori in order to generate a uniform energy level @xmath28 .
the coefficient _ a _ in function p(x ) is equal to 11 if the matching variable is normalized on the correlation length @xmath21 ( @xmath29 ) and is equal to 400 if the matching variable is normalized on @xmath30 ( @xmath31 ) .
( b ) initial distribution of the average turbulent kinetic energy .
, title="fig:",scaledwidth=49.0% ] -1 mm the relationship between intermittency and small - scale anisotropy is considered to determine whether small - scale isotropy is restored at very high reynolds numbers .
the set - up often used to test the return to isotropy is a homogeneous shear flow in which the large - scale mean velocity has a linear profile . in many previous works ( e.g. @xcite )
, researchers looked at the increments in the streamwise ( @xmath32 direction ) fluctuating velocity , @xmath33 , in the direction of the mean shear @xmath34 , where @xmath35 is the transverse direction and @xmath36 is the mean velocity . the effect of the reynolds number on turbulence small - scale anisotropy and intermittency is considered in an attempt to separate large and small scales , with respect to their size and energy , in order to reduce the probability of mutual communication , although taylor scale reynolds numbers of @xmath37 ( equivalent to a turbulence reynolds number based on the integral scale of @xmath38 ) can be achieved in the laboratory , it is argued @xcite that there is still a need to conduct experiments and computations at even higher reynolds numbers to resolve the outstanding basic and applied issues in turbulence . in our opinion , this is not the only approach that can be used to test the pli .
another approach is to focus on flows at a given taylor scale reynolds number and to slightly perturb the state of homogeneity and isotropy to see how and to what extent the perturbation is felt by the small scales during the transient .
this method is inspired by the initial value problem in stability theory @xcite , where the basic laminar flow is arbitrarily perturbed by an infinitesimal perturbation with a given wavenumber .
the mildest perturbation that can be conceived without forcing the turbulence ( i.e. without introducing a production of kinetic energy in the system ) is that of varying the correlation length of the turbulence in the space , while avoiding the generation of a mean shear .
as known , the coupling of shear with fluctuation is a source of kinetic energy . in order to achieve this situation
, we consider a numerical experiment in which we follow the time decay of a turbulent flow with an initially uniform distribution of turbulent kinetic energy , but where the integral scale has been varied slightly in two adjacent regions .
the two regions are characterized by different spectrum shapes in the low wavenumber range , one almost @xmath39 and the other close to @xmath40 @xcite .
thus , we have a condition where we can observe a different decay generated by the scale inhomogeneity .
this simple configuration allows us to investigate the generation of large - scale and small - scale anisotropy arising from the different decay rates of the two interacting isotropic fields , @xcite@xcite .
the different decay exponents shown by the two fields generate an energy step , and thus a layer where the flow is neither homogeneous nor isotropic .
thus , the system is characterized by the simultaneous presence of a scale and an energy gradient .
no mean velocity gradient is present .
no preferential direction is induced by the mean flow gradient @xcite .
the paper is organized as follows .
section 2 presents the dns method , the simulation conditions and the turbulence characteristics .
section 3 is devoted to the general lineaments of the temporal evolution of the interaction layer .
section 4 presents an analysis of the transient lifetime , in which the onset , growth , peaking and decay of anisotropy and of the kinetic energy gradient can be observed . scaling laws that link the velocity and velocity derivative skewness and kurtosis
are proposed , on the basis of best fitting of the results , produced by three simulations parametrized by the ratio of the correlation lengths of the two isotropic interacting fields ( @xmath41 ) .
we also discuss the difference with the anisotropy present in homogeneous shear flows .
the reynolds number , based on the taylor microscale of the interacting fields , varies from 70 to 150 , when the correlation length is varied from 0.8 to 2.8 cm .
section 5 provides the conclusions . , region 2 is the turbulence with the smaller correlation length .
the insert shows the initial reynolds number ( black symbols ) and decay exponent ( red symbols ) .
( b ) temporal variation of the ratio of the turbulent kinetic energy between the larger and smaller scale flows .
the inset shows the energy ratio at three different instants as function of the initial scale gradient imposed across the surface.,title="fig:",scaledwidth=60.0% ] + , region 2 is the turbulence with the smaller correlation length .
the insert shows the initial reynolds number ( black symbols ) and decay exponent ( red symbols ) .
( b ) temporal variation of the ratio of the turbulent kinetic energy between the larger and smaller scale flows .
the inset shows the energy ratio at three different instants as function of the initial scale gradient imposed across the surface.,title="fig:",scaledwidth=30.0% ] + , region 2 is the turbulence with the smaller correlation length .
the insert shows the initial reynolds number ( black symbols ) and decay exponent ( red symbols ) .
( b ) temporal variation of the ratio of the turbulent kinetic energy between the larger and smaller scale flows .
the inset shows the energy ratio at three different instants as function of the initial scale gradient imposed across the surface.,title="fig:",scaledwidth=60.0% ] + , region 2 is the turbulence with the smaller correlation length .
the insert shows the initial reynolds number ( black symbols ) and decay exponent ( red symbols ) .
( b ) temporal variation of the ratio of the turbulent kinetic energy between the larger and smaller scale flows .
the inset shows the energy ratio at three different instants as function of the initial scale gradient imposed across the surface.,title="fig:",scaledwidth=24.7% ] + , conventionally defined as the distance between the points where @xmath42 has values of 0.75 and 0.25 .
@xmath43 and @xmath44 are the energies in the two homogeneous regions .
time has been normalized with the average eddy turnover time @xmath45 .
comparison data from wind tunnel experiments by veeravalli and warhaft @xcite and from two - dimensional simulations @xcite are also shown.,title="fig:",scaledwidth=45.0% ] + , conventionally defined as the distance between the points where @xmath42 has values of 0.75 and 0.25 .
@xmath43 and @xmath44 are the energies in the two homogeneous regions .
time has been normalized with the average eddy turnover time @xmath45 .
comparison data from wind tunnel experiments by veeravalli and warhaft @xcite and from two - dimensional simulations @xcite are also shown.,title="fig:",scaledwidth=21.15% ] + -3 mm , conventionally defined as the distance between the points where @xmath42 has values of 0.75 and 0.25 .
@xmath43 and @xmath44 are the energies in the two homogeneous regions .
time has been normalized with the average eddy turnover time @xmath45 .
comparison data from wind tunnel experiments by veeravalli and warhaft @xcite and from two - dimensional simulations @xcite are also shown.,title="fig:",scaledwidth=45.0% ] , conventionally defined as the distance between the points where @xmath42 has values of 0.75 and 0.25 .
@xmath43 and @xmath44 are the energies in the two homogeneous regions .
time has been normalized with the average eddy turnover time @xmath45 .
comparison data from wind tunnel experiments by veeravalli and warhaft @xcite and from two - dimensional simulations @xcite are also shown.,title="fig:",scaledwidth=45.0% ] -1 mm with @xmath46 used to fit the whole data set .
data for a mixing with an imposed energy gradient ( @xmath47 ) from @xcite are also shown.,title="fig:",scaledwidth=48.0% ] with @xmath46 used to fit the whole data set .
data for a mixing with an imposed energy gradient ( @xmath47 ) from @xcite are also shown.,title="fig:",scaledwidth=48.0% ] + with @xmath46 used to fit the whole data set .
data for a mixing with an imposed energy gradient ( @xmath47 ) from @xcite are also shown.,title="fig:",scaledwidth=48.0% ] -4 mm in the inhomogeneous direction and @xmath48 in the homogeneous direction from the simulation with an initial scale ratio @xmath49 .
the horizontal lines indicate the measured values of the derivative moments in the larger - scale homogeneous region ( @xmath50 ) and in the smaller - scale homogeneous region ( @xmath51).,title="fig:",scaledwidth=51.0% ] in the inhomogeneous direction and @xmath48 in the homogeneous direction from the simulation with an initial scale ratio @xmath49 .
the horizontal lines indicate the measured values of the derivative moments in the larger - scale homogeneous region ( @xmath50 ) and in the smaller - scale homogeneous region ( @xmath51).,title="fig:",scaledwidth=51.0% ] + in the inhomogeneous direction and @xmath48 in the homogeneous direction from the simulation with an initial scale ratio @xmath49 .
the horizontal lines indicate the measured values of the derivative moments in the larger - scale homogeneous region ( @xmath50 ) and in the smaller - scale homogeneous region ( @xmath51).,title="fig:",scaledwidth=51.0% ] in the inhomogeneous direction and @xmath48 in the homogeneous direction from the simulation with an initial scale ratio @xmath49 .
the horizontal lines indicate the measured values of the derivative moments in the larger - scale homogeneous region ( @xmath50 ) and in the smaller - scale homogeneous region ( @xmath51).,title="fig:",scaledwidth=51.0% ] [ cols= " < " , ] + , empty symbols : longitudinal derivative along the homogeneous directions @xmath35 .
( a - c ) time evolution of the peak of the longitudinal derivative skewness and kurtosis inside the mixing layer .
( b - d ) ratio between the longitudinal moments ( inhomogeneous over homogeneous directions ) .
( e ) derivative kurtosis as a function of the derivative skewness .
the dashed lines show the power law scalings obtained by best fitting .
the fitting does not include the data obtained from the simulation with @xmath52 , which were not a sufficiently good statistical quality .
more runs would be needed to enlarge the statistical base and reduce the noise.,title="fig:",scaledwidth=50.0% ]
the scheme of the flow and of the computational domain can be seen in figure [ fig.schema ] , part _ a_. the initial conditions are obtained by matching two isotropic fields , obtained from simulations of homogeneous and isotropic turbulence , which have different correlation lengths .
these were obtained by using a high pass filter on the velocity field , see @xcite and the spectra in figure 1_c_. three simulations , with three different ratios of the initial integral scale , 1.5 , 2.1 and 2.8 , have been carried out .
the reynolds numbers , based on the taylor microscale of these fields , vary from 70 to 150 , with a variation of the correlation length from 0.8 to 2.8 cm @xcite .
the matching is made over a layer with an initial width of the same order as the correlation lengths @xmath53 @xcite .
the flow is assumed to be contained in a parallelepiped with an aspect ratio of 2 ( @xmath54 ) and cyclic boundary conditions are applied to all the spatial directions .
the navier - stokes equations are solved using a pseudo - spectral fourier - galerkin method with an explicit fourth - order time integration @xcite .
the initial condition is generated as @xmath55 , where @xmath19 and @xmath20 are the two homogeneous and isotropic fields with integral scales @xmath21 and @xmath22 , @xmath56/2 $ ] is the matching function and @xmath57 is a correction introduced to obtain a uniform energy distribution across the interaction layer , see part _ a _ of [ fig.matching ] .
this correction is the product of a spatial modulation of the matching region @xmath25 and a random field @xmath58 with zero mean and uniform variance @xmath27 .
the value of @xmath27 is chosen a posteriori in order to generate a uniform energy level @xmath28 throughout the initial matching layer , see part b ) of figure [ fig.matching ] .
length @xmath30 is the domain size in the @xmath32 direction and @xmath59 is a constant , which is chosen in order to have an initial transition layer that is no larger than the integral scale .
this kind of correction is not necessary in the experiments where the energy gradient is imposed by the initial conditions @xcite .
the turbulent field @xmath19 has a taylor microscale reynolds number equal to 150 and is the same in all the simulations , which differ according to the smaller - scale flow @xmath20 .
the three - dimensional spectra of the initial conditions are shown in figure [ fig.schema](c ) .
the flows have the same global energy and the different slopes of the large - scale range of the spectra generate different integral scales .
due to the use of periodic boundary conditions , there are two matching layers in the computational domain , see again part _ a _ of figure [ fig.schema ] .
directions @xmath9 and @xmath10 in this flow configuration remain statistically homogeneous during the decay , so that all the statistics can be computed using plane averages in these homogeneous directions .
in this experiment , we have a situation in which it can be observed that the introduction of an inhomogeneity of the correlation length in an otherwise homogeneous turbulence is sufficient to induce an inhomogeneity in the kinetic energy during the decay .
this is a general feature that can be observed in all the present simulations . in figure [ fig.schema ] ( b ) , it is possible to see the gradual time differentiation of the portions of two flows in the central region of the simulation . in this case , the integral scale ratio is 2.8 .
the flows with smaller integral scales decay faster and show higher decay exponents , see figure [ fig.decay-ratio ] .
the decay exponent of the larger - scale flow , whose low wavenumber spectrum has a @xmath39 slope , is -1.15 , a result which is close to saffman s prediction @xcite .
the other decay exponents are -1.3 , when the integral scale is reduced by 1.5 , -1.46 when it is reduced by 2.1 and -1.65 when is reduced by 2.8 , a situation where the low wavenumber spectrum approaches a @xmath40 slope ; see [ fig.schema](b ) and the inset in figure [ fig.decay-ratio ] ( _ a _ ) . as a consequence , the ratio between the energies in the two regions , initially equal to 1 , increases during the whole simulation ( figure [ fig.decay-ratio ] ( _ b _ ) ) . during the decay
, the large correlation length scale flow also becomes the most energetic one , and an energy gradient , always concurrent with the integral scale gradient , soon emerges , as shown in figure [ fig.gradient-width ] .
the concurrent presence of the gradients , the kinetic energy and the integral scale , in the absence of a mean shear , leads to intermittency and enhances turbulent transport @xcite , as discussed in the next section .
the time evolution of the kinetic energy gradient is plotted in figure [ fig.gradient-width ] ( _ a _ ) .
this gradient is maximum after about one initial eddy turnover time @xmath60 , which is defined on the initial scales of the field that has the largest correlation length ( @xmath43 is the initial turbulent kinetic energy and @xmath21 the correlation length ) .
this gradient then gradually decays because the difference between energy levels 1 and 2 also decays in time , see [ fig.decay-ratio ] ( _ a _ ) .
furthermore , the interaction length increases , see [ fig.gradient-width ] ( _ b _ ) .
it can be noticed that the morphology of this transient is very similar to a pattern that is often encountered when observing three - dimensional asymptotically stable linear perturbations in sheared steady flows @xcite .
here , however , the lifetime is much shorter than the typical lifetime of hydrodynamic stability , where this kind of transient can last hundreds , even thousands , of base flow timescales .
however , at the basic level , the generation - growth - decay - extinction cycle of intermittency can be considered common in real turbulent flows and representative at a local ( zonal ) regional level .
local perturbations , associated with more or less intense spatial and temporal variations of the correlation length , are in fact standard features in turbulence . in our opinion , they can be considered analogous to arbitrary perturbations in the hydrodynamic stability theory .
the interaction that we consider can be interpreted as a long - term interaction , since the two isotropic fields are initially separated by a layer which is as thick as the larger correlation length @xmath21 .
the first thing which should be noticed is the onset of a step variation of the turbulent energy from the initial flat distribution .
one is therefore tempted to derive a general notion of dynamics : a sufficient condition for producing gradients of energy by nonlinear interaction is the presence of variations of the correlation lengths .
the mechanism is associated with the different decay exponents that the turbulence shows in the first decade of eddy turnover times on the basis of different correlation lengths .
we should recall that the larger the correlation length , the smaller the decay exponent @xmath61 , @xmath62 .
a synthesis of the decay properties is presented in figure 3(a ) as a function of time and of the correlation scale @xmath63 . the inverse relationship between @xmath61 and the taylor microscale reynolds number
is shown in the inset .
the increase in the energy ratio across the interaction layer with time and with the gradient of the initial correlation length is presented in figure 3(b ) .
it should be mentioned that , at a constant instant , for instance @xmath64 , a doubling of this scale gradient produces a quadrupling of the energy ratio , see the inset .
the effect of this phenomenology is therefore very pronounced , and should be observable and measurable in the laboratory .
the interaction lifetime is finite and can be better described in terms of the energy gradient that starts to be generated at the beginning of the interaction , see figure 4 _ a _ and _ b_. the gradient intensity is proportional to the ratio ( gradient ) of the correlation length and the maximum value is reached in a range of time which is inversely proportional to the gradient intensity .
after the peak of the transient evolution is reached , a long decay phase appears , which is due , on one hand , to the growth in the width of the layer , and on the other , to the self - decay of the energy of the two isotropic fields . by extrapolating the instants in which the energy gradient becomes zero , on the basis of the results produced over the first 10 simulation timescales ( see in particular the log - linear version of the decay in figure 4(b ) )
, it is possible to see that the lifetime increases almost proportionally to the initial scale ratio ; see the inset in figure 4(a ) . this result can be understood and verified by observing the evolution in the diffusion thickness @xmath65 of the interaction layer . in figure 4 _ b _ , data from the present simulations are compared with data from a 3d reference simulation over the same computational domain ( the energy ratio is here 6.6 , with the two isotropic fields sharing the same integral scale @xcite ) , with data from one 2d simulation , with an energy ratio equal to 6.7 and a uniform integral scale @xcite , and with data from a wind tunnel laboratory experiments @xcite of shearfree turbulent mixings with a ratio of 6.7 .
the mixing self - diffusion thickness @xmath65 is conventionally defined as the distance between the points where @xmath42 has values of 0.75 and 0.25 .
@xmath43 and @xmath44 are the energy in the two homogeneous regions .
the time has been normalized to the average eddy turnover time @xmath45 .
we can observe that the diffusion lengths of the present simulations collapse very well with this normalization and grow in a slower manner ( @xmath66 ) than in the case where the initial energy gradient is nonzero .
data related to a case with an energy gradient of about 0.45 j/(kg m ) , that is , @xmath67 , are shown in figure 4 _
b_. @xmath68 and a good agreement between the simulation and laboratory results are observed in three dimensions .
there is therefore an evident damping of the turbulent self - diffusion when passing from two - dimensional to three - dimensional dynamics . in the latter case ,
an efficient driving is the initial presence of the energy gradient ; if this is lacking , the mild nonhomogeneity associated with the presence of an integral scale gradient still promotes diffusion , though at a lower rate . during the transient ,
the two isotropic flows interact under both the initial macroscale gradient and the newly developed energy gradient , and develop an intermittent sublayer which shows lineaments similar to those found in shearless mixings , with an initially imposed energy gradient that has already been discussed in @xcite . in two dimensions , where the inverse cascade is active , @xmath69 . it should be noted , that , at present , a comparison with laboratory results is not yet available for either the present or the 2d numerical experiment .
large - scale intermittency can be observed , looking at the velocity fluctuation @xmath33 in the @xmath32 direction , which is the component of the velocity vector that is responsible for the energy transport across the mixing .
one can observe that the position of this peak moves , with respect to the centre of @xmath65 , towards the small correlation length side , see @xcite .
the simulations start from a situation in which the statistics are closer to the isotropic ones , that is , a skewness 0 and a kurtosis close to 3 .
it is possible to observe the gradual generation of intermittency in fig.[fig.sk - vel ] .
it is interesting to note that , when the skewness and kurtosis values are plotted in the @xmath70 plane ( figure [ fig.sk-vel](c ) ) , all the three - dimensional simulations in part follow the same path , even through they individually present different growth rates and consequently tend to reach different levels of intermittency .
we observe a quadratic relationship @xmath71 in this plane . in the case in which the energy gradient drives the interaction from the very beginning , we can observe an accumulation point ( @xmath72 ) , which can be interpreted as an asymptotic temporal condition .
the intermittent behaviour is not limited to the large scales , i.e. the velocity moments .
in fact , after a few initial eddy turnover times , the longitudinal derivative skewness , which is a normalized dimensionless measure of the average rate of enstrophy generation by vortex stretching , is also increased beyond the typical isotropic turbulence range of values .
figure [ fig.sk-der ] shows an example of the spatial distribution of the longitudinal derivative skewness and kurtosis of the velocity component @xmath33 along the inhomogeneous direction @xmath32 , that is , @xmath73 and @xmath74 , in the most intermittent configuration , that is , the one in which the initial integral scale ratio is equal to 2.8 .
the most noticeable feature is that the negative value of the longitudinal derivative skewness , whose magnitude is linked to the spectral energy transfer towards small scales , reaches a magnitude above 0.5 , that is , a value which would only be present in isotropic turbulence at much higher reynolds numbers @xcite .
this occurs together with the increase in intermittency that is visible in the kurtosis distribution .
the longitudinal derivative moments in the directions normal to the interaction layer are also modified .
it should be noted that these directions are directions along which the field remains homogeneous .
if we look at @xmath75 and @xmath76 ( directions @xmath9 and @xmath10 are statistically equivalent ) , both the skewness and kurtosis show a moderate departure from the values of the neighbouring isotropic regions .
we can observe a moderate reduction in the magnitude of the skewness and an increase in the kurtosis ( figure [ fig.sk-der ] ) .
such variations of skewness and the derivative are not constant , as could be expected , because the large - scale features that generate the layer are time evolving .
their extreme values are an indication of the departure of the small scale turbulence from the isotropic state , which is observable even in this very mild instance of nonhomogeneity .
figures [ fig.peaks-der](a-c ) shows the time evolution of these peaks after the fluctuations in the statistics distributions visible in [ fig.sk-der ] have been filtered out .
it is evident that , in all the configurations , the variation in the derivative skewness in the inhomogeneous ( @xmath32 ) and homogeneous @xmath77 ) directions has an opposite sign : the modulus of @xmath78 increases , while the modulus of @xmath79 decreases .
the ratio between these two skewness factors increases with time and with the imposed initial scale ratio . for instance
, it can be observed that , within about a lapse of about 6 - 7 timescales , the ratio between the thickness of the longitudinal derivative in the inhomogeneous direction and those in the normal directions ( directions along which the field remains homogeneous ) increases from 1 to 1.35 , when the scale ratio is close to 2 , and to 1.64 , when the scale ratio is close to 3 ( see figure [ fig.peaks-der](b ) ) .
this differentiation of the behaviour of the longitudinal derivative , in the energy gradient direction and in the directions normal to it , is qualitatively the same as in the case of the interaction of two isotropic turbulent flows with an imposed energy gradient @xcite , and seems to be a general feature of an interaction layer in the absence of shear . at the same time
, the kurtosis levels also differentiate , but even through the inhomogeneous direction shows a slightly higher level of intermittency , the difference is much less marked and never exceeds 30% .
it can be noted that the maximum kurtosis variation is about 25% of the isotropic value , while the maximum skewness is about 50% of its isotropic value .
the interaction between the two decaying isotropic turbulent flows has a greater impact on the third moment than on the fourth order moment .
we can conclude that the anisotropy exists at small scales as well as at large scales in the interaction layer between these two flows ; the large - scale anisotropy introduced by the initial conditions gradually spreads to the small scales .
however , this kind of anisotropy is different from the kind that can be observed in shear flows , see for example @xcite . here
, the main effect is the differentiation of the behaviour of the transverse derivative skewness compared to the longitudinal skewness in the direction of the mean shear ( it should be pointed out that those in the normal directions were not been discussed in the previously cited papers ) . here
, on the contrary , the transport has a negligible effect on the transverse derivative moments , which always remain close to their isotropic values .
the conclusion that the postulate of local isotropy is untenable ( pli ) , both at dissipation and inertial scales , was also reached in @xcite .
this is because the odd order moments of the increments of @xmath80 are nonzero , exhibiting scaling ranges , and the skewness structure function has a value @xmath81 , indicating that in the inertial subrange significant anisotropy is evident even at the third - moment level .
moreover , the fifth - order and seventh - order inertial subrange skewness structure functions are of the order of 10 and 100 , respectively , at least to @xmath82 .
the approach is usually therefore to check the pli at various moments with an increase in the control parameter , @xmath83 . in the present work , we approach the pli question from a different perspective .
we consider the effect of milder and milder nonhomogeneity on turbulence anisotropy . for third - order and fourth - order order velocity and velocity derivative moments , we always observe that a substantial degree of anisotropy can be measured .
as a synthesis for the anisotropy structure for the small scales , one can consider the best fitting relations in figure 7_e _ that link the maxima of the kurtosis and skewness of the velocity derivatives across and along the self - interaction region ( note the opposite slopes ) .
in conclusion , the numerical experiment that we have performed shows three main results .
first , nonlinear interaction is a powerful process that is able to transform a slight inhomogeneity in the correlation length into a kinetic energy step .
the interaction can be considered as a long - range one because , initially , the two isotropic fields are separated by a matching distance which is of the order of the largest integral scale .
second , the interaction builds a sublayer of high intermittency .
in fact , the layer which is produced at the interface between the two initial fields is preceded by this sublayer , which involves not only large scales , but also small scales , and , in particular , their anisotropy .
third , the structure of the anisotropy , in this flow configuration , is rather particular .
it is easy to observe third - order and fourth - order one - point statistical moments , for either the velocity or velocity derivatives statistics , which follow the structure of the initial condition .
the longitudinal derivative skewness gradually departs from the values that can be expected in an isotropic situation .
an appreciable differentiation can also be observed between the different directions : the skewness in the direction crossing the interaction layer ( the direction of inhomogeneity , @xmath32 ) increases in magnitude , while it decreases in the direction parallel to the layer .
for the kurtosis , the peak value across the layer is higher for the longitudinal derivative along the inhomogeneous direction .
it should be noted that the anisotropy described here is different from that of most anisotropic turbulent flows investigated , that is , homogeneous sheared turbulence . in those flows the mean shear creates anisotropy at both large and small scales , which is visible on the generation of large odd transverse moments of the velocity derivative .
however , shear has little impact on longitudinal derivatives . in the present situation ,
the mean velocity is uniform and thus there is no production of turbulent kinetic energy .
departure from local isotropy can be indicated by remarkable difference between the values taken by the longitudinal derivative moments in the directions normal and parallel to the shearfree mixing .
finally , in this work we have obtained quantitative data on the lifetime and formation timescale of the transients , maximum energy gradient and interaction layer width as a function of the correlation length ratio .
the authors would like to thank cineca and caspur computing centres for supporting this work .
scarsoglio s. , tordella d. , criminale w.o .
, an exploratory analysis of the transient and long - term behavior of small three - dimensional perturbations in the circular cylinder wake , _ studies in appl . math . _ * 123*(2 ) , 153173 , ( 2009 ) .
tordella d. , iovieno m. , small scale anisotropy induced by a spatial variation of the integral scale , _
atti dellaccademia delle scienze di torino .
classe di scienze fisiche , matematiche e naturali _ , * 142 * , suppl.2009 , 108 - 111 , ( 2008 ) , proceedings of the euromech colloquium 512 `` small scale turbulence and related gradient statistics '' . ,
step onset from an initial uniform distribution of turbulent kinetic energy , _ advances in turbulence xii _
, 677 - 680 , springer ( 2009 ) . proceedings of the _
12th european turbulence conference _ , marburg ,
september 7 - 10 , 2009 . , numerical experiments on the intermediate asymptotics of the free shear turbulence mixing . _
j. fluid mech .
_ * 549 * , 441 - 454 ( 2006 ) .
ishihara t. , kaneda y. , yokokawa m. , itakura k. , uno a. small - scale statistics in high - resolution direct numerical simulation of turbulence : reynolds number dependence of one - point velocity gradient statistics , _ j.fluid mech . _
* 595 * , 335 - 366 ( 2007 ) . | we have imagined a numerical experiment to explore the onset of turbulent intermittency associated with a spatial perturbation of the correlation length .
we place two isotropic regions , with different integral scales , inside a volume where the turbulent kinetic energy is initially uniform and leave them to interact and evolve in time .
the different length scales produce different decay rates in the two regions .
since the smaller - scale region decays faster , a transient turbulent energy gradient is generated at the interface between the two regions .
the transient is characterized by three phases in which the kinetic energy gradient across the interface grows , peaks and then slowly decays .
the transient lifetime is almost proportional to the initial ratio of the correlation lengths .
the direct numerical simulations also show that the interface width grows in time .
the velocity moments inside this interaction zone are seen to depart from their initial isotropic values and , with a certain lag , the anisotropy is seen to spread to small scales .
the longitudinal derivative moments also become anisotropic after a few eddy turnover times .
this anisotropic behaviour is different from that observed in sheared homogeneous turbulent flows , where high transverse derivative moments are generated , but longitudinal moments almost maintain the isotropic turbulence values .
apart from the behaviour of the energy gradient transients , the results also show the timescaling of the interface diffusion width , and data on the anisotropy of the large and small scales , observed through one - point statistics determined inside the intermittency sublayer , which is associated with the interaction zone .
+ keywords : decay exponent , correlation length , velocity derivatives , intermittency , anisotropy . |
web based deals offering deep discounts to a group of online buyers on products and services is a fast growing market .
group - buying deals attract new customers as well as guarantee customer traffic within a stipulated expiry date for local businesses like restaurants and tour operators @xcite .
most of these group - buying deals are sold by intermediaries like gropon , groupbuy and many other daily deal providers .
though these intermediaries depended on email based marketing models in the past , banner ads in social networking and other sites are increasingly used to attract deal customers . unlike the traditional ads ,
group - buying intermediaries receive their payment only upon satisfying the minimum number of conversions before the deal expiry ( i.e. if the deal tips ) .
this implies that if the deal does not tip , advertiser loses the amount used to buy impressions , and receives no payment .
if the advertiser fulfills or exceeds the guarantee , he receives a payment equal to the product of number of conversions and pay per conversions similar to the traditional ads .
this model is used by popular group - buying advertisers like groupon , and groupbuy among many other deal providers . though most of these deals tips for sites like groupon in current email based marketing , tipping the deals will get harder with increasing competition to attract business owners and shift to the display - ad based marketing .
the proposed strategy enables the deal advertisers to offer more aggressive tipping points , hence more volume of sales to merchants .
further , this model is easy scale to other forms of group - buying campaigns like penalties for not meeting tipping similar to guaranteed display ads . to maximize the profits while bidding for group - buy ads , bidders have to minimize cost by bidding low , but still have to win sufficient number of conversions to satisfy guarantees before the deal expiry .
bidding high increases the probability of winning impressions thereby improves the chance of the deal tipping . on the contrary , higher bids increase the payment to the exchange thereby reducing the profit .
hence bids need to be optimized considering these two conflicting pulls .
this maximal profit bidding necessitates dynamic bid optimization based on the time to expiry and the number of received conversions .
we address this problem of maximizing deal bidder profits , by real - time optimization of bids to minimize the cost of impressions while satisfying the deal tipping guarantees . for group - buying deals , the traditional static bidding strategies based on optimization of expected profits of a single impression is far from optimal .
a significant difference from the traditional ads is that the optimal bid value depends on the time to expiry and number of more conversions required to satisfy the guarantees .
for example , consider a deal requiring just a few more conversions to fulfill the guarantee .
if the deal is about to expire the advertiser would have to bid higher amounts to increase the probability of winning more impressions . on the other hand , if the time to expiry is long for the same deal , he would better off bidding smaller amounts winning fewer fraction of impressions to minimize the payment to the exchange ( since there would be higher number of user visits in larger time intervals ) . evidently , the optimum bid amount is a function of the time dependent parameters like the time to expiry and the additional number of conversions needed to satisfy guarantees . due to this time dependence of optimal bids ,
any static bidding strategy will be sub - optimal , necessitating real - time bidding .
fortunately , this dynamic bid optimization is made possible by the advent of ad exchanges offering real - time auctions ( e.g. realmedia , doubleclick , adecn ) . since the revenue is conditional upon tipping the deal
, the bidding strategies are significantly harder than the traditional non - guaranteed bidding .
in addition to the dynamic quantities mentioned above , deal profit depends on a number of static quantities : pay per event , number and bid distributions of other bidders , conversion rates , and the auction mechanism @xcite .
consequently , formulating and maximizing expected profits which is a function of all these static and dynamic quantities is significantly harder . adding to the complexity , the optimization is online necessitating low computation timings . our method of optimizing profit for guaranteed deals has two steps : ( i ) formulating the expected profit ( ii ) maximizing the profit against the bid .
for the first step , we derive the expected profit as a function of the bid value , time to expiry , fulfilled conversions , amount spent to buy impressions , auction mechanism , click through rate and the number and distribution of the other bidders . since many of these parameters are dynamic as described above , the objective function value changes as the bidding progresses .
among all these parameters , the only parameter the bidder can change is his own bid amount .
hence we optimize the expected profit against the bid amount in the second step .
when the profits are optimum , the deal bids are in a symmetric bayesian nash equilibrium similar to the traditional ads @xcite . considering the complexity of the optimization , a closed form solution is unlikely .
though the optimization is against a single variable ( i.e. bid amount ) , our analysis shows that the objective function is neither convex nor quasi - convex ( unimodal ) .
consequently , an optimization method guaranteed to converge to optimal bids on every instance is unlikely .
further , the derivative of the objective function is harder to solve than the objective itself .
considering these factors , we resort to direct numerical optimization ( without using gradients ) starting from multiple points . *
running time minimization : * since the optimization is online , computation time needs to be minimized .
therefore we explore running time optimization in multiple levels .
firstly , we use a fast converging brent s optimizer .
secondly , we reformulate the objective for faster computation for typical parameter values .
further , we approximate large binomial cumulative probability expressions with a single term normal approximation .
since the changes in the optimal bids for subsequent impressions are incremental , we reuse optimal bid values of previous impressions whenever changes are likely to be negligible .
* extensions : * interestingly , the solutions of many related problems can be directly derived from the proposed objective function .
we describe the four proposed extensions below : ( i ) deal selection : : deal selection chooses the best deals to bid to maximize the bidder profits . combining optimal bidding and selection
, we derive the bidders private value and the marginal profit increase for the impression for each deal .
the deal with the highest marginal profit increase is the greedy optimal selection .
( ii ) deal admissibility : : admissibility is the problem of predicting whether bidding for a group - buying deal is likely to be profitable based on its attributes . the intermediary or the advertiser may decide to accept or reject a deal campaign based on admissibility criterion . we show that a special case of our objective function combined with the bid optimization provides effective admission control . (
iii ) non - bidding selection : : for non - bidding scenarios like the publisher directly selecting the deals to display , the proposed formulation suggests optimal selection among the inventory of deals .
( iv ) non - guaranteed ads : : we show that the real time bid optimization of traditional non - guaranteed ads is a special case of the proposed optimization .
when there are no guarantees , the proposed objective function reduces to expected profits of traditional ads , yielding known optimal static bid formulations .
thus the method serves as a unified real time bidding strategy for both guaranteed and non - guaranteed ads .
* evaluations and results : * we evaluate the proposed methods and the extensions in a query log of size 9.3 million impressions of 935 ads . in our first set of experiments ,
we compare our profits of the proposed real time strategy with the optimal static and base adaptive baselines .
the results show that the proposed strategy improves the profits over the baselines significantly .
subsequently , experiments showing improved profits in spite of violated assumptions of the competitor bids demonstrate robustness of the strategy .
further , our running time evaluations demonstrate acceptable optimization timings .
finally , our evaluations of the ad selection and admissibility demonstrate that the extensions improve profits significantly over the baselines .
rest of the paper is organized as follows .
the next section discusses related work , followed by section on notations and the formal problem definition .
section [ sec - maximizingprofit ] derives the expected profits and proposes the optimization method .
subsequently , we discuss running time minimizations .
next section presents extensions of the problem to deal selection , admissibility , and bidding of traditional ads . section [ sec - evaluations ] present the experimental evaluations and results .
finally we present our conclusions in section [ sec - conclusions ] .
grabchak _ et al . _
@xcite addressed the problem of optimal selection of guaranteed ( group buying ) ads .
our work is different , since we deal with optimal bidding , whereas grabchak _
_ does not consider the bidding and consider offline selection of deals .
further , even the non - bidding selection sub - problem discussed in this paper is different since we consider a minimum number of conversions like deals , whereas grabchak _
_ consider an exact number of required conversions . different models of group - buying auctions and bidding mechanisms has been studied @xcite .
but our problem of bidding to sell deals online mostly made popular after the emergence of dail - deal sites has not been studied for any of the group - buying auction models .
considering related problems of allocation and bidding of display ads , ghosh _ et al . _
@xcite considered allocating guaranteed display impressions matching a quality distribution representative of the market .
vee _ et al . _
@xcite analyzed the problem of optimal online matching with access to random future samples .
boutilier _ et al . _
@xcite introduced an auction mechanism for real time bidding of display ads .
there are a number of papers on optimal ranking of textual ads in presence of budget limits .
mehta _ et al . _
@xcite deal with the problem of optimal allocation of textual ads considering budget limits of the advertisers .
buchbindar _ et al . _
@xcite provided a simpler primal - dual based analysis achieving the same competitive ratio .
these papers consider ranking / allocation of textual ads than deals .
further these problems have an upper limit on number of impressions , rather than a lower limit as in our problem .
hence , unlike these problems , ours is not a generalized online bipartite matching . with the increase of ad exchanges offering real - time bidding
, there are a few papers on related problems .
chen _ et al . _
@xcite formulated the problem of supply side allocation of traditional ads with upper bounds on budgets as an online constrained optimization matching problem .
chakraborty _ et al . _
@xcite considered the problem of ad exchanges calling out a subset of ad - networks without exceeding capacity of individual networks for real time bidding . to the best of our knowledge ,
the optimal bidding problem of group - buy deals and the extensions have not been addressed .
every group - buy deal @xmath0 has a required minimum clicks @xmath1 , an expiry time @xmath2 , a cost per click ( cpc ) @xmath3 , and a click through rate ( ctr ) @xmath4 .
thus a deal may be represented as , @xmath5 for the rest of the paper our discussions are based on guaranteed number of clicks for the ease of description .
the discussions and results are equally applicable for guaranteed conversions ( refer to section [ sec - gnerelizations ] for the details ) ( by substituting conversion rates ( cvr ) for click through rates and click per action ( cpa ) for cpc ) and guaranteed displays ( by setting click through rate to one and substituting cost per impression ( cpi ) for ctr ) .
let @xmath6 be a binary indicator variable , with @xmath7 if the advertiser s bid is successful at time @xmath8i.e .
he wins the bid for impressions and pays the content owner and zero otherwise .
let @xmath9 be the number of clicks at time @xmath8 . for our discussions ,
the time @xmath8 denotes @xmath8 user visits ( impression opportunities ) rather than wall clock time . for a deal
@xmath0 the profit @xmath10 at time @xmath8 is , @xmath11 where @xmath12 is a mapping from bids to the payment whose closed form depends on the auction model , the number of other bidders and the bid distributions . for
the commonly used first price auction for display ads @xmath13 . for other auctions like second price auction
, closed forms can be derived based on order statistics @xcite . after fulfilling guarantees
( i.e. @xmath14 ) the expected profit function for the guaranteed deals are the same as that of the traditional non - guaranteed ads .
hence , the period of interest for our analysis and experiments is the time before guarantees are fulfilled . to maximize the profit in equation [ eqn - profitforumuation ] ,
the only parameter decided by the bidder is the bid amount .
hence we may state the profit maximization problem as , * bidding problem : * _ given a guaranteed ad @xmath15 , and number of received conversions @xmath9 , find the bid amount @xmath16 such that the expected profit from @xmath17 user visits is maximal , where @xmath17 is the expected number of user visits before the ad expiry time @xmath2 . _
to explain the nature of the problem , we start by finding the optimal bid based on the expected values of parameters at @xmath18 .
this is the best possible estimate at that point of time .
as time progresses , we will get better estimates of parameters based on the actual values of number of conversions , and user visits so far .
hence we keep updating the optimal bid @xmath16 based on the current state and expected numbers in the future .
we assume that @xmath17 is known , as it can be generally estimated from the traffic statistics @xcite .
we derive the expected profits of group - buy deal campaigns based on the current state of the deal .
subsequently we analyze the nature of the the profit - function , and present a method to maximize the profits in real - time by bid adjustments .
the click probability of a deal is , @xmath19 the first factor @xmath20 is equal to the ctr of the deal is a constant for static auctions considered here . the second factor probability of winning impression @xmath21is an increasing function of the bid amount .
this implies that the probability of satisfying click guarantees , and consequently the expected profit increase with the bid amount . on the contrary ,
the amount paid by the bidder to the publisher ( @xmath12 ) is an increasing function of the bid amount .
hence the profit tends to decrease with increasing bid amount .
the bids need to be optimized considering these two conflicting effects on the profit . for real - time bidding , different advertisers or intermediaries place bids for a given ad impression .
generally the highest bidder wins , and will display his ad . in general bid values of a bidder varies , either due to the bidder s private value distribution , or due to a deliberate randomization done by the bidder to avoid giving advantage to the competition @xcite .
hence , the event of winning is probabilistic , with a binary outcome .
further , winning in consecutive bids can be assumed to be independent of each other . hence bidding to win impressions are bernoulli trials with success probability increasing with the bid amount .
the users click with probabilities equal to the estimated ctr of the winning ad .
this is again a bernoulli trial with success probability equal to the ctr .
hence these two trials bidding and getting conversions may be combined as a single bernoulli trial of bidding to win clicks .
the probability of success for this composite trial is equal to the product of ctr and probability of winning an impression . for composite bernoulli trial described above
, the number of successes follows a binomial distribution .
to facilitate representing such a binomial distribution , we introduce the following two functions , @xmath22 @xmath23 where @xmath4 is the ctr of the ad , and @xmath24 is the additional number of clicks required to satisfy the guarantees . function @xmath25 is a mapping from the bid value to the probability of winning the impression . for a sealed bid auction
in which the highest bid wins ( e.g. first or second price auctions ) , this probability is @xmath26 , where @xmath27 is the cumulative probability distribution of the bids of other bidders , and @xmath28 is the total number of bidders . to get a closed form of @xmath25 we need to assume a distribution function of bids .
for example , if the bids are uniformly distributed between @xmath29 and @xmath30 , @xmath31 .
similar closed forms can be derived for other distributions , and even for cases where different competitors following different distributions @xcite . at optimal profits ,
the bids are the best responses to competitors and hence are in a symmetric bayesian - nash equilibrium @xcite .
consequently , we may limit our analysis to truthful bidding without the loss of generality as stated by the revelation principle @xcite .
hence the assumptions on bid distributions above are equivalent to the same assumptions on private value distributions of bidders at the optimal profit outcomes . now the net expected profits is given by the objective function , @xmath32 please refer to appendix [ appendix - secexpectedprofitderivation ] for the derivation of the expected profits .
the expected profit in equation [ eqn - expectedprofitsingledeal ] has to be optimized with respect to the bid amount .
an option is to differentiate the function with respect to @xmath16 and solve the derivative for zero .
but this is hard since the derivative may have large number of terms , and solving the derivative will be harder than a direct approach .
hence a direct optimization of the objective function as we do below is faster .
an example curve of variation of the objective function with bids is shown in figure [ fig - objectivenonconvex ] .
two observations significant to the numerical optimization are ( i ) the optimization is non - convex .
( ii ) the function is not even quasi - convex ( unimodal ) .
this implies that a bisection or gradient descent method may get trapped in a local optimum , and hence the convergence to the global optima is not guaranteed .
consequently , we need to start the optimization from multiple points making the problem harder . for the bidding process
, the winning probability is one if the bid is greater than the maximum bid of the competitors bid distribution ; and zero for bids less than the minimum bids . hence the optimal values will always be between the maximum and minimum even without imposing external constraints .
this allows a simpler unconstrained optimization .
the optimizer restarts the search from multiple random starting points to avoid local minima traps ( the details of the restarts are discussed in the section [ subsec - minimizingrunning ] ) .
further since the optimization is online , fast - convergence is highly desirable . considering these factors , we adapt brent s optimization method .
brent s optimization combines parabolic interpolation with golden ratio search for faster convergence .
if the parabolic interpolation fails , the search falls back to the golden ratio search . , @xmath33 , @xmath34 , @xmath35 , and single competitor with uniform random bid in @xmath36 $ ] ( i ) exact binomial for @xmath37 , @xmath38 .
( ii ) normal approximation for @xmath39 , @xmath40.,width=264,height=207 ]
the optimization of the bids has to be performed online between the impression opportunities . evaluating objective function in equation [ eqn - expectedprofitsingledeal ]
directly may involve computing hundreds of thousands of terms . since the time duration between the impression opportunities can be very small
, the optimization time must be within acceptable limits .
in addition to using fast converging brent s optimizer , we adopt several approximations and computational methods for improved running time . * reducing number of terms : * typically for deals the user visits needed to get the required number of clicks are very large compared to the clicks needed for tipping the deal .
i.e. @xmath41 ; except for a last few user visits before deal expiry . exploiting this ,
we reduce the computation time by rewriting the equations [ eqn - phiexpression ] and [ eqn - thetaexpression ] as , @xmath42 @xmath43 this rewriting may reduce computations from hundreds of thousands of terms to less than a few hundred terms . *
normal approximation : * in spite of replacing @xmath17 by @xmath24 , computing binomial cdfs in equation [ eqn - phiexpressiontuned ] and [ eqn - thetaexpressiontuned ] may involve summation of hundreds of combinatorial terms . the binomial cdf may be approximated by a single term normal cdf for large values of @xmath17 ( central limit theorem ) . exploiting this , we compute @xmath44 and @xmath45 based on a normal cdf with correction for continuity .
i.e. @xmath46 there is no analytical solution for this integral , but can be looked up from a normal cdf table or can be approximated by finite analytical forms .
similarly , approximating the standardized form of @xmath45 as @xmath47 , @xmath48 where @xmath49 @xmath50 for small @xmath17 and @xmath24 , normal approximation may diverge more from the original binomial function , and computation of the binomial is less costly .
hence it is more accurate to use actual binomial distribution for smaller @xmath24 and @xmath17 .
we depend on the common rule of thumb for approximating binomial cdf by normal cdf , i.e. if @xmath51 we use normal approximation . considering the optimization of the normal approximation ,
the sample graph of the objective function with the approximations is shown in figure [ fig - objectivenonconvex ] . like the original binomial objective , the normal approximation is neither convex nor quasi - convex .
consequently optimizing the approximation faces the same difficulties as the optimization of the original objective . *
setting the starting points : * the optimal bid values generally change only nominally for subsequent impressions of a deal .
exploiting this fact , the optimal bid for an impression is used as the starting point for optimization for the next impression .
this reusing of optimal bids expedite convergence .
* multiple starting points : * the non - convexity of the objective requires the optimal value search to start from multiple points . as optimal bids change only incrementally for successive impressions of a deal , we avoid restart from multiple points for every impression . for this instead of starting from previous optimal values as described above we chose random starting points for optimization for the first twenty impressions .
the bid corresponding to the maximum objective value among these searches is used as the optimal value .
this strategy is found to be converging to optimal values for all the deals we tested . *
re - computation frequency : * as the change in optimal bids are nominal for subsequent impression opportunities , the previous bids can be reused .
hence we re - optimize bids only after a number of impression opportunities ( every thirty two impression opportunities in our experiments ) .
further , optimal value is always recomputed if there is a click , since one more clicks may cause a non - trivial changes in the optimal bid .
the optimal bidding for group - by deals is a general problem , which on specific assumptions reduces to number of related use - cases . as a corollary
, the proposed solution reduces to solutions of these problems on the same assumptions . in figure [ fig - objecviereductions ]
we enumerate these extensions and assumptions on which the guaranteed bidding will reduce to these problems .
considering three downward branches in figure [ fig - objecviereductions ] , the guaranteed bidding easily transform to guaranteed clicks , displays and conversions .
for example by setting @xmath52 the deal definition in equation [ eqn - dealdefinition ] will define a guaranteed conversion campaign .
consequently , on the same assumption the profit formulation in equation [ eqn - expectedprofitsingledeal ] gives guaranteed click campaign profits .
similarly , by setting @xmath4 equal to 1 and cvr , the problem and the solution will be applicable for guaranteed impressions and conversions deals respectively .
four upward branches in figure [ fig - objecviereductions ] illustrate reductions to related bidding and selection problems . we discuss them in the sections below and
the the deal related extensions are evaluated separately in sections [ subsec - selectingdealsexpt ] and [ subsec - admissibilitydealsexpt ] .
deal selection is the problem of maximizing expected profits by choosing the best deal to bid for every impression . considering the online nature of the problem
, we perform a greedy deal selection .
the deal with the maximum marginal profit by the impression is selected as the winner . expected marginal profit is calculated as the difference between the expected profits of winning the impression and failing to win the impression . adding the bid amount to marginal profit to derive the marginal revenue as , @xmath53 \label{eqn - expectedtruevalue } \end{aligned}\ ] ]
derivation of the equation is given in appendix [ appendix - secprivatevaluederivation ] .
this marginal revenue is the expected private value of the impression for the deal bidder , isomorphic to the private value of the traditional ads .
similar to the traditional ads , @xmath54 gives the expected profit by displaying deal @xmath55 at time @xmath8 . to select a deal
, the bid values of deals are optimized against the expected profits as described in section [ sec - maximizingprofit ] .
these optimal bids are substituted in equation [ eqn - expectedtruevalue ] to calculate the private values .
the deal with the highest increase in the expected profit is selected for bidding .
we evaluate the proposed selection in section [ subsec - selectingdealsexpt ] .
selection considering deals groups is a harder combinatorial optimization problem , and we leave this for future research .
admissibility criterion decides if the profit from a deal is likely to be positive .
admissibility can be directly derived as a special case of optimal bidding .
more specifically , if the maximal expected profit from an deal is positive at @xmath18 the ad is admissible . at @xmath18
the equation [ eqn - expectedprofitsingledeal ] reduces to , @xmath56 where @xmath2 is the expected number of visits before the deal expiry . to evaluate admissibility ,
bids are optimized for maximal profits and substituted in equation [ eqn - admissibility ] .
if the expected profits are positive , we consider the deal as admissible .
the profit increase from admissibility is evaluated in section [ subsec - admissibilitydealsexpt ] . if there are no competing bidders i.e .
number of bidders is equal to one the problem will reduce to that of selection from competing deals . alternatively
, this scenario may be thought of as the publisher selecting deals directly . in this case @xmath57 in selection equation [ eqn - expectedtruevalue ] for all values of @xmath16 . since
this is a special case of the selection with bidding , we do not include separate evaluations , as results are directly implied . alternatively , since there is no bidding involved , selection may be modeled as an offline problem . in this case ,
selection of deal sets for maximization profits of deal combinations may be of interest .
we leave this optimal offline selection of deal sets for future research .
as shown in figure [ fig - objecviereductions ] , when the minimum click guarantee is zero , the guaranteed deal bidding reduces to the bidding of tradition non - guaranteed ads .
consequently , the expected profits for non - guaranteed ads may obtained by substituting @xmath58 in equation [ eqn - expectedprofitsingledeal ] as , @xmath59 considering the profit from a single impression ( i.e. @xmath60 ) , thereby ignoring the constants terms of past profits , the expected profit becomes , @xmath61 .
this expected profits can be maximized with respect to the bid values .
for example , for a first price auction with @xmath28 bidders and competitors having a bid distribution with cumulative distribution function of @xmath62 , @xmath63 and @xmath64 .
note that the expected profits , hence the optimal bids , derived above is the same as the existing formulations of optimal static bid profits of traditional ads @xcite .
this is a manifestation of the broader fact that the existing bidding formulations are optimal for static environments .
hence the optimal real time bids will essentially be the same .
in other words , the real - time bidding provides no advantage over static bidding for non - guaranteed ads in static environments . on the other hand
, the real - time bidding would improve profits over existing bidding for dynamic environments @xcite .
thus the proposed bidding provides a strategy to account for the dynamism in parameters of traditional ads like the changed estimates of click - through rate or competitor bid distributions .
we compared the profits by the proposed bidding strategy with baseline strategies of ( i ) optimal static bidding ( ii ) a basic real time bidding .
we evaluated the profit increase , running time , deal selection and admissibility including the robustness of our method to violation of assumptions .
* data set : * from a click log of 330 million impressions , we randomly selected 935 ads ( with 9.3 million impressions ) having a minimum of 5 clicks and 1000 impressions .
click log contains impressions and whether the impression resulted in a click or not .
* baseline bidders : * first baseline bidder is an optimal static bidder .
the bidder derives optimal bids as a function of number of competitors and their bid distributions @xcite , and is optimal if there are no guarantees .
the second baseline bidder is a basic adaptive bidder , which bids as @xmath65 .
the strategy is intuitive , as it increases bid over the static optimal bid if the required click rate is greater than the ctr and decreases the bids otherwise .
we also used the profits by a placing a random bid as a baseline in our initial experiments .
random bidding performed much worse than all of the above baselines and is not plotted in the results . to compare the profits by bidders , the proposed real - time ( rt ) and baseline bidders compete with random bidders for every ad in the replayed " click log .
for example , the rt bidder places its bid for the first impression of an ad , along with the competing random bidders .
if the rt bidders bid is the highest , the bidder wins the impression .
the deal wins a click if the click log indicates a click for that impression .
then the same process repeats for the second impression and so on .
similarly , other baseline bidders are made to compete with the same random bidders for the same set of ads , and the realized profits are compared .
this replaying reproduces online experiment , since the user action in the consecutive impressions are most likely from different users , and hence independent .
the experiments are repeated by changing every significant parameter
one at a time to analyze the effects of different parameters .
since the click logs are of traditional ads having no required clicks or expiry , we set the expiry timings as the number of impressions of the ad in the log . to compensate for using the traditional ads instead of deals ,
the required clicks are varied by a parameter sweep over the plausible range .
we do not vary expiry time , since the ratio of required clicks to the expiry time determines the profit rather than the expiry time alone and this ratio varies as we do a parameter sweep on required clicks .
further , note that different ads in the set of 935 ads have different expiry timings , effectively functioning as a parameter sweep on expiry times .
ctr of every ad is estimated as the ratio of number of clicks to total impressions .
the competing bids are selected from different random distributions ( bids are not registered in the click log ) , since the bidders generally randomize their bids to avoid the competing bidders guessing their bids @xcite . note that this randomization in bids may be achieved by randomly choosing different deals to bid for in different time slots even for the proposed optimal bidding strategy . at the bayesian - nash equilibrium in which everyone follow the optimal strategies these assumptions on bid value distributions are equivalent to the same assumptions on private values of the advertisers , as mentioned in section [ sec - expectedprofits ] .
further , the maximum entropy ( i.e. minimum assumptions ) uniform random competing bid distribution is the hardest to predict and to optimize against ( hence we use this distribution for the experiments below ) .
any other distributions , including a fixed optimized competing bid is easier to optimize against and the realized profits will be higher .
figures [ fig - biddingprofitbasic - a ] and [ fig - biddingprofitbasic - b ] show four sets of comparisons of profits realized by different bidders against required minimum clicks @xmath1 . in the first set of group ( 10 ) in figure [ fig - biddingprofitbasic - a ] , there are four bidders the bidder evaluated and three other random bidders .
random bidders bid in a uniform random distribution in the interval of @xmath66 $ ] .
pay per click ( @xmath3 ) is set to $ 10 for this group .
the remaining groups of experiments in figures [ fig - biddingprofitbasic - a ] and [ fig - biddingprofitbasic - b ] are designed to analyze the effects of changes in parameters .
analyzing common trends in all these groups , the profit of the real time bidder exceeds that of the baseline bidders for every @xmath1 in all the experiments .
as expected , the profits reduces with @xmath1 .
the increase in profit is as large as 70 - 150% ( e.g. at @xmath67 for group ( 10 ) increase is @xmath68 ) . as an exception ,
the profits of real time bidder is the same as the static optimal bidder at @xmath69 .
this is an implication of discussions in section [ subsec - ngads ] , that the current static optimal bidding of non - guaranteed ads are a special case of the proposed rt bidding . as @xmath1 increases from left to right , the absolute and percentage of difference between static and rt profit increases .
the baseline adaptive bidder performs worse than static bidder for most parameter combinations , since the bidding considers only a subset of parameters .
this is a manifestation of the fact that a simpler intuitive adaptive strategy is not likely to perform well , especially since optimal bid depends on a large number of parameters . the baseline adaptive bidder , at @xmath69 perform very similar to ( but not the same ) as the static optimal bidder for uniform competitor bids ( for normal distribution in figure [ fig - normal - robust ]
, they differs considerably at @xmath69 ) as typical ctr values are very small compared to the optimal bids .
the next three sets of experiments analyze the effects of each of the three parameters@xmath3 , bid distribution , and number of bidders .
parameters are changed one at a time with respect to group ( 10 ) in figure [ fig - biddingprofitbasic - a ] by a factor two .
figure [ fig - biddingprofitbasic - a ] group ( 20 ) ( plotted against the second @xmath70-axis ) increases the profit due to increase in pay per click @xmath3 .
the increase in profit is more than linear to @xmath3 since the revenue increases linear to @xmath3 , but the cost of impression remains the same , consequently the profit ( @xmath71 ) increases many times . for the group ( 4 ) in figure [ fig - biddingprofitbasic - b ] , reduced bids of the competing bidders result in lower optimal bids hence increase in profits . in the second set of experiments in figure [ fig - biddingprofitbasic - b ] group ( 2 )
, the reduced number of competing bidders leads to lower optimal bids and hence increase in profits .
[ fig - biddingporiftrobustness ] normally distributed competing bid distributions with @xmath72 and @xmath73 .
all other parameters are same as group ( 10 ) in figure [ fig - biddingprofitbasic - a ] .
@xmath74 robustness to deviation from the assumed distributions .
two of the four competitors bids are normally distributed with @xmath75 and @xmath73 instead of the assumed unform distribution in @xmath66 $ ] by the rt and static bidders .
, title="fig:",width=291,height=226 ] the next sets of experiment in figure [ fig - normal - robust ] further relax assumptions on competing bidders . the group ( g ) has competing bidders having gaussian bid distributions instead of the uniform random distribution in the previous experiments ; and group ( r ) evaluates the robustness of rt bidding against violation of assumptions . like the uniform distribution , for the gaussian experiments in group ( g ) as well , the rt bidder outperforms the competitors .
the profits are higher than the uniform distributions in figure [ fig - biddingprofitbasic - a ] group ( 10 ) , since the lower entropy of gaussian distribution is easier to optimize against . for the robustness experiments
if group ( r ) , the rt and static bidders assume uniform distribution in @xmath76 $ ] for the three competing bidders , but two of the competing bidders bid in normal distribution .
the rt bidder still dominates over the baseline by considerable margins .
a plausible explanation for similar profits to figure [ fig - biddingprofitbasic - a ] group ( 10 ) is that the effects of violation of assumption and the easier gaussian distribution cancel each other .
these experiments are conducted on a shared 16 gb ram machine with two dual core cpus running at 2.54 ghz .
the running time optimizations described in the section [ subsec - minimizingrunning ] are applied ( similar to other experiments ) . analyzing the objective function in equations [ eqn - phiexpressiontuned ] and [ eqn - thetaexpressiontuned ]
, the parameter having maximum effect on the running time is the require clicks @xmath1 .
the number of terms in the objective increases with @xmath1 , and other parameters will have negligible effect on computation time .
hence we evaluated mean time to optimize the bids for an impression against plausible ranges of @xmath1 .
figure [ fig - runningtime ] shows that the running time increases linearly with the required cliks as expected .
the highest running time is less than 0.2 milliseconds , which is quite acceptable .
we evaluate the proposed real time deal selection described in section [ subsec - guarenteedadselection ] .
selection is compared with the optimal static selection preferring higher private value ads computed as @xmath77 . among the 935 ads in click log , we removed ads with ctr greater than 0.02 for these experiments .
these high ctr ads make the selection easy by dominating over other ads , hence make it harder to distinguish the selection quality . among 823 remaining ads , we randomly created groups of four ads .
selection experiments are performed for each group separately , and mean profits among all groups are plotted .
we set the bidding parameters as @xmath78 , total number of user visits as 15000 , time to expire for each ad as minimum of 15000 and the number of impressions of the ad , number of bidders as four , and competing bidders bids are uniform random in @xmath76 $ ] .
for every deal , the require clicks is set uniform random between zero and the maximum value shown in the @xmath79-axis to have different required clicks values for different deals . for selecting the best ad in the group for an impression
, first the bids are optimized using the real time bidder .
these optimal bids are used in equation [ eqn - expectedtruevalue ] for real time selection , and to compute private value ( i.e. @xmath77 ) of the static selection . to separate improvement in profit by selection from improvement by bidding
, the proposed rt bidder is used for bidding after both the static and rt selections .
the mean realized profits are shown in figure [ fig - extensions - a ] . when the required clicks is zero , the real time bidding gives the same optimal profit as the static bidding ( keep in mind that the static selection is optimal when require clicks is zero ) . for higher values of required clicks ,
the real time bidder gives considerably higher profits , with percentage of increase in profit increasing with required clicks .
the profit swings at larger values of required clicks is due to the random factors in assignment of required clicks .
the admission control proposed in section [ subsec - admssibility ] is evaluated by comparing mean deal profits against the profits without admission control .
similar to selection , bids are optimized and substituted in equation [ eqn - admissibility ] .
the deals giving positive expected profits are passed to the bidder , and mean profits are plotted .
figure [ fig - extensions - b ] shows that admission control improves profits by more than six times for some values of required clicks @xmath1 .
the profit increase for both static and rt bidders , showing effectiveness of admission control independent of the bidding method ( like the previous experiments rt bidder performs considerably better than static ) . at @xmath69
all the ads in the click log have positive expected profits .
the mean profit no longer decreases monotonically , as the admission control eliminate more low profit ads with increased @xmath1 .
further the total profits of rt bidder with and without admissibility is almost exactly the same .
this shows that there are no false negatives removed .
the admission control does not increase the total profits , because for the ads with negative expected profits , the real time bidder will bid zero making the losses to zero .
with admissibility , the bidder incur the same profit from much lesser number of ads and user visits , hence he can use the remaining user visits to sell other ads .
an emerging category of the online ads are the group - buy deals requiring minimum number of purchases . for an advertiser or intermediary selling these deals ,
optimizing bids is necessary for maximal profits .
existing bidding strategies are sub - optimal for these deals , as they do not consider event minimum group - size guarantees and expiry timings . to this end
, we propose a real time bidding strategy for guaranteed deals .
we derive the expected profits as a function of the dynamic and static parameters of the deals .
these expected profits are shown to be non - convex , and numerically optimized against the bid values . to satisfy the stringent time constraints of online bidding
, we use several approximations and running time optimizations . exploiting the generality of the proposed formulation ,
we extend the solution to related problems of deal selection for bidding , admissibility , selection for non - bidding scenarios and real time bidding of non - guaranteed ads . our empirical comparisons with base adaptive and the existing static strategies on a multi - million click log show significant profit improvements .
further our evaluations show acceptable running time and robustness against the violation of assumptions .
evaluations of extensions show considerable profit improvement by the proposed deal selection and admissibility .
10 daily deals rescue local - ad market , wall street journal , june 14 , 2011 .
k. anand and r. aron .
group buying on the web : a comparison of price - discovery mechanisms .
, pages 15461562 , 2003 . c. boutilier , d. parkes , t. sandholm , and w. walsh .
expressive banner ad auctions and model - based online optimization for clearing . in _ proceedings of aaai _ , 2008 .
n. buchbinder , k. jain , and j. naor .
online primal - dual algorithms for maximizing ad - auctions revenue . , pages 253264 , 2007 .
t. chakraborty , e. even - dar , s. guha , y. mansour , and s. muthukrishnan .
selective call out and real time bidding . , 2010 .
j. chen , x. chen , and x. song .
bidder s strategy under group - buying auction on the internet .
, 32(6):680690 , 2002 .
y. chen , p. berkhin , b. anderson , and n. devanur .
real - time bidding algorithms for performance - based display ad allocation . in _ proceedings of kdd _ , 2011 .
p. dasgupta , p. hammond , and e. maskin .
the implementation of social choice rules : some general results on incentive compatibility .
, 46(2):185216 , 1979 .
d. easley and j. kleinberg . .
cambridge univ pr , 2010 .
a. ghosh , p. mcafee , k. papineni , and s. vassilvitskii .
bidding for representative allocations for display advertising .
, pages 208219 , 2009 .
m. grabchak , n. bhamidipati , r. bhatt , and d. garg .
adaptive policies for selecting groupon style chunked reward ads in a stochastic knapsack framework . in
_ proceedings of www _ , pages 167176 .
acm , 2011 .
v. krishna . . academic press , 2009
a. mehta , a. saberi , u. vazirani , and v. vazirani .
adwords and generalized online matching .
, 54(5):22es , 2007 .
m. richardson , e. dominowska , and r. ragno .
predicting clicks : estimating the click - through rate for new ads . in _ proceedings of www _
, pages 521530 , 2007 .
e. vee , s. vassilvitskii , and j. shanmugasundaram .
optimal online assignment with forecasts . in _ proceedings of ec _ , pages 109118 .
acm , 2010 .
* cost : * at time @xmath8 an amount equal to @xmath84 is paid for the impressions .
the future expected cost is the expected payment till @xmath17 .
let @xmath85 denotes the total number of displays till @xmath17 , @xmath86 these conditional expectations can be expanded as , @xmath87 \nonumber \\ & = & \sum_{j=1}^{u_t } j \left[\frac{p((d = j)\wedge g)}{p(g ) } p(g ) + \right .
\nonumber \\ & & \left .
p ( \neg g ) \frac{p((d = j)\wedge \neg g)}{p(\neg g)}\right ] \nonumber \\ & = & \sum_{j=1}^{u_t } j \left[p\left((d = j)\wedge g \right ) + p\left ( ( d = j)\wedge \neg g \right)\right ] \nonumber \\ & = & \sum_{j=1}^{u_t } j p(d = j ) \nonumber\end{aligned}\ ] ] since number of impressions follows a binomial distribution with success probability equal to probability of display @xmath88 , @xmath89 .
hence the expected cost is @xmath90 * revenue : * revenue is conditional on g , as revue in the event of @xmath91 is zero . at time @xmath8
, total expected revenue is the sum of revenues of already realized clicks and the revenue of the expected clcisk till @xmath17 .
let @xmath92 denotes the future expected revenue till @xmath17 , @xmath93 let @xmath94 denotes the number of clicks till @xmath17 , @xmath95 total expected revenue is @xmath96 .
@xmath97 as the experiments are bernaulli trials with success ( conversion ) probability of @xmath98 , @xmath99 on displaying the ad , the ad may get conversioned with a probability equal to @xmath4 , and will not be conversioned with a probability equal to @xmath103 .
hence the expected change in profit given a display is ( we ignore the minute possible change in optimal bid in a single display ) @xmath104 is , @xmath105 + ( 1-\mu)\rho_i \left[c_t \phi ( r_t , u_t-1,b_t,\mu)+ \right . \nonumber \\ & & \!\!\!\!\!\!\!\!\!\!\left .
\theta ( r_t , u_t-1,b_t,\mu ) \right]- \left(\sum_{j=1}^{t-1}\psi_jb_j + b_t+ p_d ( u_t-1)b_t \right ) \nonumber\end{aligned}\ ] ] substituting these values in equation [ eqn - expectedvalueincrease ] we get @xmath108 as , @xmath109\nonumber \\ & = & \mu \rho_i \left[c_t \left(\begin{array}{c } u_t-1\\ r_t -1 \end{array}\right ) ( \mu p_d)^{r_t-1 } ( 1-\mu p_d)^{(u_t - r_t)}+ \right .
\nonumber \\ & & \phi ( r_t-1 , u_t-1,b_t,\mu ) + \nonumber \\ & & \left .
( r_t -1 ) \left(\begin{array}{c } u_t-1 \\ r_t -1 \end{array}\right ) ( \mu p_d)^{r_t-1 } ( 1-\mu p_d)^{(u_t - r_t ) } \right ] \nonumber \\ & = & \mu \rho_i \left[(c_t+ r_t -1)\left(\begin{array}{c } u_t-1 \\ r_t -1 \end{array}\right ) ( \mu p_d)^{r_t-1 } ( 1-\mu p_d)^{(u_t - r_t ) } \right .
\nonumber \\ & & \left .
+ \phi ( r_t-1 , u_t-1,b_t,\mu ) \right ] \nonumber \end{aligned}\ ] ] | group - buying ads seeking a minimum number of customers before the deal expiry are increasingly used by the daily - deal providers . unlike the traditional web ads , the advertiser s profits for group - buying ads
depends on the time to expiry and additional customers needed to satisfy the minimum group size
. since both these quantities are time - dependent , optimal bid amounts to maximize profits change with every impression . consequently ,
traditional static bidding strategies are far from optimal .
instead , bid values need to be optimized in real - time to maximize expected bidder profits .
this online optimization of deal profits is made possible by the advent of ad exchanges offering real - time ( spot ) bidding . to this end
, we propose a real - time bidding strategy for group - buying deals based on the online optimization of bid values .
we derive the expected bidder profit of deals as a function of the bid amounts , and dynamically vary bids to maximize profits .
further , to satisfy time constraints of the online bidding , we present methods of minimizing computation timings .
subsequently , we derive the real time ad selection , admissibility , and real time bidding of the traditional ads as the special cases of the proposed method . we evaluate the proposed bidding , selection and admission strategies on a multi - million click stream of 935 ads .
the proposed real - time bidding , selection and admissibility show significant profit increases over the existing strategies .
further the experiments illustrate the robustness of the bidding and acceptable computation timings . |
the three body problem is a classical one and the formal solution is given in terms of the faddeev equations @xcite . although formally these equations are very easy and intuitive , the technical implementation is rather involved and usually some approximations are always done @xcite .
advances in this field using chiral lagrangians have also been done @xcite . a different approach , using as input unitarized amplitudes obtained from chiral lagrangians was done in @xcite for states of three mesons , in @xcite for two mesons and a baryon and in @xcite for one meson and two baryons . a different approach , using separable potentials fitted to two body data , is also done in @xcite for two mesons and a baryon
are studied . in @xcite , the @xmath7 , @xmath8 and @xmath9 systems with a chiral @xmath10 interaction and a phenomenological @xmath11 potential . in the present work we deal with the interaction of the three - particle system composed of a @xmath12 , a @xmath13 and a @xmath14 meson .
the three - body approach has helped to bring more insight in the understanding of different observed resonances , as in the case of the @xmath15 @xcite , that was discussed in ref .
@xcite as a molecular resonance of the coupled system @xmath16 and @xmath17 . in the work of ref .
@xcite the baryonic resonance @xmath18 @xcite was proposed as a @xmath19 system .
the mesonic states @xmath20 , @xmath21 , @xmath22 , @xmath23 and @xmath24 @xcite were obtained in ref .
@xcite as multi-@xmath12 states , interpreting all these resonances as molecular states made of two , three and up to six @xmath12 mesons respectively .
this strong interaction with many @xmath12 mesons with alligned spins was also investigated in ref .
@xcite , and it was found that the @xmath25-multi-@xmath12 interaction can dynamically generate the observed @xmath26 , @xmath27 , @xmath28 and @xmath29 states @xcite . in this work , as in @xcite , the three - body interaction is solved using as input the two - body interaction of the composing hadrons given by the chiral unitary approach .
this two - body interaction in the present case , is obtained from the lowest order lagrangian which describes the interaction of pseudoscalar plus vector mesons @xcite , called the local hidden gauge approach , after the unitarization of the tree level amplitudes @xcite .
the extension to the charm sector of the local hidden gauge approach @xcite and its success in the description of known @xmath30 states , plus the prediction of the @xmath31 , which was found later in the experiment of ref .
@xcite , motivated the aplication of this formalism to the study of the @xmath32-multi-@xmath12 interaction @xcite , the three - body interaction of the @xmath33 system @xcite and of the @xmath34 system @xcite .
we follow here the analogous aplication of the two - body interaction of the @xmath4 @xcite system to solve the three - body interaction of the @xmath35 system .
one of the findings of @xcite is the existence of bound states in spin @xmath6 for the @xmath36 , @xmath37 and @xmath38 systems respectively , as a consequence of the strong vector - vector interaction in this sector .
the fact that these @xmath6 states are very bound in all the cases , since the binding energy found is @xmath39 mev , justifies the fixed center approximation to the faddeev equations ( fca ) .
this approximation is often used @xcite and it was also used in the previous works of refs .
it is justified since two of the particles form a bound cluster that is not much affected by the scattering of the third particle @xcite . in the language of quantum field theory , the fca was identified with the limit of neglecting the three - momentum of the two heavier particles in the propagators , and then it is only the third , the lighter particle , the one who finally interacts with the cluster @xcite . in the following sections we will explain in detail the formalism employed to solve the interaction , and we will show the results that we obtain with this model for the case in which the pair of @xmath40 mesons are composing a cluster of spin 2 and isospin 0 .
we will evaluate the sensitivity of this model to the free parameters and some of the different assumptions , and finally , present the conclusions and scope of this work .
in the fca the @xmath41 and @xmath42 mesons will be regarded forming a cluster for which we know its wave function .
then we can construct the diagramatic series describing the @xmath12 meson interacting with the cluster , fig .
[ fig : scheme - for - t1 - 2 ] , which can be accounted for in terms of two partition functions , @xmath43 , @xmath44 , which sum all the multiple scattering terms of particle 3 beginning with particle 1 or 2 respectively .
the partition functions satisfy the following coupled equations , @xmath45 taking @xmath46 and @xmath47 as a common factor in eqs .
( [ eq : t_1-ffc - dots ] ) and ( [ eq : t_2-ffc - dots ] ) respectively , we arrive to the coupled faddeev equations in the fixed center approximation for the particle @xmath48 and @xmath49 : @xmath50 now @xmath51 is the propagator for the @xmath12 meson folded with the wave function of the cluster and we will focus on it later . solving eqs . and we will be able to construct the total three - body @xmath52 matrix , @xmath53 in the isospin basis the @xmath12 is a triplet @xmath54 and the @xmath41 and @xmath42 are doublets @xmath55 , @xmath56 .
we will calculate the @xmath57 interaction in spin @xmath58 .
the @xmath1 and @xmath59 systems have been studied previously in the formalism of the local hidden gauge approach and heavy quark spin symmetry @xcite .
one of the results of this previous work is the absence of bound states of the @xmath1 system in isospin @xmath60 , due to the weakness of the interaction in this sector .
the bound states are found in isospin 0 and we are interested in those of spin @xmath61 .
we will take the value of the @xmath1 cluster mass for @xmath62 as @xmath63 mev , obtained whithout considering the coupling to the @xmath59 channel .
the consequence of including bottomed strange mesons as a coupled channel is that the mass of the @xmath6 state is obtained as @xmath64 mev @xcite .
so the influence of this latter channel can be neglected .
these results legitimate the use of this state as a @xmath1 bound state .
we will project the scattering amplitudes of the three - body system into proper states of total isospin @xmath60 in terms of two particle isospin states , with the restriction of considering the @xmath1 system as an isospin zero state .
@xmath65 when we make the projection @xmath66 it must be taken into account that @xmath67 relates particles @xmath48 and @xmath68 , i.e. , it describes the @xmath69 system . with @xmath70
we are labelling the @xmath52 matrix for @xmath71 interaction .
so we need to consider the @xmath2 isospin states : @xmath72 combining eqs .
( [ eq : i - one - three - particle - state ] ) , ( [ eq : i - rho - bx - states ] ) into ( [ eq : t - matrix - isospin - projection ] ) , we obtain the following result , @xmath73 in eq .
( [ eq : t_j - i-1 ] ) @xmath74 and @xmath75 are the @xmath52 matrices of the @xmath69 and @xmath71 systems respectively , for isospin @xmath76 of the two particle system .
the @xmath2 interaction has already been studied in a previous work @xcite .
the @xmath71 interaction is the same as that of @xmath2 , which could be obtained by charge conjugation .
we are going to calculate the three - body interaction in terms of the two - body @xmath2 results assuming @xmath48 and @xmath49 as a cluster . in order to do it ,
let us take the first diagram of fig .
[ fig : scheme - for - t1 - 2 ] and write the @xmath77-matrix element , at first order , for the interaction of @xmath68 with @xmath48 and also of @xmath68 with @xmath49 .
we follow the convention of @xcite , @xmath78 the @xmath79 ensures the four - momentum conservation , @xmath80 and @xmath81 stand for the initial and final energies , respectively , of the @xmath82 particle , and @xmath83 is the amplitude associated to the process , while @xmath84 stands for the volume where the states are normalized to 1 .
the @xmath77-matrix of the @xmath12 meson scattering with the @xmath1 cluster of mass @xmath85 is expressed as @xmath86 the factors @xmath87 in eqs . , and are not the same .
if one writes explicitly the expression for the double scattering , one also finds a different combination of @xmath87 factors than in the two former cases . in order to sum the different terms in the fca series with the same normalization
a redefinition of amplitudes is convenient and this was done in @xcite .
one simply has to define the @xmath88 matrices , @xmath89 together with the expression for the @xmath51 function given as @xmath90 where @xmath91 is the form factor of the cluster wave function that we describe below . with the normalization of eq .
we have @xmath92 since in this case @xmath93 , the former expression can be simplified and we find : @xmath94 the @xmath95 function in eq .
( [ eq : g0 ] ) is the form factor of the resonance or cluster , which is the fourier transform of its wave function .
there @xmath96 and @xmath97 are the energy and the mass of the @xmath12 meson .
we will consider the following form factor description for s - wave functions @xcite , @xmath98 where @xmath99 and @xmath100 are defined by , @xmath101 and @xmath102 is a three - momentum cutoff used to regularize the @xmath38 @xmath103 function in the @xmath13 system in order to obtain the @xmath38 as a bound state @xcite .
( [ eq : solution - fca ] ) and have as argument the total center of momentum ( c.o.m . )
invariant energy @xmath104 and , in analogy with @xcite , we would like to share this energy with the two particle subsystems .
we follow the procedure of ref .
@xcite and we write here the expressions of the energy for the @xmath71 and @xmath2 two - body systems @xmath105 in terms of @xmath104 , @xmath106 the total energy is shared between the particles proportionally to their masses , @xmath107 where @xmath108 is the mass of the @xmath109 particle .
the total three - momentum of the two particle system appearing in eq .
( [ eq : two - body - energy-12 ] ) is estimated in terms of the binding energy of @xmath48 and @xmath49 in the cluster , i.e. , @xmath110 with eqs .
- we are ready to implement the two - body @xmath52-matrix elements @xmath111 in eq . in order to solve eq . . as we have mentioned in sec
[ sec : fca - f - e ] , the @xmath2 , so @xmath112 interaction has been studied in @xcite in the scheme of the local hidden gauge approach @xcite , with unitarization of the amplitudes in coupled channels solving the on - shell version of the factorized bethe - salpeter equation , @xmath113^{-1}v^{i , j } \label{eq : b - s - eq},\end{aligned}\ ] ] where @xmath76 is the two - body total isospin and @xmath114 the two - body spin .
the @xmath115 matrix elements are the tree level amplitudes where we label @xmath116 as @xmath117 , @xmath118 as @xmath119 and @xmath120 as @xmath121 interaction .
@xmath122 is the @xmath123 identity matrix .
the @xmath103 is a diagonal matrix which contains the loop function for a @xmath2 and an @xmath124 loop , @xmath125 we have employed a three - momentum cutoff @xmath126 to regularize the loops of .
the four vector @xmath127 is the total four - momentum of the two - particle system @xmath128 .
the cutoff was fixed around @xmath129 mev in order to reproduce an already known state , the @xmath130 . by solving eq .
we get the @xmath52-matrix element for the @xmath117 interaction in @xmath131 and @xmath6 . in the two - body interaction ,
the width of the states has been also investigated . in order to do this
we have convoluted the loop function considering the @xmath12 meson width , and also we have included the imaginary part of the box diagrams .
the parameters of the two - body interaction in the present work are set equal to those mentioned in @xcite .
further details of this section can be found in the works of refs .
@xcite . in the next section
we will show the results that we obtain when we apply our formalism to the @xmath0 system .
in sec . [ sec : fca - f - e ] we have stated that we would use the form factor of eqs .
to to describe the clustering effect in the propagator of the @xmath12 meson . as we can see in eq .
the form factor is regularized by a three - momentum cutoff @xmath102 which , as mentioned before , is the same one used in the regularization of the @xmath132 loops in the study of the @xmath38 interaction in @xcite . in that work ,
values of @xmath102 around @xmath133 mev were used .
we take a range of values here for @xmath134 mev and see how the results change with @xmath102 . for different values of the three - momentum cutoff @xmath102 . ]
propagator function of eq . for different values of the cutoff @xmath102 . ] . ] in fig .
[ fig : form - factor ] we can see the influence of the cutoff @xmath102 in the form factor .
this dependence is translated into the propagator @xmath51 of eq . .
fig : g0 ] we plot the @xmath51 as function of the energy for different values of @xmath102 .
we consider the @xmath12-@xmath135 cluster system in order to fix the energy of the @xmath12 meson in terms of the total c.o.m .
energy , @xmath136 we recall here that we have chosen the value of the cluster mass as @xmath63 mev , @xcite .
we have used these different values of the cutoff in order to solve eq . .
as we can see in fig .
[ fig : t2 - 1 ] , we find a peak in the @xmath137 energy distribution , close to @xmath138 mev using different values of @xmath102 . this result can be identified with a new @xmath58 state as a consequence of the three - body interaction . in tab .
[ table : mass - width - j3 ] we show the mass and width associated with this new state for different values of the @xmath102 parameter .
a change of @xmath139 mev in the cutoff brings up a change of the order of @xmath140 mev in the mass , and of @xmath141 mev in the width .
thus , these values are quite stable with respect to changes in the values of the @xmath102 parameter
. the range of values of @xmath102 taken , in line with those used in @xcite , are of the order as those used in studies of related systems @xcite and can be considered of natural size @xcite .
.the mass and width of the @xmath3 state found in the @xmath0 interaction in the fca to faddeev equations . [ cols="^,^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ]
in this work we have studied the three - body system which consists of a @xmath12 meson and a @xmath132 cluster .
we have solved the faddeev equations in the fixed center approximation using the two - body interaction obtained in the unitary local hidden gauge approach .
the @xmath132 cluster is assumed to be the @xmath6 , @xmath142 state found in the previous work of ref .
@xcite . taking the @xmath4 interaction in @xmath6 from @xcite
we find a @xmath3 state of mass @xmath143 mev and width @xmath144 mev .
related to this mass is the two particle c.o.m .
energy of the @xmath4 subsystem that we find close to the mass of the @xmath5 resonant state appearing in the @xmath6 @xmath4 interaction .
we can conclude that the presence of this resonance plays an important role in the dynamics of the generation of the new three - body state .
this work is partly supported by the spanish ministerio de economia y competitividad and european feder funds under the contract number fis2011 - 28853-c02 - 01 and fis2011 - 28853- c02 - 02 , and the generalitat valenciana in the program prometeo ii-2014/068 .
we acknowledge the support of the european community - research infrastructure integrating activity study of strongly interacting matter ( acronym hadronphysics3 , grant agreement n. 283286 ) under the seventh framework programme of eu .
l. d. faddeev , sov .
jetp * 12 * ( 1961 ) 1014 [ zh .
* 39 * ( 1960 ) 1459 ] .
e. o. alt , p. grassberger and w. sandhas , nucl .
b * 2 * ( 1967 ) 167 .
e. epelbaum , h. kamada , a. nogga , h. witala , w. gloeckle and u. g. meissner , phys .
* 86 * ( 2001 ) 4787 [ nucl - th/0007057 ] .
a. martinez torres , k. p. khemchandani , l. s. geng , m. napsuciale and e. oset , phys .
d * 78 * ( 2008 ) 074031 [ arxiv:0801.3635 [ nucl - th ] ] .
l. roca and e. oset , phys .
d * 82 * ( 2010 ) 054013 [ arxiv:1005.0283 [ hep - ph ] ] .
a. martinez torres , k. p. khemchandani , d. gamermann and e. oset , phys .
d * 80 * ( 2009 ) 094012 [ arxiv:0906.5333 [ nucl - th ] ] .
a. martinez torres , k. p. khemchandani , d. jido and a. hosaka , phys .
d * 84 * ( 2011 ) 074027 [ arxiv:1106.6101 [ nucl - th ] ] .
k. p. khemchandani , a. martinez torres and e. oset , eur .
j. a * 37 * ( 2008 ) 233 [ arxiv:0804.4670 [ nucl - th ] ] .
a. martinez torres , k. p. khemchandani and e. oset , phys .
c * 77 * ( 2008 ) 042203 [ arxiv:0706.2330 [ nucl - th ] ] .
b. x. sun , h. x. chen and e. oset , eur .
phys . j. a * 47 * ( 2011 ) 127 [ arxiv:1107.0209 [ nucl - th ] ] .
m. bayar , c. w. xiao , t. hyodo , a. dote , m. oka and e. oset , phys .
c * 86 * ( 2012 ) 044004 [ arxiv:1205.2275 [ hep - ph ] ] .
y. ikeda and t. sato , phys .
c * 76 * ( 2007 ) 035203 [ arxiv:0704.1978 [ nucl - th ] ] . a. gal and h. garcilazo , nucl .
a * 864 * ( 2011 ) 153 [ arxiv:1103.4757 [ nucl - th ] ] .
n. barnea , a. gal and e. z. liverts , phys .
b * 712 * ( 2012 ) 132 [ arxiv:1203.5234 [ nucl - th ] ] .
k. a. olive _ et al . _
[ particle data group collaboration ] , chin .
c * 38 * ( 2014 ) 090001 . j. yamagata - sekihara , l. roca and e. oset , phys . rev .
d * 82 * ( 2010 ) 094017 [ phys .
d * 85 * ( 2012 ) 119905 ] [ arxiv:1010.0525 [ hep - ph ] ] .
m. bando , t. kugo , s. uehara , k. yamawaki and t. yanagida , phys .
* 54 * ( 1985 ) 1215 .
m. bando , t. kugo and k. yamawaki , phys .
rept . * 164 * ( 1988 ) 217 .
r. molina , h. nagahiro , a. hosaka and e. oset , phys .
d * 80 * ( 2009 ) 014025 [ arxiv:0903.3823 [ hep - ph ] ] .
del amo sanchez _ et al . _
[ babar collaboration ] , phys .
d * 82 * , 111101 ( 2010 ) [ arxiv:1009.2076 [ hep - ex ] ] .
a. ozpineci , c. w. xiao and e. oset , phys .
d * 88 * ( 2013 ) 034018 [ arxiv:1306.3154 [ hep - ph ] ] .
r. chand and r. h. dalitz , annals phys .
* 20 * ( 1962 ) 1 .
g. toker , a. gal and j. m. eisenberg , nucl .
a * 362 * ( 1981 ) 405 .
f. mandl and g. shaw , chichester , uk : wiley ( 2010 ) 478 p. j. yamagata - sekihara , j. nieves and e. oset , phys . rev .
d * 83 * ( 2011 ) 014003 [ arxiv:1007.3923 [ hep - ph ] ] .
j. a. oller and u. g. meissner , phys .
b * 500 * ( 2001 ) 263 [ hep - ph/0011146 ] . | in this work we study the @xmath0 three - body system solving the faddeev equations in the fixed center approximation .
we assume the @xmath1 system forming a cluster , and in terms of the two - body @xmath2 unitarized scattering amplitudes in the local hidden gauge approach we find a new @xmath3 state .
the mass of the new state corresponds to a two particle invariant mass of the @xmath4 system close to the resonant energy of the @xmath5 , indicating that the role of this @xmath6 resonance is important in the dynamical generation of the new state . |
understanding how stars interact with their environment is a general problem important for many astrophysical systems .
many types of stars are surrounded by shells , envelopes or disks of material .
such circumstellar material often participates in accretion , polar outflows , winds and disk - star interactions . in young stars ,
these processes directly effect star and planet formation and evolution . even for the closest young stars we study here ,
the stellar radius is an important spatial scale that is smaller than 0.1 milliarcseconds and it will not be imaged directly even by the next generation of telescopes .
other methods like interferometry can yield useful constraints , but the indirect techniques of spectroscopy and spectropolarimetry have much to offer .
in particular , spectropolarimetric measurements can put unique constraints on the geometry of the system as well as thermodynamic and radiative environments of circumstellar material .
many models show spectropolarimetry is a useful probe of circumstellar environments at small spatial scales .
circumstellar disks , rotationally distorted winds , magnetic fields , asymmetric radiation fields ( optical pumping ) , and in general , any scattering asymmetry can produce a change in linear polarization across a spectral line such as h@xmath0 ( cf .
mclean 1979 , wood et al .
1993 , harries 2000 , ignace et al .
2004 , vink et al .
2005a , kuhn et al .
these signatures have the potential to directly constrain the density and geometry of the circumstellar material and probe the near - star environment .
there have been several studies of young stars to date using low or medium resolution spectropolarimetry .
almost every study was focused on one class of objects and is typically based on only a few nights of observations .
though spectropolarimetric signatures have been detected in every class of star a many - epoch , high - resolution survey of any class of star has yet to be done .
high - resolution linear spectropolarimetry measures the change in linear polarization across a spectral line .
typical spectropolarimetric signals are very small , often a few tenths of a percent change in polarization across a spectral line .
measuring these signals requires very high signal to noise observations and careful control of systematics to measure signals at the 0.1% level . in order to address these issues
, we built a dedicated spectropolarimeter for the hivis spectrograph on the 3.7 m aeos telescope and performed a telescope polarization calibration ( harrington et al .
2006 , harrington & kuhn 2008 ) . in the process , we have upgraded much of the hardware , calibrations , and created a dedicated reduction package in idl .
this thesis provides a very careful and thorough characterization and analysis as well as a very large collection of data taken under a wide range of conditions in order to address these instrument - related issues .
in addition to this large set of observations , a detailed comparison between instruments was performed .
this data set was combined with recent and archival observations using another instrument as well as with previously published spectropolarimetry on three other instruments .
the resulting 5-spectropolarimeter comparisons show that the hivis spectropolarimeter , as well as the results of the survey , are verified by all spectropolarimeters .
we collected a large number of stellar observations on over 100 nights from 2004 to 2008 , monitoring roughly 30 herbig ae / be stars and 30 be and emission - line stars .
these star types are thought to have very different circumstellar environments with various combinations of winds , disks , accretion , and jets ( cf .
porter & rivinius 2003 and waters 1998 ) . however , there is still much debate over these processes .
scattering theory is the one common framework used to describe linear spectropolarimetric signatures .
the scattering theory of spectropolarimetric line profiles was originally developed for be stars but did not fit many of our herbig ae / be observations . in order to present this program in a coherent manner
, there will be several independent parts .
the differences between the star - types and geometries will be discussed in chapter 1 .
then the instrument , observing procedures and the data reduction will be presented in chapters 2 and 3 .
the detailed polarization calibration of the instrument and telescope will be discussed in chapter 4 .
chapter 5 is devoted to comparing results between instruments - measurements of several stars with many different instruments will be compared .
chapter 6 presents the herbig ae / be observing campaign and the initial h@xmath0 line profiles , stellar variability , and some individual stellar properties . before showing the detailed spectropolarimetric results in chapter 8 , the current scattering paradigm will be outlined and some examples given in chapter 7 . since the herbig ae / be observations do not conform to the current scattering theories , and a clear distinction between the hae / be stars and other stellar types must be made .
chapter 9 will present a survey of some bright be and other emission - line stars which do seem to fit scattering theory much better . in chapter 10 , a new theory based on optical pumping
will be discussed .
this theory has the potential to fit the observations significantly better than scattering theory and has a number of advantages .
the entire thesis will be summarized in chapter 11 .
herbig ae / be stars ( haebe s ) are intermediate - mass pre - main - sequence stars ( see waters 1998 for a review ) .
they are young , typically @xmath310myr old , and of spectral type a or b with masses of 2 - 8 m@xmath4 and photospheric temperatures of 7000 - 20000k .
they typically are surrounded by an envelope of material and a circumstellar disk , a result of the star formation process .
the modern criteria for a star to be classified as a haebe is that the a or b type star must have emission lines in it s spectrum and must emit an excess of infrared light from circumstellar material .
the infrared excess requirement distinguishes the haebe stars from other a and b type stars , such as the classical be stars , which are not young and have infrared excesses from free - free emission .
many haebe stars show evidence for active and dynamic circumstellar environments .
images taken in either the optical or near infra - red show large gaseous disks extending to nearly 1000au from the star .
these disks have holes in the central region .
mid- and far - ir images as well as sub - mm measurements show emission from circumstellar dust from very near the star ( ir ) out to 100 s of au ( sub - mm ) .
gaseous jets have been seen extending from the inner 10au out to nearly 1000au in some systems .
spectroscopic studies have suggested evidence for hot extended chromospheres , small amounts of @xmath5100,000k gas near the star ( cf .
bouret et al .
1997 ) , azimuthal structures such as wind - streamers and bullets , polar down - flows , accretion and cool opaque shells ( cf .
bouret et al .
1997 , bouret & catala 1998 , 2000 , bohm et al .
1996 catala et al . 1999 ) .
in a typical description of an individual haebe star based on spectroscopy many , if not all of these phenomena will be evoked to explain the observations ( cf .
beskrovnaya et al .
1994 , 1995 , 1998 , 1999 , bohm & catala 1993 , 1995 , bouret & catala 1998 , 2000 , kozlova 2004 , 2006 , 2007 , pogodin 1994 , 1997 ) .
a typical conclusion from a spectroscopic study of an individual stars will conclude that an extended chromosphere is present outside the photosphere .
the presence of sporadic polar down - flows will be outlined based on red - shifted components of helium emission lines . a disk - wind or chromospheric - wind with complex and rapidly variable structure
will be described based on blue - shifted components of hydrogen and sodium emission lines .
highly ionized species , such as nitrogen v , will be used as evidence for small clouds of high - temperature ionized gas near the star . a basic model for the circumstellar environment
will be put forward that contains many of these components as well as discussions of the evidence . with so many varied effects involved that make specific statements about temperature , density , and geometry , a new and unique constraint on the circumstellar material would be most useful .
to further develop spectropolarimetry as a technique and to compare results between different stellar systems , different star types must be observed over many epochs .
a be star is simply a b type star that shows emission lines , but they are not necessarily young pre - main - sequence stars ( see porter & rivinius 2003 for a review ) .
classical be stars , as opposed to herbig be stars are main - sequence stars that rotate quickly and show emission lines from circumstellar material .
this circumstellar material is not a result of a formation process , as with the young herbig be stars .
the circumstellar material is a direct consequence of the stellar properties .
the emission from the circumstellar material itself is transient .
there are some models that show evidence for accretion and/or winds in these systems , but the geometry and density are thought to be quite different .
for instance , the region where h@xmath0 emission forms is said to be a few stellar radii in haebe stars but up to 30 in be stars . comparing the spectroscopic modeling of ab aurigae in bouret & catala 1998 with
the interferrometrically resolved h@xmath0 in @xmath6 tau from quirrenbach et al .
1994 shows a striking contrast .
given that polarization arises from asymmetries in any system , a formation region that is ten times larger in size will have an inherently different asymmetry .
when talking about polarized radiation in most astronomical settings , the stokes vector ( * i * ) is usually the best description of the polarized light .
the stokes vector is an array of 4 numbers , iquv , that gives the intensity of the the light in different polarization states .
i is the total intensity .
q is the difference in intensity between 0@xmath7 and 90@xmath7 polarization states .
u is the difference between 45@xmath7 and 135@xmath7 polarization states .
v is the difference between the left and right handed circular polarization states .
@xmath8 these parameters can be normalized to give the relative fraction of light in each polarization state .
i is always positive whereas quv can be positive or negative with values from 0 to @xmath9i .
for example , if the stokes vector is [ 2,1,0,0 ] then the intensity is 2 , the amount of light polarized at 0@xmath7 ( + q ) is 1 and there is no net polarization at 45@xmath7 , 135@xmath7 , left circular or right circular .
the fraction of light in the q state is then q = q / i=50% . in the stellar scattering context ,
once you have the stokes vector of the scattered light and the stellar light , which is typically unpolarized , you can define the normalized stokes parameters as the ratio of the flux in a particular polarization state and the total flux : @xmath10 @xmath11 the combined stellar and scattered flux , @xmath12 & @xmath13 , are the total flux received from the source while the flux in any specific stokes parameter , @xmath14 or @xmath15 , are the individual polarized fluxes .
you can also define the degree of linear polarization , p , as : @xmath16 you can also project the primary vibrational direction of this polarization into your instrument s coordinate system as : @xmath17 spectropolarimetry is a technique that measures the intensity as well as the polarization of light as a function of wavelength .
this can be done many ways .
some instruments use broad - band filters and polarizers on an imaging instrument to measure the polarization in a few different colors ( bvr ) .
others use low or high resolution spectrographs ( r@xmath18100 to 300000 ) with polarizing optics to spectroscopically measure the polarization of light .
solar physicists have been using spectropolarimetric measurements of the sun for a century now and have observed a wealth of effects . typically , high resolution ( r@xmath5 100,000 ) polarized spectra are taken of various regions of the sun ( corona , sunspot umbra , photosphere , etc ) .
this type of measurement is so common that the polarization spectrum is designated as the second solar spectrum " .
recently , people have begun to observe and model high resolution polarized spectra in a night - time astronomical setting .
the star is unresolved and the polarization signatures are an average of all the stellar components .
when you can resolve polarization changes across a single spectral line , significant effects are seen .
for example , symbiotic binaries show 10% polarization across raman - scattered lines ( cf . harries & howarth 1996 ) . recent spectropolarimetric observations of herbig ae / be stars reveal numerous small linear polarization variations across h@xmath0 ( cf .
vink et al .
2002 , 2005b , mottram et al .
2007 , harrington & kuhn 2007 ) .
our recent observations were the first to systematically search for and find spectropolarimetric variability with wavelength and epoch near obscured h@xmath0 emission ( harrington & kuhn 2007 ) . for most of our targets and a few in the literature ,
the polarization change is clearly @xmath19 coincident with the h@xmath0 emission peak .
the polarization changes only in and around the absorptive part of the line profile , and the polarization near the emission peak is nearly identical to the continuum polarization .
thomson scattering has been suggested to describe the earlier linear spectropolarimetric variations seen across emission lines in haebe stars .
electron scattering produces differential linear polarization effects centered on the h@xmath0 line center ( cf .
vink et al .
2005a ) in this _ scattering _ paradigm .
a depolarization effect , where the polarized stellar continuum is diluted by unpolarized h@xmath0 emission , is invoked with o , b , and wolf - rayet ( wr ) stars ( cf .
mclean 1979 , oudmaijer & drew 1999 , harries howarth & evans 2002 ) to account for these polarization variations .
apparent _ polarization effect near the absorptive line profile is possible via this mechanism , but it must be accompanied by a corresponding signature ( undetected in our survey ) in the emissive component .
thus , scattering models can describe some classes of stars , but not the data we will describe .
we note that magnetic effects may also contribute a linear polarization signal via the hanle or zeeman effects .
however , observations of circular polarization in herbig ae / be stars and many other star types have shown that the stellar magnetic fields are much too small to cause the observed amplitude of polarization seen in our survey ( cf .
ignace et al .
2004 , wade et al .
2007 , catala et al . 2007 ) .
the advanced electro - optical system telescope ( aeos ) is a 3.67 m , altitude - azimuth telescope , located on haleakala mountain , maui , hawaii .
the spectrograph uses the f/200 coud optical path , with seven reflections ( five at @xmath20 incidence ) before coming into the optics room .
this is illustrated in figure [ fig : aeostel ] , taken from the zemax ( industry standard optical design software ) telescope design .
the optics room , shown in figure [ fig : coude_layout ] , contains visible ( 0.5 - 1.0@xmath21 m ) and ir ( 1.0 - 2.5@xmath21 m ) cross - dispersed echelle spectrographs .
the visible spectrograph has 2 gratings which cover 500 - 770 nm and 640 - 1000 nm at resolutions of 12800 to 48900 .
the common fore optics includes three fold mirrors , collimator , k - cell ( image rotator ) , fold - mirror , tip - tilt correction mirror , and a reimaging spherical mirror .
a movable pickoff mirror allows for switching between the visible and ir channels ( thornton 2002 , thornton et al .
2003 ) .
the spectropolarimeter is designed for use with the visible arm of the aeos spectrograph .
the visible arm is shown in figure [ fig : opben ] . from the pickoff mirror , shown in the bottom
left corner of the figure , the beam passes through the slit in the slit - mirror , reflects off a fold mirror , the collimator , echelle , collimator , long mirror , collimator , fold mirror 2 , and finally the cross disperser before being imaged on the ccd by a 5 lens camera .
there is a slit - viewer camera that is focused on the slit - mirror that allows real - time monitoring of the object .
the spectropolarimeter was mounted just downstream of the slit , near the image plane .
the plate scale is 0.16@xmath22 pixel@xmath23 .
the cross disperser s red setting was used for all observations before aug 28th 2007 and the blue setting was used afterwards ( to observe other wavelengths for other projects ) .
the spectropolarimetry module for the aeos spectrograph consists of a rotating achromatic half - wave plate and a calcite savart plate .
the savart plate is two crossed and bonded calcite crystals which separate the incoming , uncollimated beam into two parallel , orthogonally polarized beams displaced laterally by 4.5 mm .
the half - wave plate allows the rotation of the linear polarization of the incoming light to allow complete measurement of the linear polarization of the incident light .
the beams are displaced in the spatial direction ( along the long axis of the slit ) . in a savart plate , the two beams ( extraordinary and ordinary rays )
are swapped between the two crossed crystals so both displaced rays have the same optical path length through the calcite .
this design contrasts with other spectropolarimeters which use fresnel rhomb retarders and a wollaston prism to separate the polarization states ( donati et al .
1999 , kawabata 99 goodrich 2003 , manset & donati 2003 ) .
the small deviation in the image plane splits each spectral order detected on the ccd into two orthogonally polarized orders separated by a few pixels , from which orthogonally polarized spectra are extracted separately . for example , the red cross disperser setting originally shows 19 orders across the ccd , but after insertion of the savart plate , it shows 38 ( 2 polarizations for 19 orders ) as illustrated in figure [ fig : chip ] .
these components were added to the zemax optical ray trace model of the spectrograph to assess the aberrations and determine the specifications of the optics in order to produce the proper polarized - order separation on the ccd ( @xmath1825 pixels ) .
the aberrations induced by the waveplate and savart plate were predicted to be smaller than half a pixel leading to insignificant image degradation . a deviation of 4.5 mm at the slit was found to separate the polarized - orders by @xmath1824.5 pixels on the ccd .
a 60 mm savart plate produces this separation ( 0.075@xmath2460 mm ) .
a halbo optics inc . savart plate with a 15 mm aperture was purchased and tested as well as a bolder vision optik inc .
achromatic half - wave plate and a motorized , closed - loop rotation stage . the half wave plate retardance was measured with a lab ocean optics spectrograph and two crossed polarizers . the retardance varied roughly quadratically with wavelength at the 5% level from 5000 to 8000 , the useful range of the polarizers ) and was 0.49 waves retardance at h@xmath0 .
the transmission of the savart plate was above 95% for these wavelengths .
after calibration and testing of the half - wave plate , savart plate , and rotation stages , the instrument was assembled on the spectrograph on a movable plate which allows for the spectropolarimeter to be inserted or removed from the beam in under a minute .
a dekker was mounted upstream of the slit to reduce the slit length from 15@xmath22 to @xmath187@xmath22 to minimize stray light introduced from the displacement .
this also assures polarized order separation on the ccd .
flat fields taken after installation showed that the polarized orders were each @xmath1830 pixels across ( @xmath184@xmath22 ) with a separation of 7 pixels between each polarized beam . to test the equipment on the telescope , a linearly polarizing filter was mounted just upstream of the the spectropolarimeter , in order to feed 100% linearly polarzed light to the spectropolarimeter .
hivis detected 95 to 98% polarization over the useful range of the polarizer ( 5000-8000 , orders 0 - 12 ) .
this shows that hivis can detect linearly polarized light with the spectropolarimeter very efficiently .
there have been many hardware changes over the last few years .
these were done to repair and improve the overall performance and reliability of hivis .
the original ccd s failed and were replaced twice .
the polarimeter has been remounted with better alignment and less vignetting .
a new tip - tilt guiding corrector has been installed and calibrated .
a hard - mounted , adjustable dekkar and polarization calibration mount was installed to allow for quick , easy , and repeatable calibration and adjustment .
the failed ccd - id20 s were replaced with similar copies , but an unresolved hardware / rfi issue caused this repair to fail .
the original science camera failed and has been replaced by a new commercial 1k@xmath25 pixis camera ( princeton instruments ) and is working wonderfully .
the calibration stage had lenses mounted that increased the flat - field flux by a factor of 6 and the thar lamp flux by a factor of 150 .
these improvements greatly increase the ease and accuracy of spectral calibration .
the spectrograph was designed to use two 2k@xmath244k lincoln labs high resistivity ccd - id20 s mounted side by side with a 15 pixel gap to form a 4k@xmath244k array .
this setup was in use until september of 2006 .
the ccd was set to rebin 2x2 so that the two output images are each 1024@xmath242048 pixels .
the original science camera ccd s were functional until october 2004 when one of the two intermittently failed . after removing and testing the camera
, it was determined that the bond - wires had come loose .
the one good ccd was replaced and the camera was remounted for a july 2005 run ( deep - impact spectropolarimetry run , harrington et al .
shortly thereafter , the ccd performance became intermittent .
the camera was then removed for repair . in the summer of 2006
, it was decided that both chips were irrepairable .
a single spare ccd - id20 was installed for the august 9 , 21 , 26 , and sept 12 - 14 , 2006 observing runs .
this ccd was quite flat and free of cosmitic defects .
the camera worked well for some observations , but an intense , non - repeatable noise - signature was present on some nights , even after many attempted hardware fixes throughout the night .
the noise remained throughout the following months testing at the telescope , but was not present in any lab tests .
it is not certain what exactly is causing this error , but the non - repeatable nature of the noise as well as the complete saturation of the affected pixels , makes data with this ccd unusable .
rfi is suspected because of the pulsed ringing nature of the noise which follows the readout direction , and the presence of this noise at the summit only .
on 9 - 18 2006 , this spare ccd - id20 , as well as the entire dewar and camera mount , was taken off and replaced with a smaller commercial pixis 1024br 1k@xmath25 array by princeton instruments on a custom - fabricated mount .
this camera , with slightly smaller pixels , performed quite well .
the peak qe is 95% with @xmath570% qe past 900 nm and 85% qe at h@xmath0 , making the performance of this camera almost identical to that of the hivis science camera , although with the loss of complete focal plane coverage .
the original ccd - id20 chips had a plate scale of 0.16@xmath22 per pixel and a dispersion of 0.07 per pixel .
the original ccd s were rebinned 2@xmath242 onboard and the read time was about 1 minute .
the pixis camera had slightly smaller pixels ( 13@xmath26 ) and thus a slightly greater plate scale and dispersion .
the pixis camera is run in a unique way - with winview software on a standard windows - xp laptop .
the camera has gui - set read - rate and gain parameters with a read rate options of 100 khz ( 11se ) or 2 mhz ( 0.5sec ) and gains of 1 , 2 , or 3 . since the stellar - spectropolarimetry program required near - saturation photon counts , the pixis camera , while slightly oversampling , was perfectly acceptable , especially with the hour - per - night of integration gained with the fast read time ( 0.5 s vs. 60 s ) . on 11 - 17 - 2006 , some lenses and a mirror were installed on the calibration stage to increase the flux of calibration light in the spectrograph .
the lens used with the flat field lamp focused the light onto the diffusion screen much more efficiently than the originally naked bulb , and caused a 6@xmath24 increase in the flat - field flux .
the same lens setup when used with the thar lamp caused a similar increase in flux , but the improved thar exposure was still faint .
originally , only 3 lines in the h@xmath0 order were seen at 100 adu above background in a 20 minute exposure with the widest ( 1.5@xmath22 ) slit !
even a factor of 6 increase was nt enough .
another mirror was installed on top of the diffusion screen to bypass the diffusion .
this increased the flux to a total of @xmath18150@xmath24 the previous levels though introducing spatial structure to the thar exposures .
this improvement allowed for accurate line identification because many lines could be detected with the smallest slit at the highest resolution , allowing for a better comparison with the line lists and much more accurate wavelength solutions . on 11 - 17 - 2006 ,
a new dekkar was fabricated and mounted on the common - optics bench .
this was initiated because of the need to be able to move the dekker in and out of the beam quickly , easily , and repeatably by a single person .
the dekker is required to reduce the length of the slit from the original 15@xmath22 to @xmath38@xmath22 to ensure complete polarized - order separation at the focal plane .
this dekkar has much more accurate mounting and an adjustable width , allowing for more precise alignment and control of the slit length , leading to proper order separation at the focal plane .
the dekkar is not a true knife - edge and is mounted roughly 2 cm away from the slit - mirror ( and hence image plane ) which leads to a @xmath188 pixel wide edge to each order .
this is completely acceptable for the purpose of the survey .
a removable polarization calibrator unit was installed on the new dekkar mount .
a polarizing calibration optic ( linear - polarizer , quarter - wave - plate , etc ) can be mounted just up - stream of the polarimeter to allow the input of a known polarization state to the spectropolarimeter for calibration .
this type of mount allows for night - to - night calibrations as well as efficient trouble - shooting to be done easily and repeatably . on 10 - 26 - 2006 , the polarizing optics ( savart plate and wave - plate ) were remounted closer to the slit with more robust and precise hardware .
the original polarimeter was mounted fixed on posts and had to be manually aligned .
the mount also required 3 cm of optical path between the image plane ( slit ) and the entrance of the savart plate . since the clear - aperture of the savart plate was only 1 cm and the beam diverges from the slit , the larger the distance from the image plane , the more difficult the alignment , and the greater the possibility of vignetting .
the savart plate was remounted on a closer - fitting , remote - controllable xy - stage that put the savart plate 2 cm closer to the image plane , reducing the vignetting .
both the savart and the waveplate were then mounted on tip - tilt controllable rotation stages , allowing for a more precise alignment between the waveplate , savart plate , and optical axis . on 11 - 28 - 2006 ,
a replacement tip - tilt corrector was installed .
the sbig stv unit takes an image of the slit - mirror , calculates the centroid position of the star on the slit , and sends a correction signal to a tip tilt mirror mounted on a piezo .
the tip - tilt piezo voltages and alignment mirrors were recentered and the unit was recalibrated over the following month .
the tip - tilt can not centroid on a target star ( v@xmath186 - 8 ) faster than once per second or so , and thus can only correct slow guiding errors . since aeos is specially designed for air - force use , it does not have a normal sidereal - guiding tracker scope and thus is subject to slow pointing drift .
depending on how recently a pointing model has been performed , the drift can be significant .
this is completely unacceptable and the tip - tilt corrector is essential to keep the star centered on the slit .
the polarimeter has been remounted with better alignment and less vignetting .
a new tip - tilt corrector has been installed and calibrated .
a hard - mounted , adjustable dekkar and polarization calibration mount have been installed and allow for quick , easy , and repeatable calibration and adjustment .
the failed ccd - id20 s were replaced with a similar copy , but an unresolved hardware / rfi issue caused this repair to fail .
a new commercial 1k@xmath25 camera was mounted in place of the old science camera and is working wonderfully .
the calibration stage had lenses mounted that increased the flat - field flux by a factor of 6 and the thar lamp flux by a factor of 150 .
these improvements greatly increase the ease and accuracy of spectral calibration .
the stokes parameters i , q and u fully specify the linear polarization state of light .
q is the difference between horizontal and vertical polarizations .
u is the difference between + 45@xmath7 and -45@xmath7 polarizations .
typically , the fractional stokes parameters , q = q / i and u = u / i , defined in chapter 1 , are presented to quantify the fraction of the light which is polarized .
these are measured by taking the fractional difference between the orthogonally polarized spectra .
the observing sequence with the aeos spectrograph is to take spectra at 0 , 22.5 , 45 , and 67.5 degrees rotation of the half - wave plate with respect to the axis of the savart plate , hereafter called orientation 1 , 2 , 3 , and 4 .
incoming light linearly polarized at some angle to the fast axis of the half wave plate will have its linear polarization states rotated by twice that angle . for example , incoming light polarized at 22.5@xmath7 with respect to the wave plate axis will exit polarized at 45@xmath7 ( stokes u @xmath27 q ) . since the axis of the savart plate remains fixed , the rotation of the wave plate by 45@xmath7 will move linearly polarized light from one polarized beam to the other , swapping their location on the ccd and reversing their sign in the fractional polarization calculation ( q @xmath27 -q ) .
this allows for cancellation of systematic errors ( derivative , misalignment , ccd response , etc . ) in the stokes q and u calculations .
the waveplate angle can be controlled to 50@xmath22 .
the observing sequence is illustrated in figure [ fig:4ori ] .
a linear polarizer was mounted upstream of the spectropolarimeter aligned with the waveplate s axis and a normal data set ( 4 waveplate orientations ) was obtained .
notice how the bright order is swapped ( top to bottom ) with a 45@xmath7 rotation of the waveplate and how the 22@xmath7 and 67@xmath7 orientations show uniform intensity for both spectra .
the rotation of the waveplate will allow the linearly polarized light to be sampled by moving it from one polarized order to the other in a systematic way .
a frame from the slit - viewer camera is shown in figure [ fig : slitqu ] with the general geometry of the slit , dekker , and room as seen by the slit viewer camera .
the slit is parallel to the floor .
+ u is parallel to the slit , and + q is rotated @xmath28 counter clockwise , pointing to the upper right . since the image rotator was not used , the projection of the slit on the sky is a function of pointing and is not constant .
the simultaneous imaging of orthogonal polarization states also allows for a greater efficiency observing sequence and for reduction of atmospheric and systematic effects .
since both polarized spectra are imaged simultaneously with identical seeing , a single polarization measurement ( q or u ) can be done with a single image , and consistency between images is easy to quantify .
the polarization angle is measured in the telescope frame and a projection of the slit s position angle onto the sky is done with the k - cell .
the aeos image rotator was not used so that the absolute angle of the slit on the sky was not fixed .
however , the orientation of the k - cell mirrors in the optics room was fixed , making the polarization calibration much easier .
the spectropolarimetric reduction pipeline can be broken into a few basic parts : 1 ) combining flats , darks , biases , 2 ) setting up the coordinate grids as x - pixel , y - pixel , slit - tilt and wavelength calibration , 3 ) calculating a first - pass spectrum by simply averaging all pixels at a given wavelength , 4 ) calculating more spectra based on optimal - extraction routines , with cosmic ray rejection and flux - weighting , 5 ) doing the polarization calculations , and 6 ) simulating and quantifying the systematic errors . the first step is to produce an individual flat field , calculated as an average of many individual flats .
typically , an average of 15 - 40 individual exposures is used with a 3-@xmath29 outlier rejection , a so - called clipped - mean , to produce an extremely accurate calibration . since this program depends on very high s / n measurements ,
a minimum of 20 flat field images of 10000 adu each is required to get good s / n .
the dark and bias calibration is done in a less - than - traditional way because of a slowly - fluctuating , non - repeatable background level .
this effect is actually quite common with most detectors .
there was an overscan region in the original science detector with the ccd - id20 s which allowed measurement of the changing dark levels throughout the night .
the replacement pixis 1024br array is significantly smaller than the focal plane and is always exposed to light .
both the original lincoln labs ccd - id20 s and the pixis detector showed a significant ( @xmath5few% ) drift in the bias levels throughout a night .
the pixis camera had no overscan region , but the median values of all the darks taken over many months did show variation in the overall background levels . these variations with this instrumental setup and background levels required a background subtraction from each frame using the unexposed pixels outside the echelle orders as this directly tracked the background levels of the chip and reduced the systematic uncertainties . for the short exposures of this program , the dark current was small , and is automatically corrected with the frame - by - frame background subtraction .
since the instrumental setup and the order locations on the ccd s has changed much over the last few years , and changed quite often during the instrument work done in the fall of 2006 , setting up the coordinate grid requires more user - input than most dedicated reduction scripts .
the user must supply a guess at the polarized order locations , typically 5 points to fit a 3rd - order polynomial , and a bright stellar image taken that night to serve as a template for the curvature of the orders .
figure [ fig : coord ] shows a bright standard with the narrow h@xmath0 line near the center of the 1k@xmath25 array .
the code sets up an initial polynomial fit to these user - input points , and then takes spatial ( vertical ) cuts at each spectral ( horizontal ) pixel .
each of these spatial cuts are then fit by a gaussian to determine a robust location of the peak intensity in the spatial direction across the ccd , reducing the influence of cosmic rays or chip defects .
these peak intensity positions allow for a robust fit to the order shape on the ccd with a slope of around 5% ( crossing a spatial pixel every 18 or 20 spectral pixels ) .
the orders are fit independently and the shapes are the same to the 0.1 pixel level ( @xmath180.2% of an order width ) .
the order separation , caused by the savart plate , is then calculated as the median difference between the two polynomial fits . for the pixis camera ,
this was 58 - 59 pixels .
the residuals between the calculated peak intensity and the polynomial fits are typically under half a pixel and show the expected oscillation with wavelength as spatial pixels are crossed .
the star observed for the template is never completely centered in the slit from small guiding errors .
once these order shapes are calculated from the stellar observations , the polynomials are adjusted to reflect the true slit center .
this is done using the flat - field .
first an average spatial profile ( slit illumination pattern ) is calculated for the flat field , centered on the stellar template by simply averaging the flat field along the order template in the spectral direction .
these spatial profiles , shown in figure [ fig : flatslice ] clearly show the slit boundaries , where the flat field intensity falls off sharply .
the flat - field spatial cuts are cross - correlated to find the offsets from the stellar order - shape polynomial fits to the center of the flat - field spatial cuts .
once the offset from the stellar template to the flat - field region is calculated , the full coordinate grid can be computed with the order centers , edges , and background locations specified . once the coordinate grid is set ,
a calibrated flat field can be created , adjusting for the blaze function , and wavelength dependence of the flat - field lamp ( blackbody ) .
a flat - field is supposed to correct for pixel - to - pixel gain and field - dependent efficiencies only .
the flat - field values should not introduce any blaze , lamp or spatial illumination dependencies .
this means that the flat - field must be corrected for it s inherent spatial and spectral structure . to correct the flat field spectral
dependence ( horizontal ) , a simple flat - field spectrum is made for each polarized order by averaging all pixels spatially at a given spectral position .
this spectrum has a bit of intrinsic noise in it so a low - order polynomial fit is made .
the flat - field image then has this spectral dependence removed .
the spatial structure is corrected by dividing each spatial cut along the flat - field spectrum by the average spatial profile , removing the spatial dependence of the flat ( inhomogeneous slit illumination ) at each wavelength .
this makes the flat - field much closer to a true pixel - to - pixel gain measurement , shown in figure [ fig : calibratedflat ] . with the recent calibration - stage improvements that have implemented , it became possible to do very accurate wavelength calibration using low order polynomial fits to at least six clear , unblended lines , just over the h@xmath0 spectral region . with a @xmath5600x increase in thar flux ( 15x without the diffuser - bypass mirror ) , the smallest slit ( 0.35@xmath22 ) at highest resolution ( r=49000 )
can be used to clearly identify lines and compare the lines with standard line - lists obtained from the noao spectral atlas .
the new hivis detector has a wavelength scale of 0.037 per pixel ( 1.7 km / sec per pixel at h@xmath0 ) .
thus the rotation of the earth can produce up to a @xmath30 pixel shift and seasonal variations can approach 40 pixels .
the wavelength calibration is done in a two - step process . a th - ar spectrum taken with the 0.35 "
slit is used to identify the lines .
the th - ar spectra are calculated with the simple average over all illuminated pixels .
the unblended lines are identified and the peaks are fit by a gaussian to determine the line centers .
a 3@xmath31 order polyomial fit to the line centers is used to create a wavelength array .
the noao atlas spectrum , hivis spectrum , and a comparison ab - auriga spectrum are shown in figure [ fig : thar - fit ] .
the triangles mark the 6 lines used in the polynomial fits .
since all stellar observations were done with the 1.5@xmath22 slit , a 4-orientation sequence of thar spectra are taken with this slit and then reduced with the simple - average method .
the resulting lower - resolution spectrum is cross - correlated with the higher - resolution spectra produced by the 0.35@xmath22 slit to determine any shifts between slits and waveplate orientations , the beam wander or wobble , which is a small fraction of a pixel .
the average slit tilt is then calculated to modify the coordinate - grid for the reduction program .
the tilt is a spectral - direction shift of @xmath187 pixels between the bottom and top of each order ( across the 50 pixel slit width ) .
this shift is calculated as a cross - correlation of a spectral cut beween each spatial position along the order .
this tilt is then implemented as a integer - pixel shift of the coordinate grid at each spatial point .
there is a very significant increase in resolution when the tilted - coordinates are used .
figure [ fig : thar - align ] shows the spectral cuts before and after alignment , as well as the intensity of the line calculated with both tilted and un - tilted coordinate grids .
the spectral resolution derived from the th - ar line fits goes from r=30000 to r=49000 when using tilted coordinates .
once the tilted coordinate grid , gain - table , and wavelength calibration is done , the data can be processed .
each exposure is divided by the gain - table ( calibrated flat ) .
a background spectrum is made by averaging the unexposed background region outside the polarized orders .
this background is noisy and so a linear fit to this background is subtracted from each order to remove the dark current and bias of the detector . to create a simple
, first - pass at the spectrum , every pixel at each wavelength was simply averaged to compute the intensity .
at the same time , an average spatial profile ( psf ) is constructed by averaging all these spatial cuts spectrally . while this method does not correct for cosmic rays , and does not weight the pixels by their fractional - stellar - flux or psf , this simple average spectrum provides a starting point for more sophisticated reductions .
in many parts of the reduction package , one curve must be shifted and scaled to another .
a least - squares routine that allows the calculation of the shift and scaling coefficients between two input curves was developed as part of the dedicated software package .
the routine assumes that the scale and shift from one curve to another can be written in terms of the input curve and it s derivative .
to illustrate the routine a shift - n - scale between the two polarized spectra , @xmath32 shifted and scaled to @xmath33 , is calculated below :
@xmath34 i is the intensity as a function of ccd pixel x , a is the relative gain between the two orthoganally polarized spectra , usually @xmath181 , and b is the spectral shift in pixels , usually @xmath350.8 pixels .
the spectrum @xmath36 can be expanded to first order as below :
@xmath37 rewriting the constants @xmath38 and @xmath39 and bringing all terms to one side gives the difference between the spectra @xmath33 and @xmath32 scaled and shifted by a and b. the least squares routine minimizes the function below for all pixels ( x ) : @xmath40 expanding the square and collecting terms while implicitly assuming that each term is a sum over all pixels gives : @xmath41 to solve for k@xmath42 and k@xmath43 two equations must simultaneously solved . setting the derivative of this equation with respect to k@xmath42 and with respect to k@xmath43 equal to zero gives two equations that can be solved for the coefficients : @xmath44 @xmath45 which gives the coupled equations that can be solved for the coefficients : @xmath46 = \left [ \begin{array } { ll } i^2_1 & i_1 \frac{di_1}{dx } \\
i_1 \frac{di_1}{dx } & \frac{di_1}{dx}^2 \end{array } \right ] \left [ \begin{array } { c } k_1 \\ k_2 \end{array } \right ] \end{aligned}\ ] ] the derived coefficients @xmath47 and @xmath48 are then used to calculate the scaled and shifted functions .
an example of this routine applied to a spectrum of ab aurigae is shown in figure [ fig : shfscl ] .
a single spectrum was replicated and shifted by one pixel and corrected by the shift - n - scale routine .
the routine uses a smoothed profile to calculate the derivative with a variable amount of smoothing .
for all the data presented here , a smoothing of 3 pixels was used .
this illustrates how a misalignment of the savart plate , simulated as a spectral shift of 1 pixel between the orthogonally polarized spectra , produces a fractional difference of order 5% from this misalignment .
this signature is a systematic error that follows the derivative of the spectral line .
the scale and shift operation minimizes this derivative signal as shown ( amplified ) in the bottom line in the plot .
the residual noise is at the level of 0.2% and has some residual systematic error in it from the smoothing used to calculate the derivative , but the derivative - based error has been attenuated by a factor of about 20 .
the shift - n - scale routine also has some nonlinearity to it , and is best used for sub - pixel shifts when cross - correlation techniques fail .
for example , when shift - n - scaling the original ab aurigae spectrum to the simulated one in figure [ fig : shfscl ] , the routine reproduced the shifts to better than 0.05% for spectral shifts of 5 pixels , far greater than the range needed . however , when shift - n - scaling the simulated curves to the original curve , the calculated shifts had an error of @xmath50.1 pixels for spectral shifts @xmath50.8 pixels .
this shows that for even strongly varying line - profiles , the shifts are reproduced very well and , when used with cross - correlation techniques for single - pixel alignment , can produce very accurate sub - pixel alignments .
this first - pass spectrum and average spatial profile are projected back on the original spatial cuts with the shift - n - scale routine .
the expected noise at each spatial pixel is found ( using estimates of the gain and read - noise ) .
the standard deviation of each spatial pixel at each wavelength is then calculated and used to find any cosmic - ray hits as cosmic - rays ( and chip defects ) which will be strong devations ( @xmath49 ) from the spatial profile .
an example of this cosmic ray removal routine is shown in figure [ fig : cosray ] .
the average spatial profile is the dashed line .
the spatial cut at this wavelength shows a few pixels deviating strongly from the average spatial profile from a cosmic ray hit .
these pixels are identified and replaced with the shifted - and - scaled spatial profile , greatly reducing the error in the spectrum .
once this cosmic ray removal routine is done , a spectrum is calculated as the simple average of all the pixels .
the corrected spatial cuts are passed on to more optimal spectral extraction routines .
the first of the optimal extraction routines uses the shift - and - scale routine to fit the average spatial profile ( psf ) to each of the individual spatial cuts at each wavelength .
this reduction process assumes that the spatial profile does not vary across the order , an assumption that all optimal reduction packages adopt ( horne 1986 , marsh 1989 , donati et al .
1997 , cushing et al .
the calculated error in the shifts between the optional fit and the calculated order location is dependent on the flux , but is typically less than 0.1 pixels .
another set of spectra are calculated as simple gaussian fits to the spatial cuts . while the psf ( spatial cut ) is not entirely gaussian , and certain seeing conditions or tracking - errors lead to not - quite - gaussian spatial cuts ( psf s ) , the fits are usually robust , and match the shift - n - scale and simple - average specta well .
the standard idl gaussfit routine is applied to each spatial cut , with the initial estimates for the gaussian as : peak intensity is twice the computed average - spectra , centered on the order , with a fwhm of one - quarter the order width and zero background .
the routine works well when the spatial cuts ( psf s ) are roughly gaussian .
thus , there are three spectra with different systematic effects ( simple - average , shift - n - scale , gaussian - fit ) - which are all included in the polarization analysis to probe the polarimetric reduction codes and systematic errors inherent in these reductions . in practice ,
the exact choice of reduction parameters is unimportant .
figure [ fig : difred ] shows the simple - average method with and without a flat field or tilted coordinates , and varying order widths from 43 to 49 pixels ( covering the edges of the illuminated region , see figure [ fig : flatslice ] ) .
the tilted coordinates do add a significant increase in resolution , but for the high signal - to - noise spectropolarimetric observations the spectra are rebinned to lower resolution anyways .
the flat field does increase the signal - to - noise of the intensity calculation , but as shown below , the basic polarization calculations are independent of the flat field .
the aeos telescope is designed for military applications and does not have a standard siderial guiding telescope .
there are significant tracking errors that can accumulate and there are offsets induced by manual telescope - operator corrections .
the operation requirements set out by the afrl require that there is no control over the guiding by any outside equipemnt , and that the telescope operators completely control the pointing of the telescope .
there are a number of data set s where an operator s manual steps caused a non - gaussian spatial profile .
this inhomogeneity requires the reduction program to calculate the spectra using all three methods .
the simple averaging and shift - n - scaling methods are the most robust .
these three different reductions vary slightly between each other .
the gaussian - fit and simple - average methods vary by only a few parts - per - million when the spatial cuts ( psf s ) are gaussian . however , the shift - n - scale method and the simple - average methods vary slightly across the line profile , giving an overall uncertainty in the shape of the intensity profile . however , the spectropolarimetry is calculated as the difference between line profiles on the same image .
since the same reduction method is applied to each polarized order in the same way for each polarization measurement , it is only any differential error that would cause systematic errors in the spectropolarimetry .
the reduction package uses all three calculation methods to assess the systematic errors that might be present in each method , or in the subsequent processing of the spectra .
the focus in this work is on spectral and spectropolarimetric variability . with the many spurious polarization effects such as instrumental , telescope , interstellar , and intrinsic continuum polarizations ,
a continuum - polarization subtraction must be done to isolate the change across any individual spectral line . since spectral variability
is always referenced to continuum , no spectrophotometric or flux calibration is done , though standards are observed for future work .
the continuum is simply subtracted as a linear fit to the intensity on either side of the spectral line .
this continuum subtraction normalizes the slope of the black - body of the star , removes the instrumental continuum polarization signals , and removes the instrumental blaze efficiency .
continuum subtraction is done by averaging 20 pixels sufficiently far enough away from chip edges .
there is a variable dark signal that curves the echelle orders near the edge of the chip , 10 - 20 pixels wide , in a non - repeatable way , so this region is avoided .
cirrus , tracking errors , and wind - bounce all influence the throughput of the spectrograph and thus , even on a spectrophotometric night , a significant variation in the continuum flux is sometimes seen . the data taken before 10 - 27 - 2006 was imaged using an echelle order which placed the h@xmath0 line near the edge of the chip .
this placement was done before some of the instrumental corrections , and before the remounting of the pixis camera to a different region of the focal plane . in the more recent data , h@xmath0 is closer to the blaze - efficiency peak and is much closer to the center of the focal plane .
the continuum fits to the ccd - id20 images were not as accurate because the red wings of the h@xmath0 line were near the order edge , giving rise to a small ( @xmath181% ) error in slope estimate .
however , this uncertainty is negligable for data taken on each night because the continuum is calculated in the same way for each instrumental setup on each night .
this only produces an uncertainty in the night - to - night variability of the line profiles , and in the overall slope of the calculated continuum and continuum polarization .
h ] at this point it is useful to demonstrate the coordinate grid and basic properties of the reduction package by demonstrating the wavelength dependence of the point - spread function properties .
several authors have argued that the change in centroid location ( brightness peak ) or full - width at half - max ( fwhm ) can be used as evidence for binarity in a set of spectroastrometric calculations ( beckers 1982 , bailey 1998 , baines et al .
spectroastrometry in principle is a technique that can detect unresolved binaries as a change in spatial profile width ( psf ) and location ( centroid ) across an emission or absorption line where the relative brightness of the two binary components significantly varies across the line .
for example , if one star has an emission line and the other an absorption line , the centroid will shift towards the emitter in the core of the emission line .
if the binary is significantly wide and bright , the fwhm would also decrease across the line as the relative contribution to the psf from the wide companion is reduced .
this method makes use of the assumption that the psf across a line or spectral order does not change much , and that any significant change can be attributed to the source .
some have even developed routines to separate the fluxes from the individual binary components ( porter et al .
hivis has been used to observe a number of young or emission - line stars as part of a large survey ( harrington & kuhn 2007 ) all of which have strong h@xmath0 lines .
given the reported detections and potential of this technique , the feasibility and capability of hivis to do similar work will be outlined .
h ] binaries are common among pre - main - sequence stars and this technique can easily be applied to every data set taken with this instrument .
it also provides a good check on any systematic errors such as order mis - fitting or psf mis - fitting since any systematic error will show up with this measurement technique .
since the centroid of a spatial profile on most spectrographs can be calculated to a fraction of a pixel accurately in high signal - to - noise data , this technique has the capability to detect binarity to less than 30mas separations or a third of a pixel when the pixels are 130mas ( cf .
bailey 1998 , baines et al .
normally , with a long - slit spectrograph , the slit of the instrument is rotated to a north - south line and an east - west line to fully sample the spatial changes .
this allows one to establish the position - angle of the binary system . for polarimetric reasons expanded below
, the image rotator is not used to keep this component fixed vertically during the observations .
this means that the position - angle rotates from exposure to exposure , but presenting the calculations is still useful to demonstrate the capability of the analysis software . for each spectrum ,
the optimal spectral extraction algorithms fit gaussians and average spatial profiles ( psf s ) to the data at each wavelength , as was shown in figure [ fig : cosray ] .
these spatial profiles can easily be compared with the nominal values for the average spatial profile s fwhm and centroid for each order . at each wavelength , a deviation for the fwhm and the centroid can be calculated .
because of small guiding errors , the centroid of each stellar observation does not coincide with the exact order center .
because of the tilt of the orders on the detector , this will cause the centroid of the spectrum to cross spatial pixels at different spectral pixel locations . in order to correct for this ,
the analysis package does a cross - correlation between the nominal order center location , as illustrated by the jig - saw shaped fit in figure [ fig : coord ] , and the centroid location of the actual observed spectrum .
once this small offset has been corrected , the centroid location and the fwhm across the order is calculated .
some sample calculations from data taken the 20@xmath50 of september , 2007 will be outlined below . in baines
2006 , the known binary mwc 166 was reported to have a centroid shift of roughly 40mas ew and 15mas nw with corresponding fwhm changes of 40mas and 5mas respectively .
this is tabulated in their paper as a total shift of 50mas at a pa of 287@xmath7 where the known separation of the binary is 0.65@xmath22 at a pa of 298@xmath7 .
figure [ fig : specast - mwc166 ] shows a hivis observation of this star .
the observations had a peak of 1500adu and a continuum of 800 - 1300adu across the order and a fwhm of 11.7 pixels ( 1.5@xmath22 ) .
the deviation detected in fwhm and centroid offset are both around 40mas , both the correct magnitude for this system .
we also present observations of @xmath51 cassiopeia in figure [ fig : specast - gmcas ] at the same scale as mwc 166 to show a non - detection across a strong line .
the peak intensity was near 26000adu with continuum from 4000 - 6000adu and a fwhm of 10.1 pixels ( 1.3@xmath22 ) .
this non - detection shows that spurious signatures caused from the 1% savart plate leak , non - linearity in the chip , spatial dependence of the focus or any other effects are not present at an observable level across an emission line .
the star p - cygni has a very strong h@xmath0 line .
this star was observed with two different exposure times separated by 90 minutes .
the reduction was performed with three separate settings : a width of 47 for tilted and untilted coordinates , as well as a width of 41 un - tilted .
this will be used to highlight changes in derived psf properties when including tilts and neglecting low - flux psf wings .
the altitude changed from 50@xmath7 to 34@xmath7 and the frame rotated by 28@xmath7 between the observations .
the line has a normalized intensity of 0.194 in the absorption trough and 18.180 in the emission peak , or a flux ratio of 94 .
this huge range in flux and the difference in seeing serves to highlight systematics .
one exposure is deliberately clipped ( near saturated ) to illustrate systematic errors that can occur with this technique .
the spatial profiles are shown in figure [ fig : psf - pcyg ] .
the peak intensity of the saturated exposure was 44000adu with continuum of 1800 - 3800adu and only 400adu in the absorptive component of the line .
the calculated fwhm was 9.6 pixels ( 1.25@xmath22 ) .
however , a gaussian fit to the unsaturated region shows that the true peak should have been closer to 48000adu , or a roughly 10% clipping . the unsaturatd exposure taken an hour and a half later and at higher airmass had considerably less flux , 16000adu peak , continuum of 600 - 1100adu , and only 130adu in the absorptive component of the line .
the calculated fwhm was 14.1 pixels ( 1.83@xmath22 ) .
the wavelength dependence of the psf properties shows striking changes across the line , shown in figure [ fig : specast - pcyg ] .
both exposures and all reductions show a similar change in centroid location , sharply changing by about 50 - 150mas in the blue side of the absorptive component , then rapidly dropping across the center of the absorptive component to -20mas to -80mas in the red side of the absorption .
the centroid then gradually rises over the emission to a peak of 5 - 15mas just before the emission line returns to continuum on the red side of the line .
the change is actually greater in magnitude for the unsaturated exposure with any reduction parameters .
there are actually six curves in the top panel of figure [ fig : specast - pcyg ] but the difference between order widths of 41 and 47 is completely negligible .
the tilted coordinates show systematically lower deviations for both saturated and unsaturated exposures .
in essence , more flux gives less deviation presumably from better fits .
the tilted coordinates more accurately reflects the true wavelength dependence of the orders and reduces the systematic error .
the fwhm also shows strong differences between the different exposures and reduction parameters .
the unsaturated exposure with only 140adu peak flux shows a 200mas decrease in fwhm in the absorptive component of the line in both tilted and untilted coordinates with a width of 47 pixels .
when the wings of the psf are neglected by using a width of 41 , the calculated fwhm deviations are significantly smaller , near 75mas , presumably because the wings of the psf have a noise contribution .
this again shows that more flux leads to better fits and that large flux differences across a line lead to systematic effects .
the saturated exposure shows an extra 50mas increase in fwhm across the emission peak in the clipped region of the line for all reductions . [ !
h ] another complication is extended emission .
a change in fwhm and centroid can be caused by significant resolved emission around an object , such as the observations of t ori .
if there is no inspection of the raw data , this emission may be falsely reported as a binary detection .
also - one can not rule out the possibility of unresolved nebulosity as a significant contribution to the psf .
the observations of t ori showed a peak signal of 600adu with a continuum of 100 - 150adu .
the fwhm was calculated as 10.1 pixels ( 1.3@xmath22 ) .
the wavelength dependence of the psf for two reductions , using tilted and untilted coordinates , is shown in figure [ fig : specast - tori ] .
the centroid shift is small , roughly 40mas and is independent of the order tilt .
the fwhm changes by roughly @xmath980mas when using untilted coordinates , but is essentially zero when using tilted coordinates .
this can arise when a vertical order geometry intersects the fully - resolved emission near the edges of the order .
when using properly tilted coordinates , the fully resolved emission becomes an added background that does not influence the gaussian fitting .
figure [ fig : tori ] shows the raw exposure and the extended emission with the tilted coordinates shown as the slanted black line across the extended emission .
further investigation of these effects is beyond the scope of this paper , but the analysis package is robust and subject to the same systematic errors as other analysis packages ( bailey 1998 , baines et al .
2006 ) . in summary
, this analysis package can reproduce results of others , specifically mwc 166 .
however , other systematic effects that always influence spectroscopic data must be considered as well .
this illustrates the robustness of the reduction geometry and coordinate grid as well as the psf fitting and analysis by the optimal spectral extraction routines .
the simultaneous imaging of orthogonal polarization states also allows for a greater efficiency observing sequence and for reduction of atmospheric effects since polarization measurement can be done with a single image and both states are imaged simultaneously , with identical systematics . using the fractional polarizations measured as the difference between the orthogonally polarized spectra divided by the sum , q and u
, the percent polarization , p , and the position angle , pa , can be calculated as follows : @xmath52 @xmath16 @xmath53 it is important to note that many people use the ratio method defined as follows : @xmath54 the ukirt infrared polarimeter has yet another difference method published to calculate the polarimetry on their irpol2 website .
@xmath55 this method is a different version of the simple average method and can be reduced to a simple average where b and c switch places : @xmath56 when all three of these methods are applied to a data set , the difference between the calculated polarized spectra is entirely negligable .
since no method was significantly different than any other , the easiest to implement was selected - a simple average applied to each individual exposure set .
the subtraction of the two terms allows for cancellation of systematic errors resulting from any misalignment of the polarized orders since each term is an independent measurement of stokes q. measurement of stokes u is the same formula , but with the waveplate at 22.5@xmath7 and 67.5@xmath7 .
since this is a fractional quantity , any correction applied equally to each order and to each chip , such as spectrophotometric calibrations , vignetting corrections or skyline subtractions , will not change the polarimetry in any significant way .
that is to say that no standard calibrations effect the spectropolarimetry .
we wish to note that no instrumental polalarization ripple of the kind mentioned in aitken & hough 2001 or seen in isis ( harries et al .
1996a ) has been seen in the data .
this ripple is thought to arise from fabry - perot type fringes in the wave - plate .
though the instrument uses a standard achromatic waveplate , a thin layer of polymer sandwiched between two glass substrates , no ripple near the 0.1% level has been seen .
the savart plate axis can never be perfectly aligned with the slit .
this produces a small shift in the dispersion direction between the two polarized orders .
this shift was calculated by cross - correlating the thar spectra from the two polarized orders .
the derived shifts were 2 pixels in 2004 , 12 pixels in 2005 before the savart plate reinstallation , and 7 pixels after the 2006 remounting .
to characterize any derivative noise , there are a number of ways to check the wavelength alignment of the spectra and calculate the polarization .
the three ways are 1 ) no alignment 2 ) cross - correlation an a simple integer - pixel shift ( which preserves noise properties ) and 3 ) integer - pixel shift followed by the shift - n - scale sub - pixel alignment .
when no alignment is done , the spectra and polarization are simply calculated where the are on the detector . in the integer - pixel - shift alignment method ,
a cross - correlation is done on the spectra to ensure that the spectra aligned to a single pixel , as a double - check on the wavelength calibration .
since any interpolation can effect and corrupt the noise statistics of the spectra , as shown by the shift - n - scale example in figure [ fig : shfscl ] , subpixel alignment must minimize the derivative signal more than it introduces extra noise . to evaluate all three alignments ,
the polarization analysis is done on the three different spectra using the three different alignment methods .
these nine polarization spectra are similar in their overall structure .
a sample of this is shown in figure [ fig : q9ways ] .
the polarization signature is around 1.5% and the differences between the different methods are very small . [ !
h ] the spectral shift ( beam wobble ) measured from th - ar lamp spectra taken at many waveplate orientations shows that there is a one pixel shift induced over a complete rotation of the waveplate with a roughly sinusoidal shape .
the alignment of the waveplate with the savart plate causes a rotational offset between the rotation - stage zero point and the waveplate zero point of 85@xmath7 .
thus the operation of the waveplate is from rotation angles of 85@xmath7 to 152.5@xmath7 near the crest of the sine - curve .
measurements of thar spectra taken each night , shown in figure [ fig : wobble - exp ] , show that the beam wobble is roughly 0.3 pixels between the four waveplate orientations .
the derivative signal caused by a 0.3 pixel shift for strong emission lines with contrast of 5 - 20 is roughly 1 - 3% .
it will be shown later that the derivative signal subtracts away in the dual - beam polarization analysis and that a beam wobble of 0.3 pixels produces a negligible error .
this beam - wobble signal follows regions of curvature and is proportional to the initial derivative error , as it is a derivative - of - derivative error ( acceleration ) .
the sub - pixel shifts between the three different intensity calculation methods is calculated using the shift - n - scale routine .
the shifts were found to be a small fraction of a pixel .
the wavelength of each pixel for each horizontal order is assigned by a 3@xmath31 order polynomial fit to thorium - argon calibration lamp images .
this analysis is done independently on each order for the two orthogonal polarization states . typically , reduction packages , such as libre esprit for the espadons spectrograph on cfht ( donati et al .
1997 ) use the wavelength fits as the alignment method between the polarized orders . the simple average without alignment method
was found to work best because the flat - field cancels in the polarization calculation , leaving the measurement independent of the pixel - to - pixel gain . since hivis uses the alt - az coud focus , the spectropolarimeter s q - u reference frame rotates on the sky as the telescope tracks .
there are two main rotations : one with altitude , between the tertiary mirror and m4 , as the altitude changes , the other with azimuth , between m6 and the coud pickoff mirror .
normally , coud spectrographs use a 3-mirror image rotator somewhere downstream ( in an pupil plane ) to compensate for the telescope s motion and keep the orientation of the slit on the sky fixed .
hivis has an image rotator , but since the rotator is 3 oblique reflections , it is not used to increase the accuracy of the polarimetric calibration of the instrument . changing
the mirror orientations would change the polarization induced by the mirrors .
the orientation of the instrument on the sky is allowed to rotate and exposures are kept short so that the rotation is small .
the analysis software takes the date and time of the exposure and calculates the altitude and azimuth of the telescope during the exposure .
knowing the orientation of the telescope mirrors , the projection of the spectropolarimeter s q and u axes on the sky can be reconstructed , and all measurements rotated to a common frame .
a common convention is to set + q to north on the sky .
the rotation angle between the 4 individual exposures in each set is calculated as position - angle + altitude - azimuth and is checked to quantify the inter - set smear .
very large rotation angles occured when an object is tracked near transit ( high azimuth rates ) . for measurements with small rotations ,
the polarization measurements are assumed to be accurate and are rotated to an absolute frame ( of the first exposure ) on the sky . for every data
set , rotation in an individual exposure causes each exposure to become a mix of stokes q and u , reducing the efficiency of the measurement and adding some uncertainty to the stokes parameters .
imagine an incident + q beam .
a rotation of the instrument of 22.5@xmath7 during an exposure would cause the projected instrument frame to go from + q to + u and the resulting measurement would only show 50% polarization , when 100% was incident .
rotation between sets also induces systematic error .
as the instrument rotates on the sky , different amounts of time will be spent sampling the polarization states . as an extreme example
, 22.5@xmath7 rotation of the instrument between each individual exposure would cause all images to be oriented on the sky with + q whereas the reduction package would assume the sequence was + q , + u , -q , -u .
the systematic error with this high rotation rate causes complete loss of information about u while doubling the time spent measuring q. it also causes the benefits of beam - swaping to be lost entirely and it causes the reduction methods to fail .
since the expected spectropolarimetric signals are subtle differences in the polarized line profile shapes of less than one percent , and the few reported detections show polarization of at most 1% , the lines must be calculated with at least s / n@xmath57 300 in all spectral - resolution - elements .
most young emission - line stars have large line / continuum ratio s ( 5 - 10 ) .
some , such as ab auriga , have pcygni profiles with emission - peak to absorption - trough ratio s of over 20 .
the star p - cygni itself has a ratio of nearly 100 .
even when saturating the peak of the emission line on the detector , a s / n of only 100 or so was achieved in the absorption trough of ab auriga .
a flux - dependent averaging procedure has been developed to bin pixels spectrally by flux , to ensure that the s / n per spectral - pixel is roughly constant over the line at the expense of regular wavelength coverage . [ !
h ] another systematic error occurs when you calculate the degree of polarization for a spectrum , a squared quantity , with noise amplitudes that vary across the line . since the average absolute values of noisy data will be higher than clean data , and there will be significant noise variations across strong emission lines , varying noise can mascarade as a systematic spectropolarimetric error .
figure [ fig : qnoise ] shows a simulated polarization measurement where flux - dependent noise was added to eight copies of a single line - profile .
the noise varied from roughly 5% at continuum to roughly 2% in the center of the emission line .
the resulting stokes q and u parameters were centered about zero , but with a varying noise amplitude .
when these measurements are converted to a degree of polarization via @xmath58 , this varying noise shows up as a systematic error in the degree of polarization .
h ] to rebin - by - flux , the average line profile for each data set ( after alignment ) is calculated .
this provides the average flux and shape of all 8 lines . a user - defined threshold , typically 1 - 5 , sets the minimum flux in each rebinned spectral pixel .
the number of physical pixels to rebin at each wavelength is calculated and each individual spectrum is then rebinned accordingly . a rebinned wavelength array for each set is also calculated since the number of pixels binned in the spectral direction is different for each star and data - set .
the polarization analysis is then performed on these rebinned data sets as specified above and the pixel - to - pixel variation in the noise is greatly attenuated .
figure [ fig : rbqu ] shows a comparision of raw and rebinned polarization spectra .
the rebinned spectra only show a few resolution elements over the absorption trough with many covering the emission peak .
the polarization signature in the absorptive part of the line stands out very clearly .
the calculated polarization is sensitive to systematic errors in the alignment of the polarized spectra .
there are two main alignments of the spectra - between the polarized spectra in each individual image , and between the specra from each individual image .
the reduction code automatically makes derivative - error simulations for every data set to show the form of the systematic errors caused by misalignments .
the misalignment between polarization states causes a derivative signal in each fractional polarization measurement @xmath59 that can be at the few % level .
as long as the shift between polarization states is constant in each image , then the derivative signal is subtracted away when the second polarization measurement , @xmath60 , is subtracted .
the polarization state shift between images is very stable .
there is never more than a 0.05 pixel shift difference between the calculated wavelength solutions of each polarized order as determined by thar line fits .
h ] an example of the polarized flux calculations for p cygni .
the top curve is the polarized flux after all the wavelength alignment systematic errors have been removed .
the polarized flux is essentially zero and this data set shows no polarization signature . the middle curve shows a 1-pixel misalignment to illustrate the magnitude of the alignment systematic errors .
the bottom curve shows the line profile , the normalized stokes i , with the scale on the right.,title="fig:",scaledwidth=35.0% ] there is also a second order effect - beam wobble - that can cause a systematic polarization error .
if the spectra are misaligned between images , from the rotating waveplate or guiding errors in un - tilted coordinate systems , then a derivative - of - derivative effect can show up when the two polarization measurements are differenced .
this effect is simulated in figure [ fig : derivsim ] by applying a orientation - dependent wavelength shift between the exposures that increases with waveplate angle .
this simulates any beam - wobble introduced by misalignment of the waveplate .
the induced error traces the curvature of the emission line , with polarization errors showing up where curvature is greatest .
using thar spectra , this wavelength shift was measured to be @xmath180.1 pixels , depending on waveplate orientation , shown in [ fig : wobble - exp ] .
this small wobble has a very small influence on the polarization , as it takes a misalignment of several pixels to produce a noticable acceleration signal .
the derivative simulations in this code show that the simple derivative error subtracts away completely if the polarized - order - separation is constant ( which it is to 0.05 pixels ) .
the code also shows that it takes a wobble of at least 2 full pixels between each orientation , for a total of a 6-pixels misalignment to produce a 1% systematic error in a stokes parameter for strong lines . using the shift - n - scale routine , as well as the thar measurements ,
this wobble was measured to be around 0.1 pixels and thus present only well below the 0.1% level proportional to the curvature ( derivative of derivative ) . in harrington & kuhn 2007 , several systematic errors common in spectral reduction packages
were explored .
there are two main sources for error when calculating fractional differences : the wavelength alignment and the continuum subtraction of the 8 spectra .
the major source of error when calculating polarization with strong lines is typically the wavelength alignment .
this was thoroughly explored in harrington & kuhn 2007 . however , in the course of the observations some additional continuum subtraction errors were noticed in some extremely strong lines .
this continuum effect will be explored in this section .
the continuum subtraction in many packages , such as libre esprit , is done by fitting a low order polynomial over a spectral order . in the single - line reduction package , a simple linear fit to the continuum on either side of the line is used , just like in the standard iraf package .
each of the 8 polarized spectra that are used to calculate q and u have the continuum normalized independently by dividing each spectrum by the linear continuum fit .
depending on the width of the line and the number of pixels selected , these fits are typically robust to 1% or better .
this continuum normalization eliminates the telescope induced continuum polarization , as well as any intrinsic polarization . also , since this subtraction centers all normalized continuum values to 1 , the polarized flux calculation no longer gives the actual flux of polarized light .
it gives the deviation from continuum which can be a positive or negative value .
figure [ fig : pfx - pcyg ] shows an example of the polarized flux calculated for a single observation on 9 - 20 - 2007 .
a 1-pixel derivative error , the middle curve in the figure , shows the amplitude of the wavelength errors that are subtracted by the reduction package , leaving the essentially featureless polarized flux as the top curve .
the systematic error arises when calculating the degree of polarization , a 1/i normalized quantity , when this polarized flux is not properly centered at zero .
this featureless curve , when normalized by a strongly varying line profile can give rise to a 1/i type of error .
when dividing a featureless line by the p cygni profile , the absorptive trough with it s small numbers , amplifies the continuum error and falsely gives a high degree of polarization while the emission peak reduces the error . in order to illustrate this effect , a simulated continuum error , @xmath61 ,
is implemented as a constant offset to a spectrum with this formula : @xmath62 we find that this error can cause significant polarization effects across strong emission lines when @xmath61 is near 0.01 . in order to be certain that a polarization effect is real , an examination of the polarized flux
must be done to ensure that the continuum fitting has been done properly .
since the widths of the emission lines in our targets varies greatly , there are atmospheric water vapor lines near h@xmath0 , and there are other emission lines in the area , this is best done by visual inspection .
the next step in the instrument development was calibrating the instrumental polarization response
. absolute polarization at all wavelengths requires very careful calibration of the telescope .
internal optics can induce or reduce polarization ( u , q@xmath27i or i@xmath27q , u ) in the beam , cause cross - talk between polarization states ( qu@xmath27v , v@xmath27qu ) , and also rotate the plane of polarization ( q@xmath27u , u@xmath27q ) . for an instrument at the coud focus of an alt - az telescope , all of these effects are functions of wavelength , altitude and azimuth . for aeos , there are five 45@xmath7 reflections before the coud room , two of which change their relative orientation .
this is particularly severe , and only a few put a spectropolarimeter after so many mirrors .
the calibration is difficult , but some have tried this for many different telescope designs ( cf .
snchez almeida et al . 1991 , kuhn et al , 1994 , giro et al . 2003 ) since there are no spectropolarimetric standards one usually observes polarimetric standards that have a precise polarization value averaged over some bandpass such as hsu & breger 1982 , gil - hutton & benavidez 2003 , schmidt & elston 1992 , fossati et al . 2007 .
the calibration and creation of a telescope model is done by measuring unpolarized standard stars and polarized sources at many pointings to calibrate the effects of the moving mirrors .
in general , the telescope will have different responses to linearly polarized light since the absorption coefficients of the optical components vary with wavelength and the differential absorption is a function of the angle of the incident polarization with respect to the mirror / component axes .
zeemax models of the telescope were used to compute the assumed polarization properties of the coud focus given various optical constants for aluminum .
these show the expected altitude - azimuth dependence of the mueller - matrix terms , though the amplitudes of the computed effects are significantly different than measured .
while the models can calculate the full mueller matrix as functions of altitude , azimuth , and wavelength , all of these terms can not be practically measured . however , certain parts of the mueller matrix can be measured given proper sources ( polarized standards , unpolarized standards , and scattered sunlight ) .
unpolarized standard stars are the first tool used to calibrate the telescope .
observations of these standards at as many pointings as possible allows one to construct a model of the telescopes response to unpolarized light and to quantify the telescope - induced polarization ( but not the depolarization , cross - talk , or plane rotation which requires a linearly polarized standard ) .
polarized sources were the last calibration source used .
using a polarized source allows one to measure the depolarization and rotation of the plane of polarization , all of which are , in principle , orientation dependent .
polarized standard stars are polarized in their continuum light , but the constant value quoted in standard references is an average polarization in a relatively line - free , wide bandpass image .
the degree of polarization is usually not more than a few percent making instrument calibration difficult with these sources .
another calibration method which will be explored below , is to use twilight ( scattered sunlight ) as a highly polarized source with a fairly well known degree and orientation of polarization .
the polarization is certainly not constant and varies with atmospheric properties , time , and pointing , but it is a very bright and highly polarized source that has given valuable insight into the polarization properties of the mirrors . [ !
h ] flat field polarization analysis , averaged 200:1 for clarity .
stokes q is shown as diamonds , u is shown as triangles , and the degree of polarization is the top curve , shown as + symbols .
the squares represent the position - angle of polarization ( pa ) with the y - axis scale on the right .
the polarization ( top curve ) begins at 650 nm being dominated by u ( triangles ) , but begins to rise as -q ( diamons ) gets larger .
the pa rotates by about 25@xmath7 , going from nearly 55@xmath7 or mostly + u to being nearly 70@xmath7 or a mix of -q and + u near ( -3,4).,title="fig:",scaledwidth=40.0%,scaledwidth=50.0% ] all of the moving mirrors in the aeos telescope are located upstream of the polarizing optics ( analyzer ) .
as an illustration of the polarization induced by the reflections between the halogen flat field lamp ( @xmath63 ) and the spectropolarimeter , figure [ fig : flatpol ] shows the polarization reduction applied to a set of flat field frames with the 4 waveplate orientations taken in 2004 .
the flat field lamp is located just after the coud room entrance port , immediately after the coud pickoff mirror ( m7 ) on the far right in figure [ fig : coude_layout ] . the lamp is reflected off a diffuser screen and is assumed to be unpolarized .
the flat fields show a significant apparent polarization which changes with wavelength . at 6500 ,
the induced polarization is dominated by the + u term , but the u and -q terms become equal at 90 .
the position angle varies almost linearly across all wavelengths while the degree of polarization rises from 2% at 650 nm to 6% at 950 nm . since the remounting of the optics and a movement of the image - rotator , the values of the flat field polarization have changed , but this illustrates the magnitude and wavelength dependence of polarization induced by the 11 fixed reflections in the optics room .
we also have tested the polarization of the common - fore - optics .
a linear polarizer was mounted ( with a mask ) at various points along the optical path , from the calibration stage just after the coud port to just in front of the slit .
a linear polarizer was measured to have 98.2% polarization at the slit , 98.3% polarization after the k - cell , 97.4% before the k - cell , and 97.7% after the flat field screen .
this shows that the linear polarization does not vary much between the beginning of the spectrograph optics and the slit .
another polarizer mounted at the slit was measured to have 98.4% .
this polarizer was put at the calibration - stage and was measured to have 98.1% 97.7% 97.8% and 98.0% polarization for angles -03.2 , 44.0 , 87.5 , and 134.2 degrees in the slit s qu frame when manually rotated to 0 , 45 , 90 , and 135 degrees .
the measured angles match completely within the manual uncertainty .
this shows that the degree of linear polarization is preserved through the spectrograph optics .
the flat field also shows a difference in beam intensities ( vignetting or beam transmission difference ) , in addition to it s intrinsic polarization .
the polarization in a single frame , computed as @xmath64 shows + q at -4.55% -q at 1.36% + u at -3.63% and -u at 0.45% .
these values are not symmetric about zero , as would be the case if there was only intrinsic polarization . to make the polarization values symmetric about zero , a constant difference of 1.59% between the two beams
is applied .
this was calculated as 1.588 and 1.591 for q and u respectively as the null polarization spectra ( sometimes called the check spectra : + q + -q ) . restated , this means the @xmath9q @xmath9u spectra are symmetric about 1.59% , ie that the rotation of the waveplate swaps polarization around a constant difference in beam intensity of 1.59% .
subtracting the 1.59% difference in beam intensities from all single - frame polarizations @xmath64 makes these polarization spectra symmetric about 0% reflecting the true polarization of the beams .
we note that the old flat field polarization measurements showed a polarization of 2.78% at a pa of 55.0@xmath7 ( -0.93% 2.57% qu ) for 6567 .
this is not what is measured with the new setup .
this is expected since the image rotator was moved , the flat - field lenses were installed , and the new dekkar is slightly smaller than the old one , changing the illumination pattern . the new flat field polarization is just as repeatable and robust as under the old configuration .
.flat field polarizations [ flatpol ] [ cols= " < , < , < , < , < " , ] when the intervening cloud is optically thin , the polarization of the absorbed spectral feature can be larger than any scattered light spectral polarization signal .
the ratio of polarized absorbed light to polarized scattered light will be computed to show in which regimes this absorbed signature is dominant .
table [ tab - pumpvar ] lists the variables .
figure [ fig : pumpgeo ] shows the geometry of a gas cloud near a star of luminosity @xmath65 .
let the cloud and star have radii @xmath66 and @xmath67 and let the cloud be a distance @xmath68 from the star .
for simplicity , set @xmath69 .
the solid angle subtended by the cloud and the observer s detector , as seen from the center of the star , are @xmath70 and @xmath71 .
the polarized scattered light signal is @xmath72 and the total continuum optical signal is @xmath73 .
when the cloud is not projected against the disk of the star , the degree of polarization is : @xmath74 where @xmath75 is the fraction of light incident on the cloud which is scattered ( assumed isotropically ) and @xmath76 is the average intrinsic polarization of this scattered light .
this equation describes the polarization of light scattered by a single cloud in terms of the amount of incident flux multiplied by the degree of polarization and amount of light scattered to the detector all normalized by the total stellar flux received at the detector . when the cloud lies directly between the disk of the star and the observer , optical pumping
can cause the absorbed light to be polarized .
all the light removed from the beam appears as a polarized absorption feature in the spectrum .
since the absorbed light is scattered into all directions the absorption can polarize a much larger fraction of the light .
this absorptive geometry can lead to a larger polarized optical signal .
if @xmath77 then the relative transmitted polarized flux , @xmath78 , normalized to the total , is : @xmath79 here @xmath80 is the absorptive ( optically pumped ) intrinsic polarization which is computed below .
it will shown explicitly how this term is computed in the following sections .
typically this term is a small fraction of a percent and is no more than 1% . the intrinsic scattered and absorbed polarization , @xmath76 and @xmath80 , are due to population differences in , respectively , the upper and lower level magnetic substates , and to the anisotropy of the scattered or pumping radiation .
the ratio of absorbed to scattered polarization is computed as : @xmath81 the relation @xmath67=1 has been used . since the cloud lies above the star , @xmath68 is always larger than @xmath67 . in general , this alone illustrates the regimes where the absorptive effects dominate .
in low optical depth cases , the fraction ( 1-f)/f is quite large .
the further away the cloud is , the larger the ratio ( though in absolute terms both scattered and absorbed fluxes decrease ) . it is shown below that the typical absorbed polarized fraction is less than 0.01 while scattered flux can approach 1 for clouds very close to the photosphere near the limb .
as an example , the ratio for a cloud one stellar radii off the photosphere ( d=2 ) absorbing near the limb will be computed at the maximum possible scattering angle . in this geometry
, the cloud will be at a scattering angle of @xmath82 and the star will subtend @xmath83 . from chapter 7 , if simple rayleigh scattering is assumed , the degree of polarization can be calculated as the ratio q / i from the scattering phase matrix : @xmath84 for simplicity , we ll assume that the absorptive polarized fraction is 0.01 and that the optical depth is 0.1 . in this case
the ratio of absorbed to scattered polarization is high : @xmath85 in this case , the optical - pumping polarization effect will dominate any scattered polarization .
now , this simple example ignores the entire spatial distribution of scattering and absorbing particles , but does provide a compelling reason to explore polarized absorption as an alternative to scattering polarization . for optically thin cases where the clouds are a few stellar radii away from the photosphere , this absorptive effect has just as much effect as the light scattered by the same cloud .
our calculation is a very simplified version the general problem .
this example assumes that the `` cloud '' has a small line - of - sight velocity with respect to the star and have ignored doppler shifts or the spectral dependence of the stellar source . in realistic cases where the absorption is on top of an emission line , it is not stellar continuum radiation , but h@xmath0 emission that optically pumps the intervening gas . in general
this enhances the optical pumping ( since h@xmath0 can be significantly brighter than continuum ) and complicates the geometry because the pumping source is a broad region with significant spectral dependence and doppler shifts induced by cloud motion . however , since this cloud only produces polarization at the specific wavelength where there is absorption , there is less difficulty explaining profiles where emission and continuum polarizations are identical .
it is not the redistribution of polarized light , taking polarization from one wavelength and moving it to another , but the absorption itself which imprints the polarization .
thus the discussion here applies directly to the outflowing wind configuration of , for example , ab aur . in this model , absorption causes polarization and the emission line can remain unpolarized . the absorptive polarization fraction from statistical equilibrium
will be computed below followed by specific solutions . in order to compute the intrinsic absorptive polarization amplitude @xmath86
, the statistical equilibrium equation must be solved for the densities of the first 28 ( ls coupled ) hydrogen energy levels corresponding to all of the principal quantum numbers @xmath87 and @xmath88 sublevels .
the ground state ( @xmath89 ) consists of an @xmath90 doublet .
the @xmath91 levels include two @xmath92 and one @xmath93 states for a total of 8 magnetic substates .
similarly there are 18 @xmath94 sublevels .
a 28@xmath2428 matrix statistical equilibrium equation is constructed from the radiative transition rates .
note that collisional transitions are unimportant in the density regimes of interest here .
the solution for the individual sublevel densities is obtained by a singular value decomposition method .
first , one must obtain theoretical line strengths and einstein @xmath95 and @xmath96 coefficients for all allowed dipole transitions between individual sublevels ( sobelman 1992 ) .
these coefficients describe fixed upper and lower magnetic substates so that stimulated emission and absorption @xmath96 coefficients for each transition are equal .
the calculations have been checked against the fine structure @xmath95 coefficents and line strengths listed by nist ( ralchenko et al .
2007 ) . as before , take the quantization axis to lie along the mean incident pumping radiation direction and take the opening half - coneangle as @xmath97 .
it is the angle - averaged mean radiation intensity that multiplies the einstein @xmath96 coefficients in the rate equation .
next , derive distinct geometrical scaling factors for @xmath98 and @xmath99 transitions that follow from integrating stenflo s ( 1994 ) eqns . 3.72 over the pumping radiation solid angle , @xmath100 .
for a limb - darkened illuminating stellar disk where @xmath101 and @xmath102 , obtain factors @xmath103 that multiply the pumping blackbody background radiation terms in the rate equation . when the theoretical limb - darkening coefficients from al - naimiy ( 1978 ) are included , the coefficients can be written as : @xmath104 @xmath105 anisotropy of the illuminating stellar light source leads to a difference in the @xmath106 and @xmath107 populations within the @xmath108 level .
this lower state anisotropy linearly polarizes the transmitted light from the star which passes through the cloud at a non - zero angle , @xmath51 , with respect to the @xmath109 ( optical pumping ) direction .
given the geometry defined by figure [ fig : pumpgeo ] , one expects the emergent light to be partially polarized with electric field perpendicular to the plane of the figure .
most astronomical observations do not resolve the fine structure of the @xmath110 to @xmath111 h@xmath0 line but they will still exhibit a diluted linear polarization signal because only some of the transitions are insensitive to the pumping radiation anisotropy such as those which start and end on @xmath92 levels .
the stokes transfer equation simplifies in the geometry where the stellar photosphere is a background light source and the intervening cloud is optically thin .
take the positive stokes @xmath112 direction to correspond to light with electric field perpendicular to the plane of figure [ fig : pumpgeo ] .
the stokes vector transfer equation ( eq . 11.1 in stenflo 1994 ) involves only @xmath113 and @xmath112 intensity , and @xmath114 and @xmath115 continuum - normalized opacity terms .
since the stellar photospheric background is the only source of photons observed through the cloud , there are no source or emissivity terms to consider in the polarized transfer equation .
typically , only small polarization ( @xmath116 ) is found in the stellar systems . in this approximation
, the transfer equation reveals that light transmitted through a diameter of the cloud will have polarization : @xmath117 where @xmath118 is the normalizing continuum opacity . here
@xmath113 is the continuum intensity and one can compute @xmath119 from the lower level population anisotropies and eqns
. 3.74 ( stenflo 1994 ) . in this case
the effective refractive index and opacity terms @xmath120 or h@xmath121 ( eqn 4.41 ) are computed from sums of the products of the @xmath96 coefficients and the sublevel populations , derived separately for the @xmath122 and @xmath98 transitions . the optical depth through the cloud
is just @xmath123 where @xmath75 is the fraction of h@xmath0 photons scattered , as defined above . for clouds with typical path lengths of roughly @xmath124 ,
the optically thin condition yields hydrogen densities that are effectively collisionless . in the low density conditions of this disk /
wind the statistical equilibrium rate equation depends on only stimulated emission , absorption , and spontaneous emission processes .
it is the difference in the angular dependence of the @xmath125 versus @xmath126 coupling to the anisotropic radiation field that leads to differences in the degenerate @xmath110 magnetic sublevel populations .
several qualitative features of the optical pumping solutions are intuitive . for intense pumping radiation fields
the anisotropy yields a decreasing absorptive polarization in this regime the spontaneous emission is negligible and the net upward and downward transition rates between any two levels must be the same . in this case
the equilibrium level populations are independent of the upward / downward transition rates .
consequently there can be no sublevel density differences and the absorptive polarization must vanish . similarly for weak radiation fields it is the spontaneous emission that defines transition rates and this term in the rate equation does not depend on the radiation anisotropy .
thus at low and high radiation backgrounds the absorptive polarization approaches zero .
the absorptive polarization near h@xmath0 is also a function of the ratio of the h@xmath0 to @xmath127 ( and @xmath128 ) pumping flux .
the ratio of the @xmath129 to @xmath92 populations affects the net h@xmath0 polarization , since only the @xmath93 states are sensitive to radiation anisotropy . in the limit of no background h@xmath0 flux ( and in the absence of collisions ) all the hydrogen ends up in the `` sterile '' @xmath130 state because it has no decay route and there is no radiative absorption to depopulate it .
conversely , as the pumping h@xmath0 flux increases the equilibrium @xmath131 levels are populated and the induced linear polarization can increase .
this results in a rather weak temperature dependence of the optical pumping source on the absorptive polarization . even though a decreasing source temperature decreases the pumping radiation , this effect on the polarization is compensated by the increasing ratio of h@xmath0 to @xmath127 flux . for a photospheric temperature between 30,000 to 3,000k , a change in pumping flux by nearly 4 orders of magnitude
, the absorptive polarization is nearly constant .
for example , a cloud located 1.1 stellar radii from a 30,000k star ( @xmath132 ) that is @xmath133 from the line - of - sight to the center of the star , yields linear polarization of 0.53% .
if the star has a temperature of 3000k the polarization only decreases to 0.48% . for fixed radiation mean intensity , as the pumping half - cone angle ( @xmath97 ) increases the anisotropy decreases and the polarization decreases .
this effect continues until @xmath134 , at which point there is no difference between the dipole coupling of radiation to @xmath98 and @xmath99 transitions .
the actual level population difference scales with @xmath97 as the difference between the derived geometric radiation coupling coefficients @xmath135 and @xmath136 defined above .
it is straightforward to see that this varies like @xmath137 at small angles .
thus the linear polarization @xmath138 for small cone angles ( even though the anisotropy increases ) .
this is because the pumping radiation intensity declines rapidly with @xmath139 ( where @xmath68 is in units of the stellar radius ) as the cloud - star distance increases .
figure [ fig : pumpdist ] shows how the linear polarization varies for a cloud seen at the projected edge of a star , as a function of the cloud - star distance , @xmath68 . at small distances from the star
the limb darkening function yields significant anisotropy that keeps the polarization from going to zero .
the absorptive polarization depends strongly on the angle between the absorbed beam and the mean pumping direction . as this angle ( @xmath51 ) increases the polarization perpendicular to the plane increases .
for example , with a @xmath140 source temperature and a @xmath141 cone angle the polarization increases almost linearly from @xmath142 to @xmath143 as @xmath51 varies from @xmath142 to @xmath141 . with this new theory ,
detailed forward solutions for a wide range of circumstellar conditions can be developed . here , the general form of the wavelength dependent linear polarization expected from intervening wind or disk structures will be computed .
consider two geometries ; a radial outflow and orbiting material illuminated ( optically pumped ) by the stellar blackbody . by calculating polarization for clouds at a fixed distance from the photosphere ,
the polarization imprinted by a ring of material , much like the examples in chapter 7 can be traced .
the geometry is essentially that defined in figure [ fig : pumpgeo ] but allow that the wind or disk s symmetry axis is inclined at an angle @xmath144 with respect to the line - of - sight to the star .
choose the common plane containing the line - of - sight direction and the symmetry axis to be oriented vertically .
then , as seen from the observer , the intervening clouds , a ring of material at a fixed distance , projects onto a horizontally oriented arc across the stellar disk .
the ring of material is both illuminated by the star and absorbs the stellar light .
each point on this arc contributes absorptive polarization at a distinct velocity with respect to the observer .
we take positive stokes @xmath112 along the vertical sky direction .
positive @xmath145 corresponds to an electric field that is @xmath146 to this .
it follows that the absorptive polarization of a symmetric radial wind should yield only @xmath147 polarization at all velocities .
every cloud producing non - vertical polarization on one side of the photosphere will have another cloud on the other side with mirrored geometry , canceling any non - vertical polarization .
figure [ fig : pumpdist ] shows the degree of polarization calculated for a cloud projected on the limb , the geometry of maximum polarization , assuming star with a simple black - body spectrum with a temperature of 10,000k .
this degree of polarization is fixed by the quantum - mechanical properties of hydrogen gas , the intensity of the radiation and the geometry .
the velocity of the cloud , in contrast to the scattering models , only influences the wavelength of the absorption and does not produce other doppler effects .
the optical - pumping effect in disky systems has symmetric and anti - symmetric components as seen in figure [ fig : pumpcalc ] .
in contrast to scattering theory , even a homogeneous rotating disk geometry must exhibit non - zero @xmath148 and @xmath149 line profiles because of the angular difference between the line of sight and radial directions in every point along the orbit .
the zero line - of - sight velocity condition for orbiting material occurs along the vertical symmetry line in the center of the h@xmath0 line . for any non - zero angle between source and line - of - sight
, there is polarization . the characteristic symmetric stokes @xmath148 and antisymmetric stokes
@xmath149 profiles for such a disk , plotted as stars and diamonds in figure [ fig : pumpcalc ] , show the degree of polarization rising towards larger velocities .
these are plotted against units of the maximum rotation or outflow velocity of the disk or wind .
the polarization is maximum when the material projects against the limb ( largest asymmetry ) on both the approaching and receding sides of the disk .
the absorptive polarization outside of these plotted points must be zero , but spectral smoothing due to the finite spectral resolution of real measurements will yield some polarization outside of the absorptive region .
the exact spectropolarimetric morphology caused by this inclined ring of material would be a double - peaked stokes @xmath148 and anti - symmetric @xmath149 . in qu - space
, this would correspond to two separate excursions , one toward ( + q,+u ) and then another toward ( + q ,- u ) .
however , the extensions would not be orthogonal because of the different amplitudes of the @xmath148 and @xmath149 signatures .
the extensions would also not be exactly linear because of the differing curvature of the signatures . in figure
[ fig : pumpcalc ] we ve plotted the stokes @xmath148 polarization across the velocity distribution of an outflow .
the ring of material is assumed to have a spatially constant radial velocity and the velocity distribution has been normalized by the maximum projected outflow velocity .
the outflow has it s maximum projected velocity ( blue - shift ) closest the line of sight .
this occurs at the left - most portion of the curve , with zero polarization and a relative velocity of -1 .
the cloud occulting the limb is at the minimum relative velocity , in this case at a velocity near -0.5 and a polarization of roughly 0.35% .
stokes @xmath149 must be zero , as can be seen in the anti - symmetry of stokes @xmath149 for the disk calculation .
clouds that had the same magnitude velocity with opposite signs in the disk case will now have the same projected velocity in the wind case .
this would leave only a stokes @xmath148 signature that would correspond to a linear extension in qu - space .
though neither of these calculations accounts for extended h@xmath0 source regions , a non - black - body stellar source , or other complications , the illustration shows that the main optical - pumping effects are easily interpreted .
the polarization is less than 1% of the continuum flux , polarization is only present where there is absorption , and the polarization is greatest where the absorbing material projects with maximum angular deviation between pumping source and line - of - sight directions . though this simple calculation is only illustrative ,
the essential characteristics of optical pumping in a wind - type geometry can be inferred .
the polarization will be highest for the parts of the h@xmath0 line occulting the source with greatest asymmetries . in stars with with extended h@xmath0 emission ,
the maximum polarization occurs at the wavelengths of highest angular difference between the incident radiation and line - of - sight .
the degree of polarization is fixed by the quantum - mechanics - this effect will not produce extremely large polarizations and has a characteristic degree of polarization around 1% continuum or less . although most real stellar disks and outflows are not expected to be homogeneous in density or velocity , in general a non - zero @xmath149 polarization is expected ( and observed ) which , of course , depends on the rotation angle between the instrument @xmath112-axis and the star as well as any geometrical complexities .
the zero polarization level is also ill - defined in the observations since a wavelength independent constant has been removed from all observations and interstellar , telescope , and intrinsic polarization are difficult to disentangle with the hivis observations .
nevertheless the polarization amplitude and simple form of many herbig ae / be detections is consistent with the diamond " symbols in figure [ fig : pumpcalc ] .
the polarization is only in the blue - shifted , absorbed component of the h@xmath0 line , and there is a characteristic amplitude of less than 1% continuum .
this new model shows how optical pumping can describe spectropolarimetric effects with the magnitude and form of the absorptive linear polarization features in some herbig ae / be systems .
the variation of stokes @xmath148 and @xmath149 near absorptive components of the h@xmath0 line originates in this model from the intervening gas cloud projected against the disk of the star and the emission region .
the polarization features contain information on the kinematics and inclination of the cloud .
one immediate conclusion from these calculations is that the absorptive region is near the surface of the star , most likely within 2 stellar radii in order to have a polarization amplitude of about 0.5% .
the main properties of this model is that polarization is present only where there is absorption .
the polarization is maximal where there is greatest asymmetry between the line - of - sight and the source direction .
the quantum - mechanical properties of hydrogen put the amplitude of the polarization effects at typically a fraction of a percent ( continuum ) . for constrained geometries
it may be possible to devise an inversion technique to extract more specific information on non - homogeneous cloud structures in the innermost regions of these stellar systems .
no attempt has been made to include extended h@xmath0 emission regions or other circumstellar complications , but the main framework for polarization - in - absorption is complete .
this thesis presented many components of stellar spectropolarimetry in a very thorough , comprehensive and complete study .
a new high - resolution spectropolarimeter was built and thoroughly calibrated .
custom processing and analysis packages were written and implemented .
a massive amount of observations were performed on many herbig ae / be , be , and emission - line targets to compile a spectropolarimetric survey far larger in scope than any performed to date .
the current spectropolarimetric theories were outlined and compared with the herbig ae / be and be / emission - line survey results .
weakness of current theory to explain the herbig ae / be observations were explained and an entirely new theory was outlined .
the new theory , based on optical pumping , has the potential to explain the polarization - in - absorption seen in the herbig ae / be systems . in chapters 2 through 4 , a discussion of the instrument , reduction and calibration was presented . with signatures of less than 0.1% amplitude , there is a great need for precise and repeatable measurement and analysis . in order to accomplish this ,
many methods of testing and verifying were used .
a dedicated reduction script was developed for this instrument .
many methods of calculating and correcting the polarized spectra were explored with most of the methods giving identical results .
this software package was cross - checked against another package , libre - esprit , in use on the espadons spectropolarimeter giving similar results .
polarization calibration optics were installed on the spectrograph and response tests were performed .
the spectrograph optics were shown to be quite stable .
polarized flat field measurements showed that the spectropolarimeter could measure linear polarization very accurately over the full useful wavelength range of the polarizer .
the spectrograph induces a small amount of polarization in unpolarized light and this polarization is wavelength - dependent .
this is automatically corrected by the reduction software .
the polarization properties of the entire optical system was explored using unpolarized standard stars and highly polarized twilight .
the polarization properties of the telescope are quite severe and complicated .
the effect of on detected spectropolarimetric signatures can be quite significant .
however , the telescope can not induce any effects across a single spectral line . in effect
, the telescope can reduce the overall magnitude of a spectropolarimetric effect , or rotate the plane of polarization .
the telescope can not create a spectropolarimetric signature or modify the morphology of the signature .
thus , any spectropolarimetric signature detected is a useful lower limit on the magnitude of the polarization change across a line . in order to verify and validate the instrument and the results of the survey , a large cross - examination campaign was performed and presented in chapter 5 .
nearly simultaneous observations of a few stars were performed with hivis and the espadons spectropolarimeters .
the results agreed quite well and illustrated the telescope polarization properties well .
archival espadons observations were also used to validate the hivis measurements .
literature results were compiled from three other lower resolution instruments : isis , hpol and wws .
the results again agreed quite well for many stars observed with hivis .
the observing campaign was outlined in chapter 6 .
the spectroscopic observations and properties of the herbig ae / be stars were discussed .
the h@xmath0 line of these stars is typically strong and quite variable .
the emission lines can be a few to thirty times continuum with many different absorptive components on top of the emission .
spectroscopic measurements for these stars over many time - scales were presented and discussed .
a few short - term variability studies were reported illustrating the dynamic circumstellar environment in these stars .
in addition to the hivis observations , many previous studies from the literature were discussed in a case - by - case basis to show stellar properties such as disk , wind , or jet phenomena or specific constraints on circumstellar material such as polar down flows or small - volume , high - temperature components in the circumstellar environment .
chapter 7 outlined the current theory of spectropolarimetric line profiles , based on disk - scattering effects .
the current theory predicts symmetric , broad , double - peaked spectropolarimetric signatures in the orbiting thin - disk case . in the case of outward radial motion ,
the disk - scattering theory predicts a shifting of the spectropolarimetric profiles toward the red side of the spectral line .
other spectropolarimetric effects , such as the depolarization effect and zeeman splitting were discussed .
the depolarization theory is essentially a dilution of continuum polarization by unpolarized emission .
this effect produces broad signatures that are roughly inversely proportional to the amount of emission .
magnetic fields produce a more complex spectropolarimetric signature and other studies have shown that the field strengths of these stars is far too small to produce any measurable spectropolarimetric effect . in chapter 8 , the spectropolarimetric survey of the herbig
ae / be stars was presented .
the survey showed that roughly two thirds of the windy - type stars showed polarization effects in the blue - shifted absorptive components of the h@xmath0 line .
half of the disky - type stars also showed this polarization - in - absorption effect in the line - center absorptive component .
the magnitude of the effect was typically 0.1% to 2% with several stars showing complex morphologies and complex geometries in qu - space .
only one star showed a spectropolarimetric effect across the entire h@xmath0 line .
these observations are difficult to explain using current scattering theories .
the depolarization effect causes a broad , monotonic change in polarization that is roughly inversely proportional to emission and is a linear extension in qu - space .
the disk - scattering effect shows broad , symmetric morphology for orbiting material and a shift toward red wavelengths for out - flowing material . both of these effects do not fit the observed morphology of polarization - in - absorption on the blue - shifted side of the windy - stars or the polarization - in - central - absorption - only for the disky - stars .
the amplitude of the observed effects is also a consideration .
scattering theory has no natural amplitude , but most of the detections fall in the 0.3% to 1.5% range . in order to solidify the case for exploring new alternative theories , a smaller survey of 30
be and emission - line stars was performed and presented in chapter 9 .
these be and emission - line stars had a very different type of spectropolarimetric morphology . in 10 of 30 stars there was a broad , monotonic spectropolarimetric signature present across the entire h@xmath0 line .
this type of signature is typical of the depolarization effect .
however , of these 10 broad detections , many showed more complex morphologies than a simple depolarization .
in another 5 of 30 stars there were antisymmetric or other more complex spectropolarimetric signatures .
this smaller survey is in good agreement with the collection of previous work done to date on these stars .
the broad morphology is reproduced , but the higher resolution , higher signal - to - noise and larger sample of this survey shows that there are still some uncertainties about the exact form of the spectropolarimetric effects . the inability of the scattering theories to fit the polarization - in - absorption " seen in the herbig ae / be stars provided the motivation for developing a new theory for spectropolarimetric line profiles .
this new theory is based on optical pumping - a method for producing polarization simply from absorption .
this new theory is outlined in chapter 10 .
an anisotropic radiation source can induce an imbalance in the sub - states of an atomic transition .
for instance , the h@xmath0 line is an n=3 to n=2 energy - level transition with many allowed sub - states in both levels .
if the hydrogen gas is optically pumped by the stellar radiation , this non - isotropic gas will have a polarization - dependent absorption . in the circumstellar environment ,
radiative transitions dominate collisions and this effect can be much larger than scattering effects .
when this theory is implemented in a simple thin - disk case , just as for scattering theory , there is only polarization where there is absorption and the spectropolarimetric effect is greatest at the wavelengths corresponding to the greatest angle between the line of sight and the absorbing material . although detailed models to predict spectropolarimetric line profiles in the complicated environments have not yet been explored , this new theory has a very clear link between spectrpolarimetric effects and absorption .
this new model shows promise in explaining the large collection of herbig ae / be observations .
this project involved many late nights on lonely mountain tops .
i have to give much thanks to those that helped me keep my sanity and stay focused .
i dedicate this to those people .
i m glad i was supported by so many and i was nt discouraged by those who think building instruments is simply engineering .
don mickey was very helpful and insightful .
he participated in many useful discussions about polarimetry and optical design and offered much insight and encouragement .
jeff kuhn was a great advisor .
though he should be hopelessly buried by work for his hundred other projects , somehow he managed to always find time and energy for this .
i do nt know how he does it .
this program was partially supported by the nsf ast-0123390 grant , the university of hawaii and the airforce research labs ( afrl ) .
some of this research used the facilities of the canadian astronomy data centre operated by the national research council of canada with the support of the canadian space agency .
the archive was very helpful by providing the archival espadons data .
this program also made use of observations obtained at the canada - france - hawaii telescope ( cfht ) which is operated by the national research council of canada , the institut national des sciences de lunivers of the centre national de la recherche scientifique of france , and the university of hawaii .
these observations were reduced with the dedicated software package libre - esprit made available by j. -f .
the simbad data base operated by cds , strasbourg , france was very useful for compiling stellar properties .
agol e. , et al . , 2004 ,
apj , 600 , 781 aitken d.k . &
hough j.h . , 2001 , pasp , 113 , 1300 al - naimiy , h. m. 1978 , ap&ss , 53 , 181a ashok n. m. , et .
al . , 1999 ,
7130 bailey j. , 1998 , mnras , 301 , 161 baines d. et al .
, 2006 , mnras , 367 , 737 bastien p. et al . , 1988 ,
aj , 95 , 900 beckers j. , 1982 , opt .
acta , 29 , 361 beskrovnaya n.g .
et al . , 1994 ,
a&a , 287 , 564 beskrovnaya n.g .
et al . , 1995 , a&a , 298 , 585 beskrovnaya n.g .
et al . , 1998 ,
a&as , 127 , 243 beskrovnaya n.g .
et al . , 1999 , a&a , 343 , 163 beskrovnaya n.g .
& pogodin m.a . , 2004 , a&a , 414 , 955 bjorkman k.s .
et al . , 1998 , apj , 509 , 904 bohm t. & catala c. , 1993 , a&as , 101 , 629 bohm t. & catala c. , 1995 , a&a , 301 , 155 bohm t. et al . , 1996 ,
, 120 , 431 bouret j .- c . ,
catala c. , & simon t. , 1997 , a&a , 328 , 606 bouret j .- c . & catala c. , 1998 , a&a , 340 , 163 bouret j .- c . & catala c. , 2000 , a&a , 359 , 1011 bouwman j. et al . , 2000 ,
a&a , 360 , 213 brannigan e. et al . , 2006 , mnras , 367 , 315 brown j.c .
& mclean i.s . , 1977 ,
a&a , 57 , 141 brown j.c .
et al . , 1978 ,
a&a , 68 , 415 brown j.c .
et al . , 1989 , a&a , 68 , 415 carmona a. et al . , 2007 ,
apj , 344 , 341 cassinelli j.p .
et al . , 1987 , apj , 317 , 290 catala c. et al . , 1986 , apj , 308 , 791 catala c. & kunasz p.b . ,
1987 , a&a , 174 , 158 catala c. et al . 1989 , a&a , 221 , 273 catala c. et al , 1993 , a&a , 275 , 245 catala c. et al .
, 1999 , a&a , 345 , 884 catala c. et al . , 2007 , a&a , 462 , 293 chavero c. et al
. , 2006 , a&a , 452 , 921 corcoran m. & ray t.p . , 1998 ,
a&a , 331 , 147 corder s. et al . , 2005 ,
apj , 622 , l133 coulson k.l . ,
a. deepak publishing , 1988 , hampton virginia cronin t.w .
et al . , 2005 ,
spie , 5888 , 204 cushing m.c .
et al . , 2004 ,
pasp , 116 , 362 davies b. et al . , 2007 ,
apj , 542 , l115 devine d. et al .
, 2000 , apj , 542 , l115 domiciano de souza a. et al . , 2003 , a&a , 407 , l47 donati j.f .
et al . , 1997 ,
mnras , 291 , 658 donati j.f .
et al . , 1999 ,
a&as , 134 , 149 drew j.e .
et al . , 1998 ,
mnras , 296 , l6 eisner j.a .
et al . , 2003 , apj , 588 , 360 eisner j.a .
et al . , 2004 ,
apj , 613 , 1049 elia d. et al . , 2004 , apj , 601 , 1000 eversberg t. et al . , 1998 , pasp , 110 , 1356 fernndez m. et al . , 1995 , a&as , 114 , 439 fossati l. et al . , 2007 , asp conf .
364 , 503 fu h .-
h et al . , 1997 , aj , 114 , 1623 fukagawa m. et al . , 2004 ,
apj , 605 , l53 gl j. et al . , 2001 ,
lond . , 457 , 1385 galazutdinov g.a .
et al . , 2003 ,
a&a , 407 , 705 garcia lopez r. et al . , 2006 , a&a , 459 , 837 gil - hutton r. & benavidez p. , 2003 ,
mnras , 345 , 97 giro e. et al .
, 2003 , proc .
spie , 4843 , 456 goodrich r.w .
et al . , 1995 ,
pasp , 107 , 179 goodrich r.w .
& cohen m.h .
, 2003 , proc .
spie , 4843 , 146 grady c.a .
et al . , 1993 ,
apj , 415 , l39 grady c.a .
et al . , 2000 ,
apj , 544 , 895 grady c.a .
et al . , 2005 ,
apj , 630 , 958 grady c.a .
et al . , 2007 ,
apj , 665 , 1391 guimaraes m.m .
et al . , 2006 ,
a&a , 457 , 581 hamaguchi k. et al . , 2005 , apj , 618 , 360 hanle , w. 1924 , z. phys . , 30 , 93 happer , w. 1972 , rev . mod .
44 , 169 harker d.e .
et al . , 2005 , apj , 622 , 430 harries t.j .
& howarth i.d . , 1996a , a&a , 310 , 533 harries t.j .
& howarth i.d .
, 1996b , a&as , 119 , 61 harries t.j .
& howarth i.d . , 1997 ,
a&as , 121 , 15 harries t.j . , 2000 ,
mnras , 315 , 722 harries t.j . & howarth i.d . , 2000 ,
a&a , 361 , 139 harries t.j .
et al . , 2002 ,
mnras , 337 , 341 - 355 harrington d.m .
et al . , 2006 , pasp , 118 , 845 harrington d.m .
et al . , 2007 ,
icarus , 187 , 177 harrington d.m . &
2007 , apjl , 667 , l89 harrington d.m . &
kuhn j.r . , 2008 , pasp , 120 , 89 hayes d.p . , 1984 , aj , 89 , 1219 henney w.j .
, raga a.c . &
, 1994 , apj , 427 , 305 henney w.j . & axon d.j . , 1995 ,
apj , 454 , 233 herbst w. & shevchenko v.s .
, 1999 , aj , 118 , 1043 hernndez j. et al . , 2004 , aj , 127 , 1682 hillenbrand l.a .
et al . , 1992 ,
apj , 397 , 613 hillier d.j . , 1990 , a&a , 231 , 116 hillier d.j . , 1991 ,
a&a , 247 , 455 hillier d.j . , 1994 ,
a&a , 289 , 492 hillier d.j .
, 1996 , a&a , 308 , 521 horne k. , 1986 , pasp , 98 , 609 horvth g. et al . , 2002 , applied optics , 41 , 543 hsu j. & breger m. , 1982 , apj , 262 , 732 hubrig s. et al . , 2004 , a&a , 428 , l1 hubrig s. et al . , 2006 ,
a&a , 446 , 1089 hubrig s. et al . , 2006b , mnras , 371 , 1953 hubrig s. et al . , 2007 , a&a , 463 , 1039 ignace r. , 2000 , a&a , 363 , 1106 ignace r. et al . , 2004 ,
apj , 609 , 1018 ikeda y. et al . , 2003 ,
spie , 4843 , 437 - 447 ikeda y. et al .
, 2004 , apj , 604:357 - 361 ismailov n.z . &
aliyeva a.a . , 2005 ,
info . bulletin . on variable stars , 5634 jaschek c. & andrillat t. , 1998 , a&as , 128 , 475 kawabata k.s .
et al . , 1999 , pasp , 111 , 898 kozlova o.v .
et al . , 2003 , astrofizika , 46 , 331 kozlova o.v .
, 2004 , astrofizika , 47 , 287 kozlova o.v . , 2006 , astrofizika , 49 , 81 kozlova o.v . , 2007 , astrofizika , 50 , 467 kuhn j.r . et al . , 1994 , solar physics , 153 , 143 kuhn j.r .
, 2007 , apjl , 668 , l63 kurosawa r. et al . , 2006 , 370 , 580 lee r. , 1998 , applied optics , 37 , 1465 liu y. & voss k. , 1997 , applied optics , 36 , 8753 liu w.m .
et al . , 2007 , apj , 658 , 1164 maheswar g. et al . , 2002 ,
a&a , 387 , 1003 mannings v. & sargent a. i. , 1997 , apj , 490 , 792 mannings v. et al . , 1997 , nature , 388 , 555 manoj p.
et al . , 2006 ,
apj , 653 , 657 manset n. & donati j .- f . , 2003 , proc .
spie , 4843 , 425 marconi m. et al .
, 2000 a&a , 355 , l35 marconi m. & palla f. , 2004 , iau symp .
224 , 69 marsh t.r . , 1989 ,
pasp , 101 , 1032 marsh k.a .
et al . , 1995 , apj , 451,777 mclean i.s .
& brown j.c . , 1978 , a&a , 69 , 291 mclean i.s . &
clarke d. , 1979 , mnras , 186 , 245 mclean i.s . , 1979 , mnras , 186 , 265 meech k.j .
et al . , 2005 , science , 310 , 265 miroshnichenko a.s .
et al . , 1998 , pasp , 110 , 883 miroshnichenko a.s .
et al . , 1999 ,
mnras , 302 , 612 miroshnichenko a.s .
et al . , 2004 ,
a&a , 427 , 937 mora a. et al . , 2001 , a&a , 446 , 1089 mottram j.c .
et al . , 2007 , mnras , 377 , 1363 oudmaijer r.d . , & drew j.e .
, 1999 , mnras , 305 , 166 oudmaijer r.d . , et al .
, 2001 , a&a , 379 , 564 oudmaijer r.d .
et al . , 2005 , mnras , 364 , 725 palik e.d . , 1991 , boston : academic .
patat f. & romaniello m. , 2006 , pasp , 118 , 146 patel m. et al .
, 2006 , mnras , 373 , 1641 pelupessy i. et al . , 2000 ,
a&a , 359 , 695 pirzkal n. et al . , 1997 , apj , 481 , 392 poeckert r. & morlborough l.m . , 1976 ,
apj , 206 , 182 poeckert r. & morlborough l.m . , 1977 , apj , 218 , 220 poeckert r. & morlborough l.m . , 1978 , apj , 220 , 940 poeckert r. & morlborough l.m . , 1979 , apj , 233 , 259 pogodin m.a . , 1994 ,
a&a , 282 , 141 pogodin m.a . , 1997 , a&a , 317 , 185 pogodin m.a . et al . , 2000 ,
a&a , 359 , 299 pogodin m.a .
et al . , 2004 ,
a&a , 417 , 715 pogodin m.a .
et al . , 2005 , 438 , 239 pomozi i. et al . , 2001 ,
j. experimental biology , 204 , 2933 pontefract m. et al . , 2000 ,
mnras , 319 , l19 porter j.m . &
rivinius r. , 2003 , pasp , 115 , 1153 porter j.m .
et al . , 2004 , a&a , 428 , 327 praderie f. et al . , 1986 ,
apj , 303 , 311 quirrenbach a. et al . , 1993 ,
apj , 416 , l25 quirrenbach a. et al . , 1994 , a&a , 283 , l13 quirrenbach a. et al . , 1997 ,
apj , 479 , 477 ralchenko , y. et al .
2007 , nist spectra database [ http://physics.nist.gov/asd3 ] reipurth b. et al . , 1996 , a&as , 120 , 229 rodgers b.m . , 2001 ,
washington phd thesis rodrguez l.f .
et al . , 2007 , rmx aa , 43 , 149 snchez almeida j. et al . , 1991 , solar physics , 134 , 1 schmidt g.d .
& elston r. , 1992 , aj , 104 , 1563 semel m. et al .
, 1993 , a&a , 278 , 231 - 237 sitko m.l . , 1981 , apj , 247 , 1024 skinner s.l .
et al . , 1993 ,
apjs 87 , 217 skinner s.l .
et al . , 2004 ,
apj , 614 , 221 strafella f. et al . , 1998 , apj , 505 , 299 suhai b. & horvth g. , 2004 , josa , 21 , 1669 swartz d.a .
et al . , 2005 , apj , 628 , 811 sobelman , i.i . 1992 , atomic spectra and radiative transitions ( berlin : springer )
stenflo , j.o .
1994 , solar magnetic fields ( dordrecht : kluwer ) stenflo , j.o .
, 1997 , a&a , 322 , 985 stenflo , j.o .
1998 , a&a , 338 , 301 tambovtseva l.v .
et al . , 1999 ,
astrophysics , 42 , 54 th p.s .
et al . , 1994 , a&as , 104 , 315 thornton r.j .
2002 , univ .
hawaii phd thesis thornton r.j . , et al . , 2003 , proc .
spie , 4841 , 1115 trujillo bueno , j. & landi deglinnocenti , e. 1997 apj 482 , l183 van den ancker m.e .
et al . , 1998 ,
a&a , 330 , 145 vieira s.l.a .
et al . , 1999 ,
a&a , 345 , 559 vieira s.l.a .
et al . , 2003 ,
aj , 126 , 2971 vink j.s .
et al . , 2002 ,
mnras , 337 , 356 vink , j.s . , 2005a , a&a , 430 , 213 vink j.s .
et al . , 2005b , mnras , 359 , 1049 wade g.a .
et al . , 2007 , mnras , 376 , 1145 wang l. et al .
, 2003 , apj , 592 , 457 warwick , j.w . & hyder , c. 1965 , ap .
j. , 141 , 1362 wassel e.j .
et al . , 2006 ,
apj , 650 , 985 waters l.b.f.m . & waelkens c. , 1998 , ann .
a&a , 36 , 233 whittet d.c.b .
et al . , 1992 ,
apj , 386 , 562 wisniewski j.p .
et al . , 2003 , apj , 588 , 486 wisniewski j.p .
et al . , 2007 , apjl , 656 , l21 wood k. et al . , 1993 ,
a&a , 271 , 492 wood k. & brown j.c . , 1994 ,
a&a , 291 , 202 wood k. et al . , 1997 , apj , 477 , 926 wolf m.j .
et al . , 1996 ,
aj , 111 , 856 yan h. & lazarian a. , 2006 , apj , 653 , 1292 yan h. & lazarian a. , 2007 , apj , 657 , 618 yan h. & lazarian a. , 2008 , apj , in press | the near - star environment around young stars is very dynamic with winds , disks , and outflows .
these processes are involved in star and planet formation , and influence the formation and habitability of planets around host stars . even for the closest young stars
, this will not be imaged by the next generation of telescopes .
other proxies must be developed to probe the circumstellar environment .
the polarization of light across individual spectral lines is such a proxy that contains information about the circumstellar material on small spatial scales .
many models have been created to relate the circumstellar environment to observable polarization changes across spectral lines . however , measuring signals at the 0.1% level requires a very careful control of systematic effects .
we have recently built a high - resolution spectropolarimeter for the hivis spectrograph on the 3.67 m aeos telescope to address these issues .
we have obtained a large number of high precision spectropolarimetric observations of herbig ae / be , classical be and other emission - line stars collected on 117 nights of observations .
many stars showed systematic variations in the linear polarization amplitude and direction as a function of time and wavelength in the h@xmath0 line .
the detected linear polarization varies from our typical detection threshold near @xmath1 up to @xmath2 .
surprisingly , in several stars this polarization effect is _ not _ coincident with the h@xmath0 emission peak but is detected only in the absorptive part of the line profile and varies with the absorption .
these detections are largely inconsistent with the traditional scattering models and inspired a new explanation of their polarization .
we developed a new spectropolarimetric model and argue that polarization in absorption is evidence of optical pumping .
we argue that , while scattering theory fits many be and emission - line star observations , this new theory has much more potential to explain polarization - in - absorption as seen in herbig ae / be and other stellar systems .
this thesis presents a large spectropolarimetric study that combines new instrumentation , custom processing software , thorough calibrations , cross - instrument comparisons , a massive observing campaign on many targets larger than most studies to date combined , comparison of current theories on multiple objects and finally the creation of a new theory . |
to date , some @xmath12 damped lyman - alpha ( dla ) absorption systems have been identified in the spectra of high - redshift qsos .
the observed absorption troughs indicate concentrations of neutral gas , with large hi column densities , @xmath13 @xmath14 .
there is considerable evidence that these objects are associated with the progenitors of present day galaxies .
this has been confirmed in several cases by direct imaging of qso fields ( steidel et al .
1994 , 1995 , 1996 ; djorgovski et al . 1996
; le brun et al . 1996 ) .
there are various indications that dlas are associated with young galaxies .
the abundances of metals at low ionization stages in dlas are comparable to those found in disk galaxies ( wolfe 1995 ) .
the metal absorption line profiles show a leading edge asymmetry ( wolfe 1995 ; prochaska & wolfe 1997 ) and are shifted relative to the ly@xmath15 emission redshift ( lu , sargent , & barlow 1997 ) , as expected from a rotating thick disk with circular velocities comparable to those seen in spiral galaxies .
furthermore , observations of redshifted 21-cm absorption and emission from dlas indicate disk - like structures of galactic dimensions ( briggs et al . 1989 ; wolfe et al . 1992 ) .
however , recent hst images ( le brun et al .
1996 ) have revealed that dla galaxies span a wide range of morphological types .
the existence of a substantial galaxy population at redshifts @xmath16 is consistent with most cdm models , which predict that galaxies should form by @xmath17 ( frenk et al .
1996 , and references therein ) . the hi column densities in dlas can be derived from the equivalent widths of the observed absorption features .
the mean hi column density along a random line of sight may then be summed up and divided by the absorption path length probed to obtain the comoving spatial hi mass density @xmath18 in galaxies at a given redshift . since dlas dominate the hi mass in the universe ( lanzetta et al .
1995 , and references therein ) , their statistics can be used to infer the evolution of the comoving hi density from high redshifts up to the present epoch .
recent work ( lanzetta et al . 1995 ; wolfe et al . 1995 ; storrie - lombardi , mcmahon , & irwin 1996 ) has shown that , for universes with a zero cosmological constant ( @xmath19 ) , the inferred comoving total gas density @xmath20 ( including hi and he ) at a redshift @xmath21 is comparable to the stellar mass density observed in present day galaxies .
this was seen to be consistent with a simple `` closed
box '' picture in which galaxies had formed by @xmath22 , and the neutral gas was subsequently converted into stars , while the total baryonic mass ( gas + stars ) is conserved .
note that the closed - box model does not assume that the baryonic masses of individual galaxies are conserved but rather that the total baryonic mass of _ all _ galaxies is not evolving with time .
this model allows for mergers which conserve the total mass of all of the galaxies involved .
evolutionary models may be generalized to include a net infall of gas from the intergalactic medium , or a net outflow of gas from dla galaxies .
the closed - box assumption represents the simplest evolutionary model , but in general we do not expect the total baryonic mass in galaxies to be conserved . in particular , the intracluster medium ( icm ) in rich clusters of galaxies contains an amount of iron which is approximately equal to the amount locked in stars in these clusters ( renzini et al .
1993 ; loewenstein & mushotzky 1996 ) .
the natural explanation is that the metals observed in the icm were produced inside of galaxies and subsequently expelled into the icm by supernova driven winds . in this paper
, we examine whether the dla data are consistent with the substantial outflow needed to account for this observation . in the subsequent discussion ,
we also allow @xmath23 , and explore the family of models with @xmath24 , where @xmath3 and @xmath1 are the mean present day cosmic mass densities in matter and vacuum energy , respectively , in units of the critical density .
the absorption path length corresponding to a given redshift interval depends on the assumed cosmological model ; it is shortest for flat , @xmath19 universes , longer for open , @xmath19 universes , and longer still for flat , @xmath23 models .
because the absorption path length is longest for @xmath25dominated cosmologies , we expect the inferred value of @xmath26 at high redshifts to be the smallest , and hence the deficit with respect to the present day stellar density to be the largest in these cosmologies .
this approach is similar to methods which constrain @xmath1 based on the statistics of gravitational lensing ( kochanek 1996 , and references therein ) .
in closed box models , one may investigate the constraints that can be placed on @xmath1 by requiring , for some suitably chosen @xmath27 , that @xmath28 be comparable to the present - day mass density in stars , @xmath29 .
however , a net infall of gas between the high - redshift epoch and @xmath30 can bring the dla data into reasonable agreement with high@xmath25 cosmologies . for our purposes , it is sufficient to parameterize the amount of accretion or expulsion as @xmath31 , where @xmath32 is the total baryonic density in galaxies , and @xmath33 .
in this paper we explore the constraints on both galaxy evolution and cosmology by constructing confidence regions in the @xmath34 plane . to simplify our analysis
, we make the reasonable assumption that the distribution of dlas in hi column density @xmath35 and redshift @xmath36 , @xmath37 , is separable into functions of @xmath35 and @xmath36 over a suitably small redshift interval .
this allows us to fit the column density distribution separately from the extraction of constraints on galaxy formation or the underlying cosmology [ the latter being exclusively related to the redshift dependence of @xmath37 ] .
the value of @xmath29 is calculated by multiplying the local luminosity density by the mean mass - to - light ratio .
recent imaging of dla galaxies ( le brun et al .
1996 ) indicates that , while some are spirals , a significant fraction of these objects have irregular morphologies , which may indicate that a substantial amount of merging was taking place at high redshifts .
since mergers may result in the formation of elliptical galaxies , we include all galaxy types in our calculation of the local luminosity density , and hence @xmath29 .
the assumption that underlies our discussion is that star formation requires _ cold _ hi gas , which must be represented in a fair sample of all ly@xmath15 absorption systems , irrespective of whether the star formation process occurs in spirals or in ellipticals . in 2.1
we show how the comoving hi density is inferred from the dla sample , and how sensitive it is to the underlying cosmology .
section 2.2 adds the impact that evolution in the hi content of galaxies might have on our analysis . in
3 we discuss the statistical methods used to compare the data to the theoretical predictions for @xmath28 . in 4 we present the derived confidence intervals for our constraints on the evolution of galaxies and the cosmological constant . finally ,
5 summarizes our conclusions .
in order to extract useful constraints from the data , we must predict some observable property of the dla sample , and show explicitly how our prediction depends on our assumptions about cosmology and evolution . in 2.1 , we show how , for a given cosmology , the comoving hi density @xmath39 is related to the total dla column density along a line of sight . in 2.2 , we introduce a parameter which characterizes the amount of evolution in the hi density .
the comoving hi density in dla systems at a redshift @xmath36 is inferred by calculating the mean hi column density in a proper length interval @xmath40 along a line of sight , and dividing the result by @xmath41 .
since the absorption path length element @xmath42 corresponding to a given redshift element @xmath43 depends on the cosmological parameters @xmath44 ( the mean present - day cosmic mass densities in matter and vacuum energy , respectively , in units of the critical density ) , the inferred value of the comoving hi density will depend on the assumed geometry of the universe .
let @xmath45 be the mean number of dlas along a line of sight with hi column densities between @xmath35 and @xmath46 and redshifts between @xmath36 and @xmath47 .
note that @xmath37 is different from the function @xmath48 usually encountered in the literature ; the latter is conventionally defined such that @xmath49 gives the number of dlas with with column densities between @xmath35 and @xmath46 and absorption distances between @xmath50 and @xmath51 .
the inferred comoving hi density is given by @xmath52 \int_{n_{\rm min}}^ \infty f(n , z)ndn , \label{eq : rhomean}\ ] ] where @xmath53 is the mass of a hydrogen atom , and @xmath54 @xmath14 is the minimum hi column density included in the sample . for a cosmological model with density parameters ( @xmath44 ) and an open or flat geometry ,
the absorption path length element @xmath42 is given by @xmath55 for a flat matter dominated universe with @xmath56 and @xmath57 , @xmath58 ; for a low density universe with @xmath59 and @xmath57 , @xmath60 ; and for a flat , @xmath25dominated universe with @xmath59 and @xmath61 , @xmath62 .
hence , because the path length corresponding to a given redshift interval @xmath43 is longest for @xmath25dominated cosmologies , the @xmath63 value inferred from equation ( [ eq : rhomean ] ) is smallest for large @xmath1 . in this paper
we will only consider the family of flat cosmologies with @xmath64 , and will write all cosmology dependent expressions in terms of the single adjustable parameter @xmath1 .
the _ observed _ distribution @xmath65 , defined as the _ total _ number of dlas observed in the sample with column densities between @xmath35 and @xmath46 and redshifts between @xmath36 and @xmath47 , depends on the sensitivity of the qso sample to the detection of a dla feature at redshift @xmath36 . in the absence of obscuration of background qsos by dust in the dlas , @xmath66 where @xmath67 here , @xmath68 the heaviside step function , @xmath69 is the total number of qsos in the sample , and @xmath70 is the redshift window over which the observations are able to detect a dla feature in the spectrum of the @xmath71th qso ( depending on the redshift of the qso and the response of the detector ) .
thus , @xmath72 is the number of lines of sight for which a dla feature is detectable at an absorber redshift @xmath36 .
we therefore have @xmath73 { 1\over{g(z ) } } \int_{n_{\rm min}}^\infty f_{\rm obs}(n , z)ndn .
\label{eq : rhotot}\ ] ] equation ( [ eq : rhotot ] ) expresses the inferred value of @xmath39 , factored into functions that depend exclusively on @xmath1 , which we treat as a free parameter , and the properties of the qso and dla samples , which are fixed by observations . given a specified galaxy evolution picture , the evolution of the gas density is obtained by solving the equations of cosmic chemical evolution ( e.g. pei & fall 1995 ) : @xmath74 @xmath75 here , the dot denotes a time derivative ; @xmath26 is the total gas density ( note that @xmath76 , since @xmath26 includes hii , h@xmath77 , he , and heavier elements ) ; @xmath78 is the mass density in stars ; @xmath79 is the net rate at which the total baryonic mass density changes ( @xmath80 corresponds to a net accretion of material from the intergalactic medium , while @xmath81 corresponds to a net expulsion of material from galaxies ) ; @xmath82 is the mean stellar yield ( mass fraction of elements heavier than he produced in stars ) , averaged over the stellar initial mass function ( imf ) ; @xmath83 is the metallicity of the gas ( mass fraction of elements heavier than he present in the gas ) in galaxies , and @xmath84 is the metallicity of the infalling gas . the simplest solution is the `` closed - box '' solution , which has no net accretion or outflow ( @xmath85 ) : @xmath86 \label{eq : closedbox}\ ] ] where we have assumed @xmath87 .
solutions which include accretion or outflow have been identified by pei & fall ( 1995 ) for the case where @xmath88 . for simplicity
, we will assume that @xmath26 is constant within a sufficiently narrow redshift interval around @xmath5 .
although the minimum width of this interval is limited to @xmath8912 by the current size of the dla sample , one can imagine that forthcoming dla surveys will allow one to narrow it much more in the future . in principle
, one could have considered the case of a constant nonzero rate of change @xmath90 ; however , even with @xmath8912 this produces results which do not differ appreciably from the case where @xmath26 is a constant , for a range of reasonable values of @xmath91 .
any deficit in the gas density @xmath28 at @xmath5 relative to the total present day baryonic mass @xmath92 will be due to a combination of two effects : ( 1 ) some material may have accreted onto existing galaxies or assembled into new galaxies since @xmath5 , and ( 2 ) some star formation may have already occurred by redshift @xmath5 , depleting part of the gas . effect ( 2 ) is expected to be sub
dominant based on the low metallicities @xmath93 observed at redshifts @xmath16 ( lu et al .
1996 ) and the relatively small star formation rates observed at such high redshifts ( madau 1996 ) .
nevertheless , we will lump the two effects together , and will parameterize the combined contribution to the deficit by defining @xmath94 where @xmath95 is the fraction of @xmath96 which was present at a redshift of 3 .
( we choose @xmath5 as the fiducial redshift for the comparison , but since we assume that @xmath26=constant over an interval containing @xmath5 , we could just as easily choose any other redshift in this interval for the definition of @xmath4 . ) hence , @xmath97 corresponds to a net accretion and @xmath98 corresponds to a net expulsion of material since @xmath5 .
star formation prior to @xmath5 is expected to produce at most @xmath9925% of @xmath29 ( approximately the fractional area under the the curve @xmath100 inferred from madau 1996 , corrected for high-@xmath25 cosmologies ) .
let @xmath101 be the fraction of the total present - day mass in stars which were produced by @xmath0 ; then , with @xmath33 , we have @xmath102 finally , we neglect the contributions of ionized hydrogen , molecular hydrogen ( see ge & bechtold 1997 ) , as well as metals , to @xmath26 at redshifts @xmath16 , but include helium , 25% by mass , resulting in a mean molecular weight of 1.3@xmath53 .
we therefore use @xmath103 .
next we describe the methods used to compare the data to the model prediction @xmath39 for the mean hi density at high redshift , at given values of @xmath4 and @xmath1 . in 3.1 , we construct the likelihood function , and in 3.2 , we use bayesian analysis to derive confidence regions in the ( @xmath104 ) plane from the likelihood function .
the likelihood function gives the probability of obtaining a particular data set , assuming the truth of a specified model .
the model should include predictions for both the expected value of the quantity in question and the _ distribution _ of measurements due to statistical ( e.g. poisson ) fluctuations about the expected value .
the most commonly used likelihood estimator in the literature is the @xmath105 statistic , which is simply proportional to the logarithm of the likelihood function for normally distributed errors .
its widespread use is due to two convenient facts : the tendency , due to the central limit theorem , of sums of many random variables to be normally distributed ; and the fact that the distribution of the @xmath105 statistic for normally distributed errors is known , which makes it easy to calculate confidence regions .
however , the @xmath105 statistic is not appropriate for our analysis here . to apply the @xmath105 statistic
, we would need to collect the data into redshift bins , and estimate the value of @xmath18 in each bin by summing up the column densities and dividing by the total absorption path length probed in the bin .
however , given our small sample size ( only 73 objects in total ) , such experimentally - measured values of @xmath18 will not be normally distributed .
in fact , the values we measured from monte - carlo - simulated data sets show a significant skewness in their distribution , and so we are not justified in using the @xmath105 statistic to calculate confidence regions
. moreover , this approach introduces arbitrariness in the choice of bin size and also loses information about the dla sample by binning the data .
we have therefore chosen to construct a likelihood function which makes use of the _ unbinned _ data , and which does not rely on the assumption of normally distributed errors .
the starting point for constructing our likelihood function is to note that any prediction for @xmath39 can be related to the observed distribution of dlas in column density and redshift using equation ( [ eq : rhotot ] ) .
since we assume that @xmath39 is approximately constant over some narrow redshift interval around @xmath5 , we may use equation ( [ eq : rhog ] ) , and rearrange ( [ eq : rhotot ] ) to write @xmath106 { \ } \rho_{\rm s0}(1+\eta - f ) , \label{eq : fndn}\ ] ] where @xmath107 .
thus , given an evolution model @xmath39 and a cosmological model @xmath1 , we have a definite prediction for a property of the dla sample , namely the total column density as a function of redshift [ given by the left hand side of equation ( [ eq : fndn ] ) ] .
note that ( [ eq : fndn ] ) gives the _ expected _ value of this quantity ; it is not obvious how the values measured from hypothetical data sets should be distributed about this expected value ( as mentioned above , given the small sample size , the values in each redshift bin will not be normally distributed ) .
specifically , the distribution of measured @xmath39 values depends on the shape of @xmath108 as a function of hi column density @xmath35 , and will be skewed toward smaller values if @xmath108 is dominated by low@xmath35 systems , and vice - versa .
unfortunately , our evolution models @xmath39 predict only the first moment of @xmath108 , not the distribution itself .
however , the dla sample _ does _ tell us something about the full distribution ; in particular , the entire sample is well fitted by a so called gamma distribution ( storrie - lombardi , irwin , & mcmahon 1996 ) : @xmath109 where @xmath110 and @xmath111 are functions of @xmath36 and @xmath112const .
if we specialize to the case where @xmath111 is a constant , then @xmath110 contains all of the @xmath36dependence , and is proportional to @xmath113 ( i.e. , all of the redshift dependence is due to the cosmological geometry ) . in this case , we may rewrite equation ( [ eq : fndn ] ) as @xmath114{\ } \rho_{\rm s0}(1+\eta - f),\nonumber\\ \label{eq : fdn}\end{aligned}\ ] ] where @xmath115 is the mean column density in the distribution , with @xmath116 the incomplete gamma function .
note that equation ( [ eq : fdn ] ) , with @xmath117 independent of redshift , applies for _ any _ distribution @xmath118 which is separable into functions of @xmath35 and @xmath36 .
the assumption of separability is particularly appealing if only a small fraction of the available hi is depleted from dla galaxies during the redshift range under consideration . suppressing the dependence on @xmath1 , @xmath4 , and @xmath119 for brevity ,
the quantity @xmath120 defined in equation ( [ eq : fdn ] ) denotes the redshift distribution of absorbers ; the number of systems observed with redshifts between @xmath36 and @xmath47 is @xmath121 . for the moment , assume that we know the _ shape _ of @xmath108 a priori , and hence that we know @xmath117 from equation ( [ eq : navg ] ) .
[ we will address the question of how to incorporate properly our incomplete knowledge of @xmath108 from the data into our analysis in the next section .
] then we may construct a likelihood function from equation ( [ eq : fdn ] ) as follows .
suppose we wish to compare the data to our prediction over a range of redshifts ( @xmath122 ) .
if we divide this range into sufficiently small intervals @xmath43 , such that @xmath123 , then there will be at most one object in each such interval .
then the probability of finding no objects in a given interval at a redshift @xmath36 is given by the poisson distribution , @xmath124 $ ] ; similarly , the probability of finding one object is @xmath125 $ ] ; and the probability of finding more than one object is negligible , @xmath126 ^ 2\}\ll 1 $ ] .
then the likelihood , or conditional probability of obtaining a particular data set @xmath127 from a distribution @xmath120 predicted by ( [ eq : fdn ] ) is @xmath128\right\ } \left\{\prod_{j=1}^{m_{\rm none } } \exp[-n(z_j)dz]\right\ } , \label{eq : likea}\ ] ] where @xmath129 is the number of dla systems observed , @xmath130 is the redshift of the @xmath71th system , @xmath131 is the number of intervals @xmath43 in our range ( @xmath122 ) which have _ no _ dlas in them [ note that @xmath132 , @xmath133 is the redshift of the @xmath134th such interval , and we have suppressed the dependence of @xmath120 on the parameters @xmath135 for brevity . for @xmath136 , equation ( [ eq : likea ] ) may be written as @xmath137 \exp\left[-\int_{z_{\rm min}}^{z_{\rm max } } n(z)dz\right ] .
\label{eq : likeb}\ ] ] this probability clearly depends on the chosen size of the interval @xmath43 , and shrinks to zero as @xmath138 .
this reflects the fact that , as @xmath43 shrinks , the number of possible distinct outcomes @xmath127 increases , so the probability of obtaining any _ particular _ data set @xmath127 goes to zero . in the next section
, we will use bayesian inference to construct confidence intervals from the likelihood function ( [ eq : likeb ] ) .
equation ( [ eq : likeb ] ) gives the probability of obtaining our dla sample as a random realization of the redshift distribution ( [ eq : fdn ] ) , _ assuming _ the truth of a particular cosmology @xmath1 and evolution model @xmath4 , and assuming knowledge of the expected average column density @xmath117 in the sample .
however , our evolution model for @xmath39 does not make any predictions about the value of @xmath117 ; we must make use of the dla sample to extract information about the range of reasonable values for @xmath117 .
in addition , the value of @xmath139 is uncertain since it is related to the uncertain mass - to - light ratios of present - day galaxies . both of these uncertainties in our knowledge may be rigorously incorporated into our analysis if we use bayesian inference .
bayes theorem follows trivially from the axioms of probability and the definition of conditional probability .
the theorem relates @xmath140 , the conditional probability distribution for values of the model parameters given the observed data and an assumed value of @xmath117 , to the likelihood @xmath141 [ cf .
( [ eq : likeb ] ) ] .
the conditional probability of obtaining the observed data set as a random realization of the model with particular parameter values is , @xmath142 here , @xmath143 is the _ prior _ probability distribution for the two parameters , which , in the absence of any previous data , we take to be uniform and independent of @xmath117 ; and @xmath144 is a normalization constant which ensures that @xmath145 . note that the value of @xmath43 in equation ( [ eq : likeb ] ) , which may be taken to be arbitrarily small , is absorbed into @xmath144 .
unfortunately , our _ a priori _ knowledge does not tell us the precise value of @xmath117 .
however , we may include as an additional component of our model the assumption that the data are well fitted by a gamma distribution ( [ eq : gammadist ] ) . by fitting a functional form ( [ eq : gammadist ] ) to the data
, we may obtain a prior distribution @xmath146 of reasonable values of @xmath147 .
the procedure for obtaining @xmath146 is identical to the procedure we employ to get @xmath148 .
note that the gamma distribution is not the only plausible functional form to fit to the data ; however , it provides a better fit than a single power law ( storrie - lombardi , irwin , & mcmahon 1996 ) , since there is a significant break in the observed slope of @xmath37 .
any other reasonable two - parameter distribution ( e.g. a broken power law ) will yield similar results to the gamma distribution .
the sum rule of probability theory allows us to `` marginalize '' the parameter @xmath117 by integrating equation ( [ eq : bayesa ] ) over values of @xmath117 , weighted by @xmath146 : @xmath149 where we have assumed that @xmath143 is uniform , and have absorbed all constants into @xmath144 such that @xmath150 is still normalized to unit area . we may incorporate the uncertainty in @xmath139 in a similar way .
as will be seen in 4.1 , the uncertainty in @xmath139 is approximately gaussian .
then , making the dependence on @xmath139 explicit , we finally have @xmath151 p(d|\omega_\lambda,\eta,\n_{\rm avg},\rho_{\rm s0 } ) , \label{eq : bayesc}\ ] ] where once again the normalization has been absorbed into @xmath144 .
we will use equation ( [ eq : bayesc ] ) , substituting equations ( [ eq : fdn ] ) and ( [ eq : likeb ] ) for @xmath152 and obtaining @xmath153 from a fit to the data , to compare the data to the models and calculate confidence regions .
we include in our sample all dla systems whose redshifts and hi column densities ( or , in a few cases , equivalent widths ) have appeared in the literature ( wolfe et al .
1986 ; lanzetta 1991 ; lanzetta 1995 ; wolfe et al . 1995 ; storrie - lombardi , mcmahon , & irwin 1996 ) , for a total of 73 systems . in the cases where no hi column density has been confirmed ,
we have calculated it from the reported equivalent width using equation ( 3 ) in wolfe et al .
this is the same sample used by storrie - lombardi , mcmahon , & irwin 1996 , with the addition of data from wolfe et al .
( 1995 ) . in our analysis
, we have concentrated on high redshifts ( @xmath27 ) , and hence have used subsets of this sample .
since our final results depend sensitively on the value and degree of uncertainty in the local stellar density @xmath139 , it is important that we obtain the most accurate and precise possible estimate of this quantity from the literature .
previous studies ( wolfe et al . 1995 ; lanzetta et al .
1995 ; storrie - lombardi , mcmahon , & irwin 1996 ) have employed the value @xmath154 ( gnedin & ostriker 1992 ) , where @xmath155 is the present - day critical density , and the hubble constant is @xmath156 km s@xmath157 mpc@xmath157 .
we make use of recent observations to refine this estimate .
we compute @xmath139 by multiplying the local luminosity density of galaxies by their mean stellar mass - to - light ratio .
since the mass - to - light ratio is in general correlated with the galaxy luminosity , we have @xmath158 where @xmath159 is the galaxy luminosity function ( lf ) and @xmath160 is the mass - to - light ratio in solar units .
we have used the luminosity function for the ns112 sample of the las campanas redshift survey ( lin et al .
1996 ) , which represents the most precise determination of the local lf to date .
we include all galaxy types , since recent identifications of dla galaxies ( le brun et al .
1996 ) have indicated that dla systems may be associated with a wide range of morphological types .
lin et al . ( 1996 ) obtain a best - fit schechter function with the parameters @xmath161 , @xmath162 , and @xmath163 mpc@xmath164 , where the photometry was done in a band very similar to the cousins @xmath165 band .
lin ( 1997 , private communication ) points out that these schechter parameters also provide a good fit for the gunn @xmath166-band luminosity function after a correction is made from isophotal to total galaxy magnitudes . using a solar absolute magnitude of @xmath167 ( broeils , 1997 , private communication )
, this yields an @xmath166-band luminosity density of @xmath168 mpc@xmath164 .
for the mass - to - light ratio @xmath160 , we use the relation measured by broeils & courteau ( 1996 ) in the gunn @xmath166-band for the disks of sbc - type galaxies : @xmath169h$ ] , where the uncertainty is the @xmath170 deviation and the distribution of observed values is approximately gaussian .
thus , the uncertainty in @xmath139 is dominated by the gaussian uncertainty in @xmath160 .
we include all galaxy types in our calculation of @xmath139 , with the assumptions that spiral galaxies are well described by maximal
disk models , and that spiral and elliptical galaxies have the same stellar mass - to - light ratio .
we take the mass - to - light ratio for sbc galaxies as typical for all galaxy types of the same luminosity .
our approach overestimates the stellar mass - to - light ratio somewhat if , as some studies suggest ( rix et al .
1997 ) , the dark matter accounts for a significant fraction of the total mass in the inner regions of galaxies . using equation ( [ eq : rhos ] ) , we obtain @xmath171 based on the uncertainty in @xmath160 , we assume that the values of @xmath139 are normally distributed with a variance @xmath172 .
the observed column density distribution of dla systems , and hence the inferred value of @xmath18 , is affected by the presence of dust in the dla galaxies ( fall & pei 1993 ) .
the dust obscures background qsos , and causes incompleteness in the dla sample .
the presence of dust leads to an underestimate of @xmath18 from the data ; accounting for this fact will lead to better agreement between data and predictions for high-@xmath25 cosmologies .
fall & pei ( 1993 ) showed that the true column density distribution is given by @xmath173 , \label{eq : ftrue}\ ] ] where @xmath174 is the power - law slope at the bright end of the qso luminosity function [ @xmath175 , and @xmath176 is the extinction optical depth , given by @xmath177 here , @xmath178 is the dimensionless dust - to - gas ratio , @xmath179 is the extinction curve , and @xmath180 is the effective wavelength of the band of the qso survey . we have used equations ( [ eq : ftrue ] ) and ( [ eq : taudef ] ) , together with an assumption that the observed distribution @xmath108 may be fit by a gamma distribution ( [ eq : gammadist ] ) , and that @xmath181 is proportional to the metallicity , to estimate the correction to the expected value of @xmath39 due to dust obscuration .
following pei & fall ( 1995 ) , we assume that @xmath182 , @xmath183 , and @xmath184 for qso surveys in the @xmath185-band .
we assume that the metallicity at redshifts @xmath0 is approximately one - tenth of the solar value , @xmath186 .
we find that the effect of dust is to reduce the inferred value of @xmath18 by a factor of @xmath187 .
our results in the next section will include this correction .
we have applied equation ( [ eq : bayesc ] ) , corrected for dust extinction , to the dla sample for two redshift ranges , @xmath188 and @xmath189 , allowing @xmath190 , @xmath191 . to compute the prior distribution @xmath146 ,
we first fit a gamma - distribution ( [ eq : gammadist ] ) to the data , and use bayesian methods similar to those described in 3.2 to obtain a probability distribution @xmath192 .
we fix @xmath110 by requiring that the integral of @xmath108 over all column densities be equal to the total number @xmath129 of dlas observed in the narrow redshift interval under consideration .
we then obtain a cumulative probability distribution for @xmath119 by integrating numerically the probability @xmath192 inside contours of constant @xmath117 in the @xmath193 plane , given by equation ( [ eq : navg ] ) .
we differentiate this cumulative distribution numerically to obtain the differential distribution @xmath146 .
finally , we use equations ( [ eq : fdn ] ) and ( [ eq : likeb ] ) to obtain the likelihood function @xmath152 . substituting these results into equation ( [ eq : bayesc ] )
gives the differential probability distribution @xmath150 .
confidence regions are obtained by integrating this distribution inside contours of constant @xmath150 .
our results do not depend strongly on the value of the hubble constant ; we assume @xmath194 km s@xmath157 mpc@xmath157 . in figure 1a , we show 68% , 95% , and 99% confidence regions in the @xmath195 plane , for the redshift range @xmath188 ; figure 1b shows the same for the redshift range @xmath189 . in both cases ,
we assume that 25% of the present - day mass density in stars had been assembled by @xmath11 ( @xmath196 ) , consistent with the star formation rate of madau ( 1996 ) .
the effect of the sample size is evident : the 99% confidence region is 20% smaller for the larger sample ( 45 objects with @xmath188 ) than for the smaller sample ( 17 objects with @xmath189 ) .
note that , for the larger sample , the data are inconsistent with a net expulsion of more than half of the gas from galaxies ( @xmath197 ) at the 99% confidence level , and also rule out @xmath198 at 95% confidence and @xmath199 at 68% , in all flat cosmologies .
one may compare this result with the observational finding that the intracluster medium ( icm ) in rich clusters of galaxies contains an amount of iron which is approximately equal to the amount found in the stellar component of galaxies in these clusters ( renzini et al .
1993 ; loewenstein & mushotzky 1996 ) .
the natural explanation is that the iron in the icm was produced in supernova explosions inside of galaxies , and was subsequently expelled into the icm by supernova driven winds .
if galaxies were fully assembled by a redshift @xmath200 , then they should have had twice as much baryonic material at such redshifts than is observed locally .
since our analysis excludes the possibility that there was more than 1.5 ( 1.75 ) times the present day baryonic density at 95% confidence for the larger ( smaller ) samples , it suggests that ( 1 ) infall of material into existing galaxies or formation of new galaxies has taken place since @xmath11 ; or ( 2 ) more than 25% by mass of the present - day stellar population formed by @xmath11 .
these possibilities will be discussed in more depth in 5 .
note also that for high@xmath1 cosmologies , outflow models are strongly ruled out , so it would be more difficult to reconcile the dla data with the intracluster iron observations in an @xmath1-dominated universe .
if we wish to specialize to a particular cosmological model , we should consider the one - dimensional probability distribution @xmath201 , since the sizes of the confidence intervals decrease when the number of free parameters is reduced .
figure 2a shows the probability distribution for @xmath4 , given @xmath57 ( the case corresponding to the most conservative constraints on @xmath4 ) .
we consider the sample with @xmath188 .
the solid curves correspond to different assumptions about the amount of star formation that took place at @xmath202 : @xmath203 , 0.25 , and 0.5 , from left to right .
upper limits on @xmath4 derived from these curves are summarized in table 1 . as @xmath101 increases , the gas density observed in dlas is supplemented by more and more stars , so the data become more consistent with the closed box model . as can be seen from figure 2a ,
the data are consistent with modest to significant amounts of infall or formation of new galaxies subsequent to @xmath204 .
significant expulsion is less likely ; for @xmath205 the data are inconsistent with @xmath206 at 95% confidence .
if one allows for half of the present - day stellar population to form before @xmath5 , then one achieves marginal consistency with @xmath207 at 95% confidence .
however , in this case the metallicity of dla systems will significantly exceed its observed value of @xmath208 at @xmath209 ( lu et al .
figure 2b shows similar results for @xmath210 ; here , no significant amount of expulsion is viable . hence , it is unlikely that all of the baryonic material seen today in galaxies and the icm of rich clusters was present in dla galaxies at redshifts @xmath27 .
the dotted curves in figure 2 are the results obtained by considering present - day galaxies + icm as a whole , with equal mass densities in stars and the icm .
hence , we attempt to account for an amount `` @xmath96''= @xmath211 .
we assume that 25% of the present - day mass in stars + enriched icm ( that is , 0.5@xmath139 ) was present in the form of stars by @xmath11 .
clearly , only about half of this amount was present in dla galaxies at @xmath27 ( see fig .
2 ) ; the rest of it must have assembled into galaxies ( and some of it subsequently expelled ) after @xmath11 .
this possibility will be discussed further in 5 .
the constraints obtained with our method will improve as the catalog of dla systems grows .
the sloan digital sky survey , which is getting underway in 1997 ( gunn & knapp 1993 ; see also http://www.astro.princeton.edu/bbook ) , will catalog @xmath212 quasars , at least an order of magnitude more than the number discovered to date ( loveday 1996 ) .
spectroscopic follow - ups on this sample could increase the dla sample size by 12 orders of magnitude . in figure 3
, we predict the effect on our results of a more modest increase in the dla sample size .
figure 3a indicates how the probability distribution @xmath213 changes when the sample size @xmath129 with @xmath189 is increased by a factor of 2 or 5 . for illustrative purposes ,
we choose @xmath57 and @xmath196 .
we take into account the observational uncertainty @xmath214 in the present day stellar density @xmath139 , and assume that the additional dlas have the same column density distribution as the existing sample . in this case
it is a simple analytical matter to determine @xmath213 for a hypothetically larger data set .
the trend is obvious : as more data are acquired , our measurement of @xmath4 becomes more precise , and @xmath213 becomes more sharply peaked .
given our assumptions , the mean value of @xmath4 will remain constant , but the confidence intervals will shrink . for the case shown in figure 3a , the 95% upper bound on @xmath4 is 0.32 , 0.10 , and -0.08 , respectively , for @xmath215 , 34 , and 85 . in reality , the mean value of @xmath4 will shift around as the column density distribution becomes better known [ i.e. @xmath119 in equation ( [ eq : navg ] ) will not remain constant as observations improve ] , but the width of @xmath213 will still behave in the same way .
the precision with which we can measure @xmath4 is limited by the 25% uncertainty in @xmath139 .
figure 3b demonstrates what would happen if @xmath139 were known exactly . in this case , the statistics are purely poisson and the width ( and hence the height ) of our normalized distribution scales as @xmath216 .
we have used the catalog of dla systems to place constraints on the amount of evolution in the baryonic content of galaxies and the value of the cosmological constant .
we compared the gas density @xmath26 at redshift @xmath0 to the present day stellar mass density @xmath217 in galaxies for a range of flat cosmologies with @xmath218 .
our underlying assumption is that cold gas is required for star formation .
we make use of the facts that dla systems dominate the hi content of the universe at redshifts @xmath27 and the correction to their baryonic mass due to molecular gas is small , @xmath219% ( ge & bechtold 1997 ) .
we defined @xmath4 to be the net fraction of the baryonic content of local galaxies which was expelled since @xmath220 , and used bayesian inference to derive confidence regions in the ( @xmath6 ) plane . in all cosmologies
we find that @xmath7 with at least 95% confidence , as long as only @xmath9 of the current stellar population formed before @xmath5 .
the most likely value of @xmath4 is negative ( cf .
2 ) , implying a net _ increase _ by several tens of percent in the baryonic mass of galaxies since @xmath204 .
the inferred value of @xmath4 is more extreme for @xmath1dominated cosmologies . on the other hand , recent observations of high metal abundances in the intracluster medium of rich clusters ( renzini et al .
1993 ; loewenstein & mushotzky 1996 ) require that metal
rich gas be _ expelled _ from galaxies in an amount approximately equal to the current mass in stars .
the possibility that a dominant fraction of the present day stellar population may have already formed by @xmath5 and resulted early on in this expulsion , is ruled out by the low metalicities ( @xmath221@xmath222 solar ) observed in dlas at @xmath223 ( lu et al .
moreover , the intergalactic medium which later accretes onto clusters of galaxies has a metallicity as low as @xmath224 solar at @xmath209 ( cowie et al .
1995 ; tytler et al .
1995 ; songaila & cowie 1996 ; cowie 1996 ) , rather than the needed value of @xmath225 solar .
most of the required metal enrichment and star formation activity must therefore have occurred at @xmath226 .
the most likely explanation to the above discrepancy is that a significant amount of gas had been assembled and partly expelled from galaxies after @xmath11 .
the increase in galactic mass could have been either in the form of accretion onto existing galaxies or through the formation of new galaxies , such as those responsible for the faint excess in deep galaxy counts ( lowenthal et al .
1996 , and references therein ) . the likely value of @xmath4 of minus several tenths ( fig .
2 ) , implies that more than half the associated baryonic mass was processed through galaxies after @xmath11 . as an example , let us assume that @xmath227 at @xmath0 , and that 150% of the current baryonic mass of galaxies had assembled after @xmath11 .
this implies that the total mass processed through galaxies is twice ( 150%+50% ) their current mass , as required by the observation that clusters contain twice the iron locked up in stars ( renzini et al .
1993 ; elbaz , arnaud , & vangioni - flam 1995 ; lowenstein & mushotzky 1996 ) .
half of the processed mass was converted into galactic stars and half expelled into the intergalactic medium . if the expelled gas is @xmath228 of the intergalactic gas it mixed with [ assuming @xmath229 and the value of @xmath217 in eq .
( [ eq : rhosvalue ] ) ] , then it could have yielded the @xmath230 solar metallicity observed in clusters as long as its original metallicity was a few times solar .
the phase of massive metal enrichment must have occurred at @xmath2312 since the iron abundance in clusters shows little evolution at @xmath232 ( mushotzky & lowenstein 1997 ) .
this inference could be tested observationally through a dedicated search for enhanced star formation activity and supernova rate at @xmath2312 .
we have employed a bayesian analysis which has the dual advantages of taking the various observational uncertainties properly into account , and making use of unbinned data .
the constraints obtained with our method will improve as the size of the quasar sample increases ( cf .
3 ) . in particular , future spectroscopic observations of the @xmath212 quasars cataloged by the sloan digital sky survey ( http://www.astro.princeton.edu/bbook ) , could increase the current sample size of dlas by 12 orders of magnitude , and improve our limits on the amount of galactic evolution considerably .
briggs , f. h. , wolfe , a. m. , liszt , h. s. , davis , m. m. , & turner , k. l. 1989 , apj , 341 , 650 broeils , a. h. , & courteau , s. 1996 , preprint astro - ph/9610264 cowie , l. 1996 , in hst and the high redshift universe , proc .
37th herstmonceux conference , eds .
tanvir , n. r. , aragn - salamanca , a. , wall , j.v .
( world scientific : singapore ) , in press , ( preprint astro - ph/9609158 ) cowie , l. l. , songaila , a. , kim , t .- s .
, & hu , e. m. 1995 aj , 109 , 1522 djorgovski , s. g. , pahre , m. a. , bechtold , j. , & elston , r. 1996 , nature , 382 , 234d elbaz , d. , arnaud , m. , & vangioni - flam , e. 1995 , a & a , 303 , 345 fall , s. m. , & pei , y. c. 1993 , apj , 402 , 479 frenk , c. s. , baugh , c. m. , cole , s. , & lacey , c. g. 1996 , to appear in `` dark matter 1996 : dark and visible matter in galaxies and cosmological implications '' , eds m. persic and p. salucci ( preprint astro - ph/9612109 ) ge , j. , & bechtold , j. 1997 , apj , 477 , l1 , in press ( preprint astro - ph/9701041 ) gnedin , n. yu .
, & ostriker , j. p. 1992 , apj , 400 , 1 gunn , j. e. , & knapp , j. 1993 , _ sky surveys _
b. t. soifer , asp conference series # 43 , 267 kochanek , c. s. 1996 , apj , 466 , 638 lanzetta , k. m. , wolfe , a. m. , & turnshek , d. a. 1995 , apj , 440 , 435 lanzetta , k. m. , et al .
1991 , apjs , 77 , 1 le brun , v. , bergeron , j. , boiss , p. , & deharveng , j. m. 1996 , a&a , in press ( preprint astro - ph/9611031 )
lin , h. , et al .
1996 , apj , 464 , 60 loewenstein , m. , & mushotzky , r. f. 1996 , apj , 466 , 695 lowenthal , j. d. , koo , d. c. , guzman , r. , gallego , j. , phillips , a. c. , faber , s. m. , vogt , n. p. , illingworth , g. d. , & gronwall , c. 1996 , apj , in press ( preprint astro - ph/9612239 ) loveday , j. 1996 , preprint astro - ph/9605028 lu , l. , sargent , w. l. w. , barlow , t. a. , churchill , c. w. , & vogt , s. 1996 , apjs , 107 , 475 lu , l. , sargent , w. l. w. , & barlow , t. a. 1997 , apj , in press ( preprint astro - ph/9701116 ) madau , p. 1996
, preprint astro - ph/9612157 madau , p. , ferguson , h. c. , dickinson , m. e. , giavalisco , m. , steidel , c. c. , & fruchter , a. 1996 , preprint astro - ph/9607172 mushotzky , r. f. , & loewenstein , m. 1997 , preprint astro - ph/9702149 pei , y. c. , & fall , s. m. 1995 , apj , 454 , 69 prochaska , j. x. , & wolfe , a. m. 1997 , apj , submitted ( preprint astro - ph/9605021 rix , h .- w . , de zeeuw , p. t. , carollo , c. m. , cretton , n. , & van der marel , r. p. 1997
, preprint astro - ph/9702126 renzini , a. , ciotti , l. , dercole , a. , & pellegrini , s. 1993 , apj , 419 , 52 songaila , a. , & cowie , l. l. 1996 , aj , 112 , 335 steidel , c. c. , pettini , m. , dickenson , m. , & persson , s. e. 1994 , aj , 108 , 2046 steidel , c. c. , bowen , d. v. , blades , j. c. , & dickenson , m. 1995 , apj,440 , l455 steidel , c. c. , giavalisco , m. , pettini , m. , dickinson , m. , & adelberger , k. l. 1996 , aj , 112 , 352 storrie - lombardi , l. j. , irwin , m. j. , & mcmahon , r. g. 1996 , preprint astro - ph/9608146 storrie - lombardi , l. j. , mcmahon , r. g. , & irwin , m. j. 1996 , preprint astro - ph/9608147 tytler , d. et al .
1995 , in qso absorption lines , eso astrophysics symposia , ed .
( heidelberg : springer ) , p.289 wolfe , a. m. 1995 , in _ qso absorption lines _ , ed
. g. meylan ( heidelberg : springer ) , pp .
13 - 22 wolfe , a. m. , turnshek , d. a. , smith , h. e. , & cohen , r. d. 1986 , apjs , 61 , 249 wolfe , a. m. , turnshek , d. a. , lanzetta , k. m. , lu , l. , & oke , j. b. 1992 , apj , 385 , 151 wolfe , a. m. , lanzetta , k. m. , foltz , c. b. , & chaffee , f. h. 1995 , apj , 454 , 698 * table 1 : * upper limits on @xmath4 for two cosmologies and different fractions @xmath233 of stars formed at redshifts @xmath0 . the last column includes the mass density in the intracluster medium ( icm ) , assuming 25% of the total ( stars+icm ) was in the form of stars at @xmath0 . see figure 2 for the differential probability distributions @xmath213 corresponding to columns 36 . | we use the existing catalog of damped lyman alpha ( dla ) systems to place constraints on the amount of evolution in the baryonic content of galaxies and on the value of the cosmological constant .
the density of cold gas at redshifts @xmath0 is obtained from the mean hi column density of dlas per cosmological path length . this path length per unit
redshift is in turn a sensitive function of the vacuum density parameter , @xmath1 .
we compare the total inferred mass of cold gas at high redshifts to that observed in stars today for cosmologies with @xmath2 , where @xmath3 is the matter density parameter .
we define @xmath4 to be net fraction of the baryonic content of local galaxies which was expelled since @xmath5 , and use bayesian inference to derive confidence regions in the ( @xmath6 ) plane . in all cosmologies
we find that @xmath7 with at least @xmath8 confidence if @xmath9 of the current stellar population formed before @xmath5 .
the most likely value of @xmath4 is negative , implying a net _ increase _ by several tens of percent in the baryonic mass of galaxies since @xmath0 . on the other hand , recent observations of high metal abundances in the intracluster medium of rich clusters ( loewenstein & mushotzky 1996 ) require that metal
rich gas be _ expelled _ from galaxies in an amount approximately equal to the current mass in stars . based on our results and
the low metallicity observed in dlas at @xmath10 , we infer that more than half of the baryonic mass processed through galaxies must have been assembled and partly expelled from galaxies _ after _ @xmath11 .
we expect our constraints to improve considerably as the size of the dla sample will increase with the forthcoming sloan digital sky survey .
submitted to _ the astrophysical journal _ , march 1997 |
a well - established observational hallmark of how galaxies evolve as a function of environment is the star formation - galaxy density relation .
a plethora of studies utilizing multi - wavelength tracers of activity have shown that star formation universally decreases with increasing galaxy density at @xmath16 ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in particular , the cores of massive galaxy clusters are galaxy graveyards full of massive spheroidal systems that are dominated by old stellar populations . however , as we approach the epoch when these quiescent behemoths should be forming the bulk of their stars ( @xmath17 ; * ? ? ? * ; * ? ? ? * ; * ? ? ?
* ) , the star formation - density relation should weaken and possibly reverse . identifying when star formation is quenched as a function of galaxy mass and environment provides strong constraints on galaxy models ( e.g. * ? ? ?
* ; * ? ? ?
* ) , i.e. is individual galaxy mass or the larger scale environment the primary driver of evolution ?
observations of field galaxies at @xmath18 indicate that the star formation - density relation turns over at this epoch such that there is an enhancement of activity in the highest density regions of the field @xcite .
studies also find an excess of dust - obscured star formation in group environments at @xmath16 @xcite , and very recent results suggest that star formation may be enhanced in the significantly richer core of a galaxy cluster at @xmath19 @xcite .
as cluster surveys push to higher redshifts ( @xmath20 ) and thus closer to the epoch when massive galaxies are forming their stars , variations in age will become evident in , e.g. a larger scatter in color , and robust star formation rates should reveal increasing levels of activity even in the cluster cores .
we report here the first confirmation of increasing star formation activity with increasing galaxy density observed in cluster clg j0218.3 - 05101 , a spitzer - selected galaxy cluster at @xmath0 ( * ? ? ?
* hereafter pap10 ) .
we use cosmological parameters @xmath21 , @xmath22 , and @xmath23 km s@xmath6 mpc@xmath6 throughout the paper ; at @xmath0 , this corresponds to an angular scale @xmath24 mpc arcmin@xmath6 .
all magnitudes are in the ab system .
clg j0218.3 - 0510 has a wealth of multi - wavelength imaging data that includes @xmath25 imaging from the subaru - xmm deep survey @xcite and @xmath26 imaging from the ukirt ir deep sky survey ( ukidss , * ? ? ?
* ) ; for these data , we utilized the @xmath27selected catalog from @xcite .
the cluster field also has deep _ spitzer _ imaging available in the four irac bands ( 3.68.0@xmath28 m ) and mips @xmath3 m as part of the _ spitzer _ public legacy survey of the ukidss ultra deep survey ( spuds , pi : j. dunlop ) . we matched the @xmath27band selected catalog to the irac data following @xcite .
clg j0218.3 - 05101 was originally detected as a strong overdensity of galaxies at @xmath29 with a mean surface density @xmath30 higher than that of galaxies in the same redshift range in the ukidss ultra - deep survey @xcite .
follow - up spectroscopy confirmed 10 members that define a narrow redshift peak at @xmath31 with additional members at @xmath32 and @xmath33 ( pap10 ; * ? ? ? * ) . in our analysis , we include 80 more members that are selected from photometric redshifts determined with eazy @xcite ; we use the limit for the integrated redshift probability defined by the spectroscopically confirmed members ( @xmath34 ; see pap10 for a detailed description of the photometric selection ) .
analysis of xmm - newton data in this field also reveals a weak detection consistent with extended emission from the cluster .
the cluster mass estimated from its x - ray luminosity is @xmath35 m@xmath5 and consistent with its estimated dynamical mass of @xmath36 m@xmath5 . to define the cluster center
, we select the massive ( spectroscopically confirmed ) member located at the peak of the x - ray emission : its j2000 coordinates are ( 2:18:21.07 , -5:10:32.84 ) , and all of the cluster galaxies lie within @xmath37 mpc of this member .
we fit the 10band galaxy photometry ( @xmath1m@xmath2 m ) , this corresponds to a rest - frame wavelength range of @xmath38m@xmath39 m . ] with the 2007 version of the @xcite stellar population synthesis models using a chabrier initial mass function ( for more details , see * ? ? ?
we find that models with solar metallicity best reproduce the rest - frame colors and scatter of the cluster red sequence galaxies ( pap10 ) , and we allow the models to range in age from @xmath40 to @xmath41 yr ; we also include dust attenuation using the @xcite law with color excess values ranging from @xmath42 .
we allow for a range of star formation histories parameterized as a decaying exponential with an @xmath43folding time @xmath44 , where at any age @xmath45 the star formation rate is @xmath46 and @xmath44 ranges from 1 myr to 100 gyr ( corresponding approximately to instantaneous bursts to constant star formation , respectively ) . in the model
fitting , we add a @xmath47% error in quadrature to the photometric uncertainties on each band to account for mismatches in the multi - band photometry , and for the fact that the models do not continuously sample the model parameter space .
our sample of 92 cluster members has an average best - fit age of @xmath48 myr and an average best - fit @xmath44 of @xmath48 myr ; 35 members have sed - derived star formation rates @xmath495 m@xmath5 yr@xmath6 .
the cluster field was imaged with spitzer mips as part of the legacy ukidss ultra deep survey ( spuds pi : j. dunlop ) .
we extracted sources from the public spuds @xmath3 m map using starfinder @xcite , an idl - based psf - fitting code designed for crowded fields .
the @xmath3 m image has 1.245pixel@xmath6 , and we derived a model psf and aperture corrections from the brightest isolated sources from the spuds image .
the catalog includes all sources detected with @xmath50 ; this corresponds to a flux of @xmath51jy . using simulated sources based on the psf and injected into the map
, we determine that the catalog is 80% complete at this flux level . the measured @xmath3 m fluxes are converted into total infrared luminosities ( l@xmath52 ) using the @xcite templates .
recent herschel studies indicate that while this technique is very accurate at @xmath53 , extrapolations from monochromatic @xmath3 m fluxes overestimate the true @xmath54 by factors of @xmath55 at @xmath56 ( e.g. * ? ? ?
finally , star formation rates are calculated from @xmath54 using the prescription of @xcite , adjusted to the chabrier imf .
by fitting seds to the 10-band photometry ( @xmath1m@xmath2 m ) , we are able to measure accurate rest - frame colors ( ab system ) and stellar masses as well as correct for dust extinction .
figure [ fig : col_smass ] shows the measured and dust - corrected rest - frame @xmath8 color versus stellar mass for the 92 cluster galaxies within @xmath37 mpc of the bcg .
as demonstrated by , e.g. @xcite , the galaxies begin to separate into the well - known bimodal distribution only when the colors are corrected for extinction .
the clg j0218.3 - 0510 members differ from their counterparts in clusters at @xmath57 ( e.g. * ? ? ?
* ; * ? ? ?
* ) in that they have a color - stellar mass distribution that is surprisingly similar to that observed in the field at @xmath11 @xcite : the clg j0218.3 - 0510members define a strong _ blue _ sequence and span a range in ( dust - corrected ) color ( fig .
2 , bottom panel ) . while these members populate a red sequence in the @xmath58 color - magnitude diagram ( pap10 ) , the correlation between rest - frame @xmath8 color and stellar mass is visibly weaker color because these filters correspond approximately to rest - frame @xmath59 at @xmath0 . ] . however , the most massive galaxies ( m@xmath60 m@xmath5 ) are still the reddest , i.e. they have the oldest stellar populations .
seventeen of the members are detected at @xmath3m(fig .
[ fig : col_smass ] ) , and the most ir luminous members include some of the most massive cluster galaxies ( fig . [ fig : sfr_smass ] ) . figure [ fig : sfr_smass ] shows that these ir luminous members follow the same trend of increasing star formation with stellar mass that is observed in the field at @xmath11 by @xcite .
note that because we can only detect members that are ir - bright ( l@xmath61 l@xmath5 ) , we are sensitive only to the upper envelope of the trend observed at @xmath11 .
three of the clg j0218.3 - 0510 members are ultra - luminous infrared galaxies ( ulirgs ; l@xmath62 l@xmath5 ; see fig .
[ fig : sfr_smass ] ) and the remaining 14 are lirgs ( @xmath63l@xmath64 l@xmath5 ) . in stark contrast , of the @xmath65 galaxies in clusters at @xmath16 studied with wide - field ( @xmath66 mpc ) mid - ir imaging ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , only one is a ulirg and it lies outside the core of the bullet cluster , a well - known cluster - cluster merger at @xmath67 @xcite .
the higher fraction of ir luminous galaxies at @xmath0 is likely to be driven by the overall evolution of the ir luminosity function in clusters ( e.g. * ? ? ?
* ) ; however , more deep ir imaging of clusters at @xmath20 is needed to confirm this trend . in our analysis
, we assume that the @xmath3 m sources are dominated by star formation and do not harbor active galactic nuclei ( agn ) . from the sed fits
, we find that two of the @xmath3 m detected members do have emission at @xmath68 m ( irac channel 4 ) that deviates from the stellar fit , i.e. have a power - law component indicative of an agn , and one of these galaxies is the most ir luminous member with @xmath69l@xmath70 [ l@xmath5]@xmath72 ( see fig .
[ fig : sfr_smass ] ) .
however , studies find that most of the emission ( @xmath73% ) in ulirgs is due to star formation @xcite , thus these members are likely to have both strong star formation and an agn component .
note that at @xmath0 , the @xmath74 m pah band lies partly in the @xmath3 m channel . because part of the ir luminosity is due to star formation
, we include both @xmath3 m members in our analysis ; repeating our analysis without these two members confirms that our overall results do not change . in the cluster core ( @xmath75 mpc ) ,
the star formation rate density from the @xmath3 m photometry alone is @xmath4 m@xmath5 yr@xmath6 mpc@xmath7 ; we stress that this is likely a lower limit given the @xmath3 m imaging can not detect any members with sfr@xmath76 m@xmath5 yr@xmath6 , i.e. with star formation rates typical for ir - detected galaxies in clusters at @xmath16 . only one other galaxy cluster at @xmath77 has a comparably high star formation rate in its core @xcite . in comparison
, studies of ir - detected galaxies in clusters at @xmath16 find that star - forming members are strongly segregated at @xmath78 mpc @xcite .
given the high star formation rate in its core , does clg j0218.3 - 0510 follow the well - established trend at @xmath16 of decreasing star - formation with increasing galaxy density ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
figure [ fig : sfr_density ] compares the relative fraction of star - forming members to passive members as a function of local galaxy density ( @xmath79 ) which is defined by distance to the 10th nearest neighbor @xcite .
we use star - formation rates derived from the @xmath3 m imaging as well as from the sed fitting because the two independent star formation tracers complement each other and provide an important check of our results : the @xmath3 m imaging is a robust measure of the dust - obscured star formation but only detects the most active members ( @xmath80 m@xmath5 yr@xmath6 ) while the sed fitting is sensitive to lower levels of unobscured activity ( @xmath815 m@xmath5 yr@xmath6 ) .
both star formation tracers confirm that _ the relative fraction of active members is highest in the regions of highest galaxy density _
[ fig : sfr_density ] ) , i.e. exactly opposite to that observed in clusters at @xmath16 .
a spearman rank test supports with @xmath82% confidence ( @xmath83 significance ) that the relative fraction of ir luminous members increases with increasing galaxy density from @xmath12% at @xmath13 gal mpc@xmath7 to @xmath14% at @xmath15 gal mpc@xmath7 .
we stress that excluding the two candidate agn does not change the trend , and the robustness of this result is underscored by the fact that we see the same trend using the sed - derived rates . while studies of galaxies in the field at @xmath18 ( @xmath84 gal mpc@xmath7 ) find that the star formation - density relation is beginning to turn over at this epoch @xcite , this is the first confirmation of such a reversal in the significantly higher density environment of galaxy clusters .
the measured ir luminosities correspond to specific star formation rates ( star formation rate divided by stellar mass ; ssfr ) per gyr of @xmath85 : these active members can more than double their stellar masses in the next gyr ( by @xmath86 ) .
however , to reproduce the relatively homogeneous stellar ages measured in massive cluster galaxies at @xmath16 ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , these ir luminous members can not maintain such a high ssfr for even a gyr .
the current star formation must be quenched rapidly and any later bursts of activity can not add a substantial amount of new stars , at least not in the massive members ( @xmath69m@xmath87 [ m@xmath5]@xmath88 ) that must populate a well - defined red sequence by @xmath89 .
we measure the rest - frame colors ( dust - corrected ) , ir luminosities , star formation rates , and stellar masses of galaxies in clg j0218.3 - 0510 , a spitzer - selected cluster at @xmath0 , by fitting spectral energy distributions to photometry in 10 bands ( @xmath1m@xmath2 m ) and with deep @xmath3 m imaging . the cluster sample ( pap10 ; * ? ? ?
* ) is composed of 12 spectroscopically confirmed members and 80 members selected from photometric redshifts measured using eazy @xcite ; all members are within @xmath90 mpc of the massive cluster galaxy located at the peak of the x - ray emission .
the 92 cluster members have a color - stellar mass distribution that is surprisingly similar to that observed in field galaxies at @xmath11 .
when corrected for dust , the cluster members define a strong blue sequence and span a range in color , indicating a substantial amount of recent and ongoing star formation in the cluster core . this dramatic level of activity
is underscored by the 17 members detected at @xmath3 m . in the cluster core ( @xmath91 mpc ) ,
the star formation rate density from the ir luminous members alone is @xmath4 m@xmath5 yr@xmath6 mpc@xmath7 ; the true value is likely to be higher given that we only include members with sfr@xmath92 m@xmath5 yr@xmath6 .
these ir luminous members also follow the same trend of increasing star formation with stellar mass that is observed in the field at @xmath11 .
we discover the striking result that the the relative fraction of star - forming galaxies increases with increasing local galaxy density in clg j0218.3 - 0510 , a reversal of the well - established trend at lower redshifts and in line with recent work at @xmath19 that suggests enhanced star formation in cluster cores .
measurements using star formation rates derived from the @xmath3 m imaging and from the sed fitting provide independent confirmation that the relative fraction of star - forming galaxies triples from the lowest to highest density regions . by pushing into the redshift desert ( @xmath93 ) , we are able to reach the epoch when massive cluster galaxies are still forming a significant number of their stars .
kt acknowledges generous support from the swiss national science foundation ( grant pp002 - 110576 ) .
jsd acknowledges the support of the royal society via a wolfson research merit award , and also the support of the european research council via the award of an advanced grant .
this work is based in part on data obtained as part of the ukirt infrared deep sky survey .
a portion of the magellan telescope time was granted by noao , through the telescope system instrumentation program ( tsip ; funded by nsf ) .
, p. l. , nichol , r. c. , miller , c. j. , balogh , m. l. , goto , t. , zabludoff , a. i. , romer , a. k. , bernardi , m. , sheth , r. , hopkins , a. m. , castander , f. j. , connolly , a. j. , schneider , d. p. , brinkmann , j. , lamb , d. q. , subbarao , m. , & york , d. g. 2003 , , 584 , 210 | we measure the rest - frame colors ( dust - corrected ) , infrared luminosities , star formation rates , and stellar masses of 92 galaxies in a spitzer - selected cluster at @xmath0 . by fitting spectral energy distributions ( seds ) to 10-band photometry ( @xmath1m@xmath2 m ) and measuring @xmath3 m fluxes for the 12 spectroscopically confirmed and 80 photometrically selected members
, we discover an exceptionally high level of star formation in the cluster core of @xmath4 m@xmath5 yr@xmath6 mpc@xmath7 .
the cluster galaxies define a strong blue sequence in @xmath8 color and span a range in color .
we identify 17 members with l@xmath9 l@xmath10 , and these ir luminous members follow the same trend of increasing star formation with stellar mass that is observed in the field at @xmath11 . using rates derived from both the @xmath3 m imaging and sed fitting
, we find that the relative fraction of star - forming members triples from the lowest to highest galaxy density regions , e.g. the ir luminous fraction increases from @xmath12% at @xmath13 gal mpc@xmath7 to @xmath14% at @xmath15 gal mpc@xmath7 . the observed increase is a reversal of the well - documented trend at @xmath16 and signals that we have reached the epoch when massive cluster galaxies are still forming a substantial fraction of their stars . |
the launch of the rossi x - ray timing explorer ( rxte ) has led to the discovery of kilohertz quasi - periodic oscillations ( khz qpos ) in neutron star ( ns ) low mass x - ray binaries ( lmxbs ) @xcite .
these high - frequency qpos usually appear in pairs ( upper @xmath5 and lower @xmath6 ) at frequencies from a few hundred hz to more than 1khz , as shown in atoll sources , z sources ( see @xcite for the definition of atoll and z sources ) and the accreting millisecond x - ray pulsars ( amxps ) ( see @xcite for a review ) .
it is suggested that khz qpos might be used to test the general relativity in a strong gravitational field regime @xcite , and help to constrain the ns @xmath7 relation @xcite .
the frequencies of the khz qpos are strongly correlated with other timing and spectral features , such as the photon indexes of the power - law component of the energy spectrum @xcite , the noise features @xcite , the positions in the x - ray color - color diagram ( e.g. , @xcite ) , the x - ray luminosities @xcite , the low frequency qpos @xcite , where the high-/low - frequency correlation is similar to those in black hole candidates and white dwarf cataclysmic variables ( e.g. @xcite ) . besides , the quality factors and @xmath8 amplitudes of the khz qpos can vary as a function of the qpo frequencies ( e.g. , @xcite ) .
there is currently no consensus as to the origin of these qpos . some interpretations , such as the beat frequency model by @xcite and the resonance model by @xcite , encounter difficulties when interpreting the following observational characteristics of the khz qpos @xcite .
while the relativistic precession model @xcite fits the @xmath9 relation to the khz qpo data quite well .
however , it requires the ns mass of 2 solar mass ( @xmath10 ) @xcite , which is relative higher than the actual measurement results @xcite .
the alfvn wave oscillation model ( awom ) by @xcite makes a good description of the khz qpo data @xcite .
the model predicts the relative emission positions of the khz qpos ( @xmath11 , the ratio between the ns radius and the khz qpo emission radius ) , which can help to infer the emission radii of khz qpos , then further infer the magnetosphere - accretion disk structure and evaluate the possibility of the khz qpo mechanism .
@lccccl@ source ( 25 ) & @xmath6 & @xmath5 & @xmath12 & @xmath13 & references + & ( hz ) & ( hz ) & ( @xmath14 ) & ( km ) & + amxp ( 2 ) + sax j1808.4 - 3658 & @xmath15 & @xmath16 & @xmath17 & & 1 + xte j1807.4 - 294 & @xmath18 & @xmath19 & @xmath20 & @xmath21 & 2 + atoll ( 15 ) + 4u 0614 + 09 & @xmath22 & @xmath23 & @xmath24 & @xmath25 & 3 + 4u 1608 - 52 & @xmath26 & @xmath27 & @xmath28 & @xmath29 & 4 + 4u 1636 - 53 & @xmath30 & @xmath31 & @xmath32 & @xmath33 & 5 + 4u 1702 - 43 & 722 & @xmath34 & @xmath35 & @xmath36 & 6 + 4u 1705 - 44 & @xmath37 & @xmath38 & @xmath39 & @xmath40 & 7 + 4u 1728 - 34 & @xmath41 & @xmath42 & @xmath43 & @xmath44 & 8 + 4u 1735 - 44 & @xmath45 & @xmath46 & @xmath47 & @xmath48 & 9 + 4u 1820 - 30 & @xmath49 & @xmath50 & @xmath51 & & 10 + 4u 1915 - 05 & @xmath52 & @xmath53 & @xmath54 & @xmath55 & 11 + aql x-1 & @xmath56 & @xmath57 & @xmath58 & @xmath59 & 12 + igr j17191 - 2821 & @xmath60 & @xmath61 & @xmath62 & @xmath63 & 13 + ks 1731 - 260 & @xmath64 & @xmath65 & @xmath66 & @xmath67 & 14 + sax j1750.8 - 2900 & @xmath68 & @xmath69 & @xmath70 & @xmath71 & 15 + xte j1701 - 407 & @xmath72 & @xmath73 & @xmath74 & @xmath75 & 16 + xte j2123 - 058 & @xmath76 & @xmath77 & @xmath78 & @xmath79 & 17 + z ( 8) + cir x-1 & @xmath80 & @xmath81 & @xmath82 & @xmath83 & 18 + cyg x-2 & @xmath84 & @xmath85 & @xmath86 & & 19 + gx 5 - 1 & @xmath87 & @xmath88 & @xmath89 & @xmath90 & 20 + gx 17 + 2 & @xmath91 & @xmath92 & @xmath93 & @xmath94 & 21 + gx 340 + 0 & @xmath95 & @xmath96 & @xmath97 & @xmath98 & 22 + gx 349 + 2 & @xmath99 & @xmath100 & @xmath39 & @xmath101 & 23 + sco x-1 & @xmath102 & @xmath103 & @xmath104 & @xmath105 & 24 + xte j1701 - 462 & @xmath106 & @xmath107 & @xmath108 & @xmath109 & 25 + [ data_all ] [ cols= " < " , ] in this paper we analyze the emission positions of khz qpos based on awom : in @xmath110 2 , we introduce the awom and show the khz qpo data used in the paper . in @xmath110 3
we infer the emission radii of khz qpos and analyze the results for the particular sources . in @xmath110
4 we make discussions and conclusions .
it is thought that khz qpos reflect the motion of matter in orbit at some preferred radius in the accretion disk around the neutron star in lmxbs @xcite . and
in order to consider the interaction between the accretion flow and the magnetic field of ns , we introduce the magnetosphere - disk radius @xmath111 , where the kinetic energy of the free - falling gas becomes comparable to the magnetic energy of the ns magnetosphere @xcite : @xmath112 where @xmath113 is a constant factor that is usually taken as 0.5 @xcite , @xmath114 is the alfvn radius , @xmath115 is the magnetic moment of ns , @xmath116 is the gravitational constant , @xmath117 is the ns mass and @xmath118 is the accretion rate at the inner disk boundary .
we assume the ns magnetic field to be dipolar : @xmath119 where @xmath120 is the ns surface dipole magnetic field strength , @xmath121 is the ns radius and @xmath122 is the radial distance refer to the center of the ns , and it can be seen that the magnetic field strength decreases with @xmath123 . then we adopt the accreting ns mass @xmath117 of 1.6@xmath10 based on its standard value of 1.4@xmath10 ( the ns mass in lmxbs is averagely increased by about 0.2@xmath10 on account of the accretion , see @xcite ) and assume the ns radius @xmath121 of 15 km , then the magnetosphere - disk radius can be written as : @xmath124 is the ns radius that is assumed to be 15 km , and @xmath122 is the radial emission radius of khz qpos with the range of @xmath125 km and @xmath126 km for atoll and z sources , respectively ( see table [ data_all ] and table [ data_mass ] for the details).,width=283 ] values .
the abscissa and ordinate represent the @xmath12 values and the cdf values respectively , and the arrows indicate the curves for amxps ( @xmath127 ) , atoll ( @xmath128 ) and z ( @xmath129 ) sources .
( b ) same as ( a ) , but for the emission radius @xmath122 of khz qpos ( the corresponding ranges of @xmath122 are @xmath130 km , @xmath131 km and @xmath132 km for amxps , atoll and z sources respectively ) , and the shaded area shows the radius range of 16 - 19 km , which contains most of the data of atoll sources ( @xmath133 ) and z sources ( @xmath134 ) ( see table [ data_all ] and table [ data_mass ] for the details ) . , title="fig:",width=283 ] values .
the abscissa and ordinate represent the @xmath12 values and the cdf values respectively , and the arrows indicate the curves for amxps ( @xmath127 ) , atoll ( @xmath128 ) and z ( @xmath129 ) sources .
( b ) same as ( a ) , but for the emission radius @xmath122 of khz qpos ( the corresponding ranges of @xmath122 are @xmath130 km , @xmath131 km and @xmath132 km for amxps , atoll and z sources respectively ) , and the shaded area shows the radius range of 16 - 19 km , which contains most of the data of atoll sources ( @xmath133 ) and z sources ( @xmath134 ) ( see table [ data_all ] and table [ data_mass ] for the details ) .
, title="fig:",width=283 ] in awom , the upper khz qpo frequency @xmath5 is assumed as the keplerian orbital frequency @xmath135 of the accretion flow at the preferred radius @xmath122 @xcite : @xmath136 while the lower khz qpo frequency @xmath6 is interpreted as the mhd alfvn wave oscillation frequency there .
the model predicts the following relations between the ns mass density , the emission position of the khz qpos and the twin khz qpo frequencies : @xmath137 @xmath138 @xmath139 @xmath140 @xmath141 where @xmath142 , @xmath143 is the parameter that is related to the ns mass density @xmath144 , @xmath117 is the ns mass , @xmath145 is the ns radius @xmath121 in unit of 10 km , @xmath12 is the position parameter and @xmath122 is the emission radius of the khz qpos refer to the center of the ns . according to awom , the ns mass density parameter @xmath143 and the position parameter @xmath12 can be inferred by the twin khz qpo frequencies .
+ + we search the literature for all published instances of khz qpo frequencies , and constrain the sample to the detection of two simultaneous khz qpo peaks .
the sample includes 416 pairs of twin khz qpos from 25 ns - lmxb sources , where the 23 pairs are taken from the accreting millisecond x - ray pulsars , and the 190/203 ones from atoll / z sources .
the adopted sources with the ranges of the lower and upper khz qpo frequencies and references are reported in table [ data_all ] .
in addition , three sources in the sample have been measured with the ns masses , which are shown in table [ data_mass ] .
we infer the emission radii of khz qpos based on awom . for the 22 sources without the measured ns masses ( see table [ data_all ] ) , we take the following calculation steps : 1 . for a certain pair of khz qpo frequencies ( @xmath6 and @xmath5 )
, we first calculate the value of the position parameter @xmath12 by solving equation ( [ nu_2_nu_1_2 ] ) : + @xmath146 .
then we infer the corresponding emission radius @xmath122 of the khz qpos with the @xmath12 value by equation ( [ x ] ) : @xmath147 , where the ns radius @xmath121 is assumed to be 15 km .
while for the three sources with the measured ns masses ( cyg x-2 , sax j1808.4 - 3658 and xte 1820 - 30 , see table [ data_mass ] ) , we first take the following steps to infer the ns radius @xmath121 based on awom : 1 . for each source
, we fit the equation ( [ nu_2_nu_1_1 ] ) : + @xmath148 + to the twin khz qpo frequencies and obtain the value of the ns density parameter @xmath143 .
then we infer the ns radius @xmath121 of this source with the @xmath143 value and the measured ns mass @xmath117 by solving equation ( [ a ] ) : @xmath149 .
next we take the following steps to infer the emission radii of khz qpos for the three sources : 1 . for a certain pair of khz qpo frequencies ,
we calculate the @xmath12 value with @xmath6 and @xmath5 by solving equation ( [ nu_2_nu_1_2 ] ) : @xmath146 .
then we infer the corresponding emission radius @xmath122 of the khz qpos with the @xmath12 value and the inferred ns radius @xmath121 by equation ( [ x ] ) : @xmath147 .
the ranges of the inferred @xmath12 and @xmath122 values for the sources without and with the measured ns masses are shown in table [ data_all ] and table [ data_mass ] , respectively .
there are some aspects should be noticed : 1 .
the emission radii of khz qpos of amxps , atoll and z sources are distributed in the range of @xmath150 km , @xmath125 km and @xmath126 km , respectively .
fig.[ns_accretion ] shows the schematic diagram of the radial ranges of the emission positions of khz qpos for atoll and z sources .
it is evidence that z sources have the larger range of the emission radii of khz qpos than atoll sources , which may result from source cir x-1 with the large emission radii of @xmath151 km ( see also fig.[r_z ] ) .
we show the cumulative distribution function ( cdf ) curves of the position parameter @xmath12 and the emission radius @xmath122 of khz qpos for amxps ( @xmath127 , @xmath152 km ) , atoll ( @xmath128 , @xmath153 km ) and z ( @xmath129 , @xmath154 km ) sources in fig.[r_atoll_z_msp ] .
it can be seen that most of the emission radii of khz qpos of atoll sources ( @xmath133 of the data ) and z sources ( @xmath134 of the data ) cluster around the region of @xmath0 km .
xte j1807.4 - 294 ( @xmath152 km ) , 4u 0614 + 09 ( @xmath155 km ) , gx 5 - 1 ( @xmath156 km ) and gx 340 + 0 ( @xmath157 km ) show the larger range of the emission radii of khz qpos from @xmath158 17 km to @xmath158 30 km ( see table [ data_all ] for the details ) than the other sources , and their corresponding cdf curves of the emission radii are shown in fig.[r_z ] .
we also analyze the innermost emission positions of khz qpos for all the sources in the sample .
fig.[x_r ] shows the cdf curve of the minimal emission radii ( @xmath159 km ) of all the 25 sources , from which it can be seen that 22 sources share the minimal emission radii around the range of @xmath0 km , while the other three sources , i.e. cir x-1 , cyg x-2 and sax j1808.4 - 3658 , show the larger ones of about 23 , 21 and 20 km , respectively .
km ) of all the 25 sources in table [ data_all ] and table [ data_mass ] .
the dashed line shows the position of the emission radius of 19 km and the arrows indicate the innermost emission positions of source cir x-1 ( @xmath160 km ) , cyg x-2 ( @xmath161 km ) and sax j1808.4 - 3658 ( @xmath162 km ) ( see table [ data_all ] and table [ data_mass ] for the details ) .
, width=283 ] we investigate the emission radii of khz qpos for the accreting millisecond x - ray pulsars , atoll and z sources based on awom , and find that most of the emission positions of khz qpos cluster around the region of several kilometers away from the ns surface .
the details of the conclusions are discussed and summarized as follows : 1 .
the inferred emission radii of khz qpos of amxps , atoll and z sources based on awom are distributed in the range of @xmath150 km , @xmath125 km and @xmath126 km respectively ( see table [ data_all ] , table [ data_mass ] and fig.[ns_accretion ] for the details ) .
cir x-1 has the obviously larger emission radii of @xmath151 km than the other sources .
it is thought that khz qpos reflect the motion of matter in orbit at some preferred radius @xmath122 in the accretion disk @xcite , where the magnetosphere - disk radius there @xmath163 satisfies the equation ( [ r_m_ns ] ) : + @xmath164 .
+ we suggest that cir x-1 may have the large ns surface magnetic field @xmath120 , which will cause the large magnetosphere - disk radius , i.e. the large ones of the emission radii of khz qpos .
it is not clear what physical parameters determine the appearance of the detectable khz qpos , nor why they only occur in the certain range of @xmath165 km away from the ns surface , and we suspect that it may be related to the accretion environment of lmxbs , which need the further study of the accretion state of the systems .
2 . xte j1807.4 - 294 , 4u 0614 + 09 , gx 5 - 1 and gx 340 + 0 show the larger range of the emission radii of khz qpos from @xmath158 17 km to @xmath158 30 km ( see table [ data_all ] and fig.[r_z ] ) , implying these sources may have the more suitable physical environment to produce khz qpos .
less bright source xte j1807.4 - 294 ( @xmath166 km ) and 4u 0614 + 09 ( @xmath167 km ) have the similar large range of the emission radii with the more bright source gx 5 - 1 ( @xmath168 km ) and gx 340 + 0 ( @xmath169 km ) , which implies the emission condition of khz qpos may not sensitive to the mean accretion rate .
it can be seen from see table [ data_all ] , table [ data_mass ] and fig.[r_atoll_z_msp ] that the emission radius @xmath122 of khz qpos for all the sources are larger than the ns radius @xmath121 ( @xmath170 km and @xmath121 is assumed to be 15 km , i.e. @xmath171 ) , which means that the accretion matter that is related to the khz qpos is at the position at least one kilometer away from the ns surface .
there may be some interpretations for these phenomena : one is the effect of the local strong magnetic spot @xcite which expels the magnetosphere - disk not to approach the ns surface , so the accretion matter may at the position away from the ns surface .
another explanation is that the accretion plasma drops onto the stellar surface due to the instability when crossing the innermost stable circular orbit ( isco ) of ns , and the isco radius ( @xmath172 ) is the inner boundary of the accretion matter , so the emission radius of the khz qpos should be larger than @xmath173 .
we try to test the isco effect scenario : three sources in the sample , i.e. cyg x-2 , sax j1808.4 - 3658 and xte 1820 - 30 , have both the detected twin khz qpo frequencies and the measured ns masses ( see table [ data_mass ] ) , which can be used to calculate both the ns radii and isco radii with the following equations : + + @xmath148 , + @xmath149 , + @xmath172 .
+ the results are shown in table [ data_mass ] , i.e. for cyg x-2 , @xmath174 km and @xmath175 km , for sax j1808.4 - 3658 , @xmath176 km and @xmath177 km , for xte 1820 - 30 , @xmath178 km and @xmath179 km .
it can be seen that the isco radii of the three sources are all smaller than their inferred ns radii , so we suggest that the isco effect may not be the main reason to cause the emission radii of the khz qpos larger than the ns radii .
it can be seen from table [ data_all ] , table [ data_mass ] and fig.[r_atoll_z_msp ] that most atoll and z sources ( 19/23 ) are inferred to share the similar emission radii of khz qpos of @xmath0 km .
we try to probe the ns magnitude field by the khz qpos : the luminosities of atoll sources are less than those of z sources in 2 - 3 magnitude orders in general , thus we assume that the accretion rates of atoll sources are about 2 magnitude orders lower than those of z sources in average .
the magnetosphere - disk radius @xmath111 where the khz qpos occur satisfy the equation ( [ r_m_ns ] ) : + @xmath164 , + the similar @xmath180 km ) and the different @xmath118 will infer the ns surface magnetic field strengths of atoll sources are about one magnitude order less than those of z sources , which is consistent with the prediction of @xcite .
source xte j1701 - 462 has been observed both atoll and z behaviors @xcite , which can go from one state to the other one in timescale of months , and @xcite suggest this source has a ns surface magnetic field strength should not change in that timescale .
we argue that xte j1701 - 462 has a steady ns surface magnetic field and should not change in the short timescale , and ascribe the change of the atoll and z behaviors to the different magnetic field strengths at the different magnetosphere - disk radii : by assuming the ns magnetic field to be dipolar ( see equation ( [ ns_b ] ) ) : + @xmath181 , + when the source has a instantaneously high accretion rate , the accretion disk will go closer to the ns surface and shows a smaller @xmath111 , where the magnetic field there will be higher , then the source shows the z source behavior .
on the contract , the system will show the atoll source behavior with the lower magnetic field at the magnetosphere - disk radius when it has a instantaneously low accretion rate .
so we suggest that the atoll and z states of the source may reflect the magnetic field information at the different magnetosphere - disk radii .
cir x-1 ( @xmath160 km ) , cyg x-2 ( @xmath161 km ) and sax j1808.4 - 3658 ( @xmath162 km ) show the large values of the minimal emission radii of khz qpos compared with the other sources in the sample ( see also table [ data_all ] , table [ data_mass ] and fig.[x_r ] ) . as for cir x-1 and cyg x-2
, we try to ascribe their larger ones of the minimal emission radii to share the stronger ns surface magnetic fields that make the averaged magnetosphere - disk radii to be larger . for sax j1808.4 - 3658
, its inferred ns radius is slightly large ( @xmath182 km with the ns mass of @xmath183 , see table [ data_mass ] ) , so its emission radii of khz qpos are relative large although it has the similar emission parameter @xmath12 values to the other sources ( @xmath147 ) .
we thank j. wang , z.b . li and h.x .
yin for helpful discussions .
this work is supported by the national basic research program of china ( 2012cb821800 and 2009cb824800 ) , the national natural science foundation of china nsfc(11173034 , 11173024 , 11303047 ) , the science and technology foundation of guizhou province ( grant no.j[2015]2113 ) , the doctoral starting up foundation of guizhou normal university 2014 and the innovation team foundation of the education department of guizhou province under grant nos .
[ 2014]35 .
van straaten s. et al . , 2000 , apj , 540 , 1049 van straaten s. et al . , 2002 , apj , 568 , 912 van straaten s. , van der klis m. , mndez m. , 2003 , apj , 596 , 1155 van straaten s. , van der klis m. , wijnands r. , 2005 , apj , 619 , 455 wang d.h .
et al . , 2013 ,
mnras , 435 , 3494 | we infer the emission positions of twin kilohertz quasi - periodic oscillations ( khz qpos ) in neutron star low mass x - ray binaries ( ns - lmxbs ) based on the alfvn wave oscillation model ( awom ) . for most sources , the emission radii of khz qpos cluster around a region of @xmath0 km with the assumed ns radii of 15 km .
cir x-1 has the larger emission radii of @xmath1 km than those of the other sources , which may be ascribed to its large magnetosphere - disk radius or strong ns surface magnetic field .
sax j1808.4 - 3658 is also a particular source with the relative large emission radii of khz qpos of @xmath2 km , which may be due to its large inferred ns radius of @xmath3 km .
the emission radii of khz qpos for all the sources are larger than the ns radii , and the possible explanations of which are presented .
the similarity of the emission radii of khz qpos ( @xmath4 km ) for both the low / high luminosity atoll / z sources is found , which indicates that both sources share the similar magnetosphere - disk radii .
[ firstpage ] stars : neutron binaries : close x - rays : binaries accretion : accretion disks |
the authors thank sergi elizalde for suggesting this line of research and for several useful conversations along the way . thanks also to julie landau for her hospitality and kick serve .
in this section we introduce some language and notation which will be used throughout the paper , and we state our first result , theorem [ cantor ] , which says that if no restriction is placed on @xmath3 , then one can always find @xmath3 realizing a given compatible sequence of permutation distributions . for a positive integer @xmath1
we denote @xmath41 by @xmath42 $ ] and the group of bijections of @xmath42 $ ] by @xmath15 .
let @xmath43 be an injective map from a finite totally ordered set @xmath44 ( where @xmath45 ) to a totally ordered set @xmath46 .
let @xmath47 .
we define the _ order pattern _ @xmath48 to be the unique permutation @xmath12 satisfying @xmath13 if and only if @xmath14 . equivalently @xmath49 .
note that if @xmath12 then @xmath50 . the _ order pattern _ of an @xmath1-tuple of distinct real numbers @xmath51 is @xmath52 where @xmath53\to\r$ ] takes @xmath54 to @xmath55 .
there is a restriction map @xmath56 given by @xmath57})$ ] . using this we define @xmath58 as @xmath59 , which is equal to the inverse limit of the maps @xmath56 .
let @xmath60 .
the set @xmath61 is graded by @xmath1 and we use notation like @xmath62 to mean @xmath63 . next , let @xmath64 be the space of probability distributions on @xmath15 . note that @xmath64 is the standard simplex in @xmath65 .
we denote by @xmath66 the vertex of @xmath64 which has mass 1 at @xmath12 and 0 elsewhere . if @xmath67 and @xmath68 we say @xmath35 and @xmath69 are _ compatible _ if @xmath70 .
then @xmath71 , and @xmath72 .
as an example , the uniform distributions from each @xmath64 form a compatible sequence , hence an element of @xmath73 . for @xmath27 and @xmath74
let @xmath75 where @xmath76 let @xmath77 be the map taking a function @xmath27 to the distribution defined by @xmath78 note that for any @xmath3 and @xmath1 , the distributions @xmath6 and @xmath79 are compatible ; thus we may define @xmath80 by @xmath81 .
we can now state our first result .
[ cantor ] for every @xmath82 there exists a function @xmath27 with @xmath83 .
that is , @xmath84 .
[ lem : cantor ] the proof is constructive , a little involved , and unnecessary for the results that follow .
therefore we defer the proof to section [ cantorproof ] .
before we end this section we make an observation about convexity .
suppose @xmath85 is a collection of functions such that whenever @xmath86 and @xmath87 $ ] , the function @xmath88 is also in @xmath89 .
then @xmath90 is a convex subset of @xmath64 .
this is because @xmath91 is the block sum " of @xmath3 scaled by @xmath92 and @xmath43 scaled by @xmath93 , and so for all @xmath1 , @xmath94 .
this will usually hold if @xmath89 has `` piecewise '' in the title , such as piecewise continuous functions , piecewise monotone functions , etc .
it also holds for ( aperiodic ) measure preserving functions .
the next several sections develop the language used in the remainder of the paper .
we begin with digraphs .
a _ digraph _ is a quadruple @xmath95 with @xmath96 the vertex set , @xmath97 the edge set , and @xmath91 and @xmath92 the head and tail maps from @xmath97 to @xmath96 .
recall that @xmath56 is defined by @xmath57})$ ] .
similarly define @xmath98 by @xmath99})$ ] .
for @xmath100 let @xmath101 denote the _ permutation digraph _ @xmath102 .
the digraphs @xmath103 and @xmath104 are shown in figure [ fig : digraphs ] . and @xmath104 .
the edges of @xmath103 are shown with labels ; the edges of @xmath104 are abbreviated and the labels omitted .
for instance two directed edges go from @xmath105 to @xmath106 , with labels @xmath107 and @xmath108 .
an edge labeled @xmath109 goes in the reverse direction.,width=432 ] a _ path _ of length @xmath110 ( where @xmath111 ) in a digraph is an alternating sequence @xmath112 with @xmath113 and @xmath114 such that @xmath115 and @xmath116 .
a path of length @xmath117 is @xmath118 such that each finite initial segment ending with a vertex is a ( finite ) path .
we write @xmath119 for the set of all paths of length @xmath110 in @xmath120 and @xmath121 for the set of all paths in @xmath120 .
note that @xmath122 . to define specific paths we sometimes abuse notation slightly by thinking of @xmath123 and @xmath124 as functions from @xmath125 to @xmath96 and @xmath97 .
for example , if @xmath126 is a path of finite length @xmath110 and @xmath127 is any path with @xmath128 then the _ concatenation _ @xmath129 of @xmath126 and @xmath127 has @xmath130 and @xmath131 for @xmath132 and @xmath133 and @xmath134 for @xmath135 .
a digraph is _ strongly connected _ if there are paths connecting every ordered pair of vertices .
a ( finite ) path is _ embedded _ if all @xmath136 vertices are distinct , except possibly @xmath137 .
a _ loop _ is a finite path with @xmath137 . for each @xmath138
we define @xmath139 as follows .
first , @xmath140 is the identity on @xmath141 . if @xmath142 , let @xmath143 be the path @xmath144 . if @xmath145 then let @xmath146 be the concatenation @xmath147 .
( the result is an infinite path if @xmath148 ; otherwise the result is a path of length @xmath136 . )
thus @xmath149 .
let @xmath150 and let @xmath140 be the union of the functions @xmath151 on @xmath152 .
finally , extend @xmath140 further by defining @xmath153 for @xmath154 to be the infinite path whose initial subpath of length @xmath110 is equal to @xmath155 .
note : if @xmath156 then @xmath157 .
the next lemma says that any path ( of length @xmath158 ) on @xmath101 can be lifted to @xmath159 ( where it becomes shorter if its length is finite ) .
note however that the edges of the lift are not determined ; only the vertices are determined , because the edges of @xmath126 do not appear in the definition of @xmath146 .
the ambiguity in the lifting process will play an important role later .
the map @xmath151 is surjective .
the image of @xmath160 is @xmath161 .
for the first part , it suffices to show that @xmath162 is surjective , as @xmath151 is a composition of maps of this form . if @xmath163 then each edge of @xmath126 is a vertex of @xmath159 .
we only need to show that if @xmath164 with @xmath165 then there is an edge @xmath166 with @xmath167 and @xmath168 .
extend the function @xmath169\to[n+1]$ ] to @xmath170\to\r$ ] by defining @xmath171 such that @xmath172})$ ] .
then @xmath173 is the desired edge . for the second part , if @xmath174 then set @xmath175 and for each @xmath176 let @xmath177 be a lift of @xmath178 .
then for @xmath179 let @xmath180 be the initial vertex of @xmath181 , and note that @xmath182 .
consider the infinite path @xmath183 that begins at the vertex @xmath184 and traverses the edges @xmath36 followed by @xmath185 repeatedly .
then @xmath126 projects to the path @xmath186 which traverses the loop labeled @xmath184 and then the loop labeled @xmath187 and then repeats .
there are infinitely many paths other than @xmath126 in @xmath188 , since the vertices must alternate between @xmath184 and @xmath187 but there are two choices for each edge .
by contrast , at the next step , @xmath189 is the singleton consisting of the infinite path on @xmath190 that starts at the vertex @xmath185 and traverses the edges @xmath191 and @xmath192 repeatedly .
in fact @xmath193 is already a singleton , being the compatible sequence @xmath194 where @xmath195 is the permutation @xmath196 .
given a path @xmath197 on @xmath101 , consider the set @xmath198\times [ n]\right ) \cup \left(\{e\}\times [ 1,\ell]\times [ n+1]\right).\ ] ] this set is ( in 1 - 1 correspondence with ) the disjoint union of the domains of all the permutations @xmath123 and @xmath124 .
they are patched together " by the equivalence @xmath199 generated by 1 .
@xmath200 2 .
. the equivalence class of @xmath202 in @xmath203 will be denoted by @xmath204 or @xmath205 if the path @xmath126 is understood ; note that ( a ) this is well - defined and ( b ) every element of @xmath206 is equal to @xmath55 for some @xmath207 . by ( a ) , if @xmath208 for @xmath209 then @xmath210 , and so @xmath211 .
the set @xmath206 is easy to visualize , but let us first define a partial ordering on it .
consider the relation @xmath212 on @xmath206 generated by 1 .
@xmath213\leq [ ( v , a , d)]$ ] if @xmath214 2 .
@xmath215\leq [ ( e , a , d)]$ ] if @xmath216 and extended by transitivity .
we will show in a moment that @xmath212 is a partial ordering on @xmath206 .
the point of @xmath206 is to keep track of all order relationships which necessarily hold among @xmath217 , if @xmath0 is a permutation in @xmath218 .
consider the path @xmath126 of length 5 in @xmath104 with edges @xmath219 .
this is a loop based at @xmath220 .
attempts to construct real numbers @xmath221 such that @xmath222 quickly lead one to draw pictures like figure [ fig : worms ] .
the top picture is a plot of the desired @xmath223 s .
note that @xmath224 could be perturbed to be larger or smaller than @xmath225 , and similarly for @xmath226 and @xmath227 .
the dotted lines indicate the duration of the influence of @xmath228 on future @xmath229 s . this information is abstracted in the middle picture ,
in which the dots are the elements of @xmath230 and equivalent elements are joined by an arc .
each arc is an element of @xmath206 .
the bottom picture shows the partial ordering : an edge pointing from @xmath55 to @xmath231 indicates that @xmath232 . is a loop of length 5 in @xmath104 traversing in this order the edges @xmath233 .
figure ( a ) is a plot of a sequence @xmath234 ( for @xmath235 ) which maps to @xmath126 under @xmath236 .
the elements of @xmath237 , shown in ( b ) as dots , are the intersections of the above plot with vertical lines at @xmath238 .
the elements @xmath239 , @xmath240 , @xmath241 of @xmath237 make up the leftmost column of dots ( read from top to bottom ) .
equivalent elements are joined by an arc .
figure ( c ) depicts the poset @xmath206 whose elements are the arcs in either of the previous pictures . ] [ liftleq ] let @xmath126 be a path on @xmath101 , and let @xmath242 . if @xmath243 then @xmath244 .
we may assume the length @xmath110 is not zero .
it suffices to prove for @xmath245 , as we can lift multiple times .
the hypothesis @xmath243 implies there is a sequence @xmath246 in @xmath247 , where each step is one of the types ( i)-(iv ) . in this sequence , if one of the inequalities has @xmath248 then @xmath249 and using ( i ) and ( ii ) we can replace @xmath228 and @xmath250 by @xmath251 and either increase @xmath252 and @xmath253 by 1 ( if @xmath254 ) or increase @xmath255 and @xmath256 by 1 ( if @xmath257 . since @xmath258 and @xmath259 , the inequality is preserved in either case .
thus we obtain a new sequence with each @xmath260 .
now note that the elements @xmath261 of @xmath237 are in one - one correspondence with the elements @xmath202 of @xmath262 .
thus if we now switch every @xmath251 to a @xmath263 and subtract 1 from each @xmath264 , we obtain a sequence in @xmath262 showing @xmath244 . the relation @xmath212 is a partial order on @xmath206 . the relation is reflexive and transitive by definition .
we must show that if @xmath232 and @xmath265 then @xmath210 .
lift @xmath126 to a path @xmath266 ( where @xmath267 ) .
there are no equivalences in @xmath268 , and @xmath232 in the poset @xmath269 if and only if @xmath270 .
by lemma [ liftleq ] , @xmath232 and @xmath265 in @xmath269 , so @xmath210 .
[ totalorder ] note that for any @xmath271 , the elements @xmath272 are totally ordered in @xmath206 .
let @xmath273 and @xmath274 .
if @xmath232 in @xmath206 then there is a sequence with each @xmath275 .
call such a sequence _ monotonic _ if the function @xmath276 is monotonic in @xmath277 .
[ monotonic ] if @xmath232 in @xmath206 then there is a monotonic sequence of the form .
choose a sequence of the form ; one exists by definition .
note that only rules ( 1 ) and ( 2 ) change @xmath278 , and that @xmath278 increases ( by @xmath279 ) if either of these rules is applied by replacing the left side with the right . suppose the given sequence is not monotonic .
specifically suppose that @xmath278 increases and later decreases ; the other case is virtually identical .
choosing an innermost such backtrack , we find a subsequence of one of the following two forms : 1 .
@xmath280 2 .
@xmath281 in case ( i ) , we have @xmath282 .
but @xmath283 so we also have @xmath214
. thus we can delete the middle two terms of ( i ) and eliminate the backtracking .
case ( ii ) is similar : @xmath214 but this time @xmath284 .
now it follows that @xmath285 so again we can eliminate the backtracking . referring again to figure [ fig : worms ] , lemma [ monotonic ] says that it is very easy to determine whether @xmath232 . if @xmath286 , one just sees whether it is possible to connect the right endpoint of @xmath55 to any point above @xmath231 with a path that passes the vertical line test . if not , then @xmath232 .
[ concat ] let @xmath287 be paths on @xmath101 of lengths @xmath288 such that the concatenation @xmath129 is defined . for @xmath289 , if @xmath290 then @xmath291 .
choose a monotone sequence of the form , and add @xmath110 to the second coordinate of each term .
the new sequence proves the result .
let @xmath126 be a path of length @xmath110 on @xmath101 , and let @xmath292 .
a choice of lift @xmath293 amounts to a choice of extension of @xmath212 to a total order on @xmath294 .
that this can be done is well - known ; the process is sometimes called a topological sort . " in particular , for a subset @xmath295 $ ] of indices , if the @xmath296 are pairwise incomparable in @xmath206 then for any permutation @xmath297 there is an extension of @xmath212 to a total order on @xmath294 satisfying @xmath298 . in the terminology of lifts
this becomes the following statement , which bears on the discussion of entropy in a later section .
[ factorialgrowth ] let @xmath126 be a path of length @xmath110 on @xmath101 , and let @xmath267 .
if the elements @xmath299 of @xmath206 are pairwise incomparable , then for any permutation @xmath297 there is a lift @xmath300 such that @xmath301 in particular @xmath302 as a special case of this we also note the following . [ driftlesscoord ] let @xmath126 be a path of length @xmath110 on @xmath101 .
then @xmath232 if and only if @xmath270 for every @xmath303 ( where necessarily @xmath304 .
if @xmath305 is a loop of length @xmath306 on @xmath101 , then the elements @xmath307 of @xmath206 are totally ordered , as are the elements @xmath308 , and if we set @xmath309 , then we have @xmath232 if and only if @xmath310 , for all @xmath311 $ ] . the notion of drift
is based on how the @xmath55 compare to the @xmath229 , as measured by the following two functions .
let @xmath312 be the totally ordered set @xmath42\cup \{-\infty,\infty\}$ ] ( with @xmath313 and @xmath138 ) , and for @xmath314 define @xmath315 , x_i\leq y_k \mbox { implies } y_j\leq y_k \\
\infty&\mbox{if } i=\infty \mbox { or } i\in[n ] \mbox { and there is no $ j$ such that } x_i\leq y_j \\ -\infty&\mbox{if } i=-\infty \end{cases } \\
\operatorname{min}_{\gamma}(i)&= \begin{cases } j & \mbox{if } x_i\geq y_j , \mbox { and for } k\in[n ] , x_i\geq y_k \mbox { implies } y_j\geq y_k\\ -\infty&\mbox{if } i=-\infty \mbox { or } i\in[n ] \mbox { and there is no $ j$ such that } x_i\geq y_j \\ \infty&\mbox{if } i=\infty \end{cases}\end{aligned}\ ] ] [ maxminordpres ] if @xmath232 then @xmath316 and @xmath317 .
this is immediate from the definitions , and from the fact that @xmath232 if and only if @xmath310 .
[ funct ] suppose @xmath126 and @xmath127 are finite paths such that @xmath129 is a path
. then @xmath318 and @xmath319 .
we give the verification for @xmath320 . for @xmath321 ,
flip the argument upside down .
let @xmath110 be the length of @xmath126 and @xmath322 the length of @xmath127 .
it is clear that the two functions are equal on @xmath323 .
let @xmath324 $ ] , let @xmath325 . if @xmath326 then by lemma [ monotonic ] it is impossible to have @xmath327 for any @xmath328 $ ] , so @xmath329 .
we may therefore assume @xmath330 , and let @xmath331 .
if @xmath332 , then again by lemma [ monotonic ] it is impossible to have @xmath333 for any @xmath334 $ ] , so @xmath329 .
thus we may assume @xmath335 .
we want to show that @xmath336 . in the poset @xmath337
, we have @xmath338 .
also , if @xmath333 then there is a monotonic sequence in @xmath339 showing @xmath333 .
this sequence must contain a point of the form @xmath340 , so @xmath341 . by definition of @xmath342
, we have @xmath343 , hence @xmath344 . by definition of @xmath277
we now have @xmath345 .
thus @xmath346 , as desired .
let @xmath305 be a loop of length @xmath110 on @xmath101 .
for @xmath311 $ ] let @xmath347 we will write @xmath348 for @xmath349 .
[ drift ] a loop @xmath305 is _ partially driftless _ if @xmath350 for some @xmath324 $ ] .
+ a loop @xmath305 is _ driftless _
if @xmath351 for all @xmath324 $ ] .
+ a loop @xmath305 is _ totally driftless _
if @xmath352 for all @xmath311 $ ] .
thus @xmath305 is totally driftless if and only if @xmath353 and @xmath354 for all @xmath324 $ ] , and there is a similar description of driftless and partially driftless loops .
the loop @xmath126 in figure [ fig : worms ] is partially driftless . in @xmath206
, we have @xmath355 and @xmath356 , so @xmath357 .
however @xmath358 and @xmath359 are incomparable , so @xmath360 .
note that the number @xmath361 is necessarily greater than @xmath362 , but @xmath363 can be chosen to be greater than or less than @xmath364 .
[ partial ] let @xmath365 and @xmath305 be ( partially driftless ) loops based at @xmath263 , with @xmath366 . then @xmath367 .
let @xmath110 be the length of @xmath365 .
suppose @xmath368 .
then there is a monotonic sequence proving this . in the sequence
there must be a representative of @xmath55 for some @xmath369 .
now @xmath370 or @xmath371 would contradict @xmath372 , and @xmath373 would contradict @xmath374 . by remark [ totalorder ]
these are the only possibilities .
thus @xmath375 .
similarly @xmath376 . [ driftsuck ]
let @xmath365 and @xmath305 be loops on @xmath101 based at the vertex @xmath263 .
if @xmath305 is totally driftless then @xmath377 is totally driftless .
suppose not ; then without loss of generality there exists @xmath324 $ ] with @xmath378 .
thus @xmath379 and by lemma [ monotonic ] there is a monotonic sequence proving this inequality .
this sequence must contain @xmath380 for some @xmath277 , where @xmath110 is the length of @xmath365 .
starting there , the remainder of the sequence ( in combination with lemma [ concat ] ) shows that @xmath381 , a contradiction .
[ cyclic ] cyclic permutations of driftless loops are driftless .
let @xmath382 be a driftless loop and let @xmath383 be a cyclic permutation of @xmath305 starting at @xmath384 .
suppose @xmath385 .
fix a monotonic sequence showing this inequality , and add @xmath110 to the second coordinate of each element to obtain a new sequence , and concatenate the original sequence with the new one .
this longer sequence shows @xmath386 but it contains a subsequence showing for some @xmath342 that @xmath387 .
a _ face subgraph _ of @xmath101 is a subgraph @xmath388 such that every edge of @xmath388 is contained in a loop in @xmath388 .
equivalently @xmath388 is a face subgraph if each connected component of @xmath388 is strongly connected .
a strongly connected subgraph @xmath389 _ drifts _ if there exist @xmath390 , @xmath391 $ ] and @xmath392 such that for every loop @xmath305 in @xmath388 based at @xmath263 , @xmath393 .
otherwise @xmath388 is _ driftless_. a face subgraph @xmath389 _ drifts _ if any of its connected components drifts ; otherwise @xmath388 is driftless . [ driftlessloop ] let @xmath388 be a strongly connected subgraph of @xmath101 .
the following are equivalent : 1 .
@xmath388 is driftless ; 2 .
there exists a totally driftless loop @xmath305 with support contained in @xmath388 ; 3 .
there exists a totally driftless loop @xmath305 with support equal to @xmath388 .
the last two statements are equivalent by lemma [ driftsuck ] : if @xmath305 is a totally driftless loop with support contained in @xmath388 , and @xmath365 is any loop with support equal to @xmath388 , then @xmath377 is a totally driftless loop with support equal to @xmath388 .
statement ( 3 ) easily implies statement ( 1 ) : for fixed @xmath263 , @xmath342 , @xmath394 let @xmath395 be a cyclic permutation of @xmath305 which starts at @xmath263 . by lemma [ cyclic ] @xmath396 .
last , we show ( 1 ) implies ( 2 ) .
let @xmath397 be a loop based at @xmath0 and supported in @xmath388 .
let @xmath398 and @xmath399 , so that @xmath400 for all @xmath328 $ ] .
suppose @xmath401 . as @xmath388 is driftless
, we may pick a loop @xmath402 based at @xmath263 and supported in @xmath388 such that @xmath403 .
. we can continue this process until we have a loop @xmath405 with @xmath406 . then
, similarly , we concatenate loops on to the end of @xmath405 to create a loop @xmath365 with @xmath407 .
note that @xmath408 ( by lemma [ funct ] ) .
now , by lemma [ maxminordpres ] @xmath365 is totally driftless , with support contained in @xmath388 .
[ enlarge ] if @xmath409 and @xmath388 are strongly connected , @xmath410 , and @xmath409 is driftless , then @xmath388 is driftless .
in this section we analyze the distributions of order patterns arising from ( almost aperiodic ) measure preserving functions @xmath411 our main theorem is that the image @xmath32 is a union of open faces of a polytope @xmath412 of dimension @xmath413 , and that there is an easily checkable combinatorial criterion for determining whether a particular face of @xmath33 is in the image .
[ othermeasures ] for most of these results it is not essential that @xmath414 be the measure preserved by @xmath3 . that is , given a function @xmath27 one could choose an invariant measure @xmath415 and proceed with this section , everywhere replacing @xmath414 with @xmath415 .
for some steps it may be necessary to assume @xmath415 has no atoms .
we start by observing that theorem [ cantor ] would not hold if @xmath416 were replaced by @xmath417 .
if @xmath12 let @xmath418 denote the distribution whose value is 1 on @xmath0 and 0 elsewhere .
[ balayage ] if @xmath419 has positive measure and @xmath420 is aperiodic and measure preserving then both @xmath421 and @xmath422 have positive measure . in particular , there is no @xmath34 such that @xmath423 or @xmath424 .
suppose @xmath425 , i.e. , @xmath426 for almost all @xmath427 .
then there is some @xmath394 such that @xmath428 , hence @xmath429 .
but @xmath3 measure preserving implies @xmath430 .
similarly for @xmath431 .
note that @xmath432 and @xmath424 are in the closure of @xmath433 since @xmath434 can be made arbitrarily close to these distributions by choosing @xmath435 .
lemma [ balayage ] notwithstanding , there is a much more serious reason for the failure of theorem [ cantor ] in the measure preserving category . for @xmath34
, there is an additional set of constraints on @xmath436 beyond compatibility of the measures @xmath437 .
namely , the order pattern of @xmath438 must be distributed in the same way as the order pattern of @xmath439 .
more precisely , if @xmath440 then @xmath441 thus if @xmath34 we necessarily have @xmath442 ( recall that @xmath99})$ ] . ) the functions @xmath443 and @xmath444 , now thought of as maps @xmath445 , induce maps @xmath446 .
explicitly , for @xmath67 , @xmath447 thus by and compatibility , @xmath448 for @xmath34 . set @xmath449 .
as each condition is linear , @xmath33 is a polytope contained in the simplex @xmath64 , and @xmath450 .
we have already proved the following lemma . if @xmath34 then @xmath451 .
[ ex : polytopes ] the polytope @xmath452 is all of @xmath453 ; this is a line segment connecting @xmath454 to @xmath455 .
the preimage of a point @xmath456 under the map @xmath457 is a 3-dimensional square pyramid with apex @xmath458 . if @xmath459 the vertices of the square base are @xmath460 where @xmath461 and @xmath462 , whereas if @xmath463 then the vertices are @xmath464 with the same choices for @xmath0 and @xmath465
. if @xmath466 then the square base is a ( 2-dimensional ) face of @xmath467 ; it corresponds to the face subgraph @xmath468 consisting of all the edges except the loops @xmath469 and @xmath470 .
the entire polytope @xmath467 is 4-dimensional ; it resembles a suspension of the ( middle ) square pyramid , except that the apex of the pyramid lies on the segment connecting the suspension points @xmath471 and @xmath472 , so that @xmath467 has six vertices rather than seven .
see figure [ fig : polytopes ] ( in which @xmath457 projects vertically ) .
( below ) and @xmath467 ( above ) .
fibers of the ( vertical ) projection are 3-dimensional square pyramids ; the preimage of the midpoint of @xmath452 is shown .
the whole polytope @xmath467 is the join of an interval and a square.,width=336 ] before we get to the main theorem we establish several connections between @xmath33 and @xmath473 . an _ edge weighting _ on a digraph @xmath120 is a map @xmath474 $ ] such that @xmath475 .
a _ flow _ on @xmath120 is an edge weighting @xmath476 such that for every @xmath477 , @xmath478 note that the set of all edge weightings on @xmath473 is exactly @xmath64 , and the set of all flows on @xmath473 is exactly @xmath33 .
a flow supported on an embedded loop in @xmath473 is a vertex of @xmath33 .
the set of all flows supported on a face subgraph @xmath479 is a face @xmath480 of @xmath33 .
the assignment @xmath481 is an inclusion - preserving bijection between the set of face subgraphs of @xmath473 and the set of faces of @xmath33 .
the dimension of @xmath480 is one less than the rank of the first homology of @xmath388 . in particular , if @xmath482 then @xmath483 has dimension @xmath413 .
if two face subgraphs @xmath484 are disjoint , then @xmath485 where @xmath486 denotes the join . by counting the face subgraphs of various ranks in , say , @xmath103 , one determines the number and structure of faces of @xmath467 of each dimension .
it is instructive to compare this with the earlier description of @xmath467 given in example [ ex : polytopes ] .
the dimension of @xmath64 is @xmath487 , and the conditions impose @xmath488 additional linear constraints .
these constraints are obviously independent , since their sum is zero ; the fact that @xmath33 has dimension @xmath413 shows that the constraints are otherwise linearly independent .
here is our main theorem , which we prove after a sequence of lemmas .
[ main ] 1 .
the set @xmath32 is a union of open faces of @xmath33 .
let @xmath489 be a face of @xmath33 and let @xmath388 be the corresponding face subgraph of @xmath473 , so that @xmath490
. then @xmath491 if and only if @xmath388 is driftless .
the closure of @xmath32 is @xmath33 .
the set @xmath433 is equal to the interior of @xmath452 .
the set @xmath492 consists of @xmath493 ( which is 4-dimensional ) together with all six of its open 3-dimensional facets , nine of its thirteen open 2-dimensional faces ( including the square face ) , and two of its thirteen open edges .
none of the six vertices of @xmath467 is in @xmath492 .
there are sometimes vertices of @xmath33 in @xmath32 .
for example the embedded loop in @xmath190 with edges @xmath494 @xmath495 @xmath496 @xmath497
@xmath498 @xmath499 @xmath500 @xmath501 @xmath502 is driftless , as is easily seen by computing its poset @xmath503 .
hence by theorem [ main ] the corresponding vertex of @xmath504 is realizable .
[ functions ] let @xmath305 be a driftless loop in @xmath473 .
then there is @xmath34 such that @xmath505 equals the counting measure induced on @xmath506 by @xmath305 .
in particular @xmath505 is in the interior of the face @xmath480 , where @xmath388 is the ( edge ) support of @xmath305 .
lift @xmath305 to a permutation @xmath507 .
let @xmath476 be a measure preserving ergodic function @xmath508 .
we build the _ permutation function _ corresponding to @xmath0 : for @xmath509 , set @xmath510 finally , let @xmath511 equal @xmath512 composed with a scaled down version of @xmath476 on the interval @xmath513 $ ]
. now @xmath511 has the desired property .
[ driftlessface ] let @xmath388 be a connected face subgraph of @xmath101 .
then @xmath388 is driftless if and only if @xmath514 .
assume @xmath388 driftless .
by lemma [ driftlessloop ] there is a totally driftless loop @xmath515 with support @xmath388 . by lemma [ functions ]
there is @xmath34 with @xmath516 .
conversely , assume @xmath388 drifts .
let @xmath34 , and suppose that @xmath517 . using the drift
, we will construct from @xmath3 a positive measure subset of @xmath518 and a measure preserving function @xmath43 such that either @xmath519 for all @xmath2 or @xmath520 for all @xmath2 .
this will contradict lemma [ balayage ] .
let @xmath521 .
note that @xmath522 , since @xmath523 and @xmath524 .
let @xmath525,{\epsilon}\in\{+,-\}$ ] be as asserted in the definition of drift .
set @xmath526 and @xmath527 note @xmath528 are positive by hypothesis .
for @xmath529 let @xmath530 be the smallest @xmath531 such that @xmath532 for @xmath533 we write @xmath534 with @xmath529 , and now define @xmath535 by @xmath536 note that @xmath43 is measure preserving . to see this consider @xmath537 measurable and write @xmath538}f^{-i}(j_{v , j})\subseteq \cup_if^{-i}(j_{v , j})-\cup_{i < n}f^{-i}(j_{v , j})$ ] a sequence with measure decreasing to @xmath539 write @xmath540 note that @xmath541 is a disjoint decomposition and for every @xmath1 there is @xmath542 so that @xmath543 now if @xmath544 , then @xmath545 for all @xmath533 , and if @xmath546 , then @xmath547 for all @xmath533 .
either case contradicts lemma [ balayage ] .
[ disconnected ] for any face subgraph @xmath388 of @xmath101 , @xmath548 if and only if @xmath549 for every connected component @xmath409 of @xmath388 .
suppose @xmath34 and @xmath550 .
let @xmath409 be a connected component of @xmath388 , and let @xmath551 .
note that @xmath552 ; defining @xmath43 to be a scaled up version of @xmath553 so that @xmath554 , we have @xmath555 .
the converse implication follows from the convexity of @xmath32 .
we now prove theorem [ main ] . to prove ( 1 ) , let @xmath556 denote the open face @xmath480 .
we will show that if @xmath557 is nonempty then for each vertex of @xmath480 there are points of @xmath557 arbitrarily close to @xmath263 . by convexity of @xmath558
it follows that @xmath559 , thus proving ( 1 ) .
suppose @xmath557 is nonempty .
if @xmath388 is connected , then by lemma [ driftlessface ] @xmath388 is driftless , and by lemma [ driftlessloop ] there is a totally driftless loop @xmath305 with support @xmath388 .
let @xmath263 be a vertex of @xmath480 and let @xmath365 be an embedded loop in @xmath388 such that @xmath560 . by lemma [ concat ] the loop @xmath561 is driftless , so by lemma [ functions ] there is @xmath34 with @xmath6 equal to the counting measure on the loop @xmath561 . as @xmath39 grows this sequence of measures approaches @xmath263 .
if @xmath388 is not connected , then by lemma [ disconnected ] , for each connected component @xmath409 of @xmath388 there is @xmath562 with @xmath563 .
we apply the argument from the previous paragraph to each face @xmath564 , obtaining points of @xmath32 close to the vertices of @xmath564 . as each vertex of @xmath480
is a vertex of one of the @xmath564 s , we are done . as for ( 2 ) , by ( 1 ) we know that @xmath565 if and only if @xmath566 .
if @xmath388 is connected , lemma [ driftlessface ] finishes it .
if @xmath388 is not connected , then for any connected component @xmath409 of @xmath388 we have @xmath567 if and only if @xmath409 is driftless .
so by lemma [ disconnected ] , @xmath565 if and only if each @xmath409 is driftless , i.e. , if and only if @xmath388 is driftless
. to prove ( 3 ) , it suffices to show that @xmath568 .
this is easy : as @xmath32 is not empty , there must exist a ( connected ) driftless face subgraph . by corollary
[ enlarge ] , the whole graph @xmath473 is driftless . since @xmath569 , the result is implied by ( 2 ) .
[ uniformdist ] for each @xmath1 , there exists @xmath34 such that @xmath6 is uniform on @xmath15 .
we have answered question 1 for @xmath31 .
however question 2 remains open .
in particular , we do not know if there is @xmath34 such that @xmath6 is uniform for all @xmath1 .
see section 8 .
in this section we change our focus from the distribution @xmath6 to a coarser statistic , namely the number of permutations of length @xmath1 realized by @xmath3 .
we relate two notions about a continuous piecewise monotone function @xmath3 : finite entropy and finite exclusion type .
the basic idea is that these two concepts imply opposite things for the number of length @xmath39 permutations realized by iterates of @xmath3 as @xmath39 gets large . roughly speaking , finite
entropy implies that the number of permutations realized by @xmath3 grows ( at most ) exponentially in the length . on the other hand
, finite exclusion type means that the only restrictions on the permutations realized by @xmath3 are given by looking at permutations of a fixed finite length .
often , this will imply that the number of realizable permutations in @xmath570 grows super - exponentially in @xmath39 .
define @xmath571 to be the image of @xmath572 in @xmath15 . for ( piecewise ) continuous functions , several classical definitions of the topological entropy @xmath573 are possible .
the reader is referred to @xcite for details .
a new notion of entropy called _ topological permutation entropy _ has been studied recently by several people ; the following combines theorem 1 of @xcite with theorem 2.1 of @xcite .
[ entropy ] if @xmath574 is piecewise continuous and piecewise monotone then @xmath575 and @xmath573 is finite .
a function @xmath27 has _ exclusion type @xmath1 _ if there exists @xmath389 such that @xmath576 for all @xmath179 and _ finite exclusion type _ if it has exclusion type @xmath1 for some @xmath1 .
note that this says not only that every path in @xmath101 realized by @xmath3 is supported on @xmath388 , but also that every lift of every path supported on @xmath388 is realized by @xmath3 .
a condition equivalent to finite exclusion type is that there are finitely many _
basic forbidden patterns _ for @xmath3 , in the language of @xcite .
this means that there are finitely many permutations @xmath577 such that any permutation @xmath0 ( of any length @xmath578 ) either occurs as @xmath579 for some @xmath2 or else satisfies @xmath580 for some interval @xmath581 $ ] and some @xmath54 .
elizalde has proposed the problem of characterizing those functions which have finite exclusion type .
we will give a necessary condition .
[ partiallydriftlessthm ] suppose @xmath3 has finite exclusion type @xmath1 , and let @xmath582 be the associated subgraph .
if @xmath388 contains a partially driftless loop then @xmath583 grows super - exponentially ; i.e. , for any @xmath584 , we have @xmath585 let @xmath305 be a loop on @xmath388 with @xmath374 for some particular @xmath586 $ ] .
let @xmath110 be the length of @xmath305 , and set @xmath587 . by the hypothesis of finite exclusion type we have @xmath588 now since @xmath374 , the @xmath54 elements @xmath589 of the poset @xmath590 are pairwise incomparable , by lemma [ partial ] .
thus the number of lifts of @xmath591 to @xmath592 is at least @xmath593 , by corollary [ factorialgrowth ] .
therefore @xmath594 for all @xmath277 and the result follows .
elizalde and liu @xcite have shown that there is no piecewise monotonic function @xmath595 of finite exclusion type with associated graph @xmath468 where @xmath596 .
this does not follow from the preceding theorem , as this @xmath388 contains no partially driftless loop .
for a given function @xmath3 , denote by @xmath597 the subgraph of @xmath101 with edge set @xmath598 .
[ ergodic ] if @xmath595 is ergodic then for every @xmath1 , @xmath597 contains a partially driftless loop .
consider the graph @xmath388 with vertex set @xmath599 , edge set @xmath600 and head and tail maps the restrictions to the initial and final segments of @xmath601 .
note that any directed cycle in @xmath388 yields a partially driftless loop in @xmath597 and that @xmath388 has finitely many vertices so it suffices to construct an infinite path in @xmath388 .
consider @xmath602 . by the compactness of @xmath518 and ergodicity of @xmath3 , @xmath603 . since @xmath604 there will be an infinite path in @xmath388 if @xmath605
; this is shown next .
for any @xmath427 choose @xmath606 so that if @xmath607 then @xmath608 .
choose @xmath609 with @xmath610 .
choose @xmath611 with @xmath612 for every @xmath613 ( so that @xmath614 ) .
such an @xmath39 exists since there is always eventually another sequence of length @xmath1 avoiding the @xmath615 ball around @xmath2 .
[ periodic ] if @xmath595 is piecewise continuous and if @xmath616 is a periodic point of period @xmath617 such that @xmath3 is continuous at every iterate of @xmath616 , then @xmath597 contains a partially driftless loop . using continuity ,
choose @xmath618 so that for any @xmath74 within @xmath394 of @xmath616 , the balls @xmath619 are pairwise disjoint for @xmath620 and the iterates satisfy @xmath621 for @xmath622 . then the image in @xmath101 of @xmath623 is a partially driftless loop .
[ entropyresult ] if @xmath595 is piecewise continuous and piecewise monotonic and either * @xmath3 is ergodic on a subinterval of @xmath518 , or *
@xmath3 has arbitrarily large finite orbits on which it is continuous , then @xmath3 does not have finite exclusion type .
recall that by sarkovskii s theorem @xcite , a continuous function has points of arbitrarily large period as long as there is a periodic point whose period is not a power of 2 .
we now give the promised proof of theorem [ cantor ] . given @xmath624 we will construct @xmath3 with @xmath83 . our construction will involve several layers of cantor sets , and the resulting functions will be nowhere near continuous or measure preserving .
recall that @xmath625})$ ] if @xmath12 .
[ l : i s ] given @xmath626 , there exist intervals @xmath627\}_{{\sigma}\in \cup_n s_n}$ ] , open at the left endpoint and closed at the right endpoint , with the properties that : 1 .
@xmath628 for all @xmath629 , 2 .
@xmath630 , 3 .
@xmath631 for all @xmath12 , 4 . for each @xmath1 , @xmath632 $ ] .
we define the @xmath633 inductively as follows .
first set @xmath634 $ ] .
now let @xmath635 and assume that intervals @xmath636 have been constructed for all @xmath637 .
since @xmath638 , summed over all @xmath12 such that @xmath639 , we may subdivide each @xmath636 into half - open intervals @xmath633 of length @xmath640 .
[ l : j s ] there exist disjoint intervals @xmath641 such that for all compatible sequences @xmath642 , and any @xmath643 with @xmath644 , @xmath645 .
again the construction is inductive .
suppose that the @xmath646 have been constructed for @xmath647 , and assume further that gaps of positive lengths exist between these intervals and at both endpoints .
order the permutations in @xmath648 arbitrarily , and for each such @xmath0 , let @xmath646 be an arbitrary open interval disjoint from the previously chosen intervals and with positive length gaps away from them , subject to the further condition that @xmath649 should lie in the correct gap as determined by the value of @xmath650 .
let @xmath82 be given ; we will construct a function @xmath27 with @xmath83 .
the construction proceeds in a sequence of steps .
_ let @xmath651 denote the ( usual ) cantor set in @xmath652 $ ] . by applying an order preserving transformation
we can assume that the @xmath653 given by lemma [ l : j s ] have the additional properties that @xmath654 $ ] and @xmath655 $ ] for all permutations @xmath0 .
for each permutation @xmath0 choose an order preserving injection @xmath656 with @xmath657 .
let @xmath633 be as in lemma [
l : i s ] . finally choose @xmath658 \rightarrow c$ ] to be an order preserving bijection .
we define the function @xmath659 on a subset of @xmath652 $ ] recursively , as follows . * first , on @xmath660=i _ { ( 1)}= i_{(12 ) } \cup i_{(21)}$ ] : for each @xmath661 , if @xmath662 then set @xmath663 .
thus for @xmath661 , we have @xmath664 . *
next , assuming @xmath659 is defined on @xmath665 , we define @xmath659 on @xmath666 as follows .
notice that @xmath667 . for all @xmath668 and for all @xmath669 define @xmath670 .
thus we have @xmath671 .
we have now defined every power of @xmath659 on @xmath672 $ ] ; note that the the domain of @xmath659 , which we will call @xmath673 , is @xmath672 $ ] union a measure zero set .
the purpose of this construction is that for any @xmath12 and @xmath662 , we now have @xmath674 .
we set @xmath675 on @xmath673 .
_ step @xmath578 , @xmath676 . _
denote by @xmath677 the measure zero set in @xmath678 \cup ( 1- 2^{-m } , 1]$ ] for which @xmath679 has already been defined .
define @xmath680 \cup ( 1- 2^{-m } , 1 ] \rightarrow ( 0,1]$ ] to be the map @xmath681 , \\ 1 - 2^{m-1}(1-x ) & \mbox { for } x\in ( 1- 2^{-m } , 1 ] .
\end{cases}\ ] ] for all @xmath682 , define @xmath683 .
note that if @xmath12 and @xmath684 , we now have @xmath685 after step @xmath578 , the domain of @xmath3 includes the interval @xmath686 $ ] , so the iterative process defines @xmath3 on @xmath687 .
it remains to show that @xmath688 for all @xmath1 . for @xmath12 ,
define @xmath689 .
by we have @xmath690 for all @xmath691 .
thus @xmath692 .
since @xmath693 is of measure zero , @xmath694 we conclude that @xmath695 for each @xmath1 , and @xmath83 .
this completes the proof .
many interesting open questions remain about the relationship between functions and their distributions of order paterns .
there is an infinite version of the polytope , @xmath697 , which consists of compatible sequences @xmath698 with @xmath699 .
we do not know if the `` interior '' of @xmath697 is realizable by some @xmath34 ( where the meaning of `` interior '' depends on the topology on @xmath697 ) , or if there is a drift condition for faces .
one concrete question is this : corollary [ uniformdist ] asserts that such an @xmath3 exists for any particular @xmath1 , and of course by theorem [ cantor ] there is @xmath27 that works for all @xmath1 .
yet there is no piecewise monotonic @xmath27 that works for all @xmath1 , because such an @xmath3 would have finite entropy ( by @xcite , or theorem [ entropy ] ) hence @xmath583 would grow at most exponentially in @xmath39 .
note that such a function might be desirable as a random number generator , since from the point of view of order patterns , its iterates would look perfectly random . in a somewhat different direction ,
if @xmath415 is a reasonably nice measure on @xmath518 then the results of section [ mainsec ] hold with @xmath31 replaced by the collection @xmath700 of functions which preserve @xmath415 .
( see remark [ othermeasures ] . )
returning to the broader questions 1 and 2 , there are several interesting classes of functions @xmath89 to study , such as ( piecewise ) continuous functions , polynomials , etc .
for example , if @xmath702 is the collection of piecewise continuous functions , then it is easy to see that the only vertices of @xmath64 contained in @xmath703 are @xmath704 and @xmath705 . | a permutation @xmath0 describing the relative orders of the first @xmath1 iterates of a point @xmath2 under a self - map @xmath3 of the interval @xmath4 $ ] is called an _
order pattern_. for fixed @xmath3 and @xmath1 , measuring the points @xmath5 ( according to lebesgue measure ) that generate the order pattern @xmath0 gives a probability distribution @xmath6 on the set of length @xmath1 permutations .
we study the distributions that arise this way for various classes of functions @xmath3 .
our main results treat the class of measure preserving functions .
we obtain an exact description of the set of realizable distributions in this case : for each @xmath1 this set is a union of open faces of the polytope of flows on a certain digraph , and a simple combinatorial criterion determines which faces are included .
we also show that for general @xmath3 , apart from an obvious compatibility condition , there is no restriction on the sequence @xmath7 .
in addition , we give a necessary condition for @xmath3 to have _ finite exclusion type _
, i.e. , for there to be finitely many order patterns that generate all order patterns not realized by @xmath3 .
using entropy we show that if @xmath3 is piecewise continuous , piecewise monotone , and either ergodic or with points of arbitrarily high period , then @xmath3 can not have finite exclusion type .
this generalizes results of s. elizalde .
given a function @xmath8 \to [ 0,1]$ ] , it is natural to examine properties of the sequence of iterates of @xmath3 beginning at some point @xmath9 $ ] : @xmath10 the _ order pattern _ for a sequence of distinct reals @xmath11 is the permutation @xmath12 that ranks the elements in increasing order ; specifically , @xmath13 if and only if @xmath14 .
a number of authors have explored the relationship between functions @xmath3 and the set of order patterns realized by the iterates of @xmath3 .
work of c. bandt , g. keller , b. pompe , j. m. amig , m. kennel , and m. misiurewicz @xcite relates the number of distinct order patterns arising from a function @xmath3 to the entropy of @xmath3 .
s. elizalde and others @xcite have examined which and how many order patterns do not arise for particular functions and classes of functions . here
we take a slightly broader view and investigate the collection of distributions of order patterns that particular classes of functions achieve . specifically ,
if @xmath4 $ ] is equipped with lebesgue measure and @xmath3 is almost aperiodic ( meaning that the set of points with finite orbit has measure zero ) then @xmath3 induces a probability distribution @xmath6 on @xmath15 in a natural way : @xmath16 we shall focus on the functions @xmath17 as well as the function @xmath18 which maps @xmath3 to the sequence @xmath19 . throughout the paper we consider functions with the property that almost all orbits are infinite : @xmath20 where @xmath21 is the set of aperiodic points ,
i.e. , points with infinite orbit .
we address the following natural questions : if @xmath22 is a collection of functions , then what is @xmath23 ?
what is @xmath24 ?
we begin by answering both questions for the class @xmath25 . for any @xmath3 ,
the distributions @xmath6 , @xmath26 must satisfy a certain compatibility condition . in theorem [ cantor ] we show that this is the only constraint on what is realizable for arbitrary @xmath27 : that is , for any sequence @xmath28 of compatible distributions on @xmath15 , there is a function @xmath27 which simultaneously satisfies @xmath29 .
we then turn our attention to the class of measure preserving functions , @xmath30 our main theorem ( theorem [ main ] ) provides a complete answer to question 1 for @xmath31 .
it is easy to see that the conclusion of theorem [ cantor ] can not hold for @xmath31 ; in fact we observe that @xmath32 is contained in a polytope @xmath33 consisting of all ( normalized ) flows on a certain digraph , which we call a _
permutation digraph_. we then show that @xmath32 is a union of open faces of @xmath33 including the top - dimensional face , and we give a combinatorial criterion for determining whether or not a given open face of @xmath33 is contained in @xmath32 .
to prove the main theorem we introduce the fundamental notion of _ drift_. naively , if one wants to construct @xmath34 realizing a given distribution @xmath35 , one might chop the interval into several subintervals and define @xmath3 to permute the intervals to produce the desired frequencies .
problems soon arise , however : for example if we want half the mass of the interval to have iterates with order pattern @xmath36 and the other half @xmath37 then we quickly realize that this is impossible , because @xmath38 would move all the mass to the right , which is impossible for a measure preserving function .
this is the essence of drift , and the upshot of theorem [ main ] is that this is the only obstruction : a face of @xmath33 either has drift or not , and the faces contained in @xmath32 are exactly those without drift . finally , we discuss the relationship between the entropy of @xmath3 and a property we call _ finite exclusion type_. the latter is equivalent to @xmath3 having finitely many _
basic forbidden patterns _ , in the language introduced by amig , elizalde , and kennel @xcite ; these properties mean that there are finitely many fixed patterns such that every permutation either arises as an order pattern of iterates of @xmath3 or contains one of the forbidden patterns
. a function with finite entropy can realize at most exponentially many permutations of length @xmath39 ( see @xcite ) , but using the notion of drift we show that quite often , a function with finite exclusion type must realize a super - exponential number of permutations .
in particular , if either @xmath3 is continuous and has points of arbitrarily large period or @xmath3 is ergodic , then @xmath3 can not have both finite entropy and finite exclusion type ; see corollary [ entropyresult ] .
this generalizes results from @xcite .
the paper is organized as follows .
we introduce some language and give our result for @xmath40 in section 1 , although we defer the proof to section [ cantorproof ] .
sections 2 - 4 develop the combinatorial ideas required for our main theorem , including several preliminary results about permutation digraphs and drift .
the main theorem is stated and proved in section [ mainsec ] .
our discussion of entropy and finite exclusion type makes up section 6 , and section [ cantorproof ] contains the proof of theorem [ cantor ] .
we close with some open questions in section 8 . |
let @xmath0 be a nonsingular quintic hypersurface .
the virtual count @xmath1 of rational algebraic curves of degree @xmath2 on @xmath3 admits a computation via gromov - witten theory and mirror symmetry .
let @xmath4 denote the genus 0 gromov - witten invariant of @xmath3 in degree @xmath5 .
the count @xmath1 is defined by the aspinwall - morrison formula @xcite , @xmath6 the connection between @xmath7 and actual curve counting on @xmath3 is discussed in @xcite . the mirror symmetry prediction of candelas , de la ossa , green , and parkes @xcite relates the genus 0 potential @xmath8 to hypergeometric series .
let @xmath9 be defined by @xmath10 the functions @xmath9 are a basis of solutions of the picard - fuchs differential equation @xmath11 let the variables @xmath12 and @xmath13 be related by @xmath14 .
the prediction , @xmath15 was later proven via localization on the space of genus 0 stable maps to @xmath16 @xcite .
let @xmath0 be a nonsingular quintic hypersurface defined over @xmath17 .
let @xmath18 be the symplectic form on @xmath3 obtained from the fubini - study metric .
complex conjugation determines an anti - holomorphic involution on @xmath3 with fixed locus equal to the set of real points @xmath19 .
the inclusion @xmath20 is lagrangian with respect to @xmath18 .
the enumeration of disks with boundaries in lagrangian submanifolds plays a basic role in open string theory and has been studied mathematically in several contexts .
the subject is not a direct extension of the theory of stable maps .
new issues such as orientation play a crucial role .
we provide here a complete calculation of the disk invariants of @xmath3 with boundary in the real lagrangian @xmath19 .
an early treatment of disk enumeration occurs in the construction of the fukaya category @xcite .
disk enumeration is required to define the differentials of the floer complex . however
, a symplectic invariant via disk enumeration is not defined in @xcite .
only the cohomology of the floer complex is invariant .
symplectic disk invariants have been defined with respect to the real lagrangian associated to an anti - holomorphic involution in @xcite .
a previous definition in the presence of a torus action preserving the real lagrangian ( not directly applicable to @xmath3 ) can be found in @xcite .
we will follow here the definitions of @xcite .
let @xmath21 for @xmath5 odd denote the degree @xmath5 disk invariant of @xmath3 with boundary in @xmath19 .
for a discussion of even degree , see sections [ ssec : ms ] and [ ssec : even ] .
in fact , @xmath21 depends on a choice of a @xmath22 structure on @xmath23 however , changing the @xmath22 structure only effects the sign of @xmath21 uniformly for all @xmath5 .
our conventions are fixed by choosing @xmath24 to be positive .
let @xmath25 denote the disk potential , @xmath26 our main result is a calculation of @xmath25 .
define @xmath27 the function @xmath28 is a solution of the picard - fuchs equation with an added inhomogenous term , @xmath29 alternatively , @xmath28 may be obtained by evaluation at @xmath30 of the ( non - truncated ) hypergeometric solution of the homogeneous picard - fuchs equation , @xmath31_{h=\frac{1}{2}}\ .\ ] ] [ tm1 ] via the mirror map @xmath32 , @xmath33 the disk invariants @xmath21 are typically fractional . following the strategy of curve enumeration , virtual disk counts @xmath34 are defined by the ooguri - vafa formula .
[ tm2 ] we define the counts @xmath34 by @xmath35 definition [ tm2 ] is justified by the multiple cover calculation of proposition [ der ] in section [ mcf ] .
the contribution of @xmath36-fold covers of a disk in the appropriate local calabi - yau geometry is @xmath37 .
we conjecture the invariants @xmath34 to be integers . each holomorphic disk with boundary in @xmath19
can be reflected by the schwartz principle to yield a real rational curve in @xmath3 .
conversely , real curves mapping to @xmath3 of odd degree may be halved to yield two disks @xcite .
the virtual number of real rational curves of odd degree in @xmath3 may be defined by @xmath38 again , @xmath39 vanishes for @xmath5 even .
a table of values can be found at the end of the paper .
let @xmath40 be a @xmath41 bundle with flat connection @xmath42 over the lagrangian submanifold @xmath43 the triple @xmath44 determines an object of the fukaya category of @xmath45 homological mirror symmetry @xcite predicts the existence of a corresponding object @xmath46 of the derived category of coherent sheaves on the mirror quintic @xmath47 the holomorphic chern - simons functional of @xmath46 is predicted to be mirror to the standard chern - simons functional of @xmath48 with corrections from disk instantons @xcite . in the following ,
we briefly explain how the mirror correspondence between functionals leads to an enumerative correspondence .
assume for simplicity that @xmath46 is a holomorphic vector bundle .
denote the underlying complex vector bundle by @xmath49 , and let @xmath50 be the connection on @xmath49 defining the holomorphic structure of @xmath51 let @xmath13 denote the complex moduli parameter of @xmath52 and let @xmath53 denote the holomorphic 3-form determined by @xmath54 the holomorphic chern - simons functional of @xmath49 depends on a second complex connection on @xmath55 which we denote @xmath56 we view @xmath57 as a connection 1-form relative to @xmath58 define the holomorphic chern - simons functional by @xmath59 critical points of @xmath60 are holomorphic connections complex connections with vanishing @xmath61 component of their curvature .
@xmath60 is constant on connected components of orbits of the complex gauge group of @xmath62 for the chern - simons functional with instanton corrections , we will require the following terminology for holonomy .
if @xmath63 is a bundle with connection @xmath64 and @xmath65 is a parametrized path in @xmath66 , we denote the holonomy of @xmath67 around @xmath65 by @xmath68 if @xmath69 and @xmath70 we write @xmath71 moreover , if @xmath72 is holomorphic , we define @xmath73 to be the sign of @xmath72 coming from the determinant line of the cauchy - riemann operator . in defining the chern - simons functional with instanton corrections
, we will not try to be entirely precise , but rather give an intuitive picture .
let @xmath74 be a totally real submanifold homologous to @xmath19 , and let @xmath75 be a chain satisfying @xmath76 .
let @xmath77 be a @xmath41 bundle , and let @xmath78 be a connection on @xmath79 let @xmath12 denote the complexified kahler moduli parameter of @xmath3 and let @xmath80 denote the associated complexified kahler form .
define the chern - simons function with instanton corrections by @xmath81 a lagrangian submanifold @xmath82 with vanishing obstruction chains in the sense of @xcite and a flat connection @xmath78 together constitute a critical point of @xmath83 @xmath84 is constant on orbits of the hamiltonian symplectomorphism group .
the corrections from closed instantons intersecting @xmath75 are necessary to compensate for codimension one bubbling where all the energy of a disk instanton is transferred to a sphere bubble @xcite .
@xmath84 and @xmath60 are lagrangians defining a pair of dual quantum field theories @xcite .
their critical values are physically significant and should be topological invariants . via the mirror transformation , which expresses @xmath12 as a function of @xmath85 the value of @xmath84 at a critical point
should be calculable from the value of @xmath60 at a mirror critical point .
however , we must find non - trivial choices of critical points @xmath57 and @xmath86 corresponding under mirror symmetry for all values of @xmath87 . in our case , such @xmath86 can be found using the geometry of the anti - holomorphic involution .
we choose @xmath88 since @xmath89 , we can choose @xmath90 to be the flat @xmath41 bundle with monodromy opposite to @xmath91 since @xmath19 is the fixed point set of an anti - holomorphic involution , @xmath19 is a critical point of @xmath84 with either flat bundle .
the mirror to such a choice of @xmath90 should be a unique up to gauge transformation holomorphic connection @xmath57 on @xmath49 not gauge equivalent to @xmath58 in @xcite , a heuristic argument is given to show @xmath92 is in fact given by @xmath28 . even after specifying which holomorphic structure @xmath57 induces , @xmath92 is only defined up to a period of @xmath93 indeed , changing the choice of @xmath57 by a complex gauge transformation not isotopic to the identity changes @xmath92 by a period of @xmath93 therefore , @xmath92 should satisfy an extension of the picard - fuchs equation for @xmath47 similarly , changing the choice of @xmath75 in the definition of @xmath84 changes @xmath84 by a multiple of the first derivative of @xmath94 from the preceding discussion , we see both @xmath84 and @xmath60 are essentially relative functionals , depending either on a pair of connections @xmath95 or a pair of totally real submanifolds with @xmath41 bundle , @xmath96 and @xmath97 therefore only disks of odd degree , for which the difference in monodromy of @xmath78 and @xmath42 cancels the negative sign in definition , contribute to the physically significant critical value of @xmath98 the contributions of even degree disks cancel due to this sign .
a priori , some other physical value may depend on the even degree disks .
however , on mathematical grounds , even degree disks appear not to lead to interesting invariants , see section [ ssec : even ] . the first number @xmath99 was calculated in @xcite .
theorem 1 was predicted in @xcite via low degree graphs sums and string heuristics .
our technique of proof uses the fully equivariant mirror correspondence of givental @xcite .
a previous application can be found in @xcite where disk enumeration for ( noncompact ) local geometries was considered .
the ooguri - vafa @xcite multiple cover formula of definition [ tm2 ] is by now established in many settings , see @xcite .
we have chosen the quintic 3-fold as a first case of study , but the methods of the paper are applicable much more generally .
it will be interesting to see which aspects of the solution persist .
we thank p. biran , t. graber , d. kazhdan , n. nekrasov , l. polterovich , p. seidel , e. shustin , g. tian , r. vakil , e. zaslow , and a. zinger for related conversations .
j. s. would like to thank e. farjoun and the hebrew university of jerusalem for their warm hospitality during the preparation of the paper .
r. p. was partially supported by a packard foundation fellowship and nsf grant dms-0500187 .
j. s. was partially supported by nsf grant dms-0111298 .
j. w. was partially supported by nsf grant phy-0503584 .
we recall the definition of the disk invariant @xmath21 of the quintic from @xcite .
our conventions for conjugation , real structures , and stable disk maps are discussed in sections [ conn ] .
the euler class approach to @xmath21 is presented in section [ ssec : q3d ] .
fix a symplectic manifold @xmath100 of real dimension less than or equal to @xmath101 with an anti - symplectic involution @xmath102 @xmath103 the fixed points @xmath104 define a lagrangian submanifold of @xmath105 a @xmath106 structure and , if @xmath84 is orientable , an orientation on @xmath84 induce a natural relative orientation on the moduli space @xmath107 of stable disk maps to @xmath108 of degree @xmath109 . since @xmath107 is an orbifold with corners ,
the definition of cohomology classes on the moduli space yielding an analog of gromov - witten theory is not immediately clear .
however , using @xmath102 certain corners of @xmath107 may be eliminated .
more precisely , the boundary of @xmath107 consists generically of stable disk maps with two disk components . replacing
_ one _ of the two components by the image under @xmath110 yields another two component map .
we define an equivalence relation @xmath111 on the boundary of @xmath107 based on this correspondence . for certain components of the boundary
, the relation @xmath111 preserves orientation .
after quotienting by @xmath111 on these components , we obtain a new moduli space @xmath112 with fewer corners which is still relatively orientable . on @xmath112 ,
many interesting cohomology classes can be defined .
consequently , a set of invariants are obtained of the triple @xmath113 reminiscent of standard gromov - witten invariants in many respects . in good situations ,
the invariants obtained from @xmath114 are actually enumerative .
for example , welschinger s signed counts of real curves @xcite arise as specializations of the theory @xcite .
let @xmath115 be homogeneous coordinates on @xmath16 .
the standard complex conjugation @xmath116 on @xmath16 is @xmath117 \stackrel{c_{\operatorname{id}}}{\mapsto } [ \overline{z}_0 , \overline{z}_1,\overline{z}_2,\overline{z}_3,\overline{z}_4].\ ] ] each @xmath118 yields an anti - holomorphic involution @xmath119 equivalent to @xmath116 . in particular , the anti - holomorphic involution @xmath120 , @xmath121 \stackrel{c}{\mapsto } [ \overline{z}_0 , \overline{z}_2,\overline{z}_1,\overline{z}_4,\overline{z}_3],\ ] ] is associated to the matrix @xmath122 let @xmath123 denote the fixed points of @xmath120 .
the involution @xmath120 will be most convenient for our calculation of disk invariants . a homogeneous polynomial @xmath124 on @xmath16
is defined over @xmath17 if @xmath125 for example , @xmath126 are both real linear polynomials .
a subvariety of @xmath127 is defined over @xmath17 if the ideal @xmath128 is generated by real homogeneous functions . the lines @xmath129\
|\ z_1,z_2\in { \mathbb{c}}\ \},\ ] ] @xmath130\ |\ z_3,z_4\in { \mathbb{c}}\ \}\ ] ] are both defined over @xmath17 .
the involution @xmath120 lifts canonically to the line bundles @xmath131 the linear polynomials are elements of @xmath132 the space of real sections .
let @xmath133 be homogeneous coordinates on @xmath134 .
let @xmath135 be the anti - holomorphic involution defined by @xmath136 \stackrel{c}{\mapsto } [ \overline{v},\overline{u}].\ ] ] the @xmath120-fixed points , @xmath137 form a circle .
a holomorphic disk map @xmath138 can be reflected by the schwartz principle to yield an algebraic map @xmath139 by definition , the degree @xmath5 of the disk map equals the degree of @xmath140 .
the map @xmath140 satisfies the following real condition @xmath141 conversely , every algebraic map @xmath142 satisfying yields two disk maps with boundary @xmath143 equal to @xmath144 .
the image of @xmath140 is a real subcurve of @xmath16 .
similarly , a stable holomorphic disk map @xmath72 reflects to a stable genus 0 map @xmath140 satisfying the real condition with respect to the natural extension of @xmath120 to degenerations of @xmath134 .
in fact , stability for @xmath72 can be defined by stability for @xmath140 .
we will use the notation @xmath145 also for the stable case where @xmath146 and @xmath143 are possibly reducible .
however , @xmath143 is always connected . in the odd degree case ,
_ every _ stable genus 0 map to @xmath16 defined over @xmath17 is obtained by reflection .
let @xmath147 denote the moduli space of unpointed disk maps of odd degree @xmath5 .
reflection yields an tale double cover of smooth orbifolds @xmath148 where @xmath149 denotes the moduli space of unpointed genus @xmath150 algebraic maps defined over @xmath17 .
the real dimension of @xmath151 is @xmath152 .
in fact , @xmath153 is an orientation double cover @xcite .
let @xmath154 denote the compactification of the moduli space @xmath155 by stable disk maps , and let @xmath156 denote the space of unpointed genus @xmath150 algebraic stable maps defined over @xmath157 the moduli space @xmath154 is a smooth orbifold with corners .
in fact , @xmath153 extends to finite smooth map @xmath158 mapping the corners of @xmath154 to the boundary divisor of @xmath159 the cardinality of the fiber over a real stable map with @xmath160 components fixed by @xmath120 is @xmath161 in section [ dj ] , the construction of the closed orbifold @xmath162 was outlined . a detailed argument
is given in section [ sec : ecf ] in the proof of proposition [ pr : co ] .
the equivalence relation @xmath111 identifies the corners of @xmath154 in such a way that the map @xmath163 descends to an tale double cover @xmath164 there is a natural inclusion @xmath165 in the space of unpointed stable genus 0 algebraic maps defined over @xmath166 .
the real dimension of @xmath156 is @xmath152 .
let @xmath167 be a nonsingular quintic hypersurface defined over @xmath168 with symplectic form obtained from the fubini - study metric .
an anti - symplectic involution @xmath169 is defined by complex conjugation .
the lagrangian @xmath170 is the real locus @xmath23 we consider maps from the holomorphic disk @xmath146 to @xmath3 of odd degree with boundary lying in @xmath19 , @xmath171 since the expected dimension of the moduli space of maps to @xmath172 is 0 , the relevant gromov - witten invariant @xmath21 is simply the virtual cardinality . in order to calculate @xmath173 following @xcite , we first reformulate @xmath21 as the integral of an euler class of an obstruction bundle over the moduli space @xmath174 such integrals may be studied via fixed point localization .
a similar approach was used by kontsevich @xcite in the closed case .
let @xmath175 be the real vector bundle over @xmath154 with fiber @xmath176 } = h^0(c,\tilde{f}^*\mathcal{o}_{{\mathbb{c}}{\mathbb p}^4}(5))_{\mathbb{r}}\ ] ] where @xmath177 \in \overline{m}_{\mathbb r}({\mathbb{c}}{\mathbb p}^4,d)\ ] ] is the stable rational map obtained from the stable disk map via reflection , and @xmath178 denotes real sections .
the vector bundle @xmath175 is of real rank @xmath152 and is oriented on @xmath179 by lemma 8.7 of @xcite .
the integral of the euler class @xmath180 over @xmath154 is not well defined because the space @xmath154 has non - empty boundary , and @xmath175 is not trivial near the boundary .
however , @xmath175 naturally descends to a vector bundle @xmath181 neither @xmath182 nor @xmath183 are orientable .
let @xmath40 denote the local system defined by the determinant of the tangent bundle of the moduli space @xmath174 in lemma [ lm : ro ] of section [ sec : ecf ] , we prove @xmath184 as topological bundles .
a @xmath22 structure on @xmath19 determines the choice of the isomorphism uniquely up to scaling by a positive constant .
hence , the euler class @xmath185 is well - defined .
since @xmath183 is a closed orbifold , the integral @xmath186 is well - defined . in section [ sec : ecf ] , we obtain the following result .
[ tm3 ] for @xmath5 odd , @xmath187 we prove theorem [ tm3 ] using the symplectic virtual moduli cycle .
the same technique can be used to prove the analogous well - known result for the closed invariants . a stable disk map of even degree may still be reflected to obtain an even degree real genus 0 stable map
. however , not all stable genus 0 maps of even degree defined over @xmath17 are so obtained .
stable maps defined over @xmath17 with domains having no real points can not be halved .
the even disk invariant @xmath21 is not well - defined without the addition of the contributions of real curves without real points .
if such contributions were incorporated , @xmath21 would be expressible as the euler class of an odd dimensional real bundle and hence would presumably vanish .
hence , the definition @xmath188 for @xmath5 even . our formula for @xmath21 is _ independent _ of the quintic @xmath0 defined over @xmath17 . since the calculation is done on @xmath16 , some information is possibly lost .
more precisely , let @xmath189 the invariant @xmath21 is an integral over all stable disk maps @xmath190 of degree @xmath5 with boundary @xmath143 determining a class in @xmath191 if @xmath153 is an isomorphism , as is the case , for example , for the fermat quintic @xmath192 then there is no loss of information .
if , however , @xmath153 has a kernel , more refined disk invariants of @xmath193 may sometimes be defined for @xmath194 in the latter case , @xmath195
let @xmath196 denote the complex numbers of unit modulus , @xmath197 the torus @xmath198 acts diagonally @xmath199 .
a @xmath198-action on @xmath16 is obtained by projectivization , and canonical lifts to the line bundles @xmath200 are obtained .
there is a canonically induced translation action of @xmath198 on @xmath201 .
let @xmath202 denote the @xmath198-fixed points , @xmath203,\ \zeta_1=[0,1,0,0,0],\ \ldots,\ \zeta_5=[0,0,0,0,1].\ ] ] the involution @xmath120 fixes @xmath204 and permutes the others , @xmath205 hence , @xmath204 is the unique real @xmath198-fixed point .
consider the rank 2 subtorus @xmath206 acting by @xmath207 = [ z_0 , \xi_1 z_1 , \overline{\xi}_1 z_2 , \xi_2 z_3 , \overline{\xi}_2 z_4].\ ] ] since @xmath208 preserves @xmath209 , translation defines a @xmath208-action on the moduli spaces @xmath210 and @xmath211 .
the algebraic torus @xmath212 acts on @xmath16 by complexifying the action of @xmath208 , @xmath207 = [ z_0 , \xi_1 z_1 , \xi^{-1}_1 z_2 , \xi_2 z_3 , \xi^{-1}_2 z_4].\ ] ] of course , @xmath212 acts on @xmath201 by translation .
we follow the equivariant weight conventions of @xcite for the torus @xmath198 .
let @xmath213 be the @xmath198-equivariant cohomology class determined by chern class of the restriction of @xmath214 to @xmath215 , @xmath216 the classes @xmath213 generate , @xmath217.\ ] ] the tangent weights of @xmath16 at the point @xmath215 are @xmath218 .
let @xmath219 be the generators of @xmath220 defined by the pull - back @xmath221 and the equations @xmath222 the pull - back @xmath223 vanishes .
for notational convenience , we will often omit the pull - back @xmath224 and write @xmath225 the genus 0 gromov - witten invariants @xmath226 have been calculated in @xcite via localization on @xmath201 with respect to the @xmath198-action .
we will calculate @xmath21 via localization on @xmath211 with respect to the @xmath208-action .
let @xmath5 be odd .
the @xmath208-action on the moduli space @xmath227 lifts canonically to the vector bundle @xmath182 .
we calculate the integral @xmath228 by localization with respect to the @xmath208-action .
the localization calculation is similar in flavor to the genus 0 gromov - witten calculation of @xmath3 in @xcite .
however , two new issues arise : 1 . the @xmath208-action has fixed loci in @xmath227 with moving images in @xmath16 .
the equivariant restriction of @xmath229 to the @xmath208-fixed locus depends upon the orientation of @xmath182 . issue ( i ) is handled by identifying the non - rigid contributions with the equivariant correlators @xmath230 studied by givental @xcite .
issue ( ii ) requires an explicit evaluation of the signs occurring in the orientation .
the derivation is presented in section [ signs ] . the sum over @xmath208-fixed point loci required for the localization formula
is executed in two steps .
subsums with fixed intersection type with @xmath231 are evaluated using givental s equivariant mirror transformation for @xmath230 .
finally , the sum over intersection types is evaluated explicitly after appropriate equivariant specialization .
the interaction of the orientation signs with the localization sum is an interesting aspect of the calculation .
the outcome is a proof of theorem 1 .
we first study the @xmath208-fixed locus of the moduli space of stable disk maps .
let @xmath232 \in{\overline{m}_d({\mathbb{c}}{\mathbb p}^4/{\mathbb{c}}{\mathbb p}^4_{\mathbb r } , d)^{\mathbf{t}^2}}\ ] ] be a @xmath208-fixed map .
the boundary @xmath143 distinguishes a minimal , @xmath120-invariant , central curve @xmath233 of the domain of the reflected map @xmath234 \in \overline{m}_{\mathbb{r}}({\mathbb{c}}{\mathbb p}^4,d)\ ] ] satisfying @xmath235 .
the _ central degree _ of @xmath72 is the degree of the restriction @xmath236 the central degree @xmath237 is positive , odd , and bounded by @xmath5 .
the moduli point @xmath238\in
\overline{m}_{\mathbb{c}}({\mathbb{c}}{\mathbb p}^4,p)\ ] ] is fixed for the full complexified action of @xmath212 on @xmath16 .
the two lines @xmath239 are the only @xmath212-invariant curves of odd degree defined over @xmath240 in @xmath16 .
a real @xmath212-invariant subcurve must lie in one of the two planes @xmath241 \
z_0,z_1,z_2 \in { \mathbb{c}}\ \},\ ] ] @xmath242 \ | \
z_0,z_3,z_4 \in { \mathbb{c}}\ \}.\ ] ] in the first case , @xmath84 is the only real @xmath212-invariant line .
moreover , all non - linear @xmath212-orbits are of degree 2 .
the argument in the second case is identical .
a node of @xmath243 must map via @xmath140 to the unique real fixed point @xmath244 .
since @xmath245 must equal either @xmath84 or @xmath246 , @xmath245 can not contain @xmath204 .
hence , @xmath243 can not contain a node .
we obtain the following result .
the central curve @xmath65 is @xmath134 and @xmath247 is a galois cover of odd degree @xmath237 .
the original disk map @xmath72 is obtained from one half of @xmath140 .
hence one half of @xmath65 is selected by @xmath146 .
a half of @xmath65 determines a pair @xmath248 where @xmath249 is a non - real fixed point and @xmath237 is the central degree .
the data @xmath248 is the termed the _ intersection type _ of @xmath72 with the real lagrangian @xmath250 .
the half of @xmath65 selected by @xmath146 is the _ intersection disk_. while we have analyzed @xmath251 , we are actually interested in @xmath252 . if fact , we have proven @xmath253 since the @xmath208-fixed maps are _ not _ corner points of @xmath154 .
the localization calculation of @xmath228 is sum over the contributions of the @xmath208-fixed loci .
we may separate the contributions by intersection type , @xmath254 the _ intersection disk term _
@xmath255 of @xmath256 is the contribution of the unique @xmath208-fixed map @xmath257 incident to @xmath215 with central degree @xmath237 and domain consisting only of the intersection disk .
define the rational function @xmath258 of degree @xmath150 by the following formula , @xmath259 [ ft1 ] for an appropriate choice of @xmath22 structure on @xmath19 , we have @xmath260 changing the @xmath22 structure changes the formulas by @xmath261 for all @xmath262 the proof of lemma [ ft1 ] is given in section [ signs ] . the most interesting aspect is the calculation of the prefactor @xmath263 obtained from the orientations of the moduli space @xmath264 and the bundle @xmath265 .
let @xmath266 be the moduli space of 2-pointed stable complex algebraic maps to @xmath16 of genus 0 and degree @xmath267 .
let @xmath268 be the evaluation at the @xmath269 marking , and let @xmath270 denote the chern class of the @xmath269 cotangent line .
let @xmath271 be the complex vector bundle with fiber @xmath272 } = h^0(c,\tilde{f}^*\mathcal{o}_{{\mathbb{c}}{\mathbb p}^4}(5))_{{\mathbb{c}}}.\ ] ] following the notation of section 2.2 of @xcite , givental s equivariant correlator @xmath230 for the torus @xmath208 is defined by @xmath273 where @xmath274 is the hyperplane class , @xmath275 the sum in is over _ all _ non - negative integers @xmath267 .
the unstable degree 0 term is defined by @xmath276 let @xmath277 \in h^*_{\mathbf{t}^2}({\mathbb{c}}{\mathbb p}^4)$ ] denote the poincar dual of the class of the fixed point . for classes @xmath278 , let @xmath279\ ] ]
denote the equivariant intersection pairing .
for example , @xmath280\rangle = \lambda_i\ ] ] following convention . the intersection pair of the equivariant correlator will arise in the localization analysis : @xmath281 \rangle & = & \frac{1}{5\lambda_i } \sum_{r\geq 0 } e^{(\lambda_i/\hbar+r ) t } \int_{\overline{m}_{0,2}({\mathbb{c}}{\mathbb p}^4,r ) } \frac { c_{\text{top}}(e_r)}{{\hbar}-\psi_2 } e_2^*([\zeta_i ] ) \\ & = & \frac{{\hbar}^{-1}}{5\lambda_i } \sum_{r\geq 0 } e^{(\lambda_i/\hbar+r ) t } \int_{\overline{m}_{0,1}({\mathbb{c}}{\mathbb p}^4,r ) } \frac { c_{\text{top}}(e_r)}{{\hbar}-\psi_1 } e_1^*([\zeta_i])\end{aligned}\ ] ] where the string equation is used in the second line . in degree @xmath150 ,
the unstable 1-pointed term is defined by the second equality .
the @xmath208-fixed loci of the moduli space @xmath282 of type @xmath283 may be quite complicated .
however , every map @xmath284\in \widetilde{m}_d ( { \mathbb{c}}{\mathbb p}^4/{\mathbb{c}}{\mathbb p}^4_{\mathbb{r}},d)^{\mathbf{t}^2}\ ] ] of type @xmath283 ends in the same intersection disk . by expanding the localization formula ,
the intersection disk term @xmath255 can be factored out of @xmath285 by removing the intersection disk from @xmath72 .
what remains after the intersection disk is removed from @xmath72 ?
in fact , _ every _ genus 0 stable complex map @xmath286\in e_1^{-1}(\zeta_i ) \subset \overline{m}_{0,1}({\mathbb{c}}{\mathbb p}^4,r)^{({\mathbb{c}}^*)^2}\ ] ] can be found .
the stable disk map @xmath72 is obtained by attaching the @xmath283-intersection disk to @xmath287 at the marking .
a direct unraveling of the localization formulas yields the following fundamental result .
let @xmath288 we have @xmath289 \rangle \cdot i(\zeta_i , p)\ ] ] for @xmath290 the sum on left side can be indexed more conveniently as @xmath291 the right side of the equality may be expanded as @xmath292 ) \cdot \frac{i(\zeta_i , p)}{(5\lambda_i)(\frac{p}{2 } \lambda_i)}.\ ] ] the lemma is obtained from the equality @xmath293 ) \cdot \frac{i(\zeta_i , p)}{(5\lambda_i)(\frac{p}{2 } \lambda_i)}.\ ] ] to prove , we apply @xmath208-localization to the integral on the right .
we do _ not _ fully expand the @xmath208-localization formula . as was discussed previously , understanding the geometry of the individual @xmath208-fixed loci
is difficult as there are positive dimensional families of @xmath208-fixed curves .
however , has a much simpler proof . since both sides are expressed as @xmath208-residue integrals by localization , we need only match the geometries .
first , the @xmath208-fixed loci of the two sides of are in bijective correspondence . given a @xmath208-fixed locus on the right , the addition of the intersection disk @xmath255 at the marking @xmath294 yields a @xmath208-fixed locus of the right side .
the reverse direction is obtained by stripping the intersection disk .
second , since the @xmath208-fixed loci on the left and right are both nonsingular , the correspondence induces an _ isomorphism _ of @xmath208-fixed loci up to the automorphism factor of the intersection disk .
finally , we must match the @xmath208-fixed obstruction theories .
let @xmath295\in \widetilde{m}_d ( { \mathbb{c}}{\mathbb p}^4/{\mathbb{c}}{\mathbb p}^4_{\mathbb{r}},p+2r)^{\mathbf{t}^2}\ ] ] be a map with @xmath296 \in \text{ev}_1^{-1}(\zeta_i)\subset \overline{m}_{0,1 } ( { \mathbb{c}}{\mathbb p}^4,r)^{\mathbf{t}^2}.\ ] ] by the normalization sequence , @xmath297 } { \rightarrow}e_{r}|_{[f_c ] } \oplus
f_{p}|_{[i(\zeta_i , p ) ] } { \rightarrow}{{\mathcal{o}}}_{{\mathbb{c}}{\mathbb p}^4}(5)|_{\zeta_i } { \rightarrow}0.\ ] ] hence , the numerator in the residue integral on the left of is @xmath298 similarly , the denominator of the residue integral on the left of is @xmath299 } ) } = \frac{1}{c_{\text{top } } ( \text{nor}_{[f_c ] } ) } \frac{c_{\text{top}}(\text{tan}_{\zeta_i } ) } { ( \frac{2}{p } \lambda_i -\psi_1 ) ( \frac{2}{p}\lambda_i ) } \frac{1 } { e(\text{nor}_{[i(\zeta_i , p)]})}.\ ] ] the middle terms are obtained from tangent bundle , node smoothing , and automorphism factors . putting and together ,
we obtain the exact matching needed for .
the factorization of properly reflects the orientation on the moduli space @xmath300 .
the orientation factorization is easily obtained from @xcite .
such arguments form the geometric basis of @xcite .
though the @xmath208-action on @xmath300 has fixed loci corresponding to moving maps , the issue is completely avoided by the lemma .
we review the mirror transforms relating @xmath301 to @xmath302 following section 4.4 of @xcite .
the mirror map @xmath32 discussed in section [ mm ] can be written explicitly .
let @xmath303 then @xmath304 is the mirror map .
exponentiating yields @xmath305 the equivariant mirror transformation for the torus @xmath208 is @xmath306 transforms for the equivariant pairings are a direct consequence , @xmath307\rangle = \frac{1}{f(e^t ) } \langle s_q^*(t,\frac{2}{p}\lambda_i ) , [ \zeta_i]\rangle.\ ] ] we now complete the calculation of @xmath25 , @xmath308 \rangle \cdot i(\zeta_i , p ) \\ & = & \sum_{i=1}^4 \sum_{p\ odd}\frac{1}{f(e^{t } ) } \langle s^*_q(t,\frac{2}{p}\lambda_i ) , [ \zeta_i ] \rangle \cdot i(\zeta_i , p)\ , .\end{aligned}\ ] ] the @xmath309 summands of the last line together yield @xmath310 written in terms of the homogeneous variable @xmath311 similarly , the @xmath312 summands together yield @xmath313 the final step is to observe the localization calculation of @xmath25 is a _
weight independent _ global integral .
hence , we may evaluate the summation in the @xmath314 limit .
only the @xmath309 terms survive the limit . @xmath315
the identity for odd @xmath5 , @xmath316 restated in lemma [ idd ] below concludes the proof of theorem 1 , @xmath317 after regrouping the factors and reindexing the sum , the identity is equivalent to the following result .
[ idd ] for @xmath5 odd , @xmath318 an elementary derivation is left to the reader . a geometric proof is obtained from the multiple cover calculations in section [ mcf ] .
we now derive the signs needed in the localization calculation and prove lemma [ ft1 ] .
the orientations of the moduli spaces and vector bundles used to calculate @xmath21 arise from the natural orientation of the determinant of the cauchy - riemann @xmath106 boundary value problem developed in @xcite .
briefly , a cauchy - riemann boundary value problem consists of topological complex vector bundle @xmath319 over a riemann surface with boundary @xmath320 a totally real sub - bundle @xmath321 over the boundary @xmath322 and a generalized cauchy - riemann operator @xmath323 on @xmath324 given a @xmath106 structure on @xmath325 and a choice of orientation on @xmath321 if orientable , one may define a canonical orientation of the determinant @xmath326 reversing the @xmath106 structure on @xmath321 reverses the canonical orientation ( * ? ? ?
* lemma 2.10 ) . for the calculations of weights below , we will only be concerned with the situation where @xmath327 is the restriction of an algebraic vector bundle @xmath328 defined over @xmath168 .
we take @xmath329 and we take @xmath323 to be the restriction to @xmath146 of the @xmath330 operator defined by the holomorphic structure on @xmath331 the identifications @xmath332 are easily obtained .
so , an orientation of @xmath333 gives an orientation of the virtual vector space @xmath334 note , however , the orientation depends on the choice of @xmath335 so , we can not entirely forget the origins of our orientation in a boundary value problem . for convenience ,
we introduce the notation @xmath336 section [ sec : ws ] interprets the symplectic geometric definition of the orientation algebraically .
section [ sec : wcalc ] calculates the localization contributions from a fixed point of the torus action using the combinatorics of exact sequences and the formula for the tensor product of real representations of @xmath337 the algebraically inclined reader may safely skip all of section [ sec : ws ] besides the statement of lemma [ lem : aor ] .
let @xmath338 denote the real @xmath339-dimensional torus and let @xmath340 denote its lie algebra .
a weight is a homomorphism of real vector spaces from @xmath340 to @xmath341 let @xmath49 be a two - dimensional real irreducible oriented representation of @xmath338 and let @xmath342 be the associated homomorphism @xmath343 let @xmath344 be an orientation preserving real linear homomorphism such that associated homorphism @xmath345 satisfies @xmath346 note that @xmath347 is defined by these conditions up to homothety and hence @xmath348 is unique
. differentiating @xmath349 we obtain the weight of @xmath62 we denote by @xmath350 the 2-dimensional real oriented representation of @xmath338 of weight @xmath351 where @xmath352 may be fractional .
if @xmath353 we denote by @xmath350 the trivial two dimensional representation .
suppose @xmath49 is an oriented real representation of @xmath338 isomorphic to @xmath354 a priori , @xmath49 consists of two data : a @xmath338 action on the vector space @xmath49 and an orientation of the vector space @xmath62 however , except in the case @xmath355 the @xmath338 action on @xmath49 and the knowledge that @xmath356 determine the orientation of @xmath49 . indeed , if @xmath357 there is a unique up to homothety @xmath338 equivariant isomorphism @xmath358 the orientation on @xmath49 must agree with the orientation induced by @xmath359 let @xmath360 be a positive odd integer . from @xcite , in order to define the canonical orientation of @xmath333 mentioned above , a @xmath106 structure @xmath361 on @xmath362 must be chosen .
we fix such a @xmath106 structure . set @xmath363 choosing a connected component of the complement of @xmath364 is equivalent to choosing an orientation on @xmath354 the action of @xmath338 on @xmath365 naturally induces an action on @xmath366 when we think of @xmath146 as the disk that induces the orientation of @xmath354 indeed , as not oriented vector spaces , clearly @xmath367 we only consider the @xmath360 odd case .
the main goal of this section is to prove the following lemma which examines when isomorphism preserves orientation .
equip the vector space @xmath366 with the @xmath338-action induced from @xmath368 let @xmath369 be a @xmath106 structure on @xmath370 since @xmath360 is odd , @xmath371 as real topological vector bundles .
[ lem : aor ] assume that @xmath369 agrees with @xmath372 with respect to the canonical orientation induced by @xmath373 @xmath374 first , the canonical orientation induced by @xmath369 can be expressed as the complex orientation induced by an explicit complex structure on @xmath375 indeed , by gluing sections , we have an exact sequence @xmath376 by definition , after equipping the latter two terms of the sequence with the complex orientation , the sequence induces the desired orientation on the first term .
note that @xmath377 the inverse of the gluing map , is not canonical , but the set of all choices is connected .
so , the induced orientation is well - defined . to calculate the orientation
, we may fix a particular choice of the gluing map and calculate the induced complex structure on @xmath375 we digress for a moment to explain the relationship between different possible descriptions of @xmath375 let @xmath378 be standard linear coordinates on @xmath365 such that @xmath378 are real precisely on the real locus of @xmath379 then , @xmath366 may be identified with the vector space of homogeneous polynomials @xmath380 now , we choose new coordinates @xmath381 such that @xmath382 then , since @xmath383 we have @xmath384 on the other hand , by definition , @xmath366 is the space of the solutions @xmath385 of the cauchy riemann equations on unit disk @xmath386 satisfying the totally real boundary conditions @xmath387 in the future , we refer to these boundary conditions as @xmath388 expanding in power series about @xmath389 it is not hard to see that the boundary conditions imply @xmath390 so , solutions @xmath385 arise naturally from homogeneous polynomials @xmath391 by trivializing @xmath392 over @xmath393 by the section @xmath394 and making the identification @xmath395 for the following argument , we view sections @xmath396 as in . referring to exact sequence , we construct a gluing map @xmath397 as follows .
suppose we identify the point at which we glue sections with @xmath398 then , by taking the standard trivialization of the sheaf @xmath399 over the standard coordinate chart on @xmath400 centered at @xmath401 we may identify @xmath402 with the set of polynomials @xmath403 of degree less than or equal to @xmath404 let @xmath405 be a cutoff function depending only on @xmath406 such that @xmath407 and @xmath408 for @xmath409 given a polynomial @xmath410 we define a pre - gluing @xmath411 let @xmath412 denote the @xmath413 projection .
we define @xmath414 by @xmath415 due to our choice of gluing map , it is easy to calculate that @xmath416 let @xmath385 be as in and let @xmath417 denote the complex structure @xmath414 induces on @xmath375 it follows that @xmath418 finally , we compute the weights of @xmath375 since , by definition of the action on @xmath419 we have identified @xmath146 with the hemisphere of @xmath400 inducing the intrinsic orientation of @xmath420 we must have that @xmath421 transforms by @xmath422 since @xmath423 it follows that in coordinates @xmath381 the action of @xmath338 takes the form @xmath424 in other words , @xmath338 acts by @xmath425 on the section @xmath426 the decomposition @xmath427 follows .
let @xmath428 and @xmath429 be two oriented real vector spaces , and let @xmath430 and @xmath431 be oriented bases .
there are different ways the tensor product @xmath432 can inherent an orientation from @xmath428 and @xmath433 we use the right - to - left lexicographical ordering convention .
we take @xmath434 as an oriented basis of @xmath435 if @xmath428 is even dimensional then the orientation of @xmath436 is independent of the orientation of @xmath437 and vice versa . as before ,
let @xmath350 be the 2-dimensional real representation of @xmath338 with weight @xmath438 the following result is a straightforward linear algebra calculation .
with respect to the right - to - left lexicographical orientation of the tensor product , @xmath439 let @xmath440\in \overline{m}_d({\mathbb{c}}{\mathbb p}^4/{\mathbb{c}}{\mathbb p}^4_{\mathbb r},d)\ ] ] denote a @xmath208-fixed disk of type @xmath441 after permuting indices , the proof given below applies to the other possible intersection types as well .
first , we calculate the equivariant euler class of the tangent space to @xmath154 at @xmath442 $ ] which we denote @xmath443 we use the deformation exact sequence @xmath444 here , @xmath445 denotes the space of first - order deformations of the map @xmath446 and @xmath447 denotes the first - order deformations of @xmath72 modulo reparametrization the tangent space to @xmath154 at @xmath442 $ ] . to carry out the corresponding closed calculation , the weights of each of the first two terms and computed and divided with cancelling 0-weights . in the open case ,
more information is needed about the weight 0 components since isomorphisms of two copies of the trivial real representation of @xmath338 need not preserve orientation .
we will use the real euler sequence to linearize the exact sequence to get a better handle on the sign .
let @xmath448 and @xmath449 let @xmath450 let @xmath451 denote the 2-dimensional linear subspace corresponding to @xmath452 and let @xmath453 denote the @xmath338-invariant complement . consider the commutative diagram in figure [ fig : lin ] .
@xmath454\ar[d ] & { h^0}(d,{\mathcal o}_{{{\mathbb{c}}}{\mathbb p}^1}(1 ) \otimes u ) \ar[r]\ar[d]^{d\!\tilde f } & { h^0}(d , t_{{{\mathbb{c}}}{\mathbb p}^1 } ) \ar[d]^{df } \\ { h^0}(d , f^*{\mathcal o}_{{{\mathbb{c}}}{\mathbb p}^4 } )
\ar[r]\ar[d ] & { h^0}(d , f^*{\mathcal o}_{{{\mathbb{c}}}{\mathbb p}^4}(1)\otimes w ) \ar[r]\ar[d ] & { h^0}(d , f^*t_{{{\mathbb{c}}}{\mathbb p}^4 } ) \ar[d ] \\ 0 \ar[r ] & \operatorname{def}(d , f ) \ar[r ] & \operatorname{def}(d , f ) } \ ] ] all rows and columns are exact .
the rows are the sections functor applied to the euler sequence .
the rightmost column is obtained from the deformation exact sequence by the identifications @xmath455 in order to discuss the orientations of the vector spaces in figure [ fig : lin ] , we must digress for a moment on the subject of @xmath106 structures . in the following , @xmath369 denotes a @xmath106 structure on @xmath456 @xmath457 denotes a @xmath106 structure on @xmath458 and @xmath459 denotes a @xmath22 structure on @xmath23 to define @xmath173 a structure @xmath459 was fixed . by lemma [ lm : sp ] of section [
sec : ecf ] , any two of @xmath460 naturally determine the third . as explained in the proof of lemma [ lm : ro ] of section [ sec : ecf ] , the choice of @xmath369 and @xmath457 compatible with @xmath459 induces the isomorphism @xmath461 used to determine the sign of @xmath462 in particular , if we orient @xmath463 and @xmath464 using @xmath369 and @xmath457 compatible with @xmath465 we will be calculating the weights of the @xmath208-action correctly . in order to facilitate calculations , we choose @xmath369 in a convenient way
. then we let @xmath459 and @xmath369 induce @xmath466 indeed , choose first a @xmath106 structure @xmath467 on @xmath468 via the natural homomorphism , @xmath469 induce a @xmath106 structure @xmath470 on @xmath471 in addition , equip @xmath472 with the canonical @xmath22 structure . by ( * ? ? ?
* lemma 8.1 ) , via the euler sequence , @xmath470 induces a @xmath106 structure @xmath369 on @xmath473 since we have chosen @xmath106 structures compatibly , by ( * ? ? ?
* lemma 8.4 ) , the middle row of the diagram in figure [ fig : lin ] respects orientation . the columns respect orientation by definition .
at this point , we focus attention on the middle column of figure [ fig : lin ] , which is the desired linearization of the deformation exact sequence .
because of the way we have induced @xmath470 from @xmath474 by an argument similar to the proof of ( * ? ? ?
* lemma 8.4 ) , we may assume that the natural isomorphism @xmath475 preserves orientation .
here , we have used the right - to - left lexicographical orientation of the tensor product . up to a degree independent sign , we may assume that the isomorphism @xmath476 also preserves sign .
so , it remains to compute the third term of the exact sequence , @xmath477 from the first two .
we assume without loss of generality that the induced @xmath208-action on @xmath478 has weights @xmath479 as opposed to their negatives .
that is , we assume the action on the underlying vector space considered in section [ sec : ws ] has weight @xmath480 this depends on the choice of @xmath481 one way or another , the opposite disk has the opposite sign , so we can always interchange @xmath482 and @xmath483 to satisfy our assumption .
note , however , that the full localization contribution of @xmath442 $ ] including the weights of the obstruction bundle is invariant under the symmetry @xmath484 let @xmath485 denote the 2-dimensional linear subspace corresponding to @xmath452 and let @xmath486 denote its @xmath208-invariant complement .
note @xmath487 we study the induced morphism , @xmath488 carefully in order to calculate the weights of the cokernel with attention to sign necessary because of the trivial representations that occur in the domain and the range . indeed , by formulas and , we have @xmath489 the map @xmath490 is determined up to homothety on the summand @xmath491 by @xmath208-equivariance .
since , homotheties of an even dimensional vector space preserve orientation , we need calculate no further . however , we need more information to determine @xmath490 on the trivial representation summand @xmath492 it is possible to explicitly write down oriented bases of @xmath428 and @xmath493 and see that @xmath494 preserves orientation on the summand @xmath492 however , that would lead to considerable notational complications .
our strategy is to modify the action of @xmath208 on @xmath428 and @xmath493 so that @xmath495 is still equivariant , but there are _ no _ copies of the trivial representation in the decomposition to irreducibles . to check equivariance
, we may work over the complex numbers , thus simplifying formulas .
choose coordinates @xmath496 and @xmath497 on @xmath498 such that under the action of @xmath208 , @xmath499 then @xmath72 is given explicitly by @xmath500 \mapsto [ z_1^p : z_2^p : 0 : 0 : 0 ] .\ ] ] let @xmath501 be a basis of @xmath502 dual to @xmath503 a section @xmath504 takes the form , @xmath505 where @xmath506 are linear functions on @xmath507 let @xmath508 be the basis of @xmath493 corresponding to @xmath509 then @xmath510 if we allow @xmath208 to act on @xmath428 by @xmath511 and on @xmath493 by @xmath512 then @xmath513 will remain equivariant .
the summands @xmath514 in decompositions both change to @xmath515 choosing @xmath153 small enough , we may assume no new trivial summands appear
. since the direct sum decomposition does nt change on the level of vector spaces as we change weights , we conclude that @xmath490 maps @xmath514 to @xmath514 preserving orientation .
hence , the cokernel of @xmath490 has equivariant euler class , @xmath516 now , @xmath517 contributes directly to @xmath518 as follows .
decomposing @xmath519 and using formulas and , we calculate @xmath520 and , @xmath521 in conclusion , @xmath522 on the other hand , by the euler class of the obstruction bundle is just @xmath523 here , we are assuming that the @xmath106 structure that @xmath457 induces on @xmath524 agrees with @xmath372 if for a given @xmath457 this is not true , reversing @xmath457 changes the sign of @xmath525 by @xmath261 for all @xmath237 by ( * ? ? ?
* lemma 2.10 ) .
so , we reverse @xmath465 thus reversing @xmath466 hence , the dependence of the total sign of @xmath526 on @xmath459 as claimed . combining everything ,
we obtain @xmath527 the extra factor of @xmath528 comes from the orbifold structure of the moduli space @xmath529 at @xmath442 $ ] .
a detailed construction was not required in @xcite to prove the invariance of @xmath21 . however , to apply the atiyah - bott localization formula as in the proof of theorem [ tm1 ] and proposition [ der ] , as well as for the obstruction bundle argument in the proof of theorem [ tm3 ]
, we need the following result .
[ pr : co ] @xmath183 is a smooth closed orbifold and @xmath182 is a smooth orbibundle .
first , we give a detailed definition of @xmath174 the moduli space @xmath179 is a smooth orbifold with corners .
a point in a corner of codimension @xmath36 corresponds to an open stable map with @xmath36 disk components .
such a stable map may have arbitrarily many sphere components .
however , since we consider open stable maps of genus @xmath401 each sphere component belongs to a tree of sphere components attached to a unique disk .
we define the _ total degree _ of a disk component to be its own degree plus the degree of all attached spheres components .
we classify corners of @xmath154 by the intersection types @xmath417 of the open stable maps .
the intersection type @xmath530 of an open stable map @xmath72 of genus zero consists of a tree @xmath531 and a labelling @xmath532 of the vertices of @xmath531 by non - negative integers .
the vertices of the tree correspond to disk maps and the edges correspond to nodes connecting two disks .
the labelling of a vertex is the total degree of the corresponding disk component .
we denote by @xmath533 the number of vertices of @xmath534 the codimension of the corresponding corner .
let @xmath417 be an intersection type , and let @xmath535 be an edge of @xmath531 connecting vertices @xmath536 and @xmath537 gluing @xmath417 at @xmath535 to obtain @xmath538 means contracting @xmath535 so that @xmath536 and @xmath539 become a single vertex @xmath540 of @xmath538 and defining @xmath541 we define a partial ordering on the set of intersection types by @xmath542 if @xmath543 may be obtained from @xmath544 by a sequence of gluings .
we use the notation @xmath545 to denote the corner of @xmath154 of intersection type @xmath546 we refer to codimension one corners as boundary components .
points of the boundary correspond to stable maps with two disk components .
the intersection type @xmath547 of such a stable map is essentially an unordered pair @xmath548 where the numbers @xmath549 are the total degrees of each disk component .
here @xmath550 for each possible such @xmath551 we define a smooth involution @xmath552 of the corresponding boundary component as follows . choose a vertex @xmath553 of @xmath554
since the total degree is odd , @xmath553 can be chosen as the unique vertex of odd degree .
for @xmath284 \in \partial_i { m}_d({\mathbb{c}}{\mathbb p}^4/{\mathbb{c}}{\mathbb p}^4_{\mathbb r},d),\ ] ] define @xmath555 to be the open stable map obtained by replacing the disk component corresponding to @xmath553 and all attached sphere maps by their conjugates .
we extend @xmath552 to the closure @xmath556 as follows .
let @xmath442 \in \partial_{i ' } m_d({\mathbb{c}}{\mathbb p}^4/{\mathbb{c}}{\mathbb p}^4_{\mathbb r},d)$ ] for some @xmath557 let @xmath558 be the set of vertices of @xmath538 that glue to form @xmath559 we define @xmath555 to be the open stable map obtained by replacing the disk - maps corresponding to the @xmath560 and all attached spheres maps by their conjugates .
next , we define an equivalence relation @xmath111 on @xmath154 as follows
. let @xmath561,[q ]
\in \partial_{i ' } m_d({\mathbb{c}}{\mathbb p}^4/{\mathbb{c}}{\mathbb p}^4_{\mathbb r},d).$ ] we write @xmath562 if there exists @xmath563 with @xmath564 such that @xmath565 finally , we can define @xmath566 we now prove @xmath183 is a closed orbifold .
we refer the reader to ( * ? ? ?
* section 2 ) for a quick review of orbifolds .
we essentially follow the notation established there .
let @xmath567 denote the quotient projection .
being a closed orbifold is a local property , so we restrict our attention to small neighborhood of a point @xmath568 \in \widetilde{m}_d({\mathbb{c}}{\mathbb p}^4/{\mathbb{c}}{\mathbb p}^4_{\mathbb r},d).\ ] ] if @xmath569)$ ] lies away from the corners of @xmath570 there is nothing to prove .
if @xmath569)$ ] meets the corners , we construct an orbifold chart in a neighborhood of @xmath561 $ ] in a canonical way .
let @xmath571 \in \pi^{-1}([p]).$ ] we assume @xmath572 \in \partial_{i_p } m_d({\mathbb{c}}{\mathbb p}^4/{\mathbb{c}}{\mathbb p}^4_{\mathbb r},d ) , \ \ k = |i_p| - 1 \geq 1.\ ] ] let @xmath573 be an orbifold chart on @xmath154 at @xmath574 here @xmath575 is a neighborhood of @xmath150 in @xmath576 @xmath577 is a finite group acting on @xmath575 , and @xmath578 is a @xmath577 invariant homomorphism from @xmath575 to a neighborhood of @xmath571.$ ] from the definition of the orbifold structure on @xmath570 the groups @xmath577 for @xmath571 \in \pi^{-1}([p])$ ] are all isomorphic .
so , we may define a group @xmath579 with isomorphisms @xmath580 we define , @xmath581 by definition of an orbifold with corners , @xmath582 is contained in a subset of @xmath583 where @xmath584 of the coordinates @xmath585 are zero . by definition of a smooth map of an orbifold , possibly shrinking the charts @xmath586 we may assume that @xmath552 induces a smooth @xmath579 equivariant involution of the disjoint union of the corners @xmath587 over all @xmath571 \in \pi^{-1}([p]).$ ] let @xmath588 enumerate the set of @xmath589 such that @xmath590 the intersection types of the boundary components adjacent to @xmath262 for a multi - index @xmath591 we define an involution @xmath592 of @xmath593 by @xmath594 so , the group @xmath595 acts on @xmath596 and by definition of @xmath597 acts transitively on @xmath569).$ ] define @xmath598 to be the stationary subgroup of @xmath571 \in \pi^{-1}([p]).$ ] the definition does not depend on the choice of @xmath571 $ ] because we are considering a transitive action of an abelian group .
define @xmath599 \in \pi^{-1}([p ] ) } v_{\tilde p } \times g_{p } , \qquad v_p = \hat v_p /\sim^*,\ ] ] where the equivalence relation @xmath600 is defined as follows .
let @xmath601 denote the multi - index with @xmath602 for @xmath603 and @xmath604 for @xmath605 define @xmath606 if @xmath607 and @xmath608 for some @xmath609 define @xmath610 the group @xmath611 acts naturally on @xmath612 finally , let @xmath613 be the @xmath611 invariant map from @xmath614 to @xmath183 naturally induced by the @xmath615 since @xmath614 is obtained by gluing together @xmath616 neighborhoods of @xmath617 on matching corners , @xmath614 is a neighborhood of @xmath618 the triple @xmath619 is easily seen to specify a natural orbifold structure in a neighborhood of @xmath262 the involutions @xmath552 lift naturally to the bundles @xmath620 then , the exact same proof extends to construct the structure of an orbibundle on @xmath621 in the following , @xmath369 denotes a @xmath106 structure on @xmath456 @xmath457 denotes a @xmath106 structure on @xmath458 and @xmath459 denotes a @xmath22 structure on @xmath23 [ lm : sp ] any two of @xmath622 determines the third . by the adjunction formula ,
the normal bundle @xmath623 of @xmath3 in @xmath624 satisfies @xmath625 so , the lemma follows from the exact sequence , @xmath626 and ( * ? ? ?
* lemma 8.1 ) .
we denote the determinant of the tangent bundle of @xmath183 by @xmath627 [ lm : ro ] there exists a topological isomorphism @xmath628 moreover , such an isomorphism is determined canonically up to homotopy by the choice of a @xmath22 structure on @xmath23 choose @xmath369 and @xmath457 compatible with @xmath629 then @xmath369 induces an orientation on @xmath179 and @xmath457 induces an orientation on @xmath630 so , there exists a unique up to homotopy isomorphism , @xmath631 preserving orientation .
we must check the identifications involved in the construction of @xmath183 are compatible with the above isomorphism : the sign of @xmath552 on @xmath632 must be the same as the sign of @xmath552 on @xmath633 this follows easily from ( * ? ? ?
* lemma 2.12 ) .
-[kur2 ] , we assume the coordinate transforms and kuranishi maps are smooth , as explained in ( * ? ? ? * appendix a1.4 ) . we anticipate that with the completion of the generalized fredholm theory currently being introduced in @xcite , the ideas of section [ kur1]-[kur2 ] will translate into a proof of theorem [ tm3 ] based on that framework . ] [ kur1 ] we denote by @xmath634 the moduli space of open stable maps to @xmath635 in general , the space @xmath634 is a compact metrizable space . in the following proof ,
we use the theory of kuranishi structures with corners developed in @xcite to define intersection theory on @xmath636 the kuranishi structures used here were shown to exist in @xcite .
see ( * ? ? ?
* appendix a and section 7 ) for a very brief summary of this theory , from which we take our notational conventions . in the following , unless explicitly noted , all kuranishi structures are kuranishi structures with corners .
we will need the following definition , which is similar to the notion of an involution of a kuranishi structure ( * ? ? ?
* definition 7.1 ) , but without property ( e1 ) .
suppose @xmath637 and @xmath638 are spaces with kuranishi structure @xmath639 let @xmath640 be a continuous map .
an _ extension _
@xmath641 of @xmath72 to a map of spaces with kuranishi structure consists of @xmath611-equivariant maps @xmath642 covering @xmath643 such that 1 .
@xmath644 2 .
@xmath645 3 .
@xmath646 maps @xmath647 to @xmath648 4 .
@xmath649 and @xmath650 now , suppose that @xmath637 has a tangent bundle given by @xmath651 and @xmath638 has a tangent bundle given by @xmath652 we say that @xmath641 is smooth if @xmath653 we say that @xmath641 is an embedding if @xmath643 are all embeddings and @xmath654 are all injective bundle maps .
the preceding definition of maps of spaces with kuranishi structure is very rigid and not likely to make a very nice category .
a better definition for a general morphism of spaces with kuranishi structure would be something like the diagram of figure [ fig : hat ] .
we will also need the notion of vector bundles over a space with kuranishi structure and their euler classes .
vector bundles over a space with kuranishi structure were constructed in a very general form in ( * ? ? ?
* section 5 ) so as to include the tangent bundle to a kuranishi structure .
the bundles we use here correspond to the special case in which @xmath655 of @xcite is taken to have rank @xmath150 everywhere .
we let @xmath637 denote a general space with kuranishi structure as above .
a _ vector bundle _
@xmath321 over @xmath637 consists of 1 . for each @xmath656 a @xmath611-equivariant vector bundle
@xmath657 2 . for each @xmath658 and @xmath659 an @xmath660-equivariant vector bundle isomorphism @xmath661 covering @xmath662 the euler class of a vector bundle over a space with kuranishi structure should determine a cohomology class in a cohomology theory for kuranishi spaces .
however , such a theory has not been developed .
since the euler classes considered here always have critical dimension , essentially all the information in the euler class is contained in a single number : the integral of the euler class over the fundamental class .
so , we focus on defining the integral of the euler class .
let @xmath321 denote a vector bundle over @xmath637 of rank equal to the expected dimension of @xmath663 let @xmath664 denote the determinant of the tangent bundle of @xmath665 i.e. , the line bundle over @xmath637 determined locally by the line bundles @xmath666 assume that an isomorphism @xmath667 has been specified .
choose a transverse perturbation of the space with kuranishi structure @xmath663 see ( * ? ? ?
* theorem a.4 ) for a brief review of notation .
choose multi - valued sections @xmath668 of @xmath669 such that multi - valued section @xmath670 of @xmath671 is transverse .
let @xmath672 denote the 0-dimensional rational simplicial complex determined by the vanishing set of @xmath673 the orientation of @xmath672 is determined by the isomorphism .
let @xmath674 denote the rational weighted cardinality of @xmath675 we define the _ euler class _ of @xmath321 by @xmath676 } e(f ) = |\sigma|.\ ] ] straightforward cobordism arguments show the definition does not depend on the choice of section @xmath385 or the perturbation of kuranishi structure .
see ( * ? ? ?
* sections 4 and 17 ) .
we continue to employ the notation of the proof of proposition [ pr : co ] .
since @xmath634 consists of open stable maps , we may define @xmath677 to be the subspace consisting of all open stable maps of intersection type @xmath546 similarly , we may define the involution @xmath552 of the corner of intersection type @xmath538 for @xmath678 and the quotient space @xmath679 let @xmath680 be a kuranishi structure with corners on @xmath636 by the arguments of ( * ? ? ?
* section 7 ) , the involutions @xmath552 extend smoothly to the kuranishi structure @xmath681 recapitulating the proof of proposition [ pr : co ] , @xmath680 induces a kuranishi structure without boundary @xmath682 on @xmath679 a transverse perturbation of the space with kuranishi structure @xmath683 defines a simplicial complex consisting of a finite number of @xmath150-simplices with rational weights .
the weighted count of the @xmath150-simplices is @xmath684 while the definition is not exactly the same as the definition given in @xcite , the equivalence is not hard to verify .
an orbifold structure is a special case of a kuranishi structure for which all the bundles @xmath685 are rank 0 .
let @xmath686 denote the kuranishi structure on @xmath687 coming from the orbifold structure , and let @xmath688 denote the kuranishi structure on @xmath689 coming from the orbifold structure constructed in proposition [ pr : co ] .
also , let @xmath690 denote the natural inclusion .
we would like to construct a kuranishi structure @xmath691 on @xmath689 for which we have the diagram of spaces with kuranishi structure shown in figure [ fig : hat ] .
the structure @xmath691 may be obtained from a kuranishi structure on @xmath687 that admits extensions of the involutions @xmath552 for which the diagram of figure [ fig : hat ] holds with tildes replaced by bars . for each @xmath561 \in \overline{m}_d(q / q_{\mathbb{r}},d),$ ] we extend the kuranishi neighborhood @xmath692 given by @xmath680 to a kuranishi neighborhood @xmath693 for the space @xmath694 we detail the construction of the extension for @xmath237 an irreducible stable map .
the construction for @xmath237 a reducible stable map is similar but notationally more complicated .
let @xmath695 and @xmath696 denote the banach manifolds of @xmath697 maps @xmath698 respectively .
let @xmath699 be defined fiberwise by @xmath700 let @xmath701 ( resp .
@xmath702 ) denote the section of @xmath703 ( resp .
@xmath704 ) given by the non - linear cauchy - riemann operator on maps to @xmath3 ( resp .
@xmath624 ) .
let @xmath705 and @xmath706 denote choices of the vertical parts of their respective linearizations .
we briefly outline the construction @xcite of the kuranishi neighborhood @xmath707 in order to explain how to extend it .
first , choose a finite dimensional subspace @xmath708 so that @xmath709 is surjective .
extend @xmath710 to a vector bundle @xmath711 over a neighborhood of @xmath237 in @xmath695 by parallel transport . over a sufficiently small neighborhood of @xmath712 @xmath705 surjects onto @xmath713 so
, we may define a smooth manifold @xmath714 we define @xmath715 to be an appropriate section of the action of the infinitesimal reparametrization group on @xmath716 then , we take @xmath717 and @xmath718 the group @xmath719 arises from the remaining discrete part of the reparametrization group of @xmath262 to extend the kuranishi neighborhood @xmath707 to a kuranishi neighborhood for @xmath720 we extend @xmath721 by parallel translation to a vector bundle @xmath722 over a neighborhood of @xmath237 in @xmath723 since @xmath624 is a homogeneous space , the operator @xmath706 is surjective onto @xmath724 at every @xmath561 \in m_d({{\mathbb{c}}}{\mathbb p}^4/{{\mathbb{c}}}{\mathbb p}^4_{\mathbb{r}},d).$ ] so , we may define a smooth manifold @xmath725 the definitions of @xmath726 and @xmath727 are just as before . in the case when @xmath237 is a reducible stable map , the extension procedure is similar except for one extra detail : we must be careful to perform parallel translations using a @xmath120-invariant metric so that the involutions @xmath552 extend to the corners of the extended kuranishi neighborhood .
see ( * ? ? ?
* section 7 ) for a discussion of this issue .
for each point in @xmath561 \in \overline{m}_d(q / q_{\mathbb{r}},d)$ ] we have just constructed a kuranishi neighborhood @xmath693 for @xmath694 it is straightforward to extend transition data @xmath728 to these extended neighborhoods . in order to complete the construction of @xmath729
it remains to define kuranishi neighborhoods of points @xmath568 \in \overline
m_d({{\mathbb{c}}}{\mathbb p}^4/{{\mathbb{c}}}{\mathbb p}^4_{\mathbb{r}},d)\setminus \overline m_d(q / q_{\mathbb{r}},d).\ ] ] for such points , we take an orbifold coordinate chart as a kuranishi neighborhood , letting @xmath685 be a trivial bundle of rank zero . because the bundles @xmath685 are rank zero , it is again easy to construct the associated transition data @xmath730 the existence of embeddings as in the diagram of figure [ fig : hat ] follows immediately from the construction of @xmath731 we proceed to extend the bundle @xmath182 of section [ ssec : q3d ] to a bundle @xmath732 over the space with kuranishi structure @xmath733 again , it suffices to construct a bundle @xmath734 over the space with kuranishi structure @xmath735 so that @xmath734 admits an extension of the involutions @xmath736 we define the fiber of @xmath737 at @xmath738 as follows . by construction , @xmath72 is a @xmath697-stable map satisfying the equation @xmath739 since @xmath72 may not be holomorphic , we have to be careful how we define the complex structure on @xmath740 choose a @xmath120-invariant metric on @xmath741 the associated complex connection on @xmath742 will also be @xmath120-invariant . equip @xmath743 with the complex structure induced from the @xmath744 part of the pull - back connection . then , we define as before , @xmath745 the transition functions @xmath746 are tautological .
because we used a @xmath120-invariant connection to induce the complex structure on @xmath747 the involutions @xmath552 lift to involutions of @xmath748 for @xmath749 hence , @xmath734 descends to a bundle @xmath732 over @xmath750 furthermore , the proof of lemma [ lm : ro ] shows that the isomorphism @xmath751 extends to an isomorphism @xmath752 let @xmath753 be the section defining the hypersurface @xmath754 for general @xmath755 the pull - back @xmath756 is not a holomorphic section of @xmath757 let @xmath758 denote the @xmath413 projection with respect to a @xmath120-invariant metric .
we define a section @xmath759 of @xmath737 by @xmath760 the local sections @xmath759 clearly match under the transition functions @xmath746 to define a global section @xmath761 of @xmath762 since @xmath385 is @xmath120-invariant and @xmath763 is defined with respect to a @xmath120-invariant metric , we conclude that @xmath761 is compatible with the involutions @xmath736 so , @xmath764 descends to a section @xmath765 of @xmath766 [ lm : tr ] if the kuranishi neighborhoods @xmath614 are chosen sufficiently small , then the sections @xmath767 vanish precisely on the image of the embedding of spaces with kuranishi structure @xmath768 moreover , the sections @xmath767 are transverse to zero .
we postpone the proof of lemma [ lm : tr ] until we complete the proof of theorem [ tm3 ] .
indeed , we calculate @xmath769}e(\widetilde f_d)\ ] ] in two different ways .
first , we construct a transverse perturbation @xmath770 of @xmath771 let @xmath772 denote the @xmath150-dimensional simplicial complex defined by @xmath773 by definition @xmath774 now , extend @xmath770 to a transverse perturbation @xmath775 of @xmath776 by lemma [ lm : tr ] , the local sections @xmath777 of @xmath778 vanish transversely exactly on @xmath779 assuming the following lemma , this implies that @xmath780}e(\widetilde f_d ) = n_d^{disk},\ ] ] [ lm : ork ] the orientation induced on each zero simplex in @xmath772 by the kuranishi structure @xmath781 agrees with the orientation induced by the kuranishi structure @xmath691 and the vector bundle @xmath766 we postpone the proof of lemma [ lm : ork ] until the end of the section . on the other hand , the sections @xmath782 of @xmath783 are transverse without any perturbation since @xmath624 is convex .
so , we may choose a transverse multi - valued section @xmath784 of @xmath785 and extend it to a multi - valued section @xmath786 of @xmath732 over @xmath787 such that the local sections @xmath788 will be transverse .
this shows that @xmath789}e(\widetilde f_d ) = \int_{{\widetilde m}_d({{\mathbb{c}}}p^4/{{\mathbb{c}}}p^4_{\mathbb{r}},d)}e(f_d).\ ] ] combining equations and , we deduce theorem [ tm3 ]
. we may focus on @xmath767 for @xmath561
\in \overline{m}_d(q / q_{\mathbb{r}},d).$ ] indeed , for other @xmath712 the kuranishi neighborhood @xmath614 is just an orbifold chart on @xmath790 so , for all @xmath791 the pull - back @xmath756 is holomorphic and non - zero .
so , @xmath792 is never zero . in case
@xmath568 \in \overline{m}_d(q / q_{\mathbb{r}},d),\ ] ] we need to show that @xmath759 vanishes transversely on @xmath793 but nowhere else .
first , we establish some notation .
let @xmath794 denote a complex @xmath120-invariant connection on @xmath795 and let @xmath796 denote its pull - back to @xmath797 let @xmath798 and @xmath799 denote the banach space bundles with fibers @xmath800 define a map of banach - space bundles @xmath801 by @xmath802 since @xmath803 and @xmath804 is surjective for all @xmath446 we have a short exact sequence @xmath805 let @xmath806 denote the unique right inverse of @xmath323 such that the image of @xmath807 is the @xmath413 complement of @xmath808 let @xmath809 denote the @xmath413 projection .
define a section @xmath810 of @xmath811 by @xmath812 throughout the following , we use @xmath813 to denote a context dependent norm .
that is , for sections of banach space bundles , @xmath814 is the appropriate banach space norm and for operators , @xmath813 is the appropriate operator norm . reformulating the definition of @xmath767 given in ,
we have @xmath815 we will argue that for @xmath72 close enough to @xmath712 we have @xmath816 for arbitrary epsilon . here , @xmath817 denotes an arbitrary distance function .
since @xmath385 vanishes transversely at @xmath818 we know that @xmath810 vanishes transversely on @xmath819 that is , we know that @xmath820 choosing @xmath821 estimate shows that after we perturb @xmath822 by @xmath823 to obtain @xmath824 it is still transverse and vanishes only on @xmath825 in order to prove estimate , we calculate @xmath826 let @xmath827 denote the complex structure on @xmath828 let @xmath829 denote the complex structure on @xmath624 and let @xmath417 denote the complex structure on @xmath830 assume that @xmath831 we calculate , @xmath832 the third equality uses equation and the fourth equality uses the holomorphicity of @xmath385 to cancel @xmath417 and @xmath833 composition with @xmath834 defines a linear map @xmath835 observe that for @xmath836 we have @xmath837 since @xmath385 vanishes on @xmath818 we see that @xmath838 maps @xmath839 to zero for @xmath840 by continuity , we have @xmath841 so , we infer from calculation that @xmath842 furthermore , we can assume a uniform bound @xmath843 for all @xmath844 for any @xmath845 we can choose @xmath614 so small that for all @xmath846 we have @xmath847 so , choosing @xmath848 such that @xmath849 and combining estimates , and , we conclude estimate .
this completes the proof of lemma [ lm : tr ] .
the following proof is a generalization of the argument given in ( * ? ? ?
* section 8 , proposition 8.8 ) . in the following ,
we will abbreviate @xmath850 we continue to use the bundles @xmath429 and @xmath851 introduced in the proof of lemma [ lm : tr ] . since @xmath852 and by lemma [ lm : tr ] , @xmath764 is transverse to @xmath401 we conclude that @xmath853 induces and isomorphism @xmath854 along @xmath819 using the fact that by construction @xmath855 we can tensor the above isomorphism with @xmath856 to obtain an isomorphism @xmath857 the lemma will follow if we show that isomorphism respects the canonical orientations of each of the three deteminant bundles @xmath858 and @xmath859 for this purpose , we introduce the commutative diagram of vector bundles over @xmath860 of figure [ fig:3d ] . here , we implicitly consider the restrictions of @xmath861 and @xmath685 to @xmath819 we denote by @xmath862 and @xmath863 the canonical projections to the quotient . in order that all squares in the diagram of figure [ fig:3d ] commute , we must choose the complex connection @xmath794 on @xmath742 carefully .
indeed , @xmath864 induces an isomorphism of complex vector bundles @xmath865 so , we are free to choose @xmath794 to be the complex connection induced by @xmath864 from the connection on @xmath866 induced by the levi - civita connection of @xmath867 with respect to the standard kahler metric .
this ensures that @xmath868 i.e. that the middle right horizontal square of the diagram commutes .
moreover , the proof of lemma [ lm : tr ] implies that @xmath869 i.e. that the front upper right vertical square of the diagram commutes .
the commutativity of the remaining squares is straightforward .
moreover , all columns and rows of figure [ fig:3d ] are exact .
we think of the diagram as an exact square of two - step complexes . applying the determinant functor to the long exact sequence of a short exact sequence of two step complexes
yields an isomorphism of determinant bundles . in particular ,
the five non - trivial rows and columns of figure [ fig:3d ] give isomorphisms @xmath870 taking determinants of the rows and columns of the diagram of figure [ fig : sn ] , we obtain isomorphisms @xmath871 clearly the last isomorphism of agrees with the last isomorphism of . putting all the isomorphisms of and together ,
we obtain the diagram of figure [ fig : det ] .
@xmath872\ar[d]^{d\check\xi } & \det(ds_p^q ) \ar[r]\ar[d ] & \det(d\bar\partial^q ) \ar[d ]
\\ { \mathcal{l}}_{\mathcal k}\otimes \det(f_d)^ * \ar[r ] & \det(ds_p)\otimes \det(f_d)^ * \ar[r ] & \det(d\bar\partial)\otimes \det(d'')}\ ] ] the left square of figure [ fig : det ] commutes by the commutativity of figure [ fig : sn ] .
the right square of figure [ fig : det ] commutes by the commutativity of figure [ fig:3d ] .
the rows of figure [ fig : det ] preserve orientation by definition .
the right column of figure [ fig : det ] preserves orientation by ( * ? ? ?
* section 8 , proposition 8.4 ) and the compatibility of @xmath873 and @xmath466 by commutativity the left column of figure [ fig : det ] also preserves orientation , completing the proof .
consider @xmath134 with the anti - holomorphic involution @xmath120 defined in section [ ccc2 ] .
the involution @xmath120 lifts canonically to @xmath874 .
the total space of the rank 2 bundle @xmath875 with the associated anti - holomorphic involution may be viewed as local model for a rational curve in @xmath3 .
the local disk invariants of @xmath134 are , by definition , @xmath876 where @xmath877 is the real vector bundle over with fiber @xmath878 } = h^1(c,\tilde{f}^*\mathcal{o}_{{\mathbb{c}}{\mathbb p}^1}(-1))_{\mathbb{r}}.\ ] ] here ,
@xmath879 \in \overline{m}_{\mathbb r}({\mathbb{c}}{\mathbb p}^1,d)\ ] ] is the stable rational map obtained from the stable disk map via reflection . as before , we consider only the @xmath5 odd case . [ der ] for @xmath5 odd , @xmath880 the factor of 2 on the right occurs since the original @xmath134 consists of 2 disks .
hence , proposition [ der ] may be viewed as the calculation of twice the multiple cover contribution of a single disk .
we emphasize that @xmath5 is the degree of the stable rational map obtained from the stable disk map by reflection .
in particular , the degree of a map @xmath72 in @xmath881 restricted to @xmath143 may be any odd integer less than or equal to @xmath882 this is necessary in order to perform the analogue of the construction of proposition [ pr : co ] for target space @xmath883 as the diffeomorphisms @xmath552 do not preserve boundary degree .
in this regard , our approach differs from the approach of @xcite .
let the torus @xmath196 act on @xmath134 by @xmath884 = [ \xi u , \overline{\xi } v].\ ] ] the fixed points are @xmath885 , \
\ \zeta_2=[0,1]\ ] ] with tangent weights @xmath886 and @xmath887 respectively .
the @xmath196-action preserves @xmath144 , and therefore determines a translation action on the moduli space of disks @xmath881 .
the @xmath196-action lifts to @xmath874 with fiber weights @xmath888 and @xmath352 over @xmath482 and @xmath483 respectively , the unique lift which respects the real structure on @xmath874 .
we give two different proofs .
the invariants @xmath889 are calculated by @xmath196-equivariant localization .
since the steps are so similar to the proof of theorem 1 , we give an abbreviated account .
as before , to each map @xmath442 \in \overline{m}_d({\mathbb{c}}{\mathbb p}^1/{\mathbb{c}}{\mathbb p}^1_{\mathbb{r}},d)^{\mathbf t}$ ] , we associate an intersection disk .
the intersection disk terms @xmath526 and @xmath890 are both equal to @xmath891 the sign @xmath263 comes from the normal bundle to the disk map just as in the proof of lemma [ ft1 ] .
we do not have to calculate the orientation of @xmath892 because it appears twice .
givental s equivariant correlator @xmath893 for the local geometry is defined by @xmath894 where @xmath895 is the evaluation map .
the calculation of @xmath893 in @xcite is much easier than @xmath230 , @xmath896 no mirror transform is needed .
the local disk potential @xmath897 is defined by summing over odd degrees , @xmath898 as before , @xmath899
\rangle \cdot i(\zeta_1,p)+ \sum_{p\ odd}\langle s_l(t,-\frac{2}{p}\lambda ) , [ \zeta_2 ] \rangle \cdot i(\zeta_2,p).\end{gathered}\ ] ] evaluation yields @xmath900 the proof of the proposition is concluded by extracting the @xmath901 terms on the right and executing the sum , @xmath902 remarkably , the binomial identity required is exactly !
although each copy of @xmath903 admits a unique real @xmath196-action , there is some freedom in the choice of action on the sum @xmath904 indeed , we can choose the action induced by the isomorphism of vector bundles with real structure , @xmath905 here , @xmath906 and @xmath350 is the real representation of the torus of weight @xmath438 with the @xmath196-action determined by the right side of , the localization contribution from any _ reducible _ open stable map vanishes .
indeed , we work this out explicitly for a torus fixed open stable map @xmath72 with reducible domain @xmath146 consisting of one disk component @xmath907 and one sphere component @xmath908 the general case is similar . denote the single node of @xmath146 by @xmath909 consider the exact sequence @xmath910 we claim that @xmath911 contains a zero weight space and therefore so does @xmath912 which immediately implies that the localization contribution of @xmath442 $ ] vanishes .
indeed , @xmath913 is just the fiber of @xmath914 at @xmath909 denoting by @xmath915 the complex representation of the torus of weight @xmath352 we have @xmath916 hence , @xmath917 as claimed .
it remains to calculate the localization contribution from the single torus fixed irreducible open stable map of degree @xmath882 this is easily seen to be @xmath918 the two proofs of proposition [ der ] together provide a geometric evaluation of the binomial sum of lemma [ idd ] .
the virtual disk counts @xmath919 of definition [ tm2 ] have not yet been proven to be integers .
however , the table in section [ xxx ] provides substantial evidence for the integrality claim .
table [ t1 ] shows the value of the disk gromov - witten invariant @xmath21 for small @xmath882 | holomorphic disk invariants with boundary in the real lagrangian of a quintic 3-fold are calculated by localization and proven mirror transforms .
a careful discussion of the underlying virtual intersection theory is included .
the generating function for the disk invariants is shown to satisfy an extension of the picard - fuchs differential equations associated to the mirror quintic .
the ooguri - vafa multiple cover formula is used to define virtually enumerative disk invariants .
the results may also be viewed as providing a virtual enumeration of real rational curves on the quintic . |
let @xmath0 denote a finite weyl group with root system @xmath5 , and let @xmath1 denote the corresponding affine weyl group , acting on a real vector space @xmath6 .
they are the most important and classical reflection groups .
the affine coxeter arrangement of @xmath0 gives a regular tessellation of @xmath6 .
define a random walk @xmath7 in the alcoves , called the _ reduced random walk_. we start at the fundamental alcove and at each step we cross one adjacent hyperplane chosen uniformly at random , subject to the condition that we never cross a hyperplane twice
. see figure [ fig : walk ] ( 0,1 ) rectangle ( 8,9 ) ; ( @xmath8 )
( @xmath9 ) ; ( @xmath10 ) ( @xmath11 ) ; ( @xmath12 ) ( @xmath13 ) ; ( @xmath14 ) ( @xmath15 ) ( @xmath16 ) ; ( 0,0 ) + + ( 90:2/3 ) + + ( 150:2/3 ) + + ( 90:2/3 ) + + ( 150:2/3 ) + + ( 210:2/3 ) + +
( 150:2/3) + + ( 210:2/3 ) + + ( 150:2/3 ) + + ( 90:2/3 ) + +
( 150:2/3 ) ; at ( 0,1 ) @xmath17 ; at ( @xmath18 ) @xmath19 ; at ( @xmath20 ) @xmath21 ; at ( 0,-1 ) @xmath17 ; at ( @xmath22 ) @xmath19 ; at ( @xmath23 ) @xmath21 ; at ( @xmath24 ) @xmath19 ; at ( @xmath25 ) @xmath21 ; at ( @xmath26 ) @xmath17 ; at ( @xmath27 ) @xmath17 ; in -2,-1,0,1 at ( @xmath28 ) @xmath21 ; at ( @xmath29 ) @xmath19 ; at ( @xmath30 ) @xmath17 ; at ( @xmath31 ) @xmath19 ; at ( @xmath32 ) @xmath21 ; at ( @xmath33 ) @xmath17 ; in 0 , 1 , ... , 10 ( @xmath34 ) ( @xmath35 ) ; in -5,-4,-3,-2,-1,0,1,2,3,4,5,6,7,8,9,10 ( @xmath36 ) ( @xmath37 ) ; in 1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16 ( @xmath36 )
( @xmath38 ) ; ( 0:0 ) ( 150:5.1 ) ; this process is a transient markov chain .
more algebraically , it is equivalent to a random infinite reduced word for @xmath1 obtained by multiplying by simple generators one at a time , subject to the condition that the length increases .
non - random infinite reduced words in the affine weyl group have a beautiful structure theory , which we recently studied in relation to factorizations in loop groups @xcite .
we prove here that : [ t : main ] let @xmath39 be a reduced random walk in @xmath1 .
there exists a unit vector @xmath40 so that almost surely we have @xmath41 where @xmath42 denotes the unit vector pointing towards the central point of @xmath43 .
thus the reduced walk has one of finitely many asymptotic directions .
the random walk we study here is different to the walks on hyperplane arrangements that we have seen in the literature , see for example @xcite . in section [ ss : w ] , we define a markov chain on the finite weyl group @xmath0 . roughly speaking , this markov chain is obtained by projecting the affine grassmannian weak order onto @xmath0 . unlike the reduced random walk on @xmath1 , this markov chain is irreducible and aperiodic ( proposition [ p : sca ] ) and thus has a unique invariant distribution @xmath44 .
the vectors @xmath45 lie in different weyl chambers @xmath46 , and we let @xmath47 denote the event that the reduced random walk @xmath48 eventually stays in @xmath46 . the probabilities @xmath49 vary depending on @xmath50 : in @xmath51 , one weyl chamber is 96 times more likely than another .
the root system notation of the next theorem is reviewed in section [ ss : notation ] .
[ t : main2 ] the vector @xmath52 of theorem [ t : main ] is given by @xmath53 where @xmath54 is the highest root of @xmath0 and @xmath55 is a normalization factor .
furthermore , @xmath56 thus the invariant distribution @xmath57 determines two apparently unrelated quantities : the coordinates of the asymptotic directions , and the probabilities of each direction .
this surprising duality is ultimately related to the associativity of the demazure , or monoidal , product in a coxeter group . in section [ ss : shi ] , we give an alternative formula for @xmath58 , expressed as a calculation involving a sum over the regions of the _ shi arrangement _ of @xmath0 .
we also conjecture ( conjecture [ conj : a ] ) that in type @xmath59 the point @xmath52 of theorem [ t : main ] is in the same direction as @xmath60 . in joint work with williams @xcite
, we conjecture that a multivariate generalization of this markov chain on the symmetric group has remarkable schubert positivity properties . in the case of @xmath61 ,
our results can be interpreted in terms of @xmath4-core partitions .
recall that a young diagram is an @xmath4-core if no @xmath4-ribbon can be removed from it .
grow a random @xmath4-core from the empty partition by randomly adding boxes to the young diagram , subject to the condition that the shape is always an @xmath4-core .
the notation in the following theorem is explained in section [ sec : cores ] .
[ t : cores ] there exists a piecewise - linear curve @xmath62 , so that for each @xmath63 , there exists an @xmath64 such that for every @xmath65 , we have @xmath66 where @xmath67 is the diagram of a random @xmath4-core of degree @xmath68 . at
( -2,1.3 ) @xmath69 { { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } \\ { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } \\ { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } \\ { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } \\ { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } \\ { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } \\ { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } \\ { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } \\ { } & { } & { } & { } & { } & { } & { } & { } & { } & { } & { } \\ { } & { } & { } & { } & { } & { } & { } & { } & { } \\ { } & { } & { } & { } & { } & { } & { } & { } & { } \\ { } & { } & { } & { } & { } & { } & { } \\ { } & { } & { } & { } & { } & { } & { } \\ { } & { } & { } & { } & { } \\ { } & { } & { } & { } & { } \\ { } & { } & { } & { } & { } \\ { } & { } & { } & { } \\ { } & { } & { } & { } \\ { } & { } & { } & { } \\ { } & { } & { } \\ { } & { } & { } \\ { } & { } & { } \\ { } & { } \\ { } & { } \\ { } & { } \\ { } \\ { } \\ { } } } } $ ] ; ( 0,0)(0.5,1.5)(1.5,2.5)(3,3 ) ; ( 0,-0.5)(0,3)(3.5,3 ) ; conjecture [ conj : a ] ( verified for @xmath70 ) gives explicit coordinates for the curve @xmath71 .
theorem [ t : cores ] puts our work in the context of vershik and kerov s @xcite work .
they show that the shape of a plancherel - measure random partition approaches a limit shape which is a continuous curve .
this result is related to a diverse number of subjects , such as the expected length of an increasing subsequence of a random permutation ( ulam s problem ) , the asymptotic representation theory of the infinite symmetric group , and the distribution of the eigenvalues of random ( gue ) hermitian matrices .
it follows ( corollary [ cor : firstrow ] ) from conjecture [ conj : a ] that the first row of a large typical random @xmath4-core is asymptotic to @xmath72 , where @xmath68 is the number of boxes in the @xmath4-core . this is less than , and
should be compared to , the value @xmath73 @xcite in the random partition case .
note however that our measure on @xmath4-cores is _ not _ the analogue of plancherel measure ( see section [ ss : plancherel ] ) .
in this project we were initially motivated by the study of families of symmetric functions which represent schubert classes in the ( @xmath74)-cohomology of the affine grassmannian @xmath75 of @xmath76 @xcite .
these symmetric functions are `` affine '' analogues of schur functions , the latter playing a key role in the theory of random partitions @xcite .
our main result may have an interpretation in terms of large products @xmath77 of an element @xmath78 in the @xmath74-homology of the affine grassmannian
it describes the asymptotics of the `` spreading out '' over the affine grassmannian of products of this class under the pontryagin multiplication of a loop group ( see section [ ss : grass ] ) .
this connection to the infinite - dimensional geometry of @xmath75 has concrete probabilistic consequences : in a separate article we plan to apply this geometry to the calculation of the boundary of the affine grassmannian weak order .
* acknowledgements .
* we benefited from john stembridge s coxeter / weyl maple package .
for affine weyl groups we use the references @xcite . we denote the simple generators of @xmath0 by @xmath79 and by @xmath80 the longest element of @xmath0 .
let @xmath81 be the additional simple generator of @xmath1 .
the weyl group acts as linear reflections in a real vector space @xmath6 , and the affine weyl group act as affine reflections in @xmath6 .
we let @xmath82 denote the set of roots of @xmath0 , and let @xmath83 denote the decomposition into positive and negative roots .
the set @xmath84 of affine roots consists of the elements @xmath85 .
the roots @xmath86 are the real affine roots , and @xmath87 is positive if and only if either ( a ) @xmath88 and @xmath89 , or ( b ) @xmath90 and @xmath91 .
we denote the positive affine roots by @xmath92 .
the simple roots are denoted @xmath93 , and we have @xmath94 , where @xmath54 is the highest root .
we let @xmath95 denote the reflection in the hyperplane perpendicular to @xmath54 . to each real affine root @xmath96 , we associate the ( affine ) hyperplane @xmath97 .
the _ affine coxeter arrangement _ is the hyperplane arrangement consisting of all such @xmath98 .
we also associate to each real affine root @xmath87 a coroot @xmath99 .
the connected components of the complement of affine coxeter arrangement are known as alcoves .
the fundamental alcove @xmath100 is bounded by the hyperplanes corresponding to the simple roots .
there is a bijection @xmath101 between the alcoves and @xmath1 , and we shall pick conventions so that @xmath102 and @xmath103 are adjacent , separated by the hyperplane corresponding to @xmath104 .
the _ weyl chambers _ are the connected components of the complement to the finite coxeter arrangement , where only the @xmath105 s are used for @xmath106 .
the _ fundamental chamber _ is the weyl chamber containing the fundamental alcove .
affine weyl group elements corresponding to alcoves inside the fundamental chamber are called _ affine grassmannian_. we shall also need the right action @xmath107 of @xmath0 on the set of alcoves .
the right action of @xmath108 takes the fundamental chamber to the weyl chamber @xmath46 labeled by @xmath50 ( the one containing the alcove @xmath109 ) .
the elements in @xmath46 are of the form @xmath110 , where @xmath111 is an affine grassmannian element .
there is an isomorphism @xmath112 , where @xmath113 denotes the coroot lattice of @xmath0 .
if @xmath114 , we denote by @xmath115 the corresponding element in @xmath1 , called a translation element . for @xmath116 ,
we have @xmath117 the inversions @xmath118 of @xmath111 are exactly the real affine roots which are sent to negative roots .
equivalently , @xmath119 consists of the roots corresponding to hyperplanes separating @xmath103 from @xmath100 .
note that with these conventions , @xmath120 is obtained from @xmath100 by translation by the vector @xmath121 .
the _ left weak order _ on @xmath1 is given by @xmath122 if and only if @xmath123 .
we shall also write @xmath124 for the weak order applied to alcoves , and write @xmath125 for the cover relations .
we say that an alcove @xmath59 is _ of type @xmath50 _
if @xmath126 .
let @xmath127 be the half - sum of positive roots .
recall that @xmath114 is anti - dominant if @xmath128 for @xmath88 .
the following result is standard @xcite .
[ lem : grass ] suppose @xmath129 .
then @xmath111 is affine grassmannian if and only if @xmath130 is anti - dominant and for every @xmath88 such that @xmath131 we have @xmath132 .
we then have @xmath133 .
we define a random walk on alcoves .
the walk begins at @xmath134 .
given @xmath135 , we pick @xmath136 uniformly at random amongst the alcoves adjacent to ( that is , sharing a facet with ) @xmath137 , with the constraint that the hyperplane separating @xmath137 and @xmath136 has not been crossed previously .
it follows easily from coxeter group theory that such walks can never `` get stuck '' .
based on the definition , somewhat surprisingly , the process @xmath39 is a markov chain .
the hyperplanes that have been crossed during the first @xmath138 steps of the walk @xmath135 are exactly the hyperplanes separating @xmath137 from @xmath134 .
we call this process the _ random walk in @xmath1 _ , ( or sometimes the _ reduced random walk in @xmath1 _ ) , starting at the fundamental alcove
. we shall also consider the process @xmath139 where the random walk is constrained to stay within the fundamental weyl chamber .
we call this the _ reduced affine grassmannian random walk _ in @xmath1 . an infinite reduced word
@xmath140 is an infinite word such that @xmath141 is a reduced word for @xmath1 , for any @xmath142 .
the coxeter - equivalence of reduced words can be extended to _
braid limits _ of infinite reduced words .
it is known that any infinite reduced word @xmath143 of @xmath1 is braid equivalent to an infinite reduced word of the form @xmath144 , where @xmath145 is the reduced word of a translation element , and @xmath146 is a finite reduced word for @xmath1 ( see @xcite ) .
sequences @xmath39 of alcoves as considered in section [ ss : random ] are tautologically in bijection with infinite reduced words .
thus theorem [ t : main ] says that a random infinite reduced word @xmath143 is not only almost surely braid equivalent to @xmath147 for one of @xmath2-many @xmath145 s , but indeed that almost surely @xmath143 and @xmath147 asymptotically converge to the same point of the boundary of the tits cone ( cf .
* remark 4.5 ) ) .
we define a markov chain with finite state space @xmath0 , which appears to be of independent combinatorial interest .
let @xmath148 be the rank of @xmath1 .
the transition probability from @xmath50 to @xmath149 is given by @xmath150 where @xmath151 is chosen so that @xmath152 .
let @xmath153 denote the transition matrix .
let @xmath154 denote the directed graph on @xmath0 with edges given by the non - zero transitions .
let @xmath155 be the markov chain on @xmath154 with transition matrix @xmath156 .
\(0 ) [ label=[red]below : @xmath157 at ( 0,0 ) @xmath19 ; ( 12 ) [ label=[red]right : @xmath157 at ( 1,2 ) @xmath158 ; ( 21 ) [ label=[red]left : @xmath157 at ( -1,2 ) @xmath159 ; ( 1 ) [ label=[red]left : @xmath160 at ( -1,1 ) @xmath161 ; ( 2 ) [ label=[red]right : @xmath160 at ( 1,1 ) @xmath162 ; ( 121 ) [ label=[red]above : @xmath160 at ( 0,3 ) @xmath163 ; \(0 ) ( 121 ) ; ( 1)(12 ) ; ( 2)(21 ) ; ( 121)(21 ) ; ( 121)(12 ) ; ( 21)(1 ) ; ( 12)(2 ) ; ( 1 ) ( 0 ) ; ( 2)(0 ) ; [ p : sca ] the markov chain @xmath164 is strongly - connected and aperiodic .
aperiodicity is clear from the definition .
strong connectedness follows from ( * ? ? ?
* theorem 4.2 ) .
it follows that @xmath164 has a unique limit stationary distribution .
explicitly describe the stationary distribution @xmath165 of @xmath164 for each @xmath0 .
this distribution appears to have remarkable enumerative properties , especially for the symmetric group @xcite .
[ conj : aeigenvector ] let @xmath166 .
then @xmath167 is an integer for all @xmath168 , and @xmath169 .
[ rem : othertypes3 ] the integrality part of conjecture [ conj : aeigenvector ] fails for other types .
for example , it is false for @xmath0 of type @xmath170 .
however , the weighted version of @xmath154 , as described in remark [ rem : othertypes ] and section [ ss : grass ] , still appears to retain these properties .
[ rem : reduced_words ] let @xmath171 be the probability measure on length @xmath68 elements of @xmath1 , where @xmath172 is proportional to the number of reduced words of @xmath111 .
define @xmath173 by setting the diagonal entries of @xmath156 to 0 .
the matrix @xmath173 is a sub - stochastic matrix , which nevertheless calculates the projected measures @xmath174 after scaling .
( the matrix @xmath173 weights each path equally regardless of the valency of the vertices that it passes through . ) after scaling , and conjugation by a suitable diagonal matrix @xmath175 , one does obtain a markov chain with transition matrix given by @xmath176 .
the methods in this section will still prove corollary [ c : main ] for the measures @xmath171 ( but with a different limit @xmath52 ) .
let @xmath139 denote the affine grassmannian random walk of [ ss : random ] .
we let @xmath177 denote the delayed random walk , where @xmath178 has probability @xmath179 of being equal to @xmath180 , where @xmath148 is the rank of the affine weyl group , and @xmath151 is the number of facets of @xmath181 which separate @xmath181 from @xmath100 .
each of the transitions in the original random walk now have probability @xmath182 .
similarly define @xmath183 .
let @xmath184 be the projection given by @xmath185 .
the following proposition is a key observation of the paper .
[ p : markov ] the projection @xmath186 of the delayed affine grassmannian random walk is the markov chain @xmath164 , with initial condition @xmath187 .
the result follows from lemma [ lem : grass ] and lemma [ l : s0 ] below .
[ lem : high ] let @xmath188 . then @xmath189 .
the sum @xmath190 can be a root only if @xmath191 .
[ l : s0 ] suppose @xmath192 is affine grassmannian .
then @xmath193 in @xmath0 if and only if @xmath194 is affine grassmannian and @xmath195 .
suppose that @xmath196 .
let @xmath197 . to show that @xmath198 , we compute @xmath199 since @xmath130 is anti - dominant by lemma [ lem : grass ] . to show that @xmath200 is grassmannian we calculate for @xmath201 @xmath202 we need to show that the root @xmath203 is positive .
first suppose that @xmath204 .
then by lemma [ lem : grass ] , we have @xmath205 , so since @xmath54 is the highest root we must have @xmath206 by lemma [ lem : high ] .
if @xmath207 we are done . if @xmath208 , we must show that @xmath209 .
we calculate that @xmath210 .
but @xmath211 , so that @xmath212
. now suppose @xmath213 .
if @xmath214 then by lemma [ lem : high ] we have @xmath215 , so we would be done . if @xmath216 we note that @xmath217 so by lemma [ lem : high ] it suffices to assume that @xmath218 and show that @xmath219 .
but @xmath220 .
for the converse , let us suppose that @xmath221 .
let @xmath222 .
we have @xmath223 but @xmath224 , so by lemma [ lem : grass ] , we have @xmath132 . if @xmath225 , then @xmath226 is a negative root , so that @xmath227 .
otherwise , we have @xmath228 . in this case
, we calculate that @xmath229 but @xmath230 , so @xmath231 , and thus @xmath200 is not grassmannian .
let @xmath232 be a random walk on @xmath154 with transition matrix @xmath156 , and @xmath233 an edge in @xmath154 .
write @xmath234 for the number of times the edge @xmath235 is used in @xmath236 .
[ lem : ergodic ] we have @xmath237 almost surely .
this follows from the ergodic theorem for markov chains , see for example ( * ? ? ?
* corollary 4.1 ) .
we first establish the statement for the delayed affine grassmannian random walk @xmath177 .
outside a set of measure 0 ( those @xmath180 that eventually stop ) , @xmath180 naturally maps ( by removing repeats ) to the random walk @xmath238 defined in section [ ss : random ] .
let the projection of @xmath180 to @xmath0 be @xmath239 , which is a markov chain on @xmath154 by proposition [ p : markov ] . write @xmath240 , where @xmath241 .
the translation element @xmath242 only changes from @xmath243 to @xmath244 if @xmath245 . by lemma [ l : s0 ] , this corresponds to transitions @xmath246 in @xmath55 , which changes @xmath242 by @xmath247 ( using @xmath248 ) . for two edges @xmath249 , by lemma [ lem : ergodic ]
, the ratio @xmath250 converges almost surely to @xmath251 .
it follows that @xmath252 almost surely . the alcove @xmath253 shares a vertex with the alcove @xmath254 , and so @xmath255 points in almost the same direction as @xmath256 .
we thus obtain theorem [ t : main ] and the first statement of theorem [ t : main2 ] for the reduced affine grassmannian random walk @xmath238 .
now , the random walk @xmath257 will eventually stay in some weyl chamber , since each weyl chamber is separated from the fundamental alcove by some hyperplanes which can be crossed at most once , and there are finitely many weyl chambers .
the asymptotic direction of @xmath238 does not depend on initial point of the random walk , but only the constraint that the walk remains inside the fundamental chamber and heads away from @xmath100 .
thus if we know that @xmath47 , we can apply the right action of @xmath0 to the part of @xmath48 lying inside @xmath46 to get a random walk in the fundamental chamber which almost surely has asymptotic direction @xmath52 , completing the proof .
the almost sure convergence of theorem [ t : main ] implies convergence in probability .
pick a norm on @xmath6 .
[ c : main ] for each @xmath258 and @xmath259 , there is a @xmath260 so that @xmath261 for @xmath65 .
it follows from the proof of theorem [ t : main ] that the point @xmath52 has rational coordinates , when written in terms of simple coroots .
this implies that there is a translation element of @xmath1 which points in the same direction as @xmath52 . in theorem [ t : main ] and corollary [ c : main ] only the limiting direction
is discussed .
the formula in lemma [ lem : grass ] for the length @xmath262 of a translation element allows us to calculate the speed that the random walk is traveling from the fundamental alcove .
we give an explicit conjecture for @xmath52 when @xmath166 . in the next result
we treat @xmath263 as a point in @xmath6 by identifying @xmath6 and @xmath264 in the usual way .
[ conj : a ] for @xmath166 , we have @xmath265 for some @xmath266 .
[ rem : othertypes ] conjecture [ conj : a ] does not hold as stated for other types .
define @xmath267 by @xmath268 , and set @xmath269 .
now , weight the transitions corresponding to left multiplication by @xmath270 by a factor of @xmath271 .
then our computations suggest that conjecture [ conj : a ] still holds for type @xmath272 , and that it is close to holding in other types .
the coefficients @xmath271 here are connected via affine dynkin diagram duality to the coefficients @xmath273 that we expected to see for reasons related to the topology of the affine grassmannian ; see section [ ss : grass ] .
the duality may be an artifact of our choice of @xmath113 instead of @xmath274 for the definition of an affine weyl group .
let @xmath7 be the reduced random walk in @xmath1 .
write @xmath47 for the event that @xmath48 eventually stays in the weyl chamber @xmath46 . write @xmath275 if @xmath276 and the type of @xmath277 is @xmath149 .
we use the same notation for the delayed random walk @xmath183 .
the reverse of the random walks @xmath48 or @xmath183 is a very different process to the original process .
for example , @xmath48 can go in many directions , at least at the beginning of the walk , but reversing @xmath48 gives a walk which heads towards the fundamental chamber .
thus the next result is very surprising .
it relies on a very special feature of coxeter groups , namely the associativity of the demazure product .
let @xmath74 denote the affine @xmath17-hecke algebra of @xmath1 ( see @xcite ) , with generators @xmath278 , a @xmath279-basis @xmath280 where @xmath281 , satisfying the multiplication formulae @xmath282 if @xmath283 , and @xmath284 otherwise , and also @xmath285 if @xmath286 , and @xmath284 otherwise . in the following we will freely identify alcoves with elements of @xmath1 .
[ l : reverse ] for each @xmath287 , we have @xmath288 , and @xmath289 . let @xmath290 .
then @xmath291(\xi)^n$ ] where @xmath292 $ ] denotes the coefficient of @xmath293 when an element of @xmath74 is written in the basis @xmath294 . but
the element @xmath78 of @xmath74 is invariant under the algebra anti - morphism @xmath295 of @xmath74 .
it follows that the coefficient of @xmath293 and @xmath296 in the product @xmath297 coincides .
restricting to elements with length @xmath68 gives the second statement .
we call @xmath298 regular if @xmath114 is regular , that is , the stabilizer subgroup of @xmath0 acting on @xmath130 is trivial .
[ l : reg ] suppose @xmath299 is regular
. then @xmath300 .
if @xmath299 is regular then @xmath301 , where @xmath302 is a regular and anti - dominant .
then @xmath303 , and @xmath304 is anti - dominant .
it is clear that @xmath305 , so we shall focus on the delayed walk .
let @xmath306 . then by the argument in section [ ss : proof
] , we have @xmath307 it follows from theorem [ t : main ] that @xmath308 as @xmath309 . thus using lemma [ l : reverse ] and [ l : reg ] , for each @xmath310 we can find @xmath68 sufficiently large so that @xmath311 it follows that @xmath312 for every @xmath313 .
we note that setting @xmath314 solves this equation , and since @xmath315 is a probability measure on @xmath0 this must be the solution .
the ideas here are related to the language of reduced words in affine coxeter groups , see for example @xcite .
shi arrangement _ is the hyperplane arrangement consisting of the hyperplanes @xmath316 .
one of the regions ( connected components of the complement ) of the shi arrangement is exactly the fundamental alcove @xmath100 .
let @xmath317 and @xmath318 be two regions of the shi arrangement .
we say that @xmath317 is less than @xmath318 if more hyperplanes of the shi arrangement
separate @xmath318 from the fundamental alcove , than for @xmath317 .
write @xmath319 for the cover relations of this partial order .
let @xmath320 denote the set of pairs @xmath321 , where @xmath317 is a region of the shi arrangement , and @xmath168 is such that @xmath317 contains an alcove of type @xmath50 .
we make @xmath320 into a directed graph by defining edges @xmath322 whenever @xmath323 , and an alcove @xmath59 of type @xmath50 in @xmath317 is adjacent ( shares a facet ) with an alcove @xmath324 of type @xmath325 in @xmath318 , satisfying @xmath125 .
[ l : shi]if @xmath322 then every alcove @xmath59 of type @xmath50 in @xmath317 shares a facet with an alcove @xmath324 of type @xmath325 in @xmath318 , and we have @xmath125 .
suppose @xmath59 and @xmath326 are both of type @xmath50 inside @xmath317 .
set @xmath327 .
let @xmath328 be a hyperplane cutting out a facet of ( the closure of ) @xmath59 , and suppose @xmath324 is on the other side of @xmath328 , adjacent to @xmath59 and satisfying @xmath125
. similarly define @xmath329 adjacent to @xmath326 , on the other side of @xmath328 . clearly , @xmath330 .
since @xmath59 and @xmath326 belong to the same region of the shi arrangement , the line segment joining the center of @xmath59 to the center of @xmath326 does not intersect the shi arrangement .
but one can go from @xmath324 to @xmath329 by crossing @xmath328 , traveling from @xmath59 to @xmath326 and crossing @xmath328 again .
thus the only hyperplane of the shi arrangement that could separate @xmath324 from @xmath329 is @xmath328 .
suppose this is indeed the case
. then @xmath130 can not be parallel to @xmath328 .
let @xmath328 be orthogonal to the root @xmath331 , so that we must have @xmath332 .
but from it is easy to see that one of the hyperplanes @xmath333 was crossed going from @xmath59 to @xmath326 .
it follows that the region @xmath317 is not bounded in the @xmath331 direction .
the hyperplane @xmath328 must thus be @xmath334 or @xmath335 . in either case
, it separates @xmath59 from @xmath100 , contradicting the assumption @xmath125 . thus @xmath329 and @xmath324 belong to the same region @xmath318 of the shi arrangement .
but if @xmath125 then @xmath328 must separate @xmath324 from @xmath100 , and thus @xmath336 .
denote by @xmath337 of the unique region of the shi arrangement that is a translation of the weyl chamber @xmath46 .
let @xmath338 be the graph obtained from @xmath320 by removing @xmath339 .
let @xmath64 be the transition matrix of @xmath320 and let @xmath340 be its restriction to @xmath338 .
let @xmath341 be the vector with components labeled by vertices of @xmath338 , given by @xmath342 .
note that for each @xmath343 , there is at most one @xmath344 for which the probability @xmath345 is non - zero .
let @xmath346 denote the unit vector corresponding to a vertex of @xmath320 , and @xmath347 denote the natural inner product on the vertex space spanned by vertices of @xmath320 .
[ t : shi ] for each @xmath168 , @xmath348 lemma [ l : shi ] guarantees that the markov chain @xmath7 projects to a markov chain on @xmath320 via @xmath349 where the alcove @xmath103 lies in the region @xmath317 .
thus the probability @xmath49 we desire is equal to the probability that a random walk in @xmath320 starting from @xmath350 , with transition matrix @xmath64 , eventually ends up at one of the vertices @xmath351 .
this immediately gives the stated formula , assuming that @xmath352 is invertible , and is equal to @xmath353 .
let @xmath317 be a region of the shi arrangement which lies between two parallel hyperplanes @xmath334 and @xmath335 .
then for each @xmath354 , there is some @xmath355 outside of @xmath317 .
it follows that the random walk @xmath39 has probability 0 of staying in a region of the shi arrangement other than one of the @xmath337 s .
thus @xmath356 must be invertible , @xmath340 must be strictly sub - stochastic , and @xmath357 .
( 1,1 ) rectangle ( 8.5,8.5 ) ; ( @xmath8 )
( @xmath9 ) ; ( @xmath10 ) ( @xmath11 ) ; ( @xmath12 ) ( @xmath13 ) ; ( @xmath358 ) ( @xmath359 ) ; ( @xmath360 ) ( @xmath361 ) ; ( @xmath362 ) ( @xmath363 ) ; ( @xmath14 ) ( @xmath15 ) ( @xmath16 ) ; in 0 , 1 , ... , 10 ( @xmath34 ) ( @xmath35 ) ; in -5,-4,-3,-2,-1,0,1,2,3,4,5,6,7,8,9,10 ( @xmath36 ) ( @xmath37 ) ; in 1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16 ( @xmath36 ) ( @xmath38 ) ; at ( 0:0 ) @xmath19 ; at ( 90:2/3 ) @xmath364 ; at ( 210:2/3 ) @xmath364 ; at ( 330:2/3 ) @xmath364 ; in 1,2,3,4,5,6 at ( :3 ) @xmath365 ; in 1,2,3 at ( + 30:3 ) @xmath366 ; at ( + 90:3 ) @xmath367 ;
in this section we suppose @xmath166 is the symmetric group . we assume basic familiarity with young diagrams .
recall that a skew young diagram @xmath368 is a ribbon if it is edge - connected and does not contain any @xmath369 square .
a young diagram @xmath130 is called an _ @xmath4-core _ if no ribbons of size @xmath4 can be removed from it ( and still leaving a young diagram ) .
the set of @xmath4-cores can be built from the empty partition by the following procedure .
take an @xmath4-core @xmath130 , and suppose @xmath370 is an addable - corner of @xmath130 on diagonal @xmath371 .
then the young diagram obtained from @xmath130 by adding all addable - corners on diagonals @xmath372 satisfying @xmath373 , is also an @xmath4-core , and recursively one obtains every @xmath4-core in this way .
figure [ fig:3 ] shows the start of the @xmath374-core graph , where the edges denote the above box adding operation .
the @xmath374-core graph is the one - skeleton of a hexagonal planar tiling .
the following result is well - known , see @xcite .
= [ minimum size=0pt , inner sep=0pt ] ( 0,0 ) node ( 0 ) [ label = left:@xmath375 node [ left=1pt , blue ] @xmath17 + + ( 90:2 cm ) node ( 1 ) [ label = below right:@xmath376{{}}}}$ ] ] node [ below=1pt , blue ] @xmath19 + + ( 150:2 cm ) node ( 2 ) [ label = left:@xmath376{{}&{}}}}$ ] ] node [ left=1pt , blue ] @xmath21 + + ( 90:2 cm ) node ( 31 ) [ label = below left:@xmath376{{}&{}&{}\\{}}}}$ ] ] node [ below=1pt , blue ] @xmath17++(150:2 cm ) node ( 42 ) [ label = left:@xmath376{{}&{}&{}&{}\\{}&{}}}}$ ] ] node [ left=1pt , blue ] @xmath19++(90:2 cm ) node ( 531 ) [ label = left:@xmath376{{}&{}&{}&{}&{}\\{}&{}&{}\\{}}}}$ ] ] ; \(1 ) node [ below=1pt , blue ] @xmath21++(30:2 cm ) node ( 11 ) [ label = right:@xmath376{{}\\{}}}}$ ] ] node [ right=1pt , blue ] @xmath19++(90:2 cm ) node ( 211 ) [ label = below right:@xmath376{{}&{}\\{}\\{}}}}$ ] ] node [ below=1pt , blue ] @xmath17++(30:2 cm ) node ( 2211 ) [ label = right:@xmath376{{}&{}\\{}&{}\\{}\\{}}}}$ ] ] node [ right=1pt , blue ] @xmath21++(90:2 cm ) node ( 32211 ) [ label = right:@xmath376{{}&{}&{}\\{}&{}\\{}&{}\\{}\\{}}}}$ ] ] ; [ p : bij ] there is a natural bijection between @xmath4-cores and the affine grassmannian elements of @xmath377 .
the edges of the @xmath4-core graph correspond to left - multiplication by simple generators .
[ l : slope ] let @xmath379 be an antidominant element of the coroot lattice
. then the @xmath4-core of the translation element @xmath380 has slope @xmath381 between diagonals @xmath382 and @xmath383 , for @xmath384 . by a random @xmath4-core we will mean an @xmath4-core generated by applying the bijection in proposition [ p : bij ] to the markov chain @xmath238 described in section [ ss : random ] . if @xmath130 is a @xmath4-core
, then we let @xmath387 denote the curve drawing out the lower - right boundary of @xmath130 , scaled by the degree @xmath388 in both directions . here
the degree is the length of the corresponding affine grassmannian element from proposition [ p : bij ] , or equivalently , the distance from the empty partition in the @xmath4-core graph . by convention ,
@xmath387 includes a vertical ray going to @xmath389 along the @xmath390-axis , and a horizontal ray going to @xmath391 along the @xmath111-axis . given two curves @xmath392 of this form , we write @xmath393 to denote the supremum of the distance between @xmath175 and @xmath394 , measured along the diagonals @xmath395 . with this notation , corollary [ c : main ] combined with lemma [ l : slope ] translates to theorem [ t : cores ] .
let us use conjecture [ conj : a ] to predict the piecewise - linear curve @xmath396 of theorem [ t : cores ] .
let @xmath302 be an anti - dominant element of @xmath113 satisfying @xmath397 ( that is , @xmath302 is in the same direction as @xmath263 ) . to calculate the correct scaling we use lemma [ lem : grass ] which says that @xmath398 , where @xmath399 .
now consider the piecewise - linear curve @xmath400 which successively connects the points @xmath401 @xmath402 using lemma [ l : slope ] , one calculates that the core @xmath130 corresponding to @xmath403 has diagram @xmath387 extremely close to @xmath396 : namely , it passes through the specified points but may not be linear in between those points .
it follows that to compare with the limit shape of a random partition in the sense of vershik and kerov , we should scale our diagrams differently .
instead of scaling by the degree of @xmath130 , we should scale by the number of boxes in @xmath130 . in the case of our limit shape @xmath396 ,
we would scale by the area between @xmath396 and the axes , which is @xmath405 scaling by @xmath406 in both directions to get a diagram with area 1 , the @xmath111-intercept would become @xmath407 , which should be compared with the the result of @xmath21 in the case of vershik and kerov .
this value @xmath21 is also the limit @xmath408 of the expected length @xmath409 of the longest increasing subsequence of a uniform random permutation in @xmath410 , divided by @xmath411 @xcite .
this work was motivated by the connections to a family @xmath412 of symmetric functions labeled by @xmath413 , known as _
affine stanley symmetric functions _
@xcite ( and also a closely related family @xmath414 called the affine stable grothendieck polynomials @xcite ) .
the coefficient @xmath415 \tf_x(x)$ ] of the square - free monomial in @xmath416 is equal to the number of reduced words of @xmath111 . whereas stanley s seminal work @xcite studies _ exact _ formulae for the number of reduced words , our approach looks for _ asymptotic _ formulae . the symmetric functions @xmath416 plays the same role for affine permutations , namely , a generating function for `` semi - standard '' objects , as the schur functions @xmath417 play for grassmannian permutations .
schur functions play a crucial role in the study of random partitions ; see for example @xcite .
the measure we obtain on the set @xmath418 of affine permutations of length @xmath68 from our random walk is not the same measure as the one obtained by letting @xmath419 be proportional to the number of reduced words of @xmath111 .
nevertheless , corollary [ c : main ] and theorem [ t : cores ] still apply ( see remark [ rem : reduced_words ] ) . in @xcite
, we proved an enumerative identity @xmath420 where the sum is over @xmath4-cores of degree @xmath421 .
_ weak tableaux _ count paths in the @xmath4-core graph .
_ strong tableaux _ are defined in terms of the strong ( bruhat ) order .
the terms on the right hand side of would give the natural analogue of the plancherel measure for partitions .
in @xcite , a symmetric function generalization of is also given , and involves affine stanley symmetric functions and _
@xmath151-schur functions_. the identity is generalized to the kac - moody case in @xcite .
recall from the proof of lemma [ l : reverse ] in section [ ss : probw ] , that the probabilities @xmath422 were given by the coefficients @xmath292 \xi^n$ ] for an element @xmath423 . in the case
@xmath166 , by ( * ? ? ?
* corollary 7.5 ) , the element @xmath78 can be interpreted ( up to a factor ) as the divisor schubert class in the @xmath74-homology @xmath424 of the affine grassmannian of @xmath76 .
the affine grassmannian @xmath75 is an infinite - dimensional space of central importance in representation theory . in the case of a complex simple algebraic group with weyl group @xmath0 , the natural element to consider from the point of view of the geometry of @xmath425
is @xmath426 where the definition of the weights @xmath273 can be found in @xcite ; see ( * ? ? ?
* proposition 2.17 ) for an explanation of these weights ( the argument in @xcite is for the homology case , but easily extends to @xmath74-homology ) .
probabilistically , this amounts to considering random walks where the allowable transitions are not taken uniformly at random , but left multiplication by @xmath270 is weighted by the @xmath273 .
note that theorem [ t : main ] and its proof still remain valid in this situation .
see also remark [ rem : othertypes ] .
xxx p. bidigare , p. hanlon , and d. rockmore , a combinatorial description of the spectrum for the tsetlin library and its generalization to hyperplane arrangements .
duke math .
j. 99 ( 1999 ) , no .
1 , 135174 .
a. bjrner and f. brenti , combinatorics of coxeter groups .
graduate texts in mathematics , 231 .
springer , new york , 2005 .
xiv+363 pp .
j. humphreys , reflection groups and coxeter groups .
cambridge studies in advanced mathematics , 29 .
cambridge university press , cambridge , 1990 .
xii+204 pp .
k. ito , parameterizations of infinite biconvex sets in affine root systems . hiroshima math . j. 35 ( 2005 ) , no .
3 , 425451 . v.g .
kac , infinite - dimensional lie algebras .
third edition .
cambridge university press , cambridge , 1990 .
xxii+400 pp .
affine stanley symmetric functions .
128 ( 2006 ) , no . 6 ,
schubert polynomials for the affine grassmannian .
( 2008 ) , no .
1 , 259281 . t. lam , l. lapointe , j. morse , and m. shimozono , affine insertion and pieri rules for the affine grassmannian .
( 2010 ) , no .
977 , xii+82 pp .
t. lam and p. pylyavskyy , total positivity for loop groups ii : chevalley generators , preprint , 2009 ; arxiv:0906.0610 .
t. lam , a. schilling , and m. shimozono , @xmath74-theory schubert calculus of the affine grassmannian .
146 ( 2010 ) , no .
4 , 811852 .
t. lam and m. shimozono , dual graded graphs for kac - moody algebras . algebra number theory 1 ( 2007 ) , no .
4 , 451488 .
t. lam and l. williams , a schubert positive markov chain on the symmetric group ? , preprint , 2011 .
logan and l.a .
shepp , a variational problem for random young tableaux .
advances in math . 26 ( 1977 ) , no .
2 , 206222 . a. okounkov , infinite wedge and random partitions .
selecta math .
( n.s . ) 7 ( 2001 ) , no .
1 , 5781 . | let @xmath0 be a finite weyl group and @xmath1 be the corresponding affine weyl group . we show that a large element in @xmath1 , randomly generated by ( reduced ) multiplication by simple generators , almost surely has one of @xmath2-specific shapes .
equivalently , a reduced random walk in the regions of the affine coxeter arrangement asymptotically approaches one of @xmath2-many directions .
the coordinates of this direction , together with the probabilities of each direction can be calculated via a markov chain on @xmath0 .
our results , applied to type @xmath3 , show that a large random @xmath4-core has a limiting shape which is a piecewise - linear graph , in a similar sense to vershik and kerov s work on the shape of a random partition . |
since a nucleon is not a point object , but has structure , it must undergo changes when placed inside a nucleus . among other properties , the meson - nucleon coupling constants may change . if this happens , it should affect the nn force .
these effects have to be small .
otherwise , traditional nuclear physics , where one uses free - space two - body force would have failed badly .
but even small changes in the nn force may have noticeable effect on the properties of nuclear matter .
the purpose of this paper is to investigate possible changes of meson nucleon coupling constants due to the quark structure of the nucleon and their ultimate effects on the saturation properties , such as the density @xmath6 and the binding energy @xmath7 of nuclear matter .
the nn force in nuclear matter may become density dependent due to a wide variety of reasons .
any time one eliminates some degrees of freedom the resulting effective interaction becomes density dependent , the brueckner @xmath8-matrix being the most widely known example .
another recent example is the work of li @xcite on the effective interaction to be used in a mean field calculation which reproduces the results of a brueckner - hartree - fock calculation .
the density dependence studied here involves excitations of n@xmath9 ( @xmath10 resonances ) degrees of freedom .
ultimately we are interested in the change of the nn force in the medium .
in relativistic nuclear physics the most important forces are mediated by a scalar - isoscalar field with @xmath11 mev and a vector - isoscalar field with @xmath12 mev .
thus the force range is short and the neighboring nucleons which can alter the internal structure of the interacting nucleons must also be close . due to the exclusion principle and short - range correlations , the effective density of the polarizing nucleons , denoted as @xmath13 ,
is less than half the normal nuclear density .
it is quite possible that the @xmath14 excitations produced by the neighboring nucleons may be treated perturbatively .
however , an earlier study of this problem @xcite established that a fairly large number of resonances contribute . the mean field approach , which generates the eigen - combination of n and @xmath14s as the lowest state in the field due to the neighboring nucleons is a more expeditious way of calculating the effect .
our quantitative studies are based on the following strategy .
we describe the structure of the nucleon with a model called the chiral confining model ( ccm ) @xcite .
specifically , we use the toy @xcite version of this model .
the role of nuclear matter is simulated with baths of external and fields , the vacuum values being @xmath15 mev and @xmath16 , respectively .
the nucleon structure problem is solved in the presence of these bath fields .
then various properties of the nucleon , including meson nucleon coupling constants , are calculated for ranges of values of the two bath fields .
the fact that the and the fields are , in turn , produced by the nucleons through field - dependent coupling constants allows us to obtain density dependences of the and the fields by solving appropriate nonlinear self - consistency equations .
once this has been done , we can use the known field dependences of various properties of the nucleon to obtain their density dependences .
we indicate the density - dependent values of various properties of the nucleon with a star .
thus @xmath17 denotes the free space nn coupling constant and @xmath18 the same quantity in nuclear matter .
the latter is always dependent on @xmath19 , the nuclear density .
we may note that the properties of nuclear matter depend principally on @xmath20 , @xmath21 , @xmath22 , @xmath23 and @xmath24 , where the last is the @xmath25 pauli coupling coefficient .
a necessary input is the field - dependence of the exchanged meson masses .
we use the simple model that the meson masses are linearly dependent on the field , keeping the pion mass fixed .
the coefficient of linear dependence of and masses on the field is chosen to give @xmath26 at normal nuclear density .
the changes , as found in the ccm , clearly will have effect on the properties of nuclear matter . to study this
we adopt the relativistic many - body approach @xcite .
specifically , we carry out a relativistic dirac - brueckner calculation of the properties of nuclear matter using the one - boson - exchange model of ref . @xcite . for the quasi - potential version of this model
the full dirac structure of the nn amplitude in free space has been constructed @xcite .
the resulting so - called ia2 representation can be used to determine the saturation properties of nuclear matter @xcite . modifying the free space t - matrix to also include pauli - blocking and introducing our density - dependent meson - nucleon coupling constants , self - consistent relativistic dirac - brueckner calculations were performed in the manner of ref .
@xcite for a range of @xmath13 the effective density of the polarizing nucleons .
our main results are that i [ ] ( i ) the effects of the density - dependent meson nucleon coupling constants , arising out of the quark structure of the nucleons , on the saturation density and @xmath7 at saturation are small but not negligible , i [ ] ( ii ) and they do tend to improve the results . the next section contain brief introduction to the chiral confining model ( ccm ) and its toy version .
section 3 describes the calculation of nucleon properties as functions of bath @xmath27 and @xmath28 fields .
these results are used in the next section to extract density dependences of nucleon properties .
section 5 describes the relativistic treatment of nuclear matter using the dirac - brueckner approach .
the last section presents the main results , discussions of these results and concluding remarks .
an early attempt at extracting density dependence of nucleon properties using ccm is described in ref .
the present paper contains two significant improvements - inclusions of the instanton induced interaction and the pion cloud contributions to @xmath0 , @xmath29 and @xmath30 coupling constants .
the ccm has been described in detail in @xcite . here
we give a brief review .
the ccm is based on the notion of color dielectric function as introduced by nielsen and ptkos @xcite . by considering the average of all possible link operators , starting from @xmath31 and ending on @xmath32 with the paths completely contained in a four - dimensional hypercube of side @xmath33
, they introduced a color singlet , lorentz scalar quantity @xmath34 and a color octet , lorentz vector , coarse grained gluon field @xmath35 : k(x)&=&_0tr , + b_=12_a _ a b^a_&=&_0i . upon integrating out the qcd gluon fields in favor of these new collective variables
one obtains the nielsen - ptkos lagrangian in the form of a derivative expansion : _
np=(x)[ik(x)12-k(x)m_q - gb/(x)](x ) -g^a_g^a+ .
npl1 the gauge field is @xmath36 and not @xmath35 .
the coarse grained field tensor is g^a_=_- _
npl2 from the gluonic term one identifies @xmath37 as the color dielectric function . conjectured that k_vac=0 .
kvac1 this conjecture , crucial for our model , has been justified from the lattice gauge point of view by lee @xcite .
a quark has ever - present interaction with the quark condensate of the vacuum .
if @xmath38 , the interaction will appear to be infinitely strong compared to quark kinetic energy and a quark can not exist in that region .
it can only reside in the region where @xmath39 .
a color singlet quark system can polarize the vacuum and change the value of @xmath40 away from zero , thus dynamically generating the _ bag _ where the quarks can stay .
the nielsen - ptkos lagrangian recognizes the existence of gluon condensate through the vanishing of @xmath41 .
however , the quark condensate is not manifest . without it
one can not develop an effective lagrangian which contains the physics of the interaction of a quark with the quark condensate .
we deal with this problem by conjecturing that one can integrate out the coarse grained gluon fields in favor of meson fields as new collective variables @xcite .
it is also necessary to introduce , following nielsen and ptkos @xcite , a new color singlet , lorentz scalar field , @xmath42 , proportional to @xmath40 by the equation k(x)=(x ) .
chi being related to @xmath40 , a purely gluonic object , the @xmath42 field is a member of the glueball family .
hence , it is a chiral singlet , a fact evident from the first term of the nielsen - ptkos lagrangian given by eq .
( [ npl1 ] ) .
large @xmath43 analysis shows that it is a hybrid field @xcite . the result ( [ kvac1 ] ) imposes the requirement that _ vac=0 .
chivac retaining minimum powers of fields and their derivatives the basic lagrangian of the ccm has the form : _
ccm&=&k(x)(x)[i12-m_q](x ) + & + & ( x)(x ) + & + & l_meson + 12 _ ^- u ( ) . lccm the quantities @xmath44 , @xmath45 and @xmath46 are the quark - meson coupling constants .
the quantity @xmath47 is the current quark mass .
its value is set at @xmath48 mev .
it contributes @xmath49 mev to the @xmath50n sigma term @xcite . because of its negligible role in the present work we
will neither refer to this term nor count it as a parameter in our subsequent discussions .
however , it is included in the actual numerical work .
we use the chiral invariant lagrangian of lee and nieh @xcite for @xmath51 .
the lagrangian ensures that @xmath52 and that the mesons have their respective physical masses when calculated at the classical ( or tree ) level .
the fields @xmath50 and @xmath27 form a @xmath53 representation of @xmath54 , while the fields @xmath55 form @xmath56 and @xmath57 representations . to complete the definition of @xmath58
one must specify the @xmath42 potential .
since nothing substantial is known about it , we try two simple forms : & & u()=12 m_^2 , + quartic : & & u()=12 m_^2(1-/_0)^2 .
uchi the quartic potential has two minima , the one at @xmath59 describes the vacuum while the other at @xmath60 is an ` excited ' state , which for simplicity we keep degenerate with the vacuum .
the value of @xmath61 , the location of the second minimum is set at @xmath62 mev .
the hybrid mass , @xmath63 is set at @xmath64 mev . as we will discuss later , the results of the calculations depend largely on one particular combination of these parameters and rather weakly on individual ones . any mean field calculation containing isovector fields
requires using states which do not have good isospin symmetry . to accomplish this we use hedgehog spinors and fields , introduced first for @xmath50 and @xmath27 fields by chodos and thorn @xcite and extended to vector meson fields by broniowski @xcite .
we closely follow the mean field analysis of ren @xcite .
it should be emphasized that the appearence of @xmath40 in the denominator of the quark - meson interaction term is not _
ad hoc_. it is obtained by matching the @xmath40 dependence of the four - quark interactions which arise on the one hand from @xmath65 by integrating out the @xmath66 fields and on the other hand from @xmath58 by integrating out the meson fields @xcite .
the presence of the factor @xmath40 with @xmath67 changes the field canonically conjugate to @xmath68 from the usual @xmath69 to @xmath70 . as a result
the quark term in all noether s currents carries the factor @xmath40 .
the transformation @xmath71 makes the pair @xmath68 and @xmath69 canonical and removes the factor @xmath40 from all noether s currents .
two changes occur in @xmath58 .
the free quark term no longer has the factor @xmath40 , while the quark meson interaction acquires the factor @xmath72 in place of @xmath40 in the denominator .
mean field treatment of @xmath58 with a variety of reasonable sets of parameters reveals @xcite that even where the quark density is large the meson fields differ only slightly from their respective vacuum values .
this suggests strongly that we introduce a simplified version of ccm .
the simplified version , which we call the toy model , consists of fixing all meson fields at their vacuum values : _
vac =- f _ , + v_vac=_vac = v_vac = _1_vac=0 , leaving only the quark fields and the @xmath42 field as dynamical variables .
the toy model lagrangian , in terms of the canonical quark field , is given below .
_ toy=(x)[i12-m_q- ] ( x)+12 _ ^- u ( ) .
ltoy not counting @xmath47 , the toy model has only two parameters , @xmath73 and @xmath63 in the pure mass version and one additional parameter , @xmath61 , in the quartic version .
formation of the bag is even more transparent in the toy model as the constituent quark mass term , @xmath74 , becomes infinite when @xmath75 .
the unusual form of the quark - meson interaction with the factor of @xmath76 in the denominator is of some importance in the present investigation . in its absence
a constant external field will merely shift the energy of the quark without exciting it to higher states .
thus the nucleon will not be polarized .
a constant field will excite the quark .
orthonormality of different eigenstates of the dirac hamiltonian ensures that @xmath77 , but not the vanishing of @xmath78 .
the latter gives the effect of a constant external field .
the presence of the factor @xmath79 enables a constant external field to excite the quark and enhances the ability of a constant external field to do the same . because of its importance it is necessary to provide some evidence for the presence of @xmath79 in the denominator of the quark - meson interaction term .
fortunately , there is a verifiable consequence of this feature .
gauge invariance ensures that there is no such modification of the photon - quark interaction .
thus contrary to the conjectures of universal coupling @xcite or current - field identity @xcite the isovector couplings of the photon and the @xmath25 meson differ by this @xmath80 factor . because of this difference , ccm predicts that the @xmath25-nucleon pauli coupling constant , @xmath30 should differ from the isovector anomalous magnetic moment , @xmath81 .
the predicted @xcite value is @xmath82 , instead of @xmath83 predicted by the two conjectures .
based on dispersion theoretic analysis of @xmath84 scattering data hhler and pietarinen @xcite had estimated the ratio to be @xmath85 .
our value is within the possible uncertainties of this result @xcite .
meson couple to the quark in different manner .
brown and rho had exploited this feature to explain the value of @xmath86 . dialing the chiral angle they fitted the value @xmath87 .
our result is approximately independent of parameters because of the scaling property mentioned later . ]
the role of the pion field on baryon properties is never negligible . in the toy model
one takes account of it perturbatively @xcite wherever appropriate . in the context of the present study such corrections are important in the calculations of the coupling constants of mesons of even @xmath8-parity , @xmath17 , @xmath88 and @xmath89 .
the toy model with perturbative pionic correction gives poor results for the masses of @xmath90 and @xmath91 and for the nucleon charge radii @xcite .
@xcite showed that there is approximate , but very good scaling behavior in the toy model .
if one chooses the parameters to get the nucleon mass right the values of most other quantities are nearly fixed .
thus the bad results for mass and size can not be improved simultaneously by varying parameters .
one needs a new interaction which is short - ranged and more attractive between singlet - singlet quark pairs than between triplet - triplet pairs .
the magnetic one - gluon exchange interaction @xcite fits the requirement exactly .
however , the physics of exchange of a color octet tower of gluons is already incorporated in meson exchanges .
we can not include it again .
we solve the problem by including the instantanton induced t hooft interaction @xcite .
in the two flavor case of @xmath92 and @xmath93 quarks , the t hooft interaction generates interactions between quark pairs in flavor antisymmetric state only and the interaction is attractive and zero - ranged .
so it fits our requirement without duplicating any mechanism already included .
using shuryak s
@xcite description of the vacuum as an instanton liquid the strength of the t hooft interaction can be estimated in qcd , but not in ccm @xcite .
so we have to treat it as a free parameter .
we fix it by fitting the @xmath94 mass splitting with the t hooft interaction together with one - pion exchange contribution @xcite .
the method of calculation with the toy model in the absence of external and has been described in detail in ref .
@xcite and will not be repeated here .
the changes due to the presence of external fields are straightforward @xcite .
going back to @xmath58 , described by eq .
( [ lccm ] ) , we interpret @xmath27 and @xmath95 as the bath fields .
the vacuum expectation values of all other fields , @xmath96 , @xmath97 ( @xmath98 ) , @xmath99 and @xmath100 are zero because nuclear matter has good isospin and good parity .
the quark field and the @xmath42 field continue to be the only dynamical variables in the extended toy model lagrangian .
the new energy functional of a nucleon in a -bath is given below .
e&= & n_cd^3r u^()u( ) + & + & d^3r , + & -&2c_sd^3rz(r)(g^2(r)+f^2(r ) ) .
nmtoy where the valence spinor , @xmath101 is u()= ( c g(r ) + if(r ) ) , [ spinor ] and @xmath102 is a 2-component pauli spinor .
the last line in eq .
( [ nmtoy ] ) represents the contribution of the instanton induced t hooft interaction with a regulating function , @xmath103 , which cuts off the interaction at a distance scale of @xmath104 fm
. the details may be found in refs .
@xcite .
note that the external fields are constant in space and time .
the vacuum value of the field is @xmath105 .
nuclear matter generates field with positive sign thus reducing the magnitude of the net .
it is customary to represent the effect as a modification of the pion decay constant :=- f_+_nmatter =- f_^*. fpistar the in eq .
( [ nmtoy ] ) is the time component of vector field .
the static nuclear matter distribution can generate only the time component .
however , a nucleon moving with velocity @xmath106 in nuclear matter does see a space component @xmath107 .
we ignore the role of this term .
fdnp we pick values of @xmath108 in the range @xmath109 mev to @xmath110 mev in steps of @xmath111 mev and values of in the range @xmath112 to @xmath62 mev in steps of @xmath113 mev . for each pair of @xmath108
and we find the spinor and the @xmath42 field which will make the energy functional stationary .
the condition of stationarity yields coupled nonlinear equations . ref .
@xcite describes the method of solving these equations .
once the solutions are obtained we can calculate the desired properties of the nucleon in the -field bath .
we introduce the dimensionless variables @xmath32 and @xmath114 to describe , in units of @xmath115 , the bath fields and : -_vac&=&-=xm_. xdef + & = & ym _ , ydef we fit the results of every physical quantity @xmath116 of interest with the quadratic form : q(x , y)=a[1+bx+cy+dx^2+fxy+gy^2]=af^q(x , y ) .
fqdef .
values of quantities a , b , c , d , f and g for the puremass case .
the quantities under the column a are the free space values . [ cols="<,^,^,^,^,^,^",options="header " , ] [ amodel ]
according to the ccm model the free space meson parameters will be modified in the presence of the nuclear medium . within the one - boson - exchange model the medium modifications found in the previous section can be simply implemented by replacing for each meson @xmath117 the coupling constant @xmath118 and the mass @xmath119 in the one - boson - exchange interaction @xmath120 by the corresponding @xmath121 and @xmath122 given by eqs .
( [ gratfn ] ) and ( [ mratfn ] ) . moreover , the pauli - blocking due to the medium has to be accounted for in eq .
( [ tm ] ) , leading to the bethe - brueckner - goldstone equations for the g - matrix .
this is done by replacing the green function in eq .
( [ tm ] ) s_2 s(p_1 ) s(p_2 ) q pbg where @xmath123 are the medium - modified nucleon propagators and @xmath116 the pauli - blocking operator , which projects out in the intermediate states nucleon momenta inside the fermi sphere . introducing the baryonic current @xmath124 , with @xmath25 being the density and
@xmath92 the unit vector @xmath125 in the nuclear - matter frame , due to lorentz covariance we may write the nucleon self - energy contribution in terms of the 3 invariants @xmath126 ( p ) = ^s -^0 .u-^v .p _ , self where @xmath127 .
the medium - modified nucleon propagator can be approximated near the dressed nucleon pole as s(p ) = [ e^ * _ 0 - p v - m^ * ] ^-1 prop with [ mst ] m^ * = ( m+^s)/(1+^v ) + e^*= ( p_0+^0)/(1+^v ) for the pauli - blocking operator @xmath116 an angular averaged approximation has been made .
the resulting relativisitic g - matrix equations [ gm ] g= v + v s(q_1 ) s(q_2 ) q g with @xmath128 the momenta of the nucleons in the intermediate states , are solved in the @xmath129 cm frame after partial wave decomposition .
this is done within a quasi - potential equation description using the helicity basis of positive and negative energy spinor states corresponding to mass @xmath130 .
relativistic nuclear matter calculations require the knowledge of the transformation properties of the g - matrix under lorentz transformations . after having determined the relativistic g - matrix in the 2-particle cm system at a fixed matter density , we use the ia2 representation @xcite to obtain it in the nuclear matter frame .
this representation gives the complete covariant form of the g - matrix in an unambiguous way .
to reconstruct it all matrix elements of the amplitude in the full dirac space are needed . from this ia2 representation , neglecting the vacuum fluctuation terms , the self - energy can be readily determined .
taking nucleon 2 to be one of the valence nucleons in the fermi sphere with fermi momentum @xmath131 , we have [ sigm ] ( p ) = -tr_2 _ |p_2|<p_f < p , p_2| g| p , p_2 > , where nucleon 2 is in a positive energy state with momentum @xmath132 and @xmath133 designates the summation over the spin index of this nucleon and @xmath134 . hence the self - energy contribution is essentially obtained by taking the diagonal matrix - elements of the g - matrix with one of the nucleons belonging to the filled fermi sphere while the other one carries a momentum @xmath135 . since the g - matrix depends implicitly on the @xmath126 s through eq .
( [ prop ] ) the calculations have to be carried out in a self - consistent way . in doing so
, the self - energy can be determined for a given matter density @xmath25 .
in particular we have assumed that @xmath136 does not vary significantly within the fermi sphere , so that its value can be taken at the fermi momentum @xmath131 .
some studies @xcite of the momentum dependence have been made , indicating that the variations in @xmath136 are of the order of 10 @xmath137 .
hence this is a reasonable approximation .
the partial wave decomposed integral equations for the g - matrix have been solved using gaussian quadratures .
typically 24 gaussian points are sufficient to get an accuracy of a few percent .
moreover , 6 gaussian points for each integration variable in eq .
( [ sigm ] ) have been used to determine the self - energies .
the self - consistent solutions can be found in an iterative way . adopting a starting value for the @xmath136 s in eq .
( [ mst ] ) , eq .
( [ gm ] ) is solved to yield through eq .
( [ sigm ] ) new values for the @xmath136 s .
varying subsequently the @xmath136 s we then can determine the solutions of eqs .
( [ gm]-[sigm ] ) such that the @xmath136 s are the same .
once the selfconsistent self - energy solutions have been found the binding energy of the ground state can readily be calculated from the energy - momentum tensor as a function of density . for more details of dirac - brueckner calculations we refer to refs @xcite and @xcite . for the various relativistic one boson interactions we find , in general , that the system exhibits saturation . as found in ref @xcite ,
an exception to this is when one assumes that only the meson - masses drop as @xmath138 @xcite .
it is interesting to note that in the case we also allow for medium - modification of the meson coupling constants as is predicted in our study of the chiral confining model , saturation does occur , that is we find that the binding energy of the ground state as a function of the matter density @xmath25 has a minimum . in the next section
we discuss our results for the saturation properties of nuclear matter as predicted for the medium modifications of the coupling constants as found in our ccm model .
as we have discussed earlier , the coupling constants are dependent not directly on the density @xmath19 of nuclear matter , but on the effective density , @xmath139 , seen by an interacting pair in its immediate neghborhood .
the pauli exclusion principle ensures that @xmath140 .
there is certainly further reduction of @xmath141 due to strong repulsion at short distances .
thus the range of @xmath141 of interest is @xmath142 .
we pick a value of @xmath141 and proceed to calculate @xmath2 and @xmath19 at saturation using the dirac - brueckner approach described in section [ dba ] . in the calculations reported here
we have used the one boson exchange interaction a specified in ref .
@xcite . during each loop of self - consistency iteration
we use coupling constants and meson masses appropriate for @xmath143 , defined earlier by eq .
( [ rhobar ] ) , @xmath19 being the nuclear density at the particular stage of the iteration . specifically , we use eqs .
( [ gratfn ] ) and ( [ mratfn ] ) with the coefficients of the rational functions listed in tables [ pratco ] and [ qratco ] .
finally , self - consistency is achieved and we have a pair of numbers , @xmath2 and density @xmath19 at saturation for the given value of @xmath141 . figures [ earp ] and [ earq ] show these results for the puremass and the quartic cases , respectively .
the value corresponding to @xmath144 represent the amorim and tjon @xcite result . from the figures we see that dependence of the nuclear saturation properties on @xmath141 is just of the right size to be of interest in the present study .
the @xmath2 vs. @xmath25 curve for the puremass case passes through the square representing experimental data for @xmath141 slightly greater than @xmath145 . for the quartic case the @xmath2 vs. @xmath25 curve comes close to the square , but does not enter it . the closest approach occurs for @xmath141 a little larger than @xmath146 .
the results for the unphysical range , @xmath147 are included in the figures to exhibit the curious looping effect . in order to study the origin of the looping effect
we have systematically switched off the density dependences of the various coupling constants and meson masses . in so doing , we find that the density dependence of @xmath18 is the main source of the looping behavior of the saturation values of @xmath2 and @xmath19 .
qualitatively such a behaviour can be understood with a simple model where the essential ingredient of the density dependence in the coupling constant @xmath18 is built in as a small modification of the saturation curve .
assuming that all meson masses and all coupling constants , except @xmath18 are independent of @xmath148 , the density dependence of @xmath2 can be modelled in the following manner : @xmath149 , \label{app1}\ ] ] where , for convenience , we have introduced the quantity @xmath150 the first two terms describe the density dependence of @xmath2 near the saturation point when none of the coupling constants and meson masses is density dependent .
the quantity @xmath151 is the nuclear incompressibility and @xmath152 is the minimum value of @xmath2 in this situation .
the last term contains the effect of the density dependence of @xmath18 , calculated perturbatively , using , in addition , the fact that @xmath153 _ i.e. , _ the average squared relative momentum of two fermi sea nucleons is much smaller than the square of the meson mass .
an inspection of the two graph in the left column of fig .
[ coupling ] shows that the quadratic form : @xmath154 can describe these graphs quite well .
the ratio @xmath155 peaks at @xmath156 and the peak value is @xmath157 .
the values of the various parameters used in the illustration are @xmath158 , @xmath159 , @xmath160 .
we consider four sets of @xmath161 and @xmath162 , namely , @xmath163 , and @xmath164 .
the first two describe the ratios @xmath165 for the quartic and the puremass cases , respectively , in fig .
[ coupling ] .
the last two are used to illustrate the origin of looping .
it is convenient to recast eq .
( [ app1 ] ) into the form : @xmath166/k_0=(y-1)^2- y\frac{3\rho_0}{4k_0}\frac{\gs(0)^2 } { m_\sigma^2 } \left[(g_{\sigma nn}(\rho)/g_{\sigma nn}(0))^2 - 1\right ] .
\label{ebya}\ ] ] because of the presence of the third term in eq .
( [ app1 ] ) the saturation density will change . for a given value of @xmath141
this can be readily determined by minimizing the expression for @xmath2 with respect to @xmath167 .
the resulting plots of the change in the binding energy at the new saturation density vs. @xmath168 are shown in fig .
[ loop ] for the four sets of where @xmath141 has been varied in the region @xmath169 starting from zero density , as @xmath170 increases with @xmath171 the quantity @xmath2 becomes more negative .
the saturation properties reflect this through a combination of increasing saturation density , @xmath25 , and decreasing @xmath2
. eventually , decrease of @xmath170 beyond its peak with increasing @xmath171 and the effect of incompressibilty take over and the saturation density stops increasing . if this occurs in the range @xmath172 , we see looping . if the peak of @xmath173 occurs at a higher density , _
i.e. , _ for a higher @xmath161 , the saturation density keeps on increasing and we see no looping for @xmath174 . in the present example
the critical value of @xmath161 for @xmath175 appears to be slightly higher than 0.7 .
thus the occurence of a peak of @xmath173 at not too high a value of the effective density is needed for looping to occur for @xmath176 .
the value of @xmath161 is lower for the quartic case than for the puremass case .
this explains why the loop turns for the former case at a density lower than that for the latter . due to the internal structure of the nucleon
, we should , in general , expect that the effective meson nucleon parameters may change in nuclear medium . using a chiral confining model
we have studied the density dependent changes of the meson - nucleon coupling parameters .
we have also used a simple ansatz for the density dependence of and masses . using the framework of a dirac - brueckner analysis we have found that their effect on the saturation properties of nuclear matter can be significant . due to the density dependence of the @xmath18 as predicted by the chiral confining model we have found , in particular , a looping behaviour in the nuclear matter binding energy at saturation density .
the looping has essentially also been verified in a simple model , where it is mainly caused by the presence of a peak in the density dependence of the medium modified @xmath1 coupling constant at a low density .
we should stress that the effect of the density dependence of the other quantities are not negligible .
however , a qualitative understanding of the looping effect can be obtained by paying attention to the density dependence of @xmath0 alone .
it appears that the density dependence of the coupling constants and the meson masses produce effects which are small but interesting . in particular the small effect tends to improve the results for nuclear matter . from the present study
we see that the relationship between binding energy and saturation density may not be as universal as found in nonrelativistic studies and that more model dependence is exhibited once medium modifications of the basic nuclear interactions are considered .
we hope that these preliminary results will encourage more detailed investigation of this particular variety of density dependence of the nn interaction .
this work was supported by doe grant doe - fg02 - 93er-40762 .
bulk of the writing and part of the work by one of us ( mkb ) were done while visiting the institute of kernphysik , forschungszentrum jlich .
he thanks his colleagues there for their hospitality .
his visit was made possible by a alexander von humboldt foundation award for senior u.s .
physicist and he thanks the foundation for it . 100 g.q .
li , r. machleidt and y.z .
zhuo , phys .
c * 48 * , 1062 ( 1993 ) .
m. k. banerjee , phys .
c * 45 * , 1359 ( 1992 ) .
m. kim and m. k. banerjee , phys .
c * 48 * , 2035 ( 1993 ) .
m. k. banerjee , in progress in particle and nuclear physics , vol .
a. faessler ( pergamon press , 1993 ) , pg .
77 . g. e. brown , prog . theor
. phys . suppl .
* 91 * , 85 ( 1987 ) ; t. l. ainsworth , g. e. brown , m. prakash and w. weise , phys . lett . *
200b * , 413 ( 1988 ) ; g.e .
brown , h. mther and m. prakash , nucl . phys . *
a506 * , 565 ( 1990 ) .
b. ter haar and r. malfliet , phys .
* 149 * ( 1987 ) 207 .
r. machleidt , adv .
. phys . * 19 * , 189 ( 1989 ) .
a. amorim and j. a. tjon , phys .
* 68 * , 772 ( 1992 ) j. fleischer and j.a .
tjon , nucl . phys . *
b84 * , 375 ( 1975 ) ; ibid .
, d * 21 * , 87 ( 1980 ) . j. a. tjon and s. j. wallace , phys . rev .
c * 35 * , 280 ( 1987 ) ; c * 36 * , 1085 ( 1987 ) .
h. b. nielsen and a. ptkos , nucl .
phys * b195 * , 137 ( 1982 ) . s. h. lee , t. d. cohen and m. k. banerjee , phys . rev .
d * 40 * , 3060 ( 1989 ) . in _ proceedings of workshop on nuclear chromodynamics , santa barbara
s. brodsky and e. moniz , world scientific , p478 ( 1985 ) .
m. k. banerjee , w. broniowski and t. d. cohen , _ chiral solitons _
keh - fei liu , world scientific , p255 ( 1987 ) .
m. k. banerjee , t. d. cohen and m. li , u. of maryland preprint pp # 89 - 183 .
b. w. lee and h. t. nieh , nucl
b166 * , 1507 ( 1968 ) .
a. chodos and c. b. thorn , phys .
d * 12 * , 2733 ( 1975 ) . w. broniowski and m. k. banerjee , phys .
158b * , 335 ( 1985 ) ; phys .
d * 34 * , 849 ( 1986 ) .
c .- y . ren and m. k. banerjee , phys .
c * 41 * , 2370 ( 1990 ) .
m. k. banerjee , _
quark , mesons and nuclei 1 _ , ed .
p. hwang and e. m.henley ( world scientific , singapore ) ( 1988 ) .
c .- y . ren , ph .
d. thesis , u. of maryland preprint doe / er/40322 - 129 .
j. j. sakurai , ann .
( n.y . ) * 11 * , 1 ( 1960 ) . m. gell - mann and f. zachariasen , phys . rev .
* 124 * , 953 ( 1961 ) ; n. m. kroll , t. d. lee and b. zumino , phys .
rev . * 157 * , 1376 ( 1967 ) .
g. hhler and e. pietarinen , nucl . phys . *
b95 * , 1 ( 1976 ) .
g. hhler , private communication ( 1991 ) .
m. kim , ph .
d. thesis , u of maryland preprint ( 1994 ) .
j. mcgovern , m. c. birse and d. spanos , j. phys . * g 16 * , 1561 ( 1990 ) .
a. de rujula , h. georgi and s. l. glashow , phys .
d * 12 * , 147 ( 1975 ) .
a. a. belavin , a. m. polyakov , a. s. schwartz and yu .
s. tyupkin , phys .
59b * , 85 ( 1975 ) . g.
t hooft , phys .
d * 14 * , 3432 ( 1976 ) .
e. v. shuryak , nucl . phys . *
b203 * , 93 ( 1982 ) .
e. hummel and j.a .
tjon , phys .
c * 49 * , 21 ( 1994 ) . | due to the internal structure of the nucleon , we should expect , in general , that the effective meson nucleon parameters may change in nuclear medium .
we study such changes by using a chiral confining model of the nucleon .
we use density - dependent masses for all mesons except the pion . within a dirac - brueckner analysis , based on the relativistic covariant structure of the nn amplitude ,
we show that the effect of such a density dependence in the nn interaction on the saturation properties of nuclear matter , while not large , is quite significant . due to the density dependence of the @xmath0 ,
as predicted by the chiral confining model , we find , in particular , a looping behavior of the binding energy at saturation as a function of the saturation density .
a simple model is described , which exhibits looping and which is shown to be mainly caused by the presence of a peak in the density dependence of the medium modified @xmath1 coupling constant at low density . the effect of density dependence of the coupling constants and the meson masses tends to improve the results for @xmath2 and density of nuclear matter at saturation . from the present study
we see that the relationship between binding energy and saturation density may not be as universal as found in nonrelativistic studies and that more model dependence is exhibited once medium modifications of the basic nuclear interactions are considered .
i v#1 @xmath3department of physics , university of maryland , college park , maryland 20742 , usa + @xmath4instute for kernphysik , forschungszentrum jlich , 52425 jlich , germany + @xmath5institute for theoretical physics , university of utrecht , 3508 ta utrecht , the netherlands |
what happens to the current carrying electronic states in a quantum hall sample when the magnetic field is turned off ? according to the prevailing view , at @xmath4 the single particle states of a disordered two - dimensional system are localized and no current can flow at zero temperatures .
@xcite a scenario , how this transition from the quantum hall liquid into the insulator may take place , was suggested by khmelnitskii @xcite and laughlin @xcite 20 years ago . they both proposed that , in the limit @xmath5 , the extended states of the n - th landau level float up in energy , @xmath6 , eventually crossing the fermi level one by one , until the lowest current carrying state gets depopulated . here
, the classical elastic collision time @xmath7 is a function of the disorder strength . over the years
, many experiments have been carried out to check this prediction , but produced only conflicting results , @xcite and a convergent picture seems to become apparent only recently ( see , e.g. , yasin _ _ et al.__@xcite and references therein ) .
in essence , the levitation of the current carrying state s energies has been observed in various samples for very low magnetic fields .
however , some find a saturation of the landau level shift as @xmath8 , for instance in high quality p - gaas samples , @xcite and even direct transitions to the hall insulator from higher ( @xmath9 ) quantum hall plateaus have been reported.@xcite therefore , the ultimate fate of @xmath10 in the limit @xmath8 remains an important and still open question .
the answer to this long lasting problem has become even more pressing through the observation of the apparent metallic behavior found at @xmath4 in experiments on dilute two - dimensional electron and hole systems.@xcite despite quite some effort made in the past to provide a microscopic theory,@xcite our understanding of the levitation scenario is still limited and mainly comes from numerical studies @xcite that are , however , afflicted by artifacts originating from the applied lattice model , and the difficulty of performing a proper limit @xmath8 .
the aim of our present work is to overcome these shortcomings and to study the levitation of the current carrying states within a continuum model that allows to take the essential limit @xmath8 . within our model , which in a first step is projected onto the lowest landau levels ( lll ) , we do not find an unbounded floating up in energy , but a saturation of the quantum hall extended states levitation that depends on the disorder strength .
the main advantages of our method are the straight forward extension to cases where additional landau levels are taken into account as well as the possible identification of those matrix elements responsible for the levitation because they are known analytically .
we consider non - interacting electrons moving in the @xmath11-@xmath12-plane of a two - dimensional continuum model with a perpendicular magnetic field and quenched disorder . we choose spatially correlated disorder potentials to be generated by a random weighted sum of gaussian shaped potential hills at random sites distributed uniformly over the area of the sample .
this disorder potential can be considered the continuum generalization of the correlated disorder potentials on the lattice used in our previous work @xcite @xmath13 the summation index @xmath14 runs over all scatterers at random positions @xmath15 , the number of which is @xmath16 .
the random variables @xmath15 and @xmath17 are both uncorrelated and uniformly distributed over the area of the system , and over the real interval [ @xmath181,1 ] , respectively .
@xmath19 is the strength of the disorder potential and @xmath20 the concentration of the scatterers .
the disorder potential is locally uniformly distributed , spatially gaussian correlated , @xmath21 $ ] with a correlation length @xmath22 , has a vanishing mean @xmath23 , and a uniform second moment @xmath24 .
thus , the square of the disorder strength @xmath25 is effectively the second moment of the disorder potential .
now we study the hamiltonian @xmath26 of free two - dimensional electrons moving in a homogeneous perpendicular magnetic field @xmath27 given by a vector potential @xmath28 with @xmath29 in the landau gauge . for a system infinite in @xmath12-direction and periodic boundary conditions in @xmath11-direction , @xmath30 possesses the well - known eigen base @xmath31 in terms of the hermite polynomials @xmath32 . here , @xmath33 is the finite system width in the periodic direction , @xmath34 the magnetic length .
the non - negative integer @xmath35 is the landau level index , and @xmath36 with integer @xmath37 represents the momentum in @xmath11-direction . for the following numerical evaluation we have to restrict the momenta @xmath38 to a finite number , @xmath39 $ ] , which effectively restricts the system to a finite area @xmath40 with open boundary conditions in the @xmath12-direction .
we rewrite the hamiltonian @xmath41 for the disordered system in the representation of the free hamiltonian s eigen base [ eq .
( [ spaghetti_base ] ) ] , @xmath42 now the projection to a finite set of landau levels is easily achieved by restricting the base and the @xmath35 and @xmath43 sums in the hamiltonian [ eq . ( [ hamiltonian_projected_1 ] ) ] to these landau levels .
because the disorder potential is a random weighted sum over gaussians , we can easily compute the matrix elements of the disorder potential in the approximation that the area where each single gaussian is non - zero is small compared to the area of the sample .
then we can replace the spatial integrations over the finite sample by infinite gaussian integrals .
further , we introduce dimensionless quantities by measuring all energies in units of @xmath44 and all lengths in units of the magnetic length @xmath45
. we will denote those quantities by attaching a hat to their symbols . this way
the hamiltonian only depends on the dimensionless correlation length parameter @xmath46 , the magnetic field only enters the energy and length scales .
it takes the form @xmath47 to study the limit of vanishing magnetic field for a given fixed disorder potential [ eq .
( [ potential_1 ] ) ] , it is equivalent to consider the limit @xmath48 in the dimensionless hamiltonian
. therefore , in some respect it corresponds to a white noise disorder model which does not discern between @xmath49 and @xmath50 .
however , the normalization of the disorder potential may be chosen differently from the @xmath51-property often used in studies of the high magnetic field limit . for the disordered system projected onto the lowest two landau levels we find the following matrix elements for a single scatterer . inside the lowest landau level
we get with the correlation parameter @xmath52 @xmath53 all higher intra and inter landau level interaction matrix elements can be expressed as a polynomial in @xmath54 , @xmath55 and @xmath56 times the intra - lowest landau level elements . for the inter landau level matrix elements we get @xmath57 and inside the second landau level @xmath58 here , @xmath59 and @xmath60 are the spatial coordinates of the particular scatterer .
we write the integer @xmath37 instead of the momenta @xmath61 and restrict them to the range @xmath62\subset z_1 $ ] , which renders the sample area finite . for a square sample
we have @xmath63 and @xmath64 , thus @xmath65\subset r_1 $ ] .
the concentration of the scatterers has been chosen in the range @xmath66 , high enough to reach the high concentration limit in the density of states .
for the hamiltonian in the restricted @xmath67-base , the energy eigenvalues were now computed numerically by exact diagonalization for @xmath68 and various correlation length @xmath69 and disorder strength @xmath70 .
the positions of the extended states have been extracted by means of the level statistics method.@xcite
a quantum hall model projected onto the lowest landau level only ( one band model ) still exhibits localization and scaling near the critical states at the band center@xcite .
this behavior was observed also in the presence of spatially correlated disorder potentials.@xcite of course , the one band model neither shows levitation of the critical states nor any movement of the density of states ( dos ) peak position when the magnetic field or the disorder strength is varied .
therefore , we consider in the following the case of a system projected onto the lowest two landau levels . versus the strength @xmath70 of the disorder potential .
the potential correlation lengths @xmath69 are 0.02 ( @xmath71 ) , 0.05 ( @xmath72 ) , 0.1 ( @xmath73 ) , 0.2 ( @xmath74 ) , 0.3 ( @xmath75 ) , 0.5 ( @xmath76 ) , 0.7 ( @xmath77 ) , 1.0 ( @xmath78 ) , and 1.5 ( @xmath79 ) , respectively . ]
( @xmath74 ) from the relation @xmath80 , which collapses the linear broadening of the lowest disordered landau band onto a single curve ( see fig .
[ a - collapse ] ) , is shown as a function of the reduced correlation length @xmath69 .
the result from a harper model ( @xmath77 ) ( @xmath81 ) deviate only for @xmath82 . to account for the lattice effects in the harper data , actually @xmath83 ( left y - axis ) and @xmath84))^{1/2}\alpha_\mathrm{h}^{}(\hat\eta)$ ] ( right y - axis )
are shown ( for @xmath85 $ ] see ref . ) , which relate the width @xmath86 of the lowest landau band to the width of the random potential distribution for the continuum and the harper model , respectively . ] versus disorder strength @xmath70 .
all data for various correlation lengths @xmath69 fall onto a straight line .
, width=283 ] with increasing strength @xmath70 of the disorder potential the landau levels broaden into bands .
their width increases , with good accuracy , linearly with the disorder strength @xmath70 for all correlation lengths @xmath46 considered .
the ratio of the broadening of the first and the second band increases with increasing correlation length , starting at nearly equal broadening for small @xmath46 . for a fixed second moment of the disorder potential
the width of the bands increases with @xmath46 . of the position of the landau band peaks ( @xmath71 , @xmath72 ) and of the extended states ( @xmath73 , @xmath74 ) as a function of disorder strength @xmath19 for the projected continuum model incorporating the lowest two unperturbed landau levels .
the lines are quadratic fits @xmath87.,width=283 ] in fig .
[ gamma0 ] the full broadening @xmath86 at half maximum is shown for the lowest landau band .
because of its uniform behavior , we can use the width of the lowest landau band as a measure for the effective disorder strength to compare various correlation lengths @xmath46.@xcite therefore we define a function @xmath88 which allows to collapse all @xmath89 for different @xmath46 onto a single curve @xmath90 . the function @xmath88 is plotted in fig . [ plot - alpha ] together with the numerical data obtained for the disordered harper model for comparison
. the continuum and the lattice model differ only for @xmath82 .
note that @xmath88 for large @xmath46 approaches @xmath91 as has been previously reported for the harper model.@xcite our data show a strictly linear dependence @xmath92 for small @xmath46 . the resulting data collapse is shown in fig .
[ a - collapse ] .
( @xmath74 ) which describes the energy shift of the dos peak of the lowest landau band , @xmath93 , and the coefficient @xmath94 ( @xmath77 ) relating @xmath95 and @xmath96 .
the corresponding data collapse onto a single curve is shown in fig .
[ plot_beta_coll ] for various correlation lengths @xmath69 .
, width=283 ] .
the data for various correlation lengths @xmath69 fall onto a straight line , @xmath97 , width=283 ] the typical effect of increasing disorder on the the peak positions of the two landau bands as well as on the positions of the extended states is shown in fig .
[ plot_fork ] .
the dos peak of the lowest landau band moves down in energy , away from the second band , which moves to higher energies with increasing disorder strength .
the extended states behave differently . while those of the second band are essentially following the dos peak , the lowest band s extended states clearly float up in energy , which can be traced approximately across 20% of the landau gap .
this behavior is qualitatively independent of the correlation length @xmath46 . as we can not expect the results for the second landau band to be physically meaningful within a model projected only onto the two llls , we concentrate in the following on the properties of the lowest landau band . throughout the investigated range @xmath98 of correlation lengths the shift of the peak position of the density of states to lower
energy is with reasonable accuracy @xmath99 and of considerable magnitude , comparable with the magnitude of the extended states shift absolute in energy .
we therefore define a function @xmath100 which collapses all @xmath101 for different @xmath69 onto a single curve @xmath102 .
the function @xmath100 is shown in fig .
[ plot_beta ] , with the collapse of the data shown in fig .
[ plot_beta_coll ] . for a fixed second moment of the disorder potential
the dos peak shift has a maximum around @xmath103 and decays quadratically for @xmath104 .
if the width of the lowest landau band @xmath86 is used as an effective disorder strength , the coefficient @xmath94 of the shift of the density of states does not diverge for small @xmath69 but rather behaves linearly , approaching a finite value for @xmath105 .
, collapsed onto a single curve by the function @xmath106 , is shown for several correlation length @xmath69 .
, width=292 ] ( @xmath77 multiplied by a factor 6 ) , which collapses the data onto a straight line , and @xmath107 ( ) are shown as a function of the correlation length @xmath69 .
the data points ( @xmath74 ) are the corresponding results from the projected harper model discussed below.,width=283 ] the absolute levitation of the lowest landau band s extended states with increasing disorder strength measured via @xmath86 was calculated for several @xmath69 .
the shift of the extended states positions in energy is @xmath108 and thus proportional to the disorder potentials second moment @xmath109 for all @xmath46 . introducing a scaling function @xmath106
, we can again collapse the data for all @xmath46 onto a single curve @xmath110 as shown in fig .
[ plot_gamma_coll ] . apart from the numerically least resolved points at large disorder strength ,
for each @xmath69 the data fit very well onto a single quadratic curve .
we have to remark that because of the stronger broadening of the landau bands for larger @xmath69 , larger shifts on this common curve can be resolved for smaller @xmath69 . the functions @xmath111 and @xmath112 relating the levitation @xmath113 to @xmath109 and @xmath96 , respectively ,
are shown in fig .
[ plot_gamma ] .
@xmath111 decays quadratically for @xmath48 , shows a maximum around @xmath114 , and decreases monotonically for larger @xmath69 .
therefore the levitation for a given broadening @xmath86 of the lowest landau band , @xmath112 , is finite at @xmath105 and decreases with increasing correlation length @xmath69 .
before we start the discussion about the important limit @xmath8 , we would like to compare our results obtained for the continuum model with those from a projected disordered harper model.@xcite the latter shows floating across almost half the landau gap . in fig .
[ plot_harper ] the floating of the energetical position of the lowest extended states are shown in a similar way as in fig .
[ plot_fork ] for two magnetic flux densities @xmath115 , and @xmath116 , respectively . for ease of comparison ,
the energy scale of the harper model has been shifted , @xmath117 , where @xmath118 is the energy of the lowest extended states in the unperturbed harper model , and @xmath119 denotes the flux quantum and @xmath120 the lattice constant . as in the projected continuum model ,
the extended states float up in energy @xmath121 .
the perfect agreement of both models can be seen from fig .
[ plot_gamma ] where the two data points from the harper model ( @xmath74 ) are shown to fall onto the respective curve for the continuum model .
the addition of further disorder broadened sub - bands reduce the floating of the extended states @xcite and produce deviations from the @xmath122 behavior which is most evident for the full disordered harper system .
we expect similar effects to occur for the continuum model when more than the two lowest landau bands are taken into account .
( @xmath77 ) , and @xmath123 ( @xmath75 ) which correspond to @xmath124 and @xmath125 , respectively .
the lines are quadratic fits @xmath126 . ]
the projection onto the lowest two landau levels is the most simple system which shows levitation of at least the lowest landau band s extended states .
previous work @xcite has suggested that the main contribution to the levitation of the lowest level results from the coupling to the second band and should have a quadratic dependence on disorder strength .
the advantages of the continuum model discussed in the present paper are _
( i ) _ that we have no interfering intrinsic lattice effects as in the projected harper models , and _
( ii ) _ that we can absorb the magnetic field dependence into the units of energy and length , such that the only parameters of the model are the disorder potential strength and its correlation length measured in units of the magnetic length .
this reduction to a dimensionless @xmath46-parameterized model is exact for an infinite system . by calculating the properties of the dimensionless model for decreasing @xmath46 we can learn about the behavior of the original system for fixed disorder potential and decreasing magnetic field .
let us now consider the limit of vanishing magnetic field . given a fixed disorder potential , as it occurs in most experiments on samples without an additional backgate , this means the limit @xmath48 .
for the width @xmath127 of the disorder broadened lowest landau band , we find from the function @xmath88 at small @xmath46 ( see figure [ plot - alpha ] ) a linear relation @xmath128 . in natural units
this reads @xmath129 , which means that the disorder induced broadening of the landau bands decreases for a given physical disorder potential with vanishing magnetic field @xmath27 , but slower than the landau gap .
hence , the overlap of adjacent landau bands increases for @xmath8 .
this is qualitatively in agreement with previous work.@xcite the levitation of the extended states @xmath130 for small @xmath27 is ruled by the behavior of the function @xmath106 near @xmath131 . since @xmath132 for small @xmath46 , the coefficient @xmath133 for the levitation of the extended states @xmath134 starts as a constant ( see fig .
[ plot_gamma ] ) .
hence , the shift of the extended states energy is confined by the broadening of the landau levels dos , @xmath135 . in experiments , where the second moment of the disorder potential is fixed in natural units ,
the relative broadening @xmath86 diverges for small magnetic field @xmath136 which leads to an infinite shift of the relative energy @xmath137 of the lowest landau band s extended states . as a consequence the critical filling factor , i.e. , the number of electrons divided by the number of flux quanta penetrating the area of the system where the fermi energy coincides with the energy of the extended states , diverges as well for @xmath8 .
however , if we consider the energy shift @xmath138 of these extended states in natural units , the @xmath27-dependence cancels for small @xmath27 and we find with @xmath139 a _ finite _ levitation @xmath140 that only depends on the properties of the disorder potential .
this result clearly conflicts with the levitation scenario proposed by khmelnitskii @xcite and laughlin @xcite which predicts a divergent energy of the extended states for vanishing magnetic field . in order to obtain such a divergency from our model , the function
@xmath107 would have to diverge like @xmath141 , hence @xmath142 , for small @xmath46 . here
, we assumed the scattering time @xmath7 in the khmelnitskii - laughlin expression for the energy level float up to be independent of @xmath27 .
there are , however , attempts to describe the classical low field magnetoresistance in terms of a @xmath27-dependent @xmath7.@xcite interpreting these results we have to consider several issues . first of all , in the present work we included only the lowest two landau bands . though this is certainly reasonable for small broadening of the landau bands because the strongest influence on the first landau band s extended states comes from the inter - band matrix elements that couple the first to the second landau level .
yet , it is not clear that the influence of the higher levels remains small for stronger overlapping bands . from projected disordered harper models including the lowest two and the lowest three harper bands,@xcite we have some evidence that the addition of the third band reduces the levitation of the first band s extended states .
the dependence on the disorder strength remains quadratic in this case . on the other hand for strong magnetic fields ,
the levitation of the lowest band s extended states in full disordered harper models has shown to be even slower and not quadratic in the disorder strength anymore .
however , for those models the levitation of the extended states could be traced at least across the landau gap , which is far beyond the resolvable range in the continuum model discussed in the present paper .
nevertheless , it remains to be seen , whether the finite levitation result obtained here will persist when taking into account more than the lowest two landau bands .
second , the possible influence of the finite system size .
the elimination of the explicite @xmath27 dependence of the system by the reduction to the @xmath46-parameterized dimensionless model is only exact for an infinite system . for the numerical simulation
we have to limit the hilbert space which is done by restricting the number of landau levels @xmath35 and , particularly , the number of momenta @xmath38 in the representation of the projected hamiltonian [ eq . ( [ hamiltonian_projected_1 ] ) ] .
the limitation of the @xmath38 effectively renders the area of the sample , @xmath40 , finite and the dimensionless model an approximation .
however , the finite size of the sample is expected to affect only large @xmath46 , or large @xmath2 for a given @xmath27 , not the opposite limit @xmath104 which we are discussing here .
note that the effective ( linear ) size of the system scales with @xmath45 .
the other approximation we made with respect to the system size is the restriction of the spatial integration domain over the gaussian scattering potentials in the calculation of the matrix elements .
we approximated the finite integration domain in the numeric simulation by indefinite gaussian integrals .
again this should not affect the limit of short correlation lengths .
in a finite sample the simulation results will depend on the concentration @xmath20 of the scatterers in the random potential [ eq .
( [ potential_1 ] ) ] albeit the local statistical properties of the potential do not . in our calculations
we have chosen increasing @xmath20 for smaller @xmath46 , high enough to reach the high - concentration limit in the density of states . because of the limited available computer power it was not feasible to directly study the anticipated corresponding convergence in the behavior of the critical states with increasing scatterer concentration .
this should be addressed in future work .
third , the behavior has been extrapolated from relatively small but non - zero @xmath69 .
although there is no indication for a discontinuous behavior when @xmath104 , there is no guarantee that we can extrapolate our data that far .
however , the range of computed @xmath69 seems to be comparable with experiments .
for instance , combining atomic force microscopy with selective etching , the image of the topology of interfaces in algaas / gaas quantum well structures was shown to exhibit smoothly varying structures with correlation length in the range 4@xmath143 m to 2@xmath144m.@xcite for the latter value , this leads to approximately @xmath145 when the magnetic flux density is varied between @xmath146 and 1 tesla .
our findings for the disorder broadening of the lowest landau level are in qualitative agreement with results obtained previously by ando and uemura.@xcite the basic difference to our work is the incompatible definition of the disorder potential : our @xmath147 is normalized to a given second moment , whereas ando and uemura normalize their disorder potential to a fixed integral @xmath148 ( @xmath51-property ) .
this leads to an additional factor @xmath46 in the potential strength , @xmath149 . in the limit of small correlation length @xmath48
, we find for the broadening of the lowest landau level a linear @xmath69-dependence , @xmath150 , which directly translates into ando and uemura s independence on the correlation length , @xmath151 . in the opposite limit of large @xmath69
, we find that the broadening of the dos just corresponds to the second moment of the disorder potential , again perfectly in agreement with their results .
for the behavior of the critical states in the limit @xmath8 , not considered by ando and uemura , the alternative normalization of the potential strength may lead to significantly different results . in our case
, we consider it physically correct to extract the @xmath27-dependence of the energy of the critical states from the @xmath69-dependence of the model with a fixed second moment @xmath152 of the disorder potential in natural units .
if we used instead the @xmath69-dependence for a preserved @xmath51-property of the potential , i.e. , @xmath153 as for the potential used by ando and uemura,@xcite we would obtain a singular dependence @xmath154 for @xmath8 which corresponds better to the levitation scenario proposed by khmelnitskii and laughlin .
we investigated a disordered two - dimensional continuum model for the integer quantum hall effect projected onto the lowest two landau levels .
the chosen dimensionless representation enabled us to make statements about the energetic shift of both , the magnetic field and correlations length dependence of the density of states peak and of the extended states .
the results were obtained from a numerical investigation of the model , parameterized by the strength @xmath70 and the correlation length @xmath69 of an effective disorder potential only .
the magnetic field @xmath27 dependence was entirely absorbed into the units of length and energy .
for the two - level continuum model we found qualitatively the same behavior as has been observed for disordered harper models projected onto the lowest two sub - bands.@xcite with increasing disorder strength @xmath70 the landau bands broaden linearly .
we clearly observed floating up in energy of the lowest landau band s extended states while the peak of the density of states moves downwards .
both effects are proportional to the square of the disorder strength . in the limit @xmath8 , which is difficult to access in numerical studies on lattice models , we find for the projected two - band continuum model that the energy of the extended states only floats up to a finite value that depends on disorder .
as this result is at variance with the diverging floating scenario suggested 20 years ago@xcite , we discuss a few limitations of our work that may be responsible for the discrepancy . on the other hand ,
our result is in accord with some experimental work .
still , the addition of higher landau levels and the investigation of the relevant matrix elements are clearly needed to settle this issue and to gain a deeper insight into the microscopic origin of the levitation of the extended states and the corresponding quantum hall to insulator transition .
we feel that to follow up our approach within a projected continuum model will render this objective possible .
we thank bodo huckestein for valuable discussions .
this research was supported by the dfg - schwerpunktprogramm `` quantum - hall - systeme '' grant no .
schw795/1 - 1 . | we investigate the fate of the quantum hall extended states within a continuum model with spatially correlated disorder potentials .
the model can be projected onto a couple of the lowest landau bands .
levitation of the @xmath0 critical states is observed if at least the two lowest landau bands are considered .
the dependence on the magnetic length @xmath1 and on the correlation length of the disorder potential @xmath2 is combined into a single dimensionless parameter @xmath3 .
this enables us to study the behavior of the critical states for vanishing magnetic field . in the two landau band limit
, we find a disorder dependent saturation of the critical states levitation which is in contrast to earlier propositions , but in accord with some experiments . |
the discovery of methane bands in the spectrum of gl 229b ( nakajima et al 1995 , oppenheimer , kulkarni , matthews & nakajima 1995 , geballe , kulkarni , woodward , & sloan 1996 ) has not only helped to identify gl 229b as a brown dwarf but also prompted the creation of the new spectral class of t dwarfs ( kirkpatrick et al 1999 ) .
the synthetic continuum spectra of the object has been obtained with and without dust particles ( marley et al .
1996 , griffith , yelle & marley 1998 , for a review see allard et al .
although the incorporation of condensates can explain the very rapid decline of the observed continuum flux in the optical region , it is shown ( tsuji , ohnaka , & aoki 1999 , burrows , marley , & sharp 1999 ) that the pressure broadened red wing of the 0.77 @xmath0 k i doublet could also account for the observed features of the continuum flux shortward of 1.1 @xmath0 .
the spectrum of the first field methane t dwarf sdss 1624 + 0029 shows a broad band absorption feature centered at 7700 @xmath7 which is interpreted ( liebert et al . 2000 ) as the k i 7665/7699 resonance doublet .
hence , it is most likely that the shape of the red spectrum is due to the broad wings of the k i and the na i doublets and not due to the presence of condensates .
all these model spectra together with the bolometric luminosity of the object ( leggett et al .
1999 ) and the evolutionary sequences ( saumon et al .
1996 ) can constrain the effective temperature of the object very tightly , however the surface gravity is still poorly constrained .
it is shown ( saumon et al . 2000 ) that a multi - parameter fit of the observed spectrum both for the k - band and for the red end is possible with different metallicities .
therefore , in order to determine the physical properties of the atmosphere of gl 229b and the other t dwarfs uniquely , more comprehensive theoretical modeling of the observed spectrum is required .
one of the most important studies is the individual line formation by the most abundant molecules , eg .
@xmath8 , @xmath9 , @xmath10 etc . in this paper
, we for the first time , attempt to model the line formed by methane at 2.3 @xmath0 and show that if individual lines of abundant molecules , in particular that of methane can be resolved observationally , then stringent constraint on the surface gravity , on the metallicity and on the temperature at the bottom of the atmosphere , where the optical depth is very high , can be obtained .
in order to model the individual line of a particular molecule one needs to solve the non - lte line transfer equations . the two level atom line transfer equation in the plane
parallel stratified medium can be written as ( mihalas 1978 ) @xmath11\ , , \end{aligned}\ ] ] where @xmath12 is the cosine of the angle made by the ray to the normal .
the total absorption coefficient @xmath13 where @xmath14 is the frequency integrated line absorption coefficient , @xmath15 , @xmath16 is the absorption coefficient for the continuum and @xmath17 is the line profile function .
the source function is given by @xmath18 where @xmath19 is the continuum source function which is set equal to the planck function @xmath20 and the line source function @xmath21 is written as @xmath22 where @xmath23 is the probability per scattering that the photon is destroyed by collisional de - excitation and is given by ( mihalas 1978 ) @xmath24 @xmath25 is the rate of collisional de - excitation and @xmath26 is the rate of spontaneous emission .
@xmath27 is the redistribution function defined according to a hybrid model prescription by rees & saliba ( 1982 ) and is given by @xmath28 where @xmath29 is the phase function and @xmath30 is the angle - averaged frequency redistribution function for isotropic scattering .
under chemical equilibrium among different atomic and molecular species in their standard states , the number density of @xmath8 for the molecular equilibrium reaction @xmath31 can be written as ( sharp 1985 ) @xmath32 where @xmath33 and @xmath34 are the partition functions for @xmath35 and @xmath36 , @xmath37 is the `` multiple '' reduced mass and @xmath38 is the dissociation energy of @xmath8 in ev , @xmath39 and @xmath40 are the number densities of @xmath35 and @xmath36 , @xmath41 where @xmath42 is in kelvin , and the other symbols have their usual meaning .
although chemical equilibrium of @xmath35 and @xmath36 should be coupled with some other molecules , dominance of @xmath8 molecule in the atmosphere would make the above equilibrium the most probable one .
hence , we ignore , for the present , chemical equilibrium of @xmath35 and @xmath36 with other molecules . for the purpose of calculating the number densities of the molecules in various energy levels ,
we assume that the lower atmosphere of the object is in local thermodynamic equilibrium ( lte ) .
then the number of particles @xmath43 , in a specified rotation level @xmath44 is related to the total number of particles in all levels , @xmath45 , according to the following relation ( larson 1994 ) : @xmath46 where ( 2j+1 ) is the degeneracy factor and @xmath47 is the boltzmann factor .
now the integrated line absorption coefficient can be written as @xmath48^{3}10^{d_o\theta(z)}\times \nonumber \\ & & \frac{(2j+1)e^{-e_j / kt(z)}}{q_cq^2_{h_2}}n_c(z)n^2_{h_2}(z)\ , , \end{aligned}\ ] ] where @xmath49 is the oscillator strength of the transition .
we calculate the q s for @xmath36 and @xmath35 using polynomial expressions given by sauval & tatum(1984 ) . for a small value of @xmath44
, the boltzmann factor reduces to 1 .
the dissociation energy @xmath38 of @xmath8 is taken to be 4.38 ev ( jorgenson 1994 ) .
we use the continuum absorption coefficient @xmath16 provided by d. saumon .
we present the synthetic continuum spectra of gl 229b ( figure 1 ) in the infra - red region where the signature of methane is clear and dominant .
the monochromatic radiative transfer equations are solved numerically by using discrete space theory ( peraiah and grant 1973 ) .
we have adopted two sets of model parameters , model ( a ) : t@xmath50=940 k , log g=5.0 ( g in cms@xmath51 ) and [ m / h]=@xmath52 ( k band ) and model ( b ) : @xmath53=1030 k , log g=5.5 and [ m / h]=@xmath54 ( k band ) .
the temperature and pressure profiles for both the models are obtained from m. marley ( private communication ) .
the values of @xmath53 and the surface gravity g are constrained by the bolometric luminosity of gl 229b and the evolutionary sequence of saumon et al ( 1996 ) .
these synthetic spectra fit the entire observed spectrum of gl 229b except in the near infra - red region .
it should be mentioned that a good fit with the observational data for the entire wavelength region is not possible with the same value of the metallicity and the surface gravity .
the above sets of model parameters are two of many optimal sets of parameters that can produce good fit . in order to match the observed flux shortward of 1.1 @xmath0 that declines very rapidly , either dust particulates or alkali metals have to be incorporated .
however , the size of the grain that is needed to explain the observed spectrum in the optical region is too small to play any role in the infra - red region we are interested in the present work . hence in this spectral region the law of mie scattering that describes the angular distribution of photons in a dusty atmosphere reduces to that of rayleigh scattering .
therefore , the synthetic spectrum in this region matches well with the observed flux when rayleigh s law of scattering is used .
recent ovservational evidence ( liebert et al . 2000 ) implies that there is no compelling reason to introduce dust or additional opacity source in the atmosphere of methane dwarf .
therefore in the present work we have not incorporated dust opacity .
figure 1 shows that there is no difference in the calculated continuum flux with the two sets of model parameters .
both the models fit the observed data very well .
the continuum opacity and the temperature - pressure profile of the atmosphere are set by matching the synthetic continuum spectrum with the observed spectrum for the entire wavelength region .
the modeling of individual lines needs the values of additional parameters . in the absence of observational data for individual lines
, we set the value of these parameters by matching the calculated flux in the continuum with the observed continuum flux at 2.3 @xmath0 .
since the line is very weak in intensity , we choose the profile function as ( mihalas 1978 ) @xmath55 ^ 2},\ ] ] where @xmath56 is the thermal doppler width and @xmath57 is the line center
. it should be worth mentioning that individual molecular lines are usually not saturated enough so that pressure broadening is less important for them as compared to strong atomic lines .
moreover , molecular lines often overlap so strongly that their wings are completely masked ( schweitzer et al .
1996 ) and only the gaussian line cores of the strongest molecular transitions are observed .
the atmosphere of a brown dwarf is therefore only weakly sensitive to the van der walls damping constant .
nothing is known , at present , about the rotation of the brown dwarf gl 229b around its own axis of rotation .
if the projected velocity ` @xmath58 ' of the object is greater than 2 to 5 @xmath59 then rotational broadening could be significant .
however , in the present work we have neglected rotational broadening in order to make the results consistent with the calculation of the evolutionary sequences by saumon et al .
( 1996 ) that constrains the surface gravity and the effective temperature of the object
. the whole purpose of the present work is to show that with different values of the surface gravity and the metallicity , the flux at the line core is significantly different although it is the same in the continuum . since rotational broadening
would affect the spectrum equally for both the models , it is not important in the context of the present work . we assume complete frequency redistribution and use rayleigh phase function for the angular redistribution . the parameters that are to be set in order to model the @xmath8 line at 2.3 @xmath0 are @xmath23 , @xmath49 , @xmath60 , @xmath61 , and the degeneracy factor @xmath62 .
we define @xmath63 guided by equation ( [ lineabs ] ) and set the values of @xmath64 and @xmath23 such that the calculated value of the flux at the continuum matches with the observed continuum flux .
we assume that @xmath23 is independent of the geometrical depth .
this is valid if the @xmath8 line formation is confined to a narrow region in the atmosphere . after testing several empirical laws
we adopt the usual inverse square law with respect to the geometrical depth for the variation of the number density of @xmath35 and @xmath36 .
this should be verified when observational data becomes available .
experimental determination of the oscillator strengths for different transitions at 2.3 @xmath0 and at the relevant temperature could further constrain the number density of @xmath35 and @xmath36 and hence the abundance of methane in the atmosphere of gl 229b provided chemical equilibrium exists between the molecule and the atoms .
we have solved the radiative line transfer equations by using discrete space theory ( peraiah 1980 ) .
the theoretical models for the methane line at 2.3 @xmath0 are presented in figure 2 . for the model ( a ) we find that the flux at the continuum matches with its observed value when @xmath65 and @xmath66 where @xmath67 and @xmath68 are the number densities of @xmath35 and @xmath36 respectively at the bottom of the atmosphere .
the moderately high value of @xmath23 is consistent with the temperature at the lower atmosphere where the lines are formed . for the model ( b )
the flux at the continuum matches with the observed flux when @xmath65 and @xmath69 . since the temperature of the lower atmosphere for
the two models does not differ much , the value of @xmath23 remains the same for both the models .
however @xmath64 differs substantially for the two models .
this is because of the fact that for the model ( a ) the continuum opacity is less and therefore one has to reduce the line opacity in order to keep the right ratio ( @xmath70 ) between them so that the calculated flux matches with the observed flux at the continuum .
however , the higher value of @xmath64 in the model ( b ) makes the line opacity higher than that for the model ( a ) and so substantial decrease in the calculated flux at the line core is obtained .
the difference in the flux reduces as we go towards the wings .
it is worth mentioning that the wings could be masked by other lines whereas the continuum is overlapped by several lines .
figure 2 shows that a spectral resolution as high as 200,000 at @xmath71 is needed in order to investigate the individual molecular lines
. this may be possible with an appropriate combination of the telescope and the instrument .
for example , the cooled grating spectrometer 4 ( cgs4 ) available in ukirt ( united kingdome infra - red telescope ) has a spectral resolution upto 40,000 .
if observation of gl 229b is possible at present by ukirt with the maximum resolution power of cgs4 then keeping the signal to noise ratio ( which is proportional to the diameter of the telescope and to the square root of the integration time of exposure ) unaltered , a 10 m telescope ( such as keck i ) can obtain the desired resolution by using a similar type of spectrometer provided the resolution of the instrument is increased by about five times and the integration time of exposure is increased by about 2.5 times .
it is found that the numerical values of @xmath23 and @xmath64 are very much sensitive to the emergent flux .
the difference in the flux at the line core is clearly due to the different values of the surface gravity and the metallicity .
theoretical modeling of the continuum flux provides a few possible combinations of the metallicity , surface gravity and the effective temperature that are appropriate in explaining the observed continuum flux .
the observational fit of the flux at the line core would decide which one of these combinations should describe the physical properties of the atmosphere .
the physical parameters that are needed to model the individual molecular line will also be fixed once observational data is available .
therefore , in conclusion we would like to emphasize that a theoretical fitting of the observed flux for the individual lines of any of the dominant molecules , especially methane , would not only provide a much better understanding on the abundance of that molecule and the temperature of the lower atmosphere but also improve the constraints on the value of the surface gravity of brown dwarfs .
we thank m. marley for kindly providing the temperature - pressure profiles and d. saumon for the continuum opacity data .
we also thank the referee for valuable suggestions and constructive comments .
thanks are due to n. k. rao , t. p. prabhu , t. n. rengarajan , k. e. rangarajan and g. pandey for useful discussions .
allard , f. , hauschildt , p. h. , alexander , d. r. & starrfield , s. 1997 , ara&a , 35 , 137 burrows , a. , marley , m. s. , & sharp , c. 1999 , astro - ph/9908088 geballe , t. r. , kulkarni , s. r. , woodward , c. e. , & sloan , g. c. 1996 , apj , 467 , l101 griffith , c. a. , yelle , r. a. , & marley , m. s. 1998 , science , 282 , 2063 jorgenson , u. g. 1994 , in molecules in the stellar environment , ed .
u. g. jorgenson ( berlin : springer ) , 29 kirkpatrick , j. d. , et al .
1999 , apj , 519 , 834 larson , m. 1994 , in molecules in the stellar environment , ed .
u. g. jorgenson ( berlin : springer ) , 271 liebert , j. , reid , i. n. , burrows , a. , burgasser , a. j. , kirkpatrick , j. d. & gizis , j. e. 2000 , apj , 533 , l155 marley , m. s. , saumon , d. , guillot , t. , freedman r. s. , hubbard , w. b. , burrows , a. & lunine , j. i. 1996 , science , 272 , 1919 mihalas , d. 1978 , stellar atmospheres , 2nd ed .
( sanfrancisco : freeman ) nakajima , t. , et al .
1995 , nature , 378 , 463 openheimer , b. r. , kulkarni , s. r. , matthews , k. , & nakajima , t. 1995 , science , 270 , 1478 peraiah , a. 1980 , in methods in radiative transfer , ed .
w. kalkofen ( cambridge : cambridge univ . ) , 281 peraiah , a. , & grant , i. p. 1973 , j. inst .
12 , 75 rees , d. , & saliba , g. 1982 , a & a , 115 , 1 saumon , d. , hubbard , w. b. , burrows , a. , guillot , t. , lunine , j. i. , & chabrier , g. 1996 , apj , 460 , 993 saumon , d. , geballe , t. r. , leggett , s. k. , marley , m. s. , freedman , r. s. , lodders , k. , fegley , b. , & sengupta , s. k. 2000 , apj ( in press ) sauval , a. j. , & tatum , j. b. 1984 , apjs , 56 , 193 schweitzer , a. , hanschildt , p. h. , allard , f. , & basri , g. 1996 , mnras , 283 , 821 sharp , c. m. 1985 , in molecular astrophysics , ed . g. h. f. diercksen , w. f. huebner & p. w. langhoff(dordrecht : reidel ) , 661 tsuji , t. , ohnaka , k. , & aoki , w. 1999 , apj , 520 , l119 | we investigate the formation of methane line at 2.3 @xmath0 in brown dwarf gliese 229b .
two sets of model parameters with ( a ) @xmath1k and @xmath2 , ( b ) @xmath3k and @xmath4 are adopted both of which provide excellent fit for the synthetic continuum spectra with the observed flux at a wide range of wavelengths . in the absence of observational data for individual molecular lines , we set the additional parameters that are needed in order to model the individual lines by fitting the calculated flux with the observed flux at the continuum .
a significant difference in the amount of flux at the core of the line is found with the two different models although the flux at the continuum remains the same .
hence , we show that if spectroscopic observation at @xmath5 with a resolution as high as @xmath6 is possible then a much better constraint on the surface gravity and on the metallicity of the object could be obtained by fitting the theoretical model of individual molecular line with the observed data . |
let @xmath0 be a number field . by a well - known result there exists a dessin denfant ( cover of the projective line unramified outside three points ) having @xmath0 as field of moduli .
the belyi degree of @xmath0 is the minimal degree of such a cover .
the present note is an invitation to the study of this invariant .
the paper is organized as follows : in the first section , we give some general results concerning the belyi degree , as for example the fact that there exist finitely many number fields of bounded ( belyi ) degree . in 2 , we obtain a lower bound for such an invariant ( theorem 4 ) , which follows from a result of s. beckmannn concerning the ramified primes in the field of moduli of a cover of the projective line defined over a number field .
although this bound seems rough , it can be shown that it is optimal. indeed , in 4 we prove that it is reached for infinitely many imaginary quadratic fields .
in 3 we use a construction of r. licanu and a result of d. roy and j. l. thunder in order to obtain an upper bound of the belyi degree ( theorem 7 ) , only depending on the degree and on the ( absolute value ) of the discriminant of the number field .
more precisely , we prove that once we have fixed the degree , the growth of the belyi degree is at most polynomial on the discriminant .
this leads to the introduction of the minimal discriminant exponent . in 4 , which is a computational approach to the belyi degree of quadratic fields
, we obtain an upper bound which is linear on the discriminant and in some cases , we are also able to completely determine this invariant . as a consequence
we prove that the minimal discriminant exponent is equal to @xmath1 for these fields .
we also include two appendices : in the first we generalize two results of h. w. lenstra which are very useful when studying the field of moduli of a genus @xmath2 dessin denfant ; in the second we give the explicit equations for the dessins denfants introduced in 4 .
even if it is a well - known fact ( see for example @xcite , or more generally @xcite for an introduction to dessins denfants ) we start this section by proving that any number field appears as the field of moduli of a dessin denfant .
let @xmath0 be a number field .
then there exists a dessin denfant having @xmath0 as field of moduli . by the primitive element theorem
, there exists an element @xmath3 such that @xmath4 .
let @xmath5 be an elliptic curve having @xmath6 as @xmath7-invariant .
following @xcite , @xmath0 is the smallest field of definition of @xmath5 .
fix a model of @xmath5 defined over @xmath0 and a rational function @xmath8 , which induces a cover @xmath9 defined over @xmath0 . following the algorithm in @xcite
, there exists a polynomial @xmath10 $ ] such that the composite cover @xmath11 induces a cover of the projective line unramified outside @xmath12 and @xmath1 .
consider now an element @xmath13 .
then @xmath14 belongs to @xmath15 if and only if @xmath16 . in this case , by construction ( since @xmath5 and @xmath17 are defined over @xmath0 and @xmath18 is defined over @xmath19 ) , we find @xmath20 and thus the cover @xmath21 is defined over @xmath0 .
if @xmath22 then the conjugate cover @xmath23 can not be isomorphic to @xmath21 , since @xmath5 and @xmath24 have different @xmath7-invariants .
this shows that @xmath0 is the field of moduli ( and even the minimal field of definition ) of @xmath21 .
the _ belyi degree _ of a number field @xmath0 is the smallest degree of a dessin denfant having @xmath0 as field of moduli .
[ prop1 ] there exist finitely many number fields of fixed belyi degree .
two isomorphic number fields have the same belyi degree .
the first assertion directly follows from the fact that there are finitely many dessins denfants of given degree .
if @xmath25 is an isomorphism of number fields and @xmath26 is a dessin denfant having @xmath0 as field of moduli , then @xmath27 has @xmath28 as field of moduli .
[ prop2 ] for any number field @xmath0 , we have the inequality @xmath29 where @xmath30 is the greatest prime which ramifies in @xmath0 . following a result of s. beckmann @xcite ,
the field of moduli @xmath0 of a dessin denfant is unramified outside the primes dividing the order of its monodromy group . in particular ,
if the degree of the dessin denfant is strictly less than a prime @xmath30 , then the order of its monodromy group divides @xmath31 and thus @xmath30 does not ramifies in @xmath0 .
this result seems quite rough but we will see in 4 that this bound is reached for infinitely many ( imaginary quadratic ) number fields .
in this section , we follow the ideas and techniques of r. licanu in @xcite and then use a result of d. roy and j. l. thunder in @xcite .
we start with some preliminary results on heights .
let @xmath0 be a number field and consider an element @xmath32\in\mathbf p^n(k)$ ] .
recall that the ( multiplicative ) height of @xmath33 is defined as @xmath34 where the product is taken over all the places of @xmath0 ( in order to avoid exponents , we distinguish two conjugated places ) .
the normalized height is defined as @xmath35 where @xmath36 $ ] . for an element
@xmath37 we set @xmath38)$ ] .
recall also that the height of a polynomial @xmath39 $ ] is defined as @xmath40)$ ] .
in particular if @xmath41 are integers with no common divisor , we simply get @xmath42 [ lemma2 ] for any polynomial @xmath43 $ ] , we have the inequality @xmath44 we clearly have the identity @xmath45).\ ] ] if @xmath46 is a finite place of @xmath0 , we obtain @xmath47 while for @xmath46 infinite we get @xmath48 and the result follows by taking the product and then @xmath49-th roots .
we will also use the fundamental inequalities @xmath50 which holds for any @xmath51 $ ] of degree @xmath52 and any @xmath37 ( see for example @xcite , p. 13 ) and @xmath53 for any polynomial @xmath54 $ ] ( cf .
@xcite , theorem 5.9 ) .
[ lemma3 ] consider a monic polynomial @xmath51 $ ] of degree @xmath52 and the factorization @xmath55 .
setting @xmath56,\ ] ] we have @xmath57 and the inequality @xmath58 following lemma [ lemma2 ] and the above inequalities , we obtain @xmath59 from which the lemma follows .
[ upper_bound ] the belyi degree of a number field @xmath0 is bounded by a constant only depending on its degree @xmath52 and on its discriminant @xmath60 .
more precisely , there exist two effective constants @xmath61 only depending on @xmath52 such that we have the inequality @xmath62 let @xmath63 be a basis of @xmath0 over @xmath19 and consider the polynomial @xmath64.\ ] ] the cover induced by @xmath65 is unramified outside the set @xmath66 , where @xmath67 are the roots of @xmath68 . following the notation of lemma [ lemma3 ] , consider the polynomial @xmath69\ ] ] of degree @xmath70 and for any @xmath71 define @xmath72 $ ] recursively by the relation @xmath73 .
set also @xmath74 so that @xmath75 has cardinality less than or equal to @xmath76 and the cover induced by the polynomial @xmath77 is unramified outside this set . since @xmath78 following lemma [ lemma3 ] , we find @xmath79^{1 + 2(n+1)^2+\cdots+(2(n+1)^2)^{m-2}}h(f_1)^{2^{m-1}(n+1)^{2(m-1)}}\\ & \leq\left[2^{(n+1)^4}(n+1)^{2(n+1)^2}h(f_1)\right]^{2^{m-1}(n+1)^{2(m-1)}}=a_0h(f_1)^{b_0 } \endaligned\ ] ] with @xmath80 and @xmath81 only depending on @xmath52 . since @xmath82
where @xmath83 and @xmath84 only depend on @xmath52 , we then obtain the inequality @xmath85 where , once again , the constants @xmath86 and @xmath87 only depend on @xmath52 . by construction , the set @xmath88 is contained in @xmath89 and we have @xmath90 with @xmath91 only depending on @xmath52 .
now , following @xcite , lemme 4.1 ( and its proof ) , there exists a rational function @xmath92 such that the cover induced by the rational function @xmath93 is unramified outside @xmath12 and @xmath1 .
moreover , we have the bound @xmath94 now , following @xcite , theorem 2 , there exist @xmath65 such that @xmath95 so that we obtain the desired inequality .
we just have to check that the field of moduli of the cover induced by @xmath17 is @xmath0 .
this directly follows from appendix a : by construction , @xmath17 is defined over @xmath0 .
let @xmath96 not belonging to @xmath97 .
we must prove that the cover induced by @xmath98 is not isomorphic to the one induced by @xmath17 .
suppose that the opposite holds .
this can be restated as @xmath99 with @xmath100 .
remark that , by construction , the point at infinity is the only ramified point of @xmath17 in its fiber ( above @xmath1 ) .
this means that @xmath101 and thus @xmath102 with @xmath103 and @xmath104 . setting @xmath105
, we have @xmath106 and lemma [ lenstra_2 ] implies that there exist @xmath107 such that @xmath108 now , since the coefficient of @xmath109 vanishes for both @xmath65 and @xmath110 , we obtain @xmath111 .
similarly , since @xmath112 , we find @xmath113 . finally ,
comparing the coefficients of @xmath114 and @xmath115 , we obtain @xmath116 which gives @xmath117 and thus @xmath118 , which is impossible .
this concludes the proof of the theorem . as it is done in @xcite
, it is possible to give an explicit bound for the constants @xmath119 and @xmath120 . in order to obtain a light proof
, we decided to omit this computation .
the above theorem asserts that , once we have fixed the degree of the number field , the growth of its belyi degree is at most polynomial on its discriminant .
it is then natural to ask which is the least exponent for which such a inequality holds .
this leads to the notion of the minimal discriminant exponent .
let @xmath121 be an integer .
the _ minimal discriminant exponent _ is the real number @xmath122 , where we have set @xmath123=n\right\}.\ ] ] we say that @xmath124 is _ effective _ if it belongs to @xmath125 .
remark that @xmath124 is effective if and only if @xmath125 is a closed set ( this is the usual problem of passing from @xmath126 do @xmath124 ) .
this is also equivalent to the existence of a global bound for the constants @xmath119 appearing in the definition of @xmath125 . as we will se in the next section , it is possible to determine the minimal discriminant exponent for quadratic fields .
we start this list of example with the study of quadratic imaginary fields . as we will see , in some cases , it is possible to completely determine the belyi degree .
[ bound1 ] let @xmath49 be a positive integer and set @xmath127 .
we then have the inequality @xmath128 the two dessins denfants in figure 1 are conjugated and have @xmath0 as field of moduli ; their degree being @xmath129 , the result follows .
.4 cm + figure 1 .4 cm the following result shows that the bound in theorem [ prop2 ] is sharp .
let @xmath130 three positive integers such that @xmath131 is prime .
setting @xmath132 , we have the identity @xmath133 in particular , there exist infinitely many ( quadratic imaginary ) number fields such that the inequality in theorem [ prop2 ] is an equality . since @xmath30
does not divide @xmath134 , it is the greatest prime which ramifies in @xmath0 .
in particular , theorem [ prop2 ] asserts that the belyi degree of @xmath0 is greater than or equal to @xmath30 .
we just have to construct a dessin denfants of degree @xmath30 having @xmath0 as field of moduli .
the two dessins denfants described in figure 2 are conjugated ( as soon as @xmath135 and @xmath136 are pairwise distinct ) of degree @xmath30 and with @xmath0 as field of moduli .
the last part of the proposition simply follows from the fact that there exist infinitely many primes .
let @xmath137 be a prime not congruent to @xmath1 modulo @xmath138 and set @xmath139 .
we then have the identity @xmath133 the congruence condition implies that @xmath30 is congruent either to @xmath140 modulo @xmath141 or to @xmath141 modulo @xmath142 . in the first case , let @xmath143 and consider the dessin denfant described in the proof of the above proposition with @xmath144 and @xmath145 .
in the second case , let @xmath146 and take @xmath147 and @xmath148 .
.4 cm + figure 2 .4 cm
we now study the case of real quadratic fields .
the following result is similar to proposition [ bound1 ] , and even a little bit sharper .
let @xmath149 be an integer and set @xmath150 .
we then have the inequality @xmath151 the two dessins denfants in figure 3 are conjugated and have @xmath0 as field of moduli ; their degree being @xmath152 , the result follows .
.4 cm + figure 3 .4 cm the results of the last two paragraphs allow us to completely determine the minimal discriminant exponent for quadratic fields .
the minimal discriminant exponent for quadratic fields is equal to @xmath153 and is effective .
set @xmath154 with @xmath49 squarefree , so that @xmath155 .
the last two bounds for the belyi degree lead to @xmath156 this implies that @xmath157 ( cf . the notation of 3 ) .
suppose that @xmath158 .
this means that there exists @xmath159 and a constant @xmath119 such that @xmath160 now , there exist infinitely many prime numbers @xmath30 not congruent to @xmath1 modulo @xmath138 and , setting @xmath161 , corollary 12 implies that @xmath162 . in particular , for large enough @xmath30 , we find @xmath163 and thus @xmath164 which is impossible .
this concludes the proof of the proposition .
in this first appendix , we generalize two results of h. w. lenstra in @xcite .
[ lenstra_1 ] let @xmath165 be a rational function of degree @xmath166 and suppose that there exists a monic polynomial @xmath167 $ ] of degree @xmath49 such that @xmath168 for a rational function @xmath169 .
then @xmath21 is unique .
set @xmath170 with @xmath171 $ ] coprime and @xmath172 with @xmath173 $ ] coprime .
we can suppose that @xmath174 .
since @xmath175 and @xmath176 are coprime , the identity @xmath177 implies that there exists @xmath178 such that @xmath179 where we have set @xmath180 $ ] .
but in this case , following @xcite , lemma ii.2 , @xmath21 is unique .
[ lenstra_2 ] let @xmath181 and @xmath182 $ ] of the same degree such that @xmath183 .
then there exist @xmath184 such that @xmath185 .
let @xmath186 be the leading coefficient of @xmath187 and consider the monic polynomial @xmath188.\ ] ] in this case , setting @xmath189 we obtain the identity @xmath190 and the above lemma implies that @xmath191 , from which the result follows .
we close this paper by giving the explicit equations for the dessins denfants used in the proofs in [ examples ] .
although they look quite different , they can be studied in the same way .
consider three integers @xmath135 and @xmath136 with @xmath192 and @xmath193 .
we also assume that @xmath194 are pairwise distinct .
we want to describe all the rational functions @xmath195 with @xmath196 such that the logarithmic derivative @xmath197 has a unique zero at the origin . in this case
, @xmath17 induces a cover @xmath198 unramified outside @xmath12 and @xmath1 .
remark that the condition @xmath193 implies that @xmath17 has a pole at infinity .
it easily turns out that the condition on the logarithmic derivative of @xmath17 is equivalent to the two equations @xmath199 which lead to @xmath200 if @xmath201 is not a perfect square then we have @xmath196 , as desired .
clearly , the rational function @xmath17 is defined over the field @xmath202 .
it is then easily shown that @xmath0 is in fact its field of moduli ( this comes from the fact that @xmath194 are pairwise distinct ) . ,
volume 200 of _ london mathematical society lecture note series_. cambridge university press , cambridge , 1994 .
papers from the conference on dessins denfant held in luminy , april 1924 , 1993 , edited by leila schneps
. leila schneps .
dessins denfants on the riemann sphere . in _ the grothendieck theory of dessins denfants ( luminy , 1993 ) _ , volume 200 of _ london math .
lecture note ser .
_ , pages 4777 .
cambridge univ .
press , cambridge , 1994 .
jean - pierre serre .
. aspects of mathematics .
vieweg & sohn , braunschweig , third edition , 1997 . translated from the french and edited by martin brown from notes by michel waldschmidt , with a foreword by brown and serre . | in this short note we introduce the belyi degree of a number field @xmath0 , which is the smallest degree of a dessin denfant having @xmath0 as field of moduli . after the description of some general properties ( for example , the fact that there exist finitely many number fields of bounded belyi degree ) , we give a lower and an upper bound for such an invariant .
we finally give some explicit examples for quadratic fields . |
leptons in @xmath0-hadron decays can be produced both from direct @xmath12 and cascade @xmath13 decays which can be separated exploiting the correlation between the lepton charge and the @xmath0-hadron flavour .
the wrong sign charm decay @xmath14 , can be isolated from the direct decay by the low momentum signature of the leptons ; the present analysis gives the first direct measurement of @xmath15 . from about @xmath16 million
hadronic @xmath17 decays collected during 1994 , a @xmath18 event sample was selected with efficiency @xmath19 and purity @xmath20 by means of the impact parameter , energy and angular distributions of the reconstructed charged tracks .
an inclusive @xmath1 reconstruction was performed from the high rapidity particles in the jet and a secondary vertex was fitted ; the energy and angular resolutions were @xmath21 and @xmath22 respectively .
the charge of the @xmath1 hadron was determined as @xmath23 where @xmath24 is the charge of the @xmath25 track and @xmath26 is the probability for the track to come from @xmath1 decay , computed using a neural network in terms of kinematical and topological variables .
a second neural network was used in order to determine the @xmath0-hadron flavour , exploiting the charge of identified @xmath27 and slow pions from @xmath6 decay , the jet charge and @xmath28 .
a flavour discrimination @xmath29 was obtained with an efficiency @xmath30 .
the lepton energy in the @xmath0-hadron rest frame @xmath31 was computed from the @xmath1 and the lepton 4-momenta with a resolution @xmath32 .
the background due to non @xmath0-events , fake @xmath0-hadrons and fake leptons was determined from the simulation and subtracted from the experimental @xmath31 distribution . from a fit to @xmath33 the following preliminary results were obtained : @xmath34 the dominant systematics came from the lepton identification , the neural networks performances and the heavy flavour semileptonic decay models .
figure [ fig : lfit ] shows the results of the fit for the same and opposite sign @xmath35 charge correlation samples obtained using the accm model . -.2
cm 3.5truein 3.5truein 3.5truein 3.5truein
the heavy quark effective theory @xcite allows a precise determination of @xmath3 , the element of the ckm mixing matrix relating the @xmath0 and the @xmath36 quarks , from the analysis of the semileptonic decay @xmath37 .
the following relation holds : @xmath38 where @xmath39 is the product of the two heavy mesons 4-velocities and is related to the momentum transfer : @xmath40 .
@xmath41 is a phase space function and @xmath42 is a model dependent form factor , which can be expressed in terms of a single unknown parameter @xmath43 @xcite : @xmath44 in the limit in which the @xmath6 meson is produced at rest in the @xmath1 reference frame @xmath45 , the theory predicts@xcite @xmath46 experimentally , the goal is to measure @xmath47 at the point of zero recoil @xmath48 . due to the vanishing of phase space ,
the product @xmath49 was determined by an extrapolation from the @xmath50 distribution .
the data collected from 1992 to 1995 were used in the analysis .
@xmath51 mesons were inclusively reconstructed from all the particles in the lepton jet , by means of their rapidity and impact parameter .
pions with charge opposite to the lepton were then added to build a @xmath6 , and events finally tagged based on the mass difference @xmath52 .
the combinatorial background shape was estimated from the @xmath53 sample with the wrong charge correlation .
a signal sample of @xmath54 events was found with a selection efficiency @xmath55 .
figure [ fig : dm1 ] shows the @xmath56 distribution for the final sample .
3.5truein 4.truein 3.5truein 4.truein -.2 cm the 4-momentum of the @xmath1 was obtained from energy and momentum conservation and from the secondary vertex reconstruction ; the energy and angular resolutions were @xmath57 and @xmath58 respectively .
the dominant physics background was due to the semileptonic decays @xmath59 .
signal events can be distinguished from the cascade ones exploiting the squared recoil mass @xmath60 which should be zero for signal events , whereas is usually greater than zero in case of background processes due to additional particles produced in the @xmath61 decay . from a fit to the @xmath62 distribution with the @xmath63 shape fixed to hqet prediction@xcite , the ratio @xmath64 was determined .
figure [ fig : dm1 ] shows the result of the fit . from the comparison of the @xmath5 distribution with the model function [ eq :
diff ] , taking into account the experimental resolution and using the result [ eq : ffval ] , the values : @xmath65 @xmath66 were obtained , where the dominant systematics come from @xmath67 , the experimental resolution and the @xmath61 fraction and decay models .
figure [ fig : vcb ] show the result of the fit together with the @xmath47 and the decay form factor unfolded distributions .
3.5truein 3.8truein 3.5truein 3.8truein -.2 cm
the measurement of the branching ratio for the decay @xmath68 provides the most precise way to determine the @xmath69 element of the ckm mixing matrix .
evidence of the non - zero value of @xmath69 has been obtained both by observing leptons produced in @xmath1 decays with momentum above the charm kinematic threshold @xcite and by the measurement of the charmless exclusive decays @xmath70 @xcite .
these analyses were dominated by the theoretical error due respectively to the model dependence in the extrapolation to low lepton momentum and to the uncertainty on the hadronic matrix elements . according to hqet predictions at @xmath71 @xcite , @xmath69 can be extracted from the inclusive width @xmath72 with a theoretical error @xmath73 , dominated by the uncertainties on the mass of the @xmath0 quark and @xmath74 .
experimentally , a way to discriminate between the signal events and the dominant background due to the decay @xmath75 is provided by the hadronic recoil mass @xmath76 which extends to higher values for the charmed final states with respect to the charmless ones @xcite . according to recent calculations @xcite
, the signal region can be defined by the cut @xmath77 which represents a good compromise between a reasonable high @xmath78 rejection and a little model dependence in the calculation of the efficiency for the signal events @xmath79 . from the data collected between 1993 and 1995 , a @xmath18 enriched sample
was selected with purity @xmath80 and efficiency @xmath81 .
the hadronic secondary system was then reconstructed with an efficiency of about @xmath82 from all the tracks classified as coming from @xmath1 decays by means of kinematical and topological variables .
the background from @xmath78 was depleted according to the lepton impact parameter w.r.t the secondary vertex , @xmath27 identification and @xmath6 inclusive reconstruction .
the lepton energy in the @xmath1 rest frame was computed with a resolution @xmath83 . in the @xmath84 enriched sample ,
an excess of @xmath85 events was found in the data compared to the expected background ; the selection efficiency for signal events was @xmath86 .
figure [ fig : vub ] shows the @xmath87 distribution after the background subtraction for the @xmath84 enriched and depleted samples .
the fraction of @xmath68 events in the selected sample is proportional to the ratio @xmath8 . from a fit to the @xmath87 distribution the preliminary value : @xmath9 was extracted , where the systematics come from the charm decay branching ratios , the @xmath0-hadron production and decay properties , the lepton and secondary system reconstruction efficiencies and the @xmath68 modelisation .
figure [ fig : vub ] shows the result of the fit .
3.5truein 3.truein 3.5truein 3.truein -.2 cm
the semileptonic branching fraction of the @xmath36 quark was determined with a rather large error in low energy experiments @xcite .
the measurement of this quantity at a different energy scale is of great interest due to its correlation with the mixture of charmed hadrons in the final state ; a recent opal measurement@xcite shows excellent agreement with the low energy results . at lep the uncertainty on the average from the low energy experiments @xmath88 is an important source of systematic error for a number of measurements based on lepton tag analyses . from an anti-@xmath0 tagged sample of hadronic @xmath17 decays collected between 1992 and 1995 , @xmath89 events were selected by means of an exclusive @xmath90 and @xmath91 reconstruction in several channels .
charm enrichement was obtained by cutting on the fraction @xmath92 of the beam energy retained by the @xmath10 mesons .
a total of @xmath93 charmed mesons were reconstructed with a purity @xmath94 .
figure [ fig : d ] shows the reconstructed @xmath10 meson spectra for the different channels . the branching fraction @xmath95 was obtained from the amount @xmath96 of the identified leptons with @xmath97 in the hemisphere opposite to the reconstructed @xmath10 meson and with the right charge correlation : @xmath98 . in the previous expression @xmath99
is the number of the true leptons coming from @xmath36 decays , either directly produced in @xmath100 events or originating from the gluon splitting process @xmath101 in @xmath102 .
@xmath103 accounts for the true leptons from @xmath0 semileptonic decays and was computed in terms of the @xmath0 semileptonic branching fractions and the effective mixing which destroys the charge correlation between the lepton and the @xmath10 meson .
@xmath104 is the sum of the contributions due to fake leptons and true leptons from light particle decays in @xmath105 events .
small corrections @xmath106 due to hard gluon radiation were obtained from the simulation .
the lepton efficiencies were computed from the simulation tuned to reproduce the data , taking into account the detector acceptances and found to be : @xmath107 and @xmath108 for electrons and muons respectively .
the number @xmath104 of background leptons was determined from the observed multiplicity of charged particles in the lepton hemishepre and the misidentification probabilities @xmath109 and @xmath110 .
3.5truein 4.truein 3.6truein 4.truein -.2 cm the fraction of leptons from the decays of @xmath0 and @xmath36 hadrons surviving the momentum cut was obtained reweighting the simulation in order to reproduce the experimental values for @xmath111 and @xmath112 .
different semileptonic decay models were considered according to the results reported by the lep heavy flavour working group@xcite .
after the background subtraction a yield of @xmath113 leptons from @xmath36 semileptonic decays was found . from the ratio @xmath114 the @xmath95
was obtained to be : @xmath11 where the dominant systematics came from the lepton identification and the @xmath36 decay models .
figure [ fig : lept ] shows the momentum and transvers momentum spectra for the selected leptons .
figure [ fig : spectra ] shows the comparison of the measured @xmath115 lepton spectrum with the predictions from the accm model with different values of the parameters used to fit delco and mark iii data@xcite .
5.truein 2.5truein -.2 cm 5.truein 4.5truein -.2 cm
i am grateful to enrico piotto , franco simonetto and jong yi for providing me very useful informations . | the inclusive direct and cascade semileptonic branching ratios of the @xmath0 quark were measured from a fit to the lepton momentum spectrum in the @xmath1-hadron rest frame , exploiting the charge correlation between the lepton and the heavy meson : @xmath2 the value of the @xmath3 element in the ckm quark mixing matrix was computed from the differential decay width of the process @xmath4 , measured as a function of the product @xmath5 of the @xmath1 and @xmath6 4-velocities , using an inclusive method of @xmath6 reconstruction : @xmath7 a measurement of the ratio @xmath8 has been obtained by using the reconstructed secondary hadronic mass in semileptonic decays of @xmath1 hadrons , and found to be : @xmath9 the charm semileptonic branching ratio was measured using a double tag method based on the detection of exclusively reconstructed @xmath10 mesons accompanied by a lepton in the opposite hemisphere , giving : @xmath11 |
the graphene is a single - layer honeycomb lattice of carbon atoms coated on the surface of some materials.@xcite the dirac cone structure in the energy spectrum is responsible for some of the unusual properties of the system.@xcite study of the behaviors of electrons in graphene is one of the currently focused areas in the condensed - matter physics . in the theoretical investigations ,
most of the calculations are based on the continuous model with simplified dirac cone dispersion,@xcite and the coulomb interactions between electrons are treated in the random - phase approximation ( rpa ) . since the interactions are correctly taken into account in the long wavelength limit , this approach should reasonably describe the low energy behaviors of the electrons within the validity of rpa .
in such a model , however , the lattice - structure effect and the short - range part of the coulomb interaction have been completely neglected .
thus , it is desirable to explore this problem by a more realistic approach including the effect of the graphene lattice and a self - consistent scheme beyond the rpa . in this paper
, we use the tight - binding model defined on the two - dimensional honeycomb lattice to formulate the green function theory of electrons in graphene . in the self - energy of electron green function ,
the coulomb interactions are taken into account with the renormalized - ring - diagram approximation ( rrda ) .
this approximation is well known to satisfy the microscopic conservation laws.@xcite our recent investigation of the two - dimensional electron system@xcite shows that the rrda can accurately reproduce the result of the fixed - node - diffusion monte carlo simulation for the ground - state energy.@xcite it is therefore expected that the rrda could give more reliable description for the behaviors of electrons in graphene .
for readers convenience , we here briefly review the structure of a honeycomb lattice and its reciprocal lattice.@xcite for the sake of numerical computation , we will also present the mapping between coordinates defined on the basic vectors of the honeycomb lattice and the orthogonal coordinates .
the graphene lattice is of the honeycomb structure as shown in fig .
a set of basic displacement vectors of the lattice is @xmath0 where @xmath1 is the lattice constant .
we will chose @xmath1 as the unit of length , and thereby set @xmath2 hereafter .
the area of the unit cell is @xmath3 in unit of @xmath4 .
the whole lattice can be viewed as a quadrilateral lattice consisting of the unit diamond cells .
there are two sites in each unit cell : black and green . with @xmath5 and @xmath6
, we then define the unit vectors of the reciprocal lattice shown in fig .
they are given by @xmath7 where * z * is the unit vector in the direction of @xmath8 .
the basic displacement vectors of the reciprocal lattice are @xmath9 and @xmath10 . the first brillouin zone ( bz )
is the hexagon .
the diamond enclosed by the red dashed lines in fig .
[ fig2 ] is an equivalent first bz that is a convenient choice for numerical calculation . for the use of numerical calculation
, we here write down the transform between the coordinates on the basis of \{@xmath11 } and the orthogonal axes in real space . consider a vector @xmath12 in the representation of \{@xmath11}. we denote this vector in the orthogonal coordinate system as @xmath13 .
then the correspondence between these two sets of coordinates is given by @xmath14 the matrix @xmath15 of the transform @xmath16 is therefore given by @xmath17 where the first and second columns are the coordinates of vectors @xmath5 and @xmath6 respectively .
analogously , in the momentum space , we obtain the transform between the coordinates of a vector @xmath18 defined in the orthogonal system and its projections @xmath19 on the basis \{@xmath20}. denoting the transform as @xmath21 , we have @xmath22 where the first and second columns are respectively the vectors @xmath23 and @xmath24 .
the two matrices @xmath15 and @xmath25 are related by @xmath26 with @xmath27 the transpose of @xmath15 . for the later use
, we here discuss the fourier transform .
the function @xmath28 defined on the honeycomb lattice sites can be expanded as @xmath29 where the @xmath18-integral is over the first bz with @xmath30 its area , the components of @xmath31 and @xmath18 are given in the orthogonal coordinate system , and @xmath32 is the fourier component of the function @xmath28 .
( a function and its fourier component are distinguished by their arguments in this paper .
we also adopt the convention that a capital vector implies its components defined in the orthogonal coordinate system , while a lowercase vector means that its components are given in the basis \{@xmath33 } in real space or \{@xmath34 } in the momentum space ) . in the basis \{@xmath11 } , the function @xmath28 is given by @xmath35 . from eq .
( [ f ] ) , we get the expansion for the function @xmath36 , @xmath37 therefore , the fourier component of @xmath38 is given by @xmath39 .
this relationship is useful for the fourier transform of the coulomb interaction .
for describing the electron system , we use the tight - binding model in which the nearest - neighbor hopping and the coulomb interaction are taken into account .
firstly , we consider the hopping term , @xmath40 where @xmath41 is the hopping parameter , @xmath42 means the nearest - neighbors , @xmath43 ( @xmath44 ) is the creation ( annihilation ) operator of electrons at site @xmath45 with spin @xmath46 . with the basis of \{@xmath11 } , we hereafter designate the coordinates of an unit cell as that of the black site at the left lower corner of the cell as shown in fig .
the position of a site can then be denoted as @xmath47 where @xmath48 implies @xmath48-th unit cell and @xmath49 corresponds to black ( green ) site .
the electron operator , for example the annihilation one , should be then denoted as @xmath50 .
expanding the operator with the plane waves , we have @xmath51 with @xmath52 as the number of total unit cells of the lattice , and @xmath53 as the position vector of the @xmath48-th unit cell .
under such a convention , we can easily rewrite @xmath54 in momentum space .
the result is @xmath55 here the electron operators are given by spinors , @xmath56 with the first and second components denoting electrons respectively at the black and green sublattices , @xmath57 with @xmath58 s as the pauli matrices , and @xmath59 with @xmath60 and @xmath61 as the components of the electron momentum @xmath62 in the basis of \{@xmath20}. the special feature of graphene is in the energy dispersion .
the eigenvalues of @xmath63 are @xmath64 with @xmath65 in fig .
[ fig3 ] , we show the energy dispersion of upper band . clearly , @xmath66 depends on @xmath62 linearly only when @xmath62 closes to @xmath67 where it vanishes . in the continuous model ,
the energy dispersion is approximated by simple cones . .
[ fig3 ] we next consider the coulomb interaction .
the interaction @xmath68 between two electrons respectively at positions @xmath69 and @xmath70 is given by @xmath71 where @xmath72 denotes the position of the @xmath45 ( @xmath48)-th unit cell , @xmath73 is the static dielectric constant due to the screening by the electrons of carbon core and the substrate , and @xmath74 is the vector from the black site to green site in an unit cell as shown in fig .
1 . the on - site interaction @xmath75 is the coulomb repulsion between electrons of anti - parallel spins , leading to the short - range antiferromagnetic correlations ( afc ) .
since the afc is not significant in graphene , @xmath75 should not be too large . in our calculation , we set @xmath76 which is double of the nearest - neighbor interaction .
for the long - range coulomb interacting system , the final result should not sensitively depend on such a small but reasonable @xmath75 . by taking into account only the charge fluctuations ( with the spin fluctuations neglected ) , the interaction term of the hamiltonian in momentum space
is given by @xmath77 where @xmath78 is the electron density operator , and @xmath79 is the fourier component of the coulomb interaction .
since the total charge of the electrons is neutralized by the background of positive charges , the @xmath80 term is excluded from the summation .
the elements of @xmath79 are given in appendix a.
the green function of electrons is defined as @xmath81 where @xmath82 is the imaginary time , and @xmath83 means the statistical average of the @xmath82-ordered product of the operators . in the frequency space , @xmath84 is given by @xmath85^{-1 } , \label{g}\ ] ] where @xmath86 with @xmath87 the chemical potential , @xmath88 is the fermionic matsubara frequency , and @xmath89 is the self - energy . for brevity ,
we hereafter will use @xmath90 for the arguments unless stated otherwise . under the rrda which
is diagrammatically shown in fig .
[ fig4 ] , the elements of @xmath91 are given by @xmath92 where @xmath93 with @xmath94 as the temperature ( in unit of @xmath95 ) of the system , @xmath96 is the element of the matrix of the screened interaction @xmath97^{-1}\hat v(q),\ ] ] and the element of the polarizability @xmath98 is given by @xmath99 with @xmath100 and @xmath101 is the bosonic matsubara frequency .
the chemical potential @xmath87 is determined by the electron density ( that is the electron number per site ) , @xmath102 from equations ( [ g])-([mu ] ) , the green functions can be self - consistently determined . using the pad
approximation,@xcite we obtain the retarded self - energy @xmath103 and then the green function @xmath104 under the analytical continuation @xmath105 . in order to do an effective numerical calculation , it is necessary to get a clear understanding of the symmetry of the green functions , and this is shown in appendix b. on the other hand , we need to pay attention to the behavior of the screened interaction @xmath106 .
since it approaches to the bear coulomb interaction @xmath107 in the limit @xmath108 , @xmath106 can be separated into two parts , @xmath109 , where @xmath110 is the induced interaction from the ring diagrams . the contribution of @xmath107 in the self - energy yields the exchange part . here ,
an important point is that the behavior of @xmath110 in the limit @xmath111 by rrda is very different from that by rpa . for the sake of illustration
, we use the descriptions for @xmath107 and @xmath98 by their pauli components . in the limit @xmath111
, we have @xmath112 where @xmath113 . for @xmath110 in the same limit , we get @xmath114 with @xmath115 $ ] . in rpa , @xmath116 vanishes for @xmath117 .
however , in rrda , @xmath118 , it means the dynamic screening effect in the long - wavelength interactions .
as is well known , the ring diagrams in the rpa include the contribution of plasmon excitations .
the plasmon frequency @xmath119 is determined by the long - wavelength behavior of the charge polarizabilty @xmath120_{rpa } \propto q^2 $ ] at @xmath111 .
this polarizability is calculated in the absence of interactions . under the rrda ,
the corresponding polarization diagram of @xmath121 is calculated with renormalized green functions , and it does not have such a behavior as in rpa .
the renormalized - ring - diagram summation does not result in the desired plasmon excitations .
the correct way to obtain the plasmon excitations is to calculate the two - particle green function in which the vortex corrections need to be considered . under the rrda that is a conserving approximation for the single - particle green function ,
the kernel of the equation for the two - particle propagator is generated from the functional derivative of the self - energy diagrams with respect to the green function.@xcite the calculation of the two - particle propagator needs a more complicated mathematical procedure and is beyond the scope of the present approach .
under the present approximation , the screening effect is negligible only at sufficient large matsubara frequencies .
this requires a considerable amount of numerical calculations . to save the computer time , the summations over the matsubara frequencies in eqs .
( [ se ] ) and ( [ chi ] ) need to be performed with special method .
we have developed a super - high - efficiency algorithm for the series summations.@xcite in the present calculation , we have used the parameters @xmath122 = [ 2,15,5]$ ] for selection of the matsubara frequencies distributed in @xmath123 successively connected blocks each of them containing @xmath124 frequencies with @xmath125 the integer - parameter that the stride in the @xmath126-th block is @xmath127 .
the total number of the frequencies selected here is @xmath128 .
the largest number @xmath129 is about @xmath130 for the cutoff frequencies @xmath131 and @xmath132 . for the lowest temperature considered here , @xmath133
, we have @xmath134 . at the frequencies larger than the cutoff
, @xmath110 is negligible small . for illustrating the numerical method , an example for calculation of the element @xmath135 is given in appendix c. on the other hand , the momentum integrals in eqs .
( [ se ] ) and ( [ chi ] ) are convolutions . these integrals can be efficiently carried out by fourier transforms . here
again , we should pay attention to @xmath110 .
it should be carefully transformed from @xmath136-space to @xmath137-space since it has a sharp peak at @xmath80 as indicated by eq .
( [ walv ] ) .
this long - wavelength singular part should be subtracted from @xmath110 and can be treated especially .
it saves computer time to perform the transform of this singular part in the orthogonal coordinate system because where it is isotropic and the @xmath138-integral within a circle close to the origin can be reduced to a one - dimensional one . only within this circle ,
we need a very fine mesh for the integral . in the rest part of the hexagon brillouin zone
, one can use a crude mesh doing the two - dimensional integral because where the integrand is not singular .
the remained part of @xmath110 should be a regular function except there may be some undulations close to @xmath80 due to the subtraction .
another point we should take care about is that the green function varies drastically around and close to the fermi surface at low temperatures .
since the band structure is not flat at the fermi energy , the mesh for @xmath136-space integral should be fine enough around and close to the fermi surface .
we here give an example for sampling the points in momentum space .
this example is for the zero - doping . under the same consideration
, the sampling for finite doping cases can be planed similarly .
we divide the range @xmath139 $ ] in each axis in momentum space into four blocks shown as in fig .
there are @xmath140 equal meshes in the @xmath45-th block and @xmath140 s are given by @xmath141 the finest mesh is for the third block @xmath142 $ ] which is centered with @xmath143 . in the brillouin zone , this leads to a very fine mesh around the dirac points @xmath67 .
the second finest mesh is for the first block , which is designed for dealing with the undulations of the remained part of @xmath110 ( after the subtraction of the sharp peak ) when it is fourier transformed from @xmath136-space to @xmath137-space .
the rest two blocks have relatively crude meshes because the green function is smooth there .
our numerical algorithm considerably saves the computer memory and time . the similar method and its accuracy
have been demonstrated by a recent study on the two - dimensional electron system with infinite band width and long - range coulomb interaction.@xcite with our numerical algorithm , we have solved the above equations by iteration .
the system is characterized by the coupling constant @xmath144 that is defined by the ratio between the overall - average interaction energy @xmath145 and the hopping energy @xmath41 , @xmath146 the parameters @xmath41 = 2.82 ev and @xmath1 = 2.4 are known from the experimental observations.@xcite by choosing @xmath147 , we have @xmath148 0.5 .
therefore , the graphene is a moderately - coupled coulomb system .
firstly , we present the result for the spectral density that is defined by @xmath149 where @xmath150 . in the non - interacting case , @xmath151 reduces to the @xmath152-functions representing the energy dispersions of the two bands . in the present case ,
the energy levels are broadened because of the many - body effect . shown on the left panel in fig .
[ fms ] is an intensity map of the spectral density in the energy - momentum plane at the doping concentration @xmath153 = 0.02 and temperature @xmath154 = 0.02 .
this map exhibits the energy distribution of the states as a function of momentum along the high symmetry directions @xmath155-m - k-@xmath155 in the brillouin zone . for comparison ,
the free - particle energy dispersion @xmath156 ( with @xmath157 the chemical potential of the non - interacting system ) is also depicted as the solid curve . clearly , because of the coulomb effect , the energy distribution has finite width , implying the finite life times of quasiparticle .
in addition , the scale of the energy band is enlarged .
the dashed curve is a rescale of the solid one . with comparing to the non - interacting dispersion
, the energy band is magnified with a factor about 1.1 .
actually , the exchange self - energy results in an additional hopping of the electrons , which renormalizes the kinetic energy.@xcite to see this , we express the exchange self - energy in real space , @xmath158 where @xmath159 is the coulomb interaction between electrons at sites @xmath45 and @xmath48 . [ here @xmath45 and @xmath48 mean the positions of the lattice sites . ] with the quantity @xmath160 , one can define the additional hopping parameter , @xmath161 .
the most sizable additional hopping is the one between the nearest - neighbor ( nn ) . at @xmath162 = 0.02 , and @xmath144 = 0.5 , we find @xmath163 , and the magnitude of the next nn one is two order less than @xmath164 . at large distance
, @xmath165 is very small . as a total effect , such the additional hopping terms
enlarge the original energy band . on the right panel of fig .
[ fms ] , we show the map of the spectral density obtained by integration over the energy window of @xmath166 around the chemical potential .
the orbits of strong intensity correspond to the fermi surfaces which are apparent not circles as compared with those from the simplified dirac cone model .
the structure of the fermi surfaces is symmetrical under any rotation of angle @xmath167 around the origin .
all these results are comparable with the arpes experimental observations.@xcite for @xmath162=0.0 , we expect that the fermi surfaces in fig .
[ fms ] shrink into dirac points .
the broadening of energy distribution of quasiparticle is described by the imaginary part of the self - energy . to be specific ,
we analyze the green function @xmath168 .
this function can be divided into two parts , @xmath169/2 $ ] with @xmath170[\epsilon_{\vec k,21}+\sigma^r_{21}(\vec k , e)\}^{1/2 } \nonumber\end{aligned}\ ] ] where @xmath171 , @xmath172 , and @xmath173 s are the components of the retarded self - energy matrix .
the two green functions @xmath174 and @xmath175 can be considered as for the electrons at the upper band and lower band , respectively .
corresponding to @xmath176 , we define the retarded self - energies for the electrons respectively at upper and lower bands as @xmath177 in fig .
[ se02 ] , the imaginary parts of the self - energies @xmath178 are presented as functions of @xmath179 at a fermi momentum @xmath180 for the system at doping concentration @xmath162 = 0.02 and the temperature @xmath154 = 0.02 . at small energy
, @xmath181 is a quadratic function of @xmath179 .
the value at @xmath179 = 0 is very small and should vanish at zero temperature . at a region of larger energy
, @xmath181 seems to be linearly depended on @xmath179 .
the magnitude of @xmath182 is , however , small compared with @xmath181 .
the reason is clear . at the fermi surface ,
the states of energy @xmath179 is far from the corresponding states of the same momentum @xmath62 at the lower band .
therefore , the self - energy @xmath183 is small . in fig .
[ se02 ] , the rpa result for the self - energy of upper band is also shown for comparison . the magnitude of @xmath184 by rpa is larger than that by the rrda .
but , the quadratic @xmath179-dependence of @xmath184 at small @xmath179 is also reflected by the rpa calculation . in fig . [ se0 ] ,
the results for @xmath184 are depicted at low temperatures and at zero doping concentration . since
each fermi surface is a dirac point in this case , the fermi momentum for fig .
[ se0 ] is chosen as @xmath185 .
firstly , at the dirac point , because the state is the common state of the upper and lower bands , the self - energies of both bands coincide . at small energy ,
apart from a small value at @xmath179 = 0 due to the finite temperature effect , all the results for @xmath184 show a quadratic dependence on @xmath179 .
out of the small energy region , @xmath184 linearly depends on @xmath179 to certain limit . for the purpose of comparison ,
the rpa results obtained from our approach at the zero doping concentration are exhibited in fig .
[ rpase0 ] . in the limit @xmath186
, @xmath187 appears to be linearly depended on @xmath188 .
this is in agreement with the analytical result obtained from the continuous model based on the rpa.@xcite therefore , the system at zero doping is referred @xcite to behave like a marginal fermi liquid@xcite with the imaginary part of the self - energy going linear in energy near the chemical potential . however
, our numerical results based on the rrda demonstrate that the imaginary parts of the self - energies fixed at the fermi momentum are always varying as quadratic in energy close to the fermi level , regardless the system is doped or not doped .
this feature indicates that the quasiparticle in graphene behave like a moderately correlated fermi liquid . with the retarded green function
, we can also calculate the density of states ( dos ) defined as @xmath189 shown in fig .
[ dos ] is the result for @xmath190 at zero doping concentration and @xmath154 = 0.02 .
the non - interacting counterpart @xmath191 is also depicted for comparison .
since the energy linearly depends on the magnitude of the momentum near the dirac points , @xmath191 is proportional to @xmath179 at small @xmath179 .
the two peaks come from the van hove singularity because the energy bands are flat at @xmath192 as shown in fig .
[ fig3 ] . under the coulomb interaction , the spectral density of quasiparticle
is broadened .
this results in lowering the density of states and smearing the peaks . with comparing to @xmath191 , the two peaks in @xmath190 shift to larger energy because of the energy band enlarged by the exchange interaction .
we next consider the distribution function @xmath193 .
for the graphene , @xmath193 should be defined as @xmath194 as we have encountered in eq .
( [ spect ] ) , @xmath193 is determined only by @xmath195 .
further more , since the green function @xmath195 can be divided into parts of the upper and lower bands , we therefore define the distribution functions for these two bands as @xmath196 the total distribution function is given by @xmath197 . in fig .
[ dist1 ] , we exhibit the result of the distribution functions @xmath198 at temperature @xmath154 = 0.02 and doping concentration @xmath162 = 0.02 . in the inner fermi area ,
the upper band is almost fully occupied with the electrons .
the distribution @xmath199 drops drastically at the fermi surface .
the occupation at the lower band is not flat .
close to each dirac point ( the corner of the hexagon brillouin zone ) , the behavior of @xmath200 looks like a sink .
the depression of @xmath200 comes from two aspects .
one is the temperature effect .
the distribution @xmath200 reaches its minimum at the dirac points because where the energy ( dirac energy ) is the highest for states in the lower band .
since the doping concentration is small , the dirac energy is close to the fermi level .
therefore , @xmath200 has an apparent drop at the dirac points .
another reason is due to the many - body effect .
because of the coulomb interaction , an amount of electrons can be redistributed from the lower band to upper band . at zero doping , each fermi surface shrinks to a point .
the distribution functions are shown in fig .
[ dist2 ] .
the distribution @xmath201 at upper band concentrates at the dirac points . at lower band
, there is a obvious depression of @xmath200 at the point .
to see the many - body and temperature effects in the zero - doping case , we show the total distribution @xmath193 at and close to the dirac points as functions of temperature .
very close to the dirac points , @xmath193 increases drastically with decreasing temperature . at zero temperature
, the total distribution at the dirac points @xmath202 can be expected .
this is very different from the non - interacting distribution .
the latter is constant 1 because the zero - energy levels of both bands are half occupied and at the momentum different from the dirac points only the lower band is fully occupied . under the coulomb interactions ,
some of the electrons around each dirac point in the lower band are gathered to the upper band close to the dirac point , resulting in a higher distribution at and close to the point . from the increasing tendencies of @xmath203 and @xmath204
, we can infer there is an abrupt drop in @xmath193 close to each dirac point .
therefore , the zero - doping distribution function at zero temperature is consistent with the fermi liquid behavior . for the negative doping ,
the fermi surface opens again . in this case , the fermi level is at the lower band .
since the upper and lower bands are symmetric about the zero energy , at zero temperature , the electron distribution at the upper band corresponds to the hole distribution at the lower band , and vice versa .
therefore , the distributions at negative doping can be obtained from that of positive doping .
for example , at @xmath162 = -0.02 , by flipping fig .
[ dist1 ] upside down , we obtain the image of distributions .
the shape of each fermi surface is the same as that at @xmath162 = 0.02 .
however , the fermi area is outside of the surface where the lower band is near fully occupied .
finally , we give the ground - state energy per electron @xmath205 . at low temperatures , @xmath206 ,
the numerical results for the energy per electron are almost a constant . by extrapolation ,
we then obtain @xmath205 .
the results for the two doping cases are , @xmath207 for @xmath208 , and @xmath209 for @xmath210 .
in summary , using the tight - binding model with long - range coulomb interactions defined in a honeycomb lattice , we have presented the green - function formulation for the electron in graphene .
the interactions between electrons are treated with the renormalized - ring - diagram approximation .
the integral equations for determining the green function are solved self - consistently using our high - efficiency numerical algorithm .
the obtained spectral densities are comparable with the experimental observations .
since the imaginary part of the self - energy of the electron green function fixed at the fermi momentum varies as quadratic in energy near the chemical potential for both doped and undoped systems , we conclude that electrons in graphene follow the fermi liquid like behavior .
in addition , we also calculated the density of states and the distribution functions . in this appendix
, we present the fourier transform of the coulomb interaction between electrons on the honeycomb lattice .
because it is long - range interaction , the accurate result can not be obtained from the direct summation by definition with numerical calculation .
we must seek a fast converging scheme for summation over the lattice sites . to do this , we separate the interaction @xmath211 into short - range and long - range parts , @xmath212 with @xmath213 where @xmath1 is a free parameter of positive quantity , and @xmath214 . at @xmath215 , @xmath216 , therefore it is a short - range interaction .
its fourier transform can be obtained by the direct summation , @xmath217 on the other hand , @xmath218 is long ranged .
direct summation for the fourier transform of it converges very slow .
however , this function can be expressed as @xmath219 where the @xmath18-integral is over the whole momentum space , and @xmath220 the integral in eq .
( [ vl ] ) can be written in a summation over the integrals each of them over a brillouin zone .
shifting all these brillouin zones to the first brillouin zone by the corresponding reciprocal lattice vectors @xmath221 s , we have @xmath222 the order of integral and summation can be changed .
express @xmath31 as @xmath223 where @xmath224 is the position of @xmath48-th unit cell of the honeycomb lattice , and @xmath225 is a vector within the unit cell .
we get @xmath226 where use of @xmath227 has been made . from this equation
, we recognize the fourier transform of the function @xmath228 defined on the honeycomb lattice with @xmath225 a parameter , @xmath229 where the factor @xmath230 comes from the ratio between the area of the first brillouin zone of the honeycomb lattice and that of the square lattice @xmath231 .
since @xmath232 decreases exponentially at large @xmath233 , the @xmath234-summation converges very fast .
we are now ready to express the elements of the interaction matrix @xmath235 . with the above functions , we have @xmath236 where @xmath237 maps the vector @xmath19 under the basis @xmath238 into the one in the orthogonal coordinate system . for doing numerical calculation , it is necessary to understand the symmetry of the green function .
we start the discussion with the definition of the green function in real space , @xmath239 which describes a particle of spin-@xmath46 propagating from position @xmath70 to @xmath69 .
firstly , this function has the following property , @xmath240 because that the configuration for a particle propagating from @xmath70 to @xmath69 is the same as in the inverse process .
therefore , the diagonal green functions are even under @xmath241 . denoting @xmath242
, we obtain @xmath243 on the other hand , for the off - diagonal part , we get @xmath244 though the off - diagonal green functions have no definite parity , they can be separated into even and odd functions . for @xmath245 , we have @xmath246 where @xmath247 and @xmath248 are even and odd functions of @xmath249 , respectively .
now , we can express the green function matrix in terms of pauli matrix , @xmath250 where the pauli components as functions of @xmath249 have their definite parities .
obviously , in momentum space , as functions of @xmath62 , they have the same parities . using the property of parity ,
the fourier expansions of these components are given by @xmath251 where the @xmath62-integrals are over the half brillouin zone : @xmath252 and @xmath253 .
further more , in the half brillouin zone , we can separate the green functions @xmath254 s into even and odd functions of @xmath61 , @xmath255 with the superscripts @xmath256 denoting the parities , @xmath257/2.\ ] ] correspondingly , in real space , @xmath258 s are separated into even and odd functions of @xmath259 .
we have @xmath260 where the @xmath62-integrals are now over the first quadrant of the first brillouin zone : @xmath252 and @xmath261 .
we next consider the property of the green functions under the exchange of the coordinates @xmath262 . since the system is symmetric under this exchange , we have @xmath263 from the above definitions
, one can obtain that the pauli components @xmath264 are symmetric about the exchange but @xmath265 for the readers convenience , we here give an example for calculation of @xmath267 .
for the briefness , we present only the result for @xmath268 .
the results for other elements and the method for calculation of the self - energy elements @xmath269 can be obtained immediately with the same consideration .
as has been mentioned in the main text , the momentum convolution of two green functions in @xmath270 can be evaluated by fourier transform . in real space , it is given by @xmath271 where the @xmath234-summation for the matsubara frequencies is over @xmath272 .
using the property of the green function , @xmath273 , we take the summation only for @xmath274 , @xmath275}_{n=1}g_0(\vec r , i\omega_m - i\omega_n)g_0(\vec r ,-
i\omega_n ) \nonumber\\ ~~+ |g_0(\vec r , i\omega_{\bar n})|^2/2|^{\bar n = ( m+1)/2}_{\rm if~m~is~odd}\}.\nonumber\end{aligned}\ ] ] where @xmath276 $ ] is the integer part of @xmath277 . applying our rule for the series summation
, we get @xmath278}_pg_0(\vec r , i\omega_m - i\omega_{p})g_0(\vec r ,- i\omega_{p } ) \nonumber\\ ~~+ |g_0(\vec r , i\omega_{\bar n})|^2/2|^{\bar n = ( m+1)/2}_{\rm if~m~is~odd}\}\nonumber\\ ~~+ \delta\chi_0(\vec r , i\omega_m)\nonumber\end{aligned}\ ] ] where the summation is over the selected points @xmath279 with @xmath280 and @xmath281}_p$ ] the corresponding weights,@xcite and the last term is the contribution from terms beyond the cutoff @xmath129 for the matsubara frequency , @xmath282 ~~~~{\rm if~m = 0}}\end{cases}.\ ] ] the present expression for @xmath283 is essentially the same as that in appendix b of ref . except an additional factor 2 stemming from the degree of spin freedom in the present case and a misprinting in ref . .
the notation of taking real part of the result should be assigned in the expression of @xmath284 in appendix b of ref . .. | using the tight - binding model with long - range coulomb interactions between electrons , we study some of the electronic properties of graphene . the coulomb interactions are treated with the renormalized - ring - diagram approximation . by self - consistently solving the integral equations for the green function , we calculate the spectral density .
the obtained result is in agreement with experimental observation .
in addition , we also compute the density of states , the distribution functions , and the ground - state energy . within the present approximation , we find that the imaginary part of the self - energy fixed at the fermi momentum varies as quadratic in energy close to the chemical potential , regardless the system is doped or not .
this result appears to indicate that the electrons in graphene always behave like a moderately correlated fermi liquid . |
titanium dioxide ( tio@xmath6 ) has a history of applications since the early 19th century . although its initial usage was confined to use as a white pigment , its applications diversified over the course of the last few decades , including but not limited to biological implants , photocatalysis and dye - sensitized solar cells ( dsscs ) , based on the refractive , biocompatible , photocatalytic and photoelectric properties of the material @xcite .
recent emphasis on the latter applications spurred research work in the surface properties and reactions of crystalline tio@xmath6 , understanding of which is crucial in their realization .
in the case of dsscs , this often involves the interactions of these surfaces with sensitizing dyes , such as triscarboxy - ruthenium terpyridine [ ru(4,4,4-(cooh)@xmath7-terpy)(ncs)@xmath7 ] , the `` black dye '' .
recent developments have also been made in bio - sensitized solar cells ( bsscs)@xcite , which are similar to dsscs in operating principles , but with the synthetic dye often used in the latter replaced by one that is biologically derived .
one class of such biomolecules are the cyclic tetrapyrroles , which encompass porphyrins and chlorins .
many of such compounds contain a ring structure , with one or more carboxylic functional groups which play importance roles in anchoring the dye to the tio@xmath6 surfaces , benzoic acid s adsorption on tio@xmath6 can thus offer an exemplar case study for how this entire class of compounds can interact with tio@xmath6 surfaces . out of all the tio@xmath6 surfaces ,
the rutile @xmath0 is the most energetically stable and naturally occurring , and thus forms the substrate surface of choice in such adsorption studies .
the rutile @xmath0 surface , though not prone to reconstruction , can undergo @xmath1 reconstruction upon annealing under uhv conditions , with the models proposed including added ti@xmath6o@xmath7 and ti@xmath7o@xmath9 rows .
as such , the focus of our studies shall be on the adsorption of benzoic acid on these two types of rutile @xmath0 surfaces , in terms of the binding mode and the overlayer patterns .
experimental studies have been performed on such adsorptions , using leed , esdiad and stm by guo et al @xcite , and later by grinter et al using stm @xcite . in guo
s studies , the optimal adsorption mode observed was in line with those of other small carboxylates i.e. dissociative bidentate bridging binding ( bb ) , involving the two carboxyl o atoms , after deprotonation on one of them , binding to two neighbouring five - coordinated ti sites ( ti@xmath2 ) along the @xmath10 $ ] rows of such sites . in terms of adsorption patterns , through leed@xcite , it was also observed that a p@xmath11 pattern was formed .
when studied using stm however , a different picture emerged , as a p@xmath12 pattern was observed instead .
such discrepancy could be accounted for by the fact that leed is a scattering technique dependent upon the charge distribution of the surface , while stm surveys the apex of the surface and the adsorbates , and the fact that the phenyl rings of the benzoates were rotated to align with the @xmath13 $ ] direction to form dimer structures .
the rotation of the phenyl rings and the resultant dimer structures are due to the interactions of the lower phenyl h atoms and the @xmath14 orbital of the neighbouring benzoate , facilitated by the bridging o@xmath15 anions . building upon these findings ,
a later stm study of these adsorptions was performed@xcite , with the aims of studying orientations of the benzene rings of the adsorbate overlayer , as well as the adsorption of benzoates on the @xmath0-@xmath1 reconstructed surface .
the binding geometry of the acid was once again confirmed , but the elongation of the phenyl rings along the @xmath16 $ ] direction was only significant in the case of the reconstructed surface , due to improved hydrogen bonding of the phenyl h atoms to the surface o atoms as a result of closer proximity .
there was also no formation of dimer structures at high coverages as reported in the much earlier study - which could be due to the higher dosing temperature used this time , nor was there any observation of adsorption of dissociated h@xmath17 ions . on the @xmath0-@xmath1 reconstructed surface
, it was also observed that benzoic acid adsorbs and binds the same way as on the unreconstructed surface .
although dft studies of the adsorption of benzoic acid on the rutile @xmath0 surface had been performed by troisi et al @xcite , these were performed with the aim of comparing the effects of using different dft implementations on the computed results on electronic structures of such benzoate - tio@xmath6 complexes , rather than the physical structure of the adsorbates on the surface which forms the main theme of experimental research work . in our work
, we will expand upon this topic by investigating the proposed models for surface adsorbate structures and patterns described in the experimental work , using dft as implemented through vasp . after presenting our methodology
, we calculate and compare the energetics of these configurations obtained using lda , gga and vdw , as well as producing high resolution simulated stm images to provide comparisons with these results , before concluding .
our dft investigations were carried out using the vienna ab initio simulations package ( vasp)@xcite@xcite . out of all the elemental pseudopotential files , oxygen has the highest @xmath18 value at 395.700 ev , hence this value was chosen as the cut - off energy @xmath18 in all our simulations .
iterative electronic relaxations are done such that energy differences ( @xmath19 ) between two successive steps should not exceed @xmath20 ev , while relaxations are said to be achieved when the rms force on each atom falls below @xmath21 ev / . in our calculations , lda , gga(pbe ) and
gga+dft - d2-based methods were used to give three different sets of results for comparison of the effects of hydrogen bonding and van der waals forces , as gga and dft - d2 methods can account for these forces much better , as the interactions as a result of these forces were implied in the experimental stm studies @xcite . in order to simulate the rutile tio@xmath6 substrate , an 8 ti layer slab terminated by two @xmath0 surfaces on each end was chosen , as studies of the rutile @xmath0 surface through dft have revealed that surface energies converge at around this thickness @xcite .
this was done after performing bulk relaxation of the rutile crystal , where k - sampling values of @xmath22 were used , in order to reflect the approximate inverse ratio of the dimensions of the real space unit cell .
the optimized lattice constants obtained through lda calculations were found to be @xmath23 and @xmath24 for rutile , reflecting a contractions of @xmath21 to @xmath25 from the experimental values , which are expected for dft using lda as reported in other research literature , as well as being within @xmath26 error margin @xcite . as for gga based calculations ,
the optimized lattice constants were @xmath27 and @xmath28 , representing an expansion of 0.04 , which is expected due to gga s underestimation of cohesive energies of insulators .
the new lattice parameters were then used to set up the slabs for surface relaxations .
8 ti layer slabs were set up using four such cells with the terminated ends autocompensated . for the @xmath1-reconstructed surface , we used the ti@xmath7o@xmath9 added - row structure as assumed in @xcite .
we then set the benzoates up such that they correspond to the following six configurations , which involve the phenyl rings being aligned either with the @xmath10 $ ] direction(denoted @xmath29 ) or the @xmath16 $ ] direction ( denoted @xmath30 ) , as well as being aligned such that they form @xmath4-periodic patterns ( denoted @xmath31 ) , or with the neighbouring @xmath10$]-rows of benzoates aligned one ti@xmath2 step out with respect to each other ( denoted @xmath32 ) .
the different monolayer patterns are denoted as follows , with the relaxed gga structures shown in the relevant figures ( lda and gga relaxed structures show little different ) : * bb@xmath33 : phenyl rings of the benzoates are aligned with the @xmath10 $ ] direction , with the benzoates of neighbouring @xmath10 $ ] rows in step with each other , as seen in figure [ paraingga ] * bb@xmath34 : phenyl rings of the benzoates are aligned with the @xmath10 $ ] direction , with the benzoates of neighbouring @xmath10 $ ] one ti@xmath2 site out of step with each other , as seen in figure [ perpingga ] * bb@xmath35 : phenyl rings of the benzoates are perpendicular to the @xmath10 $ ] direction , with the benzoates of neighbouring @xmath10 $ ] rows in step with each other , as seen in figure [ paraoutgga ] * bb@xmath36 : phenyl rings of the benzoates are perpendicular to the @xmath10 $ ] direction , with the benzoates of neighbouring @xmath10 $ ] one ti@xmath2 site out of step with each other , as seen in figure [ perpoutgga ] * bb@xmath37 : phenyl rings of the benzoates are aligned with the @xmath10 $ ] direction , using the @xmath1-reconstructed @xmath0 surface instead , as seen in figure [ pararecongga ] * bb@xmath38 : phenyl rings of the benzoates are perpendicular to the @xmath10 $ ] direction , using the @xmath1-reconstructed @xmath0 surface instead , as seen in figure [ perprecongga ] these collections of atoms are padded by 15 of vacuum between the topmost and the bottom - most atoms of the vertically adjacent simulation cells , in order to ensure lack of interaction between the atoms of these cells . to calculate the adsorption energies ( @xmath39 ) ,
the energies of three simulation cells of the above defined dimensions , consisting of the slab and the adsorbate ( @xmath40 ) , just the slab ( @xmath41 ) and just the adsorbate ( @xmath42 ) in neutral form respectively , were first calculated , obtaining @xmath39 by : @xmath43 as the latest stm studies @xcite did not report h adsorption on the protruding bridging o@xmath15s , these simulations were also run with the dissociated h removed , with stabilities of each mode compared in terms of the total energy of the simulation cell @xmath44 .
in addition , the ionic adsorption energies of @xmath45 the different modes will be compared against one another , in a similar fashion to , with @xmath46 being replaced by that corresponding to the energy of a single benzoate anion in the simulation cell .
this gives us completeness in the comparisons for adsorption energies . upon relaxations of these different modes of adsorptions under the above described set ups , simulated stm images
were then generated based on the electronic structures of the relaxed adsorbate - surface complex , at a bias voltage of + 1.5v .
for the adsorption structures , the optimized structures for the six bb modes corresponded largely with the models proposed for the stm images @xcite , in that the carboxylate group remained anchored via two ti@xmath2-o@xmath47 bonds , while the benzene rings remained aligned more or less along the @xmath10 $ ] and the @xmath16 $ ] directions .
slight deviations were observed in the bb@xmath48 and the bb@xmath37 modes , in that the benzene rings were not exactly aligned along the @xmath10 $ ] directions and instead being slightly rotated about the @xmath49 $ ] axis .
in addition to the slight rotations , in all of the bb@xmath29 modes , the benzene rings underwent slight planar rotations about the @xmath16 $ ] axis .
these are likely due to the sideways repulsions between the h atoms on the benzene rings . in terms of adsorption energetics , the adsorption energies ( @xmath39 )
were obtained ( with the values obtained through gga calculations ) as displayed in table [ dissads ] . when the dissociated hs were removed , ionic adsorption energies ( @xmath45 )
were obtained and presented in table [ ionads ] .
.adsorption energies for dissociative adsorptions of benzoic acid on the rutile @xmath0 surface , in ev / , as calculated through lda , with and without the co - adsorbed h. [ dissads ] [ cols="<,<,<,<",options="header " , ] in the experimental stm studies , the stm images produced presented the sites of benzoate adsorption as bright spots , elongated along the directions of the alignments of the phenyl rings@xcite , with no direct visual evidence of co - adsorbed hydrogens , except for low coverages on the @xmath1-reconstructed surface .
the simulated stm images produced have replicated the same arrays of bright spots representing the two - dimensional periodities of the benzoate adsorbates ( see figures [ bbipstm ] and [ bbopstm ] ) , while the rutile @xmath0 surface itself has been shown to be dark in comparison to the adsorption sites . in the limit representing high charge isosurfaces i.e. large currents , the positions and the shapes of the bright regions representing the benzoate adsorption sites were mainly determined by the positioning of the carboxylate groups , rather than the orientations of the benzene rings .
the latter only started to influence the appearances of bright regions when the charge isosurfaces are set at values sufficiently low ( i.e. limits of low currents ) , such that the rutile @xmath0 substrate surface itself loses its distinctive visual features , such as distinct dark regions as shown in figure [ bbipstm](top ) .
where the benzene rings were imaged in the limit of low currents , they appeared either as double rows of trios of bright spots , with the brightest at the centre representing the topmost c atom of the ring in the case of the benzene ring being aligned with the @xmath10 $ ] direction ; or as twin central bright spots accompanied by darker , kidney - shaped spots on either side .
it is also observed while in the limit of high currents where the carboxylate groups were distinctively imaged , positions of the co - adsorbed h can be seen from the differing orbital shapes in the @xmath10 $ ] o@xmath15 rows , with circular - shaped dim spots representing the 1s orbitals of h and dumbell - shaped ones representing the 2p orbitals of the o@xmath15 atoms . in the bb@xmath50 modes ,
whose simulated stm images are as shown in figure [ bbipstm ] , revealed orientations of rectangular rows of bright spots , with their widths aligned along the planes of the benzene rings determined through dft calculations .
the patterns of the bright spots representing the benzene rings however differed between the bb@xmath51 and the bb@xmath52 modes , with the former as parallel trios of bright spots along the @xmath10$]-direction , and the latter as two central bright spots sandwiched by two larger merged dimmer spots aligned along the @xmath16$]-direction . in the bb@xmath53 modes , the simulated stm images generated are as shown in figure [ bbopstm ] .
the bright spots representing the @xmath12 overlayer symmetry can be observed in both cases , in agreement with stm results obtained from experimental studies , where such features were observed in some regions alongside the bright dots representing the @xmath4 symmetry on the unrecontructed surface .
the orientations of the bright spots again reflected the orientations of the benzene rings as shown in figures [ paraoutgga ] and [ perpoutgga ] . however , whereas the benzene ring was imaged as duo rows of bright spots in the bb@xmath51 mode , in the bb@xmath52 mode the darker spots around the central bright spots merged , forming kidney - shaped spots .
for the case of bb@xmath54 modes , in both cases the reconstructed ti@xmath6o@xmath7 ridge appeared as large bright bands along the @xmath10 $ ] directions , with features showing the ridges being largely indistinguishable .
again , although the double rows of three bright spots were observed when the ring was aligned along the @xmath10 $ ] direction , the dimmer pairs of spots on the sides of the central bright spots became merged in the case of the benzene ring being aligned with the @xmath16 $ ] direction .
for adsorption energetics , gga and gga+dftd2 simulations have shown the bb@xmath51 to be the most energetically stable out of all the bb modes of adsorptions on the unreconstructed surface , when co - adsorbed with the dissociated hydrogens .
being the most observed pattern in the experimental stm studies of the unreconstructed surface , this is an expected result . in terms of the simulations themselves , this shows that when hydrogen bonding and/or van der waals forces are taken into account , the bb@xmath51 mode is the most stable .
examining the three pairs of bb@xmath29 and bb@xmath30 modes of adsorption , in gga and gga+dftd2-based calculations , it is clear that on the non - reconstructed surface , simulations have shown that for both @xmath1- and @xmath12-symmetric patterns , the bb@xmath29 modes were shown to be more energetically stable in both cases .
these observations , in both the experimental and computational dft studies , can be accounted for by the long distances between the o@xmath15 atoms and the nearest phenyl h atoms when the benzene ring is rotated , making hydrogen bonding between the two too weak to be a significant stablizing factor . in terms of the @xmath12-symmetric patterns on the rutile @xmath0 surface ,
the bb@xmath55 modes were shown to be more energetically stable through gga and gga+dftd2 calculations .
this again is in line with the experimental observations , which revealed alignment of the benzene rings with the @xmath10 $ ] direction .
the proposed reasons for such observations were again that the nearest distances between the o@xmath15 atoms and the phenyl h atoms being too far apart to make hydrogen bonding a significant stablizing force . in comparison the bb@xmath51 mode , the corresponding bb@xmath55 mode appeared less energetically stable ( see figures [ dissads ] and [ ionads ] ) .
this again agrees with the findings of experimental stm studies , where such patterns were comparatively rare relative to the @xmath11-symmetric patterns . in the case of the @xmath1-reconstructed rutile @xmath0 surface ,
however , the results of the dft calculations contradict the conclusions reached in the stm studies . in both the cases of gga and gga+dftd2 calculations
, it was the bb@xmath56 configuration that was more energetically stable in all the three pairs of simulations , instead of the @xmath16$]-aligned benzene rings as observed in the experimental stm images .
the likely explanation for this were the asymmetric distortions in the ti@xmath6o@xmath7 reconstruction ridges produced in the optimized structures , with the predicted energetic stablization effects from hydrogen bonding between the o@xmath15 atoms on the reconstructed ridges and the nearest phenyl h being less than predicted .
perhaps the single greatest anomaly in the results lies in the abnormally large ionic adsorption energy values for benzoates on the @xmath57-reconstructed surface - see figure [ dftd2ads ] .
we found that , for the ionic adsorption on the reconstructed surface , the energies appeared to be anomalous .
we include the values for completeness , but are unable to explain their large values , which we will continue to investigate . in terms of the simulated stm images , the images generated revealed the bright spots corresponding to sites of benzoate adsorption , with the orientations of the rings much more clearly delineated than the experimental stm images do .
the arrays of positions of the bright spots also corresponded to the 2d symmetries of the adsorption monolayers in the bb@xmath50 , bb@xmath53 and bb@xmath54 configurations .
for both the bb@xmath50 and the bb@xmath53 modes , in the limit of sufficiently low valued charge isosurfaces such that the benzene rings start to be imaged distinctively , the co - adsorbed h atoms lose their distinguishability from their surroundings , whereas in the limit of isosurfaces of high charge values , the vacant o@xmath15 sites could be distinguished from the ones bound by dissociated h by the 2p orbitals imaged .
the simulated stm images in the low charge value limits reveal features which are consistent with those observed in the experimental stm images produced of benzoates on the unreconstructed surface , most notably the elongations of the bright regions along the directions of alignments of the benzene rings , as well as the absence of features reflecting the co - adsorption of dissociated h. this suggests that perhaps even if stm images do not directly image the co - adsorbed h ions , it does not mean that such co - adsorption did not take place . on the reconstructed surfaces , the simulated stm images revealed central bright bands representing the ti@xmath6o@xmath7 reconstructed rows similar to those observed in the experimental stm images , and that no distinguishable features representing co - adsorption of dissociated h ions .
this is however still in agreement with experimental results as co - adsorbed h ions were not imaged at high coverages .
again , the lack of direct visual evidence for co - adsorption of dissociated protons does not necessarily imply complete absence of it .
we have studied the different patterns for the bidentate bridging dissociative adsorption of benzoic acid on both the unreconstructed and @xmath1-reconstructed surfaces through dft methods lda , gga and gga+dftd2 , as implemented through vasp , following both @xmath11 and @xmath12 symmetries . in all the three dft methods studied , both with and without the dissociated h , the bb@xmath51 pattern was invariably found to be the most energetically stable out all the six bb patterns of adsorption .
these results support stm studies @xcite which revealed that on the unreconstructed @xmath0 surface , where the prevalent pattern was that of bb@xmath50 , with the orientation of the benzene rings aligned along the @xmath10 $ ] direction . in the case of @xmath1-reconstructed surface however , the bb@xmath56 mode was found to be more stable , in contrast to the experimental stm studies results@xcite where the benzene rings were observed to have rotated by 90@xmath58 , with hydrogen bonding between the phenyl hydrogens and the o@xmath15 atoms along the reconstructed ridges being the reason for the stablization of this orientation . in the case of simulated stm images ,
the same overlayer patterns of bright dots was reproduced for the proposed models , elongations of the bright spots representing the alignments of benzene rings were observed , which were not obvious in the case of experimental stm studies .
these alignments were only deduced by line profiles taken along the @xmath10 $ ] and the @xmath49 $ ] directions in the experimental stm studies . in the different limits of high and low - valued charge isosurfaces ,
the simulated stm images have also revealed the carboxylate and the benzene ring groups respectively , while in experimental stm images , these structures were not clearly distinguished in the bright spots representing the sites of benzoate adsorption .
we thank dr geoff thorton from the chemistry department of university college london for providing the model for the rutile @xmath0-@xmath1-reconstructed surface , the research computing group of the internet services division of university college london , and jakub vrtny , conn orourke and umberto terranova for valuable discussions .
this research work is funded by uckg limited , china .
999 diebold , u. _ surf .
_ , * 2003 * , _ 48 _ , 53 - 229 wang , z - s , yamaguchi , t. ; sugihara , h. ; arakawa , h. _ langmuir _ , * 2005 * , _ 21 _ , 4272 - 4276 thavasi , v. ; lazarova , t. ; filipek , s. ; kolinski , m. ; querol , e. ; kumar , a. ; ramakrishna , s. ; padrs , e. ; renugopalakrishnan , v. _ j. nanosci .
_ , * 2008 * , _ 8 _ , 1 - 9 fujishima , a. ; honda , k. nature 1972 , _ 238 _ , 37 - 38 fujishima , a. ; zhang , x. ; tryk , d. _ surf .
rep . _ * 2008 * md .
k. nazeeruddin . ; p. pechy . ; m. grtzel .
_ , , 1705 - 1706 lundqvist , m. j. ; nilsing , m. ; lunell , s. _ j. phys . chem .
b _ , * 2006 * , _ 110 _ , 20513 - 20525 perron , h. ; domain , c. ; roques , j. ; drot , r. ; simoni , e. ; catalette , h. _ theor . chem . acc .
_ , * 2007 * , _ 117 _ , 565 - 574 onishi , h. ; iwasawa , y. _ surf .
_ , * 1994 * , _ 313 _ , l783-l789 fukui , k - i . ; onishi , h. ; iwasawa , y. _ surf .
_ , * 1994 * , _ 313 _ , l783-l789 fukui , k - i . ; onishi , h. ; iwasawa , y. _ chem .
_ , * 1997 * , _ 280 _ , 296 - 301 wendt , s. ; schaub , r. ; matthiesen , j. ; vestergaard , e. k. ; wahlstrm , m. d. ; rasmussen , p. ; thostrup , p. ; molina , l. m. ; lgsgaard , e. ; stensgaard , i. ; hammer , b. ; besenbacher . _ surf .
_ , * 2005 * , _ 598 _ , 226 - 245 guo , q. ; cocks , i. ; williams , e. m. _ surf .
sci . _ * 1997 * , _ 393 _ 1 11 .
grinter , d. c. ; nickels , p. ; woolcot , t. ; basahel , s. n. ; obaid , a. y. ; al - ghamdi , a. a. ; el - mossalamy , e - s . h. ; alyoubi , o. a. ; thornton , g. _ j. phys . chem .
c _ * 2012 * , _ 116 _ , 10201026 martsinovich , n. ; jones , d. r. ; troisi , a. _ j. phys . chem .
c _ * 2010 * , _ 114 _ , 22659 - 22670 _ phys .
b _ , * 1996 * , _ 54 _ , 11169 _ j. comput . chem .
_ , * 2008 * , _ 29 _ , 20442078 kckell , p. ; terakura , k. _ surf .
_ , * 2000 * , _ 461 _ , 191 - 198 kckell , p. ; terakura , k. _ app . surf .
_ , * 2000 * , _ 166 _ , 370 - 375 fukui , k - i . ; onishi , h. ; iwasawa , y. _ appl .
sci . _ * 1999 * , _ 140 _ , 259 - 264 sasahara , a. ; uetsuka , h. ; onishi , h. _ appl .
a _ , * 2001 * , _ 72 [ suppl . ] _ , s101-s103 | adsorption of benzoic acid on the @xmath0 surface of rutile , both unreconstructed and @xmath1-reconstructed ones , at saturation coverage of one benzoate per two adjacent five - coordinated ti@xmath2 sites has been studied using dft simulations , as implemented through the vienna ab initio simulations package ( vasp ) . in order to study the effects of hydrogen bonding and van der waals forces in influencing the relative stabilities of different adsorbate overlayer superstructures
, these studies were performed through local density approximation ( lda ) , generalized gradient approximation(gga ) and dft - d2 . through the calculations ,
it was found out that although the optimized structures corresponded with the proposed models for the experimental results , the relative energetic stabilities of different overlayer structures have shown some differences with the experimental results . in the gga calculations ,
the overlayer structures involving the benzene rings aligned along the @xmath3-direction were shown to be more stable than those aligned perpendicular to it , regardless of whether the benzoates were arranged in @xmath4 or @xmath1 symmetries on the @xmath4-unreconstructed surface , or on the @xmath1-reconstructed surface .
these are in contrast to the experimental studies using stm , whose observations revealed that configurations with the benzene rings aligned along the @xmath5-direction when adsorbed on the @xmath1-reconstructed surface . |
hyperon polarization in different reactions have attracted much attention recently ( see , e.g. [ 1 - 15 ] , and the references given there ) .
this is , on the one hand , because the polarization can easily be determined in experiments by measuring the angular distribution of their decay products thus provide us with a powerful tool to study the spin effects in different cases . on the other hand , it is also triggered by the surprisingly large transverse hyperon polarization discovered already in 1970s in unpolarized hadron - hadron and hadron - nucleus reactions@xcite .
understanding the origin(s ) of such puzzling transverse hyperon polarizations has been considered as a challenge to the theoretician working in this field and should provide us with useful information on spin structure of baryon and spin dependence of strong interaction . in a recent letter@xcite , triggered by the similarities of the corresponding data@xcite
, we pointed out that the transverse polarization observed@xcite in unpolarized hadron - hadron collisions should be closely related to the left - right asymmetries observed@xcite in singly polarized @xmath5 collisions .
they should have the same origin(s ) . by using the single - spin left - right asymmetries for inclusive @xmath6 production as input
, we can naturally understand the transverse polarization for hyperon which has one valence quark in common with the projectile , such as @xmath7 , @xmath8 or @xmath9 in @xmath5-collisions , or @xmath10 in @xmath11-collisions . to further understand the puzzling transverse polarization of @xmath10 in @xmath5-collisions , which has two valence - quarks in common with the projectile ,
we need to assume that the @xmath12 and @xmath13 , which combine respectively with the valence-@xmath14-diquark and the remaining @xmath15-valence quark to form the produced @xmath10 and the associatively produced @xmath16 in the fragmentation region , should have opposite spins . under this assumption
, we obtained a good quantitative fit to the @xmath17-dependence of @xmath10 polarization in @xmath5 collisions ( where @xmath18 , @xmath19 is the longitudinal component of the momentum of the produced hyperon , and @xmath20 is the total center of mass energy of the @xmath5 system . ) the obtained qualitative features for the polarizations of other hyperons are all in good agreement with the available data .
we further pointed out , in a recent brief report@xcite , that @xmath10 polarization in the exclusive process @xmath21 can be used to give a special test of the picture since @xmath22 in this channel , which is the simplest one for @xmath23 , should take the maximum among all the different channels .
it is encouraging to see that also this result is in agreement with the data obtained by r608 collaboration at cern@xcite , which shows that @xmath22 in this channel is indeed much larger than that for @xmath24 .
it should be interesting to see whether the assumption that @xmath12 and @xmath25 have opposite spins can also be applied to other reactions . in this connection , it is encouraging that the measurements on @xmath10 polarization in @xmath1 is being carried out at jefferson laboratory@xcite . in this paper
, we would like to apply the picture in [ [ letter97 ] ] to @xmath1 to give the prediction on @xmath10 polarization in reactions with longitudinally polarized electron beams .
we compare the results with those obtained in other cases for the spin states of the @xmath12 and @xmath25 needed for the production of @xmath10 and @xmath16 in this process .
these will be given in section 2 .
after that , in section 3 , we extend the calculations to other similar processes such as @xmath26 or @xmath4 and present the predictions on the polarizations of the hyperons ( or neutron ) in these reactions .
we will also discuss the influence from the sea and that due to the mass differences between the @xmath27 and @xmath28 baryons in section 4 .
we will summarize the results in section 5 .
we found out that these polarizations are very sensitive to the spin states of the @xmath12 and @xmath13 thus provide us with an ideal tool to study the spin correlations between them .
we now consider the process @xmath1 , and discuss the problem in the center of mass frame of the produced hadronic system .
we suppose that the energy and momentum transfer in the reaction are high enough so that the parton picture can be used .
we thus envisaged with the following picture . in the collision process
, a @xmath29 quark is knocked out by the virtual photon .
the knocked @xmath29 combines with a @xmath30 to form the observed @xmath16 and the remaining @xmath31-diquark combines with the @xmath32 to form the observed @xmath10 .
this process is illustrated pictorially in fig.1 .
since we are now considering the longitudinal polarizations of hyperons in @xmath0 , two points of the picture in [ [ letter97 ] ] apply : ( i ) the @xmath33 and @xmath34 that combine with the diquark from the incoming proton to form the hyperon and the remaining quark to form the associatively produced meson should have opposite spins .
( ii ) the su(6 ) wave function can be used to relate the spins of the quarks to that of the hadrons . using these two points ,
we obtain immediately the following : because @xmath16 is a spin zero object , the @xmath30 has to have the spin opposite to the scattered @xmath29-quark . hence , according to ( i ) , the @xmath32 quark should be polarized in the same direction as the scattered @xmath29-quark . according to the su(6 )
wave function , which states that @xmath35 , [ where the subscripts denote the spin and its third component of the @xmath14-diquark ] , the @xmath14-diquark has to be in the spin zero state and the @xmath10 spin just equals to that of the @xmath12 quark . since the momentum of @xmath10 is opposite to that of @xmath16 or that of the scattered @xmath29-quark , the longitudinal polarization of @xmath10 is opposite to that of the scattered @xmath29-quark in the helicity basis .
hence , we obtain , @xmath36 here , @xmath37 is the polarization of the incoming electron beam .
we use the superscript ( a ) to denote that the result is obtained in the case that the spins of @xmath32 and @xmath30 are opposite for the convenience of comparison with other cases which will be discussed in the following .
@xmath38 is the spin transfer factor from @xmath39 to @xmath40 in the scattering process @xmath41 .
this is an electromagnetic process thus @xmath38 can easily be calculated and can be found in different publications .
it is a function of only one variable @xmath42 , which is defined as @xmath43 , where @xmath44 , @xmath45 and @xmath40 are the four momenta for the incoming proton , @xmath39 and the virtual photon exchanged in the scattering .
@xmath38 is given as , @xmath46 it is a monotonically increasing function of @xmath42 which increases from 0 to 1 when @xmath42 goes from 0 to 1 . since @xmath42 is related to @xmath47 and @xmath48 by @xmath49 , ( where @xmath47 is the bjorken-@xmath50 , @xmath51
, @xmath52 is the total @xmath53 center of mass energy squared ) , it is clear that we can also express @xmath54 for @xmath10 in @xmath1 as a function of @xmath47 and @xmath48 at given @xmath20 .
eq.(1 ) is the result obtained based on the picture in ref.[2 ] under the assumption that the spins of @xmath32 and @xmath30 are opposite to each other . to compare
, we now consider the other two cases for the spins of @xmath32 and @xmath30 , i.e. , ( b ) the spins of @xmath32 and @xmath30 are parallel to each other , and ( c ) the spins of @xmath32 and @xmath30 are completely uncorrelated , i.e. , they have the same probability to be parallel or anti - parallel . apparently , in case ( b )
, we have , @xmath55 and in case ( c ) , @xmath56 we see that , the results are very much different from each other . hence ,
measuring @xmath57 in this process can distinguish these three different cases with good efficiency .
we can extend the calculations to other similar processes such as @xmath60 , @xmath61 , or @xmath4 . as we can see from fig.1 , since in all these processes ,
the produced meson @xmath62 or @xmath6 is a spin zero object , we can obtain the same relation between the polarization of the @xmath33 shown in the figure and that of the scattered @xmath63 from the incoming proton as we obtained in @xmath1 .
however , in contrast to the case for @xmath10 production where only spin zero @xmath14-diquark contributes and if the @xmath14-diquark is in the spin zero state , the produced @xmath64 can only be a @xmath10 , in reactions such as @xmath26 , the diquarks in different spin states can contribute and given the spin state of the diquark @xmath65 and that of the quark @xmath33 , the produced @xmath66 can still be a @xmath27 or a @xmath28 baryon .
all these make the calculations more complicated . to obtain the polarization of the produced hyperon
, we need to know the probabilities for the diquark which participates in the reaction to be in different spin states and the relative probabilities for the the produced @xmath66 to evolve into a @xmath27 or a @xmath28 baryon .
the former can be calculated using the su(6 ) wave - function of the proton . to do this , we note that , due to the helicity conservation , the scattered quark has the same longitudinal polarization before and after the scattering . given the polarization of the scattered quark and the wave function of the proton , we can calculate the probabilities for the remaining diquark in different spin states .
the latter can be determined by the projection of the spin wave function for the produced @xmath66 to that of the corresponding baryon , if we neglect the influence from the mass difference of the corresponding @xmath67 and @xmath68 baryons . to show how these calculations are carried out , we now take @xmath60 as an explicit example . in this reaction
, the scattered quark has to be a @xmath69 quark .
we recall that the su(6 ) wave function for proton is , @xmath70,\ ] ] where we use the up or down arrow to denote the helicity `` @xmath71 '' or `` @xmath72 '' state .
we now consider the case where the scattered @xmath73 quark is in the helicity @xmath71 state .
we see that if the incoming proton is in the helicity `` @xmath71 '' state , the remaining @xmath74-diquark should be in the @xmath75 state , and the relative weight is @xmath76 .
the incoming proton can also be in the helicity `` @xmath72 '' state .
the corresponding wave function can be obtained by reversing the spins of the quarks in eq.(5 ) . in this case , since the scattered @xmath73 is in the helicity `` @xmath71 '' state , the remaining @xmath74-diquark should be in the @xmath77 state , and the relative weight is @xmath78 .
hence , we obtain that , if the the scattered @xmath73 quark is in the helicity `` @xmath71 '' state , the @xmath74-diquark can be in @xmath79 and @xmath80 state , and the relative probabilities are @xmath76 and @xmath78 respectively . next , we consider that the produced @xmath74-diquark combine together with a @xmath32 quark to form a @xmath81 or a @xmath82 .
we first consider the case ( a ) that @xmath32 and @xmath83 have opposite spins .
since the @xmath32 should be polarized in the same direction as the scattered @xmath73 and moves in the opposite direction of the scattered @xmath73 , it should be in the helicity `` @xmath72 '' state .
we consider the case that the @xmath74-diquark is in the state @xmath75 .
the spin state of the produced @xmath84 is given by @xmath85 .
the projections of this state @xmath85 to the su(6 ) wave function of @xmath58 and that of @xmath86 are @xmath87 and @xmath88 respectively .
we thus obtain the relative probabilities for the production of these two hyperons in this case are @xmath76 and @xmath78 respectively .
the results in other cases can be obtained in the same way . the results in case ( c ) are shown in tables 1 - 3 for the different reactions .
those in case ( a ) and ( b ) can also be read out from the tables by looking only at the results for @xmath89 or @xmath90 respectively . from tables 1 - 3 , we can calculate the hyperon polarizations in different cases for the spin states of @xmath32 and @xmath30 .
the results are shown in table 4 .
we see that the produced @xmath91 , @xmath81 and @xmath59 are all polarized significantly .
we also see that , similar to @xmath10 in @xmath92 , the polarizations for @xmath59 in @xmath93 obtained in the three cases differ quite significantly from each other .
but the differences between the results for @xmath91 , @xmath81 are not so significant .
hence , to distinguish between the three cases , we should study @xmath54 and @xmath94 in the corresponding reactions .
we note that although the polarization of @xmath91 can not be measured but it implies a polarization of @xmath10 , @xmath95 in @xmath96 which can be measured experimentally .
we should emphasize that the results in table 4 are obtained under the following two assumptions and/or approximations : ( 1 ) only valence quarks are taken into account in the scattering of @xmath39 with @xmath44 while the sea contribution is neglected ; ( 2 ) the influence of the mass differences between the @xmath27 and the @xmath28 baryons on the relative production rates are neglected . in practice ,
both of them should be taken into account .
we now discuss their influences on the polarization results obtained above .
it is clear that , in deeply inelastic @xmath53-scattering , the exchanged virtual photon @xmath97 can also be absorbed by a sea - quark or an anti - sea - quark .
however , since we consider only the exclusive reactions @xmath0 or the @xmath98 process @xmath99 , both of produced hadrons have very large momentum fractions in the center of mass frame of the hadronic system .
the sea contributions to such process should be very small . to see this , we take @xmath1 as an explicit example . besides the case shown in fig.1
, this process can also happen via the absorption of the @xmath97 by a @xmath30 from the sea of the proton .
after the absorption , the @xmath30 flies in the direction of the @xmath97 while the rest of the proton moves in the opposite direction .
this @xmath30 may pick up a valence @xmath29 from the proton and combine into a @xmath16 which flies in the direction of the @xmath30 while the rest of the proton combines together into a @xmath10 .
since the valence quark @xmath29 usually takes a relatively large fraction of the momentum of the proton , the probability for this to happen should be small in particular at high energies .
furthermore , the ratio of the probability for the virtual photon to be absorbed by @xmath30 to that by @xmath29 is given by @xmath100 $ ] .
at , e.g. , the cebaf energies , it is only of the order of @xmath101 .
hence we expect that the sea contribution should be negligible .
in contrast , there should be some influences from the mass differences of the @xmath27 and the @xmath28 baryons .
the relative production weights have to be influenced by such mass differences .
however , it is very interesting to note that in case ( a ) , there is completely no influence from the mass difference on the hyperon polarizations in all the reactions discussed above and the @xmath10 polarizations in all the three cases ( a ) , ( b ) and ( c ) are not influenced by the mass effect .
this is because in case ( a ) , only @xmath102 contributes to the production of @xmath58 or @xmath59 .
it always gives rise to @xmath103 , @xmath104 or @xmath105 , i.e. completely negatively polarized @xmath58 or @xmath59 .
the mass effect can affect the production rate , but does not influence the polarization in the corresponding exclusive reaction . for @xmath10 production
, only @xmath106 contributes to @xmath10 production and it contributes only to @xmath10 production , so the @xmath10 polarization in all cases are not influenced by such mass effect . in cases
( b ) and ( c ) , there are indeed some influences from the mass difference on the polarizations of @xmath107 , @xmath108 and @xmath59 . to see how large they may influence the polarization of the hyperons ,
we multiply the production rate by a corresponding exponential factor @xmath109 for the production of each baryon , where @xmath110 is the mass of the hyperon .
we assume that the production processes can be treated as two steps , i.e. , first the production of the corresponding quarks states then the evolution to the corresponding baryons . in this way
, we need to normalize the production weights of the produced baryons to the total production weight of the corresponding quark states .
we obtain that the above polarization should be modified into the following . in case
( b ) , we have @xmath111 @xmath112 @xmath113 similarly , in case ( c ) , we have @xmath114 @xmath115 @xmath116 to get a feeling of how strongly they depend on @xmath117 , we show the results in case ( b ) as functions of @xmath117 in fig.2 .
we see that there are indeed some influences on @xmath58 or @xmath59 polarization in the corresponding reactions but the influences are not very large .
the following point should be noted here .
all the polarizations obtained above are functions of only one variable @xmath42 .
they are independent of the energies under the condition that the energies are high enough that the parton picture can be used .
in summary , we have calculated the polarizations of the baryons in the exclusive processes @xmath1 , @xmath2 , @xmath3 , and @xmath118 with longitudinally polarized electron beams .
we used the picture proposed in [ [ letter97 ] ] , where it is assumed that the @xmath33 and @xmath34 needed in these processes to combine with the valence - diquark and struck quark to form the hyperons and the associated produced meson have the opposite spins .
the results show that all these baryons are longitudinally polarized and the polarizations are functions of only one variable @xmath42 .
we compared the results with those obtained in other possible cases for the spin states of @xmath33 and @xmath34 .
we found out that the magnitudes of hyperon polarizations are considerably large in all the different cases and that they are quite different from each other .
hence , they can be used as a good probe to study the spin correlations between the @xmath33 and @xmath34 in future experiments .
we thank li shi - yuan , xie qu - bing and other members of the theoretical particle physics group in shandong university for helpful discussions .
this work was supported in part by the national science foundation of china ( nsfc ) and the education ministry of china under huo ying - dong foundation . for a review of the data , see e.g. , a. bravar , in procs .
13th inter .
symp . on high energy spin physics , eds .
et al . _ ,
world scientific , singapore , 1999 .
liang zuo - tang and c. boros , phys .
. lett . * 79 * , 3609 ( 1997 ) .
[ letter97 ] for a dedicated review , see e.g. ,
liang zuo - tang and c. boros , inter . j. mod . a*15 * , 927 ( 2000 ) .
[ asy00 ] liang zuo - tang and c. boros , phys .
rev . d*61 * , 117503 ( 2000 ) .
r608 collab . ,
t. henkes _ et al .
b*283 * , 155 ( 1992 ) .
m. mestayer , private communication ( 2001 ) .
aleph collab .
, d. buskulic _ et al .
b*374 * , 319 ( 1996 ) ; opal collab .
k. ackerstaff _
j. c*2 * , 49 ( 1998 ) .
nomad collab . , p. astier _
et al . _ ,
b*588 * , 3 ( 2000 ) ; @xmath119 b*605 * , 3 ( 2001 ) . [ nomad ]
hermes collab . ,
a. airapetian _
rev . d*64 * , 112005 ( 2001 ) ; e665 collab .
et al . _ ,
j. c*17 * , 263 ( 2000 ) .
jaffe and x. ji , phys .
lett . * 67 * , 552 ( 1991 ) ; nucl .
b*375 * , 527 ( 1992 ) .
[ jaf ] c. boros and liang zuo - tang , phys .
d*57 * , 4491 ( 1998 ) .
a. kotzinian , a. bravar and d. von harrach , eur .
j. c*2 * , 329 ( 1998 ) .
m. anselmino , m. boglione , and f. murgia , phys .
b*362 * , 164 ( 1995 ) ; phys .
rev . d*60 * , 054027 ( 1999 ) ; m. anselmino , m. boglione , u. dalesio , and f. murgia , eur .
j. c*21 * , 501(2001 ) ; m. anselmino , d. boer , u. dalesio , and f. murgia , phys . rev . d*63 * , 054029 ( 2001 ) .
[ abm00 ] b.q .
ma , i. schmidt and j.j .
yang , phys .
d*61 * , 034017 ( 2000 ) ; b.q . ma ,
i. schmidt , j. soffer and j.j .
yang , phys .
d*62 * , 114009 ( 2000 ) ; phys . rev . d*64 * , 014017 ( 2001 ) , erratum - ibid , d*64 * , 099901 ( 2001 ) ; hep - ph/0107157 .
liu chun - xiu and liang zuo - tang , phys .
d*62 * , 094001 ( 2000 ) ; liu chun - xiu , xu qing - hua and liang zuo - tang , phys .
rev . d*64 * , 073004 ( 2001 ) ; xu qing - hua , liu chun - xiu and liang zuo - tang , phys .
d*63 * , 111301(r ) ( 2001 ) ; phys .
d ( 2002 ) in press .
[ xll01 ] 0.3cmtable 1 : possible states for the produced @xmath84 , their relative production weights , the possible corresponding products and their weights in the reaction @xmath153 in the case that the spins of @xmath32 and @xmath154 have same probability to be parallel or anti - parallel if the scattered @xmath73 is in the helicity `` @xmath71 '' state .
0.3cmtable 2 : possible states for the produced @xmath155 , their relative production weights , the possible corresponding products and their weights in the reaction @xmath156 @xmath16 in the case that the spins of @xmath32 and @xmath154 have same probability to be parallel or anti - parallel if the scattered @xmath29 is in the helicity `` @xmath71 '' state .
0.3cmtable 3 : possible states for the produced @xmath157 , their relative production weights , the possible corresponding products and their weights in the reaction @xmath158 in the case that the spins of @xmath159 and @xmath160 have the same probability to be parallel or anti - parallel if the scattered @xmath29 is in the helicity `` @xmath71 '' state .
0.3cmtable 4 : polarizations of @xmath10 , @xmath91 , @xmath81 , and @xmath59 in the reactions @xmath161 , @xmath162 and @xmath118 in the three different cases for the spin states of the @xmath33 and @xmath34 , i.e. , case ( a ) : the spins of @xmath33 and @xmath34 are anti - parallel , case ( b ) : the spins of @xmath33 and @xmath34 are parallel , case ( c ) : the spins of @xmath33 and @xmath34 have equal probability to be anti - parallel or parallel . | we apply the picture proposed in a recent letter for transverse hyperon polarization in unpolarized hadron - hadron collisions to the exclusive process @xmath0 such as @xmath1 , @xmath2 , or @xmath3 , or the similar process @xmath4 with longitudinally polarized electron beams .
we present the predictions for the longitudinal polarizations of the hyperons or neutron in these reactions , which can be used as further tests of the picture . |
it is well known that group theoretical properties and number of nambu - goldstone ( ng ) bosons are determined by a given coset space @xmath1 . if we extend it to a supersymmetric ( susy ) theory , the ng bosons are always accompanied by fermions .
it is extremely interesting to identify the fermions ( called as quasi ng fermions ) with observed quarks and leptons @xcite , since it may provide us not only an origin of families of quarks and leptons but also an answer to the fundamental question why we have three families in nature . the susy @xmath2 non - linear sigma ( nls ) model based on the coset space @xmath0 @xcite is fascinating , since it can accommodate three families of quarks and sleptons as its ng chiral multiplets .
this suggests that we may predict the maximal number of families in the approach of nls based on exceptional groups , since the exceptional group is limited up to @xmath3 .
in fact , it is shown that number of the families is limited to be three even if we take the biggest exceptional group @xmath3 @xcite . in this paper
we consider the @xmath0 model , where the unbroken @xmath4 is identified with the gauge group of the grand unified theory ( gut ) .
the masses of the squarks and sleptons are highly suppressed at the tree level , since the global @xmath2 is assumed to be exact in the limit where gauge couplings and yukawa couplings vanish .
their masses are dominantly generated via radiative corrections , leading to a natural solution to the flavor changing neutral current ( fcnc ) problem .
on the other hand , higgs multiplets , @xmath5 and @xmath6 , are considered to be non - ng multiplets , and hence , their masses are not suppressed .
the purpose of this paper is to examine if the above ng boson hypothesis of all the squarks and sleptons is consistent with observations . for this purpose , we choose one of the attractive and consistent mediation scenarios of susy breaking , that is called as pure gravity mediation ( pgm ) @xcite or minimal split susy @xcite . in the pgm , gaugino masses are generated radiatively at the one - loop level via anomaly mediation @xcite . since the gravitino mass is larger than @xmath7tev , the cosmological gravitino problem is avoided easily .
the gaugino masses are generated without a gauge singlet susy breaking ( polonyi ) field ; therefore , the cosmological polonyi problem does not exist . even without the gauge singlet susy breaking field , the higgsino mass term of the order of the gravitino mass arises via an @xmath8-symmetry breaking term , i.e. the constant term in the superpotential @xcite . in the pgm ,
the lightest neutralino is the pure wino of a mass range of @xmath9-@xmath10 gev , providing us with a good dark matter candidate .
furthermore , it is expected that the susy fcnc problem is significantly relaxed . in this paper , we point out that the @xmath0 nls model is consistent with pure gravity mediation . model . in this case , the higgsino - like neutralino rather than the stau can be the lightest susy particle , reducing the fine - tuning of the electroweak symmetry breaking . ]
the present model can be regarded as an ultraviolet completion of higgs - anomaly mediation proposed in ref .
furthermore , we also show the predicted squark masses of the first and second generations are likely to be smaller than a few tev when the thermal leptogenesis @xcite ( see also @xcite for reviews ) successfully explains the observed baryon asymmetry
. those squarks can be discovered at the lhc run-2 or at the high luminosity lhc .
this paper is organized as follows . in sec.2 , we explain our setup , the @xmath2 nls model combined with pgm . it is shown that the ng multiplets of the three chiral generations have vanishing masses at the tree level . in sec.3 , we investigate low - energy phenomenology of the model , paying attention to the the lightest squark and slepton masses .
section 4 is devoted to the conclusion .
the @xmath0 model is obtained via @xmath11 model , which contains @xmath12 ng modes .
the ng multiplets are @xcite @xmath13 where @xmath14 and @xmath15 .
note that @xmath16 and @xmath17 are identified with chiral multiplets of three family leptons and quarks , whose scalar components are massless at the tree level . here , we assume @xmath18 has a large dirac mass term with another @xmath19 , with one of higgs multiplets , @xmath5 , in this paper . ] which is required to cancel the non - linear sigma model and gauge anomalies @xcite . the higgs chiral multiplets , @xmath5 and @xmath6 ,
are introduced as non - ng matter multiplets .
then , their boson components are expected to have susy breaking soft masses of the order of the gravitino mass , @xmath20 .
the @xmath0 model is obtained through the further breaking @xmath21 down to @xmath22 , leading to six more ng bosons .
these ng bosons are identified with scalar partners of three right - handed neutrinos .
after all , the three new ng chiral multiplets of @xmath23 are identified with three chiral multiplets of the right - handed neutrinos .
the k " ahler potential for the ng multiplets is constructed from a real function transforming under @xmath2 as @xmath24 where @xmath25 is a holomorphic function of @xmath26 , and @xmath27 is invariant under transformations of the unbroken symmetry , @xmath28 .
the real function @xmath27 itself can not be @xmath2 invariant , since the shift of @xmath27 with the holomorphic function does not leave the lagrangian invariant in supergravity theories : we need an chiral superfield , @xmath29 , to cancel the shift @xcite .
then , we have the @xmath2 invariant k " ahler potential as @xmath30 with @xmath31 under @xmath2 transformation . [ [ soft - masses ] ] soft masses + + + + + + + + + + + to examine a soft susy breaking mass for @xmath26 , let us consider the leading term of @xmath32 as @xmath33 where @xmath34 denotes irrelevant terms such as higher order terms of @xmath32 .
the soft susy breaking mass of @xmath26 arises from @xmath35 \frac{|w|^2}{m_p^4 } \nonumber \\ & = & e^{k / m_p^2 } \left(\frac{\partial k}{\partial x}\right)^2 \left(\frac{\partial^2 k}{\partial x^2}\right)^{-1 } \frac{|w|^2}{m_p^4 } = g(x ) , \label{eq : gs_mass}\end{aligned}\ ] ] where @xmath36 ; @xmath37 is the superpotential with @xmath38 . here , @xmath39gev is the reduced planck mass .
then , @xmath40 which vanishes at the minimum .
we see all ng bosons are massless at the tree level .
the above argument does not depend on whether there is a direct coupling between a ( pseudo ) ng multiplet and a susy breaking field @xmath41 , since the potential is just replaced as @xmath42 ( see appendix for an explicit calculation ) . on the other hand , we assume that the soft susy breaking masses for the higgs doublets , which are non - ng multiplets , are given by : @xmath43 at the tree level , where @xmath44 is a constant free parameter .
the negativeness of @xmath45 and @xmath46 is important to solve the tachyonic slepton problem @xcite and give large masses for stops @xcite , enhancing the higgs boson mass @xcite .
in fact , the soft susy breaking masses for ng bosons arise radiatively via anomaly mediation and renormalization group ( rg ) running effects from @xmath45 and @xmath46 . at the high energy scale ,
say , the gut scale ( @xmath47 ) , the scalar mass of the ng multiplet is written as @xmath48 m_{3/2}^2 \ , , \label{eq : amsb_scalar}\end{aligned}\ ] ] where @xmath20 is a gravitino mass and @xmath49 is an anomalous dimension defined by @xmath50 with @xmath51 of a renormalization scale . for @xmath52 in eq.([eq : higgs_soft ] ) , the above relation holds at any scale and the sleptons are tachyonic below the gut scale due to the positive beta - functions for the @xmath53 and @xmath54 gauge couplings . in the next section , we show that this problem can be solved if the higgs soft mass squares are negative at @xmath47 @xcite .
since there is no polonyi field in our setup , the gaugino masses are purely generated from anomaly mediation : @xmath55 above a susy particle mass scale .
here , @xmath56 and @xmath57 are bino , wino and gluino mass , respectively ; @xmath58 , @xmath59 and @xmath60 are gauge coupling constants of @xmath53 , @xmath54 and @xmath61 . below the susy particle mass scale ,
the threshold corrections @xcite need to be included .
trilinear couplings are also generated from anomaly mediation and given by @xmath62 with @xmath63
we assume , in this subsection , that the gauge and yukawa interactions are only sources of the explicit breaking of the global @xmath2 , and hence all squarks and sleptons are massless at the tree level .
the masses of squark and sleptons arise only through anomaly mediation at the high energy ( gut ) scale .
thus , sleptons are tachyonic at @xmath47 as discussed in the previous section . however , radiative corrections from the higgs loops lift up the slepton mass for @xmath64 , solving the tachyonic slepton problem .
we estimate the positive contributions to sfermion masses from higgs loops by taking into account rg running effects .
the resultant mass spectrum is hierarchical among generations since yukawa couplings of higgs doublets are hierarchical @xcite , and hence we obtain relatively large masses for stops which explains easily the observed higgs boson mass of @xmath65gev .
since the sleptons and squarks are massless at the tree level , this model has four parameters : @xmath66 where @xmath67 is the ratio of the vacuum expectation values of @xmath5 to @xmath6 .
note that from the conditions of the correct electroweak symmetry breaking , the viable range of @xmath67 is fixed to be @xmath68 ( 60 ) for @xmath69 .
this is because one needs large yukawa couplings of the tau and bottom , @xmath70 and @xmath71 .
the necessity of the large @xmath67 is explained as follows . from the electroweak symmetry breaking ( ewsb ) conditions , we have @xmath72 where @xmath73 is a mass of the cp - odd higgs . here , we have used @xmath74 . in order to avoid a tachyonic cp - odd higgs , @xmath75
, @xmath76 should be satisfied at the electroweak scale .
this is achieved via the rg running of @xmath46 , @xmath77 for large @xmath71 and @xmath70 ( remember that @xmath46 is negative ) . for large @xmath68 ( or 60 for @xmath78 ) , the above contribution is larger than that for @xmath45 , @xmath79 then , @xmath76 ( @xmath80 ) is realized at the low - energy scale , which is required for the correct ewsb . and @xmath81gev .
, title="fig : " ] and @xmath81gev .
, title="fig : " ] the dependence of the required @xmath67 on @xmath82 originates from the radiative correction to @xmath71 from sbottom - gluino loops @xcite .
since the susy contribution to the muon @xmath83 is opposite to the needed one and the required @xmath67 for correct ewsb is so large in the case of @xmath84 , we focus on the region of @xmath85 in this paper , unless otherwise noted . in fact , we can not find particular differences in the susy - particle spectra between the regions with @xmath86 and sign@xmath87 , apart from the required @xmath67 and the contribution to the muon @xmath83 .
the susy spectrum is computed using suspect2.4.3 @xcite .
we scan all allowed parameter space of @xmath44 and @xmath67 , @xmath88 and take the maximum and minimum values of @xmath89 for each @xmath20 . here , @xmath90 and @xmath91 are the lightest chargino mass and neutralino mass , respectively , and they are nearly degenerated .
the input parameter @xmath44 is set at @xmath92gev ( @xmath93 ) .
we have demanded the lightest slepton ( squark ) to be heavier than 340 ( 1000 ) gev from the null results of the lhc susy searches @xcite .
the range of @xmath44 has been determined to avoid the tachyonic squarks and sleptons : squarks become tachyonic in the regions of @xmath94 .
note that on the viable parameter space , we find that all yukawa couplings at the gut scale are smaller than @xmath95 .
we now show the mass of the lightest chargino , which is almost pure wino , in fig .
[ fig:1 ] ( left ) .
we see the wino mass is almost insensitive to @xmath67 and @xmath44 .
this is because the threshold correction @xcite @xmath96 is suppressed by large @xmath67 and @xmath97 : in contrast to usual pgm models , the relation between wino mass and gravitino mass is almost fixed . since the higgsino is heavy , the leading dark matter candidate in this model is the wino - like neutralino
this is the reason why we restrict the range of @xmath98100 - 1000tev , ( approximately ) corresponding to the mass range for the wino dark matter , [email protected] tev .
this is followed by the phenomenological constraints assuming that the wino is the lightest susy particle ( lsp ) .
the lower - bound comes from the wino search at the lhc @xcite while the upper - bound is given by the condition to avoid the over - closure of the universe with the wino dark matter @xcite . on the right panel in fig .
[ fig:1 ] , we also show the maximum and minimum values of the gluino mass . the gluino is accessible at the lhc only when the gravitino mass is smaller than about 150tev . while the gray dots do not .
the black solid lines show the minimum values of @xmath91 , title="fig : " ] while the gray dots do not .
the black solid lines show the minimum values of @xmath91 , title="fig : " ] ( 300)tev . the higgs boson masses in unit of gev are shown as dashed lines . in the blue shaded region ,
the selectron is the lsp . in the green ( pink ) shaded regions , the lightest squark ( slepton ) is the right - handed down squark ( the left - handed selectron ) .
, title="fig : " ] ( 300)tev . the higgs boson masses in unit of gev are shown as dashed lines . in the blue shaded region ,
the selectron is the lsp . in the green ( pink ) shaded regions , the lightest squark ( slepton ) is the right - handed down squark ( the left - handed selectron ) .
, title="fig : " ] ( 300)tev . the higgs boson masses in unit of gev are shown as dashed lines . in the blue shaded region ,
the selectron is the lsp . in the green ( pink ) shaded regions , the lightest squark ( slepton ) is the right - handed down squark ( the left - handed selectron ) .
, title="fig : " ] ( 300)tev . the higgs boson masses in unit of gev are shown as dashed lines . in the blue shaded region ,
the selectron is the lsp . in the green ( pink ) shaded regions , the lightest squark ( slepton ) is the right - handed down squark ( the left - handed selectron ) .
, title="fig : " ] experiment at @xmath100 ( red ) , 1.5@xmath101 ( orange ) and 2@xmath101 ( yellow ) level . in the upper two panels
, the gravitino mass is taken as @xmath102 ( 140 ) tev .
we also show the contours of the lightest squark ( slepton ) mass in the left ( right ) panels , in unit of gev . in the blue shaded regions ,
the selectron is the lsp .
, title="fig : " ] experiment at @xmath100 ( red ) , 1.5@xmath101 ( orange ) and 2@xmath101 ( yellow ) level . in the upper two panels , the gravitino mass is taken as @xmath102 ( 140 ) tev .
we also show the contours of the lightest squark ( slepton ) mass in the left ( right ) panels , in unit of gev . in the blue shaded regions ,
the selectron is the lsp .
, title="fig : " ] experiment at @xmath100 ( red ) , 1.5@xmath101 ( orange ) and 2@xmath101 ( yellow ) level . in the upper two panels ,
the gravitino mass is taken as @xmath102 ( 140 ) tev .
we also show the contours of the lightest squark ( slepton ) mass in the left ( right ) panels , in unit of gev . in the blue shaded regions ,
the selectron is the lsp .
, title="fig : " ] experiment at @xmath100 ( red ) , 1.5@xmath101 ( orange ) and 2@xmath101 ( yellow ) level . in the upper two panels ,
the gravitino mass is taken as @xmath102 ( 140 ) tev .
we also show the contours of the lightest squark ( slepton ) mass in the left ( right ) panels , in unit of gev . in the blue shaded regions ,
the selectron is the lsp .
, title="fig : " ] next , we show the stop mass , defined by @xmath103 in fig .
[ fig : mstop ] to estimate regions consistent with the higgs boson mass of 125 gev . here ,
@xmath104 and @xmath105 are the masses of the left - handed and right - handed stop , respectively .
the scanned range of the parameter space is the same as in eq.([eq : scan_pars ] ) .
the stop mass is lifted up via rg running : @xmath106 therefore , the large @xmath107 leads to the larger stop masses . on almost all points ,
the stop mass is larger than 10tev , is as small as 0.04 and @xmath108100tev , the stop mass becomes @xmath109tev .
] and hence the higgs boson mass of 125gev is easily explained in most of the parameter space . in fig .
[ fig:2 ] , the scatter plots of the masses of the lightest squark and slepton are shown . here , the lightest squark is either the right - handed up squark or down squark , while the lightest slepton is the left - handed or right - handed selectron , depending on the parameter space .
the threshold corrections to the squark masses are included utilizing one - loop rg equations ( see e.g. @xcite ) to incorporate the large mass hierarchy between the third generation squarks and first / second generation squarks .
the dark - green and red dots satisfy the constraint , @xmath110gev , while the gray dots do not . here , @xmath111 is the higgs boson mass computed using susyhd 1.0.2 @xcite .
the black solid lines show the minimum values of @xmath91 in fig .
[ fig:1 ] .
thus , the points below the lines are excluded unless the @xmath8-parity is violated .
it can be seen that the lightest squark can be as light as 1 - 2tev even for @xmath112tev .
however , the stop mass is too large in such cases and @xmath111 becomes larger than 128gev : larger @xmath107 leads to smaller ( larger ) squark ( stop ) masses . considering the higgs boson mass constraint and the sizes of the squark mass and the lightest neutralino mass
, it is expected that the regions for @xmath113 - 350tev can be tested at the high - luminosity lhc @xcite .- parity is slightly violated . in this case , the viable region is much wider , since the sleptons and squarks can be lighter than @xmath114 .
the region with @xmath115tev may be tested at the lhc by searching an @xmath8-hadron .
notably , the region for even heavier gravitino may be also tested by searching a stable slepton in both lhc and ilc .
( the sleptons can be @xmath7 gev even for @xmath20 = 1000 tev . ) ] [ [ regions - favored - by - the - thermal - leptogenesis ] ] regions favored by the thermal leptogenesis + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + if the thermal leptogenesis is responsible for the observed baryon asymmetry , the mass of the wino - like neutralino needs to be smaller than about 1tev @xcite when it is the stable dm .
this is because the neutralino produced from the gravitino decay leads to the over - closure of the universe for @xmath116tev , if the reheating temperature is higher than @xmath117gev due to the large energy density of the gravitino .
note that the reheating temperature higher than about @xmath117gev is required for the successful thermal leptogenesis @xcite .
the critical value , @xmath118tev , corresponds to @xmath119tev ; therefore the regions with @xmath120tev or smaller are especially interesting . in fig .
[ fig : cont1 ] , we show the contours of the lightest squark mass ( left ) and slepton mass ( right ) as well as @xmath111 . on the upper ( lower ) two panels , we take @xmath121 ( 300)tev . in the gray regions , the lightest slepton is lighter than 340 gev or the lightest squark is lighter than 1tev ; hence , those regions are likely to be excluded . in the blue shaded region ,
the slepton is lighter than @xmath122 and the lsp . in the green ( pink )
shaded regions , the lightest squark ( slepton ) is the right - handed down squark ( left - handed selectron ) .
[ thus , the right - handed strange squark ( the left - handed smuon ) is also light . ] in the other regions of the left ( right ) panels , the right - handed up squark ( right - handed selectron ) is the lightest squark ( slepton ) .
( the right - handed smuon is also light . ) in the wide regions , the lightest squark is lighter than 3tev , which may be discovered at the future lhc experiments .
also , the slepton can be light as 500gev for @xmath121tev .
[ [ muon - g-2 ] ] muon @xmath83 + + + + + + + + + + + + + + + + + + + + + + finally , let us briefly comment on regions where the muon @xmath83 anomaly @xcite . ]
is explained via susy contributions @xcite . since the sleptons and bino can be light as @xmath7gev and @xmath123tev ,
respectively for @xmath124tev and @xmath125 is as large as @xmath126tev , there are regions consistent with the muon @xmath83 experiment . in fig .
[ fig : g-2 ] , we show the regions consistent with the experimental value of the muon @xmath83 at @xmath100 ( red ) , 1.5@xmath101 ( orange ) and 2@xmath101 ( yellow ) level . in the left ( right ) panels ,
the contours of the lightest squark ( slepton ) mass are also shown .
we take @xmath102 ( 140 ) tev in the upper ( lower ) two panels .
we include leading two - loop corrections : the correction to the muon yukawa coupling @xcite and the logarithmic qed correction @xcite .
these regions are explored in details in ref .
@xcite , where further information such as the bottom - tau yukawa unification and bottom - tau - top yukawa unification can be found . are shown in refs .
@xcite ] .
the triangles and pentagons ( squares and circles ) show maximum ( minimum ) values .
, title="fig : " ] .
the triangles and pentagons ( squares and circles ) show maximum ( minimum ) values .
, title="fig : " ] ( gev ) and @xmath127 ( tev ) for @xmath128 and @xmath129 .
the red solid ( blue dashed ) line shows @xmath111 @xmath130 . here , @xmath127 is a mass of the lightest squark .
, title="fig : " ] ( gev ) and @xmath127 ( tev ) for @xmath128 and @xmath129 . the red solid ( blue dashed ) line shows @xmath111 @xmath130 . here , @xmath127 is a mass of the lightest squark .
, title="fig : " ] since yukawa and gauge couplings break @xmath2 explicitly , there might be small susy - breaking soft masses for the squarks and sleptons . in this subsection
, we consider the case that squark and sleptons have a common soft mass @xmath131 of @xmath132tev in addition to the contribution from anomaly mediation in eq.([eq : amsb_scalar ] ) .
notice that @xmath131 breaks the @xmath2 symmetry explicitly .
the model parameters in this setup are @xmath133 the masses of the lightest chargino and the gluino are shown in fig .
[ fig:3 ] .
we take @xmath134 .
the maximum ( minimum ) values are denoted as triangles and pentagons ( squares and circles ) .
the lightest chargino ( neutralino ) is almost pure wino as in the case of sec .
[ sec : vanish ] .
we scan the parameter space over the ranges @xmath135 with the fixed @xmath44 , @xmath136 .
the result is very similar to the one in fig .
[ fig:1 ] , where there is an approximate one - to - one correspondence between the chargino mass and the gravitino mass . [ cols="^,^,^,^,^",options="header " , ] in fig.[fig : squark_m0 ]
, we show the contours of the squark mass and @xmath111 on @xmath20-@xmath131 plane for @xmath128 ( left ) and -1 ( right ) .
we set @xmath137 .
we focus on the viable parameter space , where @xmath110 , corresponding to the gravitino mass up to 400 ( 300 ) tev for @xmath138 .
if the universal scalar mass at @xmath139 is smaller than 3 - 4 tev , the lightest squark is lighter than about 3.5tev , which may be accessible to the future lhc experiments .
we show mass spectra at some example points in the viable parameter space of the model ( table [ tab : sample ] ) . on the points * i*-*iii * , the masses of the squarks and sleptons vanish at @xmath139 , while on the point * iv * , we introduce the small universal mass @xmath131 . on the point * iii * , @xmath84 ; therefore , @xmath67 is large as @xmath140 .
on the point * ii * , the smuons are light and the susy contribution to the muon @xmath83 is @xmath141 . on that point ,
the calculated higgs boson mass using susyhd is @xmath142gev .
however , the stop mass is as large as 15tev , and the computed higgs mass using feynhiggs 2.12.0 @xcite is larger than 125gev ; therefore , the point could be consistent with the observed higgs boson mass . on the listed points ,
the mass of the lightest squark is smaller than 3tev , which may be tested at the lhc run-2 or at the high luminosity lhc .
the @xmath23 non - linear sigma model is very attractive , since it may provide us an intriguing answer to one of the fundamental questions , why we have three families of quarks and leptons . in the @xmath2 nls model ,
the masses of squarks and sleptons vanish at the tree level while the higgs doublets have soft susy breaking masses of the order of the gravitino mass . in this paper
, we have shown that the @xmath23 nls model is consistent with all observations if one adopts the pure gravity mediation or minimal split susy , where the gaugino masses arise only from anomaly mediation . the tachyonic slepton problem in anomaly mediation
is solved due to the renormalization group running effects from the negative higgs soft mass squares .
we have shown that if the observed baryon asymmetry is explained by the thermal leptogenesis , the squarks are lighter than 2 - 3 tev in a wide range of the viable parameter space , and we expect them to be discovered at the lhc run-2 or at the high luminosity running . moreover
, the sleptons may be as light as o(100)gev , giving rise to a possibility for explaining the muon @xmath83 anomaly .
the sleptons are also interesting target at the lhc and at ilc .
although we have concentrated on the case that the @xmath8-parity is conserved , one can also consider the small @xmath8-parity violation . in this case , it is easy to imagine that the viable parameter region becomes wider , since the squarks and sleptons can be lighter than the lightest neutralino .
note that the testability of the model is also enhanced : the stable squark can be checked by searching an @xmath8-hadron for heavy gravitino of @xmath143700 tev ; the stable slepton can be as light as @xmath144(100 ) gev even for @xmath20 = 1000 tev .
finally , let us comment on the susy cp - problem for @xmath102-@xmath145 tev , where the selectron , bino and first / second generation squarks are as light as a few tev . in this case
, the constraints from the electron and neutron electric dipole moment ( edm ) are severe . for instance , in the base that the @xmath146-term and gravitino mass are real , the argument of the higgs @xmath147-term needs to be as small as @xmath148-@xmath149 to avoid the constraint from the electron edm @xcite .
it is , however , remarkable that we need cp violation only in the @xmath2-breaking yukawa couplings , so far .
thus , it is very interesting to consider that violations of two independent symmetries , cp and @xmath2 , arise from an underlying common physics .
this work is supported by jsps kakenhi grant numbers jp26104009 ( t.t.y ) , jp26287039 ( t.t.y . ) , jp16h02176 ( t.t.y ) , jp15h05889 ( n.y . ) and jp15k21733 ( n.y . ) ; and by world premier international research center initiative ( wpi initiative ) , mext , japan ( t.t.y . ) .
to explicitly see that the soft mass of the ng multiplet vanishes even if there is a direct coupling to the susy breaking field , let us consider the following k " ahler potential and superpotential : @xmath150 where @xmath36 and @xmath151 is a constant term ; @xmath29 is required for the @xmath2 invariance of the lagrangian .
we take unit of @xmath152 .
@xmath153 where @xmath154 .
we see that @xmath155 \nonumber \\ & = & f(x , |z|^2),\end{aligned}\ ] ] where the dependence on @xmath156 arises only though @xmath157 .
first consider the simple case : @xmath158 , @xmath159 the inverse matrix is almost same as in the case of sec . 2 ; thus , the argument in sec .
2 is still valid , even if there is an additional term : @xmath160 so far , @xmath161 vanishes at the minimum .
next we consider the case @xmath162 .
the inverse matrix is given by @xmath163 \ , , \nonumber \\
{ \rm det}(k)\ , k^{-1}_{\phi z } & = & 0 \ , , \nonumber \\ { \rm det}(k)\ , k^{-1}_{s z } & = & -c ( 1+c |z|^2 ) z^\dag f'^2 .\end{aligned}\ ] ] therefore , only @xmath164 = g_2 ( x , |z|^2)\end{aligned}\ ] ] is relevant as in eq .
( [ eq : gs_mass ] ) .
this @xmath165 gives the vanishing soft mass to the ng multiplet .
s. irie and y. yasui , z. phys .
c * 29 * , 123 ( 1985 ) .
m. ibe , t. moroi and t. t. yanagida , phys .
b * 644 * , 355 ( 2007 ) [ hep - ph/0610277 ] .
m. ibe and t. t. yanagida , phys .
b * 709 * , 374 ( 2012 ) [ arxiv:1112.2462 [ hep - ph ] ] .
n. arkani - hamed , a. gupta , d. e. kaplan , n. weiner and t. zorawski , arxiv:1212.6971 [ hep - ph ] .
m. fukugita and t. yanagida , phys .
b * 174 * , 45 ( 1986 ) .
w. buchmuller , r. d. peccei and t. yanagida , ann .
nucl . part .
sci . * 55 * , 311 ( 2005 ) [ hep - ph/0502169 ] .
s. davidson , e. nardi and y. nir , phys .
rept . * 466 * , 105 ( 2008 ) [ arxiv:0802.2962 [ hep - ph ] ] .
z. komargodski and n. seiberg , jhep * 1007 * , 017 ( 2010 ) [ arxiv:1002.2228 [ hep - th ] ] .
t. kugo and t. t. yanagida , prog .
phys . * 124 * , 555 ( 2010 ) [ arxiv:1003.5985 [ hep - th ] ] .
t. goto and t. yanagida , prog .
phys . *
83 * , 1076 ( 1990 ) .
l. j. hall , r. rattazzi and u. sarid , phys .
d * 50 * , 7048 ( 1994 ) [ hep - ph/9306309 , hep - ph/9306309 ] .
m. carena , m. olechowski , s. pokorski and c. e. m. wagner , nucl .
b * 426 * , 269 ( 1994 ) [ hep - ph/9402253 ] .
g. aad _ et al . _ [ atlas collaboration ] , phys .
d * 88 * , no .
11 , 112006 ( 2013 ) [ arxiv:1310.3675 [ hep - ex ] ] . b. bhattacherjee , m. ibe , k. ichikawa , s. matsumoto and k. nishiyama , jhep * 1407 * , 080 ( 2014 ) [ arxiv:1405.4914 [ hep - ph ] ] .
g. f. giudice , a. notari , m. raidal , a. riotto and a. strumia , nucl .
b * 685 * , 89 ( 2004 ) [ hep - ph/0310123 ] .
w. buchmuller , p. di bari and m. plumacher , annals phys .
* 315 * , 305 ( 2005 ) [ hep - ph/0401240 ] .
g. w. bennett _ et al . _ [ muon g-2 collaboration ] , phys .
d * 73 * , 072003 ( 2006 ) [ hep - ex/0602035 ] .
b. l. roberts , chin .
c * 34 * , 741 ( 2010 ) [ arxiv:1001.2898 [ hep - ex ] ] .
k. hagiwara , r. liao , a. d. martin , d. nomura and t. teubner , j. phys .
g * 38 * , 085003 ( 2011 ) [ arxiv:1105.3149 [ hep - ph ] ] . m. davier , a. hoecker , b. malaescu and z. zhang , eur . phys .
j. c * 71 * , 1515 ( 2011 ) erratum : [ eur .
j. c * 72 * , 1874 ( 2012 ) ] [ arxiv:1010.4180 [ hep - ph ] ] .
j. l. lopez , d. v. nanopoulos and x. wang , phys .
d * 49 * , 366 ( 1994 ) [ hep - ph/9308336 ] .
t. moroi , phys . rev .
d * 53 * , 6565 ( 1996 ) erratum : [ phys .
d * 56 * , 4424 ( 1997 ) ] [ hep - ph/9512396 ] . s. marchetti , s. mertens , u. nierste and d. stockinger , phys .
d * 79 * , 013010 ( 2009 ) [ arxiv:0808.1530 [ hep - ph ] ] .
g. degrassi and g. f. giudice , phys .
d * 58 * , 053007 ( 1998 ) [ hep - ph/9803384 ] .
m. endo , k. hamaguchi , s. iwamoto and n. yokozaki , phys .
d * 84 * , 075017 ( 2011 ) [ arxiv:1108.3071 [ hep - ph ] ] .
t. moroi , r. sato and t. t. yanagida , phys .
b * 709 * , 218 ( 2012 ) [ arxiv:1112.3142 [ hep - ph ] ] .
m. endo , k. hamaguchi , s. iwamoto and n. yokozaki , phys .
d * 85 * , 095012 ( 2012 ) [ arxiv:1112.5653 [ hep - ph ] ] .
m. endo , k. hamaguchi , s. iwamoto , k. nakayama and n. yokozaki , phys .
d * 85 * , 095006 ( 2012 ) [ arxiv:1112.6412 [ hep - ph ] ] .
k. nakayama and n. yokozaki , jhep * 1211 * , 158 ( 2012 ) [ arxiv:1204.5420 [ hep - ph ] ] .
s. heinemeyer , w. hollik and g. weiglein , eur .
j. c * 9 * , 343 ( 1999 ) [ hep - ph/9812472 ] .
g. degrassi , s. heinemeyer , w. hollik , p. slavich and g. weiglein , eur .
j. c * 28 * , 133 ( 2003 ) [ hep - ph/0212020 ] .
m. frank , t. hahn , s. heinemeyer , w. hollik , h. rzehak and g. weiglein , jhep * 0702 * , 047 ( 2007 ) [ hep - ph/0611326 ]
. t. hahn , s. heinemeyer , w. hollik , h. rzehak and g. weiglein , phys .
lett . *
112 * , no .
14 , 141801 ( 2014 ) [ arxiv:1312.4937 [ hep - ph ] ] . | we point out that a hypothesis of squarks and sleptons being nambu - goldstone ( ng ) bosons is consistent with pure gravity mediation or minimal split supersymmetry ( susy ) . as a concrete example
, we consider a susy @xmath0 non - linear sigma model in the framework of pure gravity mediation .
the model accommodates three families of the quark and lepton chiral multiplets as ( pseudo ) ng multiplets of the k " ahler manifold , which may enable us to understand the origin and number of the families .
we point out that squarks in the first and second generations are likely to be as light as a few tev if the observed baryon asymmetry is explained by the thermal leptogenesis ; therefore , these colored particles can be discovered at the lhc run-2 or at the high luminosity lhc . |
more than a decade ago the measurement by lvy
_ et al . _
@xcite of persistent currents in mesoscopic normal metal rings pierced by an aharonov - bohm flux @xmath0 has triggered a lot of theoretical activity @xcite . yet , up until now a truely convincing and generally accepted theoretical explanation of the surprisingly large persistent currents observed in ref .
@xcite and in subsequent experiments @xcite has not been found .
it has become clear , however , that this effect can not be explained within a model of non - interacting electrons .
ambegaokar and eckern ( ae ) @xcite were the first to examine the effect of electron - electron interactions on mesoscopic persistent currents : they realized that , to first order in the screened coulomb - interaction , the dominant contribution to the disorder averaged persistent current can be obtained from the two diagrams shown in fig . [
fig : ae ] , representing a special correction @xmath2 to the disorder averaged thermodynamic potential which depends strongly on the aharonov - bohm flux @xmath0 .
here the overline denotes averaging over the disorder . given the grand canonical potential @xmath3
, the corresponding persistent current @xmath4 can be obtained from the thermodynamic relation @xmath5 in a bulk metal at high densities the bare coulomb - interaction @xmath6 is strongly screened .
a simple way to take the screening into account diagrammatically is the random - phase approximation ( rpa ) . following this procedure ,
ae approximated the effective interaction ( in the imaginary frequency formalism ) as follows @xmath7 for momentum transfers @xmath8 small compared with the inverse elastic mean free path @xmath9 , and for frequency transfers @xmath10 small compared with the inverse elastic lifetime @xmath11 the disorder averaged polarization is given by @xmath12 where @xmath13 is the diffusion coefficient and @xmath14 is the average density of states at the fermi energy ( per spin ) in the absence of interactions .
note that @xmath15 , where @xmath16 is the volume of the system and @xmath17 is the average level spacing ( per spin ) at the fermi energy .
it turns out that both diagrams in fig .
[ fig : ae ] are dominated by momentum transfers of the order of the fermi momentum @xmath18 , which for a metallic system is large compared with @xmath9 .
( [ eq : rpa ] ) and ( [ eq : pol ] ) are therefore not suitable for a quantitatively accurate calculation of persistent currents . to make some progress analytically ,
ae estimated the contribution from the diagrams in fig .
[ fig : ae ] by replacing the effective interaction by a constant @xmath19 where @xmath20 denotes the fermi surface average over @xmath21 and @xmath22 . for simplicity
, it is assumed that the ring is quasi one - dimensional , with transverse thickness @xmath23 in the range @xmath24 , where @xmath25 is the circumference of the ring .
then diffusive motion is only possible along the circumference . at temperature
@xmath26 the resulting average persistent current can be written as @xcite @xmath27 where @xmath28 is the flux quantum @xcite and the fourier coefficients of the current are @xmath29 \ ; .
\label{eq : imae}\ ] ] here @xmath30 is the thouless energy and @xmath31 , where at zero temperature @xmath32 is the cutoff energy that regularizes the singularity in the cooperon in a finite system @xcite , see eqs .
( [ eq : gwl ] ) and ( [ eq : pdef ] ) below .
the coupling constant @xmath33 can be identified with the dimensionless effective interaction in the cooper channel to first order in perturbation theory .
ae estimated @xmath34 , assuming that the validity of the rpa can be extended to momentum transfers of the order of @xmath18 . however , higher order ladder diagrams in the cooper channel strongly reduce the effective interaction , so that @xmath35 is a more realistic estimate @xcite for the cu - rings in the experiment @xcite . in real space eq .
( [ eq : vbar ] ) amounts to replacing the electron - electron interaction by a local effective density - density interaction , @xmath36 more precisely , this replacement means that for distances @xmath37 larger than @xmath38 , the interaction is effectively local . in a recent letter schechter , oreg , imry , and
levinson @xcite pointed out that a different type of effective interaction can possibly lead to a much larger persistent current .
specifically , they used the bcs model to calculate the leading interaction correction to the orbital linear magnetic response and found @xcite @xmath39 where @xmath40 is the attractive dimensionless interaction in the bcs model , and the coherence energy @xmath41 is the smaller energy of @xmath42 and the debye energy @xmath43 .
( [ eq : chibcs ] ) should be compared with the corresponding result for the local interaction model used by ae , which implies according to eqs .
( [ eq : iae ] ) and ( [ eq : imae ] ) , @xmath44 where we have used @xmath45 and retained only the leading logarithmic order .
note that the logarithm is due to the slow decay ( @xmath46 ) of the fourier coefficients @xmath47 of @xmath48 , so that all coefficients with @xmath49 contribute to the linear response .
for @xmath50 the linear magnetic response in the bcs model is parametrically larger than the linear response in the local interaction model .
whether or not this remains true beyond the linear response has not been clarified .
note also that in the bcs model the linear magnetic response is diamagnetic because the effective interaction is attractive ( @xmath51 ) , whereas the linear response in the local interaction model is paramagnetic , corresponding to a repulsive effective interaction ( @xmath52 .
an interesting observation made by the authors of ref .
@xcite is that an effective interaction different from the local interaction used by ae can lead to a much larger persistent current , at least for sufficiently small flux @xmath0 , where it is allowed to calculate the current from the linear response .
given the rather crude approximations in the microscopic derivation of the local interaction model , it seems worth while to explore the magnetic response for other types of effective interactions .
a possibility which so far has not been thoroughly analyzed is an interaction which is dominated by small momentum transfers .
note that the assumption that only forward scattering processes ( corresponding to vanishing momentum transfer ) have to be taken into account for a consistent description of the low - energy and long - wavelength properties of normal metals lies at the heart of the landau s fermi liquid theory .
the landau model is in a sense the opposite extreme of the local interaction model , because the effective interaction in the landau model is proportional to a kronecker - delta in momentum space , @xmath53 where the landau parameter @xmath54 can be determined from experiments .
in fact , the dimensionless landau parameter @xcite @xmath55 can be written as @xmath56 , where @xmath57 is the bulk modulus , @xmath58 is the effective mass , and @xmath59 and @xmath60 are the corresponding quantities in the absence of interactions . inserting the known bulk values for cu @xcite , @xmath61 and @xmath62 , we find @xmath63 , which is a factor of @xmath64 larger than the corresponding estimate @xmath65 in the local interaction model
. note that in real space eq .
( [ eq : f0def ] ) corresponds to a constant effective interaction , proportional to the inverse volume of the system @xmath66 given an effective interaction of the form ( [ eq : f0def ] ) , the dominant flux - dependent contributions to the average potential @xmath67 to first order in the interaction are shown in fig .
[ fig : as ] .
the fock diagrams ( b)(d ) have been discussed previously in refs .
@xcite ; as first pointed out by bal - monod and montambaux @xcite , to leading order in the small parameter @xmath68 , the three fock diagrams in fig .
[ fig : as ] ( b)(d ) cancel , so that a direct evaluation of the sum of these diagrams is rather difficult . to calculate the leading contribution of these diagrams
, we note that the fermion loops in fig .
[ fig : as ] ( b)(d ) can be identified with contributions to the disorder averaged polarization , which for general frequencies and small wavevectors can be written as @xcite @xmath69 where @xmath70 is a generalized frequency - dependent diffusion coefficient .
the crucial observation is now that the sum of the three fock diagrams in fig .
[ fig : as ] ( b)(d ) corresponds to the usual weak localization correction to the average conductance @xcite , @xmath71 \ ; , \ ] ] where @xmath72 essentially we have used the equation of continuity to replace the charge vertices in fig .
[ fig : as ] by current vertices , which can not be renormalized by singular diffusion corrections .
the fact that a gauge transformation replacing charge vertices by current vertices can be used to avoid the explicit calculation of vertex corrections has also been employed in ref.@xcite to calculate the zero bias anomaly in the tunneling density of states of two - dimensional disordered electrons interacting with coulomb forces .
the evaluation of the contribution of the three fock diagrams in fig .
[ fig : as ] to the persistent current is now straightforward .
note that for a thin ring with @xmath73 the @xmath74-summation is one - dimensional , with quantized wave - vectors @xmath75 , @xmath76 .
then we obtain for the @xmath77-th fourier component of the average current due to the fock diagrams ( b)(d ) in fig .
[ fig : as ] for the landau model @xcite , @xmath78 due to the extra factor of inverse volume , this contribution is , for experimentally relevant parameters @xcite , negligible compared with corresponding result in the local interaction model given in eq .
( [ eq : imae ] ) .
the hartree diagram in the landau model is more interesting .
the fact that the diagram with two cooperons shown in fig .
[ fig : as ] ( a ) dominates the persistent current due to electron - electron interactions with momentum transfers @xmath79 has already been pointed out in ref .
a similar diagram with two cooperons ( but without interaction line ) dominates the fluctuations of the number of energy levels in a fixed energy window centered at the fermi energy @xcite . using the approximate relation @xmath80 between the persistent current @xmath81 at constant particle number and the fluctuation @xmath82 of the particle number at constant chemical potential @xmath83 ,
several authors have realized @xcite that without interactions the two - cooperon diagram determines the average persistent current in a canonical ensemble .
note that the hartree diagram in fig .
[ fig : as ] ( a ) does not contain any vertex corrections analogous to the diffusion corrections of the vertices in the fock diagams ( b)(d ) .
this is due to the fact that the interaction line in the hartree process does not transfer any energy .
hence , the two green functions attached to the vertex of a hartree interaction are either both retarded or both advanced , so that it is impossible to attach a singular diffuson to the vertex . for the landau model the hartree diagram in fig .
[ fig : as ] ( a ) yields at finite temperature @xmath84 the following correction to the disorder averaged grand canonical potential , @xmath85 ^ 2 \nonumber \\ & & \hspace{-15 mm } \times \sum _ { { \bf{k } } } [ \overline{g}_0 ( { \bf{k } } , i \tilde{\omega}_n ) ] ^2 \overline{g}_0 ( - { \bf{k } } + { \bf{q } } , i \tilde{\omega}_{n^{\prime } } ) \nonumber \\ & & \hspace{-15 mm } \times \sum _ { { \bf{k}}^{\prime } } [ \overline{g}_0 ( { \bf{k}}^{\prime } , i \tilde{\omega}_{n^{\prime } } ) ] ^2 \overline{g}_0 ( - { \bf{k}}^{\prime } + { \bf{q } } , i \tilde{\omega}_{n } ) \ ; .
\label{eq : omegah0 } \end{aligned}\ ] ] here @xmath86 is the elastic lifetime , @xmath87 are fermionic matsubara frequencies , and @xmath88 is the disorder averaged non - interacting matsubara green function . since the cooperons ( i.e. the second line ) in eq .
( [ eq : omegah0 ] ) are only singular for @xmath89 , and because the @xmath90- and @xmath91- sums are dominated by momenta of the order of the fermi momentum , we may approximate @xmath92 and @xmath93 in eq .
( [ eq : omegah0 ] ) .
the product of the last two lines of eq .
( [ eq : omegah0 ] ) gives then rise to a factor of @xmath94 ^ 2 $ ] , so that we obtain @xmath95 with the dimensionless coefficient @xmath96 ^ 2 } \label{eq : pdef } \ ; , \ ] ] where @xmath97 are bosonic matsubara frequencies .
assuming again a thin ring with @xmath73 , we find in the limit @xmath98 for the fourier components of the persistent current , @xmath99 comparing this expression with the corresponding result ( [ eq : imae ] ) of the local interaction model , we see that in the landau model the fourier components @xmath100 are independent of @xmath77 as long as @xmath101
. therefore the linear magnetic response is determined by all fourier components up to @xmath102 , @xmath103 where we have assumed that @xmath104 , so that @xmath105 ^ 2 } \approx \frac{1}{\gamma } \approx \frac { \pi e_c}{\delta_0 } \ ; .\ ] ] note that the small energy scale @xmath17 of eq .
( [ eq : ih ] ) has disappeared on the right - hand side of eq .
( [ eq : chil ] ) , and is replaced by the much larger thouless energy @xmath106 . due to the faster decay of the fourier components ( [ eq : ilfock ] ) of the fock contribution in the landau model , the linear response due to the fock diagrams shown in fig .
[ fig : as ] ( b)(d ) is a factor of @xmath107 smaller than the corresponding hartree contribution .
interestingly , the anomalously large linear magnetic response in the bcs model given in eq .
( [ eq : chibcs ] ) is also dominated by the hartree process @xcite .
thus , the importance of hartree interactions for persistent currents is some extent independent of a specific model for the interaction . for the cu - rings used in the experiment @xcite we estimate @xmath63 , @xmath35 and @xmath108 ; with these values the linear magnetic response due to forward scattering is more than four times larger than the linear response in the local interaction model considered by ae @xcite . to take both contributions into account one
should parameterize the total effective interaction as @xmath109 which in momentum space amounts to @xmath110 using the estimate for bulk cu given above , @xmath111 and @xmath112 , we find @xmath113 , which supports our assumption that there is indeed a strong enhancement of the effective interaction in the forward scattering channel .
in summary , we have shown that a forward scattering excess interaction , involving a vanishing momentum transfer , yields the dominant contribution to the _ linear _ magnetic response of mesoscopic metal rings for experimentally relevant parameters @xcite .
on the other hand , outside the linear response regime the persistent current is dominated by the term @xmath114 involving large momentum transfers , at least if we use the bulk estimates for @xmath114 and @xmath1 for cu . however , for a mesoscopic disordered cu - ring it is not obvious that the bulk estimates are reliable .
note also that in the bulk the normal fermi liquid is stable as long as @xmath115 , so that for @xmath116 the linear magnetic response in the normal state can be diamagnetic , in spite of the fact that the effective coupling @xmath117 in the cooper channel is positive .
moreover , in the vicinity of an @xmath118-wave pomeranchuk instability @xcite , where @xmath119 and @xmath120 , one should replace @xmath1 by @xmath121 . in this case we predict a strongly enhanced diamagnetic linear response . in fact , the forward scattering channel might then dominate the persistent current even beyond the linear order in the flux @xmath0 . to clarify this point , a better microscopic theory of the effective electron - electron interaction in mesoscopic disordered metals
is necessary . in particular
, a microscopic theory should properly treat the problem of screening in a finite system and incorporate the breakdown of fermi liquid theory in quasi one - dimensional disordered metals at sufficiently low temperatures @xcite .
we prefer to express the response in terms of the normal flux quantum @xmath122 , whereas @xmath123 in ref .
@xcite is the superconducting flux quantum .
this is the reason why the numerical prefactor in our eq .
( [ eq : chibcs ] ) is @xmath124 , while the corresponding prefactor in eq .
( 1 ) of ref .
@xcite is @xmath125 .
note that we define @xmath14 to be the density of states _ per spin _
, so that @xmath126 agrees with the usual definition of the dimensionless landau parameter .
n. w. ashcroft and n. d. mermin , _ solid state physics _ , ( holt - saunders , philadelphia , 1976 ) .
d. vollhardt and p. wlfle , phys .
b * 22 * , 4666 ( 1980 ) .
p. kopietz , phys .
lett . * 81 * , 2120 ( 1998 ) .
p. kopietz , phys .
lett . * 70 * , 3123 ( 1993 ) ; erratum : * 71 * , 306 ( 1993 ) . b. l. altshuler and b. i. shklovskii , zh .
teor . fiz .
* 91 * , 220 ( 1986 ) [ sov .
jetp * 64 * , 127 ( 1986 ) ] .
a. schmid , phys .
66 * , 80 ( 1991 ) .
f. von oppen and e. k. riedel , phys .
lett . * 66 * , 84 ( 1991 ) .
b. l. altshuler , y. gefen , and y. imry , phys .
* 66 * , 88 ( 1991 ) .
i. j. pomeranchuk , zh .
35 * , 524 ( 1958 ) [ sov .
jetp * 8 * , 361 ( 1958 ) ] .
g. murthy and r. shankar , phys .
* 90 * , 066801 ( 2003 ) ; d. herman , h. mathur , and g. murthy , phys .
b * 69 * , 041301 ( 2004 ) ; g. murthy , r. shankar , d. herman , and h. mathur , _ ibid . _ * 69 * , 075321 ( 2004 ) . b. l. altshuler and a. g. aronov , in _ electron - electron interactions in disordered systems _ , edited by a. l. efros and m. pollak , ( north holland , amsterdam , 1985 ) . | we calculate the effect of electron - electron interactions involving vanishing momentum transfer ( forward scattering ) on the orbital linear magnetic response of disordered metal rings pierced by a magnetic flux @xmath0 . using the bulk value of the landau parameter @xmath1 for copper
, we find that in the experiment by lvy _
et al . _
[ phys .
rev .
lett . * 64 * , 2074 ( 1990 ) ] the forward scattering contribution to the _ linear _ magnetic response is larger than the corresponding contribution from large momentum transfers considered by ambegaokar and eckern [ phys .
rev .
lett . * 65 * , 381 ( 1990 ) ] .
however , outside the regime of validity of linear response and to first order in the effective screened interaction the persistent current is dominated by scattering processes involving large momentum transfers .
= 10000 |
recently , both the singly ionized barium ( ba@xmath0 ) and ( ra@xmath0 ) are proposed as suitable candidates for the parity non - conservation ( pnc ) experiments @xcite .
these candidates are also suitable for the atomic clock experiments @xcite .
accurate determination of the electric dipole ( e1 ) matrix elements are essential in achieving sub - one percent pnc amplitudes in the above candidates @xcite .
they are also crucial in determining the dipole polarizabilities accurately which are required to estimate shifts due to the stray electric fields applied @xcite during the measurements in the above proposed experiments .
there is no direct method to measure them experimentally .
mainly , these results are estimated from the branching ratio ( lifetime ) measurements from various states . however , when branching ratios of two or more different channels from a given state are relatively significant , it is not possible to estimate the e1 matrix elements precisely .
there are a few branching ratio @xcite and lifetime @xcite measurements of the first p - states available in ba@xmath0 , although they are not very precise to estimate the e1 matrix elements accurately . on the otherhand ,
such experiments are not available yet in ra@xmath0 .
it is also possible to calculate the above matrix elements accurately using a highly potential many - body method .
ba@xmath0 and ra@xmath0 are heavy systems implying both the electron correlation and relativistic effects will be much stronger in these systems .
variety of relativistic many - body methods have already been employed to calculate the e1 matrix elements in the considered systems @xcite .
but the results obtained by all these works differ at significant precision for which they can be used to determine sub - one percent accuracy of the pnc amplitudes or other experimentally measured quantities like light - shift ratios @xcite in the considered systems . in this work ,
we have employed the relativistic coupled - cluster ( rcc ) method to account more electron correlation effects through the non - linear terms for the determination of many e1 matrix elements in both ba@xmath0 and ra@xmath0 .
these results are then used to evaluate the light - shift ratios in ba@xmath0 at two wavelengths on which experiments were conducted @xcite . by analyzing the trends of contributions to the light - shift ratios and transition probabilities , we try to give the upper limits to some of the important matrix elements by combining with the experimental result in that system .
the matrix elements are further used to determine the lifetimes and static dipole polarizabilities in ba@xmath0 . by following the same procedure ,
we determine in ra@xmath0 the lifetimes of the first excited p - states and light - shift ratios at the same wavelengths in which experiments were carried out in ba@xmath0 .
the shift of energy in an atomic state @xmath1 , where @xmath2 represents the total angular momentum of the state with its azimuthal component @xmath3 and @xmath4 is the additional index representing other required quantum numbers , due to non - resonant ac light in an average period of light oscillations and neglecting the mixing of the magnetic sub - levels is given by @xcite @xmath5 where @xmath6 is the applied vector electric field and @xmath7 is its magnitude in the z - component . here
@xmath8 , @xmath9 and @xmath10 are the scalar , vector and tensor polarizabilities of the state @xmath1 .
this expression differs by the term containing vector polarizability when the dc electric field is applied . using the first - order time - dependent perturbation theory and assuming large detuning from resonance with the applied electric field , the scalar , vector and tensor dynamic polarizabilities of @xmath1 given as @xcite @xmath11 } \sum_{k \ne j } \frac{e_j -e_k}{(e_j -e_k)^2 - \omega^2 } \nonumber \\ & & |\langle \gamma ' k ||d|| \gamma j \rangle |^2 \ \
\phi_0(j , k ) \label{eqn21 } \\ \alpha_1(\gamma , j , m ) & = & - \frac{1}{[j ] } \sum_{k \ne j } \frac{\omega}{(e_j -e_k)^2 - \omega^2 } \nonumber \\ & & |\langle \gamma ' k ||d|| \gamma j \rangle |^2 \ \
\phi_1(j , k ) \label{eqn22 } \\
\alpha_2(\gamma , j , m ) & = & - \frac{2}{3[j ] } \sum_{k \ne j } \frac{e_j -e_k}{(e_j -e_k)^2 - \omega^2 } \nonumber \\ & & |\langle \gamma ' k ||d|| \gamma j \rangle |^2 \ \
\phi_2(j , k ) , \label{eqn23}\end{aligned}\ ] ] respectively , where @xmath12 is the e1 operator , @xmath13 $ ] is the degeneracy factor equal to @xmath14 , @xmath15 and @xmath16 represent different states than @xmath1 with opposite parity , matrix elements with double bars represent the reduced matrices , @xmath17 represent energies of the corresponding states in the absence of the external electric field and @xmath18 is the frequency of the applied electric field . the scalar factors @xmath19 s ( @xmath20 ) multiplied in the above equations are defined as @xmath21 the static dipole polarizabilities can be determined by substituting @xmath18 as zero in the above equations . in fact , the vector shift is maximal than other two shifts for the purely circular polarized light when it is aligned with any existing magnetic field @xcite . in this case
, it is possible to measure the ratio of light shifts instead of measuring the light shifts or light intensities directly .
mathematically , it is equivalent to say as the following ; the ratio of the shift in energies of two different states ( @xmath22 and @xmath23 ) is given by @xmath24 which is known as the light - shift ratio in the literature @xcite .
the probability coefficient ( in @xmath25 ) due to the allowed transition is given by @xmath26 } s_{f \rightarrow i}^{\text{e1 } } , \label{eqn5}\end{aligned}\ ] ] where @xmath27 ( ) and @xmath28 ( au ) are the wavelengths and strengths of the corresponding transitions , respectively . and
ra@xmath0 . @xmath29 and @xmath30 in ba@xmath0 and ra@xmath0 , respectively.,width=321 ] as seen in fig .
[ fig1 ] , the first excited p - states in both ba@xmath0 and ra@xmath0 can decay either to the ground s - states directly or via the excited d - states through the allowed transitions .
the lifetimes of these p - states can be evaluated from the inverse of the corresponding net transition probabilities due to all possible transition channels .
therefore , the net transition probabilities for the p - states are given by @xmath31 and hence , the lifetimes ( in @xmath32 ) of these states are given by @xmath33 where @xmath29 and @xmath30 in ba@xmath0 and ra@xmath0 , respectively . since the transition wavelengths can be determined from the excitation energies from the corresponding transitions , it is clear from the above equations that both the polarizabilities and transition probabilities depend upon the excitation energies ( or wavelengths ) and e1 matrix elements . since our motivation in the present work is to verify the accuracies of the e1 matrix elements , we consider the experimental ( observational ) excitation energies ( wavelengths ) for our purpose and analyze only the role of the e1 matrix elements from our calculations .
we employ the rcc theory for the dirac - coulomb hamiltonian in the single and double excitations approximation along with the important triple excitations ( ccsd(t ) method ) to calculate the atomic wave functions in both ba@xmath0 and ra@xmath0 . in this procedure
, we express the one valence ( @xmath34 ) configuration state function as @xcite @xmath35 where the reference state @xmath36 is constructed by appending the corresponding valence orbital to the closed - shell dirac - fock ( df ) wave function @xmath37 ; which in second quantization formalism is expressed as @xmath38 . in the above equation ,
@xmath39 and @xmath40 are the core and valence - core correlation operators . in the linear approximation ,
the above equation yields the form @xmath41 as shown in the earlier works , contributions ( electron correlation effects ) from higher triple and quadrupole excitations arise through the non - linear terms in the ccsd(t ) method and their contributions to the various properties in the heavy systems are significantly large @xcite .
the reduced transition matrix elements @xmath42 between two states @xmath43 and @xmath44 are evaluated by @xmath45 where @xmath46 and @xmath47 are the non - truncative series in the ordinary rcc theory .
however , in the ccsd(t ) method , only up to the effective three - body terms , when they are expanded using the general wick s theorem , of these series will contribute to the above expression .
these terms are calculated step by step and sandwiched between the necessary @xmath40 operators to evaluate the matrix elements . in order to estimate the errors associate with the calculated e1 matrix elements , we check the convergence of these results using both the length and velocity gauge expressions which at the single orbital level are given by @xmath48 - \frac{\alpha}{5 } ( \epsilon_i- \epsilon_f ) \left ( \frac{(\kappa_i-\kappa_f)}{2}[p_f(r ) q_i(r ) + q_f(r )
p_i(r ) ] + [ p_f(r ) q_i(r ) - q_f(r ) p_i(r ) ] \right ) \right \ } \label{eqn11}\end{aligned}\ ] ] in the length form and @xmath49 - [ p_f(r ) q_i(r ) - q_f(r ) p_i(r ) ] \right \ } , \label{eqn12}\end{aligned}\ ] ] in the velocity form , respectively . in these expressions
, @xmath50 is the racah vector with rank one , @xmath51 and @xmath52 are the large and small components of the dirac single particle orbitals , @xmath53 are the relativistic quantum numbers , @xmath54 are the single particle energies and @xmath55 is the fine structure constant .
we present our calculated e1 matrix elements of ba@xmath0 in table [ tab1 ] . the reported error bars in these results except from the matrix elements related to the d- and f- states are estimated from the differences between the length and velocity gauge results .
it was rather difficult to converge the velocity gauge results for the d- and f- states , therefore we consider the difference between the converged length gauge results using the ccsd(t ) method and just by single and double excitations approximation ( ccsd method ) , which are assumed as the upper limit to the contributions due to the neglected excitations , as the error bars .
however , these error bars are just used to estimate errors in the light - shift ratio calculations in the considered system and we analyze the accuracy of these results later by fitting with the experimentally measured values of the light - shift ratios at two different wavelengths as discussed below . lccccc transition & present & ref .
@xcite & ref .
@xcite & ref .
@xcite & ref .
@xcite + + & & & & & + @xmath56 & 3.36(1 ) & 3.310 & 3.3266 & 3.3357 & 3.300 + @xmath57 & 0.10(1 ) & 0.099 & 0.1193 & 0.0621 & + @xmath58 & 0.11(5 ) & 0.115 & 0.4696 & & + @xmath59 & 4.73(3 ) & 4.674 & 4.6982 & 4.7065 & 4.658 + @xmath60 & 0.17(5 ) & 0.035 & 0.3610 & 0.0868 & + @xmath61 & 0.11(5 ) & 0.073 & 0.5710 & & + @xmath62 & 2.44(4 ) & 2.493 & 2.3220 & & + @xmath63 & 0.66(5 ) & 0.705 & 0.7283 & & + @xmath64 & 3.80(2 ) & 3.882 & 3.6482 & & + @xmath65 & 0.97(5 ) & 1.025 & 1.0518 & & + @xmath66 & 3.11(3 ) & 3.055 & 2.9449 & & 3.009 + @xmath67 & 0.28(2 ) & 0.261 & 0.3050 & & + @xmath68 & 0.13(2 ) & 0.119 & 0.1121 & & + @xmath69 & 1.34(2 ) & 1.334 & 1.2836 & & 1.312 + @xmath70 & 0.16(1 ) & 1.472 & 0.1645 & & + @xmath71 & 0.07(2 ) & 0.070 & 0.0650 & & + @xmath72 & 3.75(11 ) & & & & + @xmath73 & 1.59(8 ) & & & & + @xmath74 & 0.17(2 ) & & & & + @xmath75 & 4.02(7 ) & 4.118 & 3.9876 & & 4.057 + @xmath76 & 0.46(1 ) & 0.432 & 0.4788 & & + @xmath77 & 0.21(2 ) & 0.206 & 0.1926 & & + @xmath78 & 1.08(4 ) & & & & + @xmath79 & 0.45(7 ) & & & & + @xmath80 & 0.15(2 ) & & & & + @xmath81 & 4.84(5 ) & & & & + @xmath82 & 2.47(6 ) & & & & + @xmath83 & 1.04(7 ) & & & & + @xmath84 & 4.89(10 ) & & & & + @xmath85 & 1.50(8 ) & & & & + @xmath86 & 2.33(7 ) & & & & + @xmath87 & 0.67(4 ) & & & & + @xmath88 & 6.91(21 ) & & & & + @xmath89 & 2.01(5 ) & & & & + [ tab1 ] lcccc intermediate & & + states ( @xmath16 ) & @xmath90 nm & @xmath91 nm & @xmath90 nm & @xmath91 nm + + & & & + @xmath92 & @xmath93978.638 & @xmath9345.080 & 48.786 & @xmath9320.419 + @xmath94 & @xmath930.014 & @xmath930.006 & @xmath930.035 & @xmath930.014 + @xmath95 & @xmath930.010 & @xmath930.004 & @xmath930.004 & @xmath930.002 + @xmath96 & 305.338 & 36.717 & 6.112 & @xmath931.121 + @xmath97 & 0.019 & 0.008 & @xmath930.004 & @xmath930.002 + @xmath98 & 0.005 & 0.002 & @xmath99 & @xmath100 + @xmath101 & & & 3.987 & 1.541 + @xmath102 & & & 0.153 & 0.162 + @xmath103 & & & 0.004 & 0.002 + others & 0.301 & 0.137 & 0.243 & 0.177 +
+ total & @xmath93672.994(1.944 ) & @xmath938.223(201 ) & 59.298(1.375 ) & @xmath9319.675(323 ) + [ tab2 ] we also compare our results with the earlier works in table [ tab1 ] .
as seen , most of the important low - lying e1 matrix elements from various works differ at the second decimal places which means the difference is mainly due to the different many - body methods employed and they may not be due to the numerical methods used in the calculations .
that is the reason why we have presented our results up to the second decimal places in the above table . as a matter of fact ,
correct results up to the second decimal places are crucial in obtaining the light - shift ratios correctly as will be shown below .
first , we would like to briefly discuss here the important differences in the various works presented in table [ tab1 ] .
geetha et al .
@xcite have employed the same approach like ours , but they had used part numerical orbitals using grasp and part analytical orbitals using gaussian type orbitals ( gtos ) . in the present work ,
we have considered pure analytical orbitals using gtos .
another difference between these two works is that @xmath104 of eq .
( [ eqn10 ] ) is truncated only at the effective one - body terms by geetha et al .
@xcite , where other higher order terms have been considered in the present calculation .
the main difference in the methods applied by dzuba et al .
@xcite and ours have been discussed by geetha et al . in their work @xcite . in summary , they have employed the green s function technique which is also an all order perturbative method like ours . in their work ,
the orbitals are also obtained analytically like the present work .
however , the level of approximation in both the works could be different which can not be obviously distinguished .
iskrenova - tchoukova and safronova @xcite have used linearized ccsd method with important triple excitations and orbitals have been constructed using the b - spline basis . as has been pointed out earlier , the non - linear terms can incorporate higher order correlation effects which are necessary @xcite to account in the considered heavy system .
again , the procedure to consider partial triple excitations in their method and in the present work is different .
we consider effects of partial triple excitations by estimating each time their contributions to the energy of the corresponding valence state and then solve the ccsd method amplitudes self - consistently which involves the above energy .
however , iskrenova - tchoukova and safronova include them directly in the amplitude determining equations .
guet and johnson have just applied the relativistic many - body method at the second order approximation ( mbpt(2 ) method ) using the b - spline basis to calculate their results in contrast to our method which is all order in nature @xcite .
we have used our calculated reduced e1 matrix elements and experimental energies @xcite in eq .
( [ eqn22 ] ) to obtain the dynamic vector dipole polarizabilities ( @xmath9 ) of the @xmath105 and @xmath106 states in ba@xmath0 at two different wavelengths ( of external electric fields ) and they are presented in table [ tab2 ]
. only @xmath107 , @xmath108 , @xmath109 , @xmath110 , @xmath111 , @xmath112 , @xmath113 , @xmath114 and @xmath115 intermediate states ( @xmath16 ) have been taken into account as they seem to be the most important contributing states as seen from this table .
it can be noticed that the magnitude of @xmath9s of the @xmath105 state are larger than the @xmath106 state at both the wavelengths , but their differences are smaller when the frequency of the external field is small .
again , the sign of these quantities are different at the smaller wavelength ( 514.53 nm ) , whereas they are opposite at the large wavelength ( 1111.68 nm ) . for both the states ,
the largest contribution comes from the @xmath107 state .
the next highest contribution arises from the @xmath110 state .
the third largest contributions to @xmath9 of the @xmath106 state comes from the @xmath113 state .
surprisingly , the rest of the contributions are dominated by the higher f - states compared to the p - states .
we have given contributions from pure core correlations , core - valence correlations and higher states those are neglected as `` others '' .
the core correlation effects are around 0.319 au and 0.147 au at the wavelengths 514.53 nm and 1111.68 nm , respectively .
these contributions are evaluated using the mbpt(2 ) method .
the possible errors due to the omission of the higher states may come from the neglected f - states which seems to be important in the determination of @xmath9 of the @xmath106 state .
therefore , an approach similar to the one applied to determine the dynamic scalar and tensor polarizabilities @xcite will be more appropriate to use for the high precision light - shift ratio studies in the present system . now considering the above results of @xmath9 for both the @xmath105 and @xmath106 states and using eq .
( [ eqn4 ] ) , we obtain the light - shift ratios in ba@xmath0 as @xmath116 and @xmath117 , where core correlations contribute around 0.6% and 1% at the wavelengths 514.53 nm and 1111.68 nm , respectively .
the corresponding experimental results are @xmath118 and @xmath119 at the wavelengths
514.53 nm and 1111.68 nm , respectively @xcite .
our results match well with the corresponding experimental results .
it shows that accounting core correlation effects are essential for the high precision results in the light - shift studies , although their magnitudes seem to be very small in comparison to the @xmath9 of the @xmath105 state .
especially , their contributions are relatively larger at the higher wavelength ( smaller frequency ) .
the errors quoted in our light - shift ratio calculations are obtained from the error bars given for the e1 matrix elements in table [ tab1 ] .
we are now in a position to analyze the accuracy of the e1 matrix elements by studying their contributions to the light - shift ratio calculations .
since contributions from the @xmath107 state seems larger , we first analyze matrix elements involving this state . from table
[ tab1 ] , we can find that different theoretical predictions of the magnitude ( sign is irrelevant for our observable ) @xmath120 range from 3.30 au to 3.37 au ( along with the error bar ) . now keeping all other matrix elements unchanged from our calculations
, we observe that light - shift ratios vary from @xmath121 and @xmath122 to @xmath123 and @xmath124 . with @xmath125 , which is reported by geetha et al .
@xcite and iskrenova - tchoukova and safronova @xcite , we observe @xmath126 and @xmath122 . keeping the value of the above matrix element constant , we consider now the range of @xmath127 as 4.66 au to 4.76 au and observe from @xmath128 and @xmath124 to @xmath129 and @xmath130 .
to improve these results , we fix now @xmath131 and vary @xmath132 result again .
we find that when we consider @xmath133 and @xmath134 for @xmath131 , we get @xmath135 and @xmath136 and @xmath137 and @xmath138 , respectively . since the former result , is little close with the experimental results for both the wavelengths , we now keep @xmath133 and increase @xmath127 result and get @xmath139 and @xmath140 and @xmath141 and @xmath142 for 4.72 au and 4.74 au , respectively . therefore , we conclude now that @xmath127 will be around 4.73 au .
now keeping @xmath133 and @xmath143 when we vary @xmath144 from 3.00 au to 3.15 au , we get from @xmath145 and @xmath146 to @xmath147 and @xmath148 . results close to the experimental measurements are able to produce only when @xmath144 is around 3.08 au .
now we keep @xmath133 , @xmath143 and @xmath149 then vary @xmath150 matrix element from 1.28 au to 1.36 au .
we get light - shift ratios close to experimental results when @xmath150 is around 1.32 au . however , light - shift results vary slowly in the given range of this matrix element while the previous matrix elements considered earlier play the crucial roles in deciding the final results .
we now vary @xmath151 matrix element from 3.50 au to 4.00 au by considering @xmath133 , @xmath143 , @xmath149 and @xmath152 , we get from @xmath153 and @xmath154 to @xmath155 and @xmath156 . by fixing @xmath157 ,
when we reshuffle other matrix elements , we get light - shift ratios close to the experimental values for @xmath133 , @xmath158 , @xmath149 and @xmath159 .
when we keep these matrix elements constant and vary the matrix element of @xmath151 , we get the best light - shift ratios for 3.65 au and they correspond to @xmath160 and @xmath161 . for any other combinations ,
the light - shift ratios vary by large amount from the experimental results ; at least for one of the wavelengths . again , there are also measurements of the transition probabilities in ba@xmath0 available from three different experiments @xcite .
we evaluate them using the experimental wavelengths ( determined from the corresponding experimental excitation energies @xcite ) and e1 matrix elements presented in table [ tab1 ] .
these results are given in table [ tab3 ] .
we have also presented both the experimental and other theoretical results in the same table .
it can be clearly noticed that our results in all the cases match well with the experimental results with smaller error bars where as some of the earlier works differ significantly .
differences between all the theoretical works are discussed by geetha et al .
@xcite .
lccccc transition(@xmath162 ) & @xmath163 & @xmath164 & @xmath165 & others & expt .
+ + & & & & & + @xmath166 & 11.290 & 4935.5 & 95.131 & 93.68@xmath167 & 95(9)@xmath168 + & & & & 91.78@xmath169 & 95.5(10)@xmath170 + & & & & & 95(7)@xmath171 + @xmath172 & 22.373 & 4555.3 & 119.889 & 119.37@xmath167 & 106(9)@xmath168 + & & & & 116.25@xmath169 & 117(4)@xmath170 + & & & & & 118(8)@xmath171 + @xmath173 & 9.672 & 6498.7 & 35.701 & 32.609@xmath167 & 33.8(19)@xmath168 + & & & & 33.42@xmath169 & 33(8)@xmath170 + & & & & & 33(4)@xmath171 + @xmath174 & 1.796 & 5855.3 & 4.531 & 4.255@xmath167 & 4.69(29)@xmath168 + & & & & 3.342@xmath169 & 4.8(5)@xmath170 + & & & & & 4.48(6)@xmath171 + @xmath175 & 16.160 & 6143.4 & 35.305 & 34.93@xmath167 & 37.7(24)@xmath168 + & & & & 35.95@xmath169 & 37(4)@xmath170 + & & & & & 37(4)@xmath171 + [ tab3 ] @xmath176s are determined from the experimental excitation energies @xcite .
+ references : @xmath167@xcite , @xmath169@xcite , @xmath168@xcite , @xmath170@xcite and @xmath171@xcite .
now if we consider @xmath120 as 3.35 au as we analyzed from the light - shift ratio studies , we get @xmath177 as @xmath178 which is at the marginal lower side error limit of reader et al experimental result @xcite .
our @xmath173 result is also at the marginal upper limit of kastberg et al experimental result @xcite .
if we consider @xmath179 from the analysis of light - shift ratio studies then we get @xmath180 which is within the experimental error bar . by gathering all the informations from the analysis of roles of various matrix elements in the light - shift ratios and transition probabilities studies , we arrive in the conclusion at this point that the magnitude of the following matrix elements will be @xmath181
the above conclusions differ from the findings of the previous work by sherman et al .
@xcite who find @xmath182 where they had combined with geetha et al @xcite results to estimate these values .
further more , light - shift ratio measurements with a large number of wavelengths are required in order to estimate the above e1 matrix elements more accurately .
lcccc state(@xmath183 ) & @xmath184 & @xmath185 & others & expt .
+ + & & & & + @xmath107 & 130.146 & 7.68(7 ) & 7.89@xmath167 & 7.92(8)@xmath171 + & & & 7.92@xmath169 & 7.90(10)@xmath186 + & & & 7.83@xmath168 & + & & & 7.99@xmath170 & + @xmath110 & 159.72 & 6.26(11 ) & 6.30@xmath167 & 6.32(10)@xmath186 + & & & 6.31@xmath169 & 6.312(16)@xmath187 + & & & 6.27@xmath168 & + & & & 6.39@xmath170 & + [ tab4 ] references : @xmath167@xcite , @xmath169@xcite , @xmath168@xcite , @xmath170@xcite , @xmath171@xcite , @xmath186@xcite and @xmath187@xcite .
there are a number of measurements on the lifetimes of the @xmath188 and @xmath189 states are available in the literature @xcite , but the associate error bars of these measurements are very large .
we evaluate the net transition probabilities of these states using the above matrix elements and experimental wavelengths @xcite .
we give these results in table [ tab4 ] and compare with the corresponding experimental results .
results obtained from other works are also given in the same table . as it can be seen from this table
that the lifetime of the @xmath107 state is more than 2% difference from the earlier findings which needs to be verified by new experiments .
although , our calculation of the lifetime of the @xmath110 state is within the error bar of the experimental results and agree with other findings , but the @xmath175 contributes around 28% in its evaluation . since the accuracy of the e1 matrix element of @xmath190
is not verified well as it has been done for the transition matrix elements involved in the determination of lifetime of the @xmath107 state , the lifetime of the @xmath110 state may be little different than what we find , but it will be within the error bar that we have given ( which is large ) .
the error bars given in our lifetime calculations come only from the error bars of the e1 matrix elements as given above .
lcccccccc intermediate & @xmath105 & @xmath107 & & & + states ( @xmath16 ) & scalar & scalar & scalar & tensor & scalar & tensor & scalar & tensor + + @xmath105 & & @xmath9340.763 & @xmath9337.280 & 37.280 & & + @xmath191 & & 19.714 & 25.889 & @xmath9325.889 & & + @xmath192 & & 0.844 & 0.954 & @xmath930.954 & & + @xmath107 & 40.763 & & & & 22.992 & @xmath9322.992 & & + @xmath108 & 0.015 & & & & 0.064 & @xmath930.064 & & + @xmath109 & 0.014 & & & & 0.011 & @xmath930.011 & & + @xmath110 & 74.560 & & & & 3.846 & 3.077 & 24.210 & @xmath9324.210 + @xmath111 & 0.043 & & & & 0.021 & 0.017 & 0.116 & @xmath930.116 + @xmath112 & 0.014 & & & & 0.003 & 0.003 & 0.019 & @xmath930.019 + @xmath106 & & @xmath9345.984 & @xmath933.846 & @xmath933.077 & & + @xmath193 & & 68.101 & 8.275 & 6.620 & & & & + @xmath194 & & 4.163 & 0.434 & 0.347 & & & & + @xmath195 & & & @xmath9336.315 & 7.263 & & & & + @xmath196 & & & 72.167 & @xmath9314.433 & & & & + @xmath197 & & & 3.895 & @xmath930.779 & & & & + @xmath113 & & & & & 11.857 & @xmath932.371 & 0.668 & 0.763 + @xmath114 & & & & & 1.643 & @xmath930.329 & 0.089 & 0.102 + @xmath115 & & & & & 0.017 & @xmath930.003 & 0.009 & 0.010 + @xmath198 & & & & & & & 13.344 & @xmath934.766 + @xmath199 & & & & & & & 2.655 & @xmath930.948 + @xmath200 & & & & & & & 0.430 & @xmath930.154 + others & 9.137 & 11.304 & 11.716 & @xmath930.508 & 8.981 & @xmath93.019 & 9.000 & @xmath93.084 + + total & 124.546(1.256 ) & 17.379(2.920 ) & 45.890(3.831 ) & 5.870(128 ) & 49.435(1.458 ) & @xmath9322.694(538 ) & 50.540(1.410 ) & @xmath9329.422(931 ) + expt .
@xcite & 123.88(5 ) & & & & & & & + @xcite & 125.5(10 ) & & & & & & & + theory @xcite & 124.15 & & & & & & & + @xcite & 123.07 & & & & & & & + @xcite & 126.2 & & & & & & & + @xcite & 124.7 & & & & & & & + [ tab5 ] since the knowledge of the static dipole polarizabilities of different states in ba@xmath0 is important to estimate shifts in the energy levels when they are subjected to high precision experiments like atomic clock , pnc and so on , we have calculated these quantities for the first five low - lying states in the given system . likewise
vector polarizabilities , the scalar and tensor static dynamic polarizabilities are evaluated using our e1 matrix elements presented in table [ tab1 ] and their experimental excitation energies @xcite .
we have considered the @xmath105 , @xmath191 , @xmath192 , @xmath106 , @xmath193 , @xmath194 , @xmath195 , @xmath196 , @xmath197 , @xmath198 , @xmath199 and @xmath200 states in these calculations along with the intermediate states considered for the above vector polarizabilities calculations .
again , we have calculated the core correlation effects using the relativistic ccsd method employed for the closed - shell system in our earlier work @xcite .
the core - valence and higher state contributions other than the matrix elements reported in table [ tab1 ] are evaluated using the mbpt(2 ) method .
our static dipole polarizability results of various states in ba@xmath0 are presented in table [ tab5 ] .
the core - correlation effects for the scalar and tensor polarizabilities are 9.582 au and @xmath930.372 au , respectively .
there are experimental results available only for the ground state @xcite .
again , a couple of theoretical calculations available for the ground state @xcite . except iskrenova - tchoukova and safronova work @xcite , others have employed molecular codes to determine them .
also , iskrenova - tchoukova and safronova have used the e1 matrix elements from the linearized ccsd(t ) method to evaluate the ground state polarizabilities . in their work , they have considered only four important matrix elements to evaluate them and we have considered two more higher excited states matrix elements in this calculation . again , their core - correlation effects are considered using the lower order many - body methods and it is larger than our finding .
there are no other results for the excited states available to compare with our results .
the error bars given in our calculations are from the errors given for the e1 matrix elements in table [ tab1 ] .
the agreement between our ground state polarizability result with its experimental measurement further supports the accuracy of the e1 matrix elements used in this calculation .
lccc transition & present & ref .
@xcite & ref .
@xcite + + & & & + @xmath201 & 3.28(2 ) & 3.224 & 3.2545 + @xmath202 & 0.08(4 ) & 0.088 & + @xmath203 & 0.09(3 ) & 0.116 & + @xmath204 & 4.54(2 ) & 4.477 & 4.5106 + @xmath205 & 0.49(2 ) & 0.339 & + @xmath206 & 0.30(2 ) & 0.095 & + @xmath207 & 3.62(5 ) & 3.550 & 3.5659 + @xmath208 & 0.06(2 ) & 0.013 & + @xmath209 & 0.02(1 ) & 0.013 & + @xmath210 & 1.54(2 ) & 1.504 & 1.5117 + @xmath211 & 0.15(2 ) & 0.127 & + @xmath212 & 0.07(2 ) & 0.057 & + @xmath213 & 4.67(2 ) & & 4.4491 + @xmath214 & 0.86(4 ) & & + @xmath215 & 0.48(11 ) & & + @xmath216 & 4.83(8 ) & 4.816 & 4.8232 + [ tab6 ] we have also followed the same procedure as above to determine the light - shift ratios and lifetimes of the first excited p - states in ra@xmath0 .
we present the various e1 matrix elements used in these calculations in table [ tab6 ] .
the error bars of different matrix elements are estimated using the same procedure followed as in ba@xmath0 .
there are also other calculations available in this system @xcite .
dzuba et al .
@xcite have used the same method as in ba@xmath0 to calculate these matrix elements . the procedure followed by safronova et al .
@xcite is based on the linearized ccsd(t ) method like iskrenova - tchoukova and safronova work in ba@xmath0 @xcite .
most of our results match with these calculations , but they still differ at the second decimal places as in ba@xmath0 .
since ra@xmath0 has been proposed for the atomic clock @xcite and pnc experiments @xcite , knowledge of the accuracies of the above e1 matrix elements are essential .
we propose also similar light - shift ratio measurements at different wavelengths for the same purpose . as a preliminary study
, we consider the same off - resonant wavelengths as have been considered in the ba@xmath0 experiments @xcite and have carried out light - shift ratio calculations . in table
[ tab7 ] , we present the dynamic vector polarizabilities for the @xmath191 and @xmath193 states at @xmath90 nm and @xmath91 nm .
we use our e1 matrix elements presented in table [ tab6 ] with the experimental energies @xcite to evaluate them . in this calculation , we have considered the @xmath108 , @xmath109 , @xmath217 , @xmath111 , @xmath112 , @xmath218 , @xmath114 , @xmath115 and @xmath219 intermediate states .
lcccc intermediate & & + states ( @xmath16 ) & @xmath90 nm & @xmath91 nm & @xmath90 nm & @xmath91 nm + + & & & + @xmath94 & @xmath93391.483 & @xmath9337.764 & 31.921 & @xmath93869.894 + @xmath95 & @xmath930.008 & @xmath930.003 & @xmath930.002 & @xmath220 + @xmath221 & @xmath930.007 & @xmath930.003 & @xmath222 & @xmath223 + @xmath97 & 94.788 & 22.384 & 3.784 & @xmath932.632 + @xmath98 & 0.144 & 0.059 & @xmath930.005 & @xmath930.002 + @xmath224 & 0.035 & 0.015 & @xmath225 & @xmath226 + @xmath102 & & & 9.452 & 3.361 + @xmath103 & & & 0.149 & 0.061 + @xmath227 & & & 0.035 & 0.015 + others & 0.323 & 0.145 & 0.795 & 0.345 + + total & @xmath93296.207(4.909 ) & @xmath9315.167(255 ) & 46.128(679 ) & @xmath93868.749(24.027 ) + [ tab7 ] as seen in table [ tab7 ] , the trend of the contributions from various intermediate states is same as in ba@xmath0 .
the core - correlations are little larger compared to ba@xmath0 and they are around 0.346 au and 0.159 au at @xmath90 nm and @xmath91 nm , respectively .
we obtain around @xmath228 and @xmath229 .
the given error bars again come from our estimated accuracies of the e1 matrix elements
. since there are no experimental data available in ra@xmath0 to check the accuracy of the e1 matrix elements or light - shift ratios , our results will be useful for the future light - shift experiments in this system .
lccc transition(@xmath162 ) & @xmath163 & @xmath230 & @xmath165 + + & & & + @xmath231 & 11.086 & 4683.6 & 106.083 + @xmath232 & 20.612 & 3815.5 & 187.960 + @xmath233 & 13.104 & 10791.2 & 10.564 + @xmath234 & 2.372 & 7080.0 & 3.385 + @xmath235 & 23.329 & 8022.0 & 22.890 + [ tab8 ] we have already reported polarizabilities of the @xmath191 , @xmath193 and @xmath196 states in our earlier work @xcite .
however , the knowledge of the transition probabilities and lifetimes of the first p - states in the considered states are not known yet .
we use our e1 matrix elements reported in table [ tab6 ] and experimental energies @xcite to evaluate them . in table
[ tab8 ] , we present the transition probabilities of various channels from the @xmath108 and @xmath111 states . as seen from this table , the relative transition probability of the @xmath236 transition is much larger than the @xmath237 transition in the present system compared to ba@xmath0 .
since the wavelength of the @xmath238 transition is one order larger here , the corresponding transition probability is comparatively smaller .
hence , the lifetime of the @xmath108 state becomes little larger than the @xmath107 state in ba@xmath0 .
again , the lifetime of the @xmath111 state becomes smaller than the @xmath110 state in ba@xmath0 due to the fact that the @xmath236 transition probability is much larger .
we present the net transition probabilities and lifetimes of the @xmath108 and @xmath111 states in table [ tab9 ] . in ra@xmath0
, @xmath235 contributes only by 12% to the final lifetime determination of the @xmath111 state in contrast to 28% in ba@xmath0 .
lcc state(@xmath183 ) & @xmath184 & @xmath185 + + & & + @xmath108 & 116.647 & 8.57(12 ) + @xmath111 & 214.235 & 4.67(5 ) + [ tab9 ]
we have employed the relativistic coupled - cluster theory to calculate the electric dipole matrix elements in the singly ionized barium and radium .
these elements are used to determine the light - shift ratios and transition probabilities in the singly ionized barium and comparing with the corresponding experimental data , we have estimated the accuracy of various low - lying matrix elements in this system .
further , we have evaluated the scalar and tensor static dipole polarizabilities and lifetimes in the same system .
we have also calculated the light - shift ratios , transition probabilities and lifetimes using our electric dipole matrix elements in the singly ionized radium .
these data will be helpful in the future experiments in the considered system .
this work is supported by nwo under veni fellowship scheme with project number 680 - 47 - 128 and put of the stichting fom physics program 48 ( tri@xmath239p ) .
we thank norval fortson , jeff sherman , bhanu pratap das and dimtry budker for useful discussions .
24 n. fortson , phys .
lett . * 70 * , 2383 ( 1993 ) .
l. w. wansbeek , b. k. sahoo , r. g. e. timmermans , k. jungmann , b. p. das and d. mukherjee , arxiv:0807.1636 ( unpublished ) .
j. a. sherman , w. trimble , s. metz , w. nagourney and n. fortson , arxiv : physics/0504013 .
b. k. sahoo , phys .
a * 74 * , 020501(r ) ( 2006 ) .
b. k. sahoo , b. p. das , r. k. chaudhuri , d. mukherjee , r. g. e. timmermans , and k. jungmann , phys .
a * 76 * , 040504(r ) ( 2007 ) .
b. k. sahoo , r. chaudhuri , b. p. das and d. mukherjee , phys .
lett . * 96 * , 163003 ( 2006 ) .
w. m. itano , j. res .
technol . * 105 * , 829 ( 2000 ) .
a. kastberg , p. villemoes , a. arnesen , f. heijkenskjld , a. langereis , p. jungner and s. linnaeus , j. opt .
b * 10 * , 1330 ( 1993 ) .
j. reader , c. h. corliss , w. l. wiese and g. a. martin , _ wavelengths and transitions probabilities for atoms and atomic ions _ , natl . bur . stand .
gpo , washington , d.c .
a. gallagher , phys .
* 157 * , 24 ( 1967 ) .
p. kuske , n. kirchner , w. wittmann , h. j. andr and d. kaiser , phys . lett . *
64a * , 377 ( 1978 ) .
e. h. pinnington , r. w. berends and m. lumsden , j. phys .
b * 28 * , 2095 ( 1995 ) .
h. j. andr , _ beam - foil spectroscopy _ , ed . by i. a. sellin and d. j. pegg plenum , new york ,
* 2 * , pg . 825
v. a. dzuba , v. v. flambaum and j. s. m. ginges , phys .
a * 63 * , 062101 ( 2001 ) .
g. gopakumar , h. merlitz , r. k. chaudhuri , b. p. das , u. s. mahapatra and d. mukherjee , phys . rev . * 66 * , 032505 ( 2002 ) . e.
iskrenova - tchoukova and m. s. safronova , phys .
a * 78 * , 012508 ( 2008 ) . c. guet and w. r. johnson , phys .
a * 44 * , 1531 ( 1991 ) .
u. i. safronova , w. r. johnson and m. s. safronova , phys .
a * 76 * , 042504 ( 2007 ) . j. a. sherman , t. w. koerber , a. markhotok , w. nagourney and e. n. fortson , phys .
94 * , 243001 ( 2005 ) .
j. a. sherman , a. andalkar , w. nagourney and e. n. fortson , arxiv:0808.1826 .
j. e. stalnaker , _ progress towards parity nonconservation in atomic ytterbium _ , ph.d .
thesis , university of california , berkeley , usa ( 2005 ) .
j. e. stalnaker , d. budker , s. j. freedman , j. s. guzman , s. m. rochester and v. v. yashchuk , phys .
a * 73 * , 043416 ( 2006 ) .
k. d. bonin and v. v. kresin , _ electric - dipole polarizabilities of atoms , molecules and clusters _ , world scientific , chap . 3 , pg . 36 ( 1997 ) .
b. s. mathur , h. tang and w. happer , phys .
rev . * 171 * , 11 ( 1968 ) . c. cohen - tannoudji and j. dupont - roc , phys .
a * 5 * , 968 ( 1972 ) .
d. mukherjee , r. moitra and a. mukhopadhyay , mol .
physics * 33 * , 955 ( 1977 ) .
i. lindgren , _ a coupled - cluster approach to the many - body perturbation theory for open - shell systems _ , in per - olov lwdin and yngve hrn , editors , atomic , molecular , and solid - state theory , collision phenomena , and computational methods , international journal of quantum chemistry , quantum chemistry symposium * 12 * , pages 33 - 58 , john wiley & sons , march 1978
. b. k. sahoo , s. majumder , r. k. chaudhuri , b. p. das , and d. mukherjee , j. phys .
b * 37 * , 3409 ( 2004 ) .
b. k. sahoo , phys .
a * 73 * , 062501 ( 2006 ) .
b. p. das , _
private communication_. c. e. moore , natl .
stand . ref .
stand . ) , * 3 * ( 1971 ) .
l. w. wansbeek , b. k. sahoo , r. g. e. timmermans , b. p. das and d. mukherjee , phys .
a * 78 * , 012515 ( 2008 ) . b. k. sahoo and b. p. das , phys .
a * 77 * , 062516 ( 2008 )
. e. l. snow and s. r. lundeen , phys .
a * 76 * , 052505 ( 2007 ) t. f. gallagher , r. kachru and n. h. tran , phys .
a * 26 * , 2611 ( 1982 ) .
i. s. lim and p. schwerdtfeger , phys .
rev . a * 70 * , 062501 ( 2004 ) .
i. miadokova , v. kell and a. j. sadlej , theor .
chem . act . * 96 * , 166 ( 1997 ) .
s. h. patil and k. t. tang , j. chem . phys .
* 106 * , 2298 ( 1997 ) . | we have employed the relativistic coupled - cluster theory to calculate the electric dipole matrix elements in the singly ionized barium ( ba@xmath0 ) and radium ( ra@xmath0 ) .
matrix elements of ba@xmath0 are used to determine the light - shift ratios for two different wavelengths on which recent experiments are carried out . by combining the measured light - shift ratios and our calculations ,
we are able to estimate possible errors associate with the used matrix elements in ba@xmath0 .
static polarizabilities of the low - lying states and lifetimes of the first excited p - states are also determined using these matrix elements in ba@xmath0 .
a similar approach has been followed to estimate the lifetimes and light - shift ratios in ra@xmath0
. 1.0 cm |
cluster algebras were introduced in 2002 by fomin and zelevinsky @xcite .
these cluster algebras are commutative rings @xmath1 that depend on the initial cluster variables @xmath2 , cluster coefficients @xmath3 and a quiver @xmath4 .
the combinatorial structure of these rings is based on a recursive operation called mutation . in this paper , we focus on cluster algebras with principal coefficients because a cluster variable in an arbitrary cluster algebra can be computed from a cluster variable in a corresponding cluster algebra with principal coefficients @xcite .
each iteration of a mutation yields a cluster variable in the form of a laurent polynomial in terms of the initial cluster variables @xmath5 .
the coefficients of the terms in this laurent polynomial are monomials in the cluster coefficients . for each cluster variable
there exists an @xmath0-polynomial obtained from the cluster variable by setting each initial cluster variable , @xmath6 , equal to one .
it was proved in @xcite that @xmath0-polynomials are polynomials in the cluster coefficients . for cluster algebras of surface type , there is a bijection between the arcs @xmath7 on the surface and the cluster variables of the cluster algebra @xcite . from an arc @xmath7 in a triangulated surface
, one can construct a labeled snake graph @xcite .
for our purposes , the labeling of the snake graph is unimportant , however each numbered tile @xmath8 corresponds to the principal coefficient with that subscript , @xmath9 .
this snake graph is useful because there is an expansion formula that gives the cluster variable @xmath10 as a sum over the perfect matchings of the snake graph @xcite .
since the cluster variable @xmath10 can be obtained from the perfect matchings , so can the @xmath0-polynomial of @xmath10 .
theorem 4.1 in @xcite tells us there is a bijection between positive continued fractions and snake graphs . in the same paper
the authors give a formula for the cluster variable of a labeled snake graph in terms of a continued fraction of laurent polynomials in @xmath11 .
however , this formula only works for cluster algebras with trivial coefficients . in this paper , we extend this result to cluster algebras with principal coefficients by giving an explicit formula for the @xmath0-polynomial as a continued fraction of laurent polynomials in @xmath12 .
a finite continued fraction @xmath13 : = a_1 + \cfrac{1}{a_{2 } + \cfrac{1}{\ddots + \cfrac{1}{a_n}}}\ ] ] + is said to be positive if each @xmath14 is a positive integer .
we define @xmath15 $ ] by the recursion @xmath15 = a_n n[a_1,\dots , a_{n-1}]+n[a_,\dots , a_{n-2}]$ ] where @xmath16=a_1 $ ] .
it will become useful for us to have the following definition . for a finite , positive continued fraction
@xmath17 $ ] , define @xmath18 for @xmath19 and @xmath20 .
we begin by recalling the definition of snake graphs and some of their properties .
first , a tile @xmath21 is a fixed length square in the plane .
the sides of a tile are orthogonal or parallel to a fixed basis , so it is natural for one to call the lower edge the south edge , etc . each tile can either be placed to the right or above the previous tile .
a snake graph @xmath22 is a connected planar graph given by a sequence of tiles @xmath23 , where @xmath23 and @xmath24 share exactly one edge , @xmath25 .
the edges @xmath25 that are contained in two tiles are called interior edges . for a snake graph made up of tiles @xmath26 ,
the interior edges are @xmath27 .
all of the other edges are called boundary edges .
a subset @xmath28 of edges of @xmath22 where each vertex in @xmath22 is incident to exactly one edge in @xmath28 is called a perfect matching .
see figure [ fig 234minmatch ] for an example of a perfect matching on a snake graph with 8 tiles , where each tile @xmath23 is labeled by @xmath8 .
see @xcite for a more detailed account regarding snake graphs . in @xcite the authors
establish a bijection between snake graphs and positive continued fractions . the snake graph obtained from the continued fraction
@xmath17 $ ] is denoted by @xmath29 $ ] and consists of tiles @xmath30 where @xmath31 .
the snake graph @xmath29 $ ] has @xmath15 $ ] many perfect matchings @xcite .
for example the snake graph @xmath32 $ ] , has @xmath33 = 5 $ ] perfect matchings as seen in figure [ fig poset22 ] .
these perfect matchings form a poset ( * ? ? ?
* section 5 ) with a unique minimal and maximal element .
these minimal and maximal perfect matchings , @xmath34 and @xmath35 respectively , are the only perfect matchings consisting entirely of boundary edges . for consistency
, we adopt the convention that the minimal perfect matching @xmath34 of @xmath29 $ ] is the perfect matching consisting of only boundary edges that also contains the south edge of the first tile . for two examples of posets of perfect matchings see figures [ fig poset22 ] and [ fig poset4 ] in example [ ex poset ] .
we can climb the poset of perfect matchings from @xmath28 to @xmath36 by turning a tile .
a tile @xmath21 can be turned if two of its edges are in the perfect matching @xmath28 . by turning the tile ,
you replace the two edges of @xmath21 in @xmath28 with the other two edges of @xmath21 and in doing so obtain the perfect matching @xmath36 . the height @xmath37 of @xmath28
is defined recursively by @xmath38 and if @xmath36 is above @xmath28 and obtained by turning the tile @xmath23 then @xmath39 .
the @xmath0-polynomial of @xmath22 is defined as @xmath40 , where the sum is over all perfect matchings of @xmath22 .
it was shown in @xcite that if @xmath22 is the snake graph of a cluster variable in a cluster algebra from a surface then the @xmath0-polynomial of that cluster variable is @xmath41 ( or a specialization of @xmath41 if the corresponding arc crosses a self - folded triangle ) .
[ ex poset ] [ fig poset22 ] [ fig poset4 ] in figures [ fig poset22 ] and [ fig poset4 ] below , we give the poset of perfect matchings on the snake graphs @xmath32 $ ] and @xmath42 $ ] respectively , where @xmath34 is at the bottom of the poset and @xmath35 is at the top .
we can use these posets to compute the @xmath0 polynomials , @xmath43 ) = 1+y_1+y_3 + y_1y_3 + y_1y_2y_3 $ ] and @xmath44 ) = 1+y_1+y_1y_2+y_1y_2y_3 $ ] .
the snake graph in figure [ fig 234minmatch ] has @xmath45 = 30 $ ] perfect matchings an the @xmath0-polynomial is given in example [ ex 234 ] using the main result of this paper .
notice that the poset for @xmath42 $ ] is linear , but the poset for @xmath32 $ ] is not .
the snake graph , @xmath42 $ ] is a zigzag snake graph , meaning that the placement of tiles alternates between above and to the right of the previous tile . in general , a continued fraction with a single entry yields a zigzag snake graph , @xmath46 $ ] , which always has a linear poset of perfect matchings . under the correspondence between continued fractions and snake graphs
each @xmath14 in the continued fraction @xmath17 $ ] corresponds to a zigzag subgraph @xmath47 of @xmath29 $ ] .
the subgraph @xmath47 is isomorphic to @xmath46 $ ] , but inherits its tile labels from @xmath29 $ ] , thus consisting of tiles @xmath48 . in figure [ fig subgraph ]
we show the subgraphs @xmath49 of the snake graph @xmath50 $ ] .
notice that @xmath51 is isomorphic to @xmath42 $ ] , shown in figure [ fig poset4 ] .
the minimal matching , @xmath52 , of @xmath47 is the completion of the matching it inherits from the minimal matching of @xmath29 $ ] . technically speaking ,
if @xmath34 is the minimal perfect matching of @xmath29 $ ] then @xmath52 is given by first collecting all edges in @xmath34 that are also in @xmath47 , denoted by @xmath53 .
then , if this is not already a perfect matching , we must include exactly one other edge as follows .
@xmath54 [ ex 234minmatch ] in figure [ fig 234minmatch ] , the minimal matching @xmath34 of @xmath50 $ ] is shown .
the snake graph of @xmath55 $ ] has three subgraphs , @xmath56 and @xmath51 . from the minimal perfect matching of @xmath29 $ ] in figure [ fig 234minmatch ]
we obtain @xmath53 .
notice that in figure [ fig subgraph ] we show these subgraphs with their completed minimal perfect matchings , @xmath57 .
[ fig 234minmatch ] [ fig subgraph ] in @xcite identities in the cluster algebra have been expressed in terms of snake graphs .
equation ( [ f eqn ] ) below follows from the grafting with a single edge formula from theorem 7.3 of @xcite , where @xmath58 and the grafting takes place at tile @xmath59 .
@xmath60 ) & = & y_{34}f({\mathcal{g}}[a_1,\dots , a_{n-1}])f({\mathcal{h}}_n ) + y_{56 } f({\mathcal{g}}[a_1,\dots , a_{n-2 } ] ) \end{aligned}\ ] ] we define the variables @xmath61 and @xmath62 as follows , where @xmath63 .
+ @xmath64 @xmath65 for the case when @xmath66 , we define the snake graph of an empty continued fraction to be a single edge .
there is only one perfect matching of a single edge , which corresponds to the @xmath0-polynomial 1 .
therefore equation ( [ f eqn ] ) for the case when @xmath66 is as follows .
@xmath67 ) & = & f({\mathcal{g}}[a_1])f({\mathcal{h}}_2 ) + \prod_{j=0}^{(\ell_2)-1 } y_j \end{aligned}\ ] ] equations ( [ f eqn ] ) , ( [ yeqn ] ) and ( [ f eqn2 ] ) will be used in the proof of our main theorem .
together they give an equation for the @xmath0-polynomial of a snake graph @xmath68 $ ] based on the @xmath0-polynomials of certain subgraphs .
[ defn correction ] for any continued fraction @xmath17 $ ] with @xmath69 , we define an associated continued fraction of laurent polynomials @xmath70 $ ] , where each @xmath71 and + again consider the continued fraction @xmath55 $ ] from the previous example [ ex 234minmatch ] . here
we consider the zigzag snake graphs @xmath47 that are subgraphs of @xmath29 $ ] .
for @xmath74 , the minimal matching shown in figure [ fig subgraph ] corresponds to the constant term of the @xmath0-polynomial , 1 , and the maximal matching we obtain by turning tile 1 corresponds to the term @xmath75 in the @xmath0-polynomial . for @xmath76
, we have the minimal matching associated to 1 , the matching obtained from turning tile 4 associated to @xmath77 and the maximal matching obtained from turning tile 4 and then tile 3 associated to @xmath78 . together these form the @xmath0-polynomial @xmath79 .
the @xmath0-polynomial of each subgraph @xmath56 and @xmath51 is given below .
let @xmath8 be odd .
the tiles @xmath48 make up @xmath47 .
the subgraph @xmath47 is the zigzag snake graph @xmath46 $ ] with the inherited labeling and completed minimal perfect matching @xmath52 discussed previously . in this completion
, the first tile of the subgraph @xmath85 can be turned immediately and is the only such tile .
if we turn tile @xmath85 , we can then turn the next tile @xmath86 and so on .
therefore @xmath87 let @xmath8 be even .
in this case , the very last tile of @xmath47 , @xmath88 has two edges in the minimal matching of the completion , @xmath52 , and can be turned . therefore in order to determine the @xmath0-polynomial of @xmath47 , we must first turn the last tile and work our way down the snake graph .
@xmath89 [ main thm ] the f - polynomial associated to the snake graph of the continued fraction @xmath17 $ ] denoted by @xmath90)$ ] is given by the equation : + @xmath91 ) = \left\ { \begin{array}{ll } n[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_n ] & \text{if } n \text { is odd},\\[.7 cm ] c_n^{-1 } { n[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_n ] } & \text{if } n \text { is even,}\ \end{array } \right.\ ] ] + where @xmath92 $ ] is defined by the recursion @xmath92 = { \mathcal{l}}_n n[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_{n-1 } ] + n[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_{n-2 } ] $ ] and @xmath93 = { \mathcal{l}}_1 $ ]
. + proof by induction .
let @xmath94 . in the case where @xmath95 is odd , by equation ( [ yeqn ] ) , we have @xmath96 and @xmath97 .
it is clear that @xmath98 ) = f({\mathcal{h}}_1)$ ] simply because @xmath99 $ ] and @xmath74 are the same snake graph .
then , by lemma [ lemma ] we have @xmath100 .
note that @xmath101 because it is an empty product .
therefore @xmath102 $ ] .
thus we have shown that @xmath98 ) = n[{\mathcal{l}}_1]$ ] .
+ let @xmath66 .
in this case we use equation ( [ f eqn2 ] ) and note that @xmath103 .
using lemma [ lemma ] and the case @xmath94 , we see that the right hand side is equal to @xmath105+c_2^{-1}$ ] and this is equal to the following , where the second equation holds by definition [ defn correction ] and the last equation holds by the definition of @xmath106 $ ] .
@xmath107 ) & = & c_2^{-1}\left(c_2\varphi_2n[{\mathcal{l}}_1 ] + 1\right)\\[.5em ] & = & c_2^{-1}\left({\mathcal{l}}_2n[{\mathcal{l}}_1 ] + 1\right)\\[.5em ] & = & c_2^{-1}n[{\mathcal{l}}_1,{\mathcal{l}}_2]\\ \end{aligned}\ ] ] now let @xmath108 be odd . assume that for all @xmath109 our statement holds .
in this situation by equation ( [ yeqn ] ) , @xmath96 and @xmath97 . additionally , we know from equation ( [ f eqn ] ) that the f - polynomial of @xmath29 $ ] is given by the following .
@xmath91 ) = y_{\ell_{n-1}}c_{n-1}^{-1}{n[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_{n-1 } ] } f({\mathcal{h}}_n ) + { n[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_{n-2}]}\ ] ] .
+ from here we can apply lemma [ lemma ] .
@xmath91 ) = y_{\ell_{n-1 } } c_{n-1}^{-1}{n[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_{n-1 } ] } \varphi_n + { n[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_{n-2}]}\ ] ] + using the fact that @xmath110 and @xmath111 we obtain our desired result as follows .
+ @xmath112 ) & = & c_n\varphi_nn[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_{n-1 } ] + n[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_{n-2}]\\ [ .5em ] & = & { \mathcal{l}}_nn[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_{n-1 } ] + n[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_{n-2}]\\ [ .5em ] & = & n[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_n ] \\ \end{aligned}\ ] ] in the case where @xmath108 is even , our argument is very similar .
assume that for all @xmath109 our statement holds .
in this case , @xmath113 and @xmath114 .
again , we make the corresponding replacements based on our induction hypothesis .
@xmath112 ) & = & n[{\mathcal{l}}_1,\dots,{\mathcal{l}}_{n-1}]f({\mathcal{h}}_n ) + \left(\prod_{j=\ell_{n-2}}^{(\ell_n)-1 } y_j \right)c_{n-2}^{-1}n[{\mathcal{l}}_1,\dots,{\mathcal{l}}_{n-2 } ] ) \\ [ .3 cm ] \end{aligned}\ ] ] then we apply lemma [ lemma ] and the rest follows similarly to the previous case .
@xmath112 ) & = & n[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_{n-1 } ] \varphi_n + \left(\prod_{j=\ell_{n-2}}^{(\ell_n)-1 } y_j\right)c_{n-2}^{-1}n[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_{n-2}]\\ [ .6em ] & = & n[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_{n-1 } ] \varphi_n + c_n^{-1}n[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_{n-2}]\\ [ .5em ] & = & c_n^{-1 } \left(\varphi_n c_nn[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_{n-1 } ] + n[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_{n-2}]\right)\\ [ .5em ] & = & c_n^{-1 } \left({\mathcal{l}}_nn[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_{n-1 } ] + n[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_{n-2}]\right)\\ [ .5em ] & = & c_n^{-1}n[{\mathcal{l}}_1,{\mathcal{l}}_2,\dots,{\mathcal{l}}_{n } ] \end{aligned}\ ] ] consider the continued fraction @xmath115 = \frac{67}{29}$ ] , then the @xmath0-polynomial has 67 terms . according to theorem [ main thm ] , since the continued fraction has an even number of entries , the f - polynomial of the snake graph @xmath116 $ ] is given by @xmath117 $ ] .
@xmath118 { \mathcal{l}}_2 & = & \dfrac{1+y_4+y_3y_4}{y_1y_2y_3y_4 } & \quad & { \mathcal{l}}_4 & = & \dfrac{1+y_{10}}{y_1y_2y_3y_4y_5y_6y_7y_8y_9y_{10 } } \end{array}\ ] ] @xmath119 ) & = & c^{-1}_nn[{\mathcal{l}}_1,{\mathcal{l}}_2,{\mathcal{l}}_3,{\mathcal{l}}_4]\\ [ 1em ] & = & \left(\prod_{j=1}^{(\ell_4)-1}y_j\right)({\mathcal{l}}_1{\mathcal{l}}_2{\mathcal{l}}_3{\mathcal{l}}_4 + { \mathcal{l}}_1{\mathcal{l}}_2+{\mathcal{l}}_1{\mathcal{l}}_4+{\mathcal{l}}_3{\mathcal{l}}_4 + 1)\\ [ 1em ] & = & ( 1+y_1)(1+y_4+y_3y_4)(1+y_6+y_6y_7+y_6y_7y_8)(1+y_{10})y_5 \nonumber \\ & & + \ : ( 1+y_1)(1+y_4+y_3y_4)y_5y_6y_7y_8y_9y_{10 } + ( 1+y_1)(1+y_{10})\nonumber \\ & & + \ : ( 1+y_6+y_6y_7+y_6y_7y_8)(1+y_{10})y_1y_2y_3y_4y_5 + y_1y_2y_3y_4y_5y_6y_7y_8y_9y_{10 } \\ \end{aligned}\ ] ] [ ex 234 ] consider the continued fraction @xmath55 $ ] .
since @xmath55 = \frac{30}{13}$ ] , the @xmath0-polynomial has 30 terms . according to theorem [ main thm ] ,
the f - polynomial of the snake graph @xmath50 $ ] is given by @xmath120 $ ] .
+ @xmath121)= & = & n[{\mathcal{l}}_1,{\mathcal{l}}_2,{\mathcal{l}}_3 ] \\ [ .5em ] & = & { \mathcal{l}}_1{\mathcal{l}}_2{\mathcal{l}}_3 + { \mathcal{l}}_1+{\mathcal{l}}_3\\ [ .5em ] & = & ( 1+y_1)(1+y_4+y_3y_4)(1+y_6+y_6y_7+y_6y_7y_8)y_5 + ( 1+y_1 ) \\ & & + \ : ( 1+y_6+y_6y_7+y_6y_7y_8)y_1y_2y_3y_4y_5\\ \end{aligned}\ ] ] s. fomin and a. zelevinsky , cluster algebras
i : foundations , _
j. amer . math .
soc . _ * 15 ( 2002 ) , 497529 .
s. fomin and a. zelevinsky , cluster algebras iv : coefficients , _ compos .
_ * 143 ( 2007 ) , 112164 . * | for cluster algebras from surfaces , there is a known formula for cluster variables and @xmath0-polynomials in terms of the perfect matchings of snake graphs .
if the cluster algebra has trivial coefficients , there is also a known formula for cluster variables in terms of continued fractions . in this paper
, we extend this result to cluster algebras with principal coefficients by producing a formula for the @xmath0-polynomials in terms of continued fractions . |
observation driven time - series models , introduced by @xcite , has a wide variety of real applications , including econometrics ( garch models ) and applied mathematics ( inferring initial conditions and parameters of ordinary differential equations ) .
the model can be described as follows .
we observe @xmath9 , @xmath10 which are associated to a dynamic system @xmath11 , @xmath12 which is potentially unknown .
define the process @xmath13 ( with @xmath14 some arbitrary point on @xmath15 ) on a probability space @xmath16 , where , for every @xmath17 , @xmath18 is a probability measure .
denote by @xmath19 .
the model is defined as , for @xmath20 @xmath21 where @xmath22 $ ] , @xmath23 and for every @xmath24 , @xmath25 ( the probabilities on @xmath26 ) . throughout
, we assume that for any @xmath27 @xmath28 admits a density w.r.t .
some @xmath29finite measure @xmath30 , which we denote as @xmath31 .
next , we define a prior probability distribution @xmath32 on @xmath33 , with lebesgue density @xmath34 . thus , given @xmath1 observations @xmath35 the object of inference is the posterior distribution on @xmath36 : @xmath37 where we have used the notation @xmath38 and @xmath39 is lebesgue measure . in most applications of practical interest
, one can not compute the posterior point - wise and has to resort to numerical methods , such as mcmc , to draw inference on @xmath40 and/or @xmath41 . in this article , we are not only interested in inferring the posterior distribution , but the scenario for which @xmath31 can not be evaluated point - wise , nor do we have access to an unbiased estimate of it ( it is assumed we can simulate from the associated distribution ) .
in such a case , it is not possible to draw inference from the true posterior , even using numerical techniques .
the common response in bayesian statistics , is now to adopt an approximation of the posterior using the notion of approximate bayesian computation ( abc ) ; see @xcite for a recent overview .
abc approximations of posteriors are based upon defining a probability distribution on an extended state - space , with the additional random variables lying on the data - space and usually distributed according the true likelihood .
the closeness of the abc posterior distribution is controlled by a tolerance parameter @xmath42 and often the approximation is exact as @xmath43 . in this paper , we introduce a new abc approximation of observation driven time - series models , which is closely associated to that developed in @xcite for hidden markov models ( hmms ) and later for static parameter inference from hmms @xcite .
this latter abc approximation is particularly well behaved and a noisy variant ( which pertrubs the data ; see e.g. @xcite ) is shown under some assumptions to provide maximum - likelihood estimators ( mle ) which asympotically in @xmath1 are the true parameters .
the new abc approximation that we develop is studied from a theoretical perspective .
relying on the recent work of @xcite we show that , under some conditions , as @xmath0 , with @xmath1 the length of the time series , the abc posterior has , almost surely , a map estimator of @xmath40 which is different from the true parameter @xmath44 say .
however , a noisy abc map of @xmath40 asymptotically converges to the true parameter , almost surely .
these results establish that the particular approximation adopted is reasonably sensible .
the other main contribution of this article is a development of a new mcmc algorithm designed to sample from the abc approximation of the posterior . due to the nature of the abc approximation
it is easily seen that standard mcmc algorithms ( e.g. @xcite ) will have an acceptance probability that will fall at an exponential rate in @xmath1 .
in addition , more advanced ideas such as those based upon the ` pseudo marginal ' @xcite , have recently been shown to perform rather poorly in theory ; see @xcite .
these latter algorithms are based upon exact approximations of marginal algorithms @xcite , which in our context is just sampling @xmath45 .
we develop an mcmc kernel , related to recent work in @xcite , which is designed to have a random running time per - iteration , with the idea of improving the exploration ability of the markov chain .
we show that the expected cost per iteration of the algorithm , under some assumptions and for reasonable performance , is @xmath2 , which compares favourably with competing algorithms .
we also show , empirically , that this new mcmc method out - performs standard pseudo marginal algorithms .
this paper is structured as follows . in section [ sec : model ]
we introduce our abc approximation and give our theoretical results on the map estimator . in section [ sec : comp ] , we give our new mcmc algorithm , along with some theoretical discussion about its computational cost and stability . in section [ sec : examples ] our approximation and mcmc algorithm is illustrated on toy and real examples . in section [ sec : summ ] we conclude the article with some discussion of future work .
the proofs of our theoretical results are given in the appendix .
as it was emphasised in section [ sec : intro ] , we are interested in performing inference when @xmath31 can not be evaluated point - wise , nor do we have access to an unbiased estimate of it .
we will instead assume it is possible to sample from @xmath46 .
in such scenaria , one can not use standard simulation based methods .
for example , in a standard mcmc approach the metropolis - hastings acceptance ratio can not be evaluated , even though it may be well - defined .
following the work in @xcite for hidden markov models , we introduce an abc approximation for the density of the posterior in as follows : @xmath47 with @xmath42 and @xmath48 where we denote @xmath49 as the open ball centred at @xmath50 with radius @xmath51 and write @xmath52 .
when @xmath30 is the lebesgue measure , @xmath53 corresponds to the volume of the ball @xmath49 . in general we will refer to abc as the procedure of performing inference for the posterior in .
in addition , we will call _ noisy abc _ the inference procedure that uses instead of the original observation sequence a perturbed one , namely @xmath54 , where each @xmath55 is given by @xmath56 with each @xmath57 is identically independently distributed ( i.i.d . ) uniformly on @xmath58 ( shorthand @xmath59 ) . in this section
we will investigate some interesting properties of the abc posterior in .
in particular , we will look at the asymptotic behaviour with @xmath1 of the resulting map estimators for @xmath40 .
the properties of the map estimator reveal information about the mode of the posterior distribution as we obtain increasingly more data . to simplify the analysis in this section
we will assume that : [ as:1 - 1 ] * @xmath60 is fixed and known , i.e. @xmath61 , where @xmath62 denotes the dirac delta measure on @xmath26 and @xmath63 is known .
* @xmath64 is bounded and positive everywhere in @xmath65 . *
the observations actually originated from the true model model for some @xmath66 , i.e. we look at a well - specified problem .
* @xmath67 and @xmath68 do not depend upon @xmath40 .
thus we have the following model recursions for the true model : @xmath69 where we will denote associated expectations to @xmath70 as @xmath71 .
in addition , for this section we will introduce some extra notations : @xmath72 is a compact , complete and separable metric space and @xmath73 is a compact metric space , with @xmath74 . for two measures @xmath75 and @xmath76 of bounded variation denote the convolution @xmath77 .
let also @xmath78 be the probability law associated to the random sequence @xmath79 , where each @xmath57 is an i.i.d .
sample from the uniform distribution defined on @xmath58 .
we proceed with some additional technical assumptions : [ as:1 ] @xmath80 is a stationary stochastic process , with @xmath81 strict sense stationary and ergodic , following .
[ as:2 ] for every @xmath82 , @xmath83 is continuous .
in addition , there exist @xmath84 such that for any @xmath85 , @xmath86
. finally @xmath87 , for every @xmath82 .
[ as:3 ] there exist a measurable @xmath88 , such that for every @xmath89 @xmath90 [ as:4 ] the following statements hold : 1 .
@xmath91 if and only if @xmath92 2 . if @xmath93 holds @xmath94a.s .
, then @xmath95 .
assumptions ( a[as:1]-[as:4 ] ) and the compactness of @xmath65 are standard assumptions for maximum likelihood estimation ( ml ) and they can be used to show the uniqueness of the maximum likelihood estimator ( mle ) ; see @xcite for more details .
therefore , if the prior @xmath96 is bounded and positive everywhere on @xmath65 it is a simple corollary that the map estimator will correspond to the mle .
in the remaining part of this section we will adapt the analysis in @xcite for mle to the abc setup .
in particular , we are to estimate @xmath40 using the log - likelihood function : @xmath97 we define the abc - mle for an @xmath1-long sequence as @xmath98 we proceed with the following proposition : [ prop1 ] assume ( a[as:1 - 1]-[as:3 ] ) .
then for every @xmath63 and fixed @xmath42 @xmath99 where @xmath100 $ ] .
the result establishes that the estimate will converge to a point , which is typically different to the true parameter .
hence there is an intrinsic asymptotic bias for the plain abc procedure . to correct this bias , consider the noisy abc procedure , of replacing the observations by @xmath101 where @xmath102 .
the noisy abc mle estimator is then : @xmath103 we have the following result : [ prop2 ] assume ( a[as:1 - 1]-[as:4 ] ) .
then for every @xmath63 and fixed @xmath42 @xmath104 the result shows that the noisy abc mle estimator is asymptotically unbiased .
therefore , given that in our setup the abc map estimator corresponds to the abc mle we can conclude that the mode of the posterior distribution as we obtain increasingly more data is converging towards the true parameter .
finally we note that our assumptions indeed pose some restrictions , but these are shown to be realistic for a few interesting models in @xcite .
in addition , the main purpose of this result is to motivate the use of the approximate posterior in when the observation sequence is long or its marginal likelihood is quite informative .
recall that we formulated in the abc posterior written in .
one can rewrite the approximate posterior in : @xmath105 with @xmath106 note we have just used fubini s theorem to rewritte the likehood @xmath107 as an integral of a product instead of a product of integrals @xmath108 shown in - . in this paper
we will focus only on mcmc algorithms and in particular on the metropolis - hastings ( m - h ) approach . in order to sample from the posterior
@xmath109 one runs an ergodic markov chain with the invariant density being @xmath109 .
then after a few iterations when the chain has reached stationarity , one can treat the samples from the chain as approximate samples from @xmath109 .
this is shown in algorithm [ alg : mcmc ] , where for convenience we denote @xmath110 .
the one - step transition kernel of the mcmc chain is usually described as the _ m - h kernel _ and follows from step 2 in algorithm [ alg : mcmc ] . 1 . * ( initialisation ) * at @xmath111 sample @xmath112 .
2 . * ( m - h kernel ) * for @xmath113 : * sample @xmath114 from a proposal @xmath115 with density @xmath116 . *
accept the proposed state and set @xmath117 with probability @xmath118 otherwise set @xmath119 .
set @xmath120 and return to the start of 2 .
unfortunately @xmath107 is not available analytically and can not be evaluated , so this rules out the possibility of using of traditional mcmc approaches like algorithm [ alg : mcmc ] .
however , one can resort to the so called _ pseudo - marginal _ approach whereby unbiased estimates of @xmath107 are used instead within an mcmc algorithm .
we will refer to this algorithm as abc - mcmc .
the resulting algorithm can be posed as one targeting a posterior defined on an extended state space , so that its marginal coincides with @xmath121 .
we will use these ideas to present abc - mcmc as a m - h algorithm which is an exact approximation to an appropriate marginal algorithm .
to illustrate an example of these ideas , we proceed by writing a posterior on an extended state - space @xmath122 as follows : @xmath123 it is clear that is the marginal of and hence the similarity in the notation .
as we will show later in this section , extending the target space in the posterior as in is not the only and certainly not the best choice .
we emphasise that the only essential requirement for each choice is that the marginal of the extended target is @xmath121 , but one should be cautious because the particular choice will affect the mixing properties and the efficiency of the mcmc scheme that will be used to sample from @xmath124 in or another variant .
we will now look at two basic different choices for extending the abc posterior while keeping the marginal fixed to @xmath121 . in the remainder of the paper
we will denote @xmath110 as we did in algorithm [ alg : mcmc ] . initially consider the abc approximation when be extended to the space @xmath122 : @xmath125 recall one can not evaluate @xmath126 and is only able to simulate from it . in algorithm
we present a natural m - h proposal that could be used to sample from @xmath127 instead of the one shown step 2 at algorithm [ alg : mcmc ] .
note that this time the state of the mcmc chain is composed of @xmath128 . here
each @xmath129 assumes the role of an auxiliary variable to be eventually integrated out at the end of the mcmc procedure .
* sample @xmath130 from a proposal @xmath131 with density @xmath132 . *
sample @xmath133 from a distribution with joint density @xmath134 * accept the proposed state @xmath135 with probability : @xmath136 however , as @xmath1 increases , the m - h kernel in algorithm [ alg : simple ] will have an acceptance probability that falls quickly with @xmath1 .
in particular , for any fixed @xmath137 , the probability of obtaining such a sample will fall at an exponential rate in @xmath1 .
this means that this basic abc mcmc approach will be inefficient for a moderate value of @xmath1 .
this issue can be dealt with by using @xmath138 multiple trials , so that at each @xmath139 , some auxiliary variables ( or pseudo - observations ) are in the ball @xmath140 .
this idea originates from @xcite and in fact augments the posterior to a larger state - space , @xmath141 , in order to target the following density : @xmath142 again , it is easy to show that the marginal of interest @xmath143 is preserved , i.e. @xmath144 in algorithm [ alg : ntry ] we present an m - h kernel with invariant density @xmath145 .
the state of the mcmc chain now is @xmath146 .
we remark that as @xmath138 grows , one expects to recover the properties of the ideal m - h algorithm in algorithm [ alg : mcmc ] .
nevertheless , it has been shown in @xcite that even the m - h kernel in algorithm [ alg : ntry ] does not always perform well .
it can happen that the chain gets often stuck in regions of the state - space @xmath147 where @xmath148 is small . * sample @xmath130 from a proposal @xmath131 with density @xmath132 .
* sample @xmath149 from a distribution with joint density @xmath150 . *
accept the proposed state @xmath151 with probability : @xmath152 we will address this shortfall detailed above , by proposing an alternative augmented target and corresponding m - h kernel .
the basic idea is that a random number of trials is used based on the value of @xmath153
. then it will be possible to use more computational effort when the chain is at regions where @xmath153 is low .
consider an alternative extended target , for @xmath154 , @xmath155 , @xmath156 : @xmath157 standard results for negative binomial distrubutions ( see @xcite for more details ) imply that @xmath158 holds and this can be used to deduce that @xmath159 is an unbiased estimator for @xmath160 .
in addition , from it follows that the marginal w.r.t .
@xmath137 is the one of interest : @xmath161 in algorithm [ alg : new ] we present a m - h kernel with invariant density @xmath162 .
the state of the mcmc chain this time is @xmath163 .
* sample @xmath130 from a proposal @xmath131 with density @xmath132 . *
for @xmath164 repeat the following : sample @xmath165 with probability density @xmath166 until there are @xmath138 samples lying in @xmath140 ; the number of samples to achieve this ( including the successful trial ) is @xmath167 . *
accept @xmath168 with probability : @xmath169 the potential benefit of this kernel is that one expects the probability of accepting a proposal is higher than the previous m - h kernel ( for a given @xmath138 ) .
this comes at a computational cost which is both increased and random .
the proposed kernel is based on the _ @xmath170hit _ kernel of @xcite , which has been adapted here to account for the data being a sequence of observations resulting from a time series .
finally , it is important to mention that in algorithms [ alg : ntry ] and [ alg : new ] generating multiple trials can be implemented very efficiently in parallel using appropriate computing hardware , such as multi - core processors or computing clusters . to implement the proposed kernel , one needs to select @xmath138 . in practice to it is difficult to know a good value this _ a priori _ , so we present a theoretical result that can add some intuition on choosing @xmath138 . let @xmath171 $ ] denote expectation w.r.t .
@xmath172 given @xmath173 .
we will also pose the assumption : [ ass : a_k_positive ] for any fixed @xmath42 , @xmath174 , we have @xmath175 . the the following result holds , whose proof can be found in the appendix . : [ prop : rel_var ] assume ( a[ass : a_k_positive ] ) and let @xmath176 , @xmath177 and @xmath178 . then for fixed @xmath179 we have @xmath180\leq\frac{cn}{n}\ ] ] where @xmath181 .
the result shows that one should set @xmath182 for the relative variance not to grow with @xmath1 , which is unsuprising , given the conditional independence structure of the @xmath183 . to get a better handle on the variance , suppose @xmath184 ,
then for @xmath137 fixed @xmath185 = \frac{\alpha_1(y_{1},\epsilon,\gamma)(1-\alpha_1(y_{1},\epsilon,\gamma))}{n}.\ ] ] for the new approach one can show @xmath186 \leq \frac{\alpha_1(y_{1},\epsilon,\gamma)^2}{(n-2)}.\ ] ] not taking into account the computational cost , one prefers this new estimate with regards to variance if @xmath187 which is likely to occur if @xmath188 is not too large ( recall we want @xmath51 to be small , so that we have a good approximation of the true posterior ) and @xmath138 is moderate - this is precisely the scenario in practice . [ rem : gamma_indep ] it is easily shown that the relative variance associated to the estimate @xmath189 $ ] is @xmath190 - 1.\ ] ] note this quantity is not uniformly upper - bounded in @xmath137 unless @xmath191 , which may not occur .
conversely , proposition [ prop : rel_var ] shows that the relative variance of the new estimator is uniformly upper - bounded in @xmath137 under minimal conditions .
we suspect that this means in practice that the kernel with random number of trials may mix faster .
as the cost per - iteration is random , we will investigate this further .
we denote the proposal of @xmath192 as @xmath193 .
let @xmath194 be the initial distribution of the mcmc chain and @xmath195 the distribution of the state at time @xmath196 .
in addition , denote by @xmath197 the proposed state for @xmath198 at iteration @xmath196 .
finally , we will write as @xmath199 the expectation of a random variable proposed by @xmath193 given the simulated state at time @xmath196 .
we will assume that the observations are fixed and known .
then we have the following result : [ prop : time ] let @xmath42 , and suppose that there exists a constant @xmath200 such that for any @xmath177 we have @xmath201 , @xmath202a.e .. then it holds for any @xmath154 , @xmath113 , that : @xmath203\leq\frac{nn}{c}.\ ] ] the expected computational cost grows linearly with @xmath1 .
thus , coupled with the result in proposition [ prop : rel_var ] , one has a cost of @xmath2 per - iteration , which is comparable to many exact approximations of mcmc algorithms ( e.g. @xcite ) , albeit in a much simpler situation .
note also that the kernel in algorithm [ alg : ntry ] is expected to require a cost of @xmath204 per iteration for reasonable performance , although this cost here is deterministic . as mentioned above
, one expects the approach with random number of trials to work better with regards to the mixing time , especially when the values of @xmath153 are not large .
we attribute this to algorithm [ alg : new ] providing a more ` targetted ' way to use the simulated auxiliary variables .
this will be illustrated numerically in section [ sec : examples ] .
a comparison of our results with the interesting work in @xcite seems relevant .
there the authors deal with a more general context and show that we should choose @xmath138 as a particular asymptotic ( in @xmath138 ) variance ; the main point is that the ( asymptotic ) variance of the estimate of @xmath160 should be the same for each @xmath137 .
we conjecture that in our set - up one should choose @xmath138 such that the actual variance of the estimate of @xmath160 is constant with respect to @xmath137 . in this scenario , on inspection of the proof of proposition [ prop : rel_var ] , for a given @xmath137 , one should set @xmath138 to be the solution of @xmath205 for some desired ( upper - bound on the ) variance @xmath206 ( whose optimal value would need to be obtained ) .
this makes @xmath138 a random variable , in addition , but does not change the simulation mechanism .
unfortunately , one can not do this in practice , as the @xmath207 are unknown . taking into account remark [ rem : gamma_indep ] , this latter approach may not be so much of a concern in practice .
we conclude this discussion by adding a related comment regarding the ergodicity of the proposed mcmc kernel .
if there exists a constant @xmath208 such that @xmath209 and the marginal mcmc kernel in algorithm [ alg : mcmc ] is geometrically ergodic , then by ( * ? ?
* propositions 7 , 9 ) the mcmc kernel of algorithm [ alg : new ] is also geometrically ergodic .
for this example let each @xmath210 be a scalar real random variable and consider the model : @xmath211 with @xmath212 and @xmath213 , where we denote @xmath214 the zero mean normal distribution with variance @xmath215 .
the prior on @xmath40 is @xmath216 .
this model is usually referred to as the standard normal means model in one dimension and the posterior is given by : @xmath217 where @xmath218 note that if @xmath219 , then the posterior on @xmath40 is consistent and concentrates around @xmath220 as @xmath0 .
the abc approximation after marginalizing out the auxiliary variables has a likelihood given by : @xmath221\ ] ] where @xmath222 is the standard normal cumulative density function .
thus , this is a scenario where we can perform the marginal mcmc .
three data sets are generated from the model with @xmath223 and @xmath224 .
in addition , for @xmath225 we perturb the data - sets in order to use them for noisy abc . for the sake of comparison
, we also generate a noisy abc data - set for @xmath226 .
we will also use a prior with @xmath227 .
we run the new mcmc kernel ( the proposal in algorithm [ alg : new ] - we will frequently use the expression ` algorithm ' to mean an mcmc kernel with the given proposal mechansim of the algorithm ) , old mcmc kernel ( algorithm [ alg : ntry ] ) and a marginal mcmc algorithm which just samples on the parameter space @xmath228 ( i.e. the posterior density is proportional to @xmath229 ) .
each algorithm is run with a normal random walk proposal on the parameter space , with the same scaling .
the scaling chosen yields an acceptance rate of around 0.25 for each run of the marginal mcmc algorithm .
the new mcmc kernel is run with @xmath230 and the old with a slightly higher value of @xmath138 so that the computational times are about the same ( so for example , the running time of the new kernel is not a problem in this example ) .
the algorithms are run for 10000 iterations and the results can be found in figures [ fig : marginal_den_normal]-[fig : marginal_acf_normal ] . in figure
[ fig : marginal_den_normal ] the density plots for the posterior samples on @xmath40 , from the marginal mcmc can be seen for @xmath231 and each value of @xmath1 .
when @xmath225 , we can observe that both abc and noisy abc both get closer to the true posterior as @xmath1 grows .
for noisy abc , this is the behavior that is predicted in section [ sec : consis ] . for the abc approximation ,
following the proof of theorem 1 in @xcite , one can see that the bias falls with @xmath51 ; hence , in this scenario there is not a substantial bias for the standard abc approximation .
when we make @xmath51 much larger a more pronounced difference between abc and noisy abc can be seen and it appears as @xmath1 grows that the noisy abc approximation is slightly more accurate ( relative to abc ) .
we now consider the similarity of the new and old mcmc kernels to the marginal algorithm ( i.e. the kernel both procedures attempt to approximate ) , the results are in figures [ fig : marginal_den_oldnew_normal]-[fig : marginal_acf_normal ] .
with regards to both the density plots ( figure [ fig : marginal_den_oldnew_normal ] ) and auto - correlations ( figure [ fig : marginal_acf_normal ] ) we can see that both mcmc kernels appear to be quite similar to the marginal mcmc . it is also noted that the acceptance rates of these latter kernels are also not far from that of the marginal algorithm ( results not shown ) .
these results are unsuprising , given the simplicity of the density that we target , but still reassuring ; a more comprehensive comparison is given in the next example .
encouragingly , the new and old mcmc kernels do not seem to noticably worsen as @xmath1 grows ; this shows that , at least for this example , the recommendation of @xmath182 is quite useful .
we remark that whilst these results are for a single batch of data , the results are consistent with other data sets .
set , for @xmath232 @xmath233 where @xmath234 ( i.e. a stable distribution , with location 0 , scale @xmath235 and asymmetry and skewness parameters @xmath236 ) .
we set @xmath237 where @xmath238 is a gamma distribution with mean @xmath239 and @xmath240 .
this is a garch(1,1 ) model with an intractable likelihood .
we consider daily log - returns data from the s&p 500 index from 03/1/11 to 14/02/13 , which constitutes 533 data - points . in the priors
, we set @xmath241 and @xmath242 , which are not overly informative .
in addition , @xmath243 and @xmath244 .
we consider @xmath245 and only a noisy abc approximation of the model .
algorithms [ alg : ntry ] and [ alg : new ] are to be compared .
the mcmc proposals on the parameters are random - walks on the log - scale and for both algorithms we set @xmath246 .
it should be noted that our results are fairly robust to changes in @xmath247 $ ] , which are the values we tested the algorithm with . in figure
[ fig : tracee05 ] we present the trace - plot of 50000 iterations of both mcmc kernels when @xmath248 .
algorithm [ alg : ntry ] took about 0.30 seconds per iteration and algorithm [ alg : new ] took about 1.12 seconds per iteration we modified the proposal variances to yield an acceptance rate around 0.3 .
the plot shows that both algorithms appear to move across the state - space in a very reasonable way .
the new algorithm takes much longer and in this situation does not appear to be required .
this run is one of many we performed and we observed this behaviour in many of our runs . in figure [ fig : tracee001 ] we can observe the trace plots from a particular ( typical ) run when @xmath249 . in this case , both algorithms are run for 200000 iterations .
algorithm [ alg : ntry ] took about 0.28 seconds per iteration and algorithm [ alg : new ] took about 2.06 seconds per iteration ; this issue is discussed below . in this scenario , considerable effort was expended for algorithm [ alg : ntry ] to yield an acceptance rate around 0.3 , but despite this , we were unable to make the algorithm traverse the state - space .
in contrast , with less effort , algorithm [ alg : new ] appears to perform quite well and move around the parameter space ( the acceptance rate was around 0.15 versus 0.01 for algorithm [ alg : ntry ] ) .
whilst the computational time for algorithm [ alg : new ] is considerably more than algorithm [ alg : ntry ] , in the same amount of computation time , it still moves more around the state space ; algorithm runs of the same length are provided for presentational purposes .
we remark that , whilst we do not claim that it is ` impossible ' to make algorithm [ alg : ntry ] mix well in this example , we were unable to do so and , alternatively , for algorithm [ alg : new ] we expended considerably less effort for very reasonable performance .
this example is typical of many runs of the algorithm and examples we have investigated and is consistent with the discussion in section [ sec : comp_consider ] , where we stated that algorithm [ alg : new ] is likely to out - perform algorithm [ alg : ntry ] when the @xmath153 are not large , which is exactly the scenario in this example . turning to the cost of simulating algorithm [ alg : new ] ; for the case @xmath248 we simulated the data an average of 148000 times ( per - iteration ) and for @xmath249 this figure was 330000 .
in this example signifcant effort is expended in simulating the @xmath183 .
this shows , at least in this example , that one can run the algorithm without it failing to sample the @xmath183 .
the results here suggest that one should prefer algorithm [ alg : new ] only in challenging scenarios , as it can be very expensive in practice . finally , we remark that the mle for a gaussian garch model , is @xmath250 .
this differs to the posterior means , which may indicate that a stable distribution could be useful for modelling the observations for this class of models .
in this article we have considered approximate bayesian inference from observation driven time series models .
we looked at some consistency properties of the corresponding map estimators and also proposed an efficient abc - mcmc algorithm to sample from these approximate posteriors .
the performance of the latter was illustrated using numerical examples .
there are several interesting extensions to this work : * the asymptotic analysis of the abc posterior in section [ sec : consis ] can be further extended .
for example , one may consider bayesian consistency or bernstein von - mises theorems , which could provide further justification to the approximation that was introduced here .
alternatively , one could look at the the asymptotic bias of the abc posterior w.r.t .
@xmath51 or the asymptotic loss in efficiency of the noisy abc posterior w.r.t .
@xmath51 similar to the work in @xcite for hidden markov models . * the geometric ergodicity of the presented mcmc sampler can be further investigated in the spirit of @xcite . *
an investigation to extend the ideas here for sequential monte carlo methods should be beneficial .
this has been initiated in @xcite in the context of particle filtering for a different class of models .
a. jasra acknowledges support from the moe singapore and funding from imperial college london .
n. kantas was kindly funded by epsrc under grant ep / j01365x/1 .
[ proof of proposition [ prop1 ] ] the proof of @xmath251 follows from ( * ? ? ? * theorem 21 ) if we can establish conditions ( b1 - 3 ) for our perturbed abc model .
clearly ( b1 ) and part of ( b2 ) holds .
( b3-i ) hold via ( * ? ? ?
* lemma 21 ) via ( a[as:3 ] ) .
we first need to show that for any @xmath252 that @xmath253 is continuous .
consider @xmath254 let @xmath255 , then , by ( a[as:2 ] ) there exists a @xmath256 such that for @xmath257 @xmath258 and hence for @xmath259 as above @xmath260 which establishes ( b2 ) of @xcite .
now , for ( b3-ii ) of @xcite , we note that as @xmath261 ( see ( a[as:2 ] ) ) the @xmath262 function is lipshitz and @xmath263 for some @xmath208 that does not depend upon @xmath264 . now @xmath265\mu(dy)|\ ] ] and @xmath266|.\ ] ]
thus , by ( a[as:3 ] ) and the fact that ( b3-i ) of @xcite holds : @xmath267 note , finally that ( b3-iii ) trivially follows by @xmath268 .
hence we have proved that @xmath269 [ proof of proposition [ prop2 ] ] this result follows from ( * ? ? ?
* proposition 23 )
. one can establish assumptions ( b1 - 3 ) of @xcite using the proof of proposition [ prop1 ] .
thus we need only prove that @xmath270 now , for any @xmath271 @xmath272\mu(dy).\ ] ] by ( a[as:4 ] ) @xmath273 @xmath274 , so @xmath275\mu(dy)\leftrightarrow x = x'\ ] ] which completes the proof .
[ proof of proposition [ prop : rel_var ] ] we have @xmath180=\frac{1}{(\prod_{k=1}^{n}\frac{\alpha_{k}(y_{1:k},\epsilon,\gamma)}{n-1})^{2}}\bigg(\prod_{k=1}^{n}\mathbb{e}_{\gamma , n}\big[\frac{1}{(m_{k}-1)^{2}}\big]-\big(\prod_{k=1}^{n}\frac{\alpha_{k}(y_{1:k},\epsilon,\gamma)}{n-1}\big)^{2}\bigg).\ ] ] now , by @xcite ( @xmath276 ) for any @xmath277 @xmath278=\frac{\alpha_{k}(y_{1:k},\epsilon,\gamma)^{2}}{(n-1)(n-2)}\ ] ] and thus clearly @xmath279\leq\frac{\alpha_{k}(y_{1:k},\epsilon,\gamma)^{2}}{(n-1)(n-2)}.\ ] ] hence @xmath180\leq(n-1)^{2n}\big(\frac{1}{(n-1)^{n}(n-2)^{n}}-\frac{1}{(n-1)^{2n}}\big).\label{eq : maineq1}\ ] ] now the r.h.s.of is equal to @xmath280}{n^{n}-2nn^{n-1}+\sum_{i=2}^{n}\binom{n}{i}n^{n - i}(-2)^{i}}.\label{eq : maineq2}\ ] ] now , we will show @xmath281\leq0.\label{eq : ineq1}\ ] ] the proof is given when @xmath1 is odd .
the case @xmath1 even follows by the proof as @xmath282 is odd and the additional term is negative .
now we have for @xmath283 that the sum of consecutive even and odd terms is equal to @xmath284\ ] ] which is negative as @xmath285 thus we have established .
we will now show that @xmath286 following the same approach as above ( i.e. @xmath1 is odd ) the sum of consecutive even and odd terms is equal to @xmath287.\ ] ] this is positive if @xmath288 as @xmath289 and @xmath290 it follows that @xmath291 ; thus one can establish .
[ proof of proposition [ prop : time ] ] we have @xmath294 & = & \int_{(\theta\times\mathsf{x})^{2}}\sum_{\mathsf{m}_{n}^{n}}\big(\sum_{k=1}^{n}m_{k}\big)\big\{\prod_{k=1}^{n}\binom{m_{k}-1}{n-1}\alpha_{k}(y_{1:k},\gamma',\epsilon)^{n}(1-\alpha_{k}(y_{1:k},\gamma',\epsilon))^{m_{k}-n}\big\ } q(\gamma,\gamma')\zeta k^{i}(d\gamma)d\gamma'\\ & = & \int_{(\theta\times\mathsf{x})^{2}}\big(\sum_{k=1}^{n}\frac{n}{\alpha_{k}(y_{1:k},\gamma,\epsilon)}\big)q(\gamma,\gamma')\zeta k^{i}(d\gamma)d\gamma'\leq\frac{nn}{c}.\end{aligned}\ ] ] where we have used the expectation of a negative - binomial random variable and applied @xmath201 , @xmath202a.e . in the inequality
10 andrieu , c. , doucet , a. & holenstein , r. ( 2010 ) .
particle markov chain monte carlo methods ( with discussion ) .
_ j. r. statist .
b _ , * 72 * , 269342 .
andrieu , c. & vihola , m. ( 2012 ) .
convergence properties of pseudo - marginal markov chain monte carlo algorithms .
arxiv:1210.1484 [ math.pr ] | in the following article we consider approximate bayesian parameter inference for observation driven time series models .
such statistical models appear in a wide variety of applications , including econometrics and applied mathematics .
this article considers the scenario where the likelihood function can not be evaluated point - wise ; in such cases , one can not perform exact statistical inference , including parameter estimation , which often requires advanced computational algorithms , such as markov chain monte carlo ( mcmc ) .
we introduce a new approximation based upon approximate bayesian computation ( abc ) . under some conditions , we show that as @xmath0 , with @xmath1 the length of the time series , the abc posterior has , almost surely , a maximum _ a posteriori _ ( map ) estimator of the parameters which is different from the true parameter .
however , a noisy abc map , which perturbs the original data , asymptotically converges to the true parameter , almost surely . in order to draw statistical inference , for the abc approximation adopted
, standard mcmc algorithms can have acceptance probabilities that fall at an exponential rate in @xmath1 and slightly more advanced algorithms can mix poorly .
we develop a new and improved mcmc kernel , which is based upon an exact approximation of a marginal algorithm , whose cost per - iteration is random but the expected cost , for good performance , is shown to be @xmath2 per - iteration .
we implement our new mcmc kernel for parameter inference from models in econometrics .
+ * key words : * observation driven time series models , approximate bayesian computation , asymptotic consistency , markov chain monte carlo . * approximate inference for observation driven time series models with intractable likelihoods * by ajay jasr@xmath3 , nikolas kanta@xmath4 , & elena ehrlic@xmath5 @xmath6department of statistics & applied probability , national university of singapore , singapore , 117546 , sg .
+ e-mail:`[email protected] ` @xmath7department of statistical science , university college london , london , wc1e 6bt , uk .
+ e-mail:`[email protected] ` @xmath8department of mathematics , imperial college london , london , sw7 2az , uk .
+ e-mail:`[email protected] ` |
compositeness @xcite is regarded as a solution to many problems which are unsolved in the standard model , like the fermion mass spectrum , the generation of masses and the large number of parameters .
the idea is that some underlying structure can explain the quark and lepton family pattern . at this stage
, we shall search for new excited fermions ( @xmath9 ) , and for new interactions originated from the constituents . a signal for
a composite structure of fermions could be the production of excited leptons or quarks , either pairwise , due to their normal gauge couplings , or singly , due to radiative transitions between normal and excited fermions @xcite @xcite .
for instance , excited fermions could be produced in @xmath10 decays ; in fact , there are strong limits coming from the @xmath11 partial widths .
mass regions up to @xmath12 gev for @xmath13 and @xmath14 as well as @xmath15 gev for @xmath16 and @xmath17 have been excluded @xcite .
however , the @xmath16 limits can be pushed to @xmath18 gev by searching the decay mode @xmath19 @xcite .
moreover , recently the possibility has been studied that the significant excess from qcd predictions found by the collider detector at fermilab ( cdf ) in the inclusive jet cross section for transverse energies @xmath20 gev could be explained by the production of excited bosons or excited quarks in the mass region of @xmath21 gev and @xmath22 gev@xcite @xcite , respectively .
other studies have already been done in order to get the low energy scale @xmath23 for compositeness by precise comparison between currently available experimental data and calculations in the composite model of quarks @xcite .
on the other hand , radiative corrections are the appropriate place to indirectly look for new physics at low energies @xcite @xcite , because when heavy fermions are coupled to longitudinal components of the gauge bosons there is no decoupling of the high energy physics from the low energy physics . in ref .
@xcite , radiative corrections at the @xmath10 scale have been considered in order to bound substructure .
our purpose in this paper is to find bounds for the masses of possible excited leptons by using the radiative corrections involved in the forward - backward and left - right asymmetries for the lepton average .
in general , the normal fermions ( @xmath24 ) and excited fermions ( @xmath25 ) at low energies can be described by an effective lagrangian like @xcite @xmath26 where @xmath27 are the field strengths for the gauge bosons and @xmath28 denotes the compositeness scale . in particular , we choose a specific model which assumes that excited leptons form a weak isodoublet , @xmath29 which couples to the ordinary left handed lepton doublet by the @xmath30 invariant interaction lagrangian @xmath31 where @xmath32 and @xmath33 are the @xmath34 and @xmath35 coupling constants respectively , @xmath36 denotes the pauli matrices and @xmath37 is the hypercharge . as for the coefficients @xmath38 and @xmath39 , in general , @xmath40 measurements imply @xmath41 , and also , the absence of dipole moments for the electron and the muon requires @xmath42 and @xmath43 to be relatively real @xcite .
thus , we have the following relations : @xmath44 now , since the excited fermions are doublets , their couplings to photon , @xmath10 and @xmath45 bosons are defined by the following renormalizable lagrangian @xmath46 where the coefficients @xmath47 from eq.(1 ) satisfy @xmath48
once we have the framework setup ( eqs . ( 1)-(6 ) ) , we can calculate the contribution to the self - energies of the gauge bosons by using dimensional regularization , where the pole @xmath49 is identified with @xmath50 @xcite . in order to simplify the analysis we assume the same mass for the excited leptons , @xmath51 ;
this is justified by the fact that we are considering the excited fermion in a linear vectorial representation of the @xmath52 gauge group .
the states with equal mass for @xmath53 and @xmath54 do not break the symmetry .
in general the self - energies due to the lagrangians of dimension 5 and 4 are : @xmath55 where the functions @xmath56 can be written as @xcite @xmath57 \nonumber \\ & + & 3m^{\ast 2}q^{2}[1+(\frac{m^{\ast 2}-q^{2}}{q^{2}})\ln ( \frac{m^{\ast 2}}{% m^{\ast 2}-q^{2 } } ) ] \nonumber \\ & -&q^{4}[8 + 3\ln ( \frac{\lambda ^{2}}{m^{\ast 2}})-3(\frac{m^{\ast 2}-q^{2}}{% q^{2}})\ln ( \frac{m^{\ast 2}}{m^{\ast 2}-q^{2}})\;\ ; , \nonumber \\
f^{(4)}(m^{\ast } , q^{2 } ) & = & q^{2}\ln \frac{\lambda ^{2}}{m^{\ast 2}}% + 4m^{\ast 2}+\frac{5}{3}q^{2 } \nonumber \\ &
-&2(2m^{\ast 2}+q^{2})\sqrt{\frac{4m^{\ast 2}}{q^{2}}-1}\arctan ( \frac{1}{% \sqrt{\frac{4m^{\ast 2}}{q^{2}}-1}})\;\;.\end{aligned}\ ] ] we have considered the ordinary lepton masses negligible . therefore , we find the vacuum polarization tensors for the gauge bosons : @xmath58 where @xmath59 and @xmath60 is obtained directly from @xmath61 . the asymptotic limits @xmath62 in the functions @xmath63are @xmath64 in such limit , @xmath65 approaches a constant value and the physics described by ( 3 ) is non - decoupled .
on the other hand , the forward - backward asymmetry @xmath0 and the left - right asymmetry @xmath1 for leptons in the @xmath10 decays are given by @xmath66 where a @xmath67 amplitude is defined as @xmath68 and the superscripts @xmath69 denote electron , lepton respectively . in our case , new physics is included in the constants @xmath70 as @xmath71 where the superindex sm denotes the sm coupling with the contribution at one loop level of the top quark and higgs scalar boson . in this way , @xmath72 only contains the new physics contribution . with this prescription
we find that @xmath73 and @xmath1 can be written as @xmath74 where @xmath75 has the contribution of new physics and @xmath76 additionally , these couplings @xmath77 can be expressed in terms of the @xmath78 parameters @xcite , @xmath79 here , @xmath80 is the @xmath2 vertex correction given in @xcite .
the other two parameters @xmath81 and @xmath82 can be written as a function of self - energies as given by eq .
( 9 ) @xcite .
we consider the asymmetries for electron and leptonic average .
the sm predicted values are @xcite @xmath83 for @xmath84 , @xmath85 , @xmath86 gev and @xmath87 gev . and the experimental values are @xcite : @xmath88 notice that the standard model value @xmath89goes 1.54@xmath90out from experimental data .
figure 1 shows the @xmath0 as a function of the scale @xmath91 for values of @xmath92 gev ( a ) and @xmath93 gev ( b ) .
different values of the parameters @xmath8 and @xmath94 have been considered .
figure 2 shows @xmath1 as a function of the scale @xmath95 for the excited lepton mass of @xmath93 gev ( a ) and @xmath96 gev ( b ) . in figure 3
we have plotted @xmath0 of the electron versus @xmath5 , as @xmath95 is the energy scale for new physics , @xmath97 must be less than @xmath95 , and we have @xmath98between @xmath99and @xmath100 . in figure 4
we show the left - right asymmetry versus the quotient @xmath101 for different values of @xmath102and @xmath103couple constants .
we can see that both forward - backward and left - right asymmetries produce bounds for such quotient when @xmath104overweight considerably to @xmath105it means that custodial symmetry is strongly broken and therefore radiative contributions become higher@xmath106
in conclusion , we have evaluated the contribution of excited lepton states , up to one - loop level , to the oblique parameters .
we have included this contribution into the leptonic asymmetries @xmath107and @xmath1 , and we have compared our results with the precise data on the electroweak observables obtained by lep and lsd collaborations @xcite @xcite . therefore , we can extract bounds on the compositeness scale and the excited lepton mass in the context of the phenomenological model under consideration .
further , in equation ( 3 ) the term proportional to @xmath108 breakdowns the custodial symmetry and it is proportional to the constant @xmath7 .
otherwise the new physics is allowed in the experimental region ( @xmath109 , which is satisfied when custodial symmetry is strongly broken . in order to get this requirement
we need that @xmath94 becomes bigger than @xmath8 . with this prescription
we obtain bounds on the rate @xmath110
we thank f. larios for reading the manuscript and useful comments .
we acknowledge the financial support from colciencias ( colombia ) and the hospitality of cinvestav ( mexico ) during the realization of this work .
r. d. express his acknowledgements to the fundacin mazda para el arte y la ciencia for his fellowship .
baur , spira , p. zerwas : phys .
d42 , ( 1990 ) 815 ; j. khun , p. zerwas : phys .
b 147 ( 1984 ) 189 ; akama , hattori : int . j. of mod . phys .
a9 ( 1994 ) 3504 ; n. cabibbo , l. maiani , y. srivastava : phys .
b 139 ( 1984 ) 459 ; f. boudjema , a. djouadi : phys . lett .
b 240 ( 1990 ) 485 ; f. boudjema , et . al . :
c 57 ( 1993 ) 425 ; o.j .
p. eboli , et .
d 53 ( 1996 ) 1253 aleph collaboration , d. decamp , et .
b 236 ( 1990 ) 501 ; b 250 ( 1990 ) 172 ; l3 collaboration , o. adriani , et .
phys . lett .
b 288 ( 1992 ) 404 ; l3 collaboration , m. acciani , et .
b353 ( 1995 ) 136 ; opal collaboration m. z. akrawy et .
b 240 ( 1990 ) 497 ; b 241 ( 1990 ) 133 ; delphi collaboration , p. abreu , et .
b 268 ( 1991 ) 296 ; b 327 ( 1994 ) 386 ; z. phys .
c 53 ( 1992 ) 4 .
@xmath0 as a function of the scale @xmath95 for values of the excited lepton mass of @xmath92gev ( a ) and @xmath111gev ( b ) . from the upper solid line we use the values ( @xmath112 ) , ( @xmath113 ) , ( @xmath114 , @xmath115 up to ( @xmath116 , @xmath117 the lower dot - dot - solid line .
the horizontal solid lines are the experimental limits .
@xmath0 as a function of the ratio @xmath101 . from the upper solid line we use the values ( @xmath116 , @xmath118 , ( @xmath119 , @xmath120 ) , ( @xmath121 , @xmath122 up to ( @xmath116 , @xmath123 ) the lower dashed line .
the horizontal solid lines are the experimental limits .
@xmath1 as a function of the ratio @xmath101 . from the upper solid line we use ( @xmath116 , @xmath118 , ( @xmath114 , @xmath124 ) , ( @xmath121 , @xmath125 ) ,
( @xmath126 , @xmath127 ) , ( @xmath116 , @xmath127 ) , up to @xmath116 , @xmath128 ( lower dashed line ) .
the horizontal solid lines are the experimental limits . | we study the effects on the leptonic asymmetries @xmath0 and @xmath1 coming from a model of compositeness . we consider the effects coming from the self - energies and the vertex correcction to @xmath2 .
thus we use the altarelli parametrization of the oblique corrections .
we get the asymptotic limits of these corrections in terms of the parameters @xmath3 @xmath4 and we get bounds for the quotient @xmath5 with different values of @xmath6 @xmath4 .
we conclude that both asymmetries produce bounds for such quotient when @xmath7 overweight to @xmath8 , and this fact is related with the breakdown of the custodial symmetry . |
over the past 30 years , much effort has been devoted to calculations of the renormalised stress - energy tensor in ground states of quantum fields on stationary background spacetimes .
many analogous calculations have been made in flat spacetime equipped with reflecting boundaries , in connection with the casimir effect .
however , it would be fair to say that only limited qualititative insight has been gained .
for example , the energy density is sometimes positive , and sometimes negative and there is no known way of predicting the sign in any general situations without performing the full calculations ( see , however , for a situation where the sign can be predicted ) .
at least analytically , these calculations are restricted to cases exhibiting a high degree of symmetry .
the aim of this paper , and a companion paper @xcite , is to point out that there are situations in which one may gain some qualitative insight into the possible magnitude of the stress - energy tensor based on simple geometric considerations .
the situation we study in this paper arises when a spacetime contains a subspacetime which is isometric to ( a subspacetime of ) another spacetime , which will usually have nontrivial symmetries . by using quantum energy inequalities ( qeis ) together with the locality properties of quantum field theory
, we are then able to use information about the second ( symmetric ) spacetime to yield information about the stress - energy tensor of states on the first spacetime ( which need have no global symmetries ) in the region where the isometry holds .
we will work on globally hyperbolic spacetimes in this paper , deferring the issue of spacetimes with boundary to a companion paper @xcite .
as well as setting out the theory behind the method , we will demonstrate it in several locally minkowskian spacetimes .
marecki @xcite has also illustrated our approach , by considering the case of spacetimes locally isometric to portions of exterior schwarzschild .
also begun here for the free massless scalar field is a similar discussion for conformally related regions of two - dimensional spacetimes . in a separate paper
we will extend this to the generalised maxwell field in higher dimensional manifolds related by conformal diffeomorphisms . to be more specific , consider a globally hyperbolic spacetime @xmath0 , consisting of a manifold of dimension @xmath1 , a lorentzian metric with signature @xmath2 , and choices of orientation and time - orientation ( which , together , are required to fulfill the demands of global hyperbolicity ) .
is globally hyperbolic if it contains a cauchy surface , i.e. , a subset intersected exactly once by every inextendible timelike curve @xcite .
the globally hyperbolic spacetimes are the most general class of spacetimes on which quantum fields are typically formulated , but one should be aware that manifolds with boundary are not included . ]
suppose an open subset of @xmath0 , when equipped with the metric and ( time-)orientation inherited from @xmath0 , is a globally hyperbolic spacetime @xmath3 in its own right .
if , moreover , any causal curve in @xmath0 whose endpoints lie in @xmath3 is contained completely in @xmath3 , then we will call @xmath3 a _ causally embedded globally hyperbolic subspacetime _
( c.e.g.h.s . ) of @xmath0 .
our main interest will be in the situation where a c.e.g.h.s .
@xmath3 of @xmath0 is isometric to a c.e.g.h.s .
@xmath4 of a second globally hyperbolic spacetime @xmath5 , with the isometry also respecting the ( time-)orientation .
( we speak of a _ causal isometry _ in this case . ) by the principle of locality , we expect that any experiment conducted within @xmath3 should have the same results as the same experiment [ i.e. , its isometric image ] conducted in @xmath4 .
no observer in @xmath3 should be able to discern , by such local experiments that she does not , in fact , inhabit @xmath4 ; in particular , energy densities in @xmath3 should be subject to the same qeis as those in @xmath4 .
we will demonstrate explicitly that these expectations are met by the qeis we employ .
( -15,1)(15,15 ) ( -8.5,2)@xmath6 ( -8.5,8.5)@xmath7 ( 8,2)@xmath8 ( 10.5,8.5)@xmath9 ( 8,5)(7,8)(9,10)(8,12.5 ) ( 7,8.75)@xmath10 ( -5,9)(-1,10)(0,10)(4.6,9 ) ( -0.2,11)@xmath11 among our results are the following , which we state for the case of a klein gordon field of mass @xmath12 in four dimensions : _ example 1 : _ suppose a timelike _ geodesic _ segment @xmath10 of proper duration @xmath13 in a globally hyperbolic spacetime @xmath0 can be enclosed in a c.e.g.h.s.@xmath3 which is causally isometric to a c.e.g.h.s . of four - dimensional minkowski space as shown in fig .
[ fig : mink_embed ] .
then any state @xmath14 of the klein gordon field ( of mass @xmath12 ) on @xmath0 obeys @xmath15 where the constant @xmath16 ( if @xmath17 , one may obtain even more rapid decay ) .
_ example 2 : _ suppose a globally hyperbolic spacetime @xmath0 is stationary with respect to a timelike killing field @xmath18 and admits the smooth foliation into constant time surfaces @xmath19 .
suppose the metric takes the minkowski form ( w.r.t . some coordinates ) on @xmath20 for some subset @xmath21 of @xmath22 with nonempty interior .
( we may suppose that @xmath21 has been taken to be maximal . ) for any @xmath23 in the interior of @xmath21 , let @xmath24 be the radius of the largest euclidean @xmath25-ball which can be isometrically embedded in @xmath21 , centred on @xmath23 , as in fig .
[ fig : example2 ] .
then any stationary hadamard state -point functions
are invariant under translations along the killing flow : @xmath26 , where @xmath27 is the group of isometries associated with the killing field . ] @xmath28 on @xmath0 obeys the bound @xmath29 for any @xmath30 , where @xmath31 is the unit vector along @xmath18 .
( -7,0)(20,11 ) ( 0,0)(0,1.6 ) ( 0,1.6)(0,8 ) ( 0,8)(0,11 ) ( 0,11.5)@xmath32 ( 0,3)(4,3.9)(8,3.5)(10,4)(8.75,5)(13,6.5)(11,7)(8.5,6.2)(6,7)(4.5,6.7)(2,7.25)(0,5.5)(-2,4)(-1.75,3)(0,3 ) ( 4,5)(3.3,1.1 ) ( 4,5)(.2,.066 ) ( 4,5)(1.5,4.4 ) ( 4.5,5)@xmath23 ( 2.7,5.2)@xmath24 ( 3,1)@xmath33 ( -.5,3.8)@xmath34 _ example 3 :
_ suppose @xmath35 is a uniformly accelerated trajectory ( parametrised by proper time ) with proper acceleration @xmath36 , and suppose @xmath10 can be enclosed within a c.e.g.h.s .
@xmath3 of @xmath0 which is causally isometric to a c.e.g.h.s . of four - dimensional minkowski space .
then , for any hadamard state @xmath14 on @xmath0 , and any smooth compactly supported real - valued @xmath37 , with @xmath38 , @xmath39 note the remarkable fact that the right - hand side is precisely the expected energy density in the rindler vacuum state along the trajectory with constant proper accleration @xmath36 .
in particular , if the energy density in some state @xmath14 is constant along @xmath10 , it must exceed or equal that of the rindler vacuum .
we emphasize that our derivation does not involve the rindler vacuum , but only the minkowski vacuum state two - point function and the qeis .
variants of these results hold in other dimensions , and also for other linear field equations such as the maxwell and proca fields ( which we will treat elsewhere ) . to prepare for our main discussion
, it will be useful to make a few general remarks about quantum energy inequalities ( qeis ) , also often called simply quantum inequalities ( qis ) .
qeis have been quite intensively developed over the past decade , following ford s much earlier insight @xcite that quantum field theory might act to limit the magnitude and duration of negative energy densities and/or fluxes , thereby preventing macroscopic violations of the second law of thermodynamics ( see for rigorous links between qeis and thermodynamical stability ) .
detailed reviews of qeis may be found in @xcite .
qeis take various forms , but we will distinguish two basic types : absolute qeis and difference qeis .
an absolute qei bound consists of a set @xmath40 of _ sampling tensors _ ,
i.e. , second rank contravariant tensor fields against which the renormalised stress - energy tensor will be averaged , a class @xmath41 of states of the theory ( which may be chosen to have nice properties ) and a map @xmath42 such that @xmath43 for all states @xmath44 be convex ( i.e. , if @xmath45 and @xmath46 are in @xmath40 then so is @xmath47 for all @xmath48 $ ] , and for @xmath49 to obey @xmath50 , but we shall not make these requirements . ] . here
@xmath51 is the renormalised stress - energy tensor defined in a manner compatible with wald s axioms @xcite , and we have adopted the convention that the same tensor may be written @xmath52 ( without indices ) or @xmath53 ( with ) .
we will permit @xmath40 to include tensors singularly supported on timelike curves or other submanifolds of spacetime , so , for example , we can treat worldline averages such as @xmath54 where @xmath10 is a smooth timelike curve parametrised by an open interval @xmath55 of proper time @xmath56 , with velocity @xmath57 , and for @xmath58 .
[ to be precise , @xmath40 is required to be a set of compactly supported distributions on smooth rank two covariant test tensor fields . ]
absolute qeis are known [ with explicit formulae for @xmath49 , and specific @xmath40 and @xmath41 ] for ( a ) the scalar field of mass @xmath12 in @xmath60-dimensional minkowski space ( see also @xcite for @xmath1 ) , ( b ) the massless scalar and fermi fields in arbitrary two - dimensional globally hyperbolic spacetimes @xcite , ( c ) general ( interacting ) conformal field theories in two - dimensional minkowski space , ( d ) a variety of higher spin linear fields in two- and four - dimensional minkowski space . for the most part only worldline bounds involving averages of the form eq .
have been studied ; it has been found that replacing @xmath37 by a scaled version @xmath61 has the effect of sending the qei bound to zero as @xmath62 ( or faster , for massive fields ) as @xmath63 , where @xmath60 is the spacetime dimension .
difference qei bounds also involve the specification of @xmath40 and @xmath41 as before , but now the bound sought takes the form @xmath64 f^{ab } \,{d{\rm vol}}\ge -{{\mathcal q}}({{\sf f}},\omega_0)>-\infty\ , , \label{eq : qidiff}\ ] ] where @xmath65 is called the reference state . if the theory were represented in a fock space built on @xmath65 ( when this is possible ) the left - hand side would be an average of the normal ordered stress - energy tensor .
however it is not always necessary to assume that @xmath14 and @xmath65 are represented in this way .
difference qeis have proved to be the easiest to establish in curved spacetimes , or where boundaries are present .
first developed in the case of ( ultra)static spacetimes with the ( ultra)static ground state chosen as the reference state @xmath65 , they are now known for scalar , spin-@xmath66 , and spin-@xmath67 fields in arbitrary globally hyperbolic spacetimes . in these general results , @xmath41 is the class of hadamard states and the bounds are sufficiently general that @xmath65 may be any element of @xmath41 , so @xmath49 becomes a function @xmath68 .
the general results do not make use of a hilbert space representation . clearly , difference and absolute qeis are quite closely related .
in particular , wald s fourth axiom requires @xmath69 to vanish identically if @xmath65 is the minkowski vacuum , so difference qeis become absolute in this case .
[ the extension of this observation to locally minkowskian spaces is a key idea in this paper . ]
more generally , we may convert a difference qei to an absolute qei by moving all the terms in @xmath65 onto the right - hand side . in cases
where the renormalised stress - energy tensor is known explicitly for the reference state , this is perfectly satisfactory .
however , there are two ( related ) drawbacks : ( i ) there is no canonical choice of reference state @xmath65 in a general spacetime ( which might have no timelike killing fields , for example ) ; ( ii ) one does not normally have available a closed form expression for @xmath70 for _ any _ state on a general spacetime , so the qei bound becomes somewhat inexplicit .
this weakens the power of qeis to constrain exotic spacetime configurations such as macroscopic traversable wormholes or ` warp drive ' .
( on sufficiently small scales , one expects that the absolute qei bounds should strongly resemble those of minkowski space as first argued in , and proven in various situations in @xcite and @xcite however one still needs to know the magnitude of @xmath69 to know on what scales this approximation holds . )
the present paper and its companion represent first steps towards absolute qeis in more general spacetimes , starting with spacetimes containing regions isometric to others where reference states are known .
work is under way on generally applicable absolute qeis and will be reported elsewhere ; however we expect the results and methods presented here to be of continuing interest , as they reduce to very simple geometrical conditions .
the paper is structured as follows . in sec .
[ sec : qeis_and_lc ] we give a brief introduction to some of the relevant notions of locally covariant quantum field theory before defining locally covariant qeis and developing their simple properties in sec . [ sec : qei_loc_cov ] .
the following two subsections show how existing qeis in the literature may be expressed in the locally covariant framework , and address some technical points along the way . in sec .
[ sec : general_apps ] we show how local covariance permits _ a priori _ bounds to be placed on energy densities in spacetimes with minkowskian subspacetimes using geometric data .
the main technique here , in addition to local covariance , is the conversion of qeis to eigenvalue problems , first introduced in .
these are applied in sec .
[ sec : specific_apps ] to specific spacetime models where the energy densities of ground- and thermal states are known , permitting comparison with our _ a priori _ bounds . in some cases these bounds
are saturated by the exact values . after a summary ,
the appendices collect various results needed in the main text .
suppose two globally hyperbolic spacetimes of the same dimension , @xmath71 and @xmath72 , are given ( we denote the corresponding manifolds and metrics by @xmath73 , @xmath74 for @xmath75 ) .
isometric embedding _ of @xmath71 in @xmath72 is a smooth map @xmath76 which is a diffeomorphism of @xmath77 onto its range @xmath78 in @xmath79 and so that the pull - back @xmath80 is everywhere equal to @xmath81 on @xmath77 . in local coordinates
, @xmath82 should hold for all @xmath83 , where @xmath84 .
we _ do _ require that all of @xmath77 is mapped into @xmath79 , but we do _ not _ require that the image of @xmath77 under @xmath11 consists of the whole of @xmath79 .
there are two possible choices of orientation and time orientation on @xmath78 : that induced by @xmath11 from the ( time-)orientation of @xmath71 , and that inherited from @xmath72 .
if these coincide and we have the further property that every causal curve in @xmath72 with endpoints in @xmath78 lies entirely in @xmath78 , then we say that @xmath11 is a _ causal isometric embedding_. an important class of examples arises where @xmath71 is a causally embedded globally hyperbolic subspacetime ( c.e.g.h.s . ) of @xmath72 as defined in sec .
[ sec : int_summary ] , in which case @xmath11 is simply the identity map .
it is also worth mentioning an example of a non - causal embedding , namely , the ` helical strip ' described by kay @xcite . in this example a long thin diamond region of two dimensional minkowski space
is isometrically embedded in a ` timelike cylinder ' which is the quotient of minkowski space by a spacelike translation .
the wrapping is arranged so that points which are spacelike separated in the original diamond are timelike separated in the geometry of the timelike cylinder .
the definition of a causal embedding is designed precisely to ensure that the induced and inherited causal structures can not differ in this way .
the relevance of local covariance to quantum field theory on manifolds has long been understood @xcite but has recently been put in a new setting by brunetti , fredenhagen and verch @xcite ( see also @xcite and ) and related work of hollands and wald ( see , e.g. , ) .
this provides a very elegant and general framework for local covariance in the language of category theory .
however , we will only need a few of the main ideas of this analysis and will not describe the whole structure . in this section
, we will restrict ourselves to the klein
gordon field of mass @xmath85 , although similar comments can be made for the dirac , maxwell , and proca fields .
there is a well - defined quantisation of the theory on any globally hyperbolic spacetime @xmath5 , in terms of an algebra of observables @xmath86 and a space of hadamard states @xmath87 which determine expectation values for observables in @xmath86
. for the purposes of this section , it suffices to know that @xmath86 is generated by smeared field objects @xmath88 labelled by smooth , compactly supported test functions @xmath89 , subject to relations expressing the field equation and commutation relations , and the hermiticity of the field .
( the structure is given in detail in appendix [ appdx : locally_covar_qft ] . )
the hadamard states of the theory are those states on @xmath86 whose two - point functions have singularities of the hadamard form , which at leading order are just those of the minkowski vacuum two - point function .
more precisely , on any causal normal neighbourhood @xmath90 in @xmath5 there is a sequence of bidistributions @xmath91 so that ( for any @xmath92 ) the two - point function of any hadamard state differs from @xmath91 on @xmath90 by a state - dependent function of class @xmath93 .
it is of key importance that @xmath94 is fixed entirely by the local metric and causal structure , through the hadamard recursion relations .
given a hadamard state @xmath14 , we may construct the expected renormalised stress - energy tensor @xmath95 by the point - splitting technique ( see , e.g. , @xcite ) : first subtract @xmath91 from the two - point function ( for @xmath96 ) , then apply appropriate derivatives before taking the points together again .
next , one subtracts a term of the form @xmath97 , where @xmath98 is locally determined ( and state - independent ) , in order to ensure that the resulting tensor is conserved and vanishes in the minkowski vacuum state .
the tensor defined in this way obeys wald s axioms mentioned above ; however , these axioms would also be satisfied if one were to add a conserved local curvature term .
such terms are sometimes described as undetermined or arbitrary ; we take the view , however , that they are part of the specification of the theory , just as the mass and conformal coupling are , even though they do not appear explicitly in the lagrangian ( a similar attitude is expressed in ) . for simplicity , and because our main applications will concern locally minkowskian spacetimes , we will assume that these terms are absent that is , we restrict to those scalar particle species for which this is the case .
the above structure is locally covariant in the following sense .
suppose a globally hyperbolic spacetime @xmath5 is embedded in a globally hyperbolic spacetimes @xmath0 by a causal isometry @xmath11 , and let @xmath99 denote the push - forward map on test functions .
that is , @xmath100 is defined by @xmath101 then there is a natural mapping of the field on @xmath5 to the field on @xmath0 given by @xmath102 ; we also write this as @xmath103 .
moreover , @xmath99 can be extended to any element of @xmath86 , respecting the algebraic relations and mapping the identity in @xmath86 to the identity in @xmath104 ; technically , it is a unit - preserving injective @xmath105-homomorphism of @xmath86 into @xmath104 . on account of the correspondence @xmath102 , we say that the field is covariant [ the transformation goes ` in the same direction ' as @xmath11 ; see the remarks below on the underlying category theory at the end of appendix [ appdx : locally_covar_qft ] ] .
by contrast , the state spaces transform in a contravariant way [ in the ` opposite direction ' to @xmath11 ] : for any state @xmath14 on @xmath104 there is a pulled - back state , which we denote @xmath107 , on @xmath86 , so that the expectation values of @xmath108 and @xmath109 are related by @xmath110 the use of pull - back notation may be justified by the observation that eq .
entails that the @xmath92-point functions of the two states are related by @xmath111 ( adopting an ` unsmeared ' notation ) .
that is , the @xmath92-point function of @xmath107 is simply the pull - back of the @xmath92-point function of @xmath14 by @xmath11 ( or more precisely , by the duplication of @xmath11 across @xmath92 copies of @xmath5 ) .
this has an important consequence when the state @xmath14 is hadamard , i.e. , @xmath112 : because the hadamard condition is based on the local metric and causal structure , both of which are preserved by @xmath11 , it is clear that @xmath107 is also hadamard .
( a more elegant proof of this @xcite is to use radzikowski s characterisation of the hadamard condition in terms of the wave - front set of the two - point function @xcite , and the transformation properties of the wave - front set under pull - backs . )
this may be expressed by the inclusion @xmath113 . as noted above , the expectation values of the stress - energy tensor
is also constructed in a purely local fashion from the two - point function of the state .
it therefore follows that @xmath114 where we have written @xmath84 : like the @xmath92-point functions , the expected stress - energy tensor in state @xmath107 is simply the pull - back of that in state @xmath14 . in coordinate - free notation
we may write @xmath115 in the above equations we have written the stress - energy tensor as if it is an element of the algebra @xmath86 , which it is not .
one may proceed in two ways : either interpreting eq . as the extension of eq . to an algebra of wick polynomials which contains @xmath86 as a subalgebra , and in which @xmath116 may be defined as a locally covariant field . for our purposes , however , it will be simpler to define the smeared stress - energy tensor only through its expectation values ; more technically , we think of it as a linear functional on the space of hadamard states , with the notation @xmath117 expressing the value of this functional applied to the state @xmath14 .
this has the advantage that one may deal with all hadamard states , rather than those which extend to the wick algebra .
we emphasise the fact that states are pulled back in this setting ; although one could push forward a state @xmath118 to obtain a state on @xmath119 , there is no guarantee that this can be extended to a hadamard state on @xmath104 , and indeed , such extensions do not always exist .
for example , the rindler vacuum state on the rindler wedge is hadamard in the interior of the wedge , but can not be extended to a hadamard state on the whole of minkowski because its stress - energy tensor diverges at the boundary of the wedge .
see @xcite for further discussion of these issues .
( -15,-5)(15,15 ) ( -8,2)@xmath6 ( 8,2)@xmath8 ( 8.5,8.5)@xmath120 ( -5,9)(-1,10)(0,10)(4.8,9 ) ( -0.2,11)@xmath11 ( -8,0 ) ( 8,0 ) ( -8,-2 ) ( 8,-2 ) ( -8,-4 ) ( 8,-4 ) we now introduce two types of locally covariant qeis . a more abstract ( and general ) definition can be given in the language of categories this will be pursued elsewhere .
recall that a set of sampling tensors on a globally hyperbolic spacetime is a set of compactly supported distributions on smooth second rank covariant tensor fields .
a _ locally covariant absolute qei _ assigns to each globally hyperbolic spacetime @xmath5 a set of sampling tensors @xmath121 on @xmath5 and a map @xmath122 such that ( i ) we have @xmath123 for all @xmath124 and @xmath118 , and ( ii ) if @xmath125 is a causal isometric embedding then @xmath126 and @xmath127 for all @xmath124 .
( we might also express this in the form @xmath128 . )
a _ locally covariant difference qei _ assigns to each globally hyperbolic @xmath5 a set of sampling tensors @xmath121 as before , and a map @xmath129 such that ( i ) @xmath130 for each @xmath124 and all @xmath131 ; ( ii ) @xmath132 holds for all @xmath124 and @xmath133
. we will shortly give examples of each type : flanagan s two - dimensional qeis for massless fields @xcite will be exhibited as a locally covariant absolute qei , while ( generalisations of ) the qei obtained in @xcite provide examples of locally covariant difference qeis . before that , let us examine some simple consequences of these definitions .
first , suppose that @xmath4 is a c.e.g.h.s .
of @xmath5 , so the identity map @xmath134 is a causal isometric embedding , and we must have @xmath135 and @xmath136 .
it is sensible to drop the identity mappings , and write the above in the form @xmath137 if @xmath138 is a causal isometric embedding we then obtain @xmath139 for all @xmath140 . as one would expect , this shows that locally covariant absolute qeis are indifferent to the larger spacetime ; one obtains the same bound whether one is in @xmath4 or its image @xmath3 in @xmath0 .
although this barely extends the original definition , it is worth isolating it as a separate result .
suppose a c.e.g.h.s .
@xmath4 of @xmath5 is causally isometric to a c.e.g.h.s .
@xmath3 of @xmath0 under the map @xmath11 .
then a locally covariant absolute qei obeys @xmath127 for all @xmath141 .
( -15,1)(15,15 ) ( -8,2)@xmath6 ( -8.5,8.5)@xmath7 ( 8,2)@xmath8 ( 8.5,9)@xmath9 ( -5,9)(-1,10)(0,10)(4.6,9 ) ( -0.2,11)@xmath11 let us now examine locally covariant difference qeis in this situation . given arbitrary hadamard states @xmath142 and @xmath143 on the parent spacetimes @xmath5 and @xmath0 , there are states @xmath144 and @xmath145 in @xmath146 , i.e. , hadamard states on @xmath147 . applying the difference qei to @xmath145 with @xmath144 as reference state , we find @xmath148 where we have used the transformation property eq . .
on the other hand , we could equally well apply the difference qei to @xmath144 , with @xmath145 as reference state , to obtain @xmath149 combining these inequalities yields @xmath150 we may also use the covariance of @xmath151 to reexpress the central member of this inequality in terms of expectation values on @xmath0 and @xmath5 , rather than @xmath4 .
the result , on dropping identity mappings from the notation , is the following .
suppose a c.e.g.h.s .
@xmath4 of @xmath5 is causally isometric to a c.e.g.h.s .
@xmath3 of @xmath0 under the map @xmath11 .
then a locally covariant difference qei obeys @xmath152 for all @xmath141 and any @xmath142 , @xmath143 .
note that the qeis used are those associated with the full spacetimes @xmath5 and @xmath0 ; similarly , the states @xmath153 , @xmath154 are states of the field on the full spacetimes .
however , the isometry @xmath11 connects only portions of the spacetime together and the restriction on the support of @xmath52 is therefore crucial : in general the above result will not hold when sampling extends outside the isometric region .
it is also worth noting that we have both lower and upper bounds . in this paper
, we will study the simplest possible setting for this result , in which @xmath5 is minkowski spacetime and @xmath153 is the minkowski vacuum state .
however other situations are possible .
for example , marecki @xcite has employed our framework in the case where @xmath5 is the exterior schwarzschild spacetime and @xmath153 is the boulware vacuum . in the minkowski case
, the result simplifies because the renormalised stress - energy tensor vanishes in the state @xmath153 , and we have the following statement . [ cor : diffmink ] suppose a c.e.g.h.s .
@xmath4 of minkowski space @xmath5 is causally isometric to a c.e.g.h.s .
@xmath3 of @xmath0 under the map @xmath11 .
then a locally covariant difference qei obeys @xmath155 for all @xmath141 and any @xmath143 , where @xmath153 is the minkowski vacuum state .
the qei we now describe was originally developed by flanagan @xcite for the massless scalar field in two dimensional minkowski space , in work which was subsequently generalised to curved spacetimes @xcite and also to arbitrary unitary positive energy conformal field theories in two dimensional minkowski space .
the results of @xcite were obtained for two - dimensional spacetimes globally conformal to the whole of minkowski space ; as noted in @xcite , however , any point of a globally hyperbolic two - dimensional spacetime has a ( causally embedded ) neighbourhood which is conformal to the whole of minkowski space , and to which flanagan s result applies .
we first state the result of @xcite , and then show that it meets our definition of a locally covariant absolute qei .
let @xmath5 be a globally hyperbolic two - dimensional spacetime , and suppose that @xmath10 is a smooth , future - directed timelike curve , parametrised by proper time @xmath156 , which is completely contained within a c.e.g.h.s .
@xmath4 of @xmath5 , such that @xmath4 is globally conformal to the whole of two - dimensional minkowski space .
then all hadamard states @xmath14 on @xmath5 obey the qei @xmath157\,d\tau \label{eq:2dcurvedqi}\ ] ] for any smooth , real - valued @xmath37 compactly supported in @xmath55 with @xmath158 , and therefore only applies in the first instance to the case where @xmath37 has connected support and no zeros in the interior thereof .
we extend the result to more general @xmath159 by choosing a nonnegative @xmath160 with no zeros in the interior of its support , assumed to be connected , and which is equal to unity on the support of @xmath37 . applying flanagan s result to @xmath161 , we may take the limit @xmath162 to obtain eq .
( cf . cor .
a.2 in , where the notation @xmath163 is used ) . ] , where @xmath57 is the two - velocity of @xmath10 , @xmath164 is its acceleration and @xmath165 is the scalar curvature on @xmath5 .
[ note that @xcite uses conventions in which @xmath166 for timelike @xmath57 ; the bound is therefore modified slightly . ] as we now describe , flanagan s bound is a locally covariant absolute qei . given @xmath55 , @xmath10 and @xmath37 as above , we may define a compactly supported distribution @xmath167 acting on smooth second rank covariant tensor fields @xmath168 , by @xmath169 our set of sampling tensors @xmath170 ( ` conf ' abbreviating ` conformal ' ) will be the set of all distributions formed in this way .
[ a distribution of the form @xmath167 is singularly supported on the curve @xmath10 ; we could also write it in the form @xmath171 where @xmath172 is the @xmath173-function at @xmath174 , obeying @xmath175 . ] the qei bound @xmath176 is then defined by @xmath177\,d\tau\ \label{eq : qdef_2d}\ ] ] for any @xmath55 , @xmath10 , @xmath37 for which @xmath178 .
for this to make sense , we must ensure that the right - hand side is unchanged if we replace @xmath55 , @xmath10 and @xmath37 by @xmath179 , @xmath180 and @xmath181 such that @xmath182 .
since @xmath10 and @xmath183 are both assumed to be parametrised by proper time , our two sampling tensors must be related in a simple way : @xmath184 is the translation of @xmath185 by some @xmath13 , so that @xmath186 , @xmath187 for all @xmath188 .
the only possible ambiguity stems from the fact that @xmath189 and @xmath190 mught differ by a relative sign which can change at zeros of @xmath181 of infinite order .
however , it is simple to show that , nevertheless , @xmath191 , and want to show that @xmath191 for all @xmath56 . differentiating and squaring yields @xmath192
from which it follows that @xmath191 except perhaps at zeros of @xmath189 .
if @xmath181 vanishes in a neighbourhood of @xmath56 then so must @xmath193 and the result holds trivially . for the remaining case , choose a sequence @xmath194 with @xmath195 ; since @xmath196 , we conclude the required result by continuity as @xmath197 . ] , ensuring that the right - hand side of eq .
is unchanged under the reparametrization of @xmath52 . the bound eq .
now takes the form of eq .
, so it remains only to verify that @xmath170 and @xmath176 have the required transformation properties .
suppose @xmath125 is a causal isometric embedding .
the push - forward @xmath99 acts on @xmath198 so that , for any smooth tensor field @xmath199 on @xmath0 , we have @xmath200 now the image curve @xmath201 can certainly be enclosed in a c.e.g.h.s . of @xmath0 which is conformal to the whole of minkowski space : namely the image under @xmath11 of that which enclosed @xmath10 . moreover ,
the image curve has velocity @xmath202 .
it is therefore clear that @xmath203 is a legitimate sampling tensor in @xmath204 , so we have shown that @xmath205 .
it is obvious that @xmath206 because all quantities involved in the bound are invariant under the isometry .
we have thus shown that two - dimensional massless fields obey a locally covariant absolute qei .
one need not restrict to worldline averages such as those described above : see @xcite for averages along spacelike or null curves , and for worldvolume averages [ in minkowski space ] .
we summarise as follows : [ thm:2d_abs_qwei ] let @xmath5 be a two - dimensional globally hyperbolic spacetime and let @xmath87 be the class of hadamard states of the massless klein
gordon field on @xmath5 .
let @xmath170 consist of all sampling tensors of the form eq . where ( i ) @xmath207 is a smooth future - directed timelike curve parametrised by proper time , with velocity @xmath208 ; ( ii ) @xmath10 which may be enclosed in a c.e.g.h.s . of @xmath5
globally conformal to the whole of minkowski space ; ( iii ) @xmath159 . then , defining @xmath209 by eq . for any @xmath55 , @xmath10 , @xmath37
for which @xmath178 , the absolute qwei @xmath210 holds for all @xmath118 and @xmath211 , and is locally covariant .
we now give two related examples of locally covariant difference qeis , based on methods first introduced in @xcite .
the first is a quantum null energy inequality ( qnei ) , constraining averages of the null - contracted stress - energy tensor along timelike curves , while the second is a quantum weak energy inequality ( qwei ) , constraining averages of the energy density along timelike curves @xcite .
suppose that @xmath5 is any globally hyperbolic spacetime of dimension @xmath1 , and @xmath212 is any smooth , future - directed timelike curve .
suppose further that @xmath213 is a smooth nonzero null vector field defined near @xmath10 .
then for any smooth , real - valued @xmath37 , compactly supported in @xmath55 , there is a difference qnei , @xmath214k^ak^bg(\tau)^2\,d\tau \ge -\int_0^{\infty}\frac{d\alpha}{\pi } \widehat{f}_{\gamma , g,{{\sf k}},\omega_0}(-\alpha,\alpha ) \label{eq : qnei}\ ] ] for all @xmath131 , where the hat denotes fourier transform and @xmath215 in which we have written @xmath216 for @xmath217 .
[ more precisely , the last factor is a distributional pull - back of the differentiated two - point function .
we also adopt the nonstandard convention @xmath218 for fourier tranforms ; for purposes of comparison , we note that the same convention was used in @xcite , but not in . ] the integral on the right - hand side of eq
. is finite as a consequence of @xmath65 being hadamard .
we emphasise that there is no necessity for @xmath14 and @xmath65 to be represented as vectors or density matrices in a common hilbert space representation in order to prove the qeis described in this section , because the proof may be phrased entirely in the algebraic formulation of qft .
the above result was derived in based on an earlier result in @xcite , described below .
however , it is slightly easier to show that it is locally covariant , which is why we have presented it first . to accomplish our task ,
we define @xmath219 to consist of all compactly supported distributions @xmath220 on smooth second rank covariant tensor fields @xmath168 on @xmath221 , such that @xmath222 for @xmath10 , @xmath213 , @xmath37 obeying the conditions already mentioned in this subsection _ and _ with @xmath37 having connected support with no zeros of infinite order in its interior , for reasons to be explained shortly .
we write @xmath223 for the set of functions @xmath37 of this type . as in the two - dimensional case
it is clear that the assignment @xmath224 is covariant in the required sense .
the qei bound is then defined by setting @xmath225 equal to minus the right - hand side of eq .
, for any @xmath55 , @xmath10 , @xmath226 and @xmath37 such that @xmath227 .
the particular parametrisation is not important , for reasons similar to those explained in the previous subsection .
however here it is important that @xmath228 : otherwise we could change @xmath37 to @xmath229 with @xmath230 changing sign from @xmath231 to @xmath232 at a zero of @xmath37 of infinite order , say at @xmath13 ; although @xmath233 , the two functions @xmath234 and @xmath235 differ when , for example , @xmath236 .
( the restriction to @xmath223 is not , however , very significant because it is dense in @xmath237 , as is shown in appendix [ appx : smooth ] . ) finally , the covariance property eq . follows because eq .
( for the case @xmath238 ) implies @xmath239 we summarise what has been proved .
[ thm:4d_qnei ] let @xmath5 be a globally hyperbolic spacetime of dimension @xmath1 and let @xmath87 be the class of hadamard states of the klein
gordon field of mass @xmath12 on @xmath5 .
let @xmath219 consist of all sampling tensors of the form eq . where @xmath207 is a smooth future - directed timelike curve parametrised by proper time
, @xmath226 is a smooth nonzero null field defined near the track of @xmath10 and @xmath228 . for each @xmath240 and reference state @xmath241 define @xmath242 for any @xmath55 , @xmath10 , @xmath226 , @xmath37 with @xmath227
. then the difference qnei @xmath214k^ak^bg(\tau)^2\,d\tau
\ge -{{\mathcal q}}^{\rm null}_{{\boldsymbol{m}}}({{\sf f}},\omega_0)\ ] ] holds for all @xmath131 and @xmath243 , and is locally covariant .
our second example of a locally covariant difference qei constrains the energy density .
we keep @xmath10 and @xmath37 as before , but replace @xmath213 by the velocity @xmath57 of the trajectory .
then the following difference qwei holds for all @xmath131 @xcite : @xmath214u^au^bg(\tau)^2\,d\tau \ge -\int_0^{\infty}\frac{d\alpha}{\pi } \widehat{g}_{\gamma , g,{{\boldsymbol{e}}},\omega_0}(-\alpha,\alpha ) \label{eq : qwei}\ ] ] where @xmath244\ ] ] and @xmath245 is a smooth @xmath60-bein defined in a neighbourhood of @xmath10 with @xmath246 on @xmath10 .
the frame @xmath247 adds a new ingredient to the discussion of covariance , which was not explored in @xcite .
subject to the condition @xmath248 , any choice of @xmath247 will give a qei bound , which may have differing numerical values .
when considering a causal isometry @xmath125 , we must therefore find a way of choosing frames in the two spacetimes so as to give equal values to the qwei bound , in accordance with covariance .
one solution would be to incorporate the frame as part of the data in the qwei , [ i.e. , writing @xmath249 and using the push - forward @xmath250 on @xmath251 but this seems rather inelegant .
fortunately , a better solution is at hand : it turns out that we can covariantly specify a subclass of frames guaranteed to yield the same numerical bound .
this is accomplished by requiring , in addition to @xmath248 , that the @xmath60-bein @xmath247 be invariant under fermi walker transport along @xmath10 , i.e. , @xmath252 for each @xmath253 , where @xmath254 is the acceleration of @xmath10 .
if @xmath255 is another @xmath60-bein also invariant under fermi walker transport and with @xmath256 , then it must be that @xmath255 is related to @xmath247 by a rigid rotation along @xmath10 , i.e. , @xmath257 for some fixed @xmath258 , because fermi
walker transport preserves inner products .
it is now easy to see that @xmath259 , because the form of @xmath247 _ off _ the curve @xmath10 is irrelevant , provided it is smooth .
accordingly this qei depends only on the smearing tensor @xmath52 [ defined by analogy with eq . ] and the reference state .
we emphasise that this is only one method of constructing a locally covariant bound in this setting , and others may be convenient in other contexts .
for example , it would be possible to simply take the infimum of the bound over all @xmath60-beins with @xmath260 ; this is certainly locally covariant , but impractical for calculational purposes . with this detail addressed , it is now straightforward to show that this qei is locally covariant by exactly the same arguments as used in the null - contracted case , and the additional observation that @xmath250 is fermi walker transported along @xmath201 if @xmath247 is along @xmath10 .
again , we summarise what has been established .
[ thm:4d_qwei ] let @xmath5 be a globally hyperbolic spacetime of dimension @xmath1 and let @xmath87 be the class of hadamard states of the klein
gordon field of mass @xmath12 on @xmath5 .
let @xmath261 consist of all sampling tensors of the form eq . where @xmath207 is a smooth future - directed timelike curve parametrised by proper time and with velocity @xmath208 , and @xmath228 .
for each @xmath262 and reference state @xmath241 define @xmath263 for any @xmath55 , @xmath10 , @xmath37 with @xmath178 , and any smooth tetrad @xmath247 defined near the track of @xmath10 with @xmath264 and which is invariant under fermi
walker transport along @xmath10
. then the difference qwei @xmath214u^au^bg(\tau)^2\,d\tau
\ge -{{\mathcal q}}^{\rm weak}_{{\boldsymbol{m}}}({{\sf f}},\omega_0)\ ] ] hold for all @xmath131 and @xmath265 , and is locally covariant .
most cases considered in the sequel will actually involve averages in static spacetimes along timelike curves which are static trajectories ( i.e. , orbits of a hypersurface orthogonal timelike killing field @xmath266 ) and with @xmath65 chosen to be a static hadamard state ( with respect to the same killing field ) . in these cases the bounds derived above simplify considerably , because the two - point function of @xmath65 obeys @xmath267 for any @xmath32 , @xmath268 , where @xmath27 is the one - parameter group of isometries obtained from @xmath266 .
we fix a particular orbit @xmath269 , which may be assumed to be a proper - time parametrisation [ as @xmath270 is constant along @xmath10 and may be set equal to unity ] .
then the two - point function , restricted to @xmath10 , can be expressed as @xmath271 where @xmath272 .
the same time - translational invariance is obtained for derivatives @xmath273 , provided that @xmath274 is invariant under the killing flow , or equivalently , has vanishing lie derivative with respect to @xmath266 on @xmath10 , i.e. , @xmath275 .
this simplifies the qwei bound as follows . if @xmath247 is lie - transported along @xmath10 then @xmath276 holds for some ` single variable ' distribution @xmath277 ; moreover , @xmath247 is also invariant under fermi
walker transport along @xmath10 ( owing to hypersurface orthogonality of @xmath266 . , and note that the curve @xmath10 has acceleration @xmath278 .
suppose @xmath279 .
then @xmath280 , which permits us to write the fermi
walker derivative as @xmath281}\log f)$ ] .
but this vanishes for hypersurface orthogonal @xmath266 ; see , e.g. , appendix c.3 in @xcite . ] ) then , as shown in , the qei eq
. becomes @xmath214u^au^bg(\tau)^2\,d\tau \ge -\int_{-\infty}^{\infty } du\ , \left| \widehat{g}(u )
\right|^2 q_{\gamma,\omega_0}(u)\ , , \label{eq : qwei_withq}\ ] ] where @xmath282 is a positive polynomially bounded function defined by @xmath283 additionally , if @xmath65 is a ground state ( as was the case in @xcite ) one may show that @xmath284 for @xmath285 , and so the function @xmath286 is supported on the positive half - line only .
more generally , it is always the case that @xmath287 decays rapidly as @xmath288 , so @xmath286 is always well - defined .
technically , @xmath287 is a measure , and may have @xmath173-function spikes which would exhibit themselves as discontinuities in @xmath282 .
since we define @xmath282 as an integral over the open interval @xmath289 , it is continuous from the left .
a similar analysis holds for the qnei eq .
, provided that the null vector field @xmath226 has vanishing lie derivative along @xmath10 , @xmath290 , because we have @xmath291 for some distribution @xmath292 . to conclude this section , we mention that more general qei bounds may be constructed along similar lines , based on other decompositions of the contracted stress energy tensor as a sum of squares .
this includes bounds averaged over spacetime volumes , see , e.g. @xcite .
however we will not need this generality here , and observe only that one would need to ensure that such decompositions are made in a canonical fashion to obtain a locally covariant bound .
in this section we develop some simple consequences of the qeis described in secs . [
sect : lcaqeis_examples ] and [ sect : lcdqeis_examples ] , specialised to minkowski space .
these will then be utilised in more general spacetimes using the local covariance properties of these bounds .
our results are obtained by converting qei bounds into eigenvalue problems which can then be solved .
for the most part , we will consider the scalar field of mass @xmath12 on @xmath60-dimensional globally hyperbolic spacetimes for @xmath1 ; special features of massless fields in two dimensions will be treated in sec . [
sect : gemassless2d ] .
accordingly , let @xmath0 be a @xmath60-dimensional globally hyperbolic spacetime , and let @xmath293 denote @xmath60-dimensional minkowski space . as illustrated in fig .
[ fig : mink_embed ] , let @xmath294 be a smooth , future - directed timelike curve , parametrised by proper time @xmath156 , and assume @xmath10 may be enclosed in a c.e.g.h.s .
@xmath3 of @xmath0 so that @xmath3 is the image of a c.e.g.h.s .
@xmath4 of @xmath293 under a causal isometric embedding @xmath138 .
thus the curve @xmath10 is the image of a curve @xmath295 in @xmath293 ; because @xmath11 is an isometry , @xmath296 is also a proper time parametrisation , and @xmath180 has the same proper acceleration as @xmath10 for each @xmath156 . given any @xmath228 , define a sampling tensor on minkowski space @xmath297 by @xmath298 on smooth covariant rank - two tensor fields @xmath168 on @xmath293 , where @xmath299 is the velocity of @xmath180 .
[ recall that @xmath228 means that @xmath37 is a real - valued smooth function whose support is compact , connected and contained in @xmath55 , and that @xmath37 has no zeros of infinite order in the interior of its support .
] under the isometry , @xmath52 is mapped to @xmath300 , with action @xmath301 where @xmath168 is now any smooth covariant rank-@xmath302 tensor field on @xmath0 .
applied to the stress - energy tensor , @xmath300 therefore provides a weighted average of the energy density along @xmath10 .
our aim is to place constraints on these averages using the locally covariant difference qwei given in theorem [ thm:4d_qwei ] . by local covariance ,
[ cor : diffmink ] guarantees that @xmath303 where @xmath304 is the minkowski vacuum state
. we will be particularly interested in the least upper bound of the energy density along @xmath10 , @xmath305 since the energy density is smooth , this value must be the maximum value taken by the field on the closure of the track of @xmath10 . using the trivial estimate @xmath306 for each @xmath156
, we have @xmath307 and , putting this together with eq .
, we obtain the inequality @xmath308 which holds , in the first place , for all @xmath228 . in the next two subsections we will analyse this in two special cases : namely , inertial motion and uniform acceleration .
when @xmath10 is inertial the qwei of theorem [ thm:4d_qwei ] takes the simpler form described in eqs . and above : @xmath309 where @xmath310 and the constant @xmath311 is @xmath312 , where @xmath313 is the area of the unit @xmath314-sphere .
( notation varies slightly from that used in . ) for all @xmath315 , it is clear that @xmath316 for all @xmath317 , while one may show that @xmath318 on the same domain and that the maximum of this expression on @xmath319 occurs at the ( unique ) solution to @xmath320 , which is numerically @xmath321 .
now @xmath322 , so we find that @xmath323 on @xmath324 as claimed . ] .
using these results , we may estimate eq . rather crudely by @xmath325 with @xmath326 for @xmath315 and @xmath327 . note that we have made two changes here :
( a ) @xmath328 has been replaced by unity ; ( b ) the lower integration limit @xmath329 has been replaced by zero . we now specialise to even dimensions @xmath330 , @xmath331 .
because @xmath37 is real - valued , @xmath332 is even and we may write @xmath333 where @xmath334 is the differential operator @xmath335 and we have used parseval s theorem , and the fact that @xmath37 vanishes outside @xmath55 . inserting the above in eq . , we have shown that @xmath336 obeys the inequality @xmath337 for all @xmath228 .
the class @xmath223 is inconvenient to work with directly ; fortunately , the same inequality holds for general @xmath58 , as we now show .
first , any @xmath159 is the limit of a sequence of @xmath338 for which @xmath339 and @xmath340 in @xmath341 ( see appendix [ appx : smooth ] ) . applying the above inequality to each @xmath342
, we may take the limit @xmath197 to conclude that it holds for @xmath37 as well .
having established the result for arbitrary real - valued @xmath58 , we extend to general complex - valued @xmath37 by applying it to real and imaginary parts separately , and then adding . accordingly the inequality eq .
holds for all @xmath58 . integrating by parts @xmath314 times , and noting that no boundary terms arise because @xmath37 vanishes near the boundary @xmath343 of @xmath55 , eq . may be rearranged to give @xmath344 where @xmath345 denotes the usual @xmath346-inner product on @xmath55 , and the operator @xmath347 on @xmath348 .
our aim is now to minimise the right - hand side over the class of @xmath37 at our disposal ( excluding the identically zero function ) . now
the operator @xmath349 is symmetric is symmetric on a domain @xmath350 if @xmath351 for all @xmath352 , which shows that the adjoint @xmath353 agrees with @xmath354 on @xmath350 , but does not exclude the possibility that @xmath353 has a strictly larger domain of definition than @xmath354 . ] and positive , i.e. , @xmath355 for all @xmath58 . by theorem x.23 in
@xcite ) , the solution to our minimisation problem is the lowest element @xmath356 of the spectrum of @xmath357 , the so - called _ friedrichs extension _ of @xmath349 .
this is a self - adjoint operator with the same action as @xmath349 on @xmath348 , but which is defined on a larger domain in @xmath341 . in particular
, every function in the domain of @xmath357 obeys the boundary condition @xmath358 at @xmath359 .
( see , where the technique of reformulating quantum energy inequalities as eigenvalue problems was first introduced , and which contains a self - contained exposition of the necessary operator theory . )
one may think of this as a precise version of the rayleigh
ritz principle .
once we have determined @xmath356 , we then have the bound @xmath360 so the problem of determining the lower bound is reduced to the analysis of a schrdinger - like equation , subject to the boundary conditions mentioned above .
the two examples of greatest interest to us are @xmath361 and @xmath362 , representing two- and four - dimensional spacetimes . starting with @xmath361 ,
let us suppose that @xmath55 is the interval @xmath363 for some @xmath364 .
we therefore solve @xmath365 subject to dirichlet boundary conditions at @xmath366 ; as is well known , the lowest eigenvalue is @xmath367 and corresponds to the eigenfunction @xmath368 .
[ a possible point of confusion is that , if @xmath37 is extended so as to vanish outside @xmath55 , it will not be smooth .
however there is no contradiction here : the point is that the infimum is not attained on @xmath348 . ]
thus we have @xmath369 because @xmath370 [ by convention , the zero - sphere has area @xmath371 .
we may infer , without further calculation , that the bound must be zero if @xmath372 , because ( returning to the ritz quotient eq . ) , the infimum over all functions in @xmath373 must be less than or equal to the infimum over all functions in @xmath348 for any bounded @xmath55 ( a similar argument applies to the semi - infinite case ) .
thus @xmath356 can be no greater than zero ; on the other hand , the minimum can not be negative either , because the original functional is nonnegative . accordingly eq .
holds in all cases , with @xmath13 equal to the length of the interval @xmath55 . in the four - dimensional case @xmath362
, we proceed in a similar way , solving @xmath374 subject to @xmath375 at @xmath343 . in the case where @xmath55 is bounded , @xmath376 [ without loss of generality ]
, the spectrum consists only of positive eigenvalues .
it is easy to see that the solutions to the eigenvalue equation @xmath377 are linear combinations of trigonometric and hyperbolic functions .
the lowest eigenfunction solution which obeys the boundary conditions is @xmath378 where @xmath379 is the minimum positive solution to @xmath380 since @xmath381 , we obtain @xmath382 if @xmath55 is semi - infinite or infinite , we may argue exactly as in the two - dimensional case that the bound vanishes , in agreement with the formal limit @xmath383 .
clearly this approach will give similar results in any even dimension , with a consequent increase in complexity in solving the eigenvalue problem .
nonetheless , it is clear that the resulting bound will always scale as @xmath62 . in fact , this is even true in odd spacetime dimensions , where the eigenvalue problem would involve a nonlocal operator and is not easily tractable .
we summarise what has been proved so far in the following way .
[ prop : emaxmassive ] let @xmath0 be a globally hyperbolic spacetime of dimension @xmath1 and suppose that a timelike geodesic segment @xmath10 of proper duration @xmath13 may be enclosed in a c.e.g.h.s . of @xmath0 which is causally isometric to a c.e.g.h.s . of minkowski space @xmath293 ,
then @xmath384 for all hadamard states @xmath14 of the klein gordon field of mass @xmath12 on @xmath0 .
the constants @xmath385 depend only on @xmath60 .
in particular , @xmath386 , while @xmath387 .
* remark : * when the field has nonzero mass , we can expect rather more rapid decay than given by this estimate . to see why , return to the argument leading to eq . .
if we reinstate @xmath329 as the lower integration limit , we have @xmath388 suppose for simplicity that @xmath376 .
if we write @xmath389 , for @xmath390 , a change of variables yields @xmath391 where the nonnegative quantity @xmath392 decays rapidly as @xmath393 , owing to the rapid decay of @xmath394 .
thus the estimate eq .
is quite crude when @xmath395 ; it is hoped to return to this elsewhere .
equipped with prop .
[ prop : emaxmassive ] , we may now address the first two examples presented in the introduction .
first , the proposition asserts that no hadamard state can maintain an energy density lower than @xmath396 for proper time @xmath13 along an inertial curve in a minkowskian c.e.g.h.s . of @xmath0 .
in particular , this justifies the claim made in example 1 in the introduction .
our bounds clearly depend only on @xmath13 , which in turn is controlled by the size of the minkowskian region @xmath3 . by choosing the curve @xmath10 and @xmath3 in an appropriate way
, fairly simple geometrical considerations can thus provide good _ a priori _ bounds on the magnitude and duration of negative energy density .
a good illustration is the following ( which includes example 2 in the introduction ) .
suppose that a @xmath60-dimensional globally hyperbolic spacetime @xmath0 with metric @xmath397 is stationary with respect to timelike killing vector @xmath18 and admits the smooth foliation into constant time surfaces @xmath398 .
suppose there is a ( maximal ) subset @xmath21 of @xmath22 , with nonempty interior , for which @xmath397 takes the minkowski form on @xmath399 .
choose any point @xmath400 in @xmath401 , with @xmath30 and suppose that we may isometrically embed a euclidean @xmath402-ball of radius @xmath403 in @xmath21 , centred at @xmath23 ( see fig .
[ fig : example2 ] ) .
then the interior of the double cone @xmath404 is a c.e.g.h.s . of @xmath0 which is isometric to a c.e.g.h.s .
of minkowski space , and contains an inertial curve sgement @xmath405 parametrised by the interval @xmath406 of proper time .
any hadamard state @xmath14 on @xmath0 therefore obeys @xmath407 along @xmath10 .
writing @xmath24 for the minimum distance from @xmath23 to the boundary of @xmath21 , it is clear that this inequality holds for all @xmath408 and hence , by continuity , for @xmath409 . moreover , if the state is stationary [ for example , if it is the ground state ] , then the energy density takes a constant value along @xmath10 and we obtain @xmath410 for any @xmath30 , where @xmath31 is the unit vector along @xmath18 . in this way
we obtain a universal bound on the fall - off of negative energy densities in such spacetimes , which could be used to provide a quantitative check on exact calculations , if these are possible , or to provide some precise information in situations where they are not .
the bound is of course very weak close to the boundary of @xmath21 : this does not imply that the energy density diverges as this boundary is approached , of course , but merely indicates that it would not be incompatible with the quantum inequalities for there to exist geometries on @xmath411 for which the stationary energy density just outside might be very negative . to conclude this subsection ,
let us briefly discuss the null - contracted qei eq . in the present context . for simplicity , we restrict ourselves to four dimensions .
suppose @xmath412 is a nonzero null vector field which is covariantly constant along @xmath180 , so , in particular , @xmath413 is also constant on @xmath180 .
our sampling tensor is now defined to be @xmath414 with action @xmath415 on smooth covariant rank-@xmath302 tensor fields @xmath168 on @xmath416 . in exactly the same way as for the qwei discussed above , we may apply local covariance to the qnei of thm .
[ thm:4d_qnei ] , so yielding @xmath417 where , as shown in , @xmath418 for the massless scalar field ( and in fact this bound also constrains the massive field too ) .
this differs from the corresponding qwei by a factor of @xmath419 [ recall that @xmath381 ] , so we may immediately deduce the following result . [ prop : nullmax4d ] let @xmath0 be a four - dimensional globally hyperbolic spacetime and suppose that a timelike geodesic segment @xmath10 of proper duration @xmath13 may be enclosed in a c.e.g.h.s . of @xmath0 which is causally isometric to a c.e.g.h.s .
of minkowski space
. if @xmath213 is a covariantly constant null vector field on @xmath10 then we have @xmath421 for any hadamard state @xmath14 of the klein
gordon field , where @xmath422 .
this result justifies the claim made above eq .
( 38 ) of , where an application is presented .
we now turn to the case where @xmath10 has uniform constant proper acceleration @xmath36 . for simplicity
we consider only massless fields in four dimensions , but expect similar results in more general cases .
we need to estimate @xmath423 where @xmath52 is supported on the uniformly accelerated worldline @xmath180 in @xmath416 .
it will be convenient to drop the tilde from @xmath180 and the subscript from @xmath416 . without loss of generality
, we may assume @xmath207 is parametrized so that @xmath424 where @xmath425 .
the first step in our calculation is to set up an orthonormal tetrad field surrounding the worldline , @xmath426 which , satisfies the two properties required : namely , that @xmath427 agrees with the velocity @xmath57 on @xmath10 , and that the frame is invariant under fermi
walker transport along @xmath10 .
the required bound is then given by @xmath428 where @xmath429 we evaluate this quantity in stages , beginning by noting that @xmath430 w_{{\boldsymbol{m}}}^{(2)}(x , x')\ , , \label{eq : rindler_edensity}\end{aligned}\ ] ] where @xmath431 where @xmath432 is the limit ( in the distributional sense ) as @xmath433 of @xmath434 thus we are in the situation of eq .
, and the bound becomes @xmath435 where @xmath436 to obtain the required fourier transform , we first use contour integration $ ] of the real axis , and its translate @xmath437 + 2\pi\xi_o i$ ] .
the contour encloses a single pole , of fourth order , at @xmath438 and the contribution of the ` short ' sides vanishes as @xmath439 .
one also exploits the fact that the contributions from the two ` long ' sides are equal up to a factor of @xmath440 . ] to find @xmath441 which decays exponentially as @xmath442 , provided @xmath443 . taking the limit @xmath433 it is easy to check that @xmath444 note that this fourier transform has support on the whole real line , not just the positive half line .
thus @xmath445 our aim is now to estimate @xmath446 in order to obtain a bound which may be analysed by eigenvalue techniques as in the previous subsection .
beginning in the half - line @xmath447 , we may estimate @xmath448 since @xmath446 is everywhere increasing . on the other hand , for @xmath449
, we may split the integral into @xmath450 and the contribution from @xmath451 $ ] to give @xmath452\ ] ] after rearranging .
now the last integral is increasing in @xmath453 , so we may bound it by its limit as @xmath454 , to yield @xmath455 for @xmath456 . using the estimates
eqs . and , and the fact that @xmath457 is even , @xmath458 applying parseval s theorem
, we arrive at @xmath459 and , together with eq . , we now have @xmath460 for any @xmath228 . as in the previous subsection
, we may extend this inequality to arbitrary @xmath58 , and then optimise over this class .
this leads to the conclusion that @xmath461 where @xmath462 is the lowest ( positive ) eigenvalue for the equation @xmath463 on @xmath55 , subject to boundary conditions @xmath375 at @xmath343 .
let us suppose that @xmath55 is bounded , writing @xmath376 without loss of generality .
it is convenient to write @xmath464 for then the eigensolutions must be scalar multiples of @xmath465 where @xmath466 and @xmath467 solves @xmath468 we shall denote the minimum solution to this equation in @xmath469 by @xmath356 ( see fig . [
fig : accn_evals ] ) ; clearly @xmath356 depends only on the ratio of the sampling time @xmath13 to the acceleration scale @xmath470 .
two limits are of interest .
firstly , when @xmath471 , one may show that @xmath472 where @xmath473 is as in eq . .
thus we regain the usual short - timescale constraint eq . .
this supports the ` usual assumption ' ( see @xcite for references ) that sampling at scales shorter than those determined by the acceleration or curvature is governed by the bound obtained for inertial curves in minkowski space . on the other hand , if we take @xmath474 , we see that @xmath475 , so @xmath476 in this limit .
in fact , more can be said for the inextendible case @xmath477 , because the approximations made to gain eq .
are rather wasteful in this limit .
choose any @xmath478 with @xmath479 and define @xmath480 , denoting the corresponding sampling tensor @xmath481 .
then a simple change of variables argument applied to eq .
shows that latexmath:[\[{{\mathcal q}}^{\rm weak}_{{\boldsymbol{m}}}({{\sf f}}_{\tau_0},\omega_{{\boldsymbol{m}}})=\int_{-\infty}^\infty dv\ , @xmath383 may be taken under the integral sign to yield @xmath483 in particular , if @xmath37 has unit @xmath346-norm , we have @xmath484 where @xmath485 is the proper acceleration of the curve , as asserted in example 3 in the introduction
. thus long term averages of the energy density measured along the curve are bounded from below , and no energy density can be less than this bound over the entire worldline .
this is an improvement by a factor of @xmath486 over the bound given in eq . .
using a more refined analysis one could presumably extract it as the limit of a result for general @xmath13 , but we will not pursue this here . to summarise , we have reached the following conclusions .
[ prop : emax_4d_accel ] let @xmath0 be a four - dimensional globally hyperbolic spacetime containing a timelike curve @xmath10 of proper duration @xmath13 and constant proper acceleration @xmath36 . if @xmath10 may be enclosed in a c.e.g.h.s . of @xmath0 which is causally isometric to a c.e.g.h.s.of minkowski space
then we have @xmath487 for any hadamard state @xmath14 of the massless klein
gordon field , where @xmath356 is the smallest solution to eq . in @xmath324 and depends on @xmath488 .
if @xmath10 has infinite proper duration , we also have the more stringent constraint eq . .
so far , we have only utilised the locally covariant difference qeis of sec .
[ sect : lcdqeis_examples ] .
for massless fields in two dimensions , however , we also have the absolute qei developed by flanagan and others , described in sec .
[ sect : lcaqeis_examples ] , which are also known to be optimal bounds . in this subsection
we briefly discuss how the results of the previous subsections may be sharpened and generalised in this context .
in fact the formula for the qei bound is sufficiently simple that we may work directly in curved spacetime , rather than in minkowskian subregions .
let @xmath489 be a smooth future - directed timelike curve , with velocity @xmath57 and accleration @xmath164 in a two - dimensional globally hyperbolic spacetime @xmath0 .
as before , @xmath55 is an open interval of proper time . in order to apply flanagan s bound , we make the additional assumption that @xmath10 may be enclosed within a c.e.g.h.s .
@xmath3 of @xmath0 , which is globally conformal to the whole of minkowski space .
then flanagan s qei asserts that @xmath490\,d\tau\ ] ] for all hadamard states @xmath14 and any smooth , real - valued @xmath37 compactly supported in @xmath55 , i.e. , @xmath159 .
we proceed as above , obtaining the estimate @xmath491\,d\tau\,,\ ] ] for all @xmath159 , where @xmath492 as usual . converting to an eigenvalue problem
, we deduce that @xmath493 where @xmath356 is the lowest element in the spectrum of the friedrichs extension of the operator @xmath494 on @xmath348 .
provided @xmath495 and @xmath496 are bounded along @xmath10 , the correct boundary conditions are dirichlet conditions @xmath497 on @xmath343 ( see e.g. , ) .
we now give two illustrative examples .
[ prop : emax2d ] suppose @xmath0 is a globally hyperbolic two - dimensional spacetime with a c.e.g.h.s .
@xmath3 which is globally conformal to the whole of minkowski space .
then the following hold for all hadamard states @xmath14 on @xmath0 : + ( a ) if @xmath10 is a curve of proper duration @xmath13 contained in @xmath3 , with @xmath498 constant along @xmath10 , then @xmath499 ( b ) if @xmath35 has ( signed ) proper acceleration growing linearly with proper time , @xmath500 , and @xmath501 on @xmath10 , then @xmath502 the proof is straightforward : for ( a ) , the eigenvalue problem is @xmath503 on an interval of length @xmath13 subject to dirichlet boundary conditions , which easily yields the stated result . for ( c )
, we may choose the origin of proper time so that @xmath504 for some constant @xmath505 .
the eigenvalue problem is then @xmath506 which is the harmonic oscillator equation [ and the friedrichs extension is also the standard harmonic oscillator hamiltonian ] .
the minimum value of @xmath507 is therefore the ` zero - point ' value @xmath508 .
[ the comparison with the usual quantum mechanical harmonic oscillator would correspond to units in which the mass and planck s constant are both set to @xmath302 . ]
thus we obtain the required result .
in this section we illustrate our general method by some concrete calculations in a variety of locally minkowskian spacetimes in both two and four dimensions .
for the most part , we focus on the lower bounds , but upper bound calculations are included where they are enlightening . for each spacetime
we consider , exact values of the renormalised stress - energy tensor are known ( or easily obtained from existing results ) for one or more states .
this permits comparison with the results of our method .
consider the massless scalar field on the two - dimensional timelike cylinder , @xmath509 , i.e. , minkowski space @xmath72 quotiented by the group of translations @xmath510 ( @xmath511 ) .
the casimir vacuum @xmath512 is the ground state of the scalar field on @xmath509 ( more precisely , it is a state on the algebra of first derivatives of the field we will ignore this subtlety , which does not modify any of our conclusions below ) .
the renormalized expectation value of the vacuum stress - tensor has the form @xmath513 where @xmath514 is a constant .
our aim is to use quantum inequalities to provide upper and lower bounds on @xmath514 .
the value of @xmath515 is , of course , well known , and will satisfy the bounds we now derive ; our aim is to demonstrate how it may be bounded without direct calculation . in order to apply our method , we must identify suitable globally hyperbolic subspacetimes of @xmath509 . for any @xmath516
, we may define a timelike geodesic @xmath517 by @xmath518 .
then the double cone @xmath519 is a causally embedded globally hyperbolic subspacetime of @xmath509 , containing @xmath10 . as this subspacetime is globally conformal to the whole of minkowski space and the energy density
is constant along @xmath10 , we have the lower bound @xmath520 from prop .
[ prop : emax2d](a ) ( in the case @xmath521 ) .
this bound clearly becomes more stringent as @xmath13 is increased , so we obtain the best bound possible ( within this method ) by taking @xmath522 . as shown in fig .
[ fig : cylinder ] , the corresponding diamond is one for which the corners of the diamond just barely fail to touch on the back of the cylinder .
this gives the final result @xmath523 diagram showing the `` largest '' causal diamond in two dimensional minkowski space that can be isometrically embedded into the two dimensional cylinder spacetime with periodicity @xmath349 in the @xmath524-direction . to an observer inside the diamond in the cylinder spacetime ,
the quantum field theory and states would be indistinguishable from that in minkowski space .
the dashed vertical line is the worldline of a stationary observer . ]
we now demonstrate how to find an upper bound on @xmath514 for which we must employ our locally covariant difference qwei .
let @xmath180 be the curve @xmath525 in @xmath72 and let @xmath526 , for some @xmath527 , which is a c.e.g.h.s . of @xmath72 .
then the quotient map @xmath528 defines a causal isometric embedding of @xmath4 in @xmath509 , with @xmath529 equal to the double cone constructed earlier in this subsection . by cor .
[ cor : diffmink ] we have @xmath530 for any sampling tensor @xmath531 .
we define @xmath52 by eq .
for @xmath532 and then use the constancy of the energy density along @xmath10 to find @xmath533 where the last inequality is derived in appendix [ sec : qi_cyl ] . as usual , this may be converted into an eigenvalue problem : here , @xmath534 where @xmath535 is the minimum eigenvalue of @xmath536 on @xmath537 subject to dirichlet boundary conditions . combining with our earlier lower bound , we thus have @xmath538 the known value of @xmath514 is exactly @xmath539 , see , which , remarkably , saturates the lower bound .
thus we have shown that , in the cylinder spacetime , the casimir vacuum energy density is the lowest possible static energy density compatible with the quantum energy inequalities .
this however is not always the case , which we will see in later examples . because the energy density is in fact negative
, the upper bound was not particularly enlightening in this example .
however the situation is different for thermal equilibrium states .
let @xmath540 be the thermal equilibrium ( kms ) state at inverse temperature @xmath473 , relative to the static time translations .
the stress - energy tensor is again diagonal @xmath541 where @xmath542 see , e.g. , sec .
4.2 of . by our general theory ,
these states should be constrained by the same lower bound as before , and this is evidently true , because the series contribution to @xmath543 is clearly positive .
the upper bound depends on the temperature : @xmath544 for any @xmath532 . in appendix
[ sec : qi_cyl ] we obtain the estimate @xmath545 and we may now immediately optimise over @xmath37 using our result for the ground state to obtain @xmath546 as shown in fig . [ fig : thermal ] this is consistent with the known value of @xmath543 .
let us now consider various quotients of four - dimensional minkowksi space by subgroups of the group of spatial translations . to begin , consider the quotient of four - dimensional minkowski space , with inertial coordinates @xmath549 by the spatial translation subgroup @xmath550 ( @xmath511 ) for some fixed periodicity length @xmath551 .
we will denote the resulting spacetime by @xmath552 , and consider the ground state @xmath553 , which has a nonzero casimir vacuum stress - energy tensor .
a calculation using the method of images and the minkowski space vacuum two - point function yields the renormalized vacuum stress - tensor for the massless scalar field in this spacetime , ( see , e.g. , ) @xmath554.\ ] ] we will now show that this is consistent with the lower bound arising from the qeis . to this end , let @xmath555 for some fixed @xmath556 .
then the double cone @xmath557 is a c.e.g.h.s . of @xmath552 which is causally isometric to a double cone in minkowski space .
thus the portion of @xmath10 parametrised by @xmath558 meets the hypotheses of prop .
[ prop : emaxmassive ] and we have @xmath559 for any hadamard state @xmath14 of the klein gordon field . in particular , the state @xmath553 obeys this bound , as @xmath560 .
in fact the energy density is about thirty times smaller than the qei bound in this case .
thus the qei bounds can be rather weak .
but this is necessary , as can be seen from the next examples , in which the same lower bound must constrain a more negative energy density .
consider the spacetime @xmath561 , which may be obtained by quotienting @xmath552 by the translation group @xmath562 ( @xmath511 ) for some nonnegative @xmath563 , which , without loss of generality , we take to be no less than @xmath564 . because @xmath565 we may apply prop . [ prop : emaxmassive ] to a double cone of the same size as before , so the lower bound is unchanged .
however the stress tensor is now @xmath566\ ] ] in the special case @xmath567 .
the sum can no longer be given in closed form , but numerically the overall prefactor ( equal to the energy density on the worldline @xmath568 ) is given in as @xmath569 .
this is still consistent with eq .
, with energy density now only around ten times smaller than the bound . in exactly the same way we may quotient @xmath570 by the translation subgroup @xmath571 ( @xmath511 ) , thereby forming @xmath572 .
if we again suppose that @xmath573 , then the bound eq . still applies to the ground state on this spacetime .
( since this spacetime supports normalisable zero modes for the massless scalar field , one must regard this as a state on the algebra of derivatives of the field , much as for massless fields in two - dimensions ) .
on the other hand , the stress - energy tensor in the natural ground state is @xmath574\ ] ] in the special case @xmath575 .
the energy density along @xmath568 in this case is numerically computed to be @xmath576 , which is again consistent with the qei constraint eq . , which is now weaker by a factor of less than @xmath577 .
let us note that the massless qei bound also provides a lower bound on the ground state energy densities of _ massive _ scalar fields in these spacetimes .
consistency here is seen from the fact that the mass diminishes the magnitude of the energy density ( note the misprints in noted in @xcite which do not , however , affect the final result ) .
our third example concerns ( the globally hyperbolic portion of ) the misner universe @xmath578 ; namely , the quotient of @xmath579 with metric @xmath580 by the translation group @xmath581 ( @xmath511 ) for some constant @xmath582 .
that is , the @xmath583 coordinate has been compactified onto a circle .
we restrict to @xmath584 to avoid the closed null geodesics which would appear at @xmath585 and the closed timelike curves appearing for @xmath586 . under the coordinate transformation @xmath587 we may , equivalently ,
regard misner space as the wedge @xmath588 of minkowski spacetime with the points @xmath589 and @xmath590 identified for each @xmath511 .
the covering space of the misner universe is the wedge @xmath591 of minkowski spacetime , shown here in cross section in the @xmath592-@xmath593 coordinates .
points in the covering space are identified , as described in the text , along the background hyperbol .
also shown is a stationary geodesic in the misner universe , which in the covering space is a constant velocity observer ( dashed line ) .
for such an observer , the largest causal diamond for a given center point that is isomorphic to a subset of minkowski space is shown in gray . identified images of this diamond
are also shown in white . ]
define a curve @xmath594 in the original coordinates , for some constant @xmath595 .
this is a timelike geodesic , with velocity @xmath596 . in the minkowski space
cover , this worldline is given by @xmath597 \tau \sinh ( x_o^1)\\[3pt ] x_o^2\\[3pt ] x_o^3 \end{array}\right ) \qquad\mbox { with } \qquad u^a(\tau ) = \left(\begin{array}{c } \cosh ( x_o^1)\\[3pt ] \sinh ( x_o^1)\\[3pt ] 0 \\[3pt ] 0 \end{array}\right),\ ] ] which is a constant velocity geodesic , as shown by the bold dashed line in figure [ fig : misner ] .
let us consider a portion of this curve , running between @xmath598 and @xmath599 , which is such that @xmath600 is a c.e.g.h.s . of misner space isometric to a double cone in minkowski space .
assuming that this region is maximal , it must be that the geodesic joining the @xmath601 image @xmath602 of @xmath603 to @xmath604 is null . setting @xmath605 and @xmath606 ,
this yields the condition @xmath607
^ 2- \left[\left(\tau_o+\frac{l}{2}\right)\sinh(x_o^1 ) - \left(\tau_o-\frac{l}{2}\right)\sinh(x_o^1+a)\right]^2 = 0,\ ] ] which entails @xmath608 this is the largest double cone of this type , centered on @xmath609 , in which an observer can not detect the compactified nature of the @xmath583-direction .
we may therefore apply prop .
[ prop : emaxmassive ] to the portion @xmath610 of @xmath10 lying between @xmath603 and @xmath604 .
this gives @xmath611 for any hadamard state @xmath14 on misner space .
in particular , an energy density @xmath612 of the form @xmath613 , for which @xmath614 , would be subject to the constraint @xmath615 by adapting the eigenvalue method , we may obtain a better bound . let us suppose that @xmath616 obeys @xmath617 on @xmath618 . then by exactly the same arguments as in sec .
[ sect : ga_inertial ] , we may deduce @xmath619 where the @xmath620 in the denominator comes from the form of @xmath616 , and the infimum is taken over all @xmath58 .
the denominator can be reinterpreted as the norm of @xmath37 in @xmath621 .
integrating by parts twice , we may rewrite the numerator as @xmath622 where the inner product is that of @xmath621 and @xmath349 is defined on @xmath348 by @xmath623 and is symmetric , i.e. , @xmath624 for all @xmath625 .
the minimisation problem is then solved by finding the lowest spectral point of the friedrichs extension of @xmath349 .
it may be shown that the friedrichs extension again amounts to the imposition of dirichlet boundary conditions @xmath375 on @xmath359 , in the norm @xmath626 , which is equivalent to the norm of the sobolev space @xmath627 since @xmath620 is bounded and bounded away from zero on @xmath55 .
accordingly , the closure of @xmath348 is precisely @xmath627 and the desired domain lies in this sobolev space , all elements of which obey @xmath375 on @xmath343 . ] and the problem now reduces to the study of the ode @xmath628 again we wish to determine the minimum eigenvalue @xmath507 for eigensolutions that satisfy the boundary conditions .
the substitution @xmath629 converts the equation to a constant coefficient linear equation , and one may determine the general solution ( e.g. , using _ mathematica _ ) as @xmath630 where @xmath631 are constants .
imposing three of the boundary conditions fixes three of the constants in terms of the fourth , which serves as an overall magnitude for the test function .
the fourth boundary condition can then be used to determine the eigenvalues .
a somewhat involved calculation leads to the transcendental equation to determine @xmath507 implicitly in terms of @xmath632 : @xmath633 we denote @xmath507 , so determined , as @xmath634 ; this constrains our original value @xmath635 by @xmath636 our interest in energy densities proportional to @xmath620 stems from the state constructed by hiscock and konkowski . this quasifree state , which we denote @xmath637 , is obtained by applying the method of images to the minkowski space two - point function in the wedge @xmath588 , and then carrying it back to the original misner coordinates to find the renormalized vacuum expectation value of the stress - tensor .
hiscock and konkowski considered the conformally coupled scalar field , but their calculations can be easily reproduced in the minimally - coupled case , to yield @xmath638,\ ] ] where @xmath639 is a negative constant depending on the @xmath583-period @xmath632 .
, replaced the numerical coefficient @xmath640 in the stress - tensor with @xmath641 $ ] . ] both the coefficient @xmath640 and the numerical evaluation of the lower bound @xmath642 are plotted in fig .
[ fig : misner_bnd ] .
it is obvious that @xmath640 , and thus the energy density obey the qei constraint for all values of @xmath632 .
the bound eq .
is still weaker .
the rindler spacetime @xmath643 is the `` right wedge '' of minkowski space , i.e. , the region @xmath644 in inertial coordinates @xmath549 .
we may also make the coordinate transformation @xmath645 x = \xi \cosh(\eta ) , & \qquad\qquad & z = z , \end{array}\ ] ] to obtain the metric in the form @xmath646 with coordinate ranges @xmath647 , @xmath648 .
lines of constant @xmath649 , when mapped into minkowski space , are worldlines for observers undergoing constant proper acceleration @xmath650 .
rindler spacetime is static with respect to @xmath651 ( corresponding to lorentz invariance in the @xmath652 plane ) and is invariant under euclidean transformations of the @xmath653 plane .
clearly any line of constant @xmath649 meets the conditions of prop .
[ prop : emax_4d_accel ] and we may immediately read off that any static hadamard state @xmath14 on @xmath643 must obey @xmath654 where @xmath57 is the unit vector parallel to @xmath655 .
in particular , this provides a constraint on the energy density @xmath656 in the ground state @xmath657 ( which is hadamard ) .
this may also be computed exactly : it was first computed for the conformally coupled scalar field by candelas and deutsch and one can easily generalize their results to the minimally coupled scalar field to obtain , replace the @xmath658 in the numerator with @xmath659 . ]
@xmath660 which is exactly the lower bound given above .
thus , remarkably , the rindler ground state saturates the qei constraints , which were obtained using local covariance and the minkowski vacuum , and nowhere involved @xmath657 .
diagram showing rindler spacetime ( with the two perpendicular space dimensions suppressed ) embedded into minkowski spacetime .
the dashed hyperbolic line , the worldline of a constantly accelerating observer , is the image of a constant @xmath649 observer s worldline in rindler coordinates .
the grey diamond is a causal region that can be isometrically identified between the two `` different '' coordinate systems . ]
let us also examine how an upper bound might be obtained .
let @xmath661 in @xmath662 coordinates and set @xmath208 as usual .
we consider sampling along @xmath10 , with sampling tensors of form @xmath663 for @xmath664 . since the energy density is constant along @xmath10 , the upper bound of cor .
[ cor : diffmink ] gives @xmath665 the right - hand side can be read off from the difference qei derived by pfenning @xcite for the electromagnetic field , because the corresponding bound for the scalar field is exactly half of the electromagnetic expression : , rather than proper time @xmath56 : our @xmath666 is related to the @xmath667 of @xcite by @xmath668 . ]
@xmath669 next consider scaling the test function , replacing @xmath37 by @xmath670 .
we find , considering the scaling behavior of the above expression , @xmath671 .
thus we find consistency with the known fact that the expectation value of the rindler ground state is bounded above by zero , i.e. @xmath672 .
in this paper we have initiated the study of interrelations between quantum energy inequalities and local covariance .
we have formulated definitions of locally covariant qeis , and shown that existing qeis obey them , modulo small additional restrictions ( sec .
[ sec : qeis_and_lc ] ) . the main thrust of our work has been directed at providing _ a priori _ constraints on renormalised energy densities in locally minkowskian regions , accomplished in sec . [
sec : general_apps ] .
the simple geometric nature of these bounds makes them easy to apply in practice , and a number of future applications are envisaged . in particular , we will discuss applications to the casimir effect in a companion paper @xcite ; at the theoretical level , it is possible to place the present discussion in the categorical language of @xcite , and this will be done elsewhere .
equally important are the specific calculations reported in sec .
[ sec : specific_apps ] .
here we saw that , in some situations , the qei bounds give best - possible constraints on the energy density , and that typical ground state energy densities are not over - estimated by the qei bound by more than a factor of about @xmath673 at worst ( in the examples so far studied ) . finally , although we confined our attention largely to locally minkowskian spacetimes in secs .
[ sec : general_apps ] and [ sec : specific_apps ] , we emphasise again that other interesting cases may be studied using our general formalism , as , for example , in the work of marecki @xcite on spacetimes with locally schwarzschild subregions .
in this appendix we describe the construction of the quantised klein gordon field within the algebraic approach to quantum field theory , and explain the construction of pulled back states used in sec .
[ sect : localcovariance ] .
the free scalar field of mass @xmath12 may be quantised on any globally hyperbolic spacetime @xmath5 in the sense that one may construct a complex unital @xmath105-algebra @xmath674 whose elements may be interpreted as ` polynomials in smeared fields ' .
a typical element of the algebra is a complex linear combination of the identity @xmath675 and a finite number of terms each of which is a finite product of a number of objects @xmath88 where @xmath676 is a test function ( i.e. , smooth and compactly supported ) on @xmath221 .
the algebra also satisfies a number of relations : 1 .
@xmath677 2 .
@xmath678 3 .
@xmath679 4 .
@xmath680= ie_{{{\boldsymbol{m}}}}(f , g){\leavevmode\hbox{\rm{\small1\kern-3.8pt\normalsize1}}}$ ] for all test functions @xmath681 on @xmath221 and complex scalars @xmath682 , where @xmath683 is the advanced - minus - retarded fundamental solution to @xmath684 on @xmath5 .
the first two axioms are necessary for compatibility with the idea of @xmath88 as a smeared hermitian field ; the third expresses the field equation in ` weak ' form ; the fourth expresses the commutation relations
. now let @xmath11 be a causal isometric embedding of @xmath71 into @xmath72 .
any test function @xmath676 on @xmath77 now corresponds to a test function @xmath685 on @xmath79 , defined by @xmath686 for @xmath687 and @xmath688 otherwise .
we may use this to define a map @xmath689 between @xmath690 and @xmath691 such that 1 .
@xmath692 2 .
@xmath693 for all test functions @xmath676 on @xmath77 3 .
@xmath689 extends to general elements of @xmath690 as a @xmath105-homomorphism , i.e. , @xmath689 is linear and obeys @xmath694 and @xmath695 for all @xmath696 . in the body of the text
we have used the notation @xmath99 for @xmath689 , relying on the context for the appropriate meaning ; here , it is convenient to distinguish the two maps .
one must check that the last statement is compatible with the axioms stated above
the only nontrivial one is the commutation relation , where the causal nature of @xmath11 plays a key role and guarantees that @xmath689 is well - defined .
what needs to be proved boils down to checking that @xmath697 for all test functions @xmath681 on @xmath77 .
this equivalence is proved as follows .
writing @xmath698 for the advanced ( @xmath699 ) and retarded @xmath700 green functions on @xmath5 , @xmath701 solves the inhomogeneous klein
gordon equation on @xmath72 with source @xmath685 and support in @xmath702 .
because @xmath11 is a causal isometry , the pull - back @xmath703 solves the inhomogeneous klein
gordon equation on @xmath71 with source @xmath676 and support in @xmath704 ; by uniqueness of solution , we have @xmath705 . accordingly @xmath706 and the required result follows . in the algebraic approach we have been pursuing , a state of the quantum field on @xmath5 is a linear map @xmath14 from @xmath674 to the complex numbers , obeying @xmath707 and @xmath708 for any @xmath709 .
one interprets @xmath710 as the expectation value of observable @xmath354 in state @xmath14 .
in particular , each state yields a hierarchy of @xmath92-point functions , i.e. , maps of the form @xmath711 we will restrict attention to those states whose corresponding @xmath92-point functions are distributions .
a state @xmath14 is hadamard if its two - point function has a particular singular structure which is determined by the local metric and causal properties of the spacetime .
note that none of the structure introduced so far invokes any particular hilbert space representation of the theory .
now suppose again that @xmath11 is a causal isometric embedding of @xmath71 into @xmath72 and let @xmath712 be a state on @xmath691 .
we obtain a state @xmath713 on @xmath691 by @xmath714 for any @xmath715 ; that is , @xmath716 , where @xmath717 is the dual map to @xmath689 ( in the body of the text , we have written @xmath718 for @xmath717 ) .
the @xmath92-point functions are therefore related by @xmath719 it is useful to write this in ` unsmeared ' notation . let @xmath720 be the @xmath92-point functions of @xmath721 . then the last equation becomes @xmath722 where the change of variables employed in the last step is justified by the fact that @xmath11 is an isometry .
as this holds for all choices of @xmath723 , we may deduce that @xmath724 that is , the @xmath92-point functions of @xmath713 are the pull - backs by @xmath11 of those of @xmath712 .
it follows that if @xmath712 is hadamard then so too is @xmath713 , because the two - point function is simply pulled back under @xmath11 and the hadamard series is constructed from the local causal and metric structure which is preserved under @xmath11 .
since the stress - energy tensor is renormalised by subtracting the first few terms of the hadamard series from the two point function , and then taking suitable derivatives before taking the coincidence limit ( making a further locally constructed correction to ensure conservation of the stress tensor ) , we have the following important consequence , which we isolate as a theorem .
suppose @xmath11 is a causal isometric embedding of globally hyperbolic spacetime @xmath71 in a globally hyperbolic spacetime @xmath72 .
any hadamard state of the massive klein
gordon quantum field on @xmath72 induces a hadamard state of the same theory on @xmath71 , whose @xmath92-point functions and renormalised expected stress - energy tensor are the pull - backs by @xmath11 of the corresponding quantities on @xmath72 .
this fits in with the principle that one should not be able to tell , by local experiments , whether one is in @xmath71 or its image within the larger spacetime @xmath72 .
it also justifies us in the abuse of notation perpetrated in sec .
[ sect : localcovariance ] , where we wrote @xmath718 in place of @xmath717 , and ( dually ) @xmath99 in place of @xmath689 .
let us conclude by briefly describing more of the structure set out by @xcite .
the key is the observation that the globally hyperbolic spacetimes of given dimension form the objects of a category in which the morphisms are causal isometric embeddings .
one may also consider a category of unital @xmath105-algebras with injective unit - preserving @xmath105-homomorphisms as morphisms .
the association of a globally hyperbolic spacetime @xmath5 with the corresponding algebra @xmath674 is then shown to be a covariant functor between these categories and gives a precise meaning to the notion of ` the same field theory on different spacetimes ' ( and the same would be true even for theories not necessarily described in terms of a lagrangian ) .
a similar functorial description may be given to the association of the state space of the theory , and quantum fields are reinterpreted as natural transformations between functors .
we refer the reader to @xcite for full details .
in this appendix , we calculate quantum weak energy inequalities for the massless , minimally coupled real scalar field in the two dimensional cylinder spacetime relative to the ground and thermal equilibrium states .
we use the notation of sec .
[ sec:2d_cyl ] .
the kms state @xmath540 at inverse temperature @xmath473 has two - point function ( see , e.g. , eq . ( 2.43 ) of ) @xmath725 where @xmath726 and @xmath727 , and the sum converges in the distributional sense ( i.e. , after smearing each term with test functions , the resulting series converges and its sum depends continuously on the test functions ) .
we exclude the zero mode @xmath728 as usual , regarding @xmath540 as a state on the derivative fields .
the two - point function of the ground state @xmath512 is obtained as the zero temperature ( @xmath729 ) limit of this expression .
we will be interested in the static curve @xmath518 , and employ the tetrad @xmath730 , @xmath731 , which is invariant under fermi
walker transport along @xmath10 . following the procedure of sec .
[ sect : lcdqeis_examples ] , we find @xmath732 and @xmath733 taking the fourier transform , we have @xmath734 and therefore obtain @xmath735 note that @xmath736 is supported on @xmath319 in the @xmath737 case , but otherwise on the whole of @xmath738 , albeit exponentially suppressed on the negative half line .
to arrive at convenient qei bounds we estimate @xmath736 by @xmath739 where @xmath740 is the heaviside step function . to see that this estimate is valid , we note that @xmath741 is clearly increasing on the negative half - line , so it is valid to bound it by @xmath742 on @xmath743 ; the second term in the estimate arises by noting that @xmath744 for all @xmath92 . using this estimate again
, we also have @xmath745 while , for @xmath746 @xmath747 where @xmath748 is the greatest integer _
strictly _ less than @xmath749 .
thus we have the estimate @xmath750 it then follows that , in the notation of sec .
[ sec:2d_cyl ] , latexmath:[\[\begin{aligned } { { \mathcal q}}^{\rm weak}_{{\boldsymbol{c}}}({{\sf f}},\omega_{{{\boldsymbol{c}}},\beta } ) & = & \int_{-\infty}^\infty du \,|\widehat{g}(u)|^2 q_{\gamma,\omega_{{{\boldsymbol{c}}},\beta}}(u ) \nonumber\\ & \le & \frac{e^{\pi\beta / l}}{4l^2\sinh^3\pi\beta / l}\int_{-\infty}^0 |\widehat{g}(u)|^2\,du + \frac{1}{2\pi^2(1-e^{-2\pi\beta / l})}\int_0^\infty u^2 |\widehat{g}(u)|^2 \,du \\ & \le & \frac{\pi e^{\pi\beta / l}}{2l^2\sinh^3\pi\beta / l}\int_{-\infty}^\infty \frac{1}{2\pi(1-e^{-2\pi\beta / l})}\int_{-\infty}^\infty theorem and the fact that @xmath332 is even for real - valued @xmath37 to convert the integral over @xmath319 into one over @xmath738 . for the ground state
, of course , this yields latexmath:[\[{{\mathcal q}}^{\rm weak}_{{\boldsymbol{c}}}({{\sf f}},\omega_{{{\boldsymbol{c } } } } ) \le \frac{1}{2\pi}\int_{-\infty}^\infty estimates are not very sharp , they have led to a very simple quantum inequality .
in fact , for the ground state , this inequality is only @xmath25 times less restrictive than the optimal quantum inequality bound found in two dimensional minkowski spacetime .
recall from sec . [
sect : lcdqeis_examples ] that we define @xmath223 to be the set of smooth compactly supported real - valued functions @xmath37 on @xmath55 whose support is connected and which have no zeros of infinite order in the interior of that support ( equivalently , @xmath37 has no zeros in @xmath753 of infinite order ) .
our aim is to prove the following result .
_ proof : _ if @xmath37 is identically zero the result is trivial , so we assume henceforth that it is not , so @xmath758 is strictly positive .
suppose the stated result is false , so there exists an @xmath759 such that @xmath760 for all @xmath761 .
choose @xmath762 sufficiently large that @xmath763 exceeds the diameter of the support of @xmath37 .
by hypothesis , for each @xmath764 the function @xmath765 has a zero of infinite order within its support ; since @xmath754 this zero must lie in the support of @xmath37 and is therefore a point at which @xmath766 vanishes , while @xmath37 takes the value @xmath767 .
we may therefore choose @xmath768 such that @xmath769 for each @xmath314 and @xmath770 runs through the values @xmath771 ( not necessarily in order ) .
using taylor s theorem with remainder at each @xmath772 , @xmath773 so @xmath774 which is a contradiction , since @xmath775 and @xmath776 must belong to the support of @xmath37 .
@xmath777 this work was initiated at the erwin schrdinger institute for mathematical physics , vienna , during a program on quantum field theory in curved spacetime .
we are grateful to the organizers of this program and to the institute for its hospitality and financial support .
cjf further thanks the isaac newton institute , cambridge , for support during the final stages of the work , and the organisers of the programme on global problems in mathematical relativity .
we also thank l.h .
ford , a. higuchi , b.s .
kay , j. loftin and t.a .
roman for many illuminating discussions , and l. osterbrink and c.j .
smith for comments on the manuscript .
this work was partially supported by epsrc grant gr / r25019/01 to the university of york and by a grant from the us army research office through the usma photonics research center .
mjp also thanks the university of york for a grant awarded under its `` research funding for staff on fixed term contracts '' scheme . | we begin a systematic study of quantum energy inequalities ( qeis ) in relation to local covariance . we define notions of locally covariant qeis of both ` absolute ' and ` difference ' types and show that existing qeis satisfy these conditions .
local covariance permits us to place constraints on the renormalised stress - energy tensor in one spacetime using qeis derived in another , in subregions where the two spacetimes are isometric .
this is of particular utility where one of the two spacetimes exhibits a high degree of symmetry and the qeis are available in simple closed form .
various general applications are presented , including _ a priori _ constraints ( depending only on geometric quantities ) on the ground - state energy density in a static spacetime containing locally minkowskian regions .
in addition , we present a number of concrete calculations in both two and four dimensions which demonstrate the consistency of our bounds with various known ground- and thermal state energy densities .
examples considered include the rindler and misner spacetimes , and spacetimes with toroidal spatial sections . in this paper
we confine the discussion to globally hyperbolic spacetimes ; subsequent papers will also discuss spacetimes with boundary and other related issues . |
certain quantum systems in @xmath2 dimensions can support quasiparticles with anyonic and non - abelian statistics .
these are experimentally realized in fractional quantum hall ( fqh ) systems . specifically , the @xmath3 fqh state is conjectured to be in the universality class of the moore - read ( mr ) pfaffian wavefunction , and it is believed that there is a fqh state at @xmath1 described by the @xmath4 read - rezayi ( rr ) state . both are non - abelian , and the latter supports universal quantum computation , making its study especially important . in this paper we consider a double point contact geometry in an assumed @xmath1 rr state , and study tunneling in the edge theory .
a similar setup is considered in [ ] .
the edge theory consists of a charged boson , and a neutral sector described by the @xmath5 parafermion cft .
the latter contains spin fields @xmath6 which are not local with respect to the other fields in the cft , and it is precisely these that enter the hamiltonian describing tunneling of quasiholes , which is the most relevant tunneling process at low temperatures .
unfortunately it is difficult to interpret such a tunneling hamiltonian , because in a chiral theory , correlation functions of spin fields are not well defined .
instead , they could be any linear combination of a finite number of _ conformal blocks_. in fact , it turns out that the choice of such linear combination is intimately related to the topology of the corresponding tunneling process occurring in the bulk ; [ ] give a prescription for determining the correct choice of linear combination , in the case of a single point contact .
we adapt their method to the case of a double point contact and use it to compute the tunneling current to first order in perturbation theory .
in contrast to the abelian case @xcite , we find that the interference term in the tunneling current is suppressed depending on the topological charge contained in the bulk region between the two point contacts .
a crucial point is that a bulk quasihole affecting the tunneling current must be prepared and annihilated on the edge in the far past and future respectively , and the prescription of [ ] must be applied to this entire process .
after one divides out the universal factor associated with preparing and annihilating the bulk quasihole , one obtains correlation functions of the original spin fields which are finite and correctly incorporate the effect of the bulk quasihole .
we assume that there exists an incompressible quantum hall fluid at @xmath7 , and that it consists of a filled lowest landau level ( of both spins ) and a first landau level described by the @xmath4 read - rezayi state , which has filling factor @xmath8 .
the read - rezayi wave function is a correlator of operators in the conformal field theory of a charged boson and @xmath5 parafermions .
the parafermions contain @xmath9 fields : @xmath10 , and their fusion rules are determined by @xmath11 and @xmath12 [ ] .
we will need the electron and quasihole operators ; they are as follows @xcite : @xmath13 we have @xmath14 , and @xmath15 .
the chiral edge theory also consists of @xmath5 parafermions and a charged mode , but they propagate with different velocities , @xmath16 and @xmath17 .
generally , one expects @xmath18 . in writing down the edge theory description of the tunneling hamiltonian , we a priori need to consider all sorts of tunneling processes , but a quick renormalization group argument @xcite shows that the most relevant term at low temperatures corresponds to an @xmath19-charge quasihole - the tunneling hamiltonian simply destroys such a quasihole on one side of the point contact and creates it on the other . here
we assume we are at temperatures higher than those at which the crossover physics of [ ] , [ ] , etc
. becomes important , so that the tunneling is weak and can be treated perturbatively .
this is an important point , because when we make reference to zero temperature results from now on we actually mean temperature small compared to @xmath20 , where @xmath21 is the bias voltage , and to @xmath22 and @xmath23 . here
@xmath24 and @xmath25 ( @xmath26 being the distance between the two contacts ) but still large compared to the crossover temperature . note that we have @xmath27 different energy scales , set by the temperature , bias voltage , @xmath28 and @xmath29 .
the last two can not be varied independently , but only through the variation of the separation @xmath26 .
the two cases we consider in this paper are zero temperature , where effectively we have one dimensionless parameter we can vary ( say @xmath30 ) , and temperature high compared to @xmath31 , where things simplify as well .
the tunneling hamiltonian is @xmath32 where @xmath33 is the tunneling amplitude at the @xmath34th point contact ( @xmath35 ) , and @xmath36 is the josephson frequency corresponding to a charge @xmath19 quasihole .
the tunneling operator @xmath37 is defined as @xmath38 where @xmath39 and @xmath40 are the spatial coordinates of the lower and upper sides of the @xmath34th point contact .
note that because the hall bar is large but finite , the upper and lower points are actually on the same edge , just very far away from each other . from now on it will be convenient to use the parametrization @xmath41 where @xmath42 is the distance around the hall bar , which we will eventually take to infinity
our discussion here closely parallels [ ] , except that we have the added complication of dealing with the parafermions and spin field operators . as mentioned before
, there is ambiguity in treating the @xmath6 as operators , and their correlators are not well defined .
indeed , to resolve the ambiguity we will need to add extra @xmath6 operators , which create and annihilate the bulk quasiholes in the far past and future . as we shall see , taking these to infinity essentially decouples them from the rest of the correlator , and their only influence is topological . in fact , together with making the hall bar really large ( that is , taking @xmath42 to infinity ) , the correlator of interest factors as a product of @xmath2-point functions , which is determined by conformal invariance up to a multiplicative constant .
all of the effects mentioned above are incorporated in that constant , denoted by @xmath43 below . in this section
we calculate the tunneling current leaving @xmath43 undetermined . in the next section
we determine @xmath43 .
we begin by writing down the appropriate operators .
the operator for the current across the point contact is the commutator of the total charge and the tunneling hamiltonian ; because only the charge mode is involved the calculation is easy and we get @xmath44 to lowest order in perturbation theory , the expectation value of the tunneling current is @xmath45 | 0 \rangle.\ ] ] we have @xmath46 because the only terms that contribute to the vacuum expectation value are ones that conserve the total charge on each side of the hall bar .
now , @xmath47}^{-2/15 } { \left [ \delta + i ( v_n ( t - t ' ) - ( x_j - x_k ) ) \right]}^{-2/15 } \nonumber \\ & & { \left [ \delta + i ( v_c ( t - t ' ) + ( x_j - x_k ) ) \right]}^{-1/15 } { \left [ \delta + i ( v_c ( t - t ' ) - ( x_j - x_k))\right]}^{-1/15}.\end{aligned}\ ] ] here @xmath43 is the constant mentioned above , which is determined by choice of conformal block ( its phase is more subtle and convention dependent ) , and @xmath48 is an infinitesimal , specifying the choice of branch cut .
we will obtain all the information about @xmath43 we need for our purposes in the next section .
for now , define @xmath49 ^{-2/15 } \left [ \delta + i ( v_n t - x ) \right ] ^ { -2/15 } \left [ \delta + i ( v_c t + x ) \right]^{-1/15 } \left [ \delta + i ( v_c t - x ) \right]^{-1/15}.\ ] ] so that @xmath50 in a similar manner we obtain @xmath51 because @xmath52 and @xmath53 commute for spacelike separation , we have @xmath54 ( for @xmath55 the argument is a little more subtle but the conclusion holds - basically you have to infinitesimally spacelike separate the two operators and then argue using continuity ) .
we therefore have @xmath56 \rangle & = & \frac{ie}{5 } \sum_{j , k } ( \gamma_j { \gamma_k}^ * e^{-i \omega_j ( t - t ' ) } c_{jk } [ p(x_j - x_k , t - t ' ) - p(x_j - x_k , -(t - t ' ) ) ] \nonumber \\ & & - { \gamma_j}^ * \gamma_k e^{i \omega_j ( t - t ' ) } c_{kj } [ p(x_k - x_j , t - t ' ) - p(x_k - x_j,-(t - t ' ) ) ] ) \end{aligned}\ ] ] doing the integral in ( [ eqn : int ] ) we obtain @xmath57\ ] ] where @xmath58 is the fourier transform of @xmath59 . for @xmath60 we can do the integral explicitly : @xmath61 when @xmath62 , we can write down an alternate integral representation for @xmath63 as follows : in the integral for the fourier transform , the denominator is a product of @xmath27 terms , with @xmath64 ( the variable being integrated over ) appearing in each .
we replace the @xmath64 integration with an integration over @xmath27 variables @xmath65 ( @xmath66 ) , replacing @xmath64 with @xmath65 in each term in the denominator , and adding in delta function constraints to make all the @xmath65 equal .
this is just a rewriting of the original integral .
we then use the integral representation of the delta functions to obtain : @xmath67 \nonumber \\ & & { | \omega + \omega_1|}^{-13/15 } { | \omega_2 - \omega_1 |}^{-13/15 } { | \omega_3 - \omega_2 |}^{-14/15 } { | \omega_3 |}^{-14/15 } \theta ( \omega + \omega_1 ) \theta ( \omega_2 - \omega_1 ) \theta ( \omega_3 - \omega_2 ) \theta ( - \omega_3)\end{aligned}\ ] ] because the integrals over @xmath68 are all convergent in the above expression , we see that @xmath69 is continuous as a function of @xmath70 .
we write @xmath71 with this notation we obtain @xmath72
the analysis of finite temperature effects again proceeds much in the same way as in [ ] .
all the nontrivial topological information is still encoded in the @xmath43 .
the only difference between the zero temperature and finite temperature case arises from the fact that the @xmath2-point functions we now wish to use are analytic continuations of @xmath2-point functions on a cylinder .
the replacement that needs to be made is @xcite : @xmath73}^{2/15 } } \rightarrow { \left [ \frac{\pi t } { \sin [ \pi t(\delta + i ( t \pm x / v_n ) ) ] } \right]}^{2/15}\ ] ] for the @xmath6 correlator , and similarly for the correlator of the exponentials of the charge bosons , except with @xmath16 replaced by @xmath17 .
we obtain the following modified @xmath74 : @xmath75}{\pi t } \right]}^{-2/15 } { \left [ \frac{\sin [ \pi t(\delta+i(t - x / v_n))]}{\pi t } \right]}^{-2/15 } \nonumber \\ & & { \left [ \frac{\sin [ \pi t(\delta+i(t+x / v_c))]}{\pi t } \right]}^{-1/15 } { \left [ \frac{\sin [ \pi t(\delta+i(t - x / v_c))]}{\pi t } \right]}^{-1/15}\end{aligned}\ ] ] as opposed to the case of [ ] , this integral is not exactly doable for arbitrary values of @xmath76 , the neutral and charge mode velocities .
however , we can extract the behavior of the current at large values of the temperature . to do this ,
simply note that the integrand above is basically a product of inverse powers of @xmath77 , which at @xmath78 large compared to @xmath22 and @xmath23 become decaying exponentials , suppressed everywhere except at @xmath79 .
noting that the integral is everywhere convergent , we see that most of its contributions come from those @xmath27 points . plugging in , we see that we have an exponential suppression of @xmath80\ ] ] of the interference term at high temperatures @xmath81 . comparing this to the result of [ ] who obtain @xmath82\ ] ]
we see that we have an effective " @xmath83 of @xmath84 .
we now come to the crux of the problem , which is to find the constants @xmath43 .
this basically comes down to determining the conformal blocks for the @xmath85 correlators , and here the topology comes into play . working in the context of the @xmath86 state and a single point contact , [ ] argued that for a given quasihole tunneling process , the two @xmath87 fields that create and destroy the quasihole should fuse to the identity . arguing along these lines
, one can uniquely determine the desired conformal block .
it is here that one uses the connection between the bulk and the boundary : the tunneling topology in the bulk determines the fusion channels of the tunneling operators in the boundary field theory . to give an actual argument for this connection would require a detailed examination of bulk tunneling , which has not been done .
we shall , as in [ ] , adopt it as part of the definition of the tunneling operator , and adapt it to our case of a double point contact .
we wish to probe the effect of test quasiholes , residing on anti - dots in the region between the two point contacts , on the tunneling current .
the main conceptual point we wish to stress is that to do the computation correctly , one must incorporate the creation and annihilation of these test quasiholes in the far past and future .
that is , we must include two more @xmath85 operators - we will take them to be on the same side of the hall bar , creating the quasihole in the far past and annihilating it in the far future ( see fig ( [ f1 ] ) ) . for simplicity
, we will analyze just two situations : one without any test quasiholes , and one with a single test quasihole , and for easy comparison we will actually create a test quasihole in both situations , but keep it far away from the region of interest in one case ( so that it has trivial braiding ) . for definiteness , we will for now analyze the following correlator , one of two that contribute to the interference term : @xmath88 here the test quasihole is created at @xmath89 , and annihilated at @xmath90 .
@xmath91 and @xmath42 are to be taken to infinity .
we find it convenient to send @xmath42 to infinity first , so the thing splits into a @xmath27 point and a @xmath2 point function by cluster decomposition .
let s see what happens when @xmath91 goes to infinity .
then @xmath92 and @xmath93 can be considered to be very close ( in comparison with @xmath94 and @xmath95 ) , so we can use the operator product expansion on them .
clearly the leading contribution comes from the identity , so that s the only one we need to retain in the limit .
the @xmath27-point function is then just the @xmath2-point function of @xmath92 and @xmath93 times an overall constant having to do with the preparation and annihilation of the test quasihole . from the above paragraph
, we see that it is very convenient to work in the basis where the fusion channels of the pairs @xmath96 , @xmath97 , and @xmath98 are known , because only in the case when all those fusion channels are the identity do we get a nonzero contribution .
but the conformal block we re interested in is specified in a different basis . to express it in the new basis ,
we proceed as in [ ] and project the worldlines of the quasiholes ( including the test quasihole ) to two dimensions ; the linear transformation necessary to enact the basis change is the matrix representing the resulting braid .
the conformal block we start with is : @xmath99 we want to expand it in the new basis above , and in particular get the coefficient of @xmath100 depending on whether or not the test quasihole resides in the region between the two point contacts , we have two possible braiding histories , shown in figures ( [ fb1 ] ) and ( [ fb2 ] ) . because , due to the considerations outlined below , we only care about the coefficients up to phase , we only need to determine the matrix for the linear transformation between the two bases up to phase . thus , even though a priori we would have to deal with the moore - seiberg braid matrix for the parafermions ( and the associated @xmath101 theory ) , we really only need the braid matrix in the associated quantum group picture . here , both @xmath102 and @xmath103 correspond to the @xmath2-dimensional spin @xmath104 representation of @xmath105 , @xmath106 . labeling conformal blocks using the bratteli diagram notation , we see that we start off with the block represented by the path @xmath107 $ ] , which for convenience we will designate via shortened notation by @xmath108 $ ] .
we see that when the braids in ( [ fb1 ] ) and ( [ fb2 ] ) act , everything happens in the subspace spanned by @xmath109 $ ] , @xmath110 , and each transposition acts either only on @xmath70 or only on @xmath111 ( they commute ) .
the relevant unitary matrix enacting the transposition is [ ] : @xmath112 where @xmath113 .
the @xmath2-element basis in which we write the matrix @xmath114 is @xmath115 , and the one corresponding to @xmath116 is @xmath117 .
note also that , using some identities involving the golden mean , we have : @xmath118 we have @xmath119 | u_{2,3 } | [ 0101010 ] \rangle = \frac{1}{\tau } \\ \label{feqn1 } \langle [ 0101010 ] | u_{2,3 } { u^2}_{4,5 } | [ 0101010 ] \rangle = \frac{1}{\tau^3}\end{aligned}\ ] ] let us now use this to obtain the @xmath43 .
first , let s see what we can surmise on general grounds . by tuning @xmath120 to @xmath121 and using the fact that @xmath122 is real
, we find that @xmath123 is real . also , by symmetry , @xmath124 . again
because the current is real we see that @xmath125 .
the phase @xmath126 can in principle be determined , but we ignored it in the calculations above .
this is because the effect of changing @xmath126 is the same as that of changing the phases of @xmath127 and @xmath120 , which are hard to determine anyway and are sensitive to the aharonov - bohm effect .
the results in ( [ feqn ] ) and ( [ feqn1 ] ) show that when a quasihole is present , @xmath128 is diminished by a factor of @xmath129 .
similar topological calculations ( with only the spatial coordinates of the points involved changing ) show that the other constants , @xmath123 and @xmath130 , are always equal , and equal to the @xmath128 in the case when there is no test quasihole in the region between the point contacts .
thus we have @xmath131 @xmath132 where @xmath133 is an unimportant constant .
we thus observe that even without looking at phases , nonabelian statistics have a robust effect of changing the magnitude of the coefficient of the interference term in the tunneling current .
this is in line with earlier arguments of [ ] for the differential conductivity .
i would like to thank eddy ardonne , michael freedman for discussions , and chetan nayak for discussions and a careful reading of the draft .
this research was supported partly by the nsf under grant no .
-0244728 , the sitp , and microsoft .
p. fendley , m. p. a. fisher , and c. nayak , _ edge states and tunneling of non - abelian quasiparticles in the @xmath3 quantum hall state and @xmath134 superconductors _ , phys . rev . b. * 75 * , 045317 ( 2007 ) , cond - mat/0607431 .
j. k. slingerland , _ hopf symmetry and its breaking _ , phd thesis , 2002 .
see also j.k .
slingerland and f.a .
bais , _ quantum groups and nonabelian braiding in quantum hall systems _ , nucl
* b 612 * ( 2001 ) 229 - 290 , cond - mat/0104035 . | we compute the dependence of the tunneling current in a double point contact in the @xmath0 read - rezayi state ( which is conjectured to describe an incompressible quantum hall fluid at filling fraction @xmath1 ) on voltage , separation between the two contacts , and temperature . using the tunneling hamiltonian of cond - mat/0607431
, we show that the effect of quasiholes in the bulk region between the two contacts is simply an overall constant multiplying the interference term .
this is the same effect as found for the differential conductivity in cond - mat/0601242 ; the difference is that we do an actual edge theory calculation and compute the full current - voltage curve at weak tunneling . |
for a long time quantum entanglement was only of philosophical interest and researchers were mainly focusing on addressing the questions that were related with the quantum mechanical understanding of various fundamental notions like reality and locality @xcite .
however , for the last two decades world had seen that quantum entanglement is not only a philosophical riddle but also a reality as far as the laboratory preparation of entangled qubits are concerned @xcite .
researches that were conducted during these decades were not all concerned about its existence but mostly about its usefulness as a resource to carry out information processing protocols like quantum teleportation @xcite , cryptography @xcite , superdense coding @xcite , and in many other tasks @xcite .
it was subsequently evident from various followed up investigations that quantum entanglement plays a pivotal role in all these information processing protocols .
therefore , understanding the precise nature of entanglement in bipartite and multiparty quantum systems has become the holy - grail of quantum information processing .
however , the precise role of entanglement as a resource in quantum information processing is not fully understood and it was suggested that entanglement is not the only type of correlation present in quantum states . this is because lately some computational tasks were carried out even in the absence of entanglement @xcite .
this provided the foundation to the belief that there may be correlation present in the system even in the absence of entanglement .
hence , researchers redefined quantum correlation from the information theoretic perspective .
this gave rise to various measures @xcite of quantum correlation , the predominant of them being quantum discord @xcite .
though there are issues that need to be addressed , in much deeper level quantum discord temporarily satisfies certain relevant questions .
subsequently , quantum discord has been given an operational interpretation in different contexts like quantum state merging @xcite and remote state preparation @xcite .
in addition , extension of the notion of quantum discord to multi qubit cases has been proposed @xcite .
many works were done in the recent past to investigate the dynamics of quantum correlation in open systems by comparing the evolution of different types of initial states in specific models .
these states are typically two qubits coupled with two local baths or one common bath . in principle , there are several factors that can affect the evolution , namely , the initial state for the system and environment , the type of system - environment interaction and the structure of the reservoir
. a more relevant question will be how robust are these measures when they are subjected to the noise in quantum channels .
it is mainly inspired by the studies of sudden death of entanglement for two qubits , having no direct interaction @xcite .
entanglement sudden death ( esd ) is said to occur when the initial entanglement falls and remains at zero after a finite period of evolution for some choices of the initial state .
esd is a potential threat to quantum algorithms and quantum information protocols and thus the quantum systems should be well protected against noisy environments .
another possible way to circumvent such resource vanishing is to make use of resources which do not suffer from sudden death . at this point
, one can ask a similar question : _ does quantum discord present similar behavior ? _ in the first study @xcite addressing this question , researchers have compared the evolution of concurrence and discord for two qubits , each subject to independent markovian decoherence ( dephasing , depolarizing and amplitude damping ) .
looking at initial states such as werner states and partially - entangled pure states , the authors find no sudden death of discord even when esd does occur ; quantum discord decays exponentially and vanishes asymptotically in all cases .
however , not much is known about the effects on multipartite correlation with time when they are transferred through noisy quantum channels . in this work ,
we study the dynamics of quantum dissension of three qubit states which happens to be a measure of multi party quantum correlation , under the effect of various quantum noisy channels .
in addition , we also study the dynamics of monogamy score of these three qubit states in presence of channel noise . in section 2 ,
we provide a detail descriptions of quantum dissension and monogamy score of quantum correlation . in section 3
, we study the effect of various noisy channels on quantum correlation and monogamy score when all the qubits are transferred through them .
finally , we conclude in section 4 by discussing future directions of explorations .
in classical information theory @xcite , the total correlation between two random variables is defined by their mutual information .
if x and y are two random variables , the mutual information is obtained by subtracting the joint entropy of the system from the sum of the individual entropies .
mathematically , this can be stated as : @xmath0 where @xmath1 defines shannon entropy function . another equivalent way of expressing mutual information is by taking into account the reduction in uncertainty associated with one random variable due to the introduction of another random variable .
stated formally as , @xmath2 or @xmath3 where @xmath4 defines conditional entropy of @xmath5 given that @xmath6 has already occurred and vice versa .
all these above expressions are equivalent in classical information theory .
when we try to quantify correlation in quantum systems from an information theoretic perspective , natural extension of these quantities will be obtained by replacing random variables with density matrices , shannon entropy with von neumann entropy and apposite definition of the conditional entropies . stated mathematically , the quantum mutual information is given by , @xmath7 where @xmath8 is the composite density matrix , @xmath9 and @xmath10 are the local density matrices and @xmath11 defines von neumann entropy function .
similarly , by applying the argument of reduction of uncertainty associated with one quantum system with introduction of another quantum system , one can have the alternative definition of mutual information as , @xmath12 and @xmath13 here @xmath14 is the average of conditional entropy and is obtained after carrying out a projective measurement on subsystem @xmath6 and vice versa .
the projective measurement is done in the general basis @xmath15 = cos@xmath16 + @xmath17 sin@xmath18 , @xmath19 = sin@xmath16 - @xmath17 cos@xmath20 , where @xmath21 and @xmath22 have the range [ 0,2@xmath23 .
hence , the quantum conditional entropy can be expressed as , @xmath14 = @xmath24 where @xmath25 $ ] , ( @xmath26 being identity operator on the hilbert space of the quantum system @xmath5 ) , gives the probability of obtaining the @xmath27the outcome . the corresponding post - measurement state of system @xmath5 is @xmath28 $ ] .
it is important to note over here that @xmath29 is different from what will be the straightforward extension of classical conditional entropy . in quantum information ,
the meaning of conditional entropy of the qubit @xmath5 given that @xmath6 has occurred is the amount of uncertainty in the qubit @xmath5 given that a measurement is carried out on the qubit @xmath6 .
consequently , the expressions @xmath30 , @xmath31 and @xmath32 are not equivalent in the quantum domain .
the differences between @xmath33 and @xmath34 are captured by quantum discord i.e. @xmath35 @xmath36 one variant of quantum discord is the geometric quantum discord which is defined as the distance between a quantum state and the nearest classical ( or separable ) state , @xcite .
quantum discord has been established as a non - negative measure of correlation for any quantum states .
subsequent researches were carried out to obtain an analytical closed form of quantum discord and was found for certain class of states @xcite . an unified geometric view of quantum correlations which includes discord , entanglement along with the introduction of the concepts like quantum dissonance was given in @xcite .
one of the natural extension of quantum discord from two qubit to three qubit systems is quantum dissension @xcite .
introduction of three qubits naturally brings in one and two - particle projective measurement into consideration .
these measurements can be performed on different subsystems leading to multiple definitions of quantum dissension . in other words
a single quantity is not sufficient enough to capture all aspects of correlation in multiparty systems .
quantum dissension in this context can be interpreted as a vector quantity with values of correlation rising because of multiple definitions as various components . however , in principle when we define correlation in multi qubit situations , measurement in one subsystem can enhance the correlation in other two subsystems and thereby making quantum dissension to assume negative values @xcite .
we emphasize on all possible one - particle projective measurements and two - particle projective measurements .
the mutual information of three classical random variables in terms of entropies and joint entropies , are given by @xmath37+h(x , y , z).\end{gathered}\ ] ] it is also possible to obtain an expression for mutual information @xmath38 that involves conditional entropy with respect to one random variable : @xmath39 one can define another equivalent expression for classical mutual information that includes conditional entropy with respect to two random variables : @xmath40 \\ -[h(x , y)+h(x , z)]+h(x|y , z).\end{gathered}\ ] ] these equivalent classical information - theoretic definitions forms our basis for defining quantum dissension in the next subsections .
let us consider a three - qubit state @xmath41 where @xmath42 and @xmath43 refer to the first , second and the third qubit under consideration .
the quantum version of @xmath38 obtained by replacing random variables with density matrices and shannon entropy with von neumann entropy reads , @xmath44+s(\rho_{xyz}).\end{gathered}\ ] ] the quantum version of @xmath45 , obtained by appropriately defining conditional entropies , is given by @xmath46 where @xmath47 refer to a one particle projective measurement on the subsystem @xmath48 performed on the basis @xmath15 = cos@xmath16 + @xmath17 sin@xmath18 , @xmath19 = sin@xmath16 - @xmath17 cos@xmath20 where @xmath21 and @xmath22 lies in the range [ 0,2@xmath23 .
quantum dissension function for single particle projective measurement is given by the difference of @xmath38 and @xmath45 , i.e. @xmath49 quantum dissension is given by the quantity @xmath50 = min(@xmath51 ) , where the minimization is taken over the entire range of basis parameters in order for @xmath52 to reveal maximum possible quantum correlation .
the natural extension of @xmath53 in the quantum domain is given by , @xmath54 \\
-[s(\rho_{xy})+s(\rho_{xz})]+s(\rho_{x|{\pi_{jyz}}}).\end{gathered}\ ] ] the two - particle projective measurement is carried out in the most general basis : @xmath55 = cos@xmath56 + @xmath17 sin@xmath57 , @xmath58 = sin@xmath56 - @xmath17 cos@xmath57 , @xmath59 = cos@xmath60 + @xmath17 sin@xmath61 , @xmath62 = sin@xmath60 - @xmath17 cos@xmath61 , where @xmath21 , @xmath22 @xmath63 [ 0,2@xmath23 . in this case , the average quantum conditional entropy is given as @xmath64 with @xmath65 $ ] and @xmath66 $ ] . to define quantum dissension for two - particle projective measurement , we once again take the difference of the equivalent expressions of mutual information , i.e. @xmath67 the discord function @xmath68 is also interpreted as quantum discord with a bipartite split of the system .
one can minimize @xmath68 over all two - particle measurement projectors to obtain dissension as @xmath69=min(@xmath70 ) .
this is the most generic expression since it includes all possible two - particle projective measurements .
both @xmath50 and @xmath69 together form the components of correlation vector defined in context of projective measurement done on different subsystems .
monogamy of quantum correlation is an unique phenomenon which addresses distributed correlation in a multiparty setting .
it states that in a multipartite situation , the total amount of individual correlations of a single party with other parties is bounded by the amount of correlation shared by the same party with the rest of the system when the rest are considered as a single entity .
mathematically , given a multipartite quantum state @xmath71 shared between n parties , the monogamy condition for a bipartite correlation measure @xmath72 should satisfy @xmath73 @xmath74 @xmath75 where @xmath76 = @xmath77 @xmath78 .
it had been shown that certain entanglement measures satisfy the monogamy inequality .
however , there are certain measures of quantum correlation , including quantum discord , which behave differently as far as the satisfying of monogamy inequality is concerned . by the term violation of monogamy inequality for certain measure
, we actually refer to a situation where we can indeed find entangled states which violates the inequality for that measure . in case of quantum discord
, it had been seen that w states violates the inequality and are polygamous in nature .
more specifically , researchers considered the monogamy score @xmath79= @xmath80+@xmath81-@xmath82 , ( where @xmath83 and @xmath84 are the traced out density matrices from @xmath85 and @xmath86 is quantum discord ) and checked whether three - qubit states violate or satisfy the inequality @xmath79 @xmath74 0 .
in this section , we investigate the dynamics of quantum dissension when three - qubit states are transferred through noisy quantum channels .
moreover , we also study the change of the monogamy score for various initial states with time and purity of the state .
we consider initial states to be mixed ghz , mixed w , classical mixture of two separable states , a mixed biseparable states and the quantum channels to be amplitude damping , phase damping and depolarizing . given an initial state for three qubits @xmath87 , its evolution in the presence of quantum noise can be compactly written as , @xmath88 where @xmath89 are the kraus operators satisfying @xmath90=i for all @xmath91 @xcite . for independent channels , @xmath92 where @xmath93 describes one - qubit quantum channel effects .
we analytically present the dynamics of each initial state with respect to the individual channels .
in other words we present the dynamics of each of @xmath94 , @xmath95 and @xmath79 . in each case
, we apply the channel for sufficient time i.e. t=10 seconds . in this subsection
, we consider the effect of generalized amplitude damping channel on various three - qubit quantum states .
the amplitude damping channel describes the process of energy dissipation in quantum processes such as spontaneous emission , spin relaxation , photon scattering and attenuation etc .
it is described by single - qubit kraus operators @xmath96=@xmath97 diag(1,@xmath98 ) , @xmath99=@xmath100 ( @xmath101)/2 , @xmath102=@xmath103 diag(@xmath98,1 ) , @xmath104=@xmath105 ( @xmath106)/2 , where @xmath107 defines the final probability distribution when @xmath108 ( q=1 corresponds to the usual amplitude damping channel ) . here
@xmath109=1-@xmath110 , @xmath111 representing the decay rate . _
_ + we consider the three - qubit mixed ghz state @xmath112 ( we universally take @xmath113 as the classical randomness ) as the initial state . the matrix elements of the density operator for a certain time @xmath91 , or for a certain value of parameter @xmath109 are given by , @xmath114,{}\nonumber\\ & & \rho_{22}=\rho_{33}=\rho_{55}=\frac{1}{8}(1-\gamma)[(1+\gamma)^2-p(1-\gamma)(3\gamma+1)],{}\nonumber\\ & & \rho_{44}=\rho_{66}=\rho_{77}=\frac{1}{8}(1-\gamma)^2[(1+\gamma)+p(3\gamma-1)],{}\nonumber\\ & & \rho_{88}=\frac{1}{8}(1 + 3p)(1-\gamma)^3,\rho_{18}=\frac{p}{2}(1-\gamma)^\frac{3}{2}.\\end{aligned}\ ] ] it is evident from fig.1 , @xmath94 and @xmath95 attains values @xmath115 at @xmath116 and @xmath117 and decays asymptotically till each of them approaches @xmath118 .
the amplitude damping channel leaves the final population state at ( diag@xmath119)^{\otimes 3}$ ] which contains no quantum dissension . in other words
, we have a steady decay of quantum dissension for the mixed ghz state with time in an amplitude damping channel .
the reduced density matrices are separable states and contain zero quantum discord for all values of time and purity .
thus , the monogamy score @xmath79 is negative of @xmath95 .
since @xmath79 is always negative in this case , the state remains monogamous through out the evolution period .
+ _ _ + for the three - qubit mixed w state @xmath120 , the dynamics of the state in terms of the matrix elements at time @xmath91 is given by , @xmath121,{}\nonumber\\ & & \rho_{22}=\rho_{33}=\rho_{55}={}\nonumber\\ & & \frac{1}{24}(1-\gamma)[3(1+\gamma)^2-p(3\gamma^2 + 6\gamma-5)],{}\nonumber\\ & & \rho_{44}=\rho_{66}=\rho_{77}=\frac{1}{8}(1-p)(1-\gamma)^2 ( 1+\gamma),{}\nonumber\\ & & \rho_{88}=\frac{1}{8}(1-p)(1-\gamma)^3,{}\nonumber\\ & & \rho_{23}=\rho_{25}=\rho_{35}=\frac{1}{3}p(1-\gamma).\end{aligned}\ ] ] the initial values of @xmath94 and @xmath95 for a pure w state are ( -1.75,0.92 ) respectively .
as shown in fig.2 , @xmath94 and @xmath95 starts asymptotic decay from ( -1.75,0.92 ) at @xmath116 and @xmath117 till they approach @xmath118 after sufficient channel action .
the final population distribution at the limit of @xmath1221 is ( diag@xmath119)^{\otimes 3}$ ] resulting in zero quantum dissension . in fig.3(a ) , we study the evolution of monogamy score with time and interestingly we find that for certain values of the parameter @xmath113 , the monogamy score @xmath79 changes from negative to positive .
this is a clear indication of the fact that the states which are initially monogamous are entering into the polygamous regime with time . _
_ + we take classical mixture of separable states @xmath123 and @xmath124 given by the density matrix @xmath125 .
the dynamics of this mixture under the action of amplitude damping channel in terms of the matrix elements is as follows @xmath126,{}\nonumber\\ & & \rho_{12}=\rho_{13}=\rho_{15}=\frac{1}{8}(1-p)\sqrt{1-\gamma}(1+\gamma)^2,{}\nonumber\\ & & \rho_{14}=\rho_{16}=\rho_{17}=\rho_{23}=\rho_{25}=\rho_{35}=\frac{1}{8}(1-p)(1-\gamma^2),{}\nonumber\\ & & \rho_{18}=\rho_{27}=\rho_{36}=\rho_{45}=\frac{1}{8}(1-p)(1-\gamma)^\frac{3}{2},{}\nonumber\\ & & \rho_{22}=\rho_{33}=\rho_{55}=\frac{1}{8}(1-p)(1-\gamma)(1+\gamma)^2,{}\nonumber\\ & & \rho_{24}=\rho_{26}=\rho_{34}=\rho_{37}=\rho_{56}=\rho_{57}={}\nonumber\\ & & \frac{1}{8}(1-p)\sqrt{1-\gamma}(1-\gamma^2),{}\nonumber\\ & & \rho_{44}=\rho_{66}=\rho_{77}=\frac{1}{8}(1-p)(1-\gamma)^2(1+\gamma),{}\nonumber\\ & & \rho_{28}=\rho_{38}=\rho_{46}=\rho_{47}=\rho_{58}=\rho_{67}=\frac{1}{8}(1-p)(1-\gamma)^2,{}\nonumber\\ & & \rho_{48}=\rho_{68}=\rho_{78}=\frac{1}{8}(1-p)(1-\gamma)^\frac{5}{2},{}\nonumber\\ & & \rho_{88}=\frac{1}{8}(1-p)(1-\gamma)^3.\end{aligned}\ ] ] at @xmath116 , the maximum values ( -1.015,0.15 ) of quantum dissensions @xmath94 and @xmath95 are obtained for @xmath127 [ fig.4 ] . in this
particular dynamics , we observe an interesting phenomenon that there is no exact asymptotic decay of quantum dissension @xmath94 .
we observe the revival of quantum correlation for a certain period of time in the initial phase of the dynamics .
this is something different from the standard intuition of asymptotic decay of quantum correlation when it undergoes dissipative dynamics .
this remarkable feature can be interpreted as that the dissipative dynamics is not necessarily going to decrease quantum correlation with passage of time . on the contrary , depending upon the initial state it can enhance the quantum correlation for a certain period of time .
we refer to this unique feature as _ revival of quantum correlation in dissipative dynamics_. however , the other dissension @xmath95 follows the standard process of asymptotic decay with time . in fig.3(b ) , we also compute the monogamy score and find that states which are initially polygamous are becoming monogamous with the passage of time .
this is contrary to what we observed in case of mixed w states . in this case , the states initially freely shareable ( polygamous ) are entering into not freely shareable ( monogamous ) regime due to channel action .
this is a remarkable feature as this helps us to obtain monogamous state from polygamous state .
this is indeed helpful as monogamy of quantum correlation is an useful tool for quantum security . _
_ + now we provide another example where action of quantum noisy channel can revive quantum dissension for a short period of time in a much smooth manner compared to our previous example . here
, we consider a mixed biseparable state : @xmath128=@xmath129@xmath130+@xmath131[@xmath132@xmath133@xmath134@xmath135+@xmath132@xmath136@xmath134@xmath137 .
the dynamics of this density matrix at time @xmath91 is given by , @xmath138,{}\nonumber\\ & & \rho_{22}=\rho_{33}=\frac{1}{8}(1-\gamma^2)[(1+\gamma)+p(1-\gamma)],{}\nonumber\\ & & \rho_{44}=\frac{1}{8}(1-\gamma)^2[(1+\gamma)+p(1-\gamma)],{}\nonumber\\ & & \rho_{55}=\frac{1}{8}(1-p)(1-\gamma)(1+\gamma)^2,{}\nonumber\\ & & \rho_{66}=\rho_{77}=\frac{1}{8}(1-p)(1-\gamma)^2(1+\gamma),{}\nonumber\\ & & \rho_{88}=\frac{1}{8}(1-p)(1-\gamma)^3,{}\nonumber\\ & & \rho_{14}=-\rho_{23}=\frac{p}{4}(1-\gamma).\end{aligned}\ ] ] at @xmath116 for this state , both @xmath94 and @xmath95 are having the value 0 . however , quite surprisingly , we find that in the initial phase both dissension @xmath94 and @xmath95 increase and attain maximum values ( 0.00133,0.00183 ) and in the subsequent phases the values lower down and finally reach 0 [ fig.5 ] .
this reiterates the fact that for certain initial states the dissipative dynamics acts as a catalyst and helps in revival of quantum correlation .
this dynamics is different from our previous dynamics in the sense that here revival of quantum correlation is much more than the quantum correlation present in the initial state .
this is indeed a strong signature that in multi - qubit cases the channel dynamics can take a zero - correlated to a correlated state .
though the rise of correlation is not very high , however , in nmr systems @xcite this rise is significant as one starting with a zero - correlated state can use the state for computation at subsequent phases of time instead of trashing it away .
the reduced density matrices are separable states for all values of time and purity , making making their discord equal to zero .
here once again we have , @xmath79 = -@xmath95 and hence channel action does not change the monogamy property of the mixed biseparable state . _ _ + the density matrix elements of mixed ghz at time @xmath91 for @xmath139 are given as , @xmath140,\rho_{18}=\frac{p}{2}(1-\gamma)^\frac{3}{2},{}\nonumber\\ & & \rho_{ii}=\frac{1}{8}[1-p(1-\gamma)^2 ] , i=2, ... ,7.\\\end{aligned}\ ] ] here @xmath94 and @xmath95 starts decaying from ( -3.00,1.00 ) at @xmath116 , @xmath117 and approaches @xmath118 after sufficient time [ fig.6 ] .
the decay of @xmath94 is not exactly asymptotic in contrast to the action of gad channel with q=1 .
the decay of @xmath95 is asymptotic as in the case of gad channel with @xmath141 .
the initial state evolves to final population distribution ( diag@xmath142)^{\otimes 3}$ ] , which contains no quantum dissension .
moreover , the reduced density matrices are separable states and do not contribute towards monogamy score .
therefore , dynamics of @xmath79 is same as that of @xmath95 , only differing by a negative sign .
+ _ _ + the density matrix evolution of the mixed w state at time @xmath91 is given by , @xmath143,{}\nonumber\\ & & \rho_{22}=\rho_{33}=\rho_{55}=\frac{1}{24}[3+p(1-\gamma)(3\gamma^2 - 7\gamma+5)],{}\nonumber\\ & & \rho_{44}=\rho_{66}=\rho_{77}=\frac{1}{24}[3-p(1-\gamma)(3\gamma^2 - 5\gamma+3)],{}\nonumber\\ & & \rho_{88}=\frac{1}{8}[1+p(1-\gamma)(\gamma^2-\gamma-1)],{}\nonumber\\ & & \rho_{23}=\rho_{25}=\rho_{35}=\frac{p}{6}(1-\gamma)(2-\gamma),{}\nonumber\\ & & \rho_{46}=\rho_{47}=\rho_{67}=\frac{p}{6}\gamma(1-\gamma).{}\nonumber\\&&\end{aligned}\ ] ] the quantum dissensions @xmath94 and @xmath95 attain values of ( -1.75,0.92 ) at @xmath116 and @xmath117 [ fig.7 ] .
as in the case of mixed ghz state , the curve for @xmath94 is not exactly asymptotic while the curve for @xmath95 is asymptotic . in the limit @xmath144 , a final population distribution of ( diag@xmath142)^{\otimes 3}$ ]
is left resulting in zero quantum dissension . for purity values closer to 1 ,
the initial states are polygamous and they enter into the monogamy regime due to action of gad channel [ fig.8(a ) ] .
the states with purity values closer to 0 are monogamous and do not experience any such transition . hence once again we have one such example where there is a useful transition from polygamous to monogamous regime . _
_ + we consider initial density matrix @xmath145 whose dynamics at time @xmath91 is as follows , @xmath146,{}\nonumber\\ & & \rho_{22}=\rho_{33}=\rho_{55}=\frac{1}{8}[1-p(1-\gamma)(\gamma^2 - 3\gamma+1)],{}\nonumber\\ & & \rho_{44}=\rho_{66}=\rho_{77}=\frac{1}{8}[1+p(1-\gamma)(\gamma^2-\gamma-1)],{}\nonumber\\ & & \rho_{88}=\frac{1}{8}[1+p(\gamma^3 - 1)],{}\nonumber\\ & & \rho_{12}=\rho_{13}=\rho_{15}=\rho_{24}=\rho_{26}=\rho_{34}=\rho_{37}=\rho_{48}=\rho_{56}{}\nonumber\\ & & = \rho_{57}=\rho_{68}=\rho_{78}=\frac{1}{8}(1-p)\sqrt{1-\gamma},{}\nonumber\\ & & \rho_{14}=\rho_{16}=\rho_{17}=\rho_{23}=\rho_{25}=\rho_{28}=\rho_{35}=\rho_{38}=\rho_{46}{}\nonumber\\ & & = \rho_{47}=\rho_{58}=\rho_{67}=\frac{1}{8}(1-p)(1-\gamma),{}\nonumber\\ & & \rho_{18}=\rho_{27}=\rho_{36}=\rho_{45}=\frac{1}{8}(1-p)(1-\gamma)^\frac{3}{2}.\end{aligned}\ ] ] once again it is evident from fig[9 ] , @xmath94 and @xmath95 achieve maximum values ( -1.015,0.15 ) at @xmath116 and @xmath127 .
however , the decay profile of @xmath95 is much smoother than that of @xmath94 . the evolution of monogamy score [ fig .
8(b ) ] is quite different for @xmath139 than that of @xmath141 . here also , all the initial polygamous density matrices enter into the monogamy regime irrespective of the values of parameter p. _ _ + we also studied the dynamics of the mixed biseparable state in presence of gad channel for @xmath139 and we found that both dissensions remain at zero starting from the initial state . in this subsection
, we consider the dephasing channel and its action on various three - qubit states .
a dephasing channel causes loss of coherence without any energy exchange .
the one - qubit kraus operators for such process are given by @xmath96=diag ( 1,@xmath98 ) and @xmath99=diag(0,@xmath147 ) .
+ _ _ + we once again consider the mixed ghz state subjected to dephasing noise .
the density matrix elements of the mixed ghz at a time @xmath91 are given by , @xmath148 here we observe that the diagonal elements are left intact whereas the off - diagonal elements undergo change as a consequence of dephasing noise .
interestingly , we find that @xmath94 is not at all influenced by dephasing channel whereas @xmath95 follows a regular asymptotic path [ fig.10 ] .
the degradation observed in @xmath94 is due to progressively lower purity levels and is unaffected by dephasing noise .
the reduced density matrices do not contribute towards monogamy score , thus making the dynamics of monogamy score just negative of @xmath95 .
+ _ _ + the dynamics of mixed w state subjected to dephasing noise is as follows : @xmath149 we noticed that for @xmath117 , @xmath94 has a slower decay rate compared with other purity values and hence a finite amount of @xmath94 is present for all @xmath91 @xmath74 10 at @xmath117 [ fig.11 ] .
the decay of @xmath95 is asymptotic . for certain values of purity ,
the initial mixed w state is monogamous .
however , they enter into the polygamous regime as a consequence of phase damping noise [ fig.12(a ) ] .
after sufficient time , @xmath79 decays down to zero for all purity values . _
_ + the dynamics of @xmath150 under the influence of phase damping channel is given by : @xmath151 here , @xmath94 exhibits a strong revival all throughout the channel . however , the decay profile of @xmath95 is perfectly asymptotic [ fig.13 ] .
prior to channel action , i.e. at @xmath116 , all density matrices are polygamous . with the action of the dephasing channel , density matrices with mixed ness closer to 1 enter into the monogamous regime [ fig.12(b ) ] .
+ _ _ + for the initial state , @xmath152 ,
the dynamics is given as : @xmath153 both @xmath94 and @xmath95 are zero throughout the channel operation time and do not show any revival .
in the final subsection of this section , we consider the effect of the depolarizing channel on three - qubit states . under the action of a depolarizing channel , the initial single qubit density matrix dynamically evolves into a completely mixed state @xmath30/2 .
the kraus operators representing depolarizing channel action are @xmath96=@xmath154@xmath155 , @xmath99=@xmath156 , @xmath102=@xmath157 , @xmath104=@xmath158 .
( where @xmath159 are pauli matrices ) + _ _ + the dynamics of a mixed ghz state when subjected to depolarizing channel is spelled out as , @xmath140,\rho_{18}=\frac{p}{2}(1-\gamma)^3,{}\nonumber\\ & & \rho_{ii}=\frac{1}{8}[1-p(1-\gamma)^2 ] i = 2, ... ,7.\end{aligned}\ ] ] both @xmath94 and @xmath95 start decaying from the initial values of ( -3.00,1.00 ) [ fig.14 ] . quite interestingly , @xmath94 exhibits smooth asymptotic decay in contrary to the anomalies observed in case of @xmath139 gad channel and dephasing channel .
this instance underlines the fact that a certain noisy environment can largely influence the dynamics of multipartite quantum correlation .
the depolarizing channel transfers the initial mixed ghz state into @xmath30/8 which contains zero quantum dissension . here
the monogamy score @xmath79 of mixed ghz state is just the negative of @xmath95 . _
_ + the dynamics of the mixed w state under the action of depolarizing channel is given by , @xmath143,{}\nonumber\\ & & \rho_{22}=\rho_{33}=\rho_{55}=\frac{1}{24}[3+p(1-\gamma)(3\gamma^2 - 7\gamma+5)],{}\nonumber\\ & & \rho_{23}=\rho_{25}=\rho_{35}=\frac{p}{6}(2-\gamma)(1-\gamma)^2,{}\nonumber\\ & & \rho_{44}=\rho_{66}=\rho_{77}=\frac{1}{24}[3-p(1-\gamma)(3\gamma^2 - 5\gamma+3)],{}\nonumber\\ & & \rho_{46}=\rho_{47}=\rho_{67}=\frac{p}{6}\gamma(1-\gamma)^2,{}\nonumber\\ & & \rho_{88}=\frac{1}{8}[1+p(1-\gamma)(\gamma^2-\gamma-1)].\end{aligned}\ ] ] here , @xmath94 and @xmath95 attain maximum values of ( -1.75,0.92 ) at @xmath116 and @xmath117 [ fig.15 ] .
the initial mixed w state evolves to @xmath30/8 in the limit of @xmath160 resulting in zero quantum dissension .
@xmath94 follows a perfect asymptotic path in contrast to the dynamics observed in case of @xmath139 gad channel and dephasing channel .
the monogamy score @xmath79 evolves as shown in fig.16 . for high purity values closer to 1 ,
the initially polygamous states enter into monogamous regime owing to depolarizing channel action . on the other hand ,
states with low purity values which are initially monogamous do not experience any such transition .
in this work , we have extensively studied the dynamics of quantum correlation ( quantum dissension ) of various three qubit states like , mixed ghz , mixed w , mixture of separable states and a mixed biseparable state when these states are transferred through quantum noisy channels such as amplitude damping , dephasing and depolarizing . in most cases ,
we find that there is an asymptotic decay of quantum dissension with time . however , in certain cases , we have observed the revival of quantum correlation depending upon the nature of initial state as well as channel .
this is quite interesting as we can explicitly see enhancement of multiqubit correlation in presence of local noise ; similar in the line of quantum discord .
+ in addition , we have studied dynamics of monogamy score of three qubit states under different quantum noisy channels .
remarkably , we have seen that there are certain states which on undergoing effects of quantum channels change itself from monogamous to polygamous states .
it is believed that monogamy property of the state is a strong signature of quantumness of the state and can be more useful security purpose compared to polygamous state .
this study is useful from a futuristic perspective where we are required to create monogamous state from polygamous state for various cryptographic protocols .
+ _ acknowledgment _ authors acknowledge prof a. k. pati of harish chandra research institute for his invaluable suggestions in the improvement of the work .
this work was initiated when the first author was at indian institute of science .
a. einstein , b. podolsky and n. rosen , phys .
, * 47 * 777 ( 1935 ) ; j.s .
bell physics * 1 * 3 , 195 , 1964 ; a.aspectet.al .
lett . , * 47 * , 460 , ( 1993 ) ; c.monroe .
lett . , * 75 * , 4714 , 1995 ; r. laflamme.et.al.arxiv:quant-ph/9709025v1 .
l.dicarlo . et .
al.nature * 467 * , 574 , 2010 ; p. neumann .
science * 320 * , 1326 , 2008 ; w.b.gao.et.al .
nature phys * 6 * , 331 , 2010 .
n.gisin .et .
phys . , 74 , 145 , 2002.a .
ekert , phys .
* 67 * , 661 ( 1991 ) ; c. h. bennett and g. brassard , proceedings of ieee international conference on computers , system and signal processing , bangalore , india , pp.175 - 179 ( 1984 ) ; p. w. shor and j. preskill , phys . rev . lett . * 85 * , 441 ( 2000 ) .
m. hillery , v. buzek and a. berthiaume , phys .
a * 59 * , 1829 ( 1999 ) ; s. k. sazim , i. chakrabarty , c. vanarasa and k. srinathan , arxiv:1311.5378 ; s. adhikari , i. chakrabarty and p. agrawal , quant .
inf . comp .
* 12 * , 0253 ( 2012 ) ; m.ray , s. chatterjee and i. chakrabarty , arxiv:1402.2383 .
s. k. sazim and i. chakrabarty , eur .
j. d * 67 * , 174 ( 2013 ) ; d. deutsch , a. ekert , r. jozsa , c. macchiavello , s. popescu and a. sanpera , phys .
77 * , 2818 ( 1996 ) ; l. goldenberg and l.vaidman , phys .
lett . * 75 * , 1239 ( 1995 ) ; a. cabello , phys . rev .
a * 61 * , 052312 ( 2000 ) ; c. li , h - s song and l. zhou , journal of optics b : quantum semiclass . opt . * 5 * , 155 ( 2003 ) ; a.k.pati phys .
a , 63 , 014320 , 2001 ; s. adhikari and b. s. choudhury , phys .
a * 74 * , 032323 ( 2006 ) ; i. chakrabarty and b. s. choudhary arxiv:0804.2568 ; i. chakrabarty , int . j. quant . inf .
* 7 * , 559 ( 2009 ) ; s. adhikari , a.s .
majumdar and n. nayak , phys .
a * 77 * , 042301 ( 2008 ) ; i. ghiu , phys . rev . a * 67 * , 012323 ( 2003 ) ; s. adhikari , i. chakrabarty and b. s. choudhary , j. phys .
a * 39 * , 8439 ( 2006 ) ; v. buzek , v. vedral , m. b. plenio , p. l. knight and m. hillery , phys .
a * 55 * , 3327 ( 1997 ) ; a. orieux , g. ferranti , a. darrigo , r. lo franco , g. benenti , e. paladino , g. falci , f. sciarrino , and p. mataloni , arxiv:1410.3678 ; a. darrigo , r. lo franco , g. benenti , e. paladino , and g. falci , ann . phys .
* 350 * , 211224 ( 2014 ) .
e. knill and r. laflamme , phys .
* 81 * , 5672 ( 1998 ) ; a. datta et al .
lett . * 100 * , 050502 ( 2008 ) ; s.l .
braunstein et al . , phys .
lett . * 83 * , 1054 ( 1999 ) ; d.a .
meyer , ibid .
* 85 * , 2014 ( 2000 ) ; s.l .
braunstein and a.k .
pati , quant .
inf . comp .
* 2 * , 399 ( 2002 ) ; a. datta et al . , phys .
a * 72 * , 042316 ( 2005 ) ; a. datta and g. vidal , ibid .
* 75 * , 042310 ( 2007 ) ; b.p .
lanyon et al . , phys .
* 101 * , 200501 ( 2008 ) .
h. ollivier , w. h. zurek , phys .
lett . * 88 * , 017901 ( 2002 ) ; l. henderson , v. vedral , j. phys .
a * 34 * , 6899 ( 2001 ) ; s. luo , phys . rev .
a * 77 * , 042303 ( 2008 ) ; a.r .
usha devi , a. k. rajagopal , phys .
lett . * 100 * , 140502 ( 2008 ) ; a. datta , a. shaji , c. m. caves , phys . rev
. lett . * 100 * , 050502 ( 2008 ) ; k. modi _ et .
lett . * 104 * , 080501 ( 2010 ) ; m. horodecki , p. horodecki , r. horodecki , j. oppenheim , a. sen(de ) , u. sen , b. synak - radtke , phys . rev . a * 71 * , 062307 ( 2005 ) .
i. chakrabarty , p. agrawal , a. k. pati , eur .
j. d * 57 * , 265 ( 2010 ) ; k. modi et al .
104 , 080501 ( 2010);m .
okrasa and z. walczak , arxiv:1101.6057 ; c. c. rulli . et .
al . phys .
rev . a , * 84 * , 042109 , 2011 . ;
ma and z .- h .
chen , arxiv:1108.4323 .
al.phys . rev .
a , * 80 * , 024103 , 2009 ; i. chakrabarty , s. banerjee and n. siddharth , quant .
11 * , 0541 ( 2011 ) ; j .- s .
xu , k. sun , c .- f .
li , x .- y .
xu , g .- c .
guo , e. andersson , r. lo franco and g. compagno , nature commun . * 4 * , 2851 ( 2013 ) ; b. bellomo , g. compagno , r. lo franco , a. ridolfo , s. savasta , int . j. quant . inf .
* 9 * , 1665 ( 2011 ) ; r. lo franco , b. bellomo , s. maniscalco , and g. compagno , int .
b * 27 * , 1345053 ( 2013 ) ; r. lo franco , b. bellomo , e. andersson , and g. compagno , phys .
a * 85 * , 032318 ( 2012 ) .
t.m.cover and j. a. thomas john wiley & sons .
, 1991 .
lett . , * 105 * , 190502 , 2010 ; k. modi , a. brodutch , h. cable , t. paterek and v. vedral , rev .
phys . * 84 * , 1655 ( 2012 ) ; j - s .
zhang and a - x .
chen , quant .
. lett . * 1 * , 69 ( 2012 ) ; d. girolami and g. adesso , phys .
a * 83 * , 052108 ( 2011 ) ; v. vedral , m. plenio , m. rippin , and p. l. knight , phys .
lett . * 78 * , 2275 ( 1997 ) ; t. r. bromley , m. cianciaruso , r. lo franco , and g. adesso , j. phys . a : math .
47 * , 405302 ( 2014 ) ; b. aaronson , r. lo franco , g. compagno , and g. adesso , new j.phys .
* 15 * , 093022 ( 2013 ) ; b. aaronson , r. lo franco , and g. adesso , phys . rev .
a * 88 * , 012120 ( 2013 ) ; m. cianciaruso , t. r. bromley , w. roga , r. lo franco , and g. adesso , arxiv:1411.2978 ( 2014 ) . | we study the dynamics of quantum dissension for three qubit states in various dissipative channels such as amplitude damping , dephasing and depolarizing . our study is solely based on markovian environments where quantum channels are without memory and each qubit
is coupled to its own environment .
we start with mixed ghz , mixed w , mixture of separable states , a mixed biseparable state , as the initial states and mostly observe that the decay of quantum dissension is asymptotic in contrast to sudden death of quantum entanglement in similar environments .
this is a clear indication of the fact that quantum correlation in general is more robust against the effect of noise .
however , for a given class of initial mixed states we find a temporary leap in quantum dissension for a certain interval of time .
more precisely , we observe the revival of quantum correlation to happen for certain time period .
this signifies that the measure of quantum correlation such as quantum discord , quantum dissension , defined from the information theoretic perspective is different from the correlation defined from the entanglement - separability paradigm and can increase under the effect of the local noise .
we also study the effects of these channels on the monogamy score of each of these initial states .
interestingly , we find that for certain class of states and channels , there is change from negative values to positive values of the monogamy score with classical randomness as well as with time .
this gives us an important insight in obtaining states which are freely sharable ( polygamous state ) from the states which are not freely sharable ( monogamous ) .
this is indeed a remarkable feature , as we can create monogamous states from polygamous states monogamous states are considered to have more signatures of quantum ness and can be used for security purpose . |
pair interactions , simd , gpu , molecular dynamics , verlet list
in most particle simulations , more than half of the computational time is spent in calculating pair interactions with limited spatial range .
when long - range interactions are present , such as electrostatics , the long - range part is usually calculated on a mesh .
certain types of analysis , such as determining particle pair correlation functions , also involve evaluating pair interactions with limited range .
many codes that compute these kind of interactions employ cpu algorithms consisting of a simple double loop to iterate through a list of particle pairs .
this nave approach has a quadratic computational complexity which makes it prohibitively expensive already for moderate numbers of particles .
however , by exploiting the limited interaction range imposed by the typically spherical cut - off , the computational cost can be reduced to linear .
this is achieved by reducing the number of neighboring particles that need to be considered .
to do so the verlet list @xcite and the linked cell @xcite algorithms as well as the combination of the two are widely used . in particular , in molecular dynamics ( md ) simulation codes @xcite these algorithms are most commonly employed .
although these algorithms suffer from limitations on modern simd architectures @xcite , there have been only a few attempts to overcome them , most of them specific to gpus @xcite without achieving generality . before the advent of cpu simd units , the performance of the simple double loop over the neighbor list was quite good as the compiler can usually unroll the inner loop . because the speed of the main memory has not kept up with the processor speed ,
caching became more important . in calculating pair interactions
this means that the location of particles in memory should correlate with their spatial location to increase cache hits .
several publications have dealt with this issue @xcite . however , as the width of the simd units increases , reordering or shuffling the input and output data for convenient access in the simd units becomes a severe bottleneck . when calculating pair interactions between all particle pairs in the system , a perfectly linear memory access pattern can be used that avoids shuffling . however ,
when a cut - off is used , a significant part of the particle neighbor list will not be ordered sequentially .
the relative cost of shuffling depends on the cost of calculating a single pair interaction and on the simd width . in molecular dynamics simulations particles
usually interact via a lennard - jones ( lj ) and a coulomb potential .
when the popular particle - mesh ewald ( pme ) electrostatics method @xcite is used , a complementary error function must be calculated .
pennycook et al .
@xcite provide a detailed analysis of the shuffling ( also called gather - scatter ) and their impact on performance with only lj interactions considered . in their work , with 8-way simd reordering instructions represent a third of the total , with 16-way simd the ratio is more than a half . in practice , the performance is affected even more . since shuffling introduces more data dependencies between instructions , reducing the instructions available for scheduling will result in low instructions per cycle ( ipc ) .
we will show that even when calculating lj and pme interactions , the shuffling ends up taking more than half of the time with 4-way simd .
on gpus , shuffling data is typically not required as the execution model allows hardware threads to access data from different memory locations . however , loading particle data requires scattered memory access which will waste gpu memory bandwidth as well as cycles ( due to instruction replay ) and will render a standard implementation memory bound . moreover
, the throughput - oriented gpu architecture requires high level of parallelism and is sensitive to memory access patterns . in order to target gpus
, some codes combine the traditional algorithms with data regularization techniques @xcite , but such approaches can still lead to inefficient execution . recasting the algorithms to a more regular data access has been shown to result in higher ipc on gpus , but not without additional trade - offs @xcite .
although on cpus the relative memory bandwidth is higher , the data dependencies can still cause bottlenecks in simd - optimized algorithms .
the main issues faced when considering data parallelization in traditional particle - pair based neighbor - lists schemes are the irregular sizes and non - contiguous nature of the neighbor lists of each particle .
we propose to address both of these issues by considering pair - interactions between clusters of particles of fixed size , similar to the work of friedrichs et al .
however , important distinguishing features of our algorithm are high parallel work - efficiency and the inherent flexibility which enables tuning for the simd width and other specifics of the hardware . by changing the size of the clusters
, our algorithm can be adapted to simd units of different widths .
adjusting the cluster size also allows tuning the number of operations `` in flight '' as well as the ratio of arithmetic to memory operations .
this flexibility , together with the high ratio of arithmetic to load / store operations , ensures that the algorithm can reach high performance on current , as well as future cpu and gpu hardware .
it is also well suited to more exotic hardware such as fpgas , but as the implementation is still ongoing , result will be reported in the future . in case of cpus , the additional major advantage is that , by matching the cluster size to the simd width , no shuffle operations are required at all .
this not only improves performance by at least a factor of 2 , but also makes the code much easier to write and read .
there is a price to pay for the improvements as the cluster pairs will contain particle pairs in addition to the ones in the original interaction sphere .
this results in extra interactions calculated between particles otherwise not within range , which we know will evaluate to zero . as we will show later ,
although this does lead to reduction in algorithmic work - efficiency , the performance gain still outweighs the extra cost .
we would like to note that the algorithm operates on the lowest level of the interaction calculation and any optimization available in the literature can be applied . for md ,
we use it together with a verlet buffer .
furthermore , all parallelization strategies developed for traditional algorithms can be used with little or no modification .
we have designed and implemented non - bonded pair interaction kernels for x86 sse2 , sse4.1 , avx and avx+fma ( amd bulldozer ) simd architectures , as well as nvidia gpus .
the kernels utilize lj interactions and monopole - monopole electrostatic interactions of general form .
we implemented analytical electrostatics kernels for reaction - field ( rf ) and pme , as well as tabulated electrostatic potentials .
we plan to support sphero - symmetric potential of arbitrary shape through tabulated interactions .
while the required additional table lookups per pair will lower the efficiency of the kernels on current cpus , on gpus and with avx2 ( which will support table lookups ) performance should be good .
the algorithms described here have been implemented in the gromacs molecular simulation package @xcite and are available in the official version 4.6 release , combined with hybrid mpi+openmp parallelization .
the source code can be obtained under the lgplv2 license from http://www.gromacs.org .
note that the cpu kernels in gromacs 4.6 have an additional optimization , not discussed in this paper , for systems where less than half of the particles have lj interactions . for water this
improves kernel performance by up to 10% .
we are looking for an algorithm that can execute single instructions on multiple data ( simd ) , while not being limited by loading and storing data from and to ( cache-)memory .
the standard implementation of the verlet - list algorithm loads a particle and calculates pair interactions by looping over its neighbors . thus a single pair interaction is calculated for each particle load and store .
the relatively cheap interactions in md simulations render this algorithm effectively memory bound .
to remedy this , our algorithm loads a cluster of @xmath0 particles and calculate @xmath0 interactions for each neighbor loaded .
this increases the data reuse by a factor of @xmath0 .
the loop over neighboring particles is replaced by a loop over clusters consisting of @xmath1 particles .
the values of @xmath0 and @xmath1 will be tuned for the simd hardware .
the standard implementation of the verlet - list algorithm can be seen as a special case of this cluster algorithm where @xmath0=1 and @xmath1=1 . in general ,
the easiest way to achieve simd parallelization is to let the compiler vectorize loops , possibly with the help of the programmer aided by feedback from the compiler . at a first glance
this might seem to be a good strategy since a particle usually has hundreds of neighbors which leads to long vectorizable loops .
for efficient loading , the order of particles in memory needs to be strongly correlated with spatial ordering to increase cache hits .
ideally , sequential particles would be loaded in groups of size equal to the simd width , but this not compatible with a spherical interaction volume . even when particles can be loaded in groups , vectorizing the inner - loop
will only give a small speed - up on wider simd units , as memory operations and data shuffling can take more time than the actual calculation .
for lj only with fixed parameters on avx 8-way simd , memory and shuffling operations account for 32 of the 70 operations @xcite ; with parameter loading , the ratio increases beyond 50% . when calculating all interactions of neighbors with one particle , we need to load 3 coordinate components , 3 parameters , as well as load and store 3 force components for each neighbor . in theory , on current cpus this should not lead to a memory - bound algorithm , but in practice performance will be far from peak due to limitations on the instruction scheduling .
the coordinates are loaded per particle as triplets of @xmath2 , @xmath3 , @xmath4 requiring data - shuffling .
the wider the cpu simd unit is , the more data shuffling is required and the longer the dependency chain gets between loading data , computation and storing forces .
hence , for efficient simd calculations it is very advantageous to use packed sequences of coordinates , e.g @xmath5 , @xmath6 and @xmath7 with 4-way simd .
on gpus , such packing is not needed as vector types are supported , but a much higher arithmetic to memory operation ratio is required to achieve peak performance . constructing the neighbor list , also called pair list , is a similar operation , but with less arithmetic , which makes it even more memory intensive .
although the pair list is usually not reconstructed every step , it involves looping over more pairs than the non - bonded kernel processes , so this can become a limiting factor . the only way to hide the latency of memory operations is to perform more calculation per load / store operation . at first sight
this might seem impossible , but this can actually be achieved with a simple scheme .
the basic idea behind our work is to spatially cluster particles in groups of fixed size and use such a cluster as the computational unit of our algorithm .
these groups can then be mapped directly to the simd hardware units , which have a fixed width .
given a 4-way simd unit , we can spatially cluster particles in groups of 4 .
we can load a cluster of 4 , so called , @xmath9-particles in simd registers and then loop over the neighboring clusters of 4 , so called , @xmath10-particles ( see fig .
[ simd ] ) . with this @xmath8 = 4@xmath114 setup ,
we compute 16 pair interactions while only performing memory load and store operations for 4 @xmath10-particles . after having looped over all neighboring @xmath10-clusters of an @xmath9-cluster , usually a few hundred , we also have to do memory operations for the @xmath9-particles , but the cost of this is negligible . in this example
the memory bandwidth is reduced by a factor of 4 , but more importantly , as we always access particles in cluster of size 4 , we can organize all data packed in groups of 4 . this eliminates the need for data shuffling which is the main performance bottleneck of the standard way of calculating non - bonded interactions on simd units .
this is the simplest version of the algorithm .
the same 4@xmath114 clusters can also be processed on 8-way simd hardware .
then two i - clusters are loaded in one simd register and each j - cluster is duplicated in one simd register .
this setup halves the number of arithmetic operations and adds a few shuffle operations . in cuda ,
memory access is more flexible .
hardware threads on nvidia gpus are organized in `` warps '' . on current gpus ,
each warp consists of 32 threads which execute the same instruction every cycle .
this results in a simd - like execution model called single instruction multiple threads ( simt ) . unlike the simd model which requires explicit programming for the simd width
, the simt architecture allows thread - level parallel programming , and the warp - based lockstep execution model needs to be considered only for performance .
this enables more flexible memory operations ( different addresses in different threads ) and divergence among threads in a warp .
the simt model allows spreading out all particle pairs in an 8@xmath114 cluster pair over the 32 threads of a warp , thus processing one particle pair on each thread .
we illustrate this in fig .
[ simd ] , for the sake of example using 16-way simt . 1 and the 4@xmath114 setups with 4-way simd and 16-way simt .
all numbers are particle indices , each black dot represents an interaction calculation and the arrows indicate the computational flow .
the simd registers for @xmath9- and @xmath10-particles are shown in green and blue , respectively .
the 4@xmath114 setup calculates 4 times as many interactions per particle load / store and requires fewer memory operations ( shown in red ) . unlike the 1@xmath111 setup , the 4@xmath114 setup does not require data shuffling in registers .
[ simd],width=302 ] we will now describe in detail how the algorithm works , starting with building the cluster pair list . the algorithmic unit of particle data representation is a cluster rather than a single particle . beside this
minor , but important difference , the overall algorithm closely follows the standard verlet or neighbor list setup .
hence , in the following , unless explicitly stated , pair list will refer to a list of cluster pairs .
note that the cluster pair list this work uses as data representation does not define a strict particle - particle in - range relationship because , as we will show later , the list by design includes particles not in - range .
moreover , the presented algorithms use newton s third law to calculate pair interactions , hence the pair list contains each pair only once , not twice .
since for each particle there is no explicit list of all particles in its neighborhood , we prefer the term `` pair list '' to the term `` neighbor list '' .
we construct a pair list using a verlet buffer ( also called `` skin '' ) which is essentially an extension of the cut - off distance to account for particle movement allowing the list to be retained for a number of steps @xcite .
the exact number depends on the relative cost of the list construction and the dependence of the buffer size on the lifetime of the list .
pair interactions are then determined for the fixed list of particle pairs defined by pairs of clusters . in the most general case ,
we need to generate a pair list of clusters of size @xmath0 particles versus clusters of size @xmath1 . in the simplest setup
, the simd width will be equal to @xmath1 , but a width of @xmath12 , where @xmath0 is divisible by @xmath13 , will also work . on a gpu
the best performance will be achieved when matching @xmath8 to the width of the simt execution model , i.e. 32 for cuda .
first we need to group the particles into clusters of fixed size . to minimize the number of additional particle pairs in the pair list , the clusters need to be as compact as possible .
a simple and efficient way of generating compact , fixed - size clusters is spatial gridding in two dimensions and spatial binning in the third dimension , see fig .
[ clustering ] .
first we construct a rectangular grid along x and y with a grid spacing of @xmath14 , where @xmath15 is the particle density .
then we sort the particles into columns of this grid . for each column
we sort the particles on z - coordinate and as a result we get the spatial clusters as consecutive groups of @xmath0 or @xmath1 particles . because the number of particles in a column is typically not a multiple of @xmath0 , we add dummy particles to the last cluster when needed .
the fraction of dummy particles is @xmath16 ; with 10000 particles and clusters of size 8 this gives 4% dummy particles in the cpu algorithm .
for the gpu we use a hierarchical cluster setup . as we can store 8 i - clusters in shared memory , we group 2@xmath112@xmath112=8 clusters of size 8 together .
this reduces the number of dummy particles to 1% with 10000 particles .
all these operations can be done efficiently in linear time .
the next step is calculating bounding boxes for each cluster , this can be done using simd instructions , as the number of particles in a cluster is constant . in the case of @xmath17 ,
adjacent pairs of bounding boxes are combined to generate clusters of double the number of particles .
a pair list can then be constructed by checking distances between the bounding boxes .
this is very efficient , as it requires one bounding box - pair distance check for @xmath18 particle pairs .
however , this results in more cluster or particle pairs than strictly necessary , as bounding boxes might be within range while none of the particle pairs falls within range . to avoid this overhead we prune pairs of clusters at distances close to the cut - off using a particle - pair distance criterion . for the gpu implementation
, the pair list construction is performed on the cpu , but the pruning is done on the gpu where this can be done more efficiently .
periodic boundary conditions can be implemented in a simple and efficient fashion by moving the @xmath9-clusters by the required periodic image shifts and storing these shifts in the cluster pair list for use during the pair interaction calculation .
-cluster list in green for the red @xmath9-cluster .
[ clustering],width=226 ] .... / * bb = cluster bounding box * / for each i - cluster ci determine grid range in x within rlist of bb[ci ] for each grid cell gx in range determine grid range in y within rlist of bb[ci ] for each grid cell gy in range determine j - clusters at gx , gy within rlist of bb[ci ] for each j - cluster cj in range if ( bbdistance(ci , cj ) < rlist ) if ( bbdistance(ci , cj ) < rbb or atomdistance(ci , cj ) < rlist ) put cj in cjlist[ci ] set forcefield exclusion masks in cjlist[ci ] .... at this point we would like to note that the cluster pair list , being simply a verlet list of particle clusters , can be seen as a generalized version of the classical neighbor list .
consequently , the neighbor list corresponds to the special case of cluster size @xmath19 and in the following we will refer to it as the 1@xmath111 scheme .
the cluster - based pair list contains inherently more particle pairs than the ones within the cut - off radius , the number of pairs for different @xmath8 is shown on fig .
[ zeros ] .
the fraction of extra interactions increases rapidly with the cluster size and decreases rapidly with the cut - off radius .
however , the increase in the efficiency of the presented algorithm should outweigh the cost of calculating these extra interactions .
as can be seen in fig .
[ zeros ] , when it comes to the number of extra pairs , lists with @xmath17 are less favorable than @xmath20 which results in clusters with close to cubic shape .
this shape minimizes the number intersecting cut - off spheres , resulting in a more compact list .
pair lists normalized by the average number of pairs in a sphere of radius @xmath21 , as a function of the pair list radius @xmath21 , for a 3-site water model , number density @xmath15=100 nm@xmath22 .
[ zeros],width=283 ] having constructed the pair list , for each cluster we now have the lists of all clusters in range .
we still need to take care of particle pairs that need to be excluded .
there are three types of exclusions .
two of those occur within cluster self - pairs . here
particle pairs occur twice , whereas we should only calculate them once and there are self interactions .
we want to calculate each pair interaction only once and skip the self interactions .
these two types of exclusions are handled in the pair interaction kernel .
additionally , there can be exclusions defined by the force field .
normal lj and electrostatic interactions should not be calculated for such excluded pairs of particles , whereas the rf or pme correction should still be applied . to treat these exclusions in the non - bonded kernels , we encode them in a bitmask stored per cluster pair in the pair list .
this compact representation saves memory and also allows for easy and fast decoding of exclusions using bitwise operations .
on cpus we sort the pair list according to the presence of exclusions so we only need to mask exclusions when really needed .
this improves performance by 15% .
the total computational cost of the pair list construction is proportional to the number of particles . to understand how the total cost scales with different parameters , it is worth looking into the details of the different tasks involved .
computational cost in terms of cycles as a function of pairs per particle is shown in fig .
[ cycles_npair ] . for the implementation of the sorting ,
after the gridding , we assume that the particle distribution is homogeneous on longer length scales , but other suitable sorting techniques can be applied in the inhomogeneous case .
then a simple pigeonhole sorting is used , which scales linearly and provides good performance .
the cost of the pair search is not proportional to the number of pairs , as only the boundary of the interaction sphere needs to be determined .
this cost is high when the number of pairs is small compared to @xmath0 and @xmath1 and it is proportional to the radius squared for large radius .
when the radius is large , the cost of search decreases proportionally with @xmath8 .
this makes the search far more efficient than a particle - pair based search .
another implication of the much lower number of ( now cluster- ) pairs , is that advanced search algorithms , such as interaction sorting @xcite , will not help and a simple search algorithm performs well . finally , the cost of interaction calculation is proportional to the total number of pair interactions calculated .
but the overhead of pairs beyond the cut - off distance decreases with increasing number of pairs .
the cost of the search and lj+pme force calculation on cpus are similar , as can been seen in fig .
[ cycles_npair ] ; rf kernels are about twice as fast , which makes the search relatively more expensive .
the optimal balance between search cost and extra cost due to the verlet buffer is usually achieved with a pair list update interval of 10 and 20 when only using a cpu . on the gpu ,
the interaction throughput is much higher which makes the cpu search relatively more expensive . as mentioned , before , we do most of the cluster - pair pruning on the gpu , which reduces the cpu search cost significantly , as can been seen in fig .
[ cycles_npair ] . depending on the speed of the cpu versus the gpu , and especially the number of cores in each ,
the optimal pair list update interval is between 10 and 50 .
furthermore , as the search algorithm maps well to gpus @xcite , we plan to port it in the near future .
, for 8@xmath118 the search cost is also shown for checking bounding box distances only .
the pair count is within the spherical volume , not including the extra pairs due to the irregular cluster - pair volume .
note that the wiggles on the curves for searching are caused by jumps in the number of grid cells fitting in a cut - off sphere .
all timings were done on an intel sandy bridge cpu with single precision 256-bit avx kernels using a single thread .
[ cycles_npair],width=283 ] when the pair list needs to be updated for every interaction calculation , the particle - pair distance based pruning should be skipped and replaced by a conditional in the interaction kernels . with very cheap interactions , such as for a pair correlation function calculation ,
no conditional should be used at all .
as in molecular simulations usually more than half of the computational time is spent in the calculating non - bonded pair interactions , it is well worth carefully optimizing these kernels .
we now have a list of cluster pairs of @xmath0 versus @xmath1 particles . as can be seen in fig .
[ code_kernel ] , writing a simd kernel for this setup is rather straightforward .
however , achieving optimal performance is not trivial .
it is often very hard to judge how close the kernel performance is to the maximum achievable performance , as it depends both on hardware characteristics , mainly the type of simd unit and the performance of the cache system and load / store units , as well as on software characteristics , mainly the compiler(s ) used .
.... for each ci cluster load m coords+params for ci for each cj cluster load n coords+params for cj / * these loops are unrolled using simd * / for j=0 to m for i=0 to n calculate interaction ci*m+i with cj*n+j store n cj - forces store m ci - forces .... for cpus we chose to write the kernels in c with extensive use of sse and avx simd - intrinsics as the current gnu and intel compilers do a good job at optimizing such code and typically achieve better performance across multiple architectures than equivalent hand - written assembly .
for gpus we chose to concentrate on the nvidia cuda programming model as the available development tools are more mature and provide higher performance than that of the alternatives .
there are two main factors that affect kernel performance and require special attention .
one is the choice of @xmath0 and @xmath1 , the other is the treatment of the exclusion and cut - off checks . for the latter the options are using conditionals , which should be avoided on cpus , or masking interactions using bitmasks .
masking is usually more efficient than conditionals .
on cpus simd bitwise and operations are used for masking , whereas on gpus we simply multiply by 0 or 1 and use a conditional for the cut - off check . using a conditional can reduce the number of instructions issued when all pairs stored in a simd register are beyond the cut - off distance . on the cpu this should only be used when all @xmath8 pairs are beyond the cut - off , as otherwise the force reduction cost increases .
this only improves the performance when an overly long pair list buffer is used , so we only use a conditional for the 1@xmath111 kernels where it helps in most cases . with the latest cpu compilers ,
not much code optimization is required , as long as the fastest possible intrinsic is used for the respective instruction set , e.g. sse2 , sse4.1 or avx . in cuda optimization is less straightforward ; as the architecture is changing rapidly , compilers and drivers are less mature .
additionally , gpus are massively parallel processors with more simple cores than cpus , which puts more burden on the programmer and compiler to pick the right optimization which might not even carry across hardware generations .
the main goal is to keep the computational units as busy as possible by avoiding stalls due to dependencies on memory operations or instruction latencies .
the ratio of compute to memory operations scales with @xmath0 , as for each loaded @xmath10-particle , @xmath0 interactions with @xmath0 @xmath9-particles are calculated in a single inner - loop iteration . on the cpu it turns out that the best performance is achieved for @xmath0=4 . while using 2 @xmath9-particles
is also possible , with 4 there seem to be enough arithmetic instructions fed to the scheduler to hide most memory operations .
using @xmath23 will lead to a marginally higher ipc and flop rate , at the expense of calculating many more zeros .
the choice of @xmath1 depends on the the simd width . here , for cpus
, we only consider value of @xmath1 equal to the full or half simd width . fitting two @xmath10-clusters in a simd - width
will simply halve the number of arithmetic operations and add a few shuffle operations .
the instruction count is largely independent of @xmath1 .
the only exception is the table lookup for tabulated electrostatics , the number of which scales with both @xmath0 and @xmath1 .
the precision , either single or double , also does nt affect the instruction count , except for the inverse square root operation , which needs an extra newton - raphson iteration in double precision .
an issue specific to cpu simd kernels is that lj pair parameter look - up is costly , as one simd load operation is required for each particle pair . with geometric or lorentz - berthelot combination rules only two loads are required per cluster pair , latency of which can be hidden with computation .
the cuda kernels turn out to be instruction latency limited , not memory limited , although this requires some tricks and a tight packing of the pair list in memory .
the pseudocode of the kernel is shown in fig .
[ code_kernel_gpu ] .
the inner loop calculates interactions between clusters of @xmath0=8 and @xmath1=4 , this way 32 threads of a warp calculate a pair interaction of an entire cluster pair simultaneously .
we chose @xmath1 smaller than @xmath0 such that we have more computation per memory operation .
we group cluster pairs two by two in the pair list , hence the pair search can be done on clusters of 8 particles and the computation on two 8@xmath114 clusters independently on two warps . additionally , we store clusters in range for 8 @xmath9-clusters in a single pair list with this further improving the data reuse .
a @xmath10-cluster interacts with half of these 8 @xmath9-clusters on average , which additionally reduces the memory pressure by a factor of 4 . as the pseudocode on fig .
[ code_kernel_gpu ] shows , non - interacting @xmath10-clusters are skipped based on an bitmask - encoded cluster interaction mask , which only causes a minor overhead .
this hierarchical grouping requires minor modifications in the pair - search code , only storing the packed exclusions masks becomes more complex . with this setup we can load the coordinates , atom types , and charges for 64 @xmath9-particles in registers on the gpu and thereby maximize the number of calculations per @xmath10-particle load .
two warps in a cuda thread block operate on a group of 8 @xmath9-clusters and their @xmath10-cluster neighbors . as the two warps by definition access different @xmath10-particles
, they can run independently and no synchronization is required during computing . on the fermi architecture partial forces are accumulated in registers and reduced in shared memory . in contrast , the kepler architecture provides a special `` warp - shuffle '' operation which can be used for efficient synchronization - free warp - level reduction .
after a lot of testing and optimization , the cuda kernels turned out to be compact and more readable than the cpu simd kernels .
more code is required for managing gpu device initialization , kernel launches and transfers between cpu and gpu . ....
/ * each of the mxn i - j pairs is assigned to a thread .
the sci i - supercluster consists of 8 ci clusters . * / sci = thread block index for each ci in sci load i - atom data into shared mem . / * loop over all cj in range of any ci in sci * / for each cj cluster load j - i cluster interaction and exclusion mask / * per warp * / if cj not masked / * non - interacting cj - sci * / load j - atom data / * loop over the 8 i - clusters * / for each ci cluster in sci if cj not masked / * non - interacting cj - ci * / load i atom data from shared mem .
r2 = sqrt(|xj - xi| ) extract excl_bit exclusion / interaction bit for j - i pair if ( ( r2 < rc_squared ) * excl_bit ) calculate i - j coulomb and lj forces accumulate i- and j - forces in registers reduce j - forces reduce i - forces .... calculating energies is only required infrequently in molecular dynamics , therefore we will concentrate on force - only kernels .
the lennard - jones force is : @xmath24 where @xmath25 is 0 for excluded particle pairs and 1 otherwise .
the lj coefficients @xmath26 and @xmath27 can be different for each atom pair @xmath9-@xmath10 . in practice
there is a limited number of atom types and often combination rules are used to obtain the parameters between two atom types . with
x86 simd instructions loading arbitrary pair parameters can be costly due to the many load and shuffle operations required .
using combination rules , either geometric or lorentz - berthelot , is more efficient .
the electrostatic interaction form we consider is : @xmath28 where @xmath29 is the long - range correction force . for reaction - field
electrostatic we have @xmath30 with @xmath31 a constant .
this can be evaluated efficiently analytically .
for pme we have @xmath32 , with @xmath33 a constant .
evaluation of the pme correction force is more costly .
but as it is bounded and very smooth , linear table interpolation can reach full single precision with a limited table size . as a second option we consider an analytical approximation using a quotient of two polynomials .
this requires 24 multiplications and additions and one division to reach full single precision .
fused multiply - add ( fma ) instructions , currently available on gpus and the amd bulldozer microarchitecture , can speed up this polynomial evaluation significantly .
achieving good performance of load and store intensive kernels requires detailed understanding of many low - level software optimization aspects : simd instruction set , throughput and latency of instructions on different processor microarchitectures , cache behavior , as well as experience with compiler - related performance issues .
unfortunately , it takes a lot of time and effort to reach optimal performance .
fortunately , this effort is required infrequently and our results can be used by anyone , as our compute - kernels are released as part of an open source project , freely available for anyone to use . to compare the different variants of the algorithm
, we focus on the intel sandy bridge cpu architecture .
the reason for this is that at the time of writing this architecture supports avx , the newest and widest simd instruction set on the x86 platform , and it also provides 256-bit operations .
this allows a direct comparison between the 4@xmath114 and 4@xmath118 setup , as well as between 128- and 256-bit avx . for comparison with other architectures
, we also show results on the amd bulldozer architecture using 128-bit avx and fma instructions as well as nvidia fermi ( gf100 ) and kepler2 ( gk110 ) @xcite gpus . as all cpu architectures in focus support
the avx instruction set , form here on 128- or 256-bit simd will refer to avx instructions of the respective type .
we report all performance data in cycles which depends only on the microarchitecture , but not the exact cpu or gpu model .
all cpu kernels were compiled with the gnu c compiler version 4.7.1 with ` -o3 ` as the only optimization .
other optimization - related compiler options did not improve the non - bonded kernel performance .
both intel c compiler versions 12.1.3 and 13.1.1 produced slightly slower code even with cpu architecture - specific optimizations enabled .
the gnu c compiler has greatly improved with recent versions , the difference between version 4.5.3 and 4.7.1 on sandy bridge with analytical ewald kernels is 22% while with recent intel compilers even slight regressions have been observed .
the cuda gpu kernel were compiled with the cuda compiler version 5.0.7 with ` -use_fast_math ` as well as the architecture - specific optimization options ` -arch = sm_20 ` and ` -arch = sm_30 ` for fermi and kepler2 gpus , respectively . on the cpu we store the properties of the @xmath0 i - particles in simd registers and loop over the list of clusters of @xmath1 j - particles .
the pair interactions for the @xmath0 different i - particles are not interdependent , except that we want to load and store the j - particle properties only once .
there are several choices to be made when transforming this algorithm into actual code . for instruction
( re-)scheduling it is advantageous to write out the @xmath0 operations for the i - particles , so it is clear to the compiler that it can reorder them . for most kernels @xmath1 matches the simd width , but for the 256-bit flavor
we also consider @xmath1=4 , which is half the simd width . on new architectures with wider simd units , such as intel mic with 16-way simd in single precision , having @xmath1 smaller than the simd width is even more important .
performance of the most important flavors of the fully optimized kernel versions is reported in table [ kernperf_rf ] and table [ kernperf_ewald ] for rf and ewald , respectively .
the metrics shown in table [ kernperf_rf ] and table [ kernperf_ewald ] represent the peak performance of the respective kernels .
the performance of cpu kernels is constant in the regime of 100 - 100000 particles .
in contrast , gpus are massively parallel multi - processors which require a high level of data - parallelism and hence many particle - pairs to reach peak performance .
the cuda kernels are within 5% of the peak performance from around 20000 particles ; the scaling depends both on generation of architecture and number of multiprocessors .
we present four different performance metrics .
the the number of pairs calculated per 1000 compute cycles ( pairs / cycle ) is the only relevant measure for the raw performance of the algorithm .
the instructions per cycle ( ipc ) provides an estimate of the hardware utilization .
the last two are the number of floating point operations per pair ( flops / pair ) and per cycle ( flops / cycle ) , where we try to minimize the former and maximize the latter . as a reference we show performance for 1@xmath111 kernels which fill the simd unit by unrolling the inner loop over @xmath10 .
these kernels do not use lj combination rules , as parameters need to be looked up either way , which saves two floating point operations per pair .
this standard way of employing simd results in low performance and low flop rates ( the theoretical peak rate for intel sandy bridge is 8 for 128-bit and 16 for 256-bit instructions , respectively ) .
the high measured ipc indicates that the instructions are scheduled very efficiently .
however , a large part of the instructions load , store and shuffle data , rather than doing computation .
the 256-bit rf kernel is only 13% faster than the 128-bit variant while it has similar ipc .
as both kernels execute the same arithmetic instructions , the observed rather small performance increase is explained by the overhead of shuffle and data load operations .
we aim to address these bottlenecks with the proposed algorithms by reducing the need for shuffles and loads . in comparison to the work of pennycook @xcite , here ,
the effect is much more pronounced as we need to load two lj parameters and a charge per @xmath10 particle , while they only implement an lj potential with fixed particle type . the large drop in performance
when using a single thread shows that the 1@xmath111 kernels are mainly limited by instruction scheduling and hyperthreading ( ht ) improves performance by offering the possibility of scheduling instructions from both threads running on the same physical core .
in double precision with 128-bit avx we can use 2-way simd and we can compare the performance for small @xmath0 and @xmath1 .
the 4@xmath112 rf kernel is 26% faster than the 2@xmath112 kernel , which outweighs the negative impact of zero interactions in most cases .
additionally , the pair search for 2@xmath112 takes significant time .
this shows that @xmath34 is not a viable option and we therefore only consider @xmath35 or larger . in the 256-bit kernels we can use the 4@xmath114 scheme which gives 50% higher performance than 4@xmath112 and even more on a single thread .
we continue with the single precision kernels for different functional forms . with 256-bit avx the @xmath8 rf kernels have a 3.3 times higher pair rate than 1@xmath111 , for ewald this factor is 2.2 .
this shows that our approach works .
256-bit is 25% to 65% faster than 128-bit depending on the interaction type and the of use ht .
the performance of the analytical ewald kernels is similar to that of the tabulated version with ht , even though the flop rate is very different . without ht
the tabulated kernels get significantly slower because of the latencies involved in reading table entries .
the amd bulldozer , in contrast with the simultaneous multi - threading intel ht implements , uses a cluster multi - threading architecture with much of the functional units , including simd units , shared between a pair of cores organized in a so called module .
therefore , we compare performance of a hyperthreaded core on intel with a module on amd , both of which support two threads . even though bulldozer has double the theoretical throughput of 4-way simd instructions and fma gives another doubling of the theoretical flop rate , the performance is only marginally higher than 128-bit simd on sandy bridge .
moreover , sandy bridge using 256-bit simd provides a 20% higher pair rate .
the cuda gpu kernels provide significantly higher performance when comparing one streaming multiprocessor with one cpu core .
the analytical and tabulated ewald kernels have similar performance , the former being slightly faster on the kepler architecture even though this kernel executes about 10% more instructions .
this is explained by the fact that the additional instructions are mainly fma - s and intrinsics which allow higher instruction level parallelism , higher ipc and better absolute performance than the texture - based table loads .
the analytical pme kernels achieve about half of the real - world peak flop rate , which is mainly because they do nt contain enough fma instructions .
also , the presence of conditionals for checking the interaction of each of the 8 i - clusters in a super - cluster deteriorates the performance by 15% .
ccccccccc precision & simd width & @xmath18 & pairs / kcycle & 1 thread & ipc & flops / pair & flops / cycle + single & 4 & 1@xmath111 & 67 & @xmath3623% & 2.32 & 38 & 2.6 + single & 8 & 1@xmath111 & 76 & @xmath3624% & 2.16 & 38 & 2.9 + + single & 4 & 4@xmath114 & 175 & @xmath3619% & 2.36 & 40 & 7.0 + single & 8 & 4@xmath114 & 223 & @xmath3627% & 1.96 & 40 & 8.9 + single & 8 & 4@xmath118 & 248 & @xmath362% & 1.68 & 40 & 9.9 + + double & 4 & 2@xmath112 & 52 & @xmath3630% & 1.74 & 45 & 2.3 + double & 4 & 4@xmath112 & 66 & @xmath3625% & 2.16 & 45 & 3.0 + double & 8 & 4@xmath114 & 98 & @xmath3610% & 1.58 & 45 & 4.4 + cccccccccc pu & simd width & ewald & @xmath18 & pairs / kcycle & 1 thread & ipc & flops / pair & flops / cycle + sb & 4 & ana . &
1@xmath111 & 51 & @xmath3631% & 2.20 & 66 & 3.4 + sb & 8 & ana . & 1@xmath111 & 63 & @xmath3616% & 1.98 & 66 & 4.2 + + sb & 4 & tab . & 4@xmath114 & 111 & @xmath3615% & 2.42 & 43 & 4.8 + sb & 8 & tab . &
4@xmath114 & 147 & @xmath3626% & 2.26 & 43 & 6.3 + sb & 8 & tab . & 4@xmath118 & 134 & @xmath366% & 1.88 & 43 & 5.8 + sb & 4 & ana . & 4@xmath114 & 110 & @xmath3614% & 2.40 & 68 & 7.5 + sb & 8 & ana . &
4@xmath114 & 139 & @xmath3611% & 1.76 & 68 & 9.5 + sb & 8 & ana . & 4@xmath118 & 137 & + 1% & 1.52 & 68 & 9.3 + + bd & 4 & ana . &
4@xmath114 & 114 & @xmath3623% & 2.16 & 68 & 7.8 + + fermi & 32 & tab . &
8@xmath114 & 549 & & 1.66 & 41 & 24 + kepler2 & 32 & tab . &
8@xmath114 & 1130 & & 3.2 & 41 & 46 + kepler2 & 32 & ana . & 8@xmath114 & 1151 & & 3.7 & 69 & 85 +
the number of pair interactions calculated in a cycle reflects how a non - bonded algorithm performs on a certain hardware .
however , this measure is not the best indicator of the effective performance , since both the buffer region and the cluster - pair scheme add interactions beyond the cut - off , which by definition evaluate to zero . to get a useful pair interaction rate which reflects the absolute performance , only the non - zero interactions should be considered . as shown on fig .
[ zeros ] , at a commonly used cut - off distance of 1 nm the 4@xmath114 setup adds 86% additional pair interactions .
however , as shown later , with a more than doubled pair interaction evaluation rate , the 4@xmath114 still proves to be faster than 1@xmath111 .
moreover , in certain cases we can actually make use of the extra pairs that the cluster scheme @xmath37 kernels add . in molecular dynamics
the standard procedure is to ensure that no interacting pair of particles is ever missed .
this is usually done by generating a pair list with a so - called verlet buffer @xcite which allows particles to move over a small distance without invalidating the pair list . at any step ,
if any of the particles has moved by more than half the buffer length , the pair list is regenerated @xcite .
this condition is sufficient , but not necessary .
the pair list needs to be invalidated only when the distance between a pair of particles decreases by the buffer length , which will happen far less frequently .
additionally , this commonly used setup is inconvenient for parallel simulations . in practice
, we can often tolerate small imperfections in the pair list . in a constant temperature ensemble
perfect energy conservation is not a requirement , as a thermostat will remove excess heat .
moreover , the amount of energy drift that can be tolerated is very problem dependent .
as there are multiple factors affecting the energy conservation in a simulation , we can allow the non - bonded interactions to cause a drift of similar magnitude like all other factors . if the buffer is too small , some particle pairs which are not in the pair list can move within the cut - off .
we can determine an upper bound to the drift caused by such events in a constant temperature ensemble , this is derived in the appendix .
the upper bound can be used to set the buffer size for simulations . with pme ,
the pair potential at the cut - off is very small , hence the effect of missing pairs will also be very small . to quantify this effect ,
we show the drift as a function of the verlet buffer size for spc / e water @xcite with a pair list lifetime of 18 fs , see fig .
[ drift ] .
this is a representative system as hydrogens in water are the fastest moving particles in nearly all atomistic simulations . with single precision floating point
coordinates , the settle @xcite and shake @xcite constraint algorithms cause an energy drift of -0.01 and 0.1 @xmath38/ns per atom , respectively .
the 4@xmath114 setup shows a drift of similar magnitude even without any additional buffering .
thus , in practice , no explicit buffer is required in single precision .
one thing to note is that at longer buffer length only repulsive hydrogen pairs contribute to the drift . at zero length ,
attractive oxygen - hydrogen pairs also contribute which leads to a cancellation of errors . at a cut - off distance of 0.9 nm .
the settle algorithm causes negative drift in single precision for large buffers .
[ drift],width=283 ] the effective performance is given by the number of interactions within the cut - off radius that can be calculated per cycle . to compare the traditional and cluster schemes we show the performance of 1@xmath111 , and 4@xmath114 256-bit avx kernels , as well as 8@xmath114 cuda gpu kernels with both rf and ewald electrostatics in table [ effperf ] .
there is one factor that complicates the comparison .
the ratio of the cost of the search and the force calculation affects the optimal list update frequency , which in turn affects the required buffer size . in our implementation ,
the pair list construction for the 1@xmath111 setup takes four times longer than calculating the interactions once , where for a 4@xmath114 setup both take about equal time .
we think there is some room for speed - up in our 1@xmath111 search implementation , which has not been fully optimized .
if we assume we can get it twice as fast , the optimal list update frequency is somewhere between 10 and 15 steps .
the optimal update frequency for 4@xmath114 and 8@xmath114 is around 10 steps . for the following comparison we will use the same update frequency of 10 steps for all setups to simplify the comparison .
the effective speed - up of the force calculation of the 4@xmath114 over the 1@xmath111 scheme on cpus is a factor of 1.8 and 1.4 for rf and ewald electrostatics , respectively .
this speedup is mainly due to higher achieved pair rate , but the smaller buffer also contributes . assuming the 1@xmath111 search cost can be brought down to twice the force calculation cost , the total performance improvement including the search cost is a factor of 2.0 and 1.5 for rf and ewald electrostatics , respectively .
the 8@xmath114 scheme results in a lower algorithmic work - efficiency due to the increase in the ratio of zero interactions calculated .
note that these results are for a cut - off of 1 nm or 210 non - zero pairs per particle . with increasing cut - off radius , the efficiency increases and the performance improvement approaches a factor of 3 .
as we run the pair search on the , slower , cpu , a longer list update interval often provides better total performance .
the gpu kernels use a conditional for skipping pairs beyond the cut - off , unlike the cpu @xmath8 kernels , which use masking .
therefore the pair - rate increases with buffer size . but calculating more pair distances always decreases the effective performance .
still , the cluster algorithm demonstrates the potential of streaming architectures with an effective performance of a factor of 5 and 7 higher than the 4@xmath114 cpu rf and ewald kernels , respectively . [ cols="^,^,^,^,^,^,^,^,^,^,^,^,^ " , ]
for calculating non - bonded interactions in molecular simulations , the standard particle - based pair interaction algorithms commonly simd - parallelized by loop unrolling have reached their limits .
kernels based on these approaches are often limited by the high memory to arithmetic operation ratio , the number of data shuffle operations required , and restrictions in instruction scheduling which reduces the potential for memory latency hiding .
the pair list construction is affected by the same issues .
we have presented a simple and flexible approach to overcome these problems .
a scheme using cluster pairs of @xmath0 versus @xmath1 atoms leads to kernels that efficiently utilize current cpu and gpu simd units .
the memory pressure is reduced by a factor @xmath0 on cpus .
we found that @xmath0=4 usually provides the best performance . on gpus
we use @xmath0=8 and the memory pressure is reduced by another factor of 4 by loading and operating on up to 8 @xmath9-clusters at once .
the algorithm reorganizes the data representation at the lowest , particle level .
therefore , any method in the literature that applies to particles , can be applied to the clusters in our method .
an example used here is the verlet buffer .
while the widely used linked cell list for reducing the search space can be applied , this does not offer any advantage as the locality of the clusters is already available through the grid used to generate the clusters .
the performance advantage of our method over traditional algorithms depends on the computational cost of the interactions , the number of particle pairs within the cut - off and the simd width . while in many cases our cluster - based algorithm significantly outperforms the particle - based algorithms , in some cases it can be less advantageous . for cheap interactions the reduction of shuffling and memory operations will favor the cluster setup , whereas for expensive interactions the extra zero interactions can outweigh the gains . for typical atomistic molecular simulations our method performs very well and is a factor 1.5 to 3 faster on 8-wide simd than traditional methods .
on intel sandy bridge cpus as well as cuda gpus the flop rate is above 60% of the peak .
most importantly ,
our scheme inherently maps well to future cpu and gpu architectures as well as existing ones not discussed here .
as the number of floating point operation per load / store operation can be tuned , a reduction of the arithmetic cycles per kernel , e.g. by introduction of fma instructions , will result in higher performance . additionally , wider simd units , for example 16-way simd in intel xeon phi , can be used efficiently with a limited amount of effort .
this work was supported by the european research council ( grants nr . 258980 and nr . 209825 ) , the swedish e - science research center and the scalalife eu fp7 project .
the authors thank erik lindahl for providing the analytical approximation of the ewald correction force and for his advice on x86 simd optimization , nvidia for advice on cuda optimization and mark abraham for thoroughly reviewing the code and this manuscript .
for a canonical ensemble , an upper bound on the average energy drift due to the finite verlet buffer size can be derived .
this depends on the atomic displacements and the shape of the potential at the cut - off .
the displacement - distribution along one dimension for a freely moving particle with mass @xmath39 over time @xmath40 at temperature @xmath41 is gaussian with zero mean and variance @xmath42 .
the variance of the distance between two non - interacting particles is @xmath43 . in practice , particles interact with each other over time @xmath40 .
these interactions make the displacement distribution narrower , since any interaction will hinder free motion of particles .
ignoring the effect of interactions on the displacements thus provides an upper bound .
we calculate interactions with a non - bonded interaction cut - off distance of @xmath44 and a pair list cut - off of @xmath45 , where @xmath46 is the verlet buffer size .
we can then write the average energy drift over time @xmath40 for pair interactions between a particle of type 1 surrounded by particles of type 2 with number density @xmath47 , when the inter - particle distance changes from @xmath48 to @xmath49 , as : @xmath50 g\!\left(\frac{r_t - r_0}{\sigma}\right ) d r_0 \ , d r_t\\ & \approx & 4 \pi ( r_\ell+\sigma)^2 \rho_2 \int_{-\infty}^{r_c } \int_{r_\ell}^\infty \big [ v'(r_c ) ( r_t - r_c ) + \\ & & \phantom{4 \pi ( r_\ell+\sigma)^2 \rho_2 \int_{-\infty}^{r_c } \int_{r_\ell}^\infty \big [ } v''(r_c)\frac{1}{2}(r_t - r_c)^2 \big ] g\!\left(\frac{r_t - r_0}{\sigma}\right ) d r_0 \ , d r_t\\ & = & 4 \pi ( r_\ell+\sigma)^2 \rho_2 \bigg\ { \frac{1}{2}v'(r_c)\left[r_b \sigma g\!\left(\frac{r_b}{\sigma}\right ) - ( r_b^2+\sigma^2)e\!\left(\frac{r_b}{\sigma}\right ) \right ] + \\ & & \phantom{4 \pi ( r_\ell+\sigma)^2 \rho_2 \bigg\ { } \frac{1}{6}v''(r_c)\left [ \sigma(r_b^2+\sigma^2)g\!\left(\frac{r_b}{\sigma}\right ) - r_b(r_b^2 + 3\sigma^2 ) e\!\left(\frac{r_b}{\sigma}\right ) \right ] \bigg\}.\end{aligned}\ ] ] here , @xmath51 is a gaussian distribution with zero mean , unit variance , and @xmath52 .
we always want to achieve small energy drift , so @xmath53 will be small compared to both @xmath44 and @xmath54 .
thus , the approximations in the above equations are good since the gaussian distribution decays rapidly . to calculate the total energy drift
, the drift needs to be averaged over all particle pairs and weighted with the particle count .
24 natexlab#1#1[1]`#1 ` [ 2]#2 [ 1]#1 [ 1]http://dx.doi.org/#1 [ ] [ 1]pmid:#1 [ ] [ 2]#2 , ( ) .
, , ( ) . , ( ) . , , , , , , , , , , ( ) . , , , ( ) . , , , , ( ) . , , , , ( ) . , , in : . , , , , in : . , , , , , ( ) . , , , , , , , , , ( ) . , , , , ( ) . , , , ( ) . , , , , , , ( ) . , , , , , , ( ) .
, , , , , , , , , , , , ( ) . , ( ) .
, , , ( ) . , , ( ) .
, , , ( ) . , , , , , , , , ( ) . | calculating interactions or correlations between pairs of particles is typically the most time - consuming task in particle simulation or correlation analysis .
straightforward implementations using a double loop over particle pairs have traditionally worked well , especially since compilers usually do a good job of unrolling the inner loop . in order to reach high performance on modern cpu and accelerator architectures , single - instruction multiple - data ( simd ) parallelization
has become essential .
avoiding memory bottlenecks is also increasingly important and requires reducing the ratio of memory to arithmetic operations .
moreover , when pairs only interact within a certain cut - off distance , good simd utilization can only be achieved by reordering input and output data , which quickly becomes a limiting factor . here
we present an algorithm for simd parallelization based on grouping a fixed number of particles , e.g. 2 , 4 , or 8 , into spatial clusters . calculating all interactions between particles in a pair of such clusters
improves data reuse compared to the traditional scheme and results in a more efficient simd parallelization .
adjusting the cluster size allows the algorithm to map to simd units of various widths .
this flexibility not only enables fast and efficient implementation on current cpus and accelerator architectures like gpus or intel mic , but it also makes the algorithm future - proof .
we present the algorithm with an application to molecular dynamics simulations , where we can also make use of the effective buffering the method introduces .
* notice : * this is the author s version of a work that was accepted for publication in computer physics communications .
changes resulting from the publishing process , such as peer review , editing , corrections , structural formatting , and other quality control mechanisms may not be reflected in this document .
changes may have been made to this work since it was submitted for publication . |
the present work on the optimal multiple stopping time problem , following the optimal single stopping time problem , involves proving the existence of the maximal reward , finding necessary or sufficient conditions for the existence of optimal stopping times and providing a method to compute these optimal stopping times .
the results are well known in the case of the optimal single stopping time problem . consider a _ reward _ given by a right - continuous left - limited ( rcll ) positive adapted process @xmath6 on @xmath7 , @xmath8 satisfying the usual conditions , and look for the maximal reward @xmath9 , \tau\in t_0\},\ ] ] where @xmath10 @xmath11 @xmath120 , \infty [ $ ] is the fixed time horizon and @xmath13 is the set of stopping times @xmath14 smaller than @xmath10 .
from now on , the process @xmath15 will be denoted by @xmath16 . in order to compute @xmath17
we introduce for each @xmath18 the _ value function _
@xmath19 , \tau\in t_s\}$ ] , where @xmath20 is the set of stopping times in @xmath13 greater than @xmath0 .
the value function is given by a family of random variables @xmath21 . by using the right continuity of the reward @xmath22
, it can be shown that there exists an adapted process @xmath23 which _ aggregates _ the family of random variables @xmath24 that is such that @xmath25 a.s . for each @xmath18 .
this process is the _ snell envelope _ of @xmath22 , that is , the smallest supermartingale process that dominates @xmath2 . moreover , when the reward @xmath22 is continuous , the stopping time defined trajectorially by @xmath26 is optimal . for details ,
see el karoui ( @xcite ) , karatzas and shreve ( @xcite ) or peskir and shiryaev ( @xcite ) . in the present work ,
we show that computing the value function for the optimal multiple stopping time problem @xmath27 , \tau _
1,\ldots,\tau_d\in t_s\},\ ] ] reduces to computing the value function for an optimal single stopping time problem @xmath28 , \theta\in t_s \},\ ] ] where the _ new reward _ @xmath2 is no longer an rcll process , but a family @xmath29 of positive random variables which satisfies some compatibility properties . for this new optimal single stopping time problem with a reward @xmath30
, we show that the minimal optimal stopping time for the value function @xmath31 is no longer given by a hitting time of processes , but by the essential infimum @xmath32 this method also has the advantage that it no longer requires any aggregation results that need stronger hypotheses and whose proofs are rather technical . by using the reduction property @xmath33 a.s .
, we give a method to construct by induction optimal stopping times @xmath34 for @xmath3 , which are also defined as essential infima , in terms of _ nested _ optimal single stopping time problems .
some examples of optimal multiple stopping time problems have been studied in different mathematical fields . in finance
, this type of problem appears in , for instance , the study of _ swing options _
[ e.g. , carmona and touzi ( @xcite ) , carmona and dayanik ( @xcite ) ] in the case of ordered stopping times . in the nonordered case ,
some optimal multiple stopping time problems appear as useful mathematical tools to establish some large deviations estimations [ see kobylanski and rouy ( @xcite ) ] .
further applications can be imagined in , for example , finance and insurance [ see kobylanski , quenez and rouy - mironescu ( @xcite ) ] . in a work in preparation [
see kobylanski and quenez ( @xcite ) ] , the markovian case will be studied in detail and some applications will be presented .
the paper is organized as follows . in section
[ 1 ] we revisit the optimal single stopping time problem for admissible families .
we prove the existence of optimal stopping times when the family @xmath2 is right- and left - continuous in expectation along stopping times .
we also characterize the minimal optimal stopping times . in section
[ 2 ] we solve the optimal double stopping time problem . under quite weak assumptions ,
we show the existence of a pair of optimal stopping times and give a construction of those optimal stopping times . in section
[ 3 ] we generalize the results obtained in section [ 2 ] to the optimal @xmath35-stopping - times problem .
also , we study the simpler case of a symmetric reward . in this case , the problem clearly reduces to ordered stopping times , and our general characterization of the optimal multiple stopping time problem in terms of nested optimal single stopping time problems straightforwardly reduces to a sequence of optimal single stopping time problems defined by backward induction .
we apply these results to _ swing options _ and , in this particular case , our results correspond to those of carmona and dayanik ( @xcite ) . in the last section ,
we prove some aggregation results and characterize the optimal stopping times in terms of hitting times of processes .
let @xmath36 be a probability space , where @xmath370 , \infty[$ ] is the fixed time horizon and @xmath38 is a filtration satisfying the usual conditions of right continuity and augmentation by the null sets of @xmath39 . we suppose that @xmath40 contains only sets of probability @xmath41 or @xmath42 .
we denote by @xmath43 the collection of stopping times of @xmath44 with values in @xmath45 $ ] . more generally , for any stopping time @xmath0 , we denote by @xmath46 the class of stopping times @xmath47 with @xmath48 a.s .
we use the following notation : for real - valued random variables @xmath49 and @xmath50 , @xmath51 , the notation `` @xmath52 '' means `` the sequence @xmath53 is nondecreasing and converges to @xmath49 a.s . ''
we first recall some classical results on the optimal single stopping time problem . the following classical results , namely the supermartingale property of the value function , the optimality criterium and the right continuity in expectation of the value function are well known [ see el karoui ( @xcite ) or karatzas and shreve ( @xcite ) or peskir and shiryaev ( @xcite ) ] .
they are very important tools in optimal stopping theory and will often be used in this paper in the ( unusual ) case of a reward given by an admissible family of random variables defined as follows .
a family of random variables @xmath29 is said to be _
admissible _ if it satisfies the following conditions : 1 . for all @xmath47
, @xmath54 is an @xmath55-measurable @xmath56-valued random variable ; 2 . for all @xmath57 , @xmath58 a.s .
on @xmath59 .
let @xmath60 be a positive progressive process .
the family defined by @xmath61 is admissible .
note also that the definition of admissible families corresponds to the notion of @xmath13-systems introduced by el karoui ( @xcite ) . for the convenience of the reader , we recall the definition of the essential supremum and its main properties in appendix .
suppose the reward is given by an admissible family @xmath62 .
the _ value function at time @xmath0 _ , where @xmath18 , is given by @xmath63.\vspace*{-2pt}\ ] ] [ p1.adm ] the value function that is the family of random variables @xmath64 defined by ( [ vs ] ) is an admissible family .
property 1 of admissibility for @xmath24 follows from the existence of the essential supremum ( see theorem [ ta ] in appendix ) .
take @xmath65 and let @xmath66 . for each @xmath67
, put @xmath68 .
as @xmath69 , we have a.s . on @xmath70 , @xmath71= e[\phi(\theta_a)| \mathcal{f}_s]=$ ] @xmath72\leq $ ] @xmath73 hence taking the essential supremum over @xmath67 , we have @xmath74 a.s . , and by symmetry of @xmath0 and @xmath75 , we have shown property 2 of admissibility .
[ p1.2a ] there exists a sequence of stopping times @xmath76 with @xmath77 in @xmath78 such that @xmath79\ \uparrow\ v(s)\qquad\mbox{a.s.}\vspace*{-2pt}\ ] ] for each @xmath0 @xmath11 @xmath13 , one can show that the set @xmath80 , \theta\in t_s \ } $ ] is closed under pairwise maximization .
indeed , let @xmath57 and @xmath81\leq e[\phi(\theta)| \mathcal{f}_s]\}$ ] .
one has @xmath82 .
let @xmath83 , a stopping time .
it is easy to check that @xmath84=$ ] @xmath85\vee$ ] @xmath86 $ ] .
the result follows by a classical result ( see theorem [ ta ] in appendix ) . recall that for each fixed @xmath18 , an admissible family @xmath87 is said to be a _ supermartingale system _ ( resp .
, a _ martingale system _ ) if , for any @xmath14 , @xmath88 @xmath11 @xmath13 such that @xmath89 a.s . , @xmath90 \leq h(\theta ' ) \qquad\mbox{a.s.}\qquad\bigl(\mathrm{resp.,}~e[h(\theta ) |\mathcal{f } _ { \theta ' } ] = h(\theta ' ) \mbox { a.s.}\bigr).\ ] ] [ p1.superm ] *
the value function @xmath91 is a supermartingale system . *
furthermore , it is characterized as the associated with @xmath92 , that is , the smallest supermartingale system which is greater ( a.s . )
than @xmath93 .
let us prove the first part .
fix @xmath94 a.s . by proposition [ p1.2a ]
, there exists an optimizing sequence @xmath95 for @xmath3 . by the monotone convergence theorem , @xmath96 = \lim_{n\to\infty } e[\phi(\theta^{n } ) |\mathcal{f } _ { s ' } ] } $ ] a.s .
now , for each @xmath97 , since @xmath98 a.s . , we have @xmath99\leq v(s')$ ] a.s . hence , @xmath100
\leq v(s')$ ] a.s . , which gives the supermartingale property of the value function .
let us prove the second part .
let @xmath101 be a supermartingale system such that for each @xmath47 , @xmath102 @xmath103 @xmath54 a.s .
fix @xmath18 . by the properties of @xmath104 , for all @xmath67 , @xmath105\geq$ ] latexmath:[$e[\phi(\theta ) @xmath67
, we have @xmath107 a.s .
now , recall the following bellman optimality criterium [ see , e.g. , el karoui ( @xcite ) ] .
[ criterium ] fix @xmath0 @xmath11 @xmath13 and let @xmath108 be such that @xmath109<\infty$ ] .
the three following assertions are equivalent : 1 .
@xmath110 is @xmath0-optimal for @xmath3 , that is , @xmath111 \qquad\mbox{a.s.};\ ] ] 2 .
@xmath112= e [ v ( \theta^*)];$ ] 3 . @xmath113= e [ \phi ( \theta^*)].$ ] note that since the value function is a supermartingale system , equality @xmath113= e [ v ( \theta^*)]$ ] is equivalent to the fact that the family @xmath114 is a martingale system .
proof of proposition [ criterium ] let us show that assertion 1 implies assertion 2 .
suppose assertion 1 is satisfied . since the value function @xmath115 is a supermartingale system greater that @xmath2 , we clearly have @xmath116 \geq e[\phi(\theta^ * ) |\mathcal{f } _ { s } ] \qquad \mbox{a.s . } \ ] ] since equality ( [ so ] ) holds , this implies that the previous inequalities are actually equalities . in particular , @xmath117 = e[\phi ( \theta^ * ) |\mathcal{f } _ { s } ] $ ] a.s .
, but as inequality @xmath118 holds a.s . , and as @xmath119<\infty$ ]
, we have @xmath120 a.s . moreover , @xmath121 $ ] a.s . , which gives @xmath113= e [ v ( \theta^*)]$ ] .
hence , assertion 2 is satisfied
. clearly , assertion 2 implies assertion 3 .
it remains to show that 3 implies 1 .
suppose that 3 is satisfied .
since @xmath122 it is clear that @xmath123= \sup_{\theta\in t_s } e [ \phi ( \theta)].\ ] ] by assertion 3 of proposition [ criterium ] , a stopping time @xmath124 such that @xmath109\!<\infty$ ] is @xmath0-optimal for @xmath3 if and only if it is optimal for the optimal stopping time problem ( [ nstp ] ) , that is , @xmath125= e[\phi(\theta^*)].\ ] ] we now give a regularity result on @xmath115 [ see lemma 2.13 in el karoui ( @xcite ) ] .
let us first introduce the following definition .
an admissible family @xmath62 is said to be _
right- ( resp . ,
left- ) continuous along stopping times in expectation [ rce ( resp . ,
lce ) ] _ if for any @xmath47 and any sequence @xmath126 of stopping times such that @xmath127 a.s .
( resp . , @xmath128 a.s . ) , one has @xmath129=\lim_{n\to\infty }
e[\phi(\theta_n ) ] . }
$ ] if @xmath22 is a continuous adapted process such that@xmath130}\phi_t ] < \infty}$ ] , then the family defined by @xmath61 is clearly rce and lce . also , if @xmath60 is an rcll adapted process such that its jumps are totally inaccessible , then the family defined by @xmath61 is clearly rce and even lce .
[ l1 ] let @xmath62 be an admissible family which is rce .
the family @xmath131 is then rce .
since @xmath132 is a supermartingale system , the function @xmath133 $ ] is a nonincreasing function of stopping times .
suppose it is not rce at @xmath18 .
if @xmath134 < \infty$ ] , then there exists a constant @xmath135 and a sequence of stopping times @xmath136 such that @xmath137 a.s . and
such that @xmath138+\alpha\leq e[v(s)].\ ] ] now , recall that @xmath139= \sup_{\theta\in t_s}e[\phi ( \theta ) ] } $ ] [ see ( [ nstp ] ) ] .
hence , there exists @xmath140 such that @xmath141+\frac { \alpha}{2 } \leq e[\phi(\theta')].\ ] ] hence , for all @xmath142 , @xmath143+\frac{\alpha}{2}\leq e[\phi ( \theta')]$ ] . as @xmath144 a.s . ,
we obtain , by taking the limit when @xmath145 and using the rce property of @xmath2 , that @xmath146+\frac{\alpha}{2 } \leq e[\phi(\theta')],\ ] ] which gives the expected contradiction in the case @xmath134
< \infty $ ] . otherwise , instead of ( [ vsna ] )
, we have @xmath147 \leq c}$ ] for some constant @xmath148 , and similar arguments as in the finite case lead to a contradiction as well . we will now give a new result which generalizes the classical existence result of an optimal stopping time stated in the case of a reward process to the case of a reward family of random variables .
[ t.1 ] let @xmath149 be an admissible family that satisfies the integrability condition @xmath150<\infty\ ] ] and which is rce and lce along stopping times .
then , for each @xmath0 @xmath11 @xmath13 , there exists an optimal stopping time for @xmath3 .
moreover , the random variable defined by @xmath151 is the minimal optimal stopping time for @xmath3 .
let us emphasize that in this theorem , the optimal stopping time @xmath152 is not defined trajectorially , but as an essential infimum of random variables . in the classical case , that is , when the reward is given by an adapted rcll process , recall that the minimal optimal stopping time is given by the random variable @xmath153 defined trajectorially by @xmath154 the definition of @xmath152 as an essential infimum allows the assumption on the regularity of the reward to be relaxed .
more precisely , whereas in the previous works ( mentioned in the ) , the reward was given by an rcll and lce process , in our setting , the reward is given by an rce and lce family of random variables .
the idea of the proof is classical : we use an approximation method introduced by maingueneau ( @xcite ) , but our setting allows us to simplify and shorten the proof .
proof of theorem [ t.1 ] the proof will be divided into two parts . _
part i _ : in this part , we will prove the existence of an optimal stopping time .
fix @xmath0 @xmath11 @xmath13 .
we begin by constructing a family of stopping times [ see maingueneau ( @xcite ) or el karoui ( @xcite ) ] . for @xmath155 @xmath11 @xmath1560,1[$ ] , define the @xmath157-measurable random variable @xmath158 by @xmath159 the following lemma holds .
[ eli ] the stopping time @xmath160 is a @xmath161-optimal stopping time for @xmath162= \sup_{\theta\in t_s } e [ \phi ( \theta)],\ ] ] that is , @xmath163 \leq e [ \phi(\theta^{\lambda}(s ) ) ] .\ ] ] suppose now that we have proven lemma [ eli ] .
since @xmath164 is nondecreasing , for @xmath165 , the stopping time @xmath166 is well defined .
let us show that @xmath167 is optimal for @xmath3 . by letting @xmath168 in inequality ( [ la ] ) , and since @xmath2 is lce , we easily derive that @xmath169 $ ] @xmath170 @xmath171 $ ] .
consequently , by the optimality criterium 3 of proposition [ criterium ] , @xmath167 is @xmath0-optimal for @xmath3 .
this completes part i. _ part ii _ :
let us now prove that @xmath172 a.s . , where @xmath152 is defined by ( [ e.te ] ) , and that it is the minimal optimal stopping time for @xmath3 . for each @xmath0 @xmath11 @xmath13 ,
the set @xmath173 is not empty ( since @xmath10 belongs to @xmath174 ) and is closed under pairwise minimization .
hence , there exists a sequence @xmath175 of stopping times in @xmath174 such that @xmath176 consequently , @xmath177 is a stopping time .
let @xmath14 be an optimal stopping time for @xmath3 . by the optimality criterium ( proposition [ criterium ] )
, and since , by assumption , @xmath178 < \infty$ ] , we have a.s . and
hence @xmath179 now , for each @xmath180 the stopping time @xmath181 defined by ( [ tls ] ) clearly satisfies @xmath182 a.s . passing to the limit when @xmath183 we obtain @xmath184 .
as @xmath185 is optimal for @xmath3 , this implies that @xmath186 optimal for @xmath187 a.s .
hence , @xmath188 which gives the desired result .
the proof of theorem [ t.1 ] is thus complete .
it now remains to prove lemma [ eli ] .
proof of lemma [ eli ] we have to prove inequality ( [ la ] ) .
this will be done by means of the following steps .
_ step 1 _ : fix @xmath1890,1[$ ] .
it is easy to check that the set @xmath190 is nonempty ( since @xmath10 @xmath11 @xmath191 ) and closed by pairwise minimization . by theorem [ ta ] in the
, there exists a sequence @xmath95 in @xmath192 such that @xmath193 a.s .
therefore , @xmath194 is a stopping time and @xmath195 a.s .
moreover , we have @xmath196 a.s . for all @xmath97 .
taking expectation and using the rce properties of @xmath115 and @xmath2 , we obtain @xmath197 ) \leq e[\phi(\theta^{\lambda } ( s ) ) ] .\ ] ] _ step 2 _ : let us show that for each @xmath1890,1[$ ] and each @xmath18 , @xmath198\qquad\mbox{a.s.}\ ] ] for each @xmath0 @xmath11 @xmath13 , let us define the random variable @xmath199.$ ] step 2 amounts to showing that @xmath200 a.s . since @xmath201 is a supermartingale system and since @xmath202 a.s .
, we have that @xmath203 \leq v(s ) \qquad\mbox{a.s.}\ ] ] it remains to show the reverse inequality .
_ step 2a _ : let us show that the family @xmath204 is a supermartingale system .
let @xmath205 be such that @xmath206 a.s . as @xmath207 a.s .
, we have @xmath208 = e [ v(\theta^{\lambda } ( s ' ) ) = e\bigl [ e \bigl [ v(\theta^{\lambda } ( s ' ) ) | \mathcal{f}_{\theta ^{\lambda } ( s ) } \bigr ] |\mathcal{f}_s \bigr]\qquad\mbox{a.s.}\ ] ] now , since @xmath209 is a supermartingale system , @xmath210 $ ] @xmath211 consequently , @xmath208 \leq e [ v(\theta^{\lambda } ( s ) ) _ step 2b _ : let us show that for each @xmath212 and each @xmath1890,1[$ ] , @xmath213 fix @xmath212 and @xmath1890,1[$ ] . on @xmath214
, we have @xmath215 a.s .
hence , on @xmath214 , @xmath216 = e [ v(s ) | \mathcal{f}_s ] = v(s ) $ ] and therefore @xmath217 furthermore , on @xmath218 , as @xmath219 is nonnegative , we have @xmath220 and the proof of step 2b is complete .
now , the family @xmath221 is a supermartingale system by convex combination of two supermartingale systems .
hence , as the value function @xmath222 is characterized as the smallest supermartingale system which dominates @xmath223 , we derive that for each @xmath224 , @xmath225 now , by the integrability assumption made on @xmath2 , we have @xmath226 a.s .
hence , we have @xmath227 a.s . consequently , for each @xmath224 , @xmath228 a.s . , which completes step 2 .
finally , step 1 [ inequality ( [ e.lvphi ] ) ] and step 2 [ equality ( [ e.u ] ) ] give @xmath229 = \lambda e [ v(\theta^{\lambda}(s ) ) ] \leq e [ \phi(\theta^{\lambda}(s ) ) ] .\ ] ] in other words , @xmath160 is a @xmath161-optimal stopping time for ( [ oie ] ) , which completes the proof of lemma [ eli ] .
recall that in the previous works [ see , e.g. , karatzas and shreve ( @xcite ) , proposition d.10 and theorem d.12 ] , the proof of the existence of optimal stopping times requires the value function to be aggregated and thus the use of some fine aggregation results such as proposition [ p.sma ] . in our work , since we only work with families of random variables , we do not need any aggregation techniques , which simplifies and shortens the proof . under some regularity assumptions on the reward , we can show that the value function family is left - continuous along stopping times in expectation .
more precisely , we have the following .
[ p.vlce ] @xmath230suppose that the admissible family @xmath231 is lce and rce , and satisfies the integrability condition .
the value function @xmath24 defined by ( [ vs ] ) is then lce .
let @xmath18 and let @xmath232 be a sequence of stopping times such that @xmath233 a.s .
note that by the supermartingale property of @xmath234 we have @xmath235\geq e[v(s)].\ ] ] now , by theorem [ t.1 ] , the stopping time @xmath236 defined by ( [ e.te ] ) is optimal for @xmath237 .
moreover , it is clear that @xmath238 is a nondecreasing sequence of stopping times dominated by @xmath152 .
let us define @xmath239 .
note that @xmath240 is a stopping time . also , as for each @xmath97 , @xmath241 a.s .
, it follows that @xmath242 a.s .
therefore , since @xmath2 is lce , @xmath243\geq e[\phi(\overline\theta)]= \lim_{n\to\infty } e[\phi ( \theta^*(s_n))]=\lim_{n\to\infty } e[v(s_n)].\ ] ] this , together with ( [ e.vsn ] ) , gives @xmath244=\lim_{n\to\infty } e[v(s_n)]}$ ] .
[ unun ] @xmath245 in this proof , we have also proven that @xmath246 is optimal for @xmath3 .
hence , by the optimality criterium , @xmath247 a.s . , which implies that @xmath248 a.s .
moreover , since for each @xmath97 , @xmath249 a.s . , by letting @xmath97 tend to @xmath250 , we clearly have that @xmath251 a.s .
hence , @xmath252
@xmath253 @xmath254 a.s .
thus , we have also shown that the map @xmath255 is left - continuous along stopping times .
we now consider the optimal double stopping time problem .
we introduce the following definitions .
the family @xmath256 is _ biadmissible _ if it satisfies : 1 . for all @xmath257
@xmath258 is an @xmath259-measurable @xmath56-valued r.v . ;
2 . for all @xmath260 a.s .
on @xmath261 .
[ biprocess ] let @xmath262 be a biprocess , that is , a function @xmath263 ^2\times\omega\rightarrow\mathbb{r}^+ ; ( t , s,\omega ) \mapsto\psi_{t , s } ( \omega)\ ] ] such that for almost all @xmath264 , the map @xmath265 is _ right - continuous _ ( i.e. , @xmath266 @xmath267 @xmath268 ) , and for each @xmath269 ^2 $ ] , @xmath266 is @xmath270-measurable . in this case , the family @xmath256 defined by @xmath271 is clearly biadmissible . for a biadmissible family @xmath272
let us consider the value function associated with the reward family @xmath256 : @xmath273.\ ] ] as in the case of the single stopping time problem , we have the following properties .
[ p2.1 ] let @xmath256 be a biadmissible family of random variables .
the following properties then hold : 1 .
the family @xmath24 is an admissible family of random variables ; 2 . for each @xmath18
, there exists a sequence of pairs of stopping times @xmath274 in @xmath275 such that @xmath276 \}_{n \in\mathbb{n}}$ ] is nondecreasing and a.s .
@xmath277 \uparrow v(s);\ ] ] 3 .
the family of random variables @xmath278 is a supermartingale system , that is , it satisfies the dynamic programming principle .
\(1 ) as in the case of single stopping time , property 1 of admissibility for @xmath24 follows from the existence of the essential supremum .
take @xmath65 and put @xmath66 , and for each @xmath279 put @xmath280 and @xmath281 . as @xmath282 , one has , a.s . on @xmath70 , @xmath283= e[\psi(\tau_1^a,\tau _
2^a)| \mathcal{f}_s]=$ ] @xmath284\leq$ ] @xmath285 hence , taking the essential supremum over @xmath286 , we have @xmath74 a.s . , and , by symmetry , we have shown property 2 of admissibility .
hence , the family @xmath24 is an admissible family of random variables .
the proofs of ( 2 ) and ( 3 ) can be easily adapted from the proofs of proposition [ p1.2a ] and proposition [ p1.superm ] .
following the case of single stopping time , we now give some regularity results on the value function .
[ rce2 ] a biadmissible family @xmath256 is said to be _ right - continuous along stopping times in expectation ( rce ) _ if , for any @xmath14 , @xmath18 and any sequences
@xmath287 @xmath11 @xmath13 and @xmath136 @xmath11 @xmath13 such that @xmath288 and @xmath289 a.s .
, one has @xmath290=\lim_{n\to\infty } e[\psi(\theta_n , s_n ) ] .
$ ] [ l2.vrce ] suppose that the biadmissible family @xmath291 is rce .
the family @xmath278 defined by ( [ vs ] ) is then rce .
the proof follows the proof of proposition [ l1 ] . in this section
, we will show that the optimal double stopping time problem ( [ vs ] ) can be reduced to an optimal single stopping time problem associated with a _ new reward family_. more precisely , for each stopping time @xmath67 let us introduce the two @xmath292-measurable random variables @xmath293,\qquad u_2({\theta } ) = \operatorname{ess}\sup_{\tau_2\in t_{\theta } } e[\psi(\theta,\tau_2 ) @xmath294 is biadmissible , for each fixed @xmath47 , the families @xmath295 and @xmath296 are admissible .
hence , by proposition [ p1.adm ] the families @xmath297 and @xmath298 are admissible . put @xmath299.\ ] ] the family @xmath300 , which is called the _ new reward family _ , is also clearly admissible . consider the value function associated with the @xmath301 \qquad\mbox{a.s.}\ ] ] [ t3 ] suppose that @xmath302 is a biadmissible family .
for each stopping time @xmath0 , consider @xmath3 defined by ( [ vs ] ) and @xmath31 defined by ( [ u12 ] ) , ( [ phi ] ) , ( [ us ] ) .
then , @xmath303 let @xmath0 be a stopping time . _ step 1 _ : first , let us show that @xmath304 let @xmath305 . put @xmath306 .
as @xmath70 is in @xmath307 , we have @xmath308= e [ { \mathbf{1}}_ae[\psi(\tau_1,\tau_2 ) |\mathcal{f}_{\tau_1 } ] |\mathcal { f}_s ] + e[{\mathbf{1}}_{a^ c}e[\psi(\tau_1,\tau_2 ) |\mathcal{f}_{\tau_2 } ] |\mathcal{f}_s].\ ] ] by noticing that on @xmath70 we have @xmath309 a.s . and , similarly , on @xmath310 we have @xmath311 a.s . , we get @xmath308 \leq e [ \phi({\tau _ 1\wedge\tau _ 2 } ) @xmath312 and @xmath313 in @xmath314 we complete step 1 .
_ step 2 _ : let us now show that @xmath315 a.s .
we clearly have @xmath316 = u_2 ( s)$ ] a.s . by similar arguments , @xmath317 a.s . and ,
consequently , @xmath318 = \phi(s)\qquad \mbox{a.s.}\ ] ] thus , @xmath319 is a supermartingale system which is greater than @xmath93 .
now , by proposition [ p1.superm ] , @xmath320 is the smallest supermartingale system which is greater than @xmath93 .
consequently , step 2 follows , which completes the proof .
note that the reduction to an optimal single stopping time problem associated with a new reward will be the key property used to construct optimal multiple stopping times and to establish an existence result for them ( see sections [ 2.3][2.5 ] ) .
@xmath245 in this section , we are given a biadmissible family @xmath321 such that .
[ p.cnopt ] let @xmath0 be a stopping time and consider the value function @xmath3 defined by ( [ vs ] ) for all @xmath67 , @xmath322 defined by ( [ u12 ] ) , @xmath54 defined by ( [ phi ] ) and @xmath31 defined by ( [ us ] ) .
suppose that the pair @xmath323 is optimal for @xmath3 and put @xmath324 then : 1 .
@xmath325 is optimal for @xmath31 ; 2 .
@xmath326 is optimal for @xmath327 a.s . on @xmath70 ;
@xmath328 is optimal for @xmath329 a.s . on @xmath330 .
moreover @xmath331 .
let @xmath0 @xmath11 @xmath13 and suppose that the pair of stopping times @xmath332 is optimal for @xmath3 .
as @xmath333 a.s . , we obtain equality in step 1 of the proof of theorem [ t3 ] . more precisely , @xmath334 = e[\phi(\tau^*_1\wedge\tau^*_2 ) |\mathcal{f}_s]=u(s)\qquad\mbox{a.s . } , \\ e[\psi(\tau^*_1,\tau^*_2)|\mathcal{f}_{\tau^*_1 } ] & = & u_2(\tau^*_1)=u_2(\tau^*_1\wedge\tau^*_2)=\phi(\tau^*_1\wedge\tau^*_2)\qquad\mbox{a.s . on }
e[\psi(\tau^*_1,\tau^*_2)|\mathcal{f}_{\tau^*_2 } ] & = & u_1(\tau^*_2)=u_1(\tau^*_1\wedge\tau^*_2)=\phi(\tau^*_1\wedge\tau^*_2)\qquad\mbox{a.s.~on } a^c,\end{aligned}\ ] ] which easily leads to ( 1 ) , ( 2 ) , ( 3 ) and @xmath335 . [ r2.1 ]
note that , in general , for a pair @xmath336 of optimal stopping times for @xmath337 the inclusion @xmath338 is strict . indeed if @xmath339 , then @xmath340 , and all pairs of stopping times are optimal .
consider @xmath341 , @xmath342 . in this case ,
@xmath343 and @xmath344 .
we now give a sufficient condition for optimality .
[ pconstruction ] using the notation of proposition [ p.cnopt ] , suppose that : 1 .
@xmath345 is optimal for @xmath31 ; 2 .
@xmath346 is optimal for @xmath347 ; 3 .
@xmath348 is optimal for @xmath349 and put @xmath350 .
the pair of stopping times @xmath351 defined by @xmath352 is then optimal for @xmath3 .
moreover , @xmath353 and @xmath354 let @xmath355 be an optimal stopping time for @xmath31 , that is , @xmath356 $ ] a.s .
let @xmath348 be an optimal stopping time for @xmath349 ( i.e. , @xmath357 $ ] a.s . ) and let @xmath346 be an optimal stopping time for @xmath347 ( i.e. , @xmath358 @xmath359 $ ] a.s . ) .
we introduce the set @xmath350 .
note that @xmath360 is in @xmath361 .
let @xmath362 be the stopping times defined by ( [ tau12 ] ) .
we clearly have the inclusion @xmath363 since @xmath364 $ ] and @xmath365 $ ] , we have @xmath366.\ ] ] the optimality of @xmath348 and @xmath346 gives that a.s .
@xmath367 \\ & = & e[{\mathbf{1}}_b \psi(\tau_1^*,\tau_2^ * ) + { \mathbf{1}}_{b^c}\psi(\tau_1^*,\tau_2^ * ) |\mathcal{f}_s]=e[\psi(\tau_1^*,\tau_2^*)|\mathcal{f}_s].\end{aligned}\ ] ] as @xmath368 a.s .
, the pair of stopping times @xmath351 is @xmath0-optimal for @xmath3 . by proposition [ p.cnopt ]
, we have @xmath369 .
hence , by ( [ incl ] ) , @xmath370 .
proposition [ pconstruction ] still holds true if condition 2 holds true on the set @xmath360 and condition 3 holds true on the set @xmath371 .
note that by remark [ r2.1 ] , we do not have a characterization of optimal pairs of stopping times .
however , it is possible to give a characterization of _ minimal optimal _ stopping times in a particular sense ( see appendix ) . before studying the problem of the existence of optimal stopping times
, we have to state some regularity properties of the new reward family @xmath62 .
let us introduce the following definition . a biadmissible family @xmath256 is said to be _
uniformly right- ( resp . ,
left- ) continuous in expectation along stopping times [ urce ( resp . ,
ulce ) ] _ if @xmath372<\infty$ ] and if , for each @xmath14 , @xmath18 and each sequence of stopping times @xmath373 such that @xmath374 a.s .
( resp . , @xmath233 a.s . ) , @xmath375 - e [ \psi ( \theta , s_n ) ] \vert & = & 0\quad \mbox{and } \\
\lim_{n\to\infty}\sup_{\theta\in t_0 } \vert e [ \psi ( s , \theta ) ] - e[\psi ( s_n,\theta ) ] \vert & = & 0.\end{aligned}\ ] ] the following right continuity property holds true for the new reward family .
[ t2.nrcadge ] suppose that the biadmissible family @xmath376 is urce ( resp .
, both urce and ulce ) .
the family @xmath377 defined by ( [ phi ] ) is then rce ( resp .
, both rce and lce ) . as @xmath378 $ ] , it is sufficient to show the rce ( resp . , both rce and lce ) properties for the family @xmath379 .
let us introduce the following value function for each @xmath380 @xmath11 @xmath13 : latexmath:[\[\label{u1 } u_1 ( \theta , s ) = \operatorname{ess}\sup_{\tau_1 \in t_{\theta } } e[\psi(\tau_1 , s ) @xmath47 , @xmath382 it is sufficient to prove that @xmath383 is rce ( resp . , both rce and lce ) , that is , if @xmath384 and @xmath385 , @xmath386 in @xmath13 are such that @xmath387 and @xmath374 a.s .
( resp . , @xmath128 and @xmath233 a.s . ) , then @xmath388 = e[u_1 ( \theta , s)]$ ] .
now , we have @xmath389 - e[u_1 ( \theta_n , s ) ] s_n ) ] |}_{\mbox{(ii)}}.\end{aligned}\ ] ] let us show that ( i ) tends to @xmath41 as @xmath145 .
for each @xmath18 , @xmath390 is an admissible family of positive random variables which is rce ( resp .
, both rce and lce ) . by proposition [ l1 ] ( resp . , proposition [ p.vlce ] ) , the value function @xmath391 is rce ( resp .
, both rce and lce ) .
it follows that ( i ) converges to @xmath41 as @xmath97 tends to @xmath250 .
let us show that ( ii ) tends to @xmath41 as @xmath145 . by definition of the value function @xmath392 ( [ u1 ] )
, it follows that @xmath393 - e[u_1 ( \theta_n , s_n ) ] \vert \leq \sup_{\tau \in t_0 } \vert e [ \psi ( \tau , s ) ] - e[\psi ( \tau , s_n ) ] \vert.\ ] ] which converges to @xmath41 since @xmath394 is urce ( resp .
, both urce and ulce ) .
the proof of theorem [ t2.nrcadge ] is thus complete .
suppose that @xmath395 < \infty$ ] .
under the same hypothesis as theorem [ t2.nrcadge ] , the family @xmath396 defined by ( [ vs ] ) is rce ( resp .
, both rce and lce ) .
this follows from the fact that @xmath397 a.s .
( theorem [ t3 ] ) , where @xmath398 is the value function family associated with the new reward @xmath399 . applying propositions [ l1 ] and [ p.vlce ] , we obtain the required properties
. we will now turn to the problem of the existence of optimal stopping times .
let @xmath302 be a biadmissible family which is urce and ulce .
suppose that @xmath400 . by theorem [ t2.nrcadge
] , the admissible family of positive random variables @xmath401 defined by ( [ phi ] ) is rce and lce . by theorem [ t.1 ] ,
the stopping time @xmath402 is optimal for @xmath403 $ ] , that is , @xmath404
= e[\phi(\theta^ { * } ) |\mathcal{f}_{s}]\qquad\mbox{a.s.}\ ] ] moreover , the families @xmath405 and @xmath406 are admissible and are rce and lce . consider the following optimal stopping time problems defined for each @xmath407 : @xmath408 \quad\mbox{and}\quad v_2(s)=\operatorname{ess}\sup_{\theta\in t_s}e[\psi(\theta^*,\theta)| \mathcal{f}_{s}].\ ] ] by theorem [ t.1 ] the stopping times @xmath348 and @xmath346 defined by @xmath409 and @xmath410 are optimal stopping times for @xmath411 and @xmath412 , respectively . note that @xmath413 and @xmath414 a.s .
let @xmath328 and @xmath326 be the stopping times defined by @xmath415 where @xmath416 . by proposition [ pconstruction ] ,
the pair @xmath417 is optimal for @xmath3 .
consequently , we have proven the following theorem .
[ topt ] let @xmath256 be a biadmissible family which is urce and ulce .
suppose that @xmath400 .
the pair of stopping times @xmath417 defined by ( [ mio ] ) is then optimal for @xmath3 defined by ( [ vs ] ) .
note that since @xmath110 , @xmath418 , @xmath419 are minimal optimal , by results in appendix , @xmath417 is minimal optimal for @xmath3 ( in the sense defined in appendix ) .
let @xmath420 , @xmath421 . in this section , we show that computing the value function for the optimal @xmath35-stopping time problem @xmath27 , \tau
_ 1,\ldots,\tau_d\in t_s\}\ ] ] reduces to computing the value function for an optimal single stopping time problem , that is , @xmath422 , \theta\in t_s \ } \qquad\mbox{a.s.,}\ ] ] for a _ new reward _ @xmath2 .
this new reward is expressed in terms of optimal @xmath423-stopping time problems .
hence , by induction , the initial optimal @xmath35-stopping time problem can be reduced to _ nested _ optimal single stopping time problems .
we say that the family of random variables @xmath424 is a @xmath35-_admissible family _ if it satisfies the following conditions : 1 . for all @xmath425
@xmath426 is an @xmath427 measurable @xmath56-valued random variable ; 2 . for all @xmath428 a.s .
on @xmath429 . for each stopping time @xmath224 , we consider the value function associated with the reward @xmath430 : @xmath431.\ ] ] as in the optimal double stopping time problem , the value function satisfies the following properties .
[ p3.1 ] let @xmath432 be a @xmath35-admissible family of random variables .
the following properties then hold : 1 .
@xmath396 is an admissible family of random variables ; 2 . for each @xmath18
, there exists a sequence of stopping times @xmath433 in @xmath434 such that the sequence @xmath435\ } _ { n\in \mathbb{n}}$ ] is nondecreasing and such that @xmath436 $ ] a.s . ; 3 .
the family of random variables @xmath437 @xmath438 defined by ( [ vsd ] ) is a supermartingale system .
the proof is an easy generalization of the optimal double stopping time problem ( proposition [ p2.1 ] ) .
following the case with single or double stopping time , we now state the following result on the regularity of the value function .
[ ld.vrce ] suppose that the @xmath35-admissible family @xmath430 is rce and that @xmath439 .
the family @xmath440 defined by ( [ vsd ] ) is then rce .
the definition of rce and the proof of this property are easily derived from the single or double stopping time case ( see definition [ rce2 ] and proposition [ l2.vrce ] ) .
the optimal @xmath35-stopping time problem ( [ vsd ] ) can be expressed in terms of an optimal single stopping time problem as follows . for @xmath441 and @xmath47 ,
consider the random variable @xmath442\\[-8pt ] & & \quad\quad\hphantom{u^{(i)}(\theta ) = \operatorname { ess } \sup_{\tau_1,\ldots,\tau_{i-1},\tau_{i+1},\ldots , \tau_{d } \in t_{\theta}^{d-1 } } e[\psi ( } \theta,\tau_{i+1},\ldots,\tau _ { d } ) is adapted to the @xmath35-dimensional case . in the two - dimensional case ( @xmath443 )
, we have @xmath444= u_2({\theta})\qquad \mbox{a.s.}\ ] ] and @xmath445 = u_1({\theta } ) \qquad\mbox{a.s.},\ ] ] by definition of @xmath446 and @xmath447 [ see ( [ u12 ] ) ] .
thus , the notation in the two - dimensional case was different , but more adapted to that simpler case . for each @xmath47 ,
define the @xmath292-measurable random variable called the _ new reward _ , @xmath448,\ ] ] and for each stopping time @xmath0 , define the @xmath157-measurable variable @xmath449 .\ ] ] [ t3.red ] let @xmath450 be a @xmath35-admissible family of random variables and for each stopping time @xmath0 , consider @xmath3 defined by ( [ vsd ] ) and @xmath31 defined by ( [ uiun ] ) , ( [ d - phi ] ) and ( [ d - us ] ) . then , @xmath451 _ step 1 _ : let us prove that for all @xmath18 , @xmath452 a.s .
let @xmath0 be a stopping time and @xmath453 .
there exists @xmath454 with @xmath455 , where @xmath456 for @xmath457 , @xmath458 a.s .
on @xmath459 and @xmath459 are in @xmath460 for @xmath461 ( for @xmath443 , one can take @xmath462 and @xmath463 ) .
we have @xmath464=\sum_{i=1}^de [ { \mathbf{1}}_{a_i}e[\psi(\tau ) |\mathcal{f}_{\tau_i } ] |\mathcal{f}_s].\ ] ] by noticing that on @xmath459 one has a.s .
@xmath465 \leq u^{(i)}({\tau_i})\leq\phi(\tau_i)= \phi(\tau_1\wedge\cdots\wedge \tau_d)$ ] , we get @xmath466 \leq e [ \phi({\tau_1\wedge\cdots\wedge \tau_d } ) |\mathcal{f}_s]\leq u(s ) \mbox { a.s.}}$ ] by taking the supremum over @xmath467 we complete step 1 .
_ step 2 _ : let us show that for all @xmath18 , @xmath468 a.s .
this follows from the fact that @xmath396 is a supermartingale system greater than @xmath399 and that @xmath469 is the smallest supermartingale system of this class .
note that the new reward is expressed in terms of optimal @xmath423-stopping time problems .
hence , by induction , the initial optimal @xmath35-stopping time problem can be reduced to nested optimal single stopping time problems . in the case of a symmetric reward
, the problem reduces to ordered stopping times and the nested optimal single stopping time problems simply reduce to a sequence of optimal single stopping time problems defined by backward induction ( see section [ swings ] and the application to _ swing options _ ) .
let @xmath450 be a @xmath35-admissible family .
let us introduce the following notation : for @xmath441 , @xmath47 and @xmath470 in @xmath13 , consider the random variable @xmath471 using this notation , note that for each @xmath472 the value function @xmath473 defined at ( [ uiun ] ) can be written @xmath474.\vspace*{-2pt}\ ] ] [ d - construction ] @xmath475suppose that : 1 .
there exists an optimal stopping time @xmath345 for @xmath31 ; 2 . for @xmath441
, there exist @xmath476 in @xmath477 such that @xmath478 with @xmath479 be such that @xmath480 for @xmath457 , @xmath481 @xmath482 a.s . on @xmath483 and @xmath483
is @xmath361-measurable for @xmath441 .
put @xmath484 then , @xmath485 is optimal for @xmath3 , and @xmath486 .
it is clear that @xmath486 , and a.s . @xmath487=\sum
_ { i=1}^de\bigl[{\mathbf{1}}_{b_i } u^{(i)}({\theta^*})|\mathcal{f}_s\bigr ] \\
& = & \sum _ { i=1}^de\bigl[{\mathbf{1}}_{b_i } e\bigl [ \psi^{(i)}\bigr(\theta^{(i)*},\theta^*\bigr)|\mathcal{f}_{\theta^*}\bigr]|\mathcal{f}_s\bigr ] \\ & = & \sum _ { i=1}^de\bigl[{\mathbf{1}}_{b_i}e\bigl[\psi\bigl(\theta^{(i)*}_1,\ldots,\theta^{(i)*}_{i-1},\theta^*,\theta^{(i)*}_{i+1},\ldots,\theta^{(i)*}_d\bigr ) \\ & = & e[\psi(\tau^*_1,\ldots,\tau^*_{i-1},\tau^*_{i},\tau ^*_{i+1},\ldots,\tau^*_d ) as in the bidimensional case , one can easily derive a necessary condition for obtaining optimal stopping times . moreover ,
for an adapted partial order relation on @xmath488 , one can also derive a characterization of minimal optimal @xmath35-stopping times .
this result is given in appendix . before studying the existence of an optimal @xmath35-stopping time for @xmath3
, we will study the regularity properties of the new reward @xmath62 defined by ( [ d - phi ] ) .
let us introduce the following definition of uniform continuity . a @xmath35-admissible family @xmath489 is said to be _
uniformly right- ( resp . ,
left- ) continuous along stopping times in expectation [ urce ( resp . ,
ulce ) ] _ if @xmath490 , and for each @xmath441 , @xmath18 and sequence of stopping times @xmath491 such that @xmath374 a.s .
( resp . , @xmath233 a.s . )
, we have @xmath492 - e\bigl[\psi^{(i ) } ( \theta , s)\bigr]\bigr| = 0 \qquad \mbox{a.s.}\vspace*{-1pt}\ ] ] [ pd.ngrce ] let @xmath489 be a @xmath35-admissible family which is urce ( resp .
, both urce and ulce ) .
the family of positive random variables @xmath493 defined by ( [ d - phi ] ) is then rce ( resp .
, both rce and lce ) .
the proof uses an induction argument . for @xmath494 and @xmath443 , the result
has already been shown .
fix @xmath495 and suppose by induction that the property holds for any @xmath35-admissible family which is urce ( resp .
, both urce and ulce ) .
let @xmath496 be a @xmath497-admissible family which is urce ( resp . ,
both urce and ulce ) . as @xmath498 $ ] ,
it is sufficient to show the rce ( resp . , both rce and lce ) properties for the family @xmath499 for all @xmath500 .
let us introduce the following value function for each @xmath380 @xmath11 @xmath13 : @xmath501\qquad\mbox{a.s.}\ ] ] as for all @xmath47 , @xmath502 it is sufficient to prove that the biadmissible family @xmath503 is rce ( resp .
, both rce and lce ) as in the bidimensional case .
let @xmath384 and @xmath385 , @xmath386 be monotonic sequences of stopping times that converge , respectively , to @xmath14 and @xmath0 a.s .
we have @xmath504 \\ & & \qquad\leq \underbrace{e\bigl[\bigl|
u^{(i ) } ( \theta , s ) - u^{(i ) } ( \theta_n , s ) \bigr|\bigr]}_{\mbox{(i ) } } + \underbrace{e\bigl[\bigl| u^{(i ) } ( \theta_n , s ) - u^{(i ) } ( \theta_n , s_n ) \bigr|\bigr]}_{\mbox{(ii)}}.\end{aligned}\ ] ] let us show that ( i ) tends to @xmath41 as @xmath145 .
note that for each @xmath18 , @xmath505 is a @xmath35-admissible family of positive random variables which is urce ( resp .
, both urce and ulce ) and @xmath506 , @xmath507 is the corresponding value function family . by the induction assumption , this family is rce ( resp .
, both rce and lce ) .
hence , ( i ) converges a.s .
to @xmath41 as @xmath97 tends to @xmath250 when @xmath508 is monotonic .
let us now show that ( ii ) tends to @xmath41 as @xmath145 . by definition of the value function @xmath509 ( [ ui ] ) , it follows that @xmath510 \leq \sup_{\theta \in t_0^{d } } \bigl| e\bigl [ \psi^{(i ) } ( \theta , s)\bigr ] - e\bigl[\psi^{(i ) } ( \theta , s_n)\bigr ] \bigr|,\ ] ] and the right - hand side tends to @xmath41 by the urce ( resp .
, both urce and ulce ) properties of @xmath5 . by theorem [ t.1 ] , the regularity properties of the new reward will ensure the existence of an optimal stopping time @xmath511 for @xmath31 . by proposition [ d - construction ] , this will allow us to show by induction the existence of an optimal stopping time for @xmath3 .
[ toptd ] let @xmath450 be a @xmath35-admissible family of positive random variables which is urce and ulce .
there then exists a @xmath512 optimal for @xmath3 , that is , such that @xmath513.\ ] ] the result is proved by induction on @xmath35 . for @xmath494
the result is just theorem [ t.1 ] .
suppose now that @xmath495 and suppose by induction that for all @xmath35-admissible families which are urce and ulce , optimal @xmath35-stopping times do exist .
let @xmath514 be a @xmath497-admissible family which is urce and ulce .
the existence of an optimal @xmath497-stopping time for the associated value function @xmath3 will be derived by applying proposition [ d - construction ] . now , by proposition [ pd.ngrce ] , the new reward family @xmath515 is lce and rce . by theorem [ t.1 ] ,
there exists an optimal stopping time @xmath110 for @xmath31 .
thus , we have proven that condition 1 of proposition [ d - construction ] is satisfied .
note now that for @xmath500 , the @xmath35-admissible families @xmath516 are urce and ulce .
thus , by the induction hypothesis , for each @xmath47 , there exists an optimal @xmath517 for the value function @xmath518 defined by ( [ ui ] ) .
noting that @xmath519 , we have proven that condition 2 of proposition [ d - construction ] is satisfied .
now applying proposition [ d - construction ] , the result follows .
suppose that @xmath520 is symmetric with respect to @xmath521 that is , @xmath522 for each permutation @xmath523 of @xmath524 .
by symmetry we can suppose that @xmath525 , that is , that the value function @xmath3 coincides with @xmath526},\ ] ] where @xmath527 s.t .
it follows that the value functions @xmath529 and the new reward @xmath54 coincide and are simply given for each @xmath530 by the following random variable : @xmath531.\ ] ] the reduction property can be written as follows : @xmath532.\ ] ] we then consider the value function @xmath533 .
the associated new reward is given for @xmath534 , @xmath535 such that @xmath536 by @xmath537.\ ] ] we then consider the value function @xmath538 and so on .
thus , by forward induction , we define the new rewards @xmath539 for @xmath540 by @xmath541\ ] ] for each @xmath542 .
the reduction property gives @xmath543.\ ] ] note that for @xmath544 , @xmath545\ ] ] for each @xmath546 .
hence , using backward induction we can now define @xmath547 by ( [ phid ] ) and then @xmath548 by the induction formula ( [ phii ] ) .
consequently , we have the following characterization of the value function and construction of a multiple optimal stopping time ( which are rather intuitive ) . [ popo ]
* let @xmath450 be a symmetric @xmath35-admissible family of random variables , and for each stopping time @xmath0 , consider the associated value function @xmath3 .
+ let @xmath549 , @xmath550 be defined by backward induction as follows : @xmath551 is given by ( [ phid ] ) for each @xmath552 . also , for @xmath553 and each @xmath554 , @xmath555 is given in terms of the function @xmath556 by backward induction formula ( [ phii ] ) .
+ the value function then satisfies @xmath557.\ ] ] * suppose that @xmath450 is urce and ulce .
let @xmath558 be an optimal stopping time for @xmath3 given by ( [ vf ] ) , let @xmath559 be an optimal stopping time for @xmath560 given by ( [ phiun ] ) and for @xmath561 , let @xmath562 be an optimal stopping time for @xmath563 given by ( [ phii ] ) .
+ then , @xmath564 is a multiple optimal stopping time for @xmath3 . first , consider the very simple additive case : suppose that the reward is given by @xmath565 where @xmath566 is an admissible family of random variables such that @xmath567\!<\ !
\infty$ ] .
we then obviously have that @xmath568 , where @xmath569 is the value function of the single optimal stopping time problem associated with reward @xmath566 .
also , if @xmath558 is an optimal stopping time for @xmath570 , then @xmath571 is optimal for @xmath3 .
let us now consider the more interesting additive case of _ swing options _
: suppose that @xmath572 and that the reward is still given by ( [ add ] ) , but the stopping times are separated by a fixed amount of time @xmath573 ( sometimes called `` refracting time '' ) . in this case , the value function is given by @xmath27 , ( \tau _ 1,\ldots,\tau_d ) \in\mathcal{s}^d_s\},\ ] ] where @xmath574 .
all the previous properties then still hold .
again , the @xmath549 satisfy the following induction equality : @xmath575.\ ] ] one can then easily derive that @xmath576 , where @xmath577 .
\\[-1pt ] % \phi_{d-2 } ( \theta_1,\ldots , \theta_{d-2})&= & y(\theta_1 ) + \cdots + y(\theta_{d-2})+ z_{d-2 } ( \theta_{d-2 } ) , \qquad \mbox{where } % \\[-1pt ] z_{d-2 } ( \theta_{d-2 } ) & = & \operatorname{ess}\sup _ { \tau\in t _
{ \theta _ { d-2}+ \delta } } e[y(\tau)+z_{d-1 } ( \tau ) |\mathcal{f } _ { \theta_{d-2 } } ] , \end{aligned}\ ] ] and so on .
hence , for @xmath578 , @xmath579 , where @xmath580.\ ] ] the value function satisfies @xmath581.\ ] ] this corresponds to proposition 3.2 of carmona and dayanik ( @xcite ) .
suppose that @xmath566 is rce and lce .
let @xmath558 be the minimal optimal stopping time for @xmath3 given by ( [ vsdeu ] ) and for @xmath582 , let @xmath562 be the minimal optimal stopping time for @xmath583 .
the @xmath35-stopping time @xmath564 is then the minimal optimal stopping time for @xmath3 .
this corresponds to proposition 5.4 of carmona and dayanik ( @xcite ) .
note that the multiplicative case can be solved similarly .
further applications to american options with multiple exercise times are studied in kobylanski and quenez ( @xcite ) .
as explained in the , in previous works on the optimal single stopping time problem , the reward is given by an rcll positive adapted process @xmath584 . moreover , when the reward @xmath22 is continuous , an optimal @xmath0-stopping time is given by @xmath585 which corresponds to the first hitting time after @xmath0 of @xmath41 by the rcll adapted process @xmath586 .
this formulation is very important since it gives a simple and efficient method to compute an optimal stopping time . in the two - dimensional case , instead of considering a reward process , it is quite natural to suppose that the reward is given by a biprocess @xmath587 ^2}$ ] such that a.s . , the map @xmath588 is continuous and for each @xmath269 ^2 $ ] , @xmath266 is @xmath589-measurable ( see remark [ biprocess ] ) .
we would like to construct some optimal stopping times by using hitting times of processes . by the existence and construction properties of optimal stopping times given in theorem [ topt ] , we are led to construct @xmath110 , @xmath348 and @xmath346 as hitting times of processes . since @xmath262 is a continuous biprocess , there is no problem for @xmath348 , @xmath346 .
however , for @xmath110 we need to aggregate the new reward @xmath62 , which requires new aggregation results .
these results hold under stronger assumptions on the reward than those made in the previous existence theorem ( theorem [ topt ] ) .
recall the classical result of aggregation of a supermartingale system [ el karoui ( @xcite ) ] .
[ p.sma ] let @xmath590 be a supermartingale system which is rce and such that @xmath591 .
there then exists an rcll adapted process @xmath592 which _ aggregates _ the family @xmath593 that is , for each @xmath18 , @xmath594 a.s .
this lemma relies on a well - known result [ see , e.g. , el karoui ( @xcite ) or theorem 3.13 in karatzas and shreve ( @xcite ) ; for details , see the proof in section [ 3.14 ] ] . classically , the above proposition [ p.sma ] is used to aggregate the value function of the single stopping time problem .
however , it can not be applied to the new reward since it is no longer a supermartingale system .
thus , we will now state a new result on aggregation .
let us introduce the following right - continuous property for admissible families .
an admissible family @xmath62 is said to be _ right - continuous along stopping times ( rc ) _ if for any @xmath595 and any sequence @xmath126 of stopping times such that @xmath127 a.s . , we have @xmath596 a.s .
we state the following result .
[ t.aggregphi ] suppose that the admissible family of positive random variables @xmath597 is right - continuous along stopping times .
there then exists a progressive process @xmath598 such that for each @xmath599 , @xmath600 a.s . and such that there exists a nonincreasing sequence of right - continuous processes @xmath601 such that for each @xmath602 $ ] , @xmath603 .
see section [ 3.14 ] . first , recall the following classical result [ el karoui ( @xcite ) ] .
[ p1.aggregv ] let @xmath604 be an admissible family of random variables which is rce .
suppose that @xmath605<\infty$ ] .
there then exists an rcll supermartingale @xmath23 which aggregates the family @xmath24 defined by ( [ vs ] ) , that is , for each stopping time @xmath0 , @xmath606 the family @xmath607 is a supermartingale system ( proposition [ p1.superm ] ) and has the rce property ( proposition [ l1 ] ) .
the result clearly follows by applying the aggregation property of supermartingale systems ( proposition [ p.sma ] ) .
[ p.prop ] suppose the reward is given by an rc and lce admissible family @xmath149 such that @xmath608<\infty$ ] .
let @xmath22 be the progressive process given by theorem 4.1 that aggregates this family .
let @xmath24 be the family of value functions defined by ( [ vs ] ) , and let @xmath23 be an rcll adapted process that aggregates the family @xmath24 .
the random variable defined by @xmath609 is the minimal optimal stopping time for @xmath3 , that is , @xmath610 a.s . as for theorem [ t.1 ] ,
the proof relies on the construction of a family of stopping times that are approximatively optimal .
the details , which require some fine techniques of the general theory of processes , are given in section [ 3.14 ] .
in the case of an rcll reward process supposed to be lce , the above theorem corresponds to the classical existence result [ see el karoui ( @xcite ) and karatzas and shreve ( @xcite ) ] . for simplicity
, we study only the case when @xmath443 .
we will now prove that the minimal optimal pair of stopping times @xmath417 defined by ( [ mio ] ) can also be given in terms of _ hitting times_. in order to do this , we first need to aggregate the value function and the new reward . [ p2.aggregv ] suppose the reward is given by an rce biadmissible family @xmath302 such that @xmath611<\infty } $ ] .
there then exists a supermartingale @xmath612 with rcll paths that aggregates the family @xmath24 defined by ( [ vs ] ) , that is , such for each @xmath0 @xmath11 @xmath13 , @xmath613 a.s .
the rce property of @xmath396 shown in proposition [ l2.vrce ] , together with the supermartingale property [ proposition [ p2.1](3 ) ] gives , by proposition [ p.sma ] , the desired result .
we will now study the aggregation problem of the new reward family @xmath62 .
let us introduce the following definition . a biadmissible family @xmath614 is said to be _ uniformly right - continuous along stopping times ( urc ) _
if @xmath615 < \infty$ ] and if for each nonincreasing sequence of stopping times @xmath373 in @xmath20 which converges a.s .
to a stopping time @xmath18 , @xmath616=0 \qquad\mbox{a.s.}\ ] ] and @xmath617=0\qquad\mbox{a.s.}\ ] ] the following right continuity property holds true for the new reward family .
[ t2.ngrc ] suppose that the admissible family of positive random variables @xmath256 is urc .
the family of positive random variables @xmath377 defined by ( [ phi ] ) is then rc . as @xmath618 $ ]
, it is sufficient to show the rc property for the family @xmath619 . now , for all @xmath47 , @xmath620 a.s . , where @xmath621 is rc .
let @xmath384 and @xmath385 , @xmath386 be nonincreasing sequences of stopping times in @xmath13 that converge to @xmath14 and @xmath0 a.s .
we have @xmath622 \(i ) _ tends to _ 0 _ as _ @xmath145 . for each @xmath18 ,
as @xmath390 is an admissible family of positive random variables which is rc , proposition [ p2.aggregv ] gives the existence of an rcll adapted process @xmath623 such that for each stopping time @xmath14
@xmath11 @xmath13 , @xmath624 ( i ) can be rewritten as @xmath625 a.s . , which converges a.s .
to @xmath41 as @xmath97 tends to @xmath250 by the right continuity of the process @xmath626 .
\(ii ) _ tends to _ @xmath41 _ as _ @xmath145 . by definition of the value function @xmath392 ( [ u1bis ] )
, it follows that @xmath627 for any @xmath628 , where @xmath629 is an rcll version of the conditional expectation .
hence , by the right continuity of this process , for each fixed @xmath630 , the sequence of random variables @xmath631 converges a.s .
to @xmath632 as @xmath97 tends to @xmath250 .
it follows that for each @xmath633 , @xmath634 now , the sequence @xmath635 converges a.s .
to @xmath41 and @xmath636 note that the second member of this inequality is integrable . by the lebesgue theorem for the conditional expectation
, @xmath637 converges to @xmath41 in @xmath638 as @xmath639 tends to @xmath250 .
the sequence @xmath635 is decreasing .
it follows that the sequence @xmath640 is also decreasing and hence converges a.s .
since this sequence converges to @xmath41 in @xmath638 , its limit is also @xmath41 almost surely . by letting @xmath639 tend to @xmath250 in ( [ uz ] ) ,
we obtain @xmath641 the proof of theorem [ t2.ngrc ] is thus complete .
[ c2.1 ] under the same hypothesis as theorem [ t2.ngrc ] , there exists some progressive right - continuous adapted process @xmath60 which aggregates the family @xmath642 @xmath643 , that is , @xmath600 a.s . for each @xmath47 , and such that there exists a decreasing sequence of right - continuous processes @xmath601 that converges to @xmath22 .
this follows from the right continuity of the new reward ( theorem [ t2.ngrc ] ) which we can aggregate ( theorem [ t.aggregphi ] ) . for the optimal @xmath35-stopping time problem , the same result holds for urc @xmath35-admissible families @xmath644 , @xmath645 , that is , families that satisfy @xmath646 < \infty$ ] and @xmath647 for @xmath648 and sequences @xmath232 in @xmath13 such that @xmath374 a.s .
the proof is strictly the same , with @xmath649 replaced by @xmath650 for @xmath14 , @xmath18 and @xmath651 with @xmath652 replaced by @xmath653 with @xmath654 and @xmath18 .
as before , for the sake of simplicity , we suppose that @xmath443 .
suppose that @xmath655 is a urc and ulce biadmissible family .
let @xmath62 be the new reward family . by theorem [ t2.nrcadge
] , this family is lce .
furthermore , by theorem [ t2.ngrc ] , this family is rc .
let @xmath22 be the progressive process that aggregates this family , given by theorem [ t.aggregphi ] .
let @xmath656 be an rcll process that aggregates the value function associated with @xmath22 . by theorem [ p.prop ] , the stopping time @xmath657 is optimal for @xmath31 . the family @xmath658 , @xmath659 is admissible , rc and lce .
let @xmath660 be the progressive process that aggregates this family given by theorem [ t.aggregphi ] .
let @xmath661 be an rcll process that aggregates the value function associated with @xmath660 . by theorem [ p.prop ] the stopping time @xmath662 is optimal for @xmath663 and @xmath664 .
the family @xmath665 , @xmath659 is admissible , rc and lce . let @xmath666 be the progressive process that aggregates this family given by theorem [ t.aggregphi ] .
let @xmath667 be an rcll process that aggregates the value function associated with @xmath666 . by theorem [ p.prop ] , the stopping time @xmath668 is optimal for @xmath669 , and @xmath670 . by proposition [ pconstruction ] , the pair of stopping times @xmath351 defined by @xmath671 where @xmath672 , is optimal for @xmath3 .
let @xmath302 be a biadmissible family which is urc and ulce .
the pair of stopping times @xmath417 defined by ( [ eto ] ) is then optimal for @xmath3 .
note that the above construction of @xmath351 as hitting times of processes requires stronger assumptions on the reward than those made in theorem [ topt ] .
furthermore , let us emphasize that it also requires some new aggregation results ( theorems [ t.aggregphi ] and [ p.prop ] ) .
we now give the proofs of proposition [ p.sma ] and theorems [ t.aggregphi ] and [ p.prop ] .
first , we give the short proof of the classical proposition [ p.sma ] which we recall here ( for the reader s convenience ) .
let @xmath673 be a supermartingale system which satisfies @xmath591 and which is right - continuous along stopping times in expectation .
there then exists an rcll adapted process @xmath592 which aggregates the family @xmath674 , that is , @xmath675 a.s .
let us consider the process @xmath676 it is a supermartingale and the function @xmath677 is right - continuous . by classical results
[ see theorem 3.13 in karatzas and shreve ( @xcite ) ] , there exists an rcll supermartingale @xmath678 such that for each @xmath679 $ ] , @xmath680 a.s .
it is then clear that for each dyadic stopping time @xmath0
@xmath11 @xmath13 , @xmath681 a.s .
( for details , see part 2 of the proof of theorem [ t.1 ] ) .
this implies that @xmath682= e[h(s)].\ ] ] since the process @xmath678 is rcll and since the family @xmath683 is right - continuous in expectation , equality ( [ eta ] ) still holds for any stopping time @xmath0
@xmath11 @xmath13 .
it then remains to show that @xmath681 a.s . , but this is classical .
let @xmath684 and define @xmath685 . since @xmath686 is a stopping time , @xmath687= e[h(s_a)]$ ] .
since @xmath688 a.s .
, it gives that @xmath689 = e[h(s){\mathbf{1}}_a ] , $ ] from which the desired result follows .
we now give the proof of theorem [ t.aggregphi ] .
suppose that the admissible family of positive random variables @xmath690 is right - continuous along stopping times .
there then exists a progressive process @xmath598 such that for each @xmath14 @xmath11 @xmath13 , @xmath691 and such that there exists a nonincreasing sequence of right - continuous processes @xmath601 such that for each @xmath692 @xmath11 @xmath693 $ ] , @xmath694 . for each @xmath97 @xmath11 @xmath695 , let us define a process @xmath696 that is a function of @xmath697 by @xmath698 t , ( [ 2^n t]+1)/2^n [ } \phi(s\wedge t)\ ] ] for each @xmath692 @xmath11 @xmath693,$ ] where @xmath699 is the set of dyadic rationals . for each @xmath700
$ ] and each @xmath701 , the process @xmath702 is @xmath703-adapted and , for each @xmath704 , the function @xmath705 is right - continuous . hence
, the process @xmath706 is also @xmath703-progressive .
moreover , the sequence @xmath707 is decreasing .
let @xmath708 be its limit , that is , for each @xmath602 $ ] , @xmath709 it follows that for each @xmath710 , the process @xmath60 is @xmath703-progressive .
thus , @xmath60 is @xmath711-progressive and consequently @xmath712-progressive since @xmath713 .
_ step 1 _ : fix @xmath47 .
let us show that @xmath714 a.s .
let us suppose , by contradiction , that the above inequality does not hold .
there then exists @xmath715 such that the set @xmath716 satisfies .
fix @xmath717 . for all
@xmath718 we have that @xmath719 where @xmath720 is defined by ( [ trois ] ) with @xmath721 replaced by @xmath722 . by definition of @xmath723
there exists @xmath724\theta(\omega ) , \frac{[2^n\theta(\omega)]+1}{2^n}[\,\cap\,\mathbb{d}$ ] such that @xmath725 we introduce the following subset of @xmath726\times\omega$ ] : @xmath727\theta(\omega ) , \frac { [ 2^n\theta(\omega)]+1}{2^n}\biggl[\,\cap\,\mathbb{d}\mbox { and } \phi ( \theta ) ( \omega ) \leq\phi(t)(\omega ) -\frac{\varepsilon}{2 } \biggr\}.\ ] ] first , note that @xmath728 is optional .
indeed , we have @xmath729 where @xmath730 + 1}{2^n}\biggr\ } \cap \biggl\ { \phi(\theta ) \leq\phi(t ) -\frac{\varepsilon}{2 } \biggr\},\ ] ] and the process @xmath731 is optional since @xmath14 and @xmath732 + 1}{2^n}$ ] are stopping times and @xmath733 is admissible .
also , @xmath70 is included in @xmath734 , the projection of @xmath728 onto @xmath735 , that is , @xmath736
\mbox { s.t . }
( t , \omega ) \in\overline a_n \}.\ ] ] hence , by a section theorem [ see dellacherie and meyer ( @xcite ) , chapter iv ] , there exists a dyadic stopping time @xmath737 such that for each @xmath264 in @xmath738 , @xmath739 and @xmath740 hence , for all @xmath264 in @xmath738 @xmath741\theta(\omega ) , \frac { [ 2^n\theta(\omega)]+1}{2^n}\biggr[\cap\mathbb{d}.\ ] ] note that @xmath742 put @xmath743 .
we have @xmath744 and @xmath745 for each @xmath97 on @xmath746 . by letting @xmath97 tend to @xmath250 in this inequality ,
since @xmath733 is right - continuous along stopping times , we derive that @xmath747 a.s . on @xmath748 , which gives the desired contradiction .
_ step 2 _ : fix @xmath14 @xmath11 @xmath13 .
let us show that @xmath749 a.s .
put @xmath750 + 1}{2^n}$ ] .
the sequence @xmath751 is a nonincreasing sequence of stopping times such that @xmath752 .
moreover , note that since the family @xmath753 is admissible , for each @xmath35 @xmath754 , for almost every @xmath755 , @xmath756 .
now , we have @xmath757\theta , t^n[\,\cap\,\mathbb{d}$ ] .
also , for each and each @xmath758\theta(\omega ) , t^n(\omega)[\,\cap\,\mathbb{d}$ ] , @xmath759\theta(\omega ) , t^n(\omega)[\cap \mathbb{d } } \phi(s)(\omega)= \phi^{n}_{\theta(\omega ) } ( \omega),\ ] ] where the last equality follows by the definition of @xmath760 [ see ( [ trois ] ) , with @xmath721 replaced by @xmath722 ] . hence , @xmath762 letting @xmath97 tend to @xmath250 , by using the right - continuous property of @xmath733 along stopping times and the convergence of @xmath760 to @xmath763 for each @xmath264 , we derive that @xmath764 a.s .
we now give the proof of theorem [ p.prop ] .
@xmath765 is an optimal stopping time for @xmath766 .
we begin by constructing a family of stopping times that are approximatively optimal . for @xmath1890,1[$ ] , define the stopping time @xmath767 the proof follows the proof of theorem [ t.1 ] exactly , except for step 1 , which corresponds to the following lemma .
[ lemo ] for each @xmath18 and @xmath1890,1[$ ] , @xmath768 by the same arguments as in the proof of theorem [ t.1 ] , @xmath769 is nondecreasing with respect to @xmath155 and converges as @xmath168 to an optimal stopping time which coincides with @xmath770 a.s .
proof of lemma [ lemo ] to simplify notation , @xmath771 will be written as @xmath772 .
for the sake of simplicity , without loss of generality , we suppose that @xmath773 is rcll for each @xmath774 .
fix @xmath775 . in the following
, we use only simple analytic arguments . by definition of @xmath776 ( [ tls ] ) , for each @xmath777 , there exists @xmath721 @xmath778 such that @xmath779 also , note that for each @xmath780 , @xmath781 .
now , fix @xmath780 and @xmath135 . by the right continuity of @xmath782 and @xmath783 ,
there exists @xmath784
@xmath11 @xmath785 such that @xmath786 note that @xmath787 and @xmath788 for any @xmath97 . again , by using the right continuity of @xmath789 and @xmath790 and by letting @xmath97 tend to @xmath250 in ( [ lemd ] ) , we derive that @xmath791 and this inequality holds for each @xmath135 , @xmath780 and @xmath264 @xmath11 @xmath735 . by letting @xmath639 tend to @xmath250 and @xmath792
tend to @xmath41 , we derive that for each @xmath264 @xmath11 @xmath735 , @xmath793 which completes the proof of the lemma .
[ app ] [ [ app1 ] ] we recall the following classical theorem [ see , e.g. , karatzas and shreve ( @xcite ) , neveu ( @xcite ) ] . [ ta ]
let @xmath794 be a probability space and let @xmath795 be a nonempty family of positive random variables defined on @xmath794 .
there exists a random variable @xmath796 satisfying : 1 . for all @xmath797 , @xmath798 a.s .
; 2 . if @xmath566 is a random variable satisfying @xmath799 a.s . for all @xmath800 , then @xmath801 a.s .
this random variable , which is unique a.s . , is called the _ essential supremum of _ @xmath802 and is denoted @xmath803 .
furthermore , if @xmath804 is closed under pairwise maximization @xmath805 implies @xmath806 ) , then there is a nondecreasing sequence @xmath807 of random variables in @xmath804 satisfying @xmath808 a.s . [ [ app2 ] ] in order to give a characterization of _ minimal optimal _ stopping times , we introduce the following partial order relation on @xmath809 : @xmath810 if and only if @xmath811.\ ] ] note that although the minimum of two elements of @xmath809 is not defined , the infimum , that is , the greatest minorant of the couple , does exist and @xmath812= { \mathbf{1}}_{\{a\wedge b < a'\wedge b'\}}(a ,
b)+ { \mathbf{1}}_{\{a'\wedge b ' < a\wedge b\}}(a',b ' ) + { \mathbf{1}}_{\{a\wedge b = a'\wedge b'\}}(a\wedge a ' , b\wedge b')$ ] . note also that if @xmath323 , @xmath813 @xmath814 are optimal for @xmath3 , then the infimum of the couple @xmath815 $ ] , in the sense of the relation @xmath816 a.s . , is optimal for @xmath3 .
the two following assertions can be shown to be equivalent : 1 .
a pair @xmath323 @xmath814 is _ minimal optimal _ for @xmath3 ( i.e , is the minimum for the order @xmath816 a.s . of the set @xmath817 , @xmath818\}$ ] ) ,
@xmath819 and @xmath348 , @xmath346 @xmath820 are such that @xmath821 on @xmath822 and @xmath823 on @xmath824 ; 2 .
a. @xmath110 @xmath820 is minimal optimal for @xmath31 ; b. @xmath346 @xmath820 is minimal optimal for @xmath347 on @xmath825 c. @xmath348 @xmath820 is minimal optimal for @xmath349 on @xmath826 , and @xmath827 , @xmath828 . consider the following partial order relation @xmath829 on @xmath488 defined by induction in the following way : for @xmath494 , @xmath830 , @xmath831 if and only if @xmath832 , and for @xmath833 , @xmath834 if and only if either @xmath835 or @xmath836 note that for @xmath443 the order relation @xmath837 is the order relation @xmath816 defined above .
one can show that a @xmath35-stopping time @xmath838 is the @xmath35-minimal optimal stopping time for @xmath3 , that is , it is minimal for the order @xmath829 in the set @xmath839 \}$ ] if and only if : 1 .
@xmath840 is minimal optimal for @xmath31 ; 2 . for @xmath841
, @xmath842 is the @xmath423-minimal optimal stopping time for @xmath843 on the set @xmath844
the authors thank gilles pags and the anonymous referee for their relevant remarks and suggestions . | we study the optimal multiple stopping time problem defined for each stopping time @xmath0 by @xmath1 $ ] .
the key point is the construction of a _ new reward _
@xmath2 such that the value function @xmath3 also satisfies @xmath4 $ ] .
this new reward @xmath2 is not a right - continuous adapted process as in the classical case , but a family of random variables .
for such a reward , we prove a new existence result for optimal stopping times under weaker assumptions than in the classical case .
this result is used to prove the existence of optimal multiple stopping times for @xmath3 by a constructive method . moreover , under strong regularity assumptions on @xmath5 , we show that the new reward @xmath2 can be aggregated by a progressive process .
this leads to new applications , particularly in finance ( applications to american options with multiple exercise times ) . , + . . |
the data we analysed were obtained with the texes ( texas echelon cross - echelle spectrograph ; see lacy et al .
2002 ) mounted on the nasa infrared telescope facility on mauna kea .
texes is capable of recording truly unique mid - infrared spectra .
we used the spectrograph in its high - resolution mode , with @xmath3 covering the regions 806 - 822 @xmath4 ( 12.2 - 12.4 @xmath5 m ) and 884 - 925 @xmath4 ( 10.8 - 11.3 @xmath5 m ) .
texes is actually a prototype for the exes spectrometer , which is a pi instrument to be used on the american - german stratospheric observatory for infrared astronomy ( sofia ) which is scheduled to fly in 2004 .
one of the largest advantages of flying an observatory above the tropopause at an altitude of 12 - 14 km compared to a ground - based one , is the smaller interference of telluric water vapor which absorbes heavily in the thermal infrared .
sofia will be flying above 99% of the telluric water vapor .
exes has a 1 meter long echelon grating and texes has a 90 cm grating .
the detectors are si : as @xmath6 pixels detectors , which are sensitive to mid - infrared light from ca .
4.5 to 28.5 microns .
we identified the unexpected absorption features in the mid - infrared spectrum of arcturus as water vapor and oh lines by comparing with sun - spot spectra .
since there does not exist any water - vapor line - lists with wavelengths accurate enough for our high - resolution work , we constructed a new one with wavelengths from recent laboratory measurements by polyansky et al .
( 1996 , 1997a , & 1997b ) .
the lines studied here are pure rotational transitions of oh of the three first vibrational levels and pure rotational transition of water within the ground vibrational state .
vibration - rotation lines of water vapor lie in the near to mid infrared region and electronic transitions in the uv .
we model our observations with the marcs model atmosphere code .
these 1d hydrostatic , spherical model photospheres are computed on the assumptions of local thermodynamic equilibrium ( lte ) , homogeneity and the conservation of the total flux ( radiative plus convective ; the convective flux being computed using the mixing length formulation ) . the radiative field used in the model generation , is calculated with absorption from atoms and molecules by opacity sampling in approximately 84000 wavelength points over the entire , relevant wavelength range considered for the star ( @xmath7@xmath8 ) .
the fundamental parameters for our standard model are @xmath9 k , @xmath10 ( cgs ) , [ fe / h]@xmath11 , and @xmath12 km s@xmath13 , and use the abundances of carbon , nitrogen , and oxygen given by decin et al .
synthetic spectra are subsequently calculated .
the overall spectral shape of a red giant as warm as arcturus is primarily determined by continuous opacities and metals , but also by a small contribution of co. water vapor is , for example , not expected in the spectrum of arcturus .
this is in stark contrast to cooler giants where molecules ( such as tio , vo , h@xmath14o , oh , co , and sio ) are totally dominant , the metals only playing a minor role .
the main continuous opacity source in the optical is due to h@xmath15 , i.e. ionization and recombination of the h@xmath16-ion . further out in the infrared , h@xmath17 opacity (
i.e. bremsstrahlung and absorption by an electron in the field of a neutral hydrogen atom ) is most important .
this opacity source grows with wavelength and dominates beyond 1.6 microns . at 1.6 @xmath5 m
there is a minimum in the opacity , which means that the continuum at 1.6 microns is formed relatively far down in the photosphere . as we go further out in the infrared the continuum
is formed successively further up in the photosphere .
we therefore expect also mid - infrared lines to be formed further out than lines at , say , 1.6 microns .
we start off by testing our model photosphere and the stellar parameters used by calculating a synthetic spectrum of the 1.6 micron region , and confronting it with the observed , high - resolution , high - signal - to - noise spectrum found in the arcturus atlas ( hinkle et al .
the oh first overtone lines found in this region are , as was noted above , formed at a deep level in the photosphere where the model assumptions are certainly valid .
we are able to model the oh lines well and are thus confident that our parameters are correct .
next , we turn our attention to the 12 micron region to study the pure rotational lines .
the continuum is formed further out at these wavelengths and the cores of moderately strong lines could be formed in shallow layers in the atmosphere where the model assumptions may be erroneous .
we find that the oh(@xmath18 ) and oh(@xmath19 ) lines can be modelled well .
however , the model can not account for the deep cores of the oh(@xmath20 ) lines .
furthermore , the model does not predict any water lines at all , which is a large fallacy .
since we are working with high - resolution data , our lines are resolved , thus providing kinematic information .
based on this piece of information , the line widths , and the excitation temperatures of the lines , we argue that all the lines are formed within the photosphere , and not above the photosphere , for example in a circumstellar envelope ( see ryde et al .
2002 ) .
we therefore construct a semi - empirical model - atmosphere based on our marcs model , by adjusting the temperature profile in the outer most , uncertain , layers of the model photosphere .
the model now violates the condition of flux constancy imposed on marcs models .
electron and gas pressures were recalculated assuming hydrostatic equilibrium .
the density and standard opacity are also consistently recalculated from this new temperature structure , as are the ionization balance , molecular equilibria , and opacities .
new synthetic spectra were calculated assuming , as before , lte for the molecular equilibrium and the line formation .
the revisions to the upper boundary layer made by us were made solely with the aim of fitting observed line profiles .
we are successful in finding a semi - empirical model that predicts the strengths of all the oh lines , the water vapor lines , and also the co vibration - rotation , 4.6 micron lines , the cores of which are also formed far up in the photosphere . over most of the modified region ,
the temperatures are a mere 300 k cooler than the original marcs model . in figure 1
a few water vapor lines are shown together with the synthetic spectrum based on the original marcs atmosphere and the spectrum based on the semi - empirical model photosphere .
what could cause the detected stellar steam in the disk - averaged spectrum of this k2 giant ? the outer layers of the atmosphere seem to be more complicated than the simple assumptions made in our models .
what physics are we missing ?
we will now discuss a few points related to this .
first , relaxation of the assumptions behind the marcs models may change the atmospheric structure .
even small alterations of the heating or cooling terms in the energy equation ( for example due to dynamic processes or uncertainties and errors in the calculations of radiative cooling ) may lead to changes in the temperature structure in the outer , tenuous regions of the photosphere where the heat capacity per volume is low .
the assumption of lte is one target of model builders seeking to go beyond the standard assumptions .
continuous and line opacity was treated in non - lte using large model atoms by short ( 2002 ) in spherical geometry , hydrostatic equilibrium , and flux constancy .
introduction of non - lte in place of lte lowered the boundary temperature by an amount quite similar to our empirical cooling of the marcs model .
this suggests that non - lte effects in atoms may suffice to account for the surprising presence of h@xmath14o molecules in the photosphere .
second , there are certainly inhomogeneities over the surface of the star .
a key assumption made in construction and application of the model atmosphere is that physical conditions are everywhere the same at a given geometrical depth .
such a model is said to be ` homogeneous ' .
arcturus , with its deep convective envelope , is certain to be inhomogeneous ; it is likely to exhibit surface granulation which may be crudely characterized as consisting of columns of warm rising and cool sinking gas .
the continuum intensity contrast in the mid - infrared will be small .
there will , however , most likely be a sharp difference in the h@xmath14o column density with the sinking column much richer in the molecules , and , as a result , the surface - averaged spectrum will show stronger h@xmath14o lines than the spectrum of the equivalent homogeneous model atmosphere .
unfortunately , inhomogeneous model atmospheres of giant stars are unavailable . yet
, the modest 300 k cooling imposed on the marcs model is surely within the range of temperature differences between rising and sinking granules . an alternative mechanism for producing inhomogeneities is variously known as ` a molecular catastrophe ' or ` temperature bifurcation ' . in warm stars , the co molecular opacity influences the temperature profile predicted by a program like marcs .
if the temperature of gas in the boundary layer is perturbed , there may be a runaway to one of two different solutions .
suppose the gas is cooled , more co molecules form and increase the cooling of the boundary layer .
the temperature drop leads to yet more co molecules and to a runaway to a ` cool ' boundary layer .
conversely , if the perturbation raises the temperature , co molecules are dissociated , the cooling rate decreased , and the temperature rise is continued . in cool o - rich stars , where the co number density is limited by the abundance of carbon , this sensitivity of opacity to temperature is greatly reduced , but other molecules ( e.g. , sio or h@xmath14o ) may act in a similar way .
third , arcturus certainly exhibits a chromosphere , which has been modelled by ayres & linsky ( 1975 ) .
the chromospheric temperature rise starts at approximately @xmath21 which is deeper down in the photosphere than the on - set of our suggested temperature decline .
this bifurcated temperature structure has been found earlier for arcturus based on co 4.6 @xmath5 m data ( wiedemann et al .
( 1994 ) ) and has also been discussed for the sun ( see , for example , ayres 2002 ) .
existence of the chromosphere around arcturus suggests a heating mechanism active in and above the upper photosphere .
cuntz & muchmore ( 1989 ) made a preliminary study of the effects of propagating acoustic waves including molecular line cooling .
when the waves lead to weak shocks , a cool atmosphere is formed . on the other hand ,
strong shocks lead to a temperature inversion , approximately matching an empirical chromospheric model for arcturus .
these calculations incorporating a driving mechanism for temperature bifurcation support the idea of an inhomogeneous upper photosphere and low chromosphere .
chromospheres are indeed dynamic , and are thought to be a wave - driven phenomenon showing spatially and temporally intermittent structures , see for example the work on the solar chromosphere by carlsson & stein ( 1995 ) .
they find that the mean temperature is low , but also that transient , high temperatures give rise to the observed enhance emission in the ultra - violet . in the 3d simulations of the solar chromosphere by wedemeyer et al .
( 2002 ) , accoustic shocks are excited by the solar convection zone and form structures with different temperatures caused by interaction of propagating wave fronts .
they show that the cool phases lasts for approximately 1 minute , on average , until a new , hot wave front passes by again .
whether the timescales are long enough and the physical and chemical conditions are right for molecular formation and how molecules react to the shocks is not clear . in this context
it is interesting to note that our suggested model that accounts for the water - vapor , oh , and co features also predicts strong tio features in the optical , which is not observed .
the abundance of titanium is more than three orders of magnitude lower than that of oxygen and carbon , perhaps suggesting that tio may not have time to form between shocks .
i have discussed our unexpected discovery of absorption lines of water vapor in the disk - averaged spectrum of arcturus ( ryde et al . 2002 ) . based on kinematic information , line widths and excitation temperature , we argue that the water vapor is photospheric and not circumstellar . until now
, h@xmath14o has not been expected to exist in photospheres of stars of arcturus effective temperature .
this is true unless the outer layers of the model photosphere do not describe these regions properly . by modifying these layers
we have succeeded in finding _ a _ model which yields a synthetic spectrum matching the observations .
we are , however , not claiming that this necessarily is the only and true model .
our exercise simply shows that it is possible to achieve a photospheric spectrum containing water vapor also for early k giants if the outer parts are cooled for some reason .
possible reasons for the unexpected water vapor include photospheric inhomogeneities ( such as convective flows , molecular catastrophes , or even star spots ) and departures from lte in the photospheric structure or line formation in the boundary layers .
3d hydrodynamic models of red giants do not as yet exist , but may be a crucial tool for the understanding of how water vapor can exist on the surface of a warm star such as arcturus .
we strongly encourage model makers to pursue the difficult task of modelling red giants hydrodynamically .
this work was supported by the swedish foundation for international cooperation in research and higher education , stiftelsen blanceflor boncompagni - ludovisi , ne bildt , and the robert a. welch foundation of houston , texas .
the construction of the texes spectrograph was supported by grants from the nsf , and observing with texes was supported by usra the texas advanced research program .
nso / kitt peak fts data used here were produced by nsf / noao .
we acknowledge the support of the irtf staff . | in this talk i presented and discussed our unexpected detection of water vapor in the disk - averaged spectrum of the k2iiip red giant arcturus [ for details , see ryde et al .
( 2002 ) ] .
arcturus , or @xmath0 botes is , with its effective temperature of @xmath1 k , the hottest star yet to show water vapor features .
we argue that the water vapor is photospheric and that its detection provides us with new insights into the outer parts of the photosphere .
we are not able to model the vater vapor with a standard , one - component , 1d , radiative - equilibrium , lte model photosphere , which probably means we are lacking essential physics in such models .
however , we are able to model several oh lines of different excitation and the water - vapor lines satisfactorily after lowering the temperature structure of the very outer parts of the photosphere at @xmath2 and beyond compared to a flux - constant , hydrostatic , standard marcs model photosphere . our new semi - empirical model is consistently calculated from the given temperature structure .
i will discuss some possible reasons for a temperature decrease in the outer - most parts of the photosphere and the assumed break - down of the assumptions made in classical model - atmosphere codes . in order to understand the outer photospheres of these objects
properly , we will , most likely , need 3d hydrodynamical models of red giants also taking into account full non - lte and including time - dependent effects of , for example , acoustic wave heating sensitive to thermal instabilities .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in |
many nuclear physics experiments realized in the last years revealed an interesting scaling property of the isotopic yield ratios obtained from the disintegration of equilibrated sources of similar sizes , with close values of excitation energy per nucleon and temperature but different isospin values as a function of the isotopic composition of the emitted cluster @xcite . this property has been called isoscaling and can be expressed mathematically by the formula , @xmath0 where @xmath1 denotes the yield of the isotope @xmath2 obtained from the decay of the excited nuclear system @xmath3 and @xmath4 , @xmath5 and @xmath6 are scaling parameters .
if one classifies the nuclear reactions which manifest isoscaling function of the projectile energy , one can say that isoscaling has been observed in reactions induced by projectiles whose energies range from a few mev / nucleon to several gev / nucleon . in terms of reaction mechanisms ,
isoscaling has been obtained in evaporation reactions ( e.g. @xmath7he + @xmath8sn , @xmath9sn at 50 mev / nucleon bombarding energy @xcite ) , deep inelastic reactions ( e.g. @xmath10o + @xmath11th , @xmath10o + @xmath12au at 8.6 mev / nucleon @xcite , @xmath13kr +
@xmath14sn and @xmath13kr + @xmath15ni at 25 mev / nucleon @xcite ) and , recently , in fission reactions ( e.g. n + @xmath16u at 14 mev / nucleon @xcite ) .
the biggest amount of experimental isoscaling data comes from multifragmentation reactions @xcite and the great interest in these studies was motivated by the possibility to infer information on the nuclear equation of state , and in particular on the asymmetry term , from the isoscaling parameters @xcite .
it is interesting to specify at this point that , relatively easy to understand for nuclear reactions involving equilibrated compound nuclear systems @xcite , isoscaling is compatible also with dynamical fragment formation as was recently demonstrated in ref .
@xcite .
the aim of the present paper is to investigate isoscaling in multifragmentation reactions and more precisely the results produced within a microcanonical framework . in order to learn how sensitive the isoscaling parameters are with respect to the observables characterizing the source s state , we considered sources whose nucleon number varies from 40 to 200 , with excitation energies between 2 and 15 mev / nucleon , freeze - out volumes ranging from @xmath17 to @xmath18 in both primary and asymptotic stages of the decay .
the relevance of such study consists in the fact that the most suitable statistical ensemble to describe the decay of an isolated excited nuclear system counting at most several hundreds of nucleons is the microcanonical ensemble .
the importance of using models which can mimic as well as possible the real physical case is demonstrated by the fact that the best description of the experimental data was obtained by the predictions of microcanonical multifragmentation models @xcite . section [ sec.iso ] makes a review of the grandcanonical argumentation of isoscaling in nuclear multifragmentation .
section [ sec.micro ] presents microcanonical results on isoscaling obtained for different mass systems with excitation energies ranging from 2 to 15 mev / nucleon at different freeze - out volumes in both primary and asymptotic stages of the decay .
the sensitivity of the isoscaling parameters to observables characterizing the state of the source is discussed in detail .
deviations of microcanonical results from grandcanonical expectations are commented .
the conclusions are drawn in section [ sec.concl ] .
isoscaling in multifragmentation was for the first time evidenced for the light emitted fragments ( @xmath19 ) obtained in the reactions @xmath20sn+@xmath20sn and @xmath9sn+@xmath9sn at 50 mev / nucleon bombarding energy @xcite and confirmed by many other reactions induced by relativistic light particles ( e.g. p , d , @xmath5+@xmath20sn and @xmath9sn at incident energies between 660 mev and 15.3 gev @xcite ) or heavy ions ( @xmath9sn+@xmath21ni and @xmath20sn+@xmath22ni at 35 mev / nucleon @xcite ) .
the theoretical interpretation of isoscaling in the multifragmentation energy domain was given using a grandcanonical ensemble @xcite .
it was demonstrated that in the grandcanonical limit the yield of a primary fragment @xmath2 obtained from the decay of a thermally equilibrated system characterized by the temperature @xmath23 can be written as @xcite : @xmath24 where @xmath25 is the binding energy of the emitted fragment @xmath2 and @xmath26 and @xmath27 are the neutron and proton chemical potentials .
taking into account that the two equilibrated systems have the same temperature , excitation energy per nucleon and similar sizes the proportionality factors from eq .
( [ eq.yield ] ) have been considered practically identical in the two cases .
thus , one obtains for the ratio of the yields of any isotope resulting from the decay of the two sources the expression of eq .
( [ eq.isoscaling ] ) where , @xmath28 and @xmath29 where @xmath30 .
despite the fact that the suitability of a grandcanonical approximation is hard to be justified for systems having at most few hundreds particles , the success of the above formulas was due to their simplicity and to the fact that no contradiction came from experimental data or various statistical multifragmentation models ( e.g. microcanonical @xcite and canonical @xcite versions of smm , expanding emitting source @xcite ) currently used in data analyzes @xcite . concerning the slope parameters @xmath5 and @xmath6 , several values have been reported in literature .
this fact is expected because is known that particle formation probabilities have explicit dependence of all observables which characterize an excited source ( mass @xmath31 , charge @xmath32 , excitation energy @xmath33 , freeze - out volume @xmath34 ) .
it is also known that the sequential particle emission from the excited primary fragments modifies the isotopic yields .
this evaporation process may also affect the values of @xmath5 and @xmath6 .
despite the fact that , contrary to the grandcanonical case , no microcanonical model provides analytical expressions for cluster yields , the occurrence of isoscaling in a microcanonical model can be understood starting from the shape of the isotopic yields corresponding to a given element @xcite .
thus , it has been observed that charge distributions of fragments with fixed values of the neutron number ( @xmath35 ) and @xmath36 distributions of fragments with fixed values of the proton number ( @xmath37 ) can be approximated by gaussian functions .
the occurrence of such gaussian - like distributions within a microcanonical model can be regarded as a natural consequence of complete equilibration . using the generic notation @xmath38 for the distribution of the observable @xmath39
when @xmath40 is kept fixed ( where @xmath41 and @xmath42 ) , one can write , @xmath43 the mean value @xmath44 gives the isotope produced with the largest probability , while the variance @xmath45 is a measure of how uniform is the population of fragments having different isospin values .
the statistical meaning of these quantities involves an obvious dependence on observables determining the ensemble s state .
since we deal with a microcanonical approximation , these observables are the mass , volume and excitation energy of the multifragmenting source . using eq .
[ yields ] , the ratio of isotopic yields writes , @xmath46 . % \end{equation}\end{aligned}\ ] ] taking into account that the two considered sources have similar sizes , temperatures and freeze - out volumes , the isotopic distributions corresponding to any cluster are expected to have equal variances ( @xmath47 ) .
thus , in eq .
( [ eq.geom ] ) the quadratic term in @xmath39 vanishes leading to a linear dependence of the yield ratio on the observable @xmath39 of the emitted cluster . from eq .
( [ eq.geom ] ) one can express the @xmath5 and @xmath6 slope parameters in terms of mean values and variances of isotopic distributions , @xmath48 @xmath49 from eqs .
( [ eq.alpha.g ] ) and ( [ eq.beta.g ] ) results that @xmath5 and @xmath6 may depend on the size of the emitted cluster . a necessary condition to obtain slope parameters independent of the size of the emitted cluster as in eq .
[ eq.isoscaling ] , is to have a constant value for the ratio of the shift between the isotopic distributions produced by the two sources and the square of the distributions variance .
as we shall see in the following sections this hold with good approximation for the light clusters emitted by relatively large equilibrated sources but does not hold for heavy fragments emitted by heavy sources and clusters emitted by light multifragmenting sources . in the following
we shall investigate isoscaling over a large range of masses , excitation energies and freeze - out volumes of the multifragmenting sources in order to determine the dependence of the slope parameters @xmath5 and @xmath6 on the source parameters , the influence of the secondary decays on @xmath5 and @xmath6 and , in the case of small sources , the dependence of @xmath5 and @xmath6 on the size of the emitted cluster . to serve this goal we use the microcanonical multifragmentation model presented in detail in refs.@xcite .
the microcanonical multifragmentation model @xcite has two distinct stages : the break - up stage and the sequential particle evaporation stage .
the most important part of the model is the so called break - up stage which aims to describe the explosion of the equilibrated nuclear source . in order to mimic as well as possible the physical situation of a small isolated system with fixed excitation energy ,
the model tries to obey sharply to microcanonical constrains .
thus , the fixed observables characterizing the state of the equilibrated source are mass , charge , total energy @xmath50 , total momentum @xmath51 ( = 0 in the center of mass ( c.m . )
frame ) and total angular momentum @xmath52 ( = 0 for non - rotating systems ) .
the freeze - out volume @xmath34 is determined by the spherical container in which fragments are generated .
the standard version of the model allows treatment of freeze - out volume according to two different scenarios @xcite : hard spherical nonoverlapping fragments placed in a spherical container and spherical container with free volume parameterization @xcite . for a faster simulation we considered the second case .
the mathematical expression of the free volume is : @xmath53 where @xmath54 is the number of fragments in a given event and @xmath55 is the volume of the nuclear system at normal density .
the price to pay for a model aiming to respect all microcanonical constrains is the impossibility to obtain formulas analytically tractable .
the key quantity of the model is the statistical weight of a configuration @xmath4 , @xmath56 where @xmath57 is the number of fragments corresponding to configuration @xmath4 , @xmath58 is the accesible volume , @xmath59 is inertial tensor of the system and @xmath60 is the available kinetic energy .
the index @xmath61 denotes the fragments in each configuration .
the average value of any observable @xmath62 is calculated via a metropolis type simulation according to the expression , @xmath63 because the observables studied in the present paper are functions of isotopic yields they are strongly dependent on all specific details of the model ( binding energy parameterization , allowed isospin asymmetry of primary fragments , employed level density formula , etc . ) .
the break - up fragments allowed to be formed are all isotopes included in the mass table of ref . @xcite .
the fragments binding energy was calculated according to the liquid - drop parameterization : @xmath64 where @xmath65 .
fragments with mass smaller than 4 are considered without internal excitation energy ; larger fragments are allowed to carry an excitation energy ( @xmath66 ) upper limited by the binding energy according to the following level density formula : @xmath67 with @xmath68 mev@xmath69 @xcite and @xmath70=9 mev .
the factor @xmath71 takes into account the decrease of the excited level lifetime with the increasing excitation energy .
the excited primary fragments obtained in the break - up stage are allowed to decay by sequential particle emission ( second stage of the model ) as in ref .
the range of the evaporated particles is considered up to @xmath72 . as already stated in literature @xcite by modifying isotopic yields
, secondary particle production mechanisms may alter isoscaling parameters .
despite the fact that the employed parameterizations are expected to affect the specific values of @xmath5 and @xmath6 they do not affect the general behavior of isoscaling parameters as a function of observables characterizing the sources state such as the conclusions of the study are general and valid . in order to verify to what extend isoscaling holds within a microcanonical framework and , if this is the case , to make a systematic study of all source parameters which might have an influence on the isoscaling parameters we considered pairs of equilibrated sources with masses ranging from @xmath73 to @xmath74 , excitation energies ranging from 2 to 15 mev / nucleon , freeze - out volumes ranging from @xmath17 to @xmath18 in both primary and asymptotic stages of the reaction . for all considered cases we obtained isoscaling in the sense that @xmath75 shows an almost perfect linear dependence as a function of @xmath36 and @xmath32 .
as general observations , one may say that the quality of the isotopic scaling is better in the break - up stage than in the asymptotic stage of the decay and the maximum size of the emitted cluster for which isoscaling still holds decreases with increasing excitation energy .
we think that the more modest quality of the isoscaling in the asymptotic stage is an artifact of the simplified manner in which the secondary decays have been implemented .
we adopt the widely used convention to denote with the index `` 2 '' the more neutron rich system and with the index `` 1 '' the more neutron poor system .
in this situation the value of @xmath5 is always positive because more neutron rich clusters will be produced by the neutron richer source and the value of @xmath6 is always negative . to avoid possible effects of different magnitudes of coulomb interaction on isotopic distributions , the considered pairs of sources have the same proton number and different mass numbers . to illustrate how isoscaling looks like in the case of this microcanonical model , in fig .
1 we present the isotopic yield ratios as a function of @xmath36 and @xmath32 for the pair of equilibrated sources ( 200 , 80 ) and ( 185 , 80 ) at 5 mev / nucleon excitation energy and a freeze - out volume 6 times larger than the nuclear volume at normal density . as results from the figure ,
isoscaling holds in the strict sense of eq .
[ eq.isoscaling ] , namely linear behavior of @xmath76 as a function of @xmath36 and @xmath32 and constant value of @xmath4 .
special attention should be drawn to the fact that in fig .
[ fig : z=80 ] distances from the points corresponding to different values of @xmath77 is not equal because the considered sequence of @xmath77 does not contain exclusively consecutive numbers ( see legend ) .
it is interesting to notice that a linear behavior of @xmath76 is obtained also for emitted clusters much heavier than those for which the experimental analysis has been possible . as a technical detail ,
we stress that despite for a given element the linear behavior of @xmath78 holds for almost all obtained isotopes , for the fitting procedure only the most stable fragments have been selected .
since , as known from the early days of multifragmentation , the excitation energy of the nuclear source induces strong modifications on all fragment multiplicities , it is expected to affect also isotopic yield ratios . in fig .
[ fig : ab_ex ] are represented @xmath5 and @xmath6 for the equilibrated sources ( 200 , 80 ) and ( 185 , 80 ) , with @xmath79 as a function of the excitation energy .
it must be specified that , if not mentioned in a different manner , along this paper @xmath5 and @xmath6 are calculated such as to obtain the best fit for fragments with @xmath80 and respectively @xmath81 . this choice is made in order to realize an analysis as much as possible similar to the way in which the slope parameters have been obtained from experimental data and to have the same amount of considered data for all excitation energies and all considered source sizes .
one can see that both parameters have a monotonic dependence on the excitation energy , their absolute value decreasing with the temperature .
thus , increasing the excitation energy from 2 mev / nucleon to 15 mev / nucleon @xmath5 diminishes by a factor of 2.5 and @xmath82 diminishes by a factor of 3.5 this effect can be understood having in mind that an increase in excitation energy will result in a more uniform population of the isotopes corresponding to a given element thus washing out the effect of the isotopic composition of the source . in terms of eqs .
( [ eq.alpha.g ] ) and ( [ eq.beta.g ] ) , an increasing excitation energy will lead to a shift of the @xmath38 distributions toward smaller values of @xmath83 such as @xmath84 slightly diminishes together with an enhancement of the distributions variance @xmath85 .
since both numerator and denominator from eqs .
( [ eq.alpha.g ] ) and ( [ eq.beta.g ] ) act in the same sense , the decrease of @xmath5 and @xmath82 with the excitation energy of the sources is obvious .
these results are in qualitative agreement with those of ref .
@xcite .
an important problem that manifests whenever we need to infer the break - up information from the available experimental multifragmentation data is the reconstruction of primary observables affected by the secondary particle emission .
this problem may be even more important when one operates with very sensitive quantities like the slope parameters @xmath5 and @xmath6 .
it was stated in the literature that when the two considered sources have almost identical parameters excepting the isospin , the effects of sequential evaporation cancel one another leading to small modifications of the break - up stage results @xcite . on the other hand ,
estimations performed with accurate models like smm @xcite indicate that modifications of the slope parameters due to evaporation may manifest with different magnitudes depending on how sequential evaporation is implemented @xcite . to check the evaporation effect on isoscaling within the present model in fig .
[ fig : ab_ex ] are plotted @xmath5 and @xmath6 in the break - up and asymptotic stages of the decay for the equilibrated sources ( 200 , 80 ) and ( 185 , 80 ) , with @xmath79 as a function of the excitation energy . as displayed by the figure , over the entire energy range the absolute values of @xmath5 and @xmath6 are larger in the asymptotic stage of the decay in comparison to the break - up stage . to get an insight on what happens during the sequential particle emission from primary excited fragments is necessary to focus on the evaporation process .
given the fact that for all excitation energies neutrons are emitted with the most important probability , the final cold fragments are more symmetric than their break - up ancestors .
this means that the isotopic distribution of a given @xmath32 shifts toward smaller values of @xmath86 .
since the amplitude of this shift is larger for the neutron richer source , it means that the value @xmath87 in the asymptotic stage is smaller than the one corresponding to the break - up stage .
this apparent trend of isoscaling parameters to decrease after particle evaporation is annihilated by a strong narrowing of isotopic distributions ( the more asymmetric nuclei have small survival probabilities after secondary decays ) such as in the end the values of @xmath5 and @xmath82 increase with respect to their break - up values . this result is counterintuitive because at a first glance one could expect secondary decays to wash out the isotopic differences of the sources . from fig .
[ fig : ab_ex ] is obvious that secondary decays affect more @xmath5 than @xmath6 .
this result can be understood taking into account that the decrease of the width of @xmath37 distributions caused by neutron evaporation is more important than the narrowing of @xmath35 distributions caused by less important charged particle emission .
one can see from fig .
[ fig : ab_ex ] that the difference between the asymptotic and break - up values increases up to 5 mev/ nucleon excitation energy and then stays roughly constant , but taking into account that the increase in excitation energy results in smaller slope parameters , this means that the relative influence of the secondary decays on @xmath5 and @xmath6 grows with energy . to illustrate the relative modifications brought by
the secondary decays to the slope parameters in fig .
[ fig : rap_ab_ex ] is plotted the deviation between asymptotic and primary values relative to the break - up results versus excitation energy .
the increasing rate is roughly constant for whole considered domain . to identify a possible freeze - out volume effect on the slope parameters ,
isoscaling was analyzed for the previously mentioned pair of sources ( ( 200 , 80 ) and ( 185 , 80 ) ) at excitation energies increasing from 2 to 15 mev / nucleon at @xmath88 , @xmath79 , @xmath89 in both primary and asymptotic stages of the decay . despite the fact that all isotopic distributions present a strong dependence on the volume of the excited source , for the considered cases no significant freeze - out volume dependence of @xmath5 and @xmath6 was identified .
the impossibility to get an analytic expression of isotopic yields within a microcanonical model makes very hard to estimate the effect of freeze - out volume ( or any other observable characterizing the state of the source ) on isotopic ratios .
however , if true , the independence of the slope parameters on the freeze - out volume is of particular importance for experiments aiming to extract the symmetry term of the binding energy from isotopic yields .
indeed , no matter is the method for selecting collisions , is hard to believe that the ensemble of equilibrated nuclear systems formed in heavy ion reactions correspond to one value of the freeze - out volume , but rather to a distribution . since there is no method to select fragmentation events according to their volume , volume dependent @xmath5 and @xmath6 would make impossible the extraction of the asymmetry term . ref .
@xcite makes use of a canonical version of smm @xcite and reaches the conclusion that for all considered temperatures ( @xmath23=4 , 5 , 6 mev ) @xmath5 increases monotonically with the freeze - out density : by increasing @xmath90 from 0.1 to 1 @xmath5 increases by a factor of two .
the opposite conclusions of ref . @xcite and the present study are produced by the in principle different basic hypothesis of each model . taking into account the implications of the volume dependence of isoscaling parameters ,
a definite answer to this problem would be of much interest .
all the above discussion suggests a strong dependence of the fragment production on the parameters characterizing the equilibrated state of the sources .
a natural question to rise at this point is how sensitive are the slope parameters on the considered sources once one keeps fixed the atomic number , the freeze - out volume and the excitation energy . to answer this question @xmath5 and @xmath6
have been calculated for different pairs of sources ( chosen from the set ( 180,80 ) , ( 185,80 ) , ( 190,80 ) and ( 200,80 ) ) at @xmath91 mev / nucleon and @xmath88 .
the result we obtained is that the slope parameters are not dictated by the difference of isospin asymmetry between the two sources , but by the difference between the binding energy of the sources . to illustrate this result we plot in fig .
[ fig : ab_b ] the monotonic dependence of @xmath5 and @xmath6 as a function of the difference between the binding energies of the equilibrated sources .
this result is in qualitative agreement with some experimental data .
@xcite reports a modification of the absolute values of @xmath5 and @xmath6 by a factor of two when calculated using ( @xmath20sn+@xmath9sn/ @xmath20sn+@xmath20sn ) nuclear reactions instead of ( @xmath9sn+@xmath9sn/ @xmath20sn+@xmath20sn ) at 50 mev / nucleon .
similar results were obtained recently studying isoscaling in ( @xmath22fe+@xmath22ni/ @xmath22ni+@xmath22ni ) and ( @xmath22fe+@xmath22fe/ @xmath22ni+@xmath22ni ) reactions at 30 , 40 and 47 mev / nucleon @xcite .
contrary to the cases of a grandcanonical model or a sequential particle emission model where one can easily express isotopic multiplicities as a function of the binding energy of the parent nucleus , within a microcanonical model the effect of binding energy can be estimated only in an intuitive way .
thus , more similar are the two considered equilibrated sources from the point of view of their binding energies , closer are the values of the total available energy of each system for a given value of the excitation energy .
this will lead to similar populations of the configuration space of each system . in this case , all observables ( including isotopic yields ) are expected to have close values .
close values of isotopic yields will obviously produce small values of @xmath5 and @xmath6 .
the only parameter characterizing the equilibrated source which was kept fixed thorough the above discussion is the source s size . to check the size effect of isoscaling parameters we choose for fragmenting sources two pairs of smaller nuclei , namely ( ( 112 , 50 ) , ( 124 , 50 ) ) and ( ( 112 , 50 ) , ( 119 , 50 ) ) with @xmath91 mev / nucleon and @xmath88 .
the values of the slope parameters obtained in this case ( ( ( 112 , 50 ) , ( 124 , 50 ) ) : @xmath92=0.37 , @xmath93=0.50 , @xmath94=-0.49 and @xmath95=-0.55 ; ( ( 112 , 50 ) , ( 119 , 50 ) ) : @xmath92=0.23 , @xmath93=0.32 , @xmath94=-0.30 and @xmath95=-0.34 ) are very different from the values obtained for the @xmath96 sources and suggest a strong dependence of the isoscaling parameters on source size .
this result is in contradiction with ref .
@xcite where the canonical version of smm @xcite predicted @xmath5 independent of the size of the sources .
the discrepancy between the cited canonical results and the present microcanonical results focuses the attention on the importance of using appropriate models for describing finite systems .
taking into account that isoscaling has been argued on grandcanonical basis and above presented microcanonical results suggest important deviations from grandcanonical expectations , it would be interesting to check whether isoscaling is valid also for smaller sources . the results obtained for ( ( 40 , 20 ) and ( 50 , 20 ) ) at @xmath91 mev / nucleon and @xmath79 proved a linear behavior of the yield ratios with the isotopic composition of the emitted fragment together with a monotonic increase of the absolute value of the slope parameters with the size of the emitted fragment as displayed in fig .
[ fig : z=20 ] where we plot the results obtained for fragments with @xmath97 and @xmath98 .
this means that for small sources we obtain isoscaling in the sense of eqs .
( [ eq.alpha.g ] ) and ( [ eq.beta.g ] ) without the constancy of the slope parameters with the size of the cluster .
this result makes us go back to large fragmenting sources in order to evidence the dependence of @xmath5 and @xmath6 on the size of the emitted cluster . from our analysis results that by considering large emitted fragments isoscaling is obtained with slope parameters dependent on the size of the cluster . to illustrate this result , in fig .
[ fig : fit_z=80 ] we plot @xmath99 for ( ( 200 , 80 ) and ( 185 , 80 ) ) at @xmath79 and @xmath91 mev / nucleon .
one can see that while for @xmath100 @xmath5 is constant , for larger emitted fragments @xmath5 increases monotonically with @xmath32 .
it is interesting to mention that isoscaling slope parameters dependent on the emitted cluster size have been evidenced also in the framework of a dynamical stochastic mean field model @xcite . because present detectors are not able to isotopically separate fragments with charge larger than 8 , the only case when one may hope to evidence this effect corresponds to light fragments emitted by light sources .
the fragment dependence of the slope parameters can be interpreted as a finite size effect which , as expected , enhances with the diminish of the source and with the increase of the emitted fragment .
to summarize , the present paper aims to investigate isoscaling in nuclear multifragmentation . in order to offer a complete and as accurate as possible image on the subject
, the study was done using a microcanonical multifragmentation model and covered a large range of source sizes ( @xmath31=40 - 200 ) , excitation energies ( @xmath33= 2 - 15 mev / nucleon ) and freeze - out volumes ( @xmath34= 3@xmath55 - 10@xmath55 ) .
primary and asymptotic stages of the decay have been considered .
the results indicate that despite in all considered cases the logarithm of the isotopic yields corresponding to the two different sources have a linear behavior as a function of @xmath32 and @xmath36 , the slope parameters are strongly dependent on most observables characterizing the state of the sources .
the only observable not affecting @xmath5 and @xmath6 is the freeze - out volume .
this result could be of particular importance for experiments aiming to extract the asymmetry term of the binding energy using multifragmentation data , since most probably equilibrated excited systems formed in heavy ion collisions have a distribution of volumes and the event separation as a function of volume is not by far possible .
the isoscaling slope parameters are affected also by sequential particle emission .
deviations from strict isoscaling have been observed for heavy clusters originating from large multifragmenting systems and light clusters emitted by light systems and can be interpreted as finite size effects .
such deviations from grandcanonical predictions are expected to manifest especially for small systems and stress the importance of using approximations suitable to the physical situation .
since the microcanonical ensemble is by principle the most appropriate statistical tool to describe the disintegration of small isolated systems and microcanonical multifragmentation models in general proved to be able to describe with accuracy experimental data , the most reliable description of isoscaling is expected to be obtained with this kind of models .
the accuracy of isoscaling parameters is essential for obtaining correct information on fundamental quantities like the asymmetry term of the binding energy .
h. s. xu _ et al . _ ,
lett . * 85 * , 716 ( 2000 ) .
m. b. tsang , w. a. friedman , c. k. gelbke , w. g. lynch , g. verde and h. s. xu , phys .
c * 64 * , 041603(r ) ( 2001 ) . j. brzychczyk et al . , phys . rev .
c * 47 * , 1553 ( 1993 ) .
v. volkov , phys .
rep . * 44 * , 93 ( 1978 ) .
g. a. souliotis et al . , phys .
c * 68 * , 024605 ( 2003 ) .
m. veselsky , g. a. souliotis and m. jandel , phy .
c * 69 * , 044607 ( 2004 ) .
m. b. tsang , w. a. friedman , c. k. gelbke , w. g. lynch , g. verde and h. s. xu , phys .
86 * , 5023 ( 2001 ) .
a. s. botvina , o. v. lozhkin and w. trautmann , phys .
c 65 , 044610 ( 2002 ) .
e. geraci _
_ , nucl . phys . *
a732 * , 173 ( 2004 ) .
d. v. shetty , s. j. yennello , a. s. botvina , g. a. souliotis , m. jandel , e. bell , a. keksis , s. soisson , b. stein and j. iglio , nucl - ex/0406008 . m. b. tsang _
et al . _ ,
phy . rev .
c * 64 * , 054615 ( 2001 ) .
a. ono , p. danielewicz , w. a. friedman , w. g. lynch and m. b. tsang , phys .
c * 68 * , 051601 ( 2003 ) .
t. x. liu _
et al . _ ,
phys . rev .
c * 69 * , 014603 ( 2004 ) .
j. p. bondorf , a. s. botvina , a. s. iljinov , i. n. mishustin and k. sneppen , phys . rep . * 257 * , 133 ( 1995 ) .
d. h. e. gross , rep .
* 53 * , 605 ( 1990 ) .
al . h. raduta and ad .
r. raduta , phys . rev .
c * 65 * , 054610 ( 2002 ) .
s. albergo , s. costa , e. costanzo and a. rubbino , nuovo cimento soc .
fis . a * 89 * , 1 ( 1985 ) . a. majumder and s. das gupta , phys
c * 61 * , 034603 ( 2000 ) .
w. a. friedman , phys .
c * 42 * , 667 ( 1990 ) .
al . h. raduta and ad .
r. raduta , phys .
c * 55 * , 1344 ( 1997 ) .
g. fai and j. randrup , nucl
a381 * , 557 ( 1982 ) .
g. audi and a. h. wapstra , nucl . phys . *
a595 * , 409 ( 1995 ) .
a. s. iljinov _
et al . _ , nucl . phys . *
a543 * , 517 ( 1992 ) .
s. r. souza _ et al .
c * 67 * , 051602(r ) ( 2003 ) .
w. p. tan _
et al . _ ,
c * 68 * , 034609 ( 2003 ) . | the exponential scaling of isotopic yields is investigated for sources of different sizes over a broad range of excitation energies and freeze - out volumes , in both primary and asymptotic stages of the decay in the framework of a microcanonical multifragmentation model .
it was found that the scaling parameters have a strong dependence on the considered pair of equilibrated sources and excitation energy and are affected by the secondary particle emission of the break - up fragments .
no significant influence of the freeze - out volume on the considered isotopic ratios has been observed .
deviations of microcanonical results from grandcanonical expectations are discussed . |
can star capture account for the growth of the largest supermassive black holes in the universe ?
could this process account for the growth of abundant smaller supermassive black holes , or intermediate mass seeds ?
the answer to this question appears to be yes , as we argue in this paper .
the central few parsecs of most standard ( @xmath4 ) galaxies contain either supermassive black holes ( smbhs ) , nuclear star clusters ( nscs ) , or both of these components . in some cases
the nscs are distinct entities and in other cases they appear to be a continuation of the inner spheroidal component . in either case ,
nscs typically represent regions of high stellar density , often with a steep , cusp - like profile .
it is well known that , for the more massive galaxies ( @xmath5 ) , the mass of the bh component is roughly linearly related @xcite to the mass in the spheroidal stellar component of the galaxy , but lower mass systems can also contain smbhs with masses below the linear relation , and for very low mass galaxies the smbhs are apparently absent in general , with notable exceptions @xcite .
however , for galaxies with masses below this transition ( @xmath6 ) , the central regions typically contain nscs having masses roughly proportional to the spheroidal stellar mass , with values similar to the smbh mass at the transition point , and with the nscs being less prominent than the smbhs above the transition .
nscs are typically absent in the largest galaxies , and only coexist with smbhs in galaxies from an intermediate mass range @xcite .
[ fig : nscdichotomy2 ] illustrates this dichotomy with observational data from nearby galactic nuclei , plotted against the velocity dispersion of the host galaxy . an overly simple explanation for this behavior would postulate that all standard galaxies start out with nuclear star clusters proportional to the spheroidal mass , but that above some critical mass ( or escape velocity ) , dynamical processes transform a larger and larger fraction of the stars in the nsc to massive seed central bhs .
that would explain in a simple fashion the continuity observed between centrally located nscs and central bhs .
the purpose of this paper is to outline the dynamical processes that could make this cartoon scenario a plausible physical process .
we focus in particular on tidal capture and disruption of stars as a runaway growth channel for stellar mass black holes ( sbhs ) .
the origin of nscs is not the primary topic of this paper , but the hypothesis that they form from the infall of globular clusters ( gcs ) due to dynamical friction @xcite is consistent with the dynamical processes that we will be treating in this paper , and is generally capable of growing nscs to @xmath7 in less than one billion years .
an alternative formation channel is _ in situ _ growth through star formation @xcite .
it is not clear which of these two mechanisms dominates nsc formation , and the data is often consistent with contributions from both channels @xcite .
however , it is clear that _ in situ _ star formation continues at some level in most nuclear star clusters ago ( * ? ? ? * and references therein ) . ] , and therefore provides a natural source of sbhs in nscs . the primary physical mechanism that we will consider , tidal capture ,
is well known .
one star can capture another into a bound orbit if a close encounter between the two puts more energy into tidal perturbations than the positive relative energy of these two stars at large separation .
the tidal capture mechanism was first pointed out by @xcite , with an early computation by @xcite and a more refined modal analysis given by @xcite .
later work analyzed subsequent orbital evolution both prior @xcite and subsequent @xcite to the onset of internal dissipation .
is visible , below which nscs dominate and above which smbhs dominate . in low @xmath8 galaxies , nsc
masses appear to follow a power law that turns over for @xmath9 ; this power law is resumed in larger galaxies by the smbh masses , which fall off sharply for @xmath10 .
smbh dynamical mass measurements and nsc mass data are taken from @xcite , and two local galaxies of interest ( the milky way and m31 ) are marked with circles ; these datasets give one smbh mass ( black dots ) for every nsc mass .
we also show a different set of smbh masses estimated with maser disk measurements @xcite ; for these galaxies associated nscs are unconstrained and the smbh masses are shown with green dots .
the green line shows the `` saturation mass '' predicted for smbhs formed through the mechanism in this paper , described in more detail in eq .
[ eq : msat ] .
growth of smbhs above the green line occurs through standard forms of gas accretion @xcite .
the black arrow at @xmath11 indicates the rough transition below which nscs are insufficiently dense to produce runaway sbh growth.,width=321 ] if sbhs are retained , or are subsequently formed , in dense stellar systems such as gcs or nscs , then mass segregation will bring them to the central , densest part of the system , whereupon they will be well situated to tidally capture the much more abundant low mass , normal stars . the dramatic influence that the resulting tidal capture binaries can have on the dynamical evolution of such clusters has been studied extensively @xcite .
but the specific possibility that a runaway of successive tidal captures would lead to the formation of a massive black hole , first proposed by @xcite , has not been quantified in detail .
we will provide some first calculations of the circumstances required for such a runaway to occur in this paper , as well as the expected endpoint of the runaway , returning to a more detailed calculation in subsequent work .
the preliminary analytic and numerical treatment presented here indicates that the physical requirements for runaway tidal captures are first satisfied for galactic nuclei near the nsc / smbh transition at black hole masses of @xmath12 and stellar velocity dispersions @xmath13 .
supermassive black hole ( smbh ) seed formation is generally studied in the context of the high redshift universe , and three leading candidate scenarios currently exist ( see * ? ? ?
* ; * ? ? ?
* for general reviews ) : stellar remnant bhs left over following the deaths of pop iii stars , the direct collapse of gas in small halos , and runaway stellar collisions in dense star clusters .
the pop iii scenario has the advantages of concreteness and ubiquity , but has been cast into doubt by recent simulations that find pop iii stars may be much less massive than was previously thought due to fragmentation during their formation @xcite . even if simpler estimations for the masses of the first stars ( @xmath0 ) are correct @xcite , this scenario produces the lowest mass smbh seeds .
direct collapse of gas in early halos has the advantage of producing much larger seeds , @xmath12 , but will be strongly suppressed by small amounts of coolants , primarily molecular hydrogen @xcite .
the mechanism in this paper is a specific example of the third channel for smbh seed formation : runaway collisions of stellar mass objects in dense stellar environments @xcite , such as gcs or nscs .
past studies of this channel predicted smbh seeds of widely varying mass , depending on which variant of the stellar runaway proceeds : @xmath14 .
we will focus primarily on a relatively unexplored variant of the runaway collision scenario , specifically the process of runaway tidal capture and the resulting growth of a massive object , although we comment briefly on more traditional versions of this scenario as well .
our work focuses on tidal capture because that process has the largest cross - section of any of the dynamical mechanisms we consider .
it is therefore vital in initiating the runaway growth of black holes that start with roughly @xmath15 sbh seeds .
but , as we will note later , such processes may not ultimately dominate the total mass growth of our black holes ; and consideration of conventional tidal destruction events as initially envisaged by @xcite under the same circumstances would also grow sbhs to comparable sizes provided a runaway can begin . in 2
we discuss the different star capture channels available to grow a sbh in a dense star cluster , and summarize existing observations of nscs , which appear to provide the most favorable environments for runaway sbh growth . in
3 we analytically explore the collisional growth of sbhs in nscs . in
4 we use idealized numerical studies to validate our analytic estimates and to test the importance of sbh multiplicity . in 5
we examine the slowing and eventual termination of sbh growth through star capture . in 6 we discuss future observational tests of our model for massive black hole growth , and in 7 we offer conclusions .
to set the stage for our discussion it is useful to summarize the properties of the dense ( primarily nuclear ) star clusters that we will be considering , and then to outline the primary physical processes occurring in these systems .
these clusters can , with a great degree of simplification , be characterized by three parameters : mass @xmath16 , core radius @xmath17 , and half mass radius @xmath18 .
we show distributions of some of these quantities in actual nscs ( taken from @xcite ) in fig .
[ fig : nscdata ] , along with mean density @xmath19 , mean relaxation time @xmath20 , and average velocity dispersion @xmath21 . here
we take the coulomb logarithm @xmath22 and assume mean stellar mass @xmath23 for simplicity .
we provide more detail on these data sets , as well as best fit relations and variances between them , in appendix [ app : data ] .
although all of these quantities have significant variance from their means , we see that nscs typically have @xmath24 .
other work has established that this is generally correlated with , and comparable to , the velocity dispersion of the host galaxy @xcite .
the characteristic sizes of nscs are typically @xmath25 , giving a wide range in @xmath26 .
relaxation times at the half - mass radius @xmath27 are under a hubble time for @xmath28 of our sample . in reality , nscs are not simple two - parameter systems , and in particular exhibit a wide range of concentration parameters @xmath29 , where @xmath17 is the core radius of a roughly isothermal system , and @xmath30 @xcite , implying central densities @xmath31 that are a few orders of magnitude larger than @xmath26 .
we plot the distributions of `` cluster - central '' quantities in fig .
[ fig : nscdata2 ] .
we note here that the definition of an nsc is not a completely rigorous one , and whether or not a galaxy s central surface brightness profile contains an nsc can be a subjective question .
the data quality of the innermost observed isophotes , the number of components used in fitting the surface brightness profile , and assumptions about the radial dependence ( or lack thereof ) of a mass - to - light ratio are all important ingredients in determining whether there is an inner light excess that can be identified as an nsc .
to some extent , however , this debate is a semantic one from the perspective of our calculations , which depend primarily on two measurable quantities : stellar density and stellar velocity dispersion .
whether or not the center of a galaxy ( on @xmath32 scales ) is formally an nsc or is merely an inward continuation of a larger - scale surface brightness power law will not affect our results , so long as it has the same density and velocity dispersion in both scenarios .
if a nsc lacks a massive black hole , how can it acquire one ? in the remainder of this section we consider two different `` collisional runaway '' channels .
the first is the classical prompt collisional runaway that is sometimes theorized to occur in denser star clusters at high redshift ; we summarize past literature on this channel and explain why it is unlikely to occur for the nscs we observe at low redshift .
the second is a relatively unexplored delayed collisional runaway channel , which forms the basis for the remainder of this paper .
after the formation of a star cluster , its constituents will evolve toward energy equipartition : heavy cluster members , such as massive young stars or sbhs , will lose orbital energy to lighter members , and will mass segregate to the center of the cluster .
recent n - body simulations indicate that star clusters may never reach a state of full energy equipartition , but nonetheless partial mass segregation does occur @xcite .
if we consider a two component cluster made up of light masses @xmath33 and heavy masses @xmath34 , this occurs on a timescale @xmath35 , where @xmath36 is the energy relaxation time .
extremely dense star clusters can undergo a prompt collisional runaway among massive young stars if the mean time between collisions , @xmath37 , becomes significantly less than the main sequence lifetime of very massive stars , @xmath38 .
if this condition is satisfied , runaway collisions between main sequence stars will form a supermassive star of mass @xmath39 @xcite , which will eventually collapse to an intermediate mass black hole ( imbh ) due to the onset of gr instability ( e.g. * ? ? ?
the approximate central number densities required for formation of a supermassive star in this type of prompt collisional runaway are @xmath40 , roughly two orders of magnitude greater than the highest seen in fig .
[ fig : nscdata ] .
clusters could form at this density at high redshift due to dissipation in gas flows @xcite , or alternatively could reach central densities of this order via spitzer instability @xcite or a broader core collapse . however , in the low redshift universe , this process is derailed by the nonzero metallicity of stellar participants .
collision products will shed most of their mass through line - driven winds , thus forming stellar - mass compact remnants in supernova explosions ; even very low - metallicity progenitors of modest size ( @xmath41 ; @xmath42 ) lose enough mass via winds to prevent imbh formation @xcite .
let us assume that there is no prompt collisional runaway ; instead , massive stars segregate to the center of the star cluster and die in supernovae there , leaving behind a smaller population of compact object remnants ( most neutron stars and likely some sbhs will escape in natal kicks ) .
these sbhs will swap into primordial binaries in binary - single encounters , and the larger interaction cross - sections of these binaries will eventually lead to ejections of many black holes . as this process repeats ,
the number of sbhs declines until either @xmath43 remain , or the sbh - sbh interaction time has become very long ( this competition between sbhs is analyzed through n - body integrations in [ sec : numerics ] ) . even if our nsc is unlucky and loses all its sbhs in this process , the growth of nscs over time ( either through _ in situ _ or cluster accretion processes ) implies that it will eventually regain one .
several different channels exist for the growth of an sbh in a gas - free star cluster , which we describe in quantitative detail in the appendices of this paper .
the four growth channels we consider are the following , listed in order of increasing cross - section and probability : 1 . gravitational wave capture of compact objects ( appendix [ sec : gw ] ) : two compact objects on hyperbolic orbits can capture into a bound orbit through emission of gws .
2 . direct collision with main sequence stars : a sbh can directly collide with a main sequence star if the pericenter of their encounter , @xmath44 , is less than the stellar radius @xmath45 .
the end product of such a collision is highly uncertain .
tidal disruption of main sequence stars ( appendix [ sec : tde ] ) : a sbh will tidally disrupt a main sequence star of mass @xmath46 if @xmath47 , where we have defined the tidal radius @xmath48 .
up to half the stellar mass can accrete onto the sbh .
4 . tidal capture of main sequence stars ( appendix [ sec : tc ] ) : a sbh will tidally capture an unbound star into a bound orbit ( by dumping excess orbital energy into tidally - excited oscillation modes ) if @xmath49 , where @xmath50 is a function of stellar structure and cluster @xmath2 .
depending on subsequent orbital evolution of this tidal capture ( tc ) binary , the star may be consumed by the sbh .
collisional cross - sections are often enhanced by gravitational focusing of orbits , so that the impact parameter @xmath51 for a close - approach distance of @xmath44 is given by @xmath52 where @xmath53 and @xmath54 are the masses of the impactors .
generic encounter rates can be calculated at leading order as @xmath55 , where @xmath56 is the density of targets , @xmath57 is the relative velocity between sbhs and their targets ( typically @xmath58 ) , and @xmath59 . of these processes ,
gw capture of compact objects has the smallest cross - section by far , and will generally be disregarded in this paper as a growth mechanism for sbhs ( see appendix [ sec : gw ] for a more detailed discussion ) . in
the gravitationally focused limit , the rate of growth due to direct collisions is at most @xmath60 , but we neglect this channel both because it is unclear whether significant stellar mass will accrete , and because the exponential growth that is theoretically possible from direct collisions is slower than the super - exponential growth allowed by tidal processes .
both tidal capture and tidal disruption scale as @xmath61 , a curve of super - exponential growth that formally goes to infinite mass in finite time . because the tidal capture process has the largest cross - section , tidal capture runaways are the most promising process for growing an imbh in nsc - like clusters that lack one .
] , and we explore the onset and development of these runaways in the next section .
in this section , we summarize our theoretical models for realistic nscs , and then apply simple analytic models for collisional growth of sbhs in these environments .
although in principle tidal capture ( or other types of ) runaways can happen in open or globular clusters as well , runaway sbh growth is ultimately density - limited , and we therefore focus on the densest observed stellar systems in the universe : nscs .
a worthwhile caveat is that more dense star clusters may have existed at high redshift , with central subclusters driven to even higher densities through gas dissipational processes @xcite .
we use the three parameter potential - density pair described in @xcite to approximate the structure of all of these star clusters . specifically , the stellar density profile of a cluster with central density @xmath62 , core radius @xmath17 , and halo radius @xmath18 ( equivalent to the half - mass radius when @xmath63 ) is @xmath64 and the gravitational potential @xmath65,\end{aligned}\ ] ] where the total cluster mass @xmath66 .
this potential - density pair is designed as an analytically tractable alternative to single - mass king models @xcite ; it is appropriate for use here because observed nscs are reasonably well fit by king models , with varying degrees of concentration @xmath67 .
@xcite find that over half ( @xmath68 ) of their large nsc sample is best fit by highly concentrated king models , with @xmath69 . when @xmath63 , the central 1d velocity dispersion is @xmath70 an approximation we shall use for the remainder of this paper ( when @xmath71 , the velocity dispersion is slightly higher than this value ) .
finally , the central relaxation time is @xmath72 at distances @xmath73 , @xmath74 .
the plausibility of collisional runaways in star clusters of fixed mass @xmath16 and halo ( roughly speaking , half mass ) radius @xmath18 depends critically on the size of the core , as @xmath75 .
clusters lacking central heat sources will , over @xmath76 central relaxation times ( * ? ? ?
* under the assumption of equal - mass stars in the cluster ) , enter a state of core collapse , as heat is slowly conducted out of the core , and @xmath17 approaches zero . in practical terms , the presence of primordial binaries in the core can arrest this collapse ( scattering of single stars off a hard binary will harden the binary further and heat the core , offsetting conductive heat losses ) . @xcite
have argued that star clusters that possess @xmath77 can avoid this problem by `` burning through '' their entire population of primordial binaries . to better quantify this argument ( and the possible existence of a @xmath78 threshold for strong susceptibility to core collapse )
, we consider an equal mass binary with total mass @xmath79 in the core of a cluster with mean stellar mass @xmath80 .
if this binary has a binding energy greater in magnitude than @xmath81 , the outcome of many cumulative three - body interactions will be to statistically harden it @xcite , heating the cluster and supporting it against core collapse .
the rate at which the binary loses its binding energy @xmath82 to the cluster core is @xcite @xmath83 here the cluster central number density @xmath84 .
notably , this hardening rate is independent of binary semimajor axis @xmath85 .
slight gradients in the cluster temperature profile will conduct excess heat from the core to the outskirts at a rate @xmath86 , so if the cluster core contains @xmath87 binaries at any point in time , it will adjust to a steady - state size where heating balances conductive cooling and @xmath88 the dimensionless conductivity constant @xmath89 for single - mass clusters @xcite . using this approximation and taking @xmath90 , we find a central cluster density in steady state of @xmath91 this formula represents an upper limit on the central density of any collisional stellar system , and it decreases rapidly with the mass of central binaries ( note , however , that it assumes an equal - mass binary ) .
the burn time for an individual binary to reach its minimum separation @xmath92 in such a cluster core is @xmath93 assuming a fraction @xmath94 of all cluster stars are born into primordial binaries , we find a total burn time of @xmath95 for a cluster to burn every one of its primordial binaries ; in the last equation we set @xmath96 for simplicity .
we note that our choice of @xmath97 is likely quite conservative , as primordial binaries will tend to eject in strong three - body encounters once they reach a separation where their orbital speed is comparable to @xmath2 .
therefore we expect most nscs to be able to burn through their primordial binaries in a hubble time , as argued by @xcite , provided they can reach the core on shorter timescales . in reality , however , binaries are continuously transported to the core by the slow process of dynamical friction in an extended nsc . assuming that the orbital speed for binaries is at most mildly supersonic ,
we can write an approximate dynamical friction timescale for a binary on a circular orbit : @xmath98 where we use the 1d velocity dispersion , take @xmath99 , and define an average cluster density @xmath100 .
we therefore conclude that only relatively dense star clusters , with @xmath101 , have both sufficient energy to burn through their primordial binaries and the ability to do this in a hubble time .
another complication to the simple cluster model presented above is the possible presence of a `` cavity '' in the center of the cluster , excavated by the scattering interactions of main sequence stars with a dense subcluster of compact objects .
we can estimate the size of this cavity by calculating how long it takes for a star to have a sufficiently close ( focused ) encounter with a sbh such that it receives a kick velocity greater than the core escape velocity @xmath102 here we assume that we are looking at older populations of spitzer - stable sbhs ; this analysis does not apply at very early times in the cluster s history when a spitzer - unstable sbh subcluster may exist . by assumption , the @xmath103 sbhs are in energy equipartition , with @xmath104 ; this implies that they inhabit a characteristic radius @xmath105 .
a star - sbh scattering event will likely result in the star being ejected from the cluster core if the closest approach is within @xmath106 . by
taking the gravitationally focused limit and the number density of sbhs , we can compute a per - orbit probability for a star to be ejected , @xmath107 .
the ejection timescale @xmath108 is , for stars within the cluster core , @xmath109 in fig .
[ fig : cavity ] , we make a contour plot of the ratio of @xmath110 .
when this ratio is less than unity , a cavity can be evacuated in the central regions of the cluster core , but when it is greater than unity , stars will diffuse inward to refill the core faster than they are ejected . interestingly , this ratio does not depend on @xmath17 or @xmath18 independently , but only on @xmath111 . in general , we find that cavities only exist for small and highly concentrated clusters . however , such clusters can not exist in a steady state so long as at least one binary exists as a heat source in the cluster core , as is shown in fig .
[ fig : cavity ] .
furthermore , it is actually large and highly concentrated clusters that are most relevant for tidal capture runaways , so we neglect the effect of a cavity for the following discussion ( but see [ sec : outcomes2 ] for how a different type of cavity can form at late times ) .
are labeled and shown as black solid lines .
here we assume that there are @xmath112 sbhs all of mass @xmath113 ; even with this unrealistically large late - time sbh population , generally @xmath114 ( i.e. a central cavity in the stellar profile does not form ) .
we also plot contours of equilibrium @xmath111 for a cluster of a given mass heated by @xmath87 binaries of mass @xmath115 ( eq . [ eq : rceq ] ) .
these are shown as dashed red lines with associated labels , and the shaded red region is a portion of parameter space forbidden provided there is at least one binary in the cluster core .
cavity formation is disfavored.,width=321 ] in conclusion , we expect high central densities to be achievable through core collapse in many large nscs ( @xmath77 ) , although binary burning may persist for greater than a hubble time unless @xmath116 . to achieve the high central densities required for runaway sbh growth ,
we therefore require high @xmath2 and high ( mean ) density clusters , although our requirements are satisfied by a large portion of the parameter space in fig . [
fig : nscdata ] . in the following subsection
, we quantify better the central density threshold required for the onset of runaway growth . of the processes described in
2 to grow stars or compact objects ( gravitational wave capture , direct collision , tidal disruption , and tidal capture ) , tidal capture and tidal disruption are in general the most promising .
growth of bhs through gw capture is only possible in extremely deep potential wells where the loss of merger products due to gw recoil can be mitigated @xcite .
the large cross - sections for tidal processes give super - exponential growth ( @xmath61 ) more conducive to a runaway than the merely exponential growth provided by direct collisions ( @xmath60 ) , and without the sizable uncertainty involved in the outcome of a direct collision .
we therefore examine the rates of tidal capture and tidal disruption relative to other processes that could sabotage the growth of an smbh seed .
these processes themselves have relative rates given by @xmath117 ; in appendix [ sec : tc ] we argue that @xmath50 , but with a weak @xmath2-dependence . for the remainder of this subsection we focus on the rate of tidal captures , which we shorthand as @xmath118 for notational clarity , but removal of factors of @xmath119 enable easy comparison to tidal disruption rates .
we also assume here that a tidally captured star is consumed whole by the sbh ; this assumption is explored in much greater detail in [ sec : outcomes ] .
initially sbhs can grow at a linear rate through tidal captures , reaching a super - exponential pace after they have roughly doubled in size .
if we assume that typical sbhs have @xmath120 and mean stellar masses in an evolved cluster are @xmath121 , we require @xmath122 tidal capture events to reach runaway growth , which corresponds to a tidal capture rate @xmath123 if we wish to reach the runaway regime in a hubble time @xmath124 .
two processes competing with tidal capture for the affection of solitary black holes are binary - single interactions and gw capture .
the former is most relevant in low-@xmath2 clusters , and the latter in the ( possibly short - lived ) sbh subclusters produced by mass segregation . in the gravitationally focused limit relevant for us , the rate of tidal captures by a single sbh is @xmath125 here we have used for convenience the normalized variables @xmath126 , @xmath127 , @xmath128 , @xmath129 , @xmath130 , and @xmath131
. single sbhs will also interact with primordial binaries of semimajor axis @xmath85 , which can result in capture into the binary or the ejection of some members of this process from the cluster .
we assume these strong interactions occur when pericenter @xmath132 , and that the logarithmically flat field distribution @xcite of primordial binary semimajor axes @xmath133 for @xmath134 is truncated within the cluster so that @xmath135 .
integrating the bh - binary interaction rate over all @xmath85 then gives the total bh - binary interaction rate , @xmath136 , which can be expressed fractionally as @xmath137 in the approximate equality above , we have assumed @xmath138 , @xmath139 , and have defined @xmath140 .
we plot the relative rates of these two processes across the parameter space of star clusters in fig .
[ fig : relativerates ] , by combining eqs . and with expressions for the central density ( @xmath141 ) and velocity dispersion ( @xmath142 ) of realistic clusters .
we see that in order to achieve the conditions necessary for a tidal capture runaway ( in less than a hubble time ) , it is necessary that the cluster halo ( or half - mass ) radius @xmath18 be very small , typically @xmath143 .
even when this criterion is satisfied , it is likely that binary interactions will occur with greater frequency than tidal captures , unless the cluster is also quite massive @xmath144 .
( cluster velocity dispersion ) and @xmath145 ( stellar density ) .
the green lines ( @xmath146 ) show curves of constant tidal capture rate @xmath118 , while the purple lines ( @xmath147 ) show curves of constant binary interaction rate @xmath136 .
solid lines show rates @xmath148 , dashed lines show rates @xmath149 , and dotted lines show rates @xmath150 . in all curves
we have assumed interactions with solar - type main sequence stars and sbh masses @xmath151 .
the shaded parameter space above the solid green curve is conducive to runaway growth through tidal capture , although areas above this curve and also above the solid purple curve may see this growth inhibited by more frequent binary interactions .
binary interactions are less important in the darkly shaded parameter space .
data points show the @xcite sample of nscs ; the large blue dots indicate best estimates for central cluster density , while the small black dots illustrate the more robust stellar density at the half mass radius.,width=321 ] binary - single encounters can prevent a tidal capture runaway from ever beginning , by ejecting all sbhs from the cluster ( or at least , the cluster core ) . if we imagine an encounter between a main sequence star and a sbh - star binary with semimajor axis @xmath85 , one of the stars will typically be ejected at a speed @xmath152 , and momentum conservation dictates that the surviving sbh - star system recoils with @xmath153 .
this process clearly sets a lower limit on the cluster @xmath2 capable of retaining sbhs , as the central cluster escape velocity is @xmath102 the largest binary ejection velocities come from the hardest possible sbh - star binaries , i.e. @xmath154 if @xmath155 , then it becomes impossible to retain sbhs in the cluster and tidal capture runaways are prevented .
if we take an angle - averaged distribution of ejection speeds @xmath156 @xcite , we find that encounters between sbhs and primordial binaries eject the newly bound sbh - star pair with speed @xmath157 at a rate @xmath158 clearly , frequent ejections with @xmath159 will endanger any runaway , but even smaller values of @xmath57 may abort runaway growth , as sbhs ejected to the low - density cluster halo will not grow through star capture and may take very long times to sink back to the center @xcite .
we therefore compute a `` core escape velocity '' @xmath160 , and we find that @xmath161 when @xmath162 in the above , we have assumed @xmath163 ; we see that primordial binaries are unlikely to efficiently eject sbhs from nsc cores , though they will be much more effective at this in open and globular clusters .
even though @xmath2 and @xmath164 are usually large enough for the sbh to survive ejection in binary - single encounters , these interactions can still endanger a tidal capture runaway because tides capture stars onto highly eccentric orbits .
if the sbh is part of a relatively wide ( but still hard , by cluster standards ) binary system at the time of capture , then chaotic three - body interactions can ensue , and it is likely that the captured star will quickly scatter to a larger pericenter where it no longer has the chance to undergo strong tidal interactions or circularize , as we discuss further in the next section . in a conservative sense , therefore , it is best to treat only the darkly shaded part of fig .
[ fig : relativerates ] as the part of parameter space where runaway tidal growth is favored . after
a star has been tidally captured by a bh , it will return to pericenter and will suffer further strong tidal encounters .
however , unlike on the first ( unbound ) passage , when tidal forces excite oscillatory modes in a previously quiescent star , repeated pericenter passages on a bound orbit can both excite and de - excite oscillation modes . in the absence of dissipation , the exchange of energy between modes and orbit will be either quasi - periodic and bounded ( if the pericenter is large relative to a `` chaos boundary '' produced by the overlap of mode - orbit resonances - see @xcite ) , or will undergo an unbounded random walk ( if the pericenter is small relative to the chaos boundary ) . in practice ,
tidal capture in the high-@xmath2 environment of a nsc requires a small pericenter , and so the mode amplitude will follow a chaotic random walk in the absence of dissipation , often reaching nonlinear sizes @xcite . the first passage between the sbh and an unbound star transfers the following orbital energy @xmath165 into oscillation modes , binding the star to an orbit with initial period @xmath166 .
here @xmath167 is a spherical harmonic modenumber and @xmath168 is a sum over all radial modenumbers with oscillation frequencies @xmath169 ( see appendix [ sec : tc ] for more details ) . in general , the lowest order @xmath170 mode dominates the energy budget for mechanical oscillations .
the instability of initially linear oscillations to stochastic growth can now be understood by taking the mode phase @xmath171 , and examining the per - orbit phase shift @xmath172 , where @xmath173 , and @xmath174 is the binding energy of the tidally captured orbit . if @xmath175 , as is generally the case for tc in nscs , stochastic instability and a random walk in mode amplitude will develop @xcite . however ,
real stars possess internal dissipational mechanisms , and the presence of these qualitatively alters the evolution of a tidally captured star .
observed circularization rates of tidally interacting stars imply an internal quality factor @xmath176 @xcite , much lower than the @xmath177 values predicted by linear dissipation theories .
this may be explicable by nonlinear mode - mode couplings ; once the energy in a mode exceeds a threshold amplitude , the development of parametric instability dumps its energy into a chain of daughter modes @xcite .
these nonlinear mode - mode couplings serve as a relatively efficient source of dissipation , preventing random walks in mode amplitude from proceeding far , and leading to a steady state mode energy @xcite comparable to the `` capture '' value after the first pericenter passage . because @xmath178 greatly exceeds the threshold energy for parametric instability
, orbital energy will be dissipated , and circularization will proceed on a timescale @xmath179 where the @xmath180th orbit has a period @xmath181 and we sum over @xmath182 orbits .
once a single star has circularized into a tight orbit of energy @xmath183 , dynamical tides will be replaced by a quasi - equilibrium tide and mass transfer will occur on the ( different ) dissipation timescale for equilibrium tidal friction .
we note that circularization is itself a runaway process , because the transfer of orbital energy into the star reduces the time for the next pericenter encounter . although the sum in eq .
[ eq : tcirc ] can be written explicitly as a hurwitz zeta function for @xmath184 , in practice @xmath185 provided @xmath186 .
however , successful circularization requires the star to radiate at least its own binding energy .
the ratio @xmath187 at late times when @xmath188 , and @xmath189 when @xmath190 .
if energy is dissipated into the star faster than it can be radiated , then the star will expand in response .
this process will itself rapidly run away ( as the lowest - order @xmath170 modes have @xmath191 , where @xmath192 ) , likely culminating in disruption @xcite .
the criterion for this is whether or not the kelvin - helmholtz time @xmath193 is greater than the inflation timescale @xmath194 here @xmath195 is the number of orbits required to double @xmath196 , assuming crudely that @xmath197 .
if @xmath198 , the star will disrupt on the inflation timescale , likely through a series of partial tidal disruptions that run away . in practice , @xmath199 is generally much longer than other timescales relevant for the tidal capture problem , so that inflation is a plausible outcome .
although @xmath200 , @xmath201 for all but the smallest ( @xmath202 ) sbhs , so stars tidally captured around bhs are unlikely to circularize prior to runaway inflation @xcite .
the star begins to disrupt at a characteristic semimajor axis @xmath203 in general @xmath204 .
the one caveat to this analysis is that the classical kelvin - helmholtz time may severely overestimate how long the heat requires to radiatively diffuse out of the star if it is deposited in less optically thick outer regions .
if the effective kelvin - helmholtz time can be shrunk by a factor @xmath205 , then it will shrink below @xmath206 and stars can be consumed by stable mass transfer @xcite rather than in a sequence of runaway partial disruptions . the final outcome is generally similar : the sbh gains an order unity fraction of the star s mass , although over a much longer timescale if tidal heating radiates away .
crucially , the above picture assumes a tidal capture binary evolving in isolation .
both circularization and runaway consumption can be derailed ( or expedited ) by orbital perturbations from other , unbound , stars .
such a tidally captured star can be ionized due to strong perturbations ( i.e. another star passing within the secondary s hill sphere ) after a time @xmath207 here we have taken the limit of gravitationally unfocused encounters , as the orbital apocenter @xmath208 , typically . on much shorter timescales than this
, the angular momentum of a tidally captured star can be perturbed into weaker or stronger tidal interactions with the bh .
the timescale for an order unity change in orbital angular momentum due to two - body stellar perturbations is @xmath209 here @xmath210 is the specific angular momentum of the tidally captured star , and @xmath211 is the specific angular momentum diffusion coefficient ; we take @xmath212 and @xmath213 .
importantly , stochastic changes in orbital angular momentum can halt mode excitation and prevent a captured star from inflating in radius , provided the pericenter moves outward . alternatively , a pericenter that random walks inward could see an acceleration in the inflation of its envelope .
the first of these processes is favored because of the larger available phase space away from @xmath214 in a 3d random walk , so a captured star can not be quickly consumed if @xmath215 is short .
we illustrate the relative timescales for these processes in fig .
[ fig : tctimes ] .
when @xmath164 is small , the shortest timescale is the orbital period @xmath166 , followed by @xmath206 .
when @xmath216 , thermalization of mode energy will result in repeated partial disruptions and eventually the full disruption of the star ; when @xmath217 , this can still occur , but most stars will also random walk to large , non - interacting values of angular momentum , a possibility we address momentarily . in general , the kelvin - helmholtz time is too long to affect the evolution of a tidally captured star , and while @xmath218 , the fact that @xmath201 prevents circularization and stable mass transfer from occurring .
generally , @xmath219 . , shown in blue , although for large values of @xmath164 it is the angular momentum relaxation time @xmath215 , shown in green .
the time for the star to enter a runaway inflation regime , @xmath206 ( black solid ) , and the time for orbital energy loss to circularize the orbit of the captured star , @xmath220 ( black dotted ) are each small multiples of @xmath166 . because @xmath221 , only the former process is relevant .
the binary ionization timescale @xmath222 is shown in purple , and is generally greater than @xmath215 .
the longest timescale in the problem ( not shown ) is the kelvin - helmholtz time for a captured star to radiate dissipated mode energy .
when @xmath216 ( regime i ) , captured stars are efficiently consumed via a series of runaway partial tidal disruptions . when @xmath223 ( regime ii ) , stars will quickly random walk to larger pericenters where tidal forces are initially irrelevant , but may still inflate and be disrupted through collisions with other tidally captured stars .
when @xmath224 ( regime iii ) , captured stars can not inflate significantly before being ejected or swapped out by encounters with unbound stars . in this diagram , @xmath225 , @xmath226 , and we have used a kroupa imf truncated at a maximum mass of @xmath227.,width=321 ] the preceding discussion has focused on the tidal capture of a single star by a sbh ; however , once the black hole has entered the runaway growth regime , it will begin tidally capturing stars at intervals less than @xmath206 ( especially if angular momentum relaxation delays stellar consumption ) . at this point multi - body interactions will become important .
if the bh is orbited by @xmath228 tidally captured stars that have been unable to circularize or expand through mode thermalization ( i.e. have random walked to slightly larger pericenter values ) , most will possess orbital apocenters @xmath229 , and they will map out a 3d density profile @xmath230 in general , this `` mini - cusp '' will have a much higher density than the surrounding star cluster , and will be prone to direct physical collisions between the stars it contains .
these collisions will destroy the involved stars if they occur at a radius @xmath231 , which is also the radius at which gravitational focusing of encounters between mini - cusp stars becomes important .
destructive collisions primarily happen at orbital pericenter at a per - star rate @xmath232 in contrast , weaker ( soft ) coagulative collisions happen preferentially at apocenter , with a per - star rate @xmath233 generally , coagulative collisions happen at rates orders of magnitude higher than destructive ones , which we hereafter neglect .
these stellar mergers will nonetheless deposit huge amounts of kinetic energy into the merger product ; much of this will thermalize promptly due to shocks , inflating the star to the point where it can likely be tidally disrupted on a subsequent pericenter passage , or grow to a size where it could engulf the central bh .
the few - body dynamics of these tidal capture mini - cusps are likely complex , as scalar resonant relaxation of stellar orbits should be suppressed by gr apsidal precession @xcite , even vector resonant relaxation may be likewise suppressed if the bh is spinning @xcite , and the number of stars may be too few for standard two - body relaxation approximations to hold . nonetheless , basic density considerations indicate that direct collisions should be frequent , and the aftermath of these will be inflated stars that are easily disrupted by the bh .
once the bh has grown to a very large size , typically @xmath234 , ionization of tidally captured stars in encounters with unbound stars will become frequent ( @xmath235 ) . at this point
tidal capture may turn off as a growth channel , leaving only full tidal disruptions to grow the bh .
we therefore conclude that tidal capture events generally result in delayed tidal disruptions due to inflation of the captured star : this occurs either as a result of runaway mode excitation ( in analogy to * ? ? ?
* ) , or following the thermalization of direct soft collision kinetic energy .
these delayed tidal interactions might be quite different from standard tdes in that they involve interaction of the bh with stars that have radially inflated and are possibly more massive than normal stars .
we will assume that the end point of such events is the consumption of the bulk of the mass of the normal star by the sbh .
this will likely occur at super - eddington rates while the black hole is small .
such super - eddington accretion disks may lose some mass in outflows , though the exact amount is still debated in the simulation literature ( * ? ? ?
* for example , find @xmath236 of the inflowing mass lost in a wind ) .
we neglect mass loss in super - eddington outflows , but emphasize that it could become important for sbh growth if it reaches an order unity fraction of the mass inflow rate .
for sbhs in unconcentrated plummer spheres , where @xmath237 . for the analytic predictions , we use eq .
[ eq : tcrate ] , while we measure this from simulation data by following the time evolution of the most massive surviving black hole .
the general agreement between analytic theory and few - body integrations ( which all employ 10 sbhs ) validates our single - body analytic growth rate approximation for dense star systems.,width=321 ] the analytic description of runaway sbh growth presented in the prior section neglects some basic dynamical processes that are difficult to account for in a closed form . for this reason ,
we have performed two sets of simple numerical simulations to test the validity of our analytical prescriptions in a more realistic , dynamical setting .
the first set are one - body orbital integrations of single sbhs in an analytic background potential representing an nsc .
the sbh s diffusion through phase space is treated analytically through calculation of the diffusion coefficients ( e.g. , @xcite ) at each time step and respective modification of its velocity vector .
the background potential is taken as static and only evolves if the sbh `` feeds '' on it , i.e. , by ( probabilistically ) capturing or disrupting a star , which reduces the mass of the nsc .
moreover , the scale radius of the background potential can expand due to dynamical heating from the sbh .
these highly idealized simulations are valuable for testing our approximate `` @xmath238 '' estimates , and the influence of phase - space diffusion on the growth rates of the sbhs
. the second set of simulations are few - body integrations of a small number of sbhs in the same type of background potential .
these simulations are significantly more realistic in that they include mergers of sbhs and their dynamical ejections through gravitational - wave recoils or few - body interactions . they provide a more physical test of the analytic model of [ sec : clusterobs ] .
comparisons of our analytic prescriptions with results from both sets of simulations are presented in the following . for a detailed description of the numerical simulations
see appendix [ sec : simulations ] . in fig .
[ fig : fewbody ] , we show how analytic , single - body and few - body estimates of the growth rates of sbhs in nscs compare . for this purpose , we define a `` runaway time scale '' @xmath239 such that eq .
[ eq : tcrate ] can be written as @xmath240 with @xmath239 having dimension myrm@xmath241 .
solving this equation for @xmath164 gives us the mass of the sbh with time @xmath242 the runaway time scale can therefore be calculated from any two snapshots of a numerical simulation using @xmath243 as we can see in fig .
[ fig : fewbody ] , the deviations between the analytic predictions and the numerical simulations is small across a wide range of nsc velocity dispersions , @xmath244 , relevant for our study .
moreover , few - body dynamical effects , such as ejections and mergers , have a low but highly stochastic effect on the runaway time scales . in general
, it appears that our analytic prescriptions reliably describe the rate at which sbhs gain mass from interactions with unbound stars in both the one - body and few - body regimes . as noted earlier , these prescriptions likely break down
if a self - gravitating , spitzer - unstable subcluster of sbhs can form , but such a phase of cluster evolution is likely short - lived , eventually culminating in the few - body regime probed by our simulations . in reality ,
the assumptions underlying the initial phase of super - exponential growth will break down once the an sbh seed has grown large enough to dominate the potential surrounding it , as we quantify in the following section .
the importance of multi - sbh effects ( which already appears to be small , based on the results in fig .
[ fig : fewbody ] ) will also diminish , as remaining sbhs are efficiently ejected from the centers of clusters with a massive imbh @xcite .
although two effects ( binary ionization , angular momentum diffusion ) will begin impeding tidal capture runaways once the sbh has grown to a mass @xmath245 , neither of these can cleanly terminate the runaway , as each becomes much smaller if the initial pericenter of capture is moved somewhat inward .
they can therefore be thought of as modest reductions in the tidal capture cross - section .
even if tidal capture were completely deactivated , imbh growth through tidal disruptions would continue so long as debris streams were not deflected by perturbations from cluster stars .
thus tidal disruption continues to grow the black hole provided @xmath246 tidal disruptions therefore can continue to grow the imbh up to very large masses in any realistic star cluster , and the factor of @xmath119 reduction in their cross - section ( relative to tidal capture ) will matter little in the runaway growth regime .
since accretion and binary physics seem unlikely to deactivate tidal growth channels , the termination of the runaway must instead have a stellar dynamical origin . in this section ,
we identify three stages of black hole growth through star capture that correspond to three different bh mass ranges .
the first stage is the one we have considered so far in this paper , where the sbh drifts through a cluster core whose stars have not yet been depleted through tidal capture / disruption . in this stage , the rate of sbh growth is @xmath247 . during this ( and later ) stages , the black hole moves in the almost flat potential of the cluster core .
its orbit can be thought of as brownian motion : damped ( by dynamical friction ) and stochastically driven ( by diffusive scatterings ) movements in a simple harmonic oscillator @xcite .
the rms value of the black hole s
orbital radius is @xmath248 there is an order unity prefactor in @xmath249 which depends on the details of the star cluster distribution function ; we neglect it for simplicity . for the remainder of this section
we take @xmath46 to be the average stellar mass in the cluster s present day stellar mass function . because even the gravitationally focused tidal capture radius @xmath250 , it is unlikely that the bh can ever deplete its `` wandersphere . ''
however , as the bh grows , very quickly the wandersphere will recede inside the influence radius @xmath251 .
once @xmath252 , the bh is no longer swimming through a sea of stars but is instead orbited by a bound minicluster and a much larger cluster of unbound stars . at this point
the rate of tidal consumption may no longer be well approximated by an @xmath238 calculation .
the end of the first stage of sbh growth occurs once it reaches a mass @xmath253 which is generally small , @xmath254 .
if large stellar mass black holes ( e.g. if star formation occurs at low metallicity ; * ? ? ?
* ) exist at the birth of a small or highly concentrated cluster , the first growth stage may be bypassed all together . in the second stage of black hole growth , the rate of star consumption is still confined to the nsc core but is now diffusion - limited , as in the classical loss cone problem @xcite .
we estimate diffusion - limited consumption rates using the simplified formalism of @xcite , who find that the stellar consumption rate is @xmath255 , where @xmath256 and @xmath257 are the stellar consumption rates in the empty and full loss cone regimes , respectively .
we note that @xmath258 in these integrals ; the angular size of the loss cone @xmath259 .
we have used the second moment of the present day stellar mass function , @xmath260 .
the critical radius @xmath261 is the location where the per - star consumption rate is equally partitioned between the full and empty loss cone regimes .
it can be found by solving the equation @xmath262 for @xmath261 ( which can not be done in closed form , as @xmath263 is a function of @xmath261 ) .
the dimensionless constant @xmath264 is a numerical prefactor from the relaxation time ( eq .
[ eq : trelax ] ) . during the second stage of bh growth , @xmath265 .
initially , @xmath266 and most of the consumed stars come from the full loss cone regime ; the total consumption rate here is comparable to our earlier @xmath238 estimate , and still scales as @xmath61 . as the black hole grows further ,
the two regimes of loss cone fueling become comparable in magnitude , slowing the runaway .
however , the imbh growth rate remains roughly exponential until the entire cluster core has been consumed , at which point @xmath267 becomes subdominant and @xmath268 quickly decelerates .
neglecting the full loss cone regime , we can find an asymptotic late time ( large @xmath164 ) solution . in the empty loss cone regime , most of the stellar flux into the loss cone originates from the smaller of @xmath261 and @xmath269 .
generally , @xmath270 , in which case we find that @xmath271 ( here we have also neglected the weak dependence of @xmath268 on logarithmic functions of @xmath164 ) .
more precisely , we find under this approximation ( and assuming that @xmath63 ) that @xmath272 star capture predominantly in the empty loss cone regime ( after the imbh has reached a mass @xmath273 ) marks the third and final phase of black hole growth in nscs .
integrating @xmath274 , we see that the black hole mass converges to a late - time value that only depends strongly on @xmath2 : @xmath275 in the second of these equations we have taken typical stellar masses and radii from a kroupa imf , and have used @xmath276 inside the logarithmic terms ( though the final prefactor depends only weakly on this choice ) .
we have also used the shorthand @xmath277 .
this `` saturation mass '' is reached at late times after the entire cluster core has been consumed , provided the imbh does not grow through other means ( e.g. agn activity ) .
our model therefore makes clear predictions for the mass of imbhs grown through the mechanisms in this paper , predictions which to first order depend only on @xmath2 .
we note that because of the approximate one - to - one correspondence between @xmath2 and the velocity dispersion of the host galaxy @xcite , this is in principle testable even when the nsc can not be resolved .
we compare these predictions to observations in [ sec : observations ] .
when @xmath164 is relatively small , @xmath270 and the above approximations are valid for the saturation mass ; in this regime @xmath278 . for high-@xmath2 nscs , it is possible for smbhs in the diffusion - limited growth regime to grow to the point where @xmath279 , in which case the asymptotic growth rate formula changes somewhat . in this limit
we find a similar but simpler version of eq .
[ eq : msat ] , @xmath280 this extremely simple formula ( @xmath281 ) represents the final saturation mass for an imbh growing through tidal capture of stars in a cluster with @xmath282 , but in practice it is only reached for high-@xmath2 systems ; lower ones saturate at the similar mass given by eq .
[ eq : msat ] , which has slight differences in the prefactor and scaling exponents . the overall growth history of example imbhs is shown in fig .
[ fig : mdotedd ] .
notably , sbhs that are able to enter the runaway regime within a hubble time often require periods of super - eddington growth ; throughout this paper we assume that these rates of mass inflow are possible , but we emphasize here that such an assumption is still an open question in the accretion literature @xcite . in as a function of black hole mass for clusters with initial central densities @xmath283 ( i.e. on the cusp of runaway growth for a @xmath284 seed .
the purple , blue , and green curves have initial core radii @xmath285 , @xmath286 , and @xmath287 , respectively , corresponding to different velocity dispersions which are labeled in the plot .
the three dots along each curve label @xmath164 at times @xmath288 , @xmath289 , and @xmath290 after the sbh enters the cluster core with a mass of @xmath291 .
, width=321 ] another notable feature of our scenario is that smbh seeds formed through tidal capture runaways are likely born with negligible spin .
although the onset of the runaway may involve tidal captures of the innermost stars , once diffusion - limited growth begins , it will come primarily from stars on radial orbits taken from more distant regions of the star cluster .
even if the cluster is endowed with net rotation , such rotation is unlikely to be imprinted on the almost radial orbits vulnerable to tidal capture or disruption .
in contrast , smbhs that have grown substantially through comparable - mass mergers or through accretion from gas inflows on larger scales will generally approach larger values of spin @xcite , though this can be prevented if mass accretion is dominated by short - lived episodes of gas inflow from random directions @xcite .
so far , we have seen that clusters of sufficiently high central density and velocity dispersion will enter a runaway ( super - exponential ) phase of bh growth .
this runaway regime begins slowing as the loss cone depletes , and then putters out into a slower diffusion - limited growth rate once the bh has eaten the entire core of its host cluster . even at this more modest diffusion - limited rate , however , bhs can grow significantly over a hubble time by tidal capture and disruption . in this section , we compare the predictions of our model to the observed demographics of nscs , and to signatures of massive black holes in galactic nuclei .
these two data sets present several tests of our model , best formulated as questions .
do we see nscs unstable to a tidal capture runaway that lack massive central black holes ? and
do we see a smbh mass distribution in galactic nuclei that falsifies our model of smbh seed formation ? using the simple analytical estimates for tidal capture rates that we derived in [ sec : clusterobs ] and validated in
[ sec : numerics ] , we can now examine observed nscs to determine their tidal capture runaway timescales .
[ fig : tcrates ] plots the tidal capture rate @xmath292 as a function of 1d cluster velocity dispersion @xmath293 ( as before , this is estimated using the fitted cluster mass and radius ) for an sbh with @xmath294 , assuming that there is no imbh or smbh to modify the central potential .
we note that estimates of the tidal capture rate are uncertain by one order of magnitude because of a factor @xmath295 uncertainty in the fitted cluster concentration parameter .
specifically , all clusters in this sample were fit to a grid of king models , but this grid was coarsely sampled in the dimension of @xmath17 : the only possible concentrations were @xmath296 and @xmath297 ( georgiev , private communication ) .
many of the nscs in fig .
[ fig : tcrates ] have tidal capture rates @xmath298 , indicating that they are formally unstable to runaway sbh growth through tidal capture .
these nscs generally have @xmath299 , and in some cases have truly enormous tc rates ( due to their high central densities ) .
we highlight here two systematic uncertainties that may cause substantial error in our estimates of central density and derived quantities such as @xmath292 : first , the grid of @xmath17 only extends up to concentrations @xmath300 , meaning that many clusters may be even more concentrated and have even higher @xmath292 .
secondly , the estimates of cluster mass @xmath16 assume a constant mass - to - light ratio , and do not account for color gradients due to dynamical or primordial mass segregation .
both of these effects concentrate surface brightness in the center of the nsc and cause an _
overestimation _ of cluster concentration @xmath301 .
observations of individual nscs often find color gradients , but the direction of the gradient can vary from cluster to cluster @xcite .
a careful resolution of these two uncertainties is important , but will have to wait for future observational work . , plotted against 1d nsc velocity dispersion @xmath293 . in all of these calculations @xmath292
is presented for @xmath151 .
the dashed and dotted black ellipses show the @xmath302 and @xmath303 contours .
notably , many of the more massive nscs are above the solid red line and have tidal capture runaway timescales @xmath304 , indicating that they may already harbor unresolved massive black holes .
almost no nscs exist in the runaway regime that have @xmath305.,width=321 ] we also can compare our predictions to the smaller number of galaxies which have both nsc mass measurements ( or upper limits ) and dynamical smbh mass measurements ( or upper limits ) . in fig .
[ fig : nscdichotomy2 ] we show our prediction for saturation mass @xmath306 for massive black holes that have been growing through star capture for a hubble time .
we plot @xmath306 , and measured nsc / smbh masses , against the host galaxy s effective dispersion @xmath8 , which is comparable to the nsc internal velocity dispersion @xcite .
we find that the smallest smbhs with dynamically measured masses ( @xmath307 ) are within a factor of a few of @xmath306 . in galaxies with a larger @xmath8 ,
smbh masses are much larger than @xmath306 , implying that other mechanisms dominate the growth of these smbhs , in agreement with the soltan argument . in smaller galaxies with @xmath308
, dynamical mass measurements only put upper limits on the presence of an smbh , and these upper limits are generally well below the predicted saturation mass .
this implies that , as expected , a tidal capture runaway has not occurred in these low velocity dispersion systems .
this low-@xmath8 result is in reasonable agreement with our prediction that nscs must have a minimum central density in order to begin growing sbhs through runaway star capture . as we see in fig .
[ fig : tcrates ] , it is only nscs with @xmath299 that possess high enough central density to enter the runaway regime within a hubble time .
this impressive pair of observational coincidences - the lack of smbhs in galaxies with @xmath309 _ and _ the unfavorably low central densities of nscs in galaxies with @xmath310 - is a nontrivial piece of circumstantial evidence in favor of our model for the birth of massive black holes in galaxies of moderate velocity dispersion .
the evidence for imbhs in the nuclei of dwarf or other galaxies is mixed , but these observations are a crucial discriminant between models of smbh seed formation , which predict very different minimum imbh masses : @xmath311 for pop iii supernovae , @xmath312 for traditional `` runaway collision '' scenarios , and @xmath313 for direct collapse of high-@xmath314 minihalos . in our scenario , we expect a distribution of black hole masses ; if the runaway time is @xmath315 , then typically @xmath316 .
it is possible for our scenario to produce smaller imbhs in clusters where the runaway time is @xmath317 , but because this requires some tuning we expect these to be rare .
our model therefore predicts a general minimum imbh mass at @xmath318 , and the distribution or even existence of small imbhs offers a promising way to falsify our predictions . due to the difficulty of dynamical mass measurements in dwarf galaxies , the most common avenues for imbh detection are indirect .
searches for agn in dwarf galaxies have returned many hundreds of promising candidates @xcite .
the smallest of these have lower mass limits @xmath319 based on luminosity arguments @xcite
. a smaller number of dwarf galaxies have dynamical mass estimates for central bhs ; one particularly exciting recent result found a @xmath320 @xcite .
we note that other , future , avenues for imbh detection do exist . for example , the halo of the milky way galaxy may host many imbhs left over from the interplay between hierarchical structure formation scenarios and gw recoil during imbh - imbh mergers @xcite .
while these wandering imbhs are in principle detectable by the hypercompact stellar systems that surround them , none have been found to date @xcite .
the existence of a large population of imbhs with @xmath321 would not be expected in the context of our model , and there is not yet clear evidence that such a population exists .
a final consistency check for our model is the rate of stellar tidal disruption events . from fig .
[ fig : mdotedd ] we see that an nsc hosting a massive black hole of mass @xmath276 should be tidally disrupting stars at a rate @xmath322 , depending on the host nsc @xmath2 .
these numbers are roughly consistent with empirically calculated rates of stellar tidal disruption @xcite , although we caution that these empirical tde rates are based on extrapolating directly determined tde rates in larger galaxies down to smaller masses ( where surface brightness profiles can not be resolved on the relevant scales ) .
a more useful constraint is the volumetric tidal disruption rate , as this is dominated by the smallest galaxies which host central massive bhs @xcite .
although our current sample of observed tidal disruption flares is small and suffers from selection effects ( roughly a dozen optically selected events , and a similar number detected through soft x - ray emission ) , it appears that the volumetric event rate inferred from observations is at the extreme low end of theoretical predictions @xcite , which may imply an absence of low mass ( @xmath323 ) black holes in the low redshift universe @xcite .
although individual tdes may arise from smaller imbhs @xcite .
] , a high occupation fraction of imbhs in dwarf galaxies would be hard to reconcile with the observed low volumetric tde rate .
we note here that past works have shown that star capture can contribute significantly both to the mass growth of small smbhs @xcite , and to the lower end of the x - ray agn luminosity function @xcite .
however , these past arguments have been based on present - day observations of low - mass smbhs and their host galaxies .
the most novel contribution of this paper is the self - consistent evolutionary picture we have presented , which shows how these low - mass smbhs must emerge from nscs above a certain @xmath62 and @xmath2 threshold .
we have shown that stellar mass black holes in nuclear star clusters will undergo runaway growth into imbhs through tidal interactions with cluster stars . at early times ( @xmath324 )
this process is super - exponential and is dominated by unbound stars . at intermediate times ( @xmath325 ) black hole growth is dominated by bound stars from the full loss cone regime and is initially super - exponential , but slows as the cluster core depletes . at later times , the imbh has eaten the entire mass of the cluster core and its diffusion - limited growth slows substantially , saturating at a finite value @xmath326 , as derived in eq .
[ eq : msat ] ( or , for high-@xmath2 systems , at a value @xmath327 ) .
thus the final mass of a massive black hole grown solely through tidal interactions with stars in an nsc depends primarily on the nsc velocity dispersion @xmath2 .
we emphasize that theoretical uncertainties remain concerning both the processes of tidal disruption and tidal capture . in the simplest models for tidal disruption events , exactly @xmath328 of the stellar mass
is bound to and accreted by the black hole @xcite ; however , it is possible that the dissipative processes required to form an accretion disk may unbind a large fraction of the initially bound gas @xcite . whether this occurs in practice is an open question in tde research . even if this type of `` leakiness '' deactivates tidal disruption as a growth channel
, tidal captures will continue to grow the black hole . while this channel may deactivate at higher ( @xmath312 ) imbh masses , this is not certain , and the buildup of a tight cluster of bound stars may lead to a cascade of soft collisions , inflating the stars and allowing them to be consumed .
we also note that both channels imply that a sbh seed that grows into an imbh or smbh in a hubble time will likely pass through phases of super - eddington growth ; this is commonly assumed to be possible in the tidal disruption literature but the exact accretion physics remains uncertain . in order for runaway growth to occur , the star cluster must be dense , with a sbh - star tidal interaction rate @xmath329 .
this corresponds to physical ( cluster core ) densities @xmath330 . from a theoretical perspective ,
such densities are achievable in a nsc that enters a state of core collapse .
although we consider additional physical processes such as dynamical friction and the finite time required for binary burning , we are in general agreement with the argument of @xcite that primordial binaries should not be able to halt core collapse in clusters with @xmath77 .
the scenario presented in this paper differs from analogous `` collisional runaway '' scenarios in the literature by its delayed nature .
previous runaway scenarios typically invoke much higher densities @xmath62 so that either a supermassive star can be formed on a timescale short compared to the stellar evolution time @xmath331 @xcite , or so that runaway gw capture can occur in a dark subcluster of compact remnants without termination due to ejection @xcite .
unlike these scenarios , ours unfolds over longer timescales and is therefore not as promising a candidate for production of high-@xmath314 quasars .
however , the central stellar densities required by our scenario appear achievable in many current low-@xmath314 nscs , making a process like the one described in this paper relevant for the evolution of small galactic nuclei .
more specifically , we expect massive black holes to be largely absent from nscs with @xmath332 , and generally present in larger nscs .
the general physical considerations outlined above appear to be consistent with current observational data on nscs .
firstly , the distribution of dynamically measured smbh masses and mass upper limits is in good agreement with the predicted transition @xmath2 from the @xcite argument , and smbh masses above this transition @xmath2 agree with our predicted saturation mass @xmath306 . in significantly larger galaxies , @xmath333 , implying smbh growth in this regime
is dominated by non - tidal processes , as is expected from the soltan - paczynski argument .
secondly , we analyze the nsc sample of @xcite and find that essentially no nscs with @xmath332 are dense enough to be unstable to a tidal runaway . for a sufficiently large velocity dispersion system ,
we predict either a smbh grown up to the saturation mass via the tidal processes we have discussed , or one which has passed this point and has grown to still larger mass via large scale gas accretion . in this paper
, we have shown that tidal capture and disruption of stars will inevitably grow stellar mass black holes into imbhs or smbhs in less than a hubble time provided the host star cluster is sufficiently dense .
although large smbhs in the universe ultimately grow via other mechanisms @xcite , the processes outlined in this paper appear sufficient to explain the bottom end of the smbh mass function . the critical density required to initiate runaway growth _
is not observed to occur _ in the smallest nscs , as we expect from simple theoretical considerations first outlined in @xcite .
however , larger nscs often exceed this central density threshold , and many of these nscs likely possess unresolved massive black holes .
the predictions of our model would be improved by future dynamical work to determine the evolution of tight stellar clusters formed by tidal capture , as well as hydrodynamical simulations of tdes that robustly determine the fraction of stellar mass accreted . on the observational side , x - ray observations ( * ? ? ?
* for example ) of the densest nscs may be useful for determining the present - day rate of black hole growth in these dynamically rich systems .
we thank torsten bker , iskren georgiev , dong lai , nathan leigh , cole miller , and scott tremaine for useful discussions .
we thank jenny greene for discussions and for sharing several maser smbh mass estimates .
we also thank seppo mikkola for feedback and for providing his ar - chain code .
ncs acknowledges support through nasa from einstein postdoctoral fellowship award number pf5 - 160145 , through nsf grant ast-1410950 , and from the alfred p. sloan foundation to brian metzger .
ahwk acknowledges support from nasa through hubble fellowship grant hst - hf-51323.01-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 .
this work was aided by the hospitality of the aspen center for physics . 99 abel , t. , bryan , g. l. , & norman , m. l. 2002 , science , 295 , 93 alexander , t. , & natarajan , p. 2014 , arxiv:1408.1718 antonini , f. , capuzzo - dolcetta , r. , mastrobuono - battisti , a. , & merritt , d. 2012 , apj , 750 , 111 antonini , f. , barausse , e. , & silk , j. 2015 , apj , 812 , 72 ayal , s. , livio , m. , & piran , t. 2000 , apj , 545 , 772 bahcall , j. n. , & wolf , r. a. 1976 , apj , 209 , 214 baldassare , v. , reines , a. , gallo , e. , & greene , j. 2015 , arxiv:1506.07531 begelman , m. c. , & rees , m. j. 1978 , mnras , 185 , 847 belczynski , k. , bulik , t. , fryer , c. l. , et al .
2010 , apj , 714 , 1217 berti , e. , & volonteri , m. 2008 , apj , 684 , 822 binney , j. , & tremaine , s. 2008 , galactic dynamics : second edition , by james binney and scott tremaine .
isbn 978 - 0 - 691 - 13026 - 2 ( hb )
. published by princeton university press , princeton , nj usa , 2008 .
bker , t. , sarzi , m. , mclaughlin , d. e. , et al .
2004 , aj , 127 , 105 bker , t. 2008 , apjl , 672 , l111 carson , d. j. , barth , a. j. , seth , a. c. , et al .
2015 , aj , 149 , 170 chatterjee , p. , hernquist , l. , & loeb , a. 2002 , physical review letters , 88 , 121103 chatterjee , p. , hernquist , l. , & loeb , a. 2002 , apj , 572 , 371 clark , p. c. , glover , s. c. o. , klessen , r. s. , & bromm , v. 2011 , apj , 727 , 110 cohn , h. , & kulsrud , r. m. 1978 , apj , 226 , 1087 cohn , h. 1980 , apj , 242 , 765 ct , p. , piatek , s. , ferrarese , l. , et al . 2006 , apj supplements , 165 , 57 dai , l. , & blandford , r. 2013 , mnras , 434 , 2948 davies , m. b. , miller , m. c. , & bellovary , j. m. 2011 , apjl , 740 , l42 french ,
k. d. , arcavi , i. , & zabludoff , a. 2016 , apjl , 818 , l21 devecchi , b. , & volonteri , m. 2009 , apj , 694 , 302 do , t. , kerzendorf , w. , winsor , n. , et al .
2015 , arxiv:1506.07891 ebisuzaki , t. , makino , j. , tsuru , t. g. , et al .
2001 , apjl , 562 , l19 erwin , p. , & gadotti , d. a. 2012 , advances in astronomy , 2012 , 946368 fabian , a. c. , pringle , j. e. , & rees , m. j. 1975 , mnras , 172 , 15p frank , j. , & rees , m. j. 1976 , mnras , 176 , 633 genzel , r. , eisenhauer , f. , & gillessen , s. 2010 , reviews of modern physics , 82 , 3121 georgiev , i. y. , bker , t. 2014 , mnras , 441 , 3570 georgiev , i. y. , bker , t. , leigh , n. , ltzgendorf , n. , & neumayer , n. 2016 , mnras , 457 , 2122 glebbeek , e. , gaburov , e. , de mink , s. e. , pols , o. r. , & portegies zwart , s. f. 2009 , a&a , 497 , 255 gnedin , o. y. , ostriker , j. p. , & tremaine , s. 2014 , apj , 785 , 71 goswami , s. , umbreit , s. , bierbaum , m. , & rasio , f. a. 2012 , apj , 752 , 43 graham , a. w. , & spitler , l. r. 2009 , mnras , 397 , 2148 greene , j. e. , & ho , l. c. 2004 , apj , 610 , 722 greene , j. e. , seth , a. c. , kim , m. , et al .
2016 , arxiv:1606.00018 greif , t. h. , springel , v. , white , s. d. m. , et al .
2011 , apj , 737 , 75 guillochon , j. , manukian , h. , & ramirez - ruiz , e. 2014 , apj , 783 , 23 grkan , m. a. , freitag , m. , & rasio , f. a. 2004 , apj , 604 , 632 hayasaki , k. , stone , n. , & loeb , a. 2013 , mnras , 434 , 909 hayasaki , k. , stone , n. c. , & loeb , a. 2015 , arxiv:1501.05207 heggie , d. c. 1975 , mnras , 173 , 729 heggie , d. c. , & hut , p. 2003
, the gravitational million - body problem , cambridge university press , cambridge , uk hopman , c. , & alexander , t. 2006 , apj , 645 , 1152 ivanov , p. b. , & papaloizou , j. c. b. 2004 , mnras , 347 , 437 ivanov , p. b. , & papaloizou , j. c. b. 2007 , a&a , 476 , 121 jiang , y .- f . , stone , j. m. , & davis , s. w. 2014 , apj , 796 , 106 katz , h. , sijacki , d. , & haehnelt , m. g. 2015 , mnras , 451 , 2352 kim , s. s. , lee , h. m. , & goodman , j. 1998 , apj , 495 , 786 king , i. r. 1966 , aj , 71 , 64 king , a. r. , & pringle , j. e. 2006 , mnras , 373 , l90 kormendy , j. , & mcclure , r. d. 1993 , aj , 105 , 1793 kormendy , j. , & ho , l. c. 2013 , ara&a , 51 , 511 kulier , a. , ostriker , j. p. , natarajan , p. , lackner , c. n. , cen , r. , 2015 , apj , 799 , 178 kpper , a. h. w. , maschberger , t. , kroupa , p. , baumgardt , h. , 2011 , mnras , 417 , 2300 lai , d. 1996 , apjl , 466 , l35 lai , d. 1997 , apj , 490 , 847 lee , h. m. , & ostriker , j. p. 1986 , apj , 310 , 176 lee , h. m. 1987 , apj , 319 , 772 lee , h. m. , & ostriker , j. p. 1993 , apj , 409 , 617 leigh , n. , bker , t. , & knigge , c. 2012 , mnras , 424 , 2130 leigh , n. w. c. , mastrobuono - battisti , a. , perets , h. b. , bker , t. 2014a , mnras , 441 , 919 leigh , n. w. c. , ltzgendorf , n. , geller , a. m. , et al . 2014b , mnras , 444 , 29 leigh , n. w. c. , georgiev , i. y. , bker , t. , knigge , c. , & den brok , m. 2015 , mnras , 451 , 859 li , g. , & loeb , a. 2013 , mnras , 429 , 3040 lightman , a. p. , & shapiro , s. l. 1977 , apj , 211 , 244 lousto , c. o. , zlochower , y. , dotti , m. , volonteri , m. , 2012 , phrvd , 85 , 084015 maksym , w. p. , ulmer , m. p. , roth , k. c. , et al . 2014 ,
mnras , 444 , 866 magorrian , j. , & tremaine , s. 1999 , mnras , 309 , 447 marconi , a. , & hunt , l. k. 2003 , apjl , 589 , l21 mardling , r. a. 1995a , apj , 450 , 722 mardling , r. a. 1995b , apj , 450 , 732 matthews , l. d. , gallagher , j. s. , iii , krist , j. e. , et al . 1999 , aj , 118 , 208 mcconnell , n. j. , & ma , c .- p .
2013 , apj , 764 , 184 meibom , s. , & mathieu , r. d. 2005 , apj , 620 , 970 merritt , d. , schnittman , j. d. , & komossa , s. 2009 , apj , 699 , 1690 merritt , d. , & vasiliev , e. 2012 , phys .
d. , 86 , 102002 metzger , b. d. , & stone , n. c. 2015 , arxiv:1506.03453 mikkola , s. , merritt , d. , 2006 , mnras , 372 , 219 mikkola , s. , merritt , d. , 2008 , aj , 135 , 2398 miller , m. c. , & davies , m. b. 2012 , apj , 755 , 81 milosavljevi , m. 2004 , apjl , 605 , l13 milosavljevi , m. , merritt , d. , & ho , l. c. 2006 , apj , 652 , 120 moran , e. c. , shahinyan , k. , sugarman , h. r. , vlez , d. o. , & eracleous , m. 2014 , aj , 148 , 136 morscher , m. , pattabiraman , b. , rodriguez , c. , rasio , f. a. , & umbreit , s. 2015 , apj , 800 , 9 mortlock , d. j. , warren , s. j. , venemans , b. p. , et al .
2011 , nature , 474 , 616 neumayer , n. , & walcher , c. j. 2012 , advances in astronomy , 2012 , 709038 novikov , i. d. , pethick , c. j. , & polnarev , a. g. 1992 , mnras , 255 , 276 oleary , r. m. , kocsis , b. , & loeb , a. 2009 , mnras , 395 , 2127 oleary , r. m. , & loeb , a. 2012 , mnras , 421 , 2737 quinlan , g. d. , & shapiro , s. l. 1987 , apj , 321 , 199 pandya , v. , mulchaey , j. , & greene , j. e. 2016 , apj , 819 , 162 peters , p. c. 1964 , physical review , 136 , 1224 poveda , a. , allen , c. , & hernndez - alcntara , a. 2007 , binary stars as critical tools & tests in contemporary astrophysics , 240 , 417 press , w. h. , & teukolsky , s. a. 1977 , apj , 213 , 183 rees , m. j. 1988 , nature , 333 , 523 reines , a. e. , greene , j. e. , & geha , m. 2013 , apj , 775 , 116 sadowski , a. , narayan , r. , tchekhovskoy , a. , et al .
2015 , mnras , 447 , 49 sanders , r. h. 1970 , apj , 162 , 791 sesana , a. 2012 , advances in astronomy , 2012 shapiro , s. l. , & teukolsky , s. a. 1983 , research supported by the national science foundation .
new york , wiley - interscience , 1983 , 663 p. , shiokawa , h. , krolik , j. h. , cheng , r. m. , piran , t. , & noble , s. c. 2015 , apj , 804 , 85 soltan , a. 1982 , mnras , 200 , 115 spitzer , l. , jr .
1969 , apjl , 158 , l139 spitzer , l. 1987 , princeton , nj , princeton university press , 1987 , 191 p. , statler , t. s. , ostriker , j. p. , & cohn , h. n. 1987 , apj , 316 , 626 stone , n. , sari , r. , & loeb , a. 2013 , mnras , 435 , 1809 stone , n. c. , & ostriker , j. p. 2015 , arxiv:1503.03479 stone , n. c. , & metzger , b. d. 2016 , mnras , 455 , 859 syer , d. , & ulmer , a. 1999 , mnras , 306 , 35 tremaine , s. d. , ostriker , j. p. , & spitzer , l. , jr .
1975 , apj , 196 , 407 trenti , m. , & van der marel , r. 2013 , mnras , 435 , 3272 valtonen , m. , & karttunen , h. 2006 , the three - body problem , by mauri valtonen and hannu karttunen , pp . .
isbn 0521852242 .
cambridge , uk : cambridge university press , 2006 van velzen , s. , & farrar , g. r. 2014 , apj , 792 , 53 wang , j. , & merritt , d. 2004 , apj , 600 , 149 weinberg , n. n. , arras , p. , quataert , e. , & burkart , j. 2012 , apj , 751 , 136 willott , c. j. , mclure , r. j. , & jarvis , m. j. 2003 , apjl , 587 , l15 visbal , e. , haiman , z. , & bryan , g. l. 2014 , mnras , 442 , l100 volonteri , m. 2010 , a&arv , 18 , 279 yajima , h. , & khochfar , s. 2015 , arxiv:1507.06701
in this paper , we frequently compare simple theoretical predictions to distributions of observed properties in a large sample of real nscs .
our sample includes 22 nscs from @xcite , 51 nscs from @xcite , and 151 nscs from @xcite . after discarding 10 nscs that overlap between these samples
we are left with 214 unique nscs in our broader sample , 207 of which have been fit to three - parameter king models @xcite for which the total mass @xmath16 , half - mass radius @xmath18 , and concentration parameter @xmath301 are measured . because of computational limitations , only a small number of values are considered for the concentration parameter : @xmath334 . for this reason `` cluster - averaged '' quantities such as @xmath26 should be considered much more reliable than `` cluster - central '' quantities such as @xmath62 , and for many of these clusters we only have lower limits on their concentration and central density . in figs .
[ fig : nscdata ] and [ fig : nscdata2 ] we fit 2d gaussians to various combinations of derived and measured parameters . in the hope that these may be of use to the reader , we provide our best fit relations in table [ tab : distributions ] . .the
2d gaussian fits for joint distributions of @xmath335 and @xmath336 . here
@xmath337 represents a number of different measured and derived parameters of interest .
the covariance matrix for the data set has two eigenvectors whose eigenvalues are @xmath338 and @xmath339 .
the first eigenvector makes an angle @xmath340 with the positive @xmath2-axis . [
cols="^,^,^,^,^",options="header " , ]
in a core - collapsed cluster , the densities of compact objects will be much higher than those of main sequence stars . by compact objects ,
we refer only to neutron stars and other stellar - mass black holes , as white dwarfs are large enough to be tidally disrupted by black holes up to masses of @xmath341 .
denser objects will emit significant energy in gravitational radiation during close encounters .
an unbound pair of compact objects can thus become bound via a gravitational brehmsstralung process ; the maximum impact parameter for this to occur is @xcite @xmath342 corresponding to a maximum pericenter distance of @xmath343 where the symmetric mass ratio @xmath344 and @xmath345 is the velocity of the two compact objects at infinity . in the second line of this equation
we have set @xmath346 . following this capture event ,
the compact object binary will evolve from an extremely high - eccentricity orbit under the influence of gravitational radiation . the very small cross - section for gw capture disfavors this sbh growth mechanism relative to tidal interactions with stars .
gw capture is further disfavored by the large recoil velocities the sbhs will receive due to anisotropic emission of gws during their merger .
these velocities are typically @xmath347 , but with a long tail going up to @xmath348 @xcite .
this is large enough to eject the merged sbh from not just its nsc , but frequently even its host galaxy and halo .
at intermediate impact parameters , stars passing close to stellar mass black holes will be tidally disrupted .
pericenters vulnerable to tidal disruption will lie interior to the tidal radius , @xmath349 following disruption , the stellar debris will follow ballistic trajectories with a frozen - in spread of specific energy @xcite , @xmath350 generally , @xmath351 , meaning that half the star will be unbound and fly off to infinity , while the other half will eventually return to the sbh on highly eccentric orbits .
these bound debris streams will collisionally shock each other and circularize into a super - eddington accretion disk after order unity pericenter returns , although the details of circularization are unclear : although they are likely mediated by gr precession around smbhs @xcite , near stellar mass bhs the tidal radius is very non - relativistic and purely hydrodynamic effects may prevail @xcite .
regardless , a sbh will likely accrete @xmath352 of the tidally disrupted star .
recently , @xcite suggested that the true accretion fraction may be significantly lower than @xmath353 because of the large energy hierarchy in the problem : @xmath351 , so the absorption of even a small fraction of the accretion luminosity will suffice to unbind a majority of the stellar debris bound to the black hole .
if this argument is correct , it would drastically reduce the ability of tdes to contribute to sbh mass growth .
however , the analytic argument of @xcite was designed to apply to smbh tdes , where the energy hierarchy is more than an order of magnitude stronger than it is for sbh tdes .
realistic numerical simulations capable of testing this argument have not yet been run for smbh tdes , but at lower mass ratios ( @xmath354 ) they have , and there is no evidence for the ejection of the bound debris @xcite .
at larger separations , tidal excitation of modes can bind a sbh to a star initially on a mildly hyperbolic orbit .
the excess orbital energy is deposited into the mechanical oscillation of normal modes in the star , putting the two onto an eccentric bound orbit . in order for successful capture to occur the energy deposited into modes , @xmath196 , must exceed the hyperbolic orbital energy @xmath355 . if we decompose the modes into spherical harmonics @xmath356 with radial mode numbers @xmath56 , we can follow @xcite to write @xmath357 here the dimensionless mode - orbit coupling constant @xmath358 is broken down into two distinct components : a dimensionless overlap integral @xmath359 measuring normal modes inside a given stellar structure model , and a dimensionless tidal coupling integral @xmath360 .
we calculate the second of these directly , using @xmath361 in this coupling integral @xmath362 is orbital phase and @xmath363 is the mode frequency .
values of @xmath364 have been tabulated in other papers @xcite , and we use these tabulations for the low-@xmath167 modes which dominate energy deposition . sample tidal coupling curves @xmath365 and @xmath366 are shown in fig .
[ tlcurve ] . in practice , the vast majority of energy deposited in typical tidal capture events comes from @xmath367 oscillations .
numerically , one can calculate @xmath368 to find the critical pericenter @xmath369 that results in tidal capture ; if we parametrize @xmath370 we find that @xmath50 in a cluster with @xmath371 .
this is shown in more detail in fig .
[ deltaecurve ] , and quantified in fig .
[ chifit ] , where we show how @xmath119 scales with the dimensionless number @xmath372 , the squared ratio of the star s surface escape velocity to the cluster @xmath2 . included in this plot
are power - law best fits to exact solutions ; if we cut off the best fit at @xmath373 ( an approximation for where the linear tides approximation is valid ) we have @xmath374 , for both @xmath170 ( blue ) and @xmath375 ( orange ) modes in a @xmath376 polytropic star .
calculation of @xmath377 and @xmath378 involves a summation over the lowest - order p - modes , but in general it is the f - mode that dominates .
these curves are independent of perturber mass ratio.,width=321 ] star perturbed by a @xmath379 bh , as a function of the pericenter radius @xmath380 in units of tidal radii .
the blue curve shows the contribution of @xmath170 modes , the orange curve the contribution of @xmath375 modes , and the purple curve ( overlapping with blue ) their combined effect .
the horizontal black lines show the typical energy at infinity of clusters with @xmath381 ( solid ) , @xmath382 ( dashed ) , and @xmath383 ( dotted).,width=321 ] parameter required for tidal capture shown as a function of @xmath372 .
the black line is an exact numerical solution , and we also show power law best fits for @xmath384 ( dotted green line ) and @xmath385 ( dashed purple line ) .
the former , fiducial curve is given in the text as eq .
[ eq : chifit ] , while the alternate one ( which may extend too far into the nonlinear tides regime , beyond which our treatment of mode excitation breaks down ) is @xmath386 . ,
for the numerical tests presented in
[ sec : numerics ] , we used two different integrators : one for single - body integrations and one for few - body integrations . the single - body code is a simple hermite integrator with a fixed time step , whereas the few - body code is a modified version of the algorithmic chain integrator ar - chain developed by @xcite .
the latter uses algorithmic chain regularization for high - precision integration of few - body dynamics , and is capable of handling velocity - dependent forces efficiently .
it includes relativistic post - newtonian terms up to order pn2.5 @xcite .
both codes integrate the orbits of sbhs in the gravitational field of a background nsc . for computational convenience ,
this star cluster is approximated by a plummer sphere with total mass , @xmath387 , and plummer scale radius , @xmath388 .
following , e.g. , @xcite , its mass density profile can be written as @xmath389 having a core of size @xmath390 with a central density of @xmath391 .
its half - mass radius is given by @xmath392 .
the velocity dispersion at radius @xmath380 can be written as @xmath393 with a central velocity dispersion of @xmath394 .
these quantities , density and velocity dispersion , are of key importance for the runaway time scale of the sbhs .
we relate these three - dimensional quantities to observables by defining the nsc s ( projected ) effective radius as @xmath395 , and the velocity dispersion at this radius to be @xmath396 weak encounters with background stars will let the sbhs diffuse through phase space while they are orbiting within the gravitational potential of the nsc .
the diffusion can be expressed as change in velocity of an sbh by @xmath397 per unit time .
we can split this change into a component along the direction of motion of the sbh , and one perpendicular to that .
following @xcite , the diffusion coefficients can be expressed as @xmath398 & = & -\frac{4\pi g^2\rho(r)m_\bullet\ln\lambda}{\sigma^2}f(\chi),\label{eq : df}\\ d[(\delta v_\parallel)^2 ] & = & \frac{4\sqrt{2}\pi g^2\rho(r)m_\bullet\ln\lambda}{\sigma}\frac{f(\chi)}{\chi},\\ d[(\delta \vec{v}_\bot)^2 ] & = & \frac{4\sqrt{2}\pi g^2\rho(r)m_\bullet\ln\lambda}{\sigma}\left[\frac{\mbox{erf}(\chi)-f(\chi)}{\chi}\right],\end{aligned}\ ] ] where @xmath399 is the velocity change in direction of motion , and @xmath400 is the velocity change perpendicular to the direction of motion . here , @xmath164 is the mass of the black hole , and @xmath401 . the function @xmath402 is given by @xmath403 we approximate the factor @xmath404 in the coulomb logarithm as @xmath405 we can identify eq . [
eq : df ] as the dynamical friction term , that is , if we assumed @xmath406 = d[(\delta \vec{v}_\bot)^2 ] = 0 $ ] , we would get chandrasekhar s dynamical friction formula .
the second term introduces a variance of the friction term , and even allows the sbhs to be accelerated when the velocity of a sbh gets sufficiently small .
the third term introduces a change in velocity perpendicular to the direction of motion of the sbh .
it is a randomly oriented vector , and hence causes the sbhs to execute a random walk in phase space .
the last two terms will establish that the sbhs are ultimately in energy equipartition with the background stars .
the velocity changes @xmath407 and @xmath408 per unit time @xmath409 can be computed with the above equations .
both changes are normal distributed , where the mean , @xmath410 , and the variance , @xmath411 , of the distributions are given by @xmath412\delta t,\\ \sigma_\parallel & = & d[(\delta v_\parallel)^2]\delta t,\\ \mu_\bot & = & 0,\\ \sigma_\bot & = & d[(\delta \vec{v}_\bot)^2]\delta t.\end{aligned}\ ] ] we compute the diffusion coefficients for each black hole at each time step , and modify its velocity on a monte carlo basis . for each time step
we draw a random orientation before adding the perpendicular velocity change to the respective sbh . hence , the sbh s modified velocity , @xmath413 , is computed using @xmath414 the change of energy , @xmath415 , of the orbiting black hole due to phase - space diffusion is given back to the stellar background potential , with @xmath416 .
as a consequence of this energy transfer , inspiralling black holes will cause an expansion of the nsc . for this purpose
we calculate the change in potential energy , @xmath417 , of the stellar system using @xmath418 where we made use of the virial theorem @xmath419 . with this change in potential energy
we can calculate a new radius for the stellar background potential at each integration step . for the plummer sphere
the new scale radius can be calculated as @xmath420 for each sbh we compute the probability of capturing or disrupting one or many background stars during each time step .
according to eq .
[ eq : tcrate ] , the tidal - capture rates are estimated as @xmath421 where the number density of stars is @xmath145 = @xmath422 with a mean stellar mass of 0.45m@xmath423 . the tidal - capture cross section , @xmath411 , is given by @xmath424 where @xmath425 is the total mass of sbh plus star , and @xmath426 is the relative velocity between the two ( which we approximate as the local velocity dispersion @xmath2 ) .
the tidal radius is given by @xmath427 , with @xmath428 being the mean stellar radius approximated as @xmath429 . from these rates we calculate tidal capture probabilities by assuming they are constant throughout one time step .
since our time steps are small fractions of the nsc s crossing time , this assumption is a good approximation . using this tidal capture probabilities , we let the sbhs grow on a monte carlo basis .
we distinguish between tidal capture and disruption events , which have related cross sections , by assuming that a fraction of the capture events are disruptions .
each tidal capture results in the respective sbh gaining one mean stellar mass , @xmath46 , whereas disruptions increase the sbh mass by @xmath430 ( see 3 ) .
however , there remains a notable shortcoming of this numerical approach : the plummer sphere has a fixed ratio of core radius to half - mass radius ( @xmath431 ) ; realistic clusters exhibit wide variance in this ratio , and the large central densities of highly concentrated clusters are more favorable to tidal capture runaways . in this sense
our simulations are conservative with regard to the onset of runaway sbh growth , and we are currently performing analogous few - body simulations using the potential - density pair of @xcite to generalize these results .
the code ar - chain includes pn terms up to order 2.5 .
the sbhs can therefore merge via gravitational wave emission .
we include gravitational wave recoils following the prescription outlined in @xcite , which is based on the fitting formula by @xcite . to save computational time ,
we assume that a merger will be inevitable when the separation between two sbhs gets smaller than 10000 schwarzschild radii . at the moment of the merger ,
we assume that the spin vectors of the two sbhs are randomly aligned .
this results in kick velocities of up to several thousand kms@xmath432 , with a median kick of @xmath433kms@xmath432 .
since our simulations focus on nsc with relatively low escape velocities , this implies that a majority of the merging sbhs escape from the nscs .
black holes can also eject each other via strong three - body interactions .
we remove sbhs from the simulations once they move beyond 1kpc from the nsc , assuming that it will take them more than a hubble time to find their way back into the center of the host galaxy .
the simulation results presented in 4 are based on a set of 9 runs with the single - body code and 9 runs with the few - body code .
the characteristics of the nscs in both sets are the same , but while the one - body simulations show us the evolution of a single sbh in these nscs , the few - body code follows 10 sbhs as they grow and interact .
we use the mcluster code to generate initial conditions for these simulations @xcite .
the 10 sbh follow a simple mass function with slope @xmath434 between 5m@xmath423 and 20m@xmath423 , resulting in the most massive sbh having 20m@xmath423 and the next - most massive sbh having a mass of 11m@xmath423 .
the single - body simulations are using only the most - massive sbh of these initial conditions .
we make five simulations with a fixed nsc mass of 10@xmath435m@xmath423 and effective radii varying between 0.1pc and 2pc .
another four simulations have a fixed effective radius of 0.2pc and nsc masses between @xmath436m@xmath423 and @xmath437m@xmath423 .
the sbhs are set up randomly within these nscs with velocities typical for their initial nsc positions .
the results are described in 4 and summarized in fig .
[ fig : fewbody ] . | nuclear star clusters ( nscs ) and supermassive black holes ( smbhs ) both inhabit galactic nuclei , coexisting in a range of bulge masses , but excluding each other in the largest or smallest galaxies .
we propose that the transformation of nscs into smbhs occurs via runaway tidal captures , once nscs exceed a certain critical central density and velocity dispersion . the bottleneck in this process , as with all collisional runaways , is growing the first e - fold in black hole mass .
the growth of a stellar mass black hole past this bottleneck occurs as tidally captured stars are consumed in repeated episodes of mass transfer at pericenter .
tidal captures may turn off as a growth channel once the black hole reaches a mass @xmath0 , but tidal disruption events will continue and appear capable of growing the seed smbh to larger sizes . the runaway slows ( becomes sub - exponential ) once the seed smbh consumes the core of its host nsc . while the bulk of the cosmic mass density in smbhs is ultimately produced ( via the soltan - paczynski argument ) by episodic gaseous accretion in very massive galaxies ,
the smallest smbhs have probably grown from strong tidal encounters with nsc stars .
smbh seeds that grow for a time @xmath1 entirely through this channel will follow simple power law relations with the velocity dispersion , @xmath2 , of their host galaxy . in the simplest regime
it is @xmath3 , but the exponents and prefactor can differ slightly depending on the details of loss cone refilling .
current tidal disruption event rates predicted from this mechanism are consistent with observations . |
the discrimination between a finite number of quantum states , prepared with given prior probabilities , is essential for many tasks in quantum information and quantum cryptography .
since nonorthogonal states can not be distinguished perfectly , discrimination strategies have been developed which are optimal with respect to various figures of merit . based on the outcome of a measurement , in these strategies
a guess is made about the actual state of the quantum system .
several strategies admit a certain probability , or rate , @xmath2 , of inconclusive results , where the measurement outcomes do not allow to infer the state . in the strategy of minimum - error discrimination @xcite inconclusive results
are not permitted , and the overall probability @xmath3 of making a wrong guess is minimized with @xmath4 , which corresponds to maximizing the overall probability @xmath5 of getting a correct result .
apart from studying some general features of the optimal measurement @xcite , in the beginning mainly the minimum - error discrimination of states obeying certain symmetry properties or of two mixed states was investigated , see e. g. @xcite .
the minimum - error discrimination of more than two states that are arbitrary has gained renewed interest only recently @xcite .
generalizing the concept of minimum - error discrimination , an optimal discrimination strategy has been studied which maximizes the total rate of correct results , @xmath5 , with a fixed rate @xmath2 of inconclusive results @xcite .
this strategy also maximizes the relative rate @xmath6 for the fixed value @xmath2 , that is it yields the maximum achievable fraction of correct results referred to all conclusive results , which in most cases is larger for @xmath7 than for @xmath8 necessary and sufficient operator conditions for the optimal measurement have been derived @xcite , and it was shown @xcite that under certain conditions the strategy for maximizing @xmath5 with a fixed rate @xmath2 contains the optimal strategies for unambiguous discrimination @xcite or for discrimination with maximum confidence @xcite as limiting cases for sufficiently large values of @xmath2 .
moreover , it was found @xcite that when the maximum of @xmath5 is known as a function of the fixed value @xmath2 , then from this function one can also obtain the maximum of @xmath5 in another optimal discrimination strategy where the error probability is fixed @xcite .
optimal discrimination with a fixed rate @xmath2 has been also investigated for a modified optimization problem where in contrast to the usual assumption the prior probabilities of the states are unknown @xcite .
solutions for optimal state discrimination with a fixed rate @xmath2 of inconclusive results and with given prior probabilities of the states have been recently derived in three independent papers : starting from the operator equations for the optimality conditions , in ref .
@xcite we obtained analytical solutions for the discrimination of symmetric states and of two states occurring with arbitrary prior probabilities , where the two states are either pure or belong to a certain class of mixed qubit states . in refs . @xcite and
@xcite the respective authors showed that the optimization problem with fixed @xmath2 can be solved by reducing it to a resulting minimum - error problem the solution of which is known and to an additional optimization .
they derived analytical solutions for discriminating between two pure states occurring with arbitrary prior probabilities and between the trine states @xcite , and also for the discrimination of geometrically uniform states @xcite .
the present paper goes beyond these previous investigations in several respects .
based on the ideas of our earlier work @xcite we study general properties of the optimal measurement for discriminating with fixed @xmath2 between @xmath0 arbitrary qudit states in a @xmath1-dimensional hilbert space .
specializing on the case @xmath9 , we develop a method for treating the optimal discrimination with fixed @xmath2 between @xmath0 arbitrary qubit states , occurring with arbitrary prior probabilities .
in contrast to the previous papers @xcite we take into account that often the optimal measurement is not unique , which means that the maximal probability of correct results with fixed @xmath2 can be obtained by a number of different measurements . in the special case @xmath4 , where the problem corresponds to minimum - error discrimination , our method differs from the approaches developed previously for treating the minimum - error discrimination of @xmath0 arbitrary qubit states @xcite .
we obtain explicit analytical results for several problems that have not been solved before , mostly for the discrimination with fixed @xmath2 between qubit states which posses a certain partial symmetry , but also for discriminating @xmath0 equiprobable qubit states and for the discrimination between a pure state and a uniformly mixed state in a @xmath1-dimensional hilbert space .
the paper is organized as follows . in sec .
ii we start by considering the optimal discrimination with fixed @xmath2 for @xmath0 qudit states . after this
we specialize on the discrimination of @xmath0 qubit states in a two - dimensional hilbert space and develop a method for solving the problem . sec .
iii is devoted to the discrimination of @xmath0 partially symmetric qubit states .
we conclude the paper in sec .
iv with discussing the relation of our method to previous studies of the minimum - error discrimination of arbitrary qubit states and with a brief summary of results .
the detailed derivations referring to sec .
iii are presented in the appendix .
we consider the discrimination between @xmath0 qudit states , given by the density operators @xmath10 ( @xmath11 ) and prepared with the respective prior probabilities @xmath12 , where @xmath13 .
a complete measurement performing the discrimination is described by @xmath14 positive detection operators @xmath15 with @xmath16 where @xmath17 is the identity operator in the @xmath1-dimensional hilbert space @xmath18 jointly spanned by the @xmath0 density operators @xmath10 .
the conditional probability that a quantum system prepared in the state @xmath19 is inferred to be in the state @xmath10 is given by @xmath20 , while @xmath21 is the conditional probability that the measurement yields an inconclusive outcome which does not allow to infer the state .
the overall probability of inconclusive results is then given by @xmath22 and the probability of correct results , @xmath5 , reads @xmath23 where @xmath3 denotes the total rate of errors .
our task is to maximize @xmath5 , or minimize @xmath3 , respectively , under the constraint that @xmath2 is fixed at a given value . upon introducing a hermitian operator
@xmath24 and a scalar real amplifier @xmath25 the necessary and sufficient optimality conditions take the form @xcite @xmath26 ( @xmath27 ) . for @xmath28 and @xmath29
these conditions refer to minimum - error discrimination @xcite , where @xmath4 .
provided that eqs .
( [ optz1 ] ) and ( [ optz2 ] ) are fulfilled , the rate of correct results takes its maximum value , given by @xmath30 the detection operators @xmath31 ( @xmath27 ) satisfying the optimality conditions need not always be unique , which means that different measurements can be optimal , yielding the same value @xmath32 .
this will be outlined later in this section and will be illustrated by examples in sec .
it is useful to introduce the ratio @xmath33 of the maximal rate of correct results , @xmath34 and the rate of conclusive results , @xmath35 by taking the derivative we find with the help of eq .
( [ pcmax- ] ) that @xmath36 here we used the fact that due to the positivity constraint @xmath37 the condition @xmath38 is equivalent to @xmath39 since @xmath40 for @xmath41 , the operator @xmath42 must turn into a full - rank operator when @xmath2 approaches unity .
( [ optz1 ] ) therefore implies that for sufficiently large values of @xmath2 the condition @xmath43 is necessarily fulfilled @xcite .
when for the given states the optimality conditions can be satisfied with @xmath44 then due to eq .
( [ lim3 ] ) the ratio @xmath33 grows with increasing @xmath2 if @xmath45 and stays constant for @xmath46 .
( [ lim3a ] ) defines @xmath47 as the smallest value of @xmath2 at which further increasing @xmath2 does not yield any advantage , since the relative rate of correct results , given by @xmath33 , remains constant .
if @xmath48 the operator @xmath42 can not be a full - rank operator , due to the equality condition in eq .
( [ optz1 ] ) . for @xmath49
however , this condition does not put any restriction on the rank of @xmath42 .
( [ lim3a ] ) takes into account that for @xmath50 the expressions for @xmath24 and @xmath25 determining the optimal solution are not unique , see the example at the end of sec .
ii a. let us first review the limiting case of large @xmath51 where @xmath33 does not change when @xmath2 is increased , which means that @xmath52 after substituting this expression for @xmath24 into eq .
( [ optz2 ] ) we obtain the conditions @xmath53 and @xmath54 for @xmath27 , where @xmath55 and @xmath56 .
these conditions require that @xcite @xmath57 the constant @xmath58 , that is the largest eigenvalue of @xmath59 , characterizes the maximum confidence that can be achieved for the individual measurement outcome @xmath60 .
in other words , @xmath58 is equal to the maximum achievable value of the ratio @xmath61 which denotes the conditional probability that the outcome @xmath60 is correct given it was obtained @xcite . due to eq .
( [ a - opt ] ) the conditions for optimality following from eq .
( [ optz2 ] ) require that @xmath62 for states where @xmath63 that is where the support of @xmath64 spans the whole hilbert space @xcite .
this means that only the states for which the maximum confidence @xmath58 is largest , @xmath65 , are guessed to occur in an optimal measurement where @xmath43 , that is where @xmath66 according to eq .
( [ lim3a ] ) . using eq .
( [ pcmax- ] ) with @xmath43 and @xmath67 we arrive at @xmath68 where @xmath47 is defined by eq .
( [ lim3a ] ) .
the actual value of @xmath2 depends on the given states . in special cases where @xmath69 the maximum relative probability of correct results @xmath70
can not be increased at all by admitting inconclusive results but stays always equal to @xmath71 , see for instance the example of equiprobable states resolving the identity operator discussed in ref .
@xcite .
if @xmath72 the explicit expression for @xmath24 can be easily determined with the help of eq .
( [ a - opt ] ) . the equality conditions in eq .
( [ optz2 ] ) then restrict the supports of the non - zero operators @xmath31 to certain known subspaces .
@xmath47 is the smallest value of @xmath73 that satisfies @xmath74 on the condition that the operators @xmath31 have their supports in the given subspaces .
@xmath42 is a full - rank operator for @xmath75 when @xmath2 decreases starting from 1 , at a certain value @xmath76 one of the @xmath1 positive eigenvalues of @xmath42 may become zero .
if an eigenvalue immediately turns negative when @xmath2 is decreased beyond @xmath76 , then @xmath76 is equal to @xmath47 . in this case
@xmath42 is a full - rank operator in the whole region @xmath77 , see the example at the end of sec .
ii a , but this need not be valid in general .
now we exploit the optimality conditions without the restriction to the limiting case of large @xmath2 . if @xmath78 , the equality in eq .
( [ optz1 ] ) can be only fulfilled when the support of the operator @xmath79 does not span the full hilbert space .
this implies that at least one of its eigenvalues is equal to zero and the determinant therefore vanishes , yielding the condition @xmath80 without lack of generality we suppose that in the optimal measurement the detection operators @xmath31 are different from zero for @xmath81 states ( @xmath82 ) and vanish for the remaining @xmath83 states , which means that the remaining states , if any , are never guessed to occur , @xmath84 in analogy to eq .
( [ det1 ] ) we then arrive at the condition @xmath85 the requirement that the determinants in eqs .
( [ det1 ] ) and ( [ det2 ] ) vanish yields a system of @xmath86 real equations with only @xmath87 real variables , namely the parameter @xmath25 and the @xmath88 real quantities determining the matrix elements of the hermitean operator @xmath24 acting in @xmath18 .
when the quantum states and their prior probabilities are completely arbitrary , the system of equations therefore does not have a solution for @xmath89 .
in special cases , however , a solution can exist where more than @xmath88 states are guessed to occur , for instance when the states have a certain symmetry . in these cases
we obtain the same operator @xmath24 and the same parameter @xmath25 when for some of the states we drop eq .
( [ det2 ] ) , keeping it for @xmath88 or less states , and when we put the detection operators corresponding to the dropped states equal to zero , see the examples treated in sec .
the measurement then differs from the one described by eq .
( [ op ] ) , but yields the same value @xmath90 when an optimal measurement exists where more than @xmath88 states are guessed , this measurement is therefore not unique .
the above considerations lead to the conclusion that we never need to make a guess for more than @xmath88 states in order to discriminate optimally between @xmath0 states in a @xmath1-dimensional hilbert space with a fixed rate @xmath2 of inconclusive results .
this conclusion generalizes previous results @xcite that have been obtained in a different way in the context of minimum - error discrimination , where @xmath4 .
it follows that for discriminating optimally with fixed @xmath2 between more than @xmath88 arbitrary states we have to consider all possible subsets @xmath91 containing @xmath88 states separately , where @xmath92 , and we have to find the maximal rates of correct results @xmath93 for discriminating the states within each subset .
the largest of these rates then determines the optimal solution for discriminating the @xmath0 states , in analogy to the findings obtained for the minimum - error discrimination of qubit states @xcite .
the operator @xmath24 and the parameter @xmath25 satisfying eqs .
( [ det1 ] ) - ( [ det2 ] ) only determine the optimal solution if the optimality conditions , eqs .
( [ optz1 ] ) and ( [ optz2 ] ) , can be fulfilled under the constraint that @xmath94 .
if for @xmath95 this is not the case , we have to try whether for @xmath96 a solution exists , and so on until @xmath97 where always the same state is guessed to occur when a conclusive outcome is obtained .
the described procedure will be applied in sec .
ii b for the case of qubit states .
an interesting special case arises when one of the states , say the first , has the property that @xmath98 for @xmath99 , implying that @xmath100 .
it is obvious that eqs .
( [ optz1 ] ) and ( [ optz2 ] ) and also the completeness relation , eq .
( [ compl-0 ] ) , are then fulfilled if @xmath101 the explicit expressions for @xmath42 and @xmath25 follow from replacing @xmath24 by @xmath102 in eq .
( [ optz1 ] ) , taking into account that @xmath103 and that the positivity constraint @xmath104 has to be satisfied . the condition @xmath105 restricts the validity of this solution to a certain region , and for this region it follows from eq .
( [ pcmax- ] ) that @xmath106 , see the example treated below .
clearly , for minimum - error discrimination , where @xmath28 , eq . ( [ ma ] ) yields @xmath107 , which means that the maximum probability of correct results can be obtained without performing any measurement , simply by always guessing the state @xmath108 with the largest prior probability to occur @xcite . in general
, it is very hard to obtain analytical solutions of the optimization problem with fixed @xmath2 for @xmath109 in our previous paper @xcite we derived the solution for a special class of linearly independent symmetric pure qudit states . in the following
we treat another example , which contains also the special case discussed above .
let us consider the optimal discrimination between a uniformly mixed qudit state @xmath110 and a pure qudit state @xmath111 , both living in the same hilbert space @xmath18 and occurring with the prior probabilities @xmath112 and @xmath113 , respectively . taking into account that the spectral representation of @xmath114 is given by @xmath115 where @xmath116 is a @xmath117-dimensional projector , we find that the optimality conditions , eqs .
( [ optz1 ] ) - ( [ optz2 ] ) , are satisfied for @xmath118 , or , equivalently , for @xmath119 , if @xmath120 while for @xmath121 , corresponding to @xmath122 , we obtain in agreement with eq .
( [ ma ] ) the solution @xmath123 in both cases we get @xmath124 . since the detection operators have to be positive , this solution , where @xmath125 , only holds true if @xmath126 with the help of eq .
( [ pcmax- ] ) we find that @xmath127 for @xmath4 this result agrees with the solution derived earlier for minimum - error discrimination @xcite .
note that for @xmath128 where @xmath129
the optimal measurement is not unique , as becomes obvious from comparing eqs .
( [ umix2 ] ) and ( [ umix3 ] ) .
it is easy to check that the relative rate of correct results , @xmath130 , grows for both lines of eq .
( [ umix1 ] ) when the fixed rate @xmath2 is increased , until it reaches unity for @xmath50 .
since according to eq .
( [ a - opt ] ) the maximum confidences for discriminating @xmath108 and @xmath131 are given by @xmath132 and @xmath133 respectively , eq .
( [ umix1 ] ) yields @xmath134 hence for @xmath50 the optimal measurement can be alternatively obtained when @xmath135 with @xmath136 , as becomes obvious from eq .
( [ lim2 ] ) . using eqs .
( [ optz1 ] ) - ( [ optz2 ] ) we find that the optimality conditions are satisfied if @xmath137 since the detection operators have to be positive , this result only holds true if @xmath138 that is @xmath139 we therefore arrive at @xmath140 which because of eq .
( [ pc ] ) expresses the fact that errors do not occur in the optimal measurement for @xmath141 hence in our example @xmath47 is the smallest rate of inconclusive results for which an unambiguous discrimination is possible .
( [ umix2 ] ) or ( [ umix3 ] ) , as well as eq .
( [ umix5 ] ) , show that for @xmath50 the optimal measurement is a projection measurement with @xmath142 , @xmath143 , and @xmath144 , which discriminates @xmath108 unambiguously , that is with the maximum confidence @xmath132 , and yields an inconclusive result when the state @xmath131 is present .
while the operator @xmath42 is a rank - one operator for @xmath145 , it follows from eq .
( [ umix5 ] ) that @xmath42 has the rank @xmath1 if @xmath146 that is in this example @xmath42 turns into a full - rank operator at @xmath50 . as mentioned after eq .
( [ lim3a ] ) , for @xmath147 the solution does not have any practical relevance since the relative rate of correct results remains constant with growing @xmath2 and its absolute rate @xmath148 decreases . in the rest of the paper we do not provide the explicit expressions of the detection operators for @xmath149 from now on we specialize our investigations on the optimal discrimination of qubit states in a two - dimensional joint hilbert space @xmath150
here it is sufficient to treat the case where the optimality conditions are satisfied with @xmath151 ( @xmath27 ) and @xmath152 when the first condition does not hold , the optimal measurement is determined by eq .
( [ ma ] ) , while the violation of the second condition corresponds to the limiting case of large @xmath2 where eq .
( [ lim2 ] ) applies .
following our earlier paper @xcite we conclude that for @xmath153 and @xmath154 the equalities in the optimality conditions , eqs .
( [ optz1 ] ) and ( [ optz2 ] ) , imply that the non - zero optimal detection operators @xmath42 and @xmath31 are proportional to the projector onto the eigenstates @xmath155 and @xmath156 belonging to the eigenvalue zero of the operators @xmath157 and @xmath158 , respectively .
hence if @xmath78 and @xmath159 for @xmath160 with @xmath82 , see eq .
( [ op ] ) , the optimality conditions require that @xmath161 @xmath162 , where @xmath163 and @xmath164 . since we supposed that @xmath165 for @xmath166 the completeness relation , eq .
( [ compl-0 ] ) , is given by @xmath167 due to eqs .
( [ detec0 ] ) and ( [ detec1 ] ) the projectors onto the normalized states @xmath168 and @xmath169 that are orthogonal to @xmath170 and @xmath171 , respectively , are determined by the operator @xmath24 and the parameter @xmath25 via the relations @xmath172 @xmath162 , where @xmath173 .
( [ compl ] ) we find that the completeness relation takes the alternative form @xmath174 for solving the optimization problem we have to determine the parameter @xmath25 and the operator @xmath24 which satisfy eqs .
( [ detec0 ] ) - ( [ compl ] ) , or , equivalently , eqs .
( [ perp ] ) - ( [ compl1 ] ) , for a certain value of @xmath81 on the condition that the constants @xmath175 are positive . for this purpose
we can proceed in two steps : in the first step we use eq .
( [ det1 ] ) together with the positivity constraint @xmath176 in order to express @xmath25 and @xmath42 in dependence of the matrix elements of @xmath177 taken with respect to an orthonormal basis in @xmath150 .
it is advantageous to choose the particular basis that is the eigenbasis of @xmath114 . introducing the spectral representation @xmath178 where @xmath179
, we represent the operator @xmath24 as @xmath180 the requirement @xmath181 see eq .
( [ det1 ] ) , then leads to @xmath182= |z_{01}|^2.$ ] it turns out that from the two possible solutions only the result @xmath183 ^ 2+\frac{|z_{01}|^2}{r(1-r)}}\ ] ] satisfies the inequality @xmath184 which implies that also @xmath176 since the determinant vanishes . after substituting this result for @xmath25 into eq .
( [ perp ] ) , we can express the completeness relation of the detection operators , eq .
( [ compl1 ] ) , as a function of @xmath2 and of the matrix elements of @xmath24 . in the second step
we apply eqs .
( [ det2 ] ) and ( [ compl1 ] ) , together with the positivity constraints @xmath185 the requirement @xmath186 valid for @xmath187 , leads to @xmath188=| z_{10}-\eta_j \rho_{10}^{(j)}|^2 \quad(j=1,\ldots , m),\ ] ] where we used the representation @xmath189 after introducing the abbreviations @xmath190 and after decomposing the non - diagonal matrix elements into their real and imaginary parts , eq .
( [ m - det ] ) can be rewritten as @xmath191 which yields for @xmath192 the equations @xmath193 when the completeness relation , given by eq .
( [ compl1 ] ) together with eq .
( [ perp ] ) , is written in matrix form we arrive at four real equations .
our task is to find the matrix elements of @xmath24 , determined by four real parameters , and the @xmath81 constants @xmath175 which together fulfill the @xmath194 equations resulting from eqs .
( [ m - det ] ) and ( [ compl1 ] ) .
provided that the solution satisfies the positivity constraints given by eq .
( [ constr ] ) , the optimization problem is solved . when the states are completely arbitrary , the system of @xmath81 equations given by eq .
( [ m - det ] ) can only have a solution if @xmath195 . therefore ,
when more than four completely arbitrary qubit states are to be optimally discriminated with fixed @xmath2 , we need to separately investigate all possible subsets containing four states , as described in sec .
ii for the discrimination of qudit states in @xmath196 we now focus on one such subset containing @xmath197 qubit states and discuss the different possible cases . _
( i ) guessing of all four states is optimal_. first we assume that @xmath198 .
( [ arb4 ] ) with @xmath199 then represents a system of three linear equations for determining the variables on their left - hand sides , @xmath200 , @xmath201 and @xmath202 , as linear functions of the variable @xmath203 occurring on the right - hand sides . if we insert the resulting expressions into eq .
( [ arb2 ] ) with @xmath204 we arrive at a quadratic equation for @xmath205 . after solving this equation
we get explicit expressions for the matrix elements of @xmath24 and , due to eq .
( [ a - sol ] ) , also for the parameter @xmath25 . by inserting these expressions into the four algebraic equations resulting from the matrix form of the completeness relation , we arrive at a linear system of equations for explicitly determining the constants @xmath206 .
provided that the conditions in eq .
( [ constr ] ) are satisfied , we have succeeded in determining the optimal solution . if one of the conditions is violated , we next investigate the case where @xmath207 . _
( ii ) guessing of three states is optimal .
we now assume that @xmath197 and @xmath207 , which means that in the optimal measurement one of the four states in the set , say the state with @xmath208 , is never guessed to occur , corresponding to @xmath209 . using eq .
( [ arb4 ] ) with @xmath210 together with eq .
( [ arb2 ] ) we can determine three of the four matrix elements of @xmath24 in dependence of the remaining one , say @xmath205 .
after inserting the resulting expressions into the matrix form of eq .
( [ compl1 ] ) we again arrive at four equations .
using three of them , the constants @xmath211 , and @xmath212 are obtained as functions of @xmath205 , and after inserting the results into the fourth we obtain an equation for determining @xmath205 which is , however , highly nonlinear for @xmath213 .
provided that the positivity conditions given by eq .
( [ constr ] ) with @xmath214 are satisfied , the solution has been obtained . _
( iii ) guessing of two states or of one state is optimal . _ if the positivity conditions can not be satisfied for @xmath215 or @xmath207 , a corresponding procedure has to be applied for @xmath216 .
it may happen that also in this case the positivity conditions can not be fulfilled and therefore @xmath97 in the optimal measurement , which means that it is a projection measurement where either a particular state is guessed to occur or an inconclusive result is obtained .
we mention at this place that in our previous paper @xcite the explicit solution for discriminating two mixed qubit states with fixed @xmath2 has been derived for cases where simple algebraic expressions result .
the density operators of the qubit states can be represented in the general form @xmath217 where @xmath218 is a normalized state vector , and where the parameters @xmath219 with @xmath220 depend on the purity of the respective states .
when the qubit states have equal prior probabilities and equal purities , that is when @xmath221 it is easy to check that the operator @xmath222 leads to @xmath223 .
hence the optimality condition given by eq .
( [ optz2 ] ) is satisfied if @xmath224 for @xmath27 . in order to exploit also the first optimality condition , eq .
( [ optz1 ] ) , we use the spectral representation of @xmath114 , given by eq .
( [ rho - diag ] ) , where we denote its eigenstates in such a way that @xmath225 . with @xmath226
it follows that eq .
( [ optz1 ] ) is fulfilled if @xmath227 and @xmath228 .
taking the completeness relation in the form of eq .
( [ compl ] ) into account and using eq .
( [ pcmax- ] ) , we find that the maximal probability of correct results is given by @xmath229 with @xmath230 . here
@xmath231 denotes the largest eigenvalue of @xmath232 , and @xmath233 is the corresponding eigenstate .
we note that for @xmath4 and @xmath234 our result agrees with the result obtained previously for the minimum - error discrimination of equiprobable pure states @xcite . when eq .
( [ compl - rel ] ) can not be satisfied with non - negative values of @xmath175 and eq .
( [ pmax ] ) therefore does not apply , @xmath235 depends on the given states @xmath218 not only via the eigenvalue @xmath231 , but also in a direct way , as will become obvious from the example investigated in sec .
iii c. moreover , even if for @xmath4 a solution with non - negative coefficients @xmath175 does exist , at least one of these coefficients may become negative if @xmath2 exceeds a certain critical value @xmath236 , which means that for @xmath237 eq .
( [ pmax ] ) is not valid anymore , see sec .
iii c. we mention that in our previous paper @xcite we considered the special case where @xmath238 , that is where @xmath239 and @xmath240 , for which we obtained the complete solution @xmath241 , valid for arbitrary values of @xmath2 .
we wish to study a discrimination problem that yields simple analytical solutions but is nevertheless sufficiently general to demonstrate the basic features of the method developed in sec .
ii . for this purpose
we consider the discrimination of @xmath0 mixed qubit states , represented by eq .
( [ rhoj - gen ] ) , which fall into two groups containing @xmath242 and @xmath243 states , respectively .
we suppose that within each group the qubit states have equal prior probabilities and purities , @xmath244 and that @xmath245 with @xmath246 where @xmath247 .
the diagonal density matrix elements @xmath248 of @xmath249 are then given by @xmath250 in addition , we assume that @xmath251 , and we consider the cases where @xmath252 eqs .
( [ rho - jj ] ) - ( [ zero - i ] ) imply that the non - diagonal density matrix elements of @xmath253 vanish and yield the spectral representation @xmath254 the state vectors @xmath218 in eq .
( [ psi - j ] ) can be visualized with the help of a bloch sphere where @xmath233 and @xmath255 are the north and south pole , respectively , that is @xmath256 with @xmath257 and @xmath258 .
clearly , in each group the end points of the bloch vectors representing the underlying states @xmath218 have the same latitude , lying in the upper hemisphere if @xmath259 and in the lower hemisphere if @xmath260 , see fig .
the phases can be represented as @xmath261 where @xmath262 is an arbitrary phase shift .
clearly , the density operators obey the symmetry conditions @xmath263 where the unitary operator @xmath264 with @xmath265 characterizes the symmetry in the first group of states , and @xmath266 with @xmath267 refers to the second group . in the case
( ii ) we put @xmath268 . the cases ( i ) and
( ii ) belong to the specific kind of partially symmetric states introduced in our earlier paper , ref .
@xcite , while in the case ( iii ) the @xmath0 states are symmetric . for later purposes we still calculate the maximum confidence @xmath58 for discriminating an individual state @xmath60 , using eq .
( [ a - opt ] ) .
we get @xmath269 for @xmath270 , and @xmath271 for @xmath272 , where @xmath273 ^ 2}}\right],\ ] ] and where @xmath274 follows from eq .
( [ maxconf1 ] ) when @xmath275 , and @xmath276 are replaced by @xmath277 , and @xmath278 , respectively . in the special case where @xmath279 , meaning that @xmath280 and @xmath281 the expression for @xmath282 is in agreement with our earlier result @xcite for the maximum - confidence discrimination of symmetric mixed qubit states . in order to solve our optimization problem
we apply the method outlined in sec .
ii . due to the partial symmetry of the states , the treatment can be considerably simplified .
in fact , assuming that at least from one of the two groups , say the first , two or more states are guessed to occur , we get for these states from eq .
( [ m - det ] ) the conditions @xmath283=| z_{10}-\eta \rho_{10}^{(j)}|^2 $ ] with @xmath284 since these conditions can only be simultaneously satisfied for two or more different phases @xmath285 if @xmath286 , we have to search for an operator @xmath24 where @xmath287=0.\ ] ] we note that the requirement @xmath288=0 $ ] was derived already in our earlier paper @xcite , using the fact that there always exists an optimal measurement where the detection operators @xmath31 obey the same symmetry as the density operators @xmath10 of the partially symmetric states . as mentioned in the beginning of sec .
ii b , it is sufficient to perform the derivation for the case where @xmath38 and @xmath289 @xmath290 . in the first step
we make use of eq .
( [ det1 ] ) . since @xmath286 ,
the condition @xmath291 is equivalent to @xmath182= 0 $ ] . taking the positivity constraint @xmath292 and the relation @xmath103 into account
, we find that the optimality condition in eq .
( [ optz1 ] ) is satisfied if @xmath293 where the expression for @xmath25 is in accordance with eq .
( [ a - sol ] ) .
note that if @xmath294 both eigenvalues of @xmath79 vanish which means that @xmath135 and which corresponds to the case @xmath295 , yielding eq .
( [ lim2 ] ) . due to eq .
( [ pcmax- ] ) the maximum probability of correct results for @xmath48 that is for @xmath296 takes the form @xmath297 in the second step we apply eq .
( [ det2 ] ) and the completeness relation , represented by eq .
( [ compl1 ] ) , together with the positivity constraints given by eq .
( [ constr ] ) .
( [ m - det ] ) we arrive at the conditions @xmath298=\eta^2 p^2\ , b(1\!-b)\,\;\;\;{\rm if}\;a_1>0,\qquad\\ ( z_{00}\!-\!\eta^{\prime } s^{\prime})[z_{11}\ !
-\!\eta^{\prime}(1\!-\!s^{\prime})]=\eta^{\prime 2}p^{\prime 2}c(1\!-\!c)\;\;\,\,{\rm if}\;a_2>0,\quad \end{array}\ ] ] where we introduced @xmath299 because of eq .
( [ detec1 ] ) the condition @xmath300 is equivalent to requiring that at least one state from the first group is guessed , and @xmath301 means that at least one state from the second group is inferred to occur .
next we make use of eqs .
( [ compl1 ] ) and ( [ perp ] ) .
because of the explicit form of the operator @xmath42 , given by eq .
( [ pcmax1 ] ) , the diagonal matrix elements of eq .
( [ compl1 ] ) lead to the conditions @xmath302 where the constants @xmath303 and @xmath304 read @xmath305 since @xmath306 the non - diagonal elements of eq .
( [ compl1 ] ) yield the condition @xmath307 due to the symmetry properties of the partially symmetric states , eq .
( [ compl2c ] ) is certainly fulfilled if @xmath308 that is if in the optimal measurement the detection operators for the individual states obey the same symmetry as their density operators , as can be seen with the help of eqs .
( [ detec11 ] ) and ( [ perp ] ) .
however , other choices of the coefficients @xmath175 and therefore of the detection operators are also possible .
taking into account that @xmath309 for @xmath310 see eq .
( [ phi ] ) , we find for instance that @xmath311 if for even numbers @xmath242 @xmath312 and for odd numbers @xmath242 @xmath313 while all other coefficients @xmath175 @xmath314 vanish .
similarly , various solutions with @xmath315 exist for the coefficients @xmath175 that belong to the states in the second group and depend on @xmath316 . moreover , if @xmath317 eq .
( [ compl2c ] ) may be also solved in a way where the terms referring to the two groups of states do not vanish separately . in order to find an optimal measurement we have to determine the values of @xmath205 , @xmath200 , @xmath318 and @xmath316 that satisfy eqs .
( [ two1a ] ) - ( [ f ] ) with @xmath319 and @xmath320 , where according to eq .
( [ optz2 ] ) the positivity constraints @xmath321 have to be fulfilled . for @xmath317 ,
where both lines of eq .
( [ two1a ] ) apply , the positivity constraints are satisfied provided that @xmath322 while for @xmath300 and @xmath323 the additional condition @xmath324\geq \eta^{\prime 2 } p^{\prime 2 } c(1-c)\ ] ] has to be taken into account .
a similar condition arises when @xmath301 and @xmath325 since in general the resulting solutions are rather involved , we treat the general case only in implicit terms and provide explicit derivations of special solutions in the appendix .
two different cases have to be considered : _ ( i ) guessing states from both groups is optimal .
_ for @xmath326 eq .
( [ two1a ] ) represents a system of two coupled equations from which we obtain expressions for @xmath205 and @xmath200 after solving a quadratic equation . when these expressions are inserted into eqs .
( [ compl2a ] ) and ( [ compl2b ] ) , the resulting linear system of equations yields the values of @xmath318 and @xmath316 , which depend on @xmath2 due to eq .
( [ f ] ) . provided that these values are indeed positive and that eq .
( [ constr2 ] ) is satisfied , we have obtained the optimal solution .
if this is not the case , a measurement where states from both groups are guessed can not be optimal . _
( ii ) guessing states from only one group is optimal . _ to be specific , let us assume that only the states from the first group are guessed , which means that @xmath300 and @xmath323 . from eqs .
( [ compl2a ] ) and ( [ compl2b ] ) with @xmath323 we obtain the two equations @xmath327.\ ] ] using in addition the first line of eq .
( [ two1a ] ) and taking eq .
( [ f ] ) into account , we arrive at three equations for determining @xmath328 and @xmath318 in dependence of @xmath2 . provided that indeed @xmath300 and that eqs .
( [ constr2 ] ) and ( [ constr2a ] ) are satisfied , the optimal solution has been determined . for the case where only the states from the second group are guessed we can proceed in an analogous way . in the next two subsections we present the complete solutions for two simplified but non - trivial discrimination problems for partially symmetric states . in our first problem
we suppose that the purity of the states in the two groups is the same , @xmath329 , and that the states occur with the same prior probability , @xmath330 .
we denote the eigenstates of @xmath114 in such a way that @xmath231 , given by eq .
( [ rhodiag ] ) , is the largest eigenvalue , @xmath331 \geq \frac{1}{2}.\ ] ] to be specific , we suppose that @xmath332 which implies that @xmath333 if @xmath334 . with the help of eq .
( [ maxconf1 ] ) this yields the relation @xmath335 for the maximum achievable confidences for discriminating the states in the two groups , where the equality sign holds if @xmath240 . as derived in the appendix , the maximum probability of correct results with fixed @xmath2
is then given by the solution @xmath336 where @xmath337 and @xmath338 , in agreement with eq .
( [ pmax ] ) , while @xmath47 and @xmath339 are defined by eqs .
( [ solq3 ] ) and ( [ a-2 ] ) in the appendix , respectively .
three different regions of @xmath2 have to be distinguished : for @xmath340 , corresponding to the full lines in fig .
2 , states from both groups are guessed in the optimal measurement
. clearly , this only applies if @xmath341 , that is if @xmath342 .
@xmath32 is then determined by @xmath343 , in agreement with the general result derived in eq .
( [ pmax ] ) . in this region
the optimum detection operators are @xmath344 and @xmath345 for @xmath346 , where the states @xmath218 are given by eq .
( [ psi - j ] ) .
the values of the constants @xmath175 can be determined by applying the method described in sec .
iii b , using the explicit expressions for @xmath318 and @xmath316 given in the appendix . in the region where @xmath347 , corresponding to the dashed lines in fig .
2 , only the states of the first group are guessed in the optimal measurement .
the optimal detection operators follow from the expression for @xmath24 , see eq .
( [ z0 ] ) , and from eqs .
( [ perp ] ) and ( [ detec11 ] ) .
the constants @xmath175 can be obtained as described sec .
iii b , with the help of the value of @xmath318 given by eq .
( [ a1 ] ) . if @xmath348 and therefore @xmath349 the second line of eq .
( [ solq ] ) holds true in the whole region @xmath350 . since @xmath332 , the bloch vectors of the states
are then confined to the upper hemisphere of the bloch sphere .
the region where @xmath295 corresponds to the limiting case of large @xmath2 described in sec .
ii a. in this region the ratio @xmath351 , does not increase anymore with growing @xmath2 but stays equal to @xmath282 , see the dotted lines in fig .
2 . the squares in fig . 2 indicate the points where @xmath50 for different numbers of @xmath242 and @xmath352 .
two special cases are worth mentioning . for @xmath4 ,
that is for minimum - error discrimination , we obtain from eq .
( [ solq ] ) @xmath353 when the first line applies , states from both groups are guessed in the optimal measurement , while otherwise only states from the first group are guessed to occur .
interestingly , the solution only depends on the total number @xmath354 of the states and is independent of their distribution over the two groups .
the second special case refers to @xmath279 , that is to the discrimination of @xmath355 equiprobable symmetric mixed qubit states .
( [ r1 ] ) then reduces to @xmath356 and requires that @xmath357 .
this means that the first line of eq .
( [ solq ] ) does not apply , and that @xmath339 is determined by the first line of eq .
( [ a-2 ] ) .
this solution coincides with our earlier result @xcite that was derived for symmetric detection operators .
in contrast to this , the present derivation does not impose any restriction on the choice of the detection operators .
in particular , it shows that for @xmath0 symmetric mixed qubit states an optimal discrimination with fixed @xmath2 can be always accomplished when not more than three ( for @xmath0 odd ) or two ( for @xmath0 even ) states are guessed to occur , as follows from eqs .
( [ even ] ) and ( [ odd ] ) . for @xmath4 , that is for minimum - error discrimination , the optimum measurement therefore can be always realized by a simple projection measurement if @xmath0 is even .
now we assume that the states are pure , @xmath358 , and that the respective prior probabilities @xmath276 and @xmath278 in the two groups of states can be different , where @xmath359 . in order to obtain the complete solution for arbitrary values of @xmath276 we have to consider both the cases where the relations @xmath360 and @xmath361 hold true between the maximum confidences for the states in the two groups , see eqs .
( [ pure4a ] ) and ( [ pcmax2 ] ) in the appendix .
we present explicit results for the simplest case , where @xmath362 , @xmath363 and @xmath364 , that is where the states are given by latexmath:[\[\label{mirror } and occur with the prior probabilities @xmath276 for @xmath366 and @xmath367 , and @xmath368 for @xmath369 respectively .
these states correspond to the three mirror - symmetric pure states the minimum - error discrimination of which has been previously investigated @xcite .
the condition @xmath360 holds true provided that @xmath370 , where for @xmath371 we get @xmath372 as shown in the appendix , for @xmath360 the maximum probability of correct results reads @xmath373 where @xmath47 is determined by eqs .
( [ solq3 ] ) and ( [ solq2a ] ) with @xmath234 , and where @xmath236 , @xmath343 , and @xmath339 are given by eqs .
( [ crit ] ) - ( [ mirror1a ] ) with @xmath374 .
on the other hand , for @xmath375 we obtain @xmath376 where @xmath377 , @xmath378 and @xmath379 are determined by eqs .
( [ mirror1a ] ) - ( [ mirror3 ] ) . in the regions of @xmath2 where the solution is given by @xmath339 , corresponding to the dashed lines in fig .
3 , only the states @xmath380 and @xmath381 are guessed to occur in the optimal measurement . on the other hand ,
all three states are guessed if @xmath382 for @xmath383 and if @xmath384 for @xmath375 , which corresponds to the full lines in fig .
3 . for @xmath385
we find that @xmath375 if @xmath386 as outlined in the appendix , for @xmath387 the first line of eq .
( [ pure3prime ] ) does not apply , and the second line is valid in the whole range @xmath388 since @xmath389 the dotted lines in fig .
3 correspond to the limiting case of large @xmath2 , and the squares indicate the points where this limiting case is reached . for minimum - error discrimination , where @xmath4 , the maximum probability of correct results does not depend on the relation between @xmath282 and @xmath274 .
( [ solq1 ] ) and ( [ pure3prime ] ) yield @xmath390 \ ; & \mbox{if $ \ ; \eta\leq \eta_{cr } $ } \quad\\ \!\!\eta\left(1 + 2\sqrt{b(1-b}\right ) \ ; & \mbox{if $ \ ; \eta \geq \eta_{cr } $ } , \quad\\ \end{array } \right.\ ] ] which coincides with the result obtained already in ref .
when @xmath391 all three states are guessed in the measurement performing minimum - error discrimination , while otherwise only the states @xmath380 and @xmath381 are guessed to occur .
before concluding the paper , we briefly discuss the relation of our method to previous investigations of the minimum - error discrimination of arbitrary qubit states , where @xmath4 . from eq .
( [ perp ] ) we obtain the representation @xmath392 @xmath393 , where due to eq .
( [ compl1 ] ) for minimum - error discrimination the condition @xmath394 has to be satisfied with non - negative values of @xmath175 . upon eliminating the operator @xmath24 , we obtain from the first equality in eq .
( [ mx ] ) for any pair of states the equation @xmath395 ( @xmath396 ) .
( [ my ] ) has been recently derived in an alternative way @xcite , and has been applied to study the minimum - error discrimination of qubit states using a geometric formulation @xcite .
in contrast to this , in our method , which refers to the general case @xmath397 , the operator @xmath24 is not eliminated . rather , our approach essentially rests on determining @xmath24 and is therefore for @xmath4 related to the treatments of minimum - error discrimination in refs . @xcite,@xcite and @xcite . in this paper
we investigated the discrimination of @xmath0 mixed quantum states by an optimal measurement that yields the maximum probability of correct results , @xmath32 , while the probability of inconclusive results is fixed at a given value @xmath397 .
for the discrimination of qudit states in a @xmath1-dimensional hilbert space , we discussed the general properties of the optimal measurement .
moreover , we derived the analytical solution for optimally discriminating with fixed @xmath2 between a uniformly mixed and a pure qudit state . in the main part of the paper we specialized on the optimal discrimination of qubit states in a two - dimensional hilbert space and developed a general method to obtain the solution .
we studied the special case where the prior probabilities of the qubit states are equal , and we also treated the discrimination between four or less arbitrary qubit states with fixed @xmath2 .
as an illustrative application of our method , we derived explicit analytical results for discriminating qubit states which posses a partial symmetry .
we emphasize that apart from determining @xmath32 , our method also allows to consider the various possible realizations of the optimal measurement for a given discrimination problem .
in particular , we found that for discriminating @xmath0 symmetric qubit states the maximum probability of correct results with fixed @xmath2 can be for instance also achieved by a measurement where only three of the states are guessed to occur when @xmath0 is odd , and only two of the states when @xmath0 is even , instead of guessing all @xmath0 states .
note added : after submitting this work a related paper @xcite appeared .
in this appendix we provide the detailed derivations for the results presented in sec .
we start by treating the case where only states from the first group are guessed , that is @xmath398 using eq .
( [ first ] ) and the first line of eq .
( [ two1a ] ) , we obtain @xmath399 with @xmath400 and @xmath401 . from eq .
( [ z0 ] ) it follows that the condition @xmath402 is equivalent to @xmath403 where @xmath404^{2}\ ] ] with @xmath405 for @xmath234 . making use of eq .
( [ f ] ) we therefore arrive at @xmath406 and due to eq .
( [ first ] ) we obtain the explicit result @xmath407 in the upper line we took into account that @xmath408 corresponds to @xmath409 which implies that @xmath410 , and we used a similar relation for the lower line . since @xmath397 , eq .
( [ solq2 ] ) can only be fulfilled and hence eq . ( [ z0 ] ) can only determine the optimal solution if @xmath411 where @xmath50 corresponds to @xmath412 from eqs .
( [ z0 ] ) and ( [ sol ] ) we obtain the maximum probability of correct results @xmath413 with @xmath414 since the expressions in the upper and lower lines of eqs .
( [ solq2 ] ) - ( [ a-2 ] ) are identical for @xmath415 and @xmath416 we can replace the conditions @xmath417 by @xmath418 , and similarly @xmath419 by @xmath420 , if we extend the restriction @xmath45 to @xmath145 .
( [ a-2 ] ) describes the optimal solution provided that @xmath205 and @xmath421 determined by eq .
( [ z0 ] ) , satisfy the positivity constraints given by eq .
( [ constr1 ] ) .
analogous results can be obtained for the case where only states from the second group are guessed to occur , that is where @xmath422 .
we still need to study the case where states from both groups are guessed in the optimal measurement , that is where @xmath300 and @xmath301 .
we shall do this in the following , where we derive the complete solutions for the problems discussed in secs .
iii c and iii d. _ equiprobable states with equal purity .
_ in accordance with eq .
( [ r1 ] ) we assume that @xmath334 and @xmath332 , which for @xmath423 and @xmath329 corresponds to @xmath335 . supposing that states from both groups are guessed , that is @xmath326 , we obtain from eq .
( [ two1a ] ) the solution @xmath424 which clearly satisfies the positivity constraints given by eq .
( [ constr1 ] ) . taking into account that @xmath425 , eqs .
( [ compl2a ] ) - ( [ f ] ) yield the constants @xmath426 and @xmath427 . since @xmath333 , cf .
( [ r1 ] ) , @xmath318 can not be negative .
on the other hand , the condition @xmath320 requires that @xmath2 does not exceed the critical value @xmath428 . with the help of eq .
( [ sol ] ) we thus arrive at the first line of eq .
( [ solq ] ) . the second line of eq .
( [ solq ] ) refers to the case where the solution determined by eqs .
( [ z0 ] ) and ( [ a-2 ] ) is optimal , as will be shown below by verifying that eq .
( [ constr1 ] ) is satisfied .
the third line corresponds to eq .
( [ lim2 ] ) . with the help of eq .
( [ solq2a ] ) we find after a little algebra that @xmath429
if @xmath334 and @xmath332 . hence for @xmath430 it follows that @xmath431 which means that for @xmath432 always the first line of eq .
( [ a-2 ] ) applies .
it remains to be shown that for @xmath347 the positivity constraints given by eq .
( [ constr1 ] ) are fulfilled when @xmath205 and @xmath200 are determined by eq .
( [ z0 ] ) . from eqs . ( [ constr2 ] ) and ( [ constr2a ] ) with @xmath433 and @xmath329
we find after minor algebra that eq .
( [ constr1 ] ) is satisfied if @xmath434 , which because of eq . ( [ solq2 ] ) yields the two conditions @xmath435 if @xmath418 and @xmath436 if @xmath437 the first condition requires that @xmath438 .
the second condition is fulfilled for @xmath439 , as becomes obvious from the second line of eq .
( [ solq2 ] ) and from the validitiy of the relation @xmath440 hence eq . ( [ constr1 ] )
is indeed satisfied and we have derived eq .
( [ solq ] ) . _ pure states with different prior probabilities . _ for @xmath358 we obtain with the help of eq .
( [ maxconf1 ] ) the relation @xmath441 where @xmath442 and @xmath443 , while @xmath444 with @xmath359 .
supposing that states from both groups are guessed , we get from eq .
( [ two1a ] ) the solutions @xmath445 and @xmath446 which have to positive when eq .
( [ constr2 ] ) holds . to be specific
, we again assume that @xmath332 .
the condition @xmath447 then implies that @xmath448 , and a corresponding relation is valid when @xmath449 . from eq .
( [ sol ] ) we obtain the solution @xmath450 which holds true for certain regions of @xmath2 where the positivity constraint given by eq .
( [ constr1 ] ) is satisfied and where eqs .
( [ compl2a ] ) - ( [ f ] ) , resulting from the completeness relation , can be fulfilled with positive values of @xmath318 and @xmath316 . outside these regions
only states from one of the groups will be guessed in the optimal measurement .
now we specialize on the discrimination of three mirror - symmetric pure states , given by eq .
( [ mirror ] ) , where @xmath368 and @xmath451 .
( [ maxconf1 ] ) yields the respective maximum confidences @xmath452 for @xmath380 and @xmath381 , and @xmath453 for the state @xmath454 .
provided that all three states are guessed , that is @xmath455 , we obtain from eq .
( [ two1a ] ) the solution @xmath456 and @xmath457 . for @xmath360 ,
that is for @xmath458 , it follows that the relation @xmath459 holds true when all three states are guessed .
( [ compl2a ] ) - ( [ f ] ) then yield the solutions @xmath460 and @xmath461 . here
we introduced a critical value @xmath236 , given by @xmath462 ^ 2 } , \quad{\rm where } \;\ ; q_{cr}\geq0\;\ ; { \rm if}\;\;\eta \leq \eta_{cr}\ ] ] with @xmath463^{-1}$ ] . while @xmath318 is always positive , the condition @xmath320 requires that @xmath464 due to eq .
( [ sol ] ) we find that @xmath465 for @xmath466 where @xmath467.\ ] ] clearly , this result only applies when @xmath468 or @xmath469 , respectively , which implies that @xmath470 and that eq .
( [ constr1 ] ) is therefore satisfied .
as will be shown below , for @xmath471 a measurement is optimal where @xmath323 , that is where eq .
( [ z0 ] ) determines the solution . extending eq .
( [ a-2 ] ) to the case @xmath415 , we obtain @xmath472 for @xmath473 , where @xmath474 hence we have obtained the first two lines of eq .
( [ solq1 ] ) , and the third line follows again from eq .
( [ lim2 ] ) . in eq .
( [ mirror1a ] ) we took into account that with @xmath475 the condition @xmath476 is equivalent to the condition @xmath477 .
we note that for @xmath478 the condition @xmath479 , or @xmath480 , respectively , can be only fulfilled if @xmath481 , which requires that @xmath482 for @xmath483 we therefore get @xmath484 , or @xmath485 , respectively , which means that the solution is given by the lower line of eq .
( [ mirror1a ] ) in the whole range @xmath486 next we treat the case where @xmath375 , or @xmath487 , respectively , which means that @xmath488 if all three states are guessed . from eqs .
( [ compl2a ] ) - ( [ f ] ) we obtain @xmath489 and @xmath490 .
it is useful to introduce @xmath378 with @xmath491 , that is @xmath492 ^ 2}{\eta^2b(1- b ) } , \quad{\rm where } \;\ ; q_{cr}^{\prime}\geq 0\;\ ; { \rm if}\;\;\eta \geq \eta_{cr}.\ ] ] provided that @xmath493 the conditions @xmath319 and @xmath494 hold true .
( [ sol ] ) we therefore get @xmath495 for @xmath496 , where @xmath497.\ ] ] thus we have obtained the second line of eq .
( [ pure3prime ] ) . for @xmath498 , where @xmath422 , only the state @xmath454 is guessed . from eq .
( [ mirror3 ] ) it follows that @xmath499 when @xmath500 this means that in eq .
( [ lim2 ] ) @xmath47 corresponds to @xmath501 , and we arrive at the third line of eq .
( [ pure3prime ] ) . on the other hand , for @xmath502
we get @xmath398 as will be shown below , when @xmath503 a measurement is optimal where @xmath323 and where the solution is therefore determined by @xmath339 , given by eq .
( [ mirror1a ] ) .
this yields the first line of eq .
( [ pure3prime ] ) . for @xmath504
the condition @xmath505 , or @xmath506 , respectively , can be only fulfilled if @xmath507 , which requires that @xmath508 for @xmath509 we therefore get @xmath510 , which means that the solution is given by eq .
( [ mirror3 ] ) in the whole range @xmath511 .
let us finally consider the positivity constraints given by eq .
( [ constr1 ] ) for the case where the solution is described by @xmath339 , that is where @xmath205 and @xmath200 are determined by eq .
( [ z0 ] ) .
making use of eqs .
( [ constr2 ] ) and ( [ constr2a ] ) with @xmath234 and @xmath364 we find after minor algebra that eq .
( [ constr1 ] ) is satisfied for @xmath512 ^ 2 } .$ ] taking the definition of @xmath236 into account , we therefore obtain from eq . ( [ solq2 ] ) the two conditions @xmath513 if @xmath418 and @xmath514 if @xmath515 for @xmath360 , or @xmath516 , respectively , the first condition requires that @xmath438 , while the second condition is always fulfilled for @xmath517 , as follows from the second line of eq .
( [ solq2 ] ) and from the relation @xmath518 which holds for @xmath516 . for @xmath375
the second condition requires that @xmath519 since @xmath520 , which means that it only applies if @xmath521 , or @xmath522 , respectively , where the first condition is always satisfied .
( [ constr1 ] ) indeed restricts the range of validity of eq .
( [ z0 ] ) to the regions @xmath523 if @xmath375 and @xmath473 if @xmath524
the author acknowledges partial financial support by deutsche forschungsgemeinschaft dfg ( sfb 787 ) . | we study the discrimination of @xmath0 mixed quantum states in an optimal measurement that maximizes the probability of correct results while the probability of inconclusive results is fixed at a given value .
after considering the discrimination of @xmath0 states in a @xmath1-dimensional hilbert space , we focus on the discrimination of qubit states .
we develop a method to determine an optimal measurement for discriminating arbitrary qubit states , taking into account that often the optimal measurement is not unique and the maximum probability of correct results can be achieved by several different measurements .
analytical results are derived for a number of examples , mostly for the discrimination between qubit states which possess a partial symmetry , but also for discriminating @xmath0 equiprobable qubit states and for the dicrimination between a pure and a uniformly mixed state in @xmath1 dimensions . in the special case where the fixed rate of inconclusive results is equal to zero
, our method provides a treatment for the minimum - error discrimination of arbitrary qubit states which differs from previous approaches . |
distributions of galaxies in the sky were found useful in studying large - scale structures of the universe ( seldner et al .
1977 ; gellar & huchra 1989 ; loveday et al .
owing to redshift surveys available in different eras , many investigations were made to detect large - scale structures in both the projected map and the three - dimensional space .
there were many discoveries of clustering features in 80 s , including voids , galaxy clusters , strings , great attractors , and various great walls ( kirshner et al . 1981 ; bahcall & soneira 1983 ; haynes & giovarelli 1986 ; lynden - bell et al . 1988 ; gellar & huchra 1989 ) . in as early as 1989
, the cfa `` great wall '' was found to extend to @xmath9 ( gellar & huchra 1989 ) , while in recent years the sloan great wall was observed to be stretched to @xmath10 in length by gott iii et al .
( 2005 ) . ] and by deng et al .
hereafter in this section , when a cosmological model is adopted in a cited paper , the parameter will be @xmath11 unless otherwise specified .
] respectively .
the latter is the largest structure of the universe observed so far .
observations also revealed that large - scale structures could be found deep in the space where redshift is high ( shimasaku et al .
2003 ; matsuda et al . 2005
; ouchi et al .
investigation of large - scale structures of the universe meets a new chance owing to two recent redshift surveys , the 2-degree field galaxy redshift survey ( 2dfgrs ) ( colless et al .
2001 ) and the sloan digital sky survey ( sdss ) ( york et al . 2000 ) which provide large amount and reliable data .
large - scale structures of the local universe are of particular importance since theories of structure formation could be directly checked by redshift surveys .
it was predicted that large - scale filamentary or sheet - like mass overdense regions would preferentially be formed in the early universe and later evolve into dense clusters of galaxies which would be expected in the local universe , and gravitational amplification of the matter density fluctuations that are generated in the early universe is assumed to be responsible for the formation of the clustering structure of the present universe ( see peebles 1982 ; blumenthal et al .
1984 ; davis et al . 1985 , 1992 ; governato et al . 1998 ; kauffmann et al . 1999 ; cen & ostriker 2000 ; benson et al . 2001 ; colberg et al .
filamentary features were found to be real when some redshift surveys were carefully checked ( bhavsar & ling 1988 ; bharadwaj et al .
2000 ; bagchi et al .
2002 ; ebeling et al .
2004 ; pimbblet et al .
2004 ; porter & raychaudhury 2005 ) .
the detection of a large concentration of primeval galaxies at redshift @xmath12 ( steidel et al .
1998 ) and the reported discovery of a large - scale coherent filamentary structure of ly@xmath13 emitters in a redshift space at @xmath14 favor the prediction that galaxies preferentially formed in large - scale filamentary or sheet - like mass overdensities in the early universe ( matsuda et al .
2005 ) . in agreement with these ,
protoclusters which have been presumed not to have virialized due to their short cosmological ages ( e.g. venemans 2005 ) were identified at as far as @xmath15 ( venemans et al .
2002 ; intema et al .
2006 ) .
the great success of cold dark matter ( cdm ) model in recent years makes it a leading model in current studies of cosmology .
the charm of the cdm model is its predictive power on the formation of large - scale structures .
however , in this model , large - scale structures up to @xmath16 are rarely formed , which are challenged by recent observations ( see yoshida et al .
2001 ; yoshida 2005 ) .
it suggests that the cdm model predicts smaller homogeneity scale of the universe than what the observations have revealed ( note that the large - scale structure of the universe was found to be several hundred @xmath17 ) . in addition
, the model faces difficulty in explaining the observed structure on length scales @xmath18 as well ( see a brief review in cembranos et al . 2005 ) .
statistical methods are useful in studying large - scale structures of the universe . as mentioned in martinez & saar ( 2002a ) , most of the statistical analysis of the galaxy distribution proposed so far are based on second order methods ( correlation functions and power spectra ) ( see also diggle 1983 ) . among them , the two - point correlation function ( totsuji & kihara 1969 ; peebles 1974 , 1980 ; davis & peebles 1983 ) and the power spectrum ( fisher et al . 1993 ; feldman et al . 1994 ; park et al . 1994 ; tadros & efstathiou 1996 ) are the tools most often used in both observational and theoretical analyses .
the latter is the fourier transform of the former .
since the two quantities are a fourier transform pair , complete knowledge of one is equivalent to complete knowledge of the other .
but this is not true for their estimators when samples employed are finite and noisy ( feldman et al .
the two - point correlation function @xmath19 , which was first used to measure the strength of galaxy clustering for redshift surveys by davis & peebles ( 1983 ) , describes the excess probability of finding a galaxy in a volume element at a separation @xmath20 from another randomly chosen galaxy above that for an un - clustered distribution .
there are various estimators of @xmath21 in literature .
the main difference is their corrections for the edge effect ( martinez & saar 2002a ) .
the quantity was found to obey a power law : @xmath22 on small scales ( @xmath23 ) ( for early works , see totsuji & kihara 1969 ; davis & geller 1976 ; davis & peebles 1983 ) . in recent investigations ,
the indexes were found to be @xmath24 and @xmath25 for the sdss data and @xmath26 and @xmath27 for the 2dfgrs data ( zehavi et al .
2002 ; hawkins et al . 2003 ) . as pointed out by yadav et al .
( 2005 ) , the power law relation of @xmath21 does not hold on large scales and it breaks down at @xmath28 for sdss and at @xmath29 for 2dfgrs , and thus it would not violate the homogeneous nature of the universe . an advantage of the two - point correlation function is its application to galaxy groups to reveal the matter distribution in the universe , since the occupation of haloes by galaxies is believed to depend on halo mass ( see benson et al .
2000 ; berlind et al .
2003 ) . as each dark matter halo
is expected to give birth to a single group of galaxies , applying the two - point correlation function to different groups of galaxies one can reveal the clustering nature of these groups and this in turn can tell how dark matter is distributed ( see , e.g. , jing & zhang 1988 ; merchan et al .
2000 ; zandivarez et al . 2003
; padilla et al . 2004 ) . in measuring the scale of homogeneity of the universe , the power spectrum is applicable ( see einasto & gramann 1993 ) .
martinez et al . ( 1998 ) introduced the @xmath30 function @xmath31 , related to the integral of the two - point correlation function , as a tool .
one of their main claims is that the estimators for @xmath31 are more reliable than the most currently used estimators for @xmath21 and that makes the use of @xmath31 recommendable ( especially in three - dimensional processes and at large scales ) despite its somewhat less informative character ( see martinez et al .
1998 ) . in the following one will find that the concept of the @xmath30 function is also useful in describing the crowded nature of the neighborhood of individual sources and hence able to reveal the density property of a group of sources .
based on the previous methods , we develop in this paper a statistical approach to sort out clustering sources .
our main concern includes : crowded degrees of sources in their neighborhoods are quantified so that one can tell how different crowded sources form a cluster ; redshift dependent densities of samples are taken into account to sort out clusters so that the relevant statistical significance could be determined ; via simulation , probabilities for forming clusters by chance are calculated .
the method is applied to the 2dfgrs data set . as the results ,
some fine maps of large - scale structures derived from the sample are available and probabilities of a list of clusters are obtained .
the paper is organized as follows . in section 2 ,
we propose the method , where a statistic called the individual neighborhood function @xmath32 is introduced to describe the crowded nature of individual sources . in section 3
, we describe the selection of the sample and study the redshift distribution it obeys .
density distributions of the sample , in terms of @xmath32 , are presented in section 4 .
clustering probabilities obtained by simulation are studied in section 5 .
coherent clusters are identified and shown in section 6 . in section 7 , we show the structures of some large prominent coherent clusters . in section 8 ,
the largest structures sorted out from the whole 2df data set are shown and discussed . in the last section ( section 9 ) , a summary of the paper and a brief discussion are presented .
there are two well - known methods for identifying galaxy groups or clusters .
the most popular one is used to create galaxy group catalogues , where a projected separation together with a velocity difference are employed to search for companion sources ( huchra & geller 1982 ; eke et al .
2004 ; diaz et al . 2005 ; merchan & zandivarez 2005 ) .
a product of this method is the property associated with the virial theorem ( e.g. , the velocity dispersion of a set of sources ) , and it is suitable to identify virialized group of galaxies .
the other is used to pick out large - scale structures , where a single scale , which is applied to all sources of the sample , is employed to identify a neighborhood source ( einasto et al .
1984 ; deng et al . 2006 ) .
the main concern of this method is the statistical property ( e.g. , the length or density ) of the identified clusters . in both approaches , the so - called `` friends - of - friends '' ( fof ) algorithm , the claim `` any friend of my friend is my friend '' , is applied .
the latter approach is somewhat suitable for our analysis .
however , besides picking out clustering sources , we need to know in what confidence level a group of sources could be regarded as a cluster or with what probability the cluster could be identified by chance . here , based on the second method
, we try to establish a statistical approach to identify large - scale structures , with which the confidence level of selecting a cluster is able to be informed . in order to develop such a method
, we need to address several questions : ( 1 ) how do we pick out a cluster ? ( 2 ) in what confidence level could a cluster with a certain number of galaxies and crowded degree be identified by chance ? to find answers to these questions , at least the following requirement should be satisfied : the probability for identifying a group of galaxies as a cluster by chance should be well determined . in doing so , it is essential that all selection effects should be removed or avoided .
it is plain that , if the scale used to find coherent clusters is larger than the area concerned , then all sources within that area must be included in a single coherent cluster ; if the scale is as small as possible , then all sources within that area must be separated from each other and no coherent clusters could be identified .
it seems that , in identifying a group of clustering sources , one needs to take several steps : determine a criterion clustering scale , @xmath33 ; sort out clustering sources with @xmath33 by applying the fof algorithm ; calculate the probability for picking out a cluster by chance . in calculating the probability
, the monte - carlo simulation will be applicable . in the case of the two - point correlation function
, the correlation length @xmath34 is regarded as a characteristic clustering scale . according to the well - known power law relation @xmath35 , the probability of finding another source , relative to any source concerned , at @xmath36 is twice the probability produced by random data sets .
if the sample is large enough , the twice probability suggests that the density measured at that scale is two times of the mean density .
as mentioned above , in the two recent surveys , @xmath37 was found to range from @xmath38 to @xmath39 .
a criterion associated with the sorting out scale comparable to these scales will be adopted in this paper .
let @xmath40 be the mean density of a flux - limited but homogenous data set without the effect of clustering , expected at redshift @xmath41 .
similar to the commonly used selection function ( see , e.g. , martinez & saar 2002b ; coil et al .
2004 ; padilla et al . 2004 ) , @xmath40 as a function of redshift is determined by the real distribution of the observed galaxies at @xmath41 , which depends on the evolutionary effect , and is affected by the flux - limited effect as well as other selection effects such as masks in given fields , fiber collisions in the spectrographs , etc ( martinez & saar 2002b ) . as pointed out by martinez & saar ( 2002b ) , many selection effects are directional , with some being due to the construction of the sample .
the absorption by the dust of the milky way also gives rise to a selection effect which is also directional .
the best way to take this effect into account is to consider the well defined maps of the distribution of galactic dust ( schlegel , finkbeiner and davis 1998 ) .
when directional selection effects are considered , we should deal with @xmath42 instead of @xmath40 , where @xmath43 and @xmath44 are coordinates of direction . for the sake of simplicity ,
we consider in this paper only the case of @xmath40 which corresponds to the commonly used radial selection function .
the difference between @xmath40 and the radial selection function is that , when measuring the former , we divide the count of sources within a redshift interval by the real volume confined by that interval . the criterion clustering scale adopted in this paper
is taken as @xmath45^{-1/3}$ ] .
the reason for adopting this criterion clustering scale is that the scale so adopted is generally smaller than , but comparable to , the mentioned typical scales @xmath37 which range from @xmath46 to @xmath39 in the two recent surveys .
the mean of @xmath33 in the sample adopted below ( sample 2 ) is @xmath47 .
the number of galaxies in the sample with their @xmath33 being less than @xmath48 is 172265 , 76% of the total .
that means that sources of clusters will generally be sorted out with the scales less than @xmath37 ( note also that , in deng et al .
2006 , @xmath49 was adopted to sort out clusters , and in hoyle et al .
2002 , @xmath50 was used to plot the median density contour ) .
we find that when @xmath51 is adopted to sort out clusters one will get a much larger number of galaxies for the largest coherent cluster sorted out with the method proposed below .
adopting @xmath52^{-1/3}$ ] will provides us a conservative result .
( note that when adopting @xmath53^{-1/3}$ ] one will get a more conservative result since @xmath33 becomes smaller , but this scale will be much less comparable to @xmath54 . ) it is known that in the case of the two - point correlation function , @xmath37 depends on the clustering property of samples . if adopting @xmath37 to identify clusters of a sample , we are going to measure the structure of the sample with a ruler concluded from the structure itself .
if the structure changes then the ruler changes . unlike @xmath37 , the criterion clustering scale @xmath55 proposed here depends only on the property of the background sample from which the concerned probability will be derived ( see what presented below ) . this quantity is entirely independent of the properties of the adopted sample and then is a rather objective ruler . due to the flux - limited effect
, some distant galaxies will be missed .
this makes the average distance between distant sources that are observed become larger .
if we take a fixed @xmath33 to identify clusters by applying the fof algorithm , we might probably include all nearby sources in a single large - scale structure when @xmath33 is large enough , while it might be possible that for those far away galaxies only very few of them are included to form a cluster since @xmath33 is not as large as to connect most of them .
this is unfair .
for some nearby sources , they might in fact be relatively far away from the local structure but they are included as members of the structure since their distances to the structure are shorter than the adopted @xmath33 ; for distant sources , since the average distances are large they have less chance to be identified as members of clusters according to this fixed @xmath33 .
when we adopt @xmath45^{-1/3}$ ] we deal with a varying @xmath55 since @xmath40 is a function of redshift . in this case
, we will get a large value of @xmath55 for large redshifts and obtain a smaller one for small redshifts . as part of out method
, a varying @xmath56 will be adopted to identify clusters ( see what proposed below ) .
the result of simulation would be acceptable when the background sample so created has no bias or has only very insignificant bias .
it is essential that all selection effects and the evolutionary effect have been removed or avoided .
there is a straightforward technique to avoid these effects .
assume that , within a spherical space centered at the observer , there are @xmath57 galaxies in total , which form various structures due to dynamics and other known or unknown factors .
when no clustering mechanisms are at work , these sources are expected to be homogeneously distributed within the area , and they constitute a global homogenous background sample .
in fact , what one can observe at present are galaxies of different cosmological ages . due to the evolutionary effect
, number densities might evolve with redshift , and then the total number , @xmath58 , of galaxies whose photons could reach us at present would be different from @xmath57 ( e.g. , the formation of galaxies and the merging of galaxies might occur somewhere at some time ) .
in this situation , the distribution of the observable background sample would only be homogeneous locally .
let us consider a survey over the whole sky .
due to the flux - limited effect and other selection effects , only @xmath59 ( @xmath60 ) galaxies could be observed in both real and background samples , which we call the whole observed real sample and the whole observed background sample respectively .
( note that these two samples have the same number of galaxies , but the galaxies are assumed to be distributed randomly in the latter sample . )
it is obvious that the mean number densities of the whole observed real sample and the whole observed background sample would be the same in different redshifts , from which we can deduced a mean number density law with respective to redshift .
now we perform a survey over a limited area of the sky .
according to the isotropic nature of the universe , it is expected that the mean number densities of this smaller observed real sample and the corresponding observed background sample are the same as well in different redshifts . keeping this in mind , a background sample created by the simulation following this mean
number density law would have no bias .
statistical significance derived from this sample would be safely acceptable .
note that selection effects would differ from survey to survey due to different techniques and instruments .
the proposed simulation method is simple due to the fact that it does not refer to the details of selection effects ( in fact , all selection effects have been taken into account ) .
besides the selection effects , there exists the boundary effect for most samples .
it is desired that problems arising from this effect could be overcome . the most effective way of solving this problem is to consider the real volume involved . in the case of estimating @xmath19 , the count of sources of the adopted sample lying inside the shell @xmath61 $ ] relative to a point
is divided by the volume of the intersection of that shell with the whole sample volume ( rivolo 1986 ; martinez et al .
1998 , 2001 ) .
this technique will be adopted in our analysis .
based on these arguments , we propose : a ) for a survey concerned , search for the mean number density law @xmath40 with the whole data set ; b ) draw from the survey a subsample for which the borders are regularly cut so that within these borders a sub background sample could be easier created according to @xmath40 ; c ) for a given redshift , determine the sorting out coherent clustering scale by @xmath62^{-1/3}$ ] ; d ) apply the fof algorithm with @xmath63 to identify clustering sources ; e ) calculate the probability of creating a cluster with a given number of galaxies and density by chance . in doing so , the real volume confined by @xmath64 and the whole area of the subsample will be considered when calculating the relevant density ( see what presented below ) so that the boundary effect would be avoided ( see rivolo 1986 ; martinez et al .
1998 , 2001 ) .
what effects does varying @xmath63 have on the analysis when the above method is applied ? as discussed above , the missing of objects due to the flux - limited effect will make @xmath40 becoming smaller and thus make @xmath63 becoming larger .
two possibilities will occur in this situation : some intrinsic clustering sources are missed and some intrinsic unclustered objects are included in a cluster identified with the method .
this will make the sorting out clusters less convincing at large distances where galaxies are rare .
the probability for mis - identifying sources is currently unavailable since the real distribution of galaxies is unaware .
the concept of clusters indicates the phenomenon of a number of sources gathering within a relatively small volume .
it is directly associated with high density regions .
the two - point correlation function is not a direct measurement of number density of the neighborhood of individual sources and can not describe the crowded nature ( the degree of density ) of their neighborhoods .
the same number found within the same volume of shells relative to two sources does not guarantee that the same amount of neighborhood galaxies would be detected within the area enclosed by the shells ( this could be observed in the galaxy distribution plot of many surveys ) . to measure the crowded nature of individual sources as well as the density of a cluster ,
we need a proper quantity which could be applicable to a discrete source sample .
there is a quantity suitable to describe the crowded nature of the neighborhood of discrete sources .
it is the conditional density in spheres @xmath65 which is an ensemble of realizations of a given point process ( see coleman et al . 1988 ; joyce et al . 1999 , 2005 ; vasilyev et al .
motivated by this quantity and the @xmath30 function mentioned above in the introduction section , we introduce in the following a new statistic to describe the crowded nature of the neighborhood of individual sources , where the mean density as a function of redshift and the boundary effect are taken into account .
consider a real sample with number of galaxies @xmath58 and a background sample of the same number of galaxies , confined within the same volume .
the latter sample is created by simulation under the assumption that galaxies are randomly distributed according to @xmath66 over the space concerned .
assign @xmath67 as a density function defined by @xmath68 in the co - moving coordinates of the universe , where @xmath69 is the number density of the sample concerned at position @xmath70 , and @xmath40 is the mean density of the background sample expected at the redshift corresponding to @xmath70 . concerning the crowded nature of the neighborhood ( the neighborhood density ) of a source , let us introduce an individual neighborhood function @xmath32 for object @xmath71 to describe the number density of the sample relative to the object within scale @xmath20 , which is defined by @xmath72 where @xmath73 is the redshift of object @xmath71 , and @xmath74 is the volume of the intersection of the sphere with radius @xmath20 , centered at object @xmath71 , with the whole volume of the adopted sample ( here , @xmath20 does not represent @xmath75 ) .
the mean of @xmath76 of a sample ( called the neighborhood function of the sample ) , @xmath77 is one of its statistical properties ( a spatial distribution property ) , which reflects the mean relative number density of the sample measured within the mentioned scale ( note that , in the case of adopting @xmath78 , @xmath79 will differ from source to source ) . for a subset of the sample ( namely , a group of sources of the sample )
, we also use the mean of @xmath32 to describe its spatial distribution property ( its density nature ) , and this quantity is also denoted by @xmath80 .
the only difference is that , in equation ( 3 ) , @xmath58 will be replaced by @xmath59 , where @xmath59 is the number of galaxies of the subset . it is obvious that the larger the @xmath80 , the more crowded the sample or the group . in terms of @xmath67 ,
the two quantities could be expressed by @xmath81 and @xmath82 ( note that , when different weights for the particles are adopted and taken into account , one might deal with other definitions of @xmath32 and @xmath83 .
in that situation , one might wish to modify definition ( 1 ) so that the forms of other equations would maintain . ) but deals with a sample for which some galaxies are missed , one might wish to introduce a weight to account for the missing sources . in this way
, @xmath84 would be replaced by @xmath85 , where @xmath86 is the weight which is asumed to be known , and then equation ( 1 ) is modified .
absorbing @xmath86 into @xmath40 , @xmath87 , and replacing @xmath40 with @xmath88 , one will get the same form of equations for @xmath32 and @xmath83 . ] in the case of the two - point correlation function , @xmath89 suggests that , for any given source , the probability of finding another source in distance @xmath20 is larger than the probability produced by the random data set and the former is @xmath90 times of the latter . in the case of the individual neighborhood function ,
the probability of finding a source within the scale of @xmath20 relative to position @xmath91 is proportional to the integral of the number density @xmath84 over the volume confined by @xmath92 for the adopted sample , and it is @xmath93 times of that of the background sample which is created randomly according to @xmath94 , where @xmath95 suggests the larger probability of the sample than that of the background sample . in the case of the neighborhood function ,
the probability of finding a source within the scale of @xmath20 is @xmath96 times of that of the background sample . in practice ,
@xmath32 and @xmath80 can be computed by @xmath97 and @xmath98 - 1,\ ] ] where @xmath99 is the @xmath100-function giving the position of particle @xmath101 ; @xmath102 is the portion of the whole sample volume satisfying @xmath103 for any position @xmath104 ; the integral takes over @xmath102 ; the sum over @xmath101 includes only particles in the position within @xmath105 in the sample , which yields the number of sources found within @xmath106 . for a volume limited sample ,
@xmath102 would differ from source to source due to different positions of the objects relative to the borders of the sample volume , even @xmath20 is the same .
one can check that , when ignoring the variation of @xmath40 with respect to redshift , then quantity @xmath96 would be similar to the ripley @xmath30 function which is an integral of @xmath107 over the spherical ball with radius @xmath20 ( ripley 1981 ; martinez et al .
there is not a direct relation between @xmath80 estimator and commonly used @xmath108 estimators since we calculate @xmath109 in a slightly different manner [ e.g. , no random samples are required to compute @xmath80 ] .
however , when the difference between @xmath96 and the ripley @xmath30 function is ignored , then the differential of @xmath96 would come to @xmath107 .
the main difference between the @xmath30 function , or the two - point correlation function , and the neighborhood function is that , with @xmath32 we can tell how crowded is the neighborhood of a source and can even show how different crowded sources are distributed in the space and how this distribution is related to the structures observed ( see what presented below ) .
quantity @xmath80 could also be closely related to the conditional density in spheres @xmath65 ( vasilyev et al .
in applying equation ( 7 ) , when one takes @xmath111 and makes the sum over the whole sample , @xmath96 would become @xmath65 , as long as the volume of the sample concerned is large enough so that for any point in the sample the sphere of radius @xmath20 centered at the point is well inside the volume . in the following analysis , cosmological distances as well as co - moving volumes will be expressed in terms of the co - moving coordinates and in calculating the latter the standard cosmological model , @xmath112 , will be adopted through out this paper , leaving the hubble constant parameter @xmath113 ( @xmath114 ) serving as a unit .
the analysis will be performed in redshift space .
that is , when calculating the co - moving coordinates of galaxies of samples drawn from redshift surveys , the peculiar velocity of individual sources will be ignored and the measured redshift will be taken as the real distant indicator of the object .
the data studied are taken from `` the 2df galaxy redshift survey ( 2dfgrs ) : final data release '' ( colless et al . 2003 ) .
two main regions of the survey data located in distinct areas of the sky are visible . to calculate @xmath76 as well as @xmath80
, we must know the volume @xmath115 for each source and for any given @xmath20 .
this would be realizable if the area concerned is well defined where its borders are sharp
. the computation would be easier performed if the edges of a region are regularly cut .
the following criterions are adopted to select a sample from the survey data set : @xmath116 , @xmath117 , @xmath118 , @xmath119 , where @xmath120 is the redshift quality parameter for best spectrum ( @xmath121 ; reliable redshifts have @xmath119 ) .
we thus get a sample of 59497 sources ( called sample 1 ) .
table 1 presents the description of this sample as well other samples mentioned below , where column 1 is the name of samples , columns 2 , 3 and 4 denote the number , area and region of the samples respectively , column 5 shows the @xmath40 that is used to calculate @xmath122 for the samples and to create the corresponding background samples by simulation ( see what mentioned below ) , and column 6 denotes the type of data ( observation or simulation , with the former being that of a real sample and the latter being that of a background sample ) .
the sky region of sample 1 is illustrated in figure 1 , where one of the two distinct areas , which encloses this sample , is also presented .
we notice that there are shortcomings of the 2df sample . as shown in cannon et al .
( 2006 ) , there are fibre collisions within 2df , where many objects around the edge of a 2 degree field are missed ( see figure 3 in cannon et al .
. this would give rise to a directional incompleteness of a survey .
however , the provided data are from such a set of 2 degree fields where most regions of the sky inside the survey boundary are covered by several overlapping fields and the potential incompleteness has been significantly reduced ( see figure 5 in colless et al .
in addition , we cut away the edge of the concerned region and this reduces the directional effects of fibre collisions as well .
we thus presume in this paper that the effect of this is minor ( i.e. ignorable ) .
another shortcoming is the existence of `` missed galaxies '' ( see pimbblet et al .
2001 ; cross et al .
as shown in cross et al .
( 2004 ) , the incompleteness of the sample could reach about 14 per cent , which varies slightly with magnitude ( see figure 11 of the paper ) .
this shortcoming is a selection effect . as a function of redshift
, it is absorbed into @xmath40 ( see what adopted below ) .
the main influence of this effect on our results is that when we identify a large - scale structure with the current 2df data we might obtain a larger one when all missed galaxies are discovered .
the third shortcoming is due to the fact that the @xmath123 magnitude limit is not constant across the entire area of the survey ( see colless et al .
2001 ; norberg et al .
this might affect the results of the analysis .
we will discuss this effect in the last section of this paper . and
the parent area where sample 1 is taken from is also presented ( both dark and light grey colors ) . ]
since the volume of the area concerned is well defined , we are now able to estimate the mean density of the background sample , @xmath66 , from the observational data . as pointed out by padilla et al .
( 2004 ) , there are several approaches to do so .
the most straightforward way is to use the luminosity function of galaxies to obtain an estimate of @xmath40 .
this is not adopted in this paper since any bias from the real luminosity function would cause systematic errors ( see padilla et al .
2004 and the references therein ) . in the following ,
let us consider three other approaches .
the first way is to estimate @xmath40 directly from the observational data . in this approach
we assume that the redshift distribution of the whole 2df survey data set could serve as a parent population of the background sample .
of course , to match the data of sample 1 , the data set serving as the parent population , which covers the whole area of the 2df survey , should meet these criterions : @xmath118 , @xmath119 .
a set of 226302 sources ( called sample 2 ) is obtained , which is almost four times of the number of galaxies of sample 1 . as illustrated below , there are two massive superclusters in 2df and these superclusters are comparable to the survey volume ( see also erdogdu et al . 2004 ; lahva and suto 2004 ; porter & raychaudhury 2005 ; einasto et al .
this suggests that 2df might not be a fair sample of the universe .
the reason for assuming the whole 2df survey data set being a fair sample of the universe is that it is indeed very large ( based on it , many statistical analyses have been made ) and it is the largest available sample of this kind ( detected by the same means of the 2df survey ) . under this assumption , we simply measure @xmath40 from sample 2 and then apply it to our simulation analysis .
the advantage of this approach is that any possible evolutionary and selection effects , known or unknown , will be accounted for .
there are two disadvantages of the method .
one is due to the measurement uncertainty which will cause an unreal distribution within the uncertainty scale .
this will be eased by assigning a random value smaller than the uncertainty to the background data when randomly drawing them from sample 2 ( the following application shows that this technique is indeed at work ; see figure 3 ) . the other is due to the fact that the 2df survey does
not covers the whole sky .
if there exist strong clusters which are comparable to the regions concerned ( yes , as pointed out above , there are some ) , the redshift distribution of the presumed parent population will be affected : bumps will be seen in the very dense place and troughs will exist in the very sparse space .
in this situation , some redshift distribution feature due to clustering ( padilla et al . 2004 ) will be mistaken as the background property and then clusters corresponding to this feature might be missed ( at least in the redshift space this clustering property will be ignored ) .
thus , the number of galaxies of some very large clusters would become smaller . to get a safer number of galaxies
, one might prefer this approach . the second way of estimating @xmath40 is to fit parametric forms to the observational data ( see padilla et al . 2004 ) , which uses sample 2 as well . in this approach
, we do not directly apply @xmath66 that is measured from sample 2 to create background samples .
instead , we fit it with an empirical curve .
as long as the probability obtained from the fit is small enough , the fitting curve will serve as @xmath40 and then will be applied to our simulation analysis .
the mean density curve so obtained is nothing but a deep smoothing of that directly measured from sample 2 .
we have a risk in employing this approach . in the process of smoothing , while bumps and troughs caused by clustering will be smoothed ,
features arising from selection effects will be smoothed as well . in the following analysis ,
results arising from the first and the second approaches will be presented and compared . for the sake of comparison , the third way of estimating @xmath40 is also adopted . in this approach , the galaxies observed are assumed to be homogenously distributed over the whole area of sample 1 .
thus we simply take @xmath125 to create background samples .
although the redshift distribution of this kind of background sample is not real , but the analysis might reveal some aspects of clustering which are unfamiliar before ( see what presented below ) .
let us derive @xmath40 from sample 2 .
the count observed within any sky region of the 2df survey is assumed to be proportional to the corresponding sky area according to the isotropic property of the universe .
let @xmath73 be the redshift of source @xmath71 in sample 2 .
we calculate the count within the redshift interval @xmath126 @xmath127 from sample 2 as well as calculate the volume @xmath128 confined by this redhift interval in the area of sample 1 .
according to the isotropic assumption , the volume confined by this redshift interval in the whole area of sample 2 should be @xmath129 times of @xmath130 .
thus the count divided by @xmath131 will be taken as the estimated value of @xmath40 for this source , which is denoted by @xmath132 . in doing so ,
when the redshift interval is less than the measurement uncertainty of the survey , the latter will be taken as the width of the redshift interval .
the estimated values of @xmath132 of sample 2 are shown in figure 2 .
the data could be well fitted by the following polynomial function : @xmath133 ( @xmath134 ) , where @xmath66 is in units of @xmath135 .
we create a background sample ( sample 3 ) randomly from sample 2 with the first approach . the second background sample ( sample 4 )
is produced with the second approach by simulation according to the empirical mean number density function , the polynomial function . the third background sample ( sample 5 )
is yielded by simulation under the assumption that the galaxies are homogenously distributed , where the mean number density is obtained by dividing 59497 with the whole volume of sample 1 , which gives @xmath136 .
the numbers of galaxies of all these background samples are the same as that of sample 1 , @xmath137 .
redshift distributions of samples 1 - 5 are shown in figure 3 .
panel ( a ) of the figure shows that there indeed exist some bumps and troughs in both samples 1 and 2 .
compared with other panels , we suspect that this phenomenon is unlikely to be caused by random . instead , it might probably be due to intrinsic distributions such as large - scale structures ( see what discussed above in this section and see fig .
22 presented below ) or selection effects . in addition , we find some significant features in sample 1 , which are much less prominent in sample 2 .
this must be due to the local property of redshift distribution and might probably be a consequence of the existence of local clustering ( see figs .
17 , 20 and 21 presented below ) .
panel ( b ) plainly demonstrates that samples created with the first approach well follow the redhsift distribution of sample 2 .
in addition , it reveals that the chaos of the observational samples is largely caused by the measurement uncertainty ( since adding an additional random value smaller than the uncertainty significantly reduces the fluctuation observed , as panel b shows ) . from panel ( c )
we observe that , when ignoring the fluctuation , which might be caused by local clustering , the polynomial function could indeed well represent the real distribution of the observational samples .
panel ( d ) illustrates that , as expected , the sources observed are not homogenously distributed at all ( the flux - limited effect and the evolutionary effect would be the main factors accounting for the non - homogenous distribution observed in the survey data ) .
although spatial distributions of sources of background samples 3 and 4 differ significantly from that of the real homogenous one , sample 5 , a distribution in agreement with the former is still regarded as a homogenous one in the following analysis .
this is because that background samples 3 and 4 suffer from the flux - limited effect and the evolutionary effect as well as other known or unknown selection effects , and it is these effects that lead to the apparent in - homogenous distribution .
when all galaxies of the same cosmological age are available , the spatial distribution of background sample sources , which are randomly created , must be homogenous due to the principle of cosmology .
therefore , in the analysis below , those in agreement with the spatial distribution of sources of the adopted background sample , or in agreement with that of the parent population of background samples , will be regarded as an intrinsic homogenous one , and those obviously deviating from that distribution will be considered as an in - homogenous one .
functions estimated from samples 1 ( open circles ) , 3 ( pluses ) , 4 ( crosses ) , and 5 ( filled circles ) . the solid line is the best linear fit to the @xmath30 function of sample 5 , which obeys @xmath138 . the correlation dimension could be easily figured out from the plot , where @xmath139 is obtained for sample 5 and @xmath140 could be derived from the other three samples within the maximum scale of the samples , @xmath141 . ]
shown in figure 2 , one could observe a deviation , of the polynomial provided for @xmath40 , from the mean number density function directly measured from sample 2 .
in addition , a deviation of the redshift distribution of the polynomial function from that of the @xmath40 of sample 2 is seen in figure 3 ( see panels b and c ) .
this suggests that the polynomial provided for @xmath40 and the @xmath40 measured from sample 2 are not the same .
however , as one will see below , this difference does not cause big problems in the clustering analysis performed in this paper . since sample 2 suffers from the clustering effect , as mentioned above and revealed below , it might be possible that samples 3 and 4 , which come from sample 2 , would be inhomogeneous not only due to the flux limit but also due to clustering .
thus , if the deviation observed in figure 2 leads to an observable difference in the homogeneity properties of the two corresponding background samples needs to be checked .
let us check this by examining their @xmath30 functions , from which the corresponding correlation dimension @xmath142 could be derived ( see martinez et al .
the @xmath30 function adopted to investigate this issue
is that provided in equation ( 8) of martinez et al .
( 1998 ) , the standard estimator introduced by doguwa and upton ( 1989 ) to account for the boundary effect . as is generally known , regardless of the possible slight bias
, this estimator , @xmath143 , has good properties . displayed in figure 4 one could find the @xmath30 functions estimated from samples 3 and 4 , where , for a comparison , that from samples 1 and 5 are also presented . as expected , the @xmath30 function estimated from sample 5 well follows the homogeneity curve , the curve for which @xmath139 . the three other samples are seen not to be homogeneous at all ( for all of them , @xmath140 holds within the maximum scale of the samples ) .
it is interesting that the @xmath30 functions estimated from samples 1 , 3 and 4 follow almost the same trend ( in this case , they will have almost the same @xmath142 ) .
while that from sample 1 is slightly larger than that of samples 3 and 4 , the difference between the latter two can not be detected .
this suggests that , the @xmath30 function , and hence the derived @xmath142 , suffers mainly from the redshift distribution which is influenced by the flux limit , the possible evolutionary effect , and the spatial distribution of large - scale structure members ( see fig .
the role that the locally clustering ( sources in sample 1 are strongly clustered while those in samples 3 and 4 are not ; see the spatial clustering plots presented below ) plays in producing the correlation dimension @xmath142 seems very insignificant .
here , we use estimator ( 6 ) to calculate the individual neighborhood function of sources in the scale of @xmath33 ( i.e. , we take @xmath78 ) which is determined by @xmath144^{-1/3}$ ] .
we apply the @xmath132 directly measured from sample 2 to calculate @xmath145 for each source of samples 1 and 3 .
in addition , we determine @xmath40 by the polynomial function obtained above and then calculate @xmath145 for each source of samples 1 and 4 with this @xmath40 .
also , we adopt @xmath136 , the mean number density of sample 1 , to calculate @xmath146 for each source of samples 1 and 5 . in this way
, we calculate @xmath146 for sample 1 with three kinds of @xmath40 ( and hence with three kinds of @xmath33 ) estimated with three approaches .
directly estimated from sample 2 ( the first approach ) ; those in panels ( c ) and ( d ) are calculated with the values of @xmath147 determined by the empirical polynomial function ( the second approach ) ; and that in panels ( e ) and ( f ) are calculated with the constant mean density @xmath148 ( the third approach ) . ]
distributions of the individual neighborhood function @xmath149 of sample 1 , calculated with three kinds of @xmath150 are presented in panels ( a ) , ( c ) and ( e ) of figure 5 respectively .
meanwhile , distributions of @xmath145 of samples 3 , 4 and 5 are displayed in panels ( b ) , ( d ) and ( f ) of the figure , respectively .
one might observe from panels ( b ) , ( d ) and ( f ) that the distributions of @xmath145 are almost the same for background samples created with different approaches , although the adopted @xmath40 , and hence @xmath33 , differs in different cases .
this is not surprising since @xmath145 reflects only a relative density and background samples are randomly created by simulation , which would be locally homogeneous .
in fact , the number density of a background sample is created in accordance with a provided mean density @xmath40 , and quantity @xmath151 is that number density divided by the mean density . in the case of a real sample ,
the situation is much different . in panel ( e ) , when calculating @xmath145 , the number density of sample 1 is divided by a constant mean density , @xmath152 , but the real number density of sample 1 differs significantly for sources with different values of redshift .
thus , one observes a much different @xmath145 distribution of sample 1 in panel ( e ) from those in panels ( a ) and ( c ) .
the values of @xmath146 could be found as large as several ten thousands ( the figure covering the whole range of value of the quantity is omitted ) .
in fact , @xmath153 in panel ( e ) reflects absolute number densities rather than locally relative ones . due to several selection factors mentioned above , there is an enormous difference of observed density .
in contrast , distributions of @xmath145 presented in panels ( a ) and ( c ) are almost the same .
this suggests that the difference of @xmath40 shown in figure 2 does not significantly influence the distribution of @xmath146 for the adopted real sample , sample 1 .
( see also the discussion below in the case of spatial distributions . )
one might observe that , as histogram plots , distributions in figure 3 look like `` continuums '' since the total number involved is very large .
however , with the same number , those in figure 5 do not look like `` continuums '' but look like `` histograms '' .
we have checked that this difference is not due to the adopted bins .
the distribution plots shown in figure 5 look like `` histograms '' because in some separated ranges of @xmath145 the counts are relatively small , while in the nearby @xmath145 ranges the corresponding counts are large ( thus , the former counts look like to be zero ) .
this phenomenon is due to the small scale we adopt ( i.e. , @xmath33 ) .
since the scale is relatively small ( see the definition of @xmath33 presented above ) , one could only find small numbers of sources within the corresponding volume .
thus , the difference of number would lead to obvious difference of the individual neighborhood function .
for example , the volumes confined by @xmath33 would differ slightly for two closely located sources .
if the number of neighborhood galaxies relative to one source is 3 and that relative to the other is 4 , then this difference would cause a 30 percent of change ( in this situation it would be hard to get 3 percent of change for this two sources ) . at least within a small region enclosing these two sources , the change of a smaller percent
would not be observed from the difference of numbers other than 3 .
for the background samples , the individual neighborhood function is mainly distributed within the range of @xmath154 ( see panels b , d and f ) , while for sample 1 , a large amount of sources have their @xmath145 larger than @xmath155 ( see panels a , c and e ) . motivated by figure 1 of governato et al .
( 1998 ) , let us symbolize a region within which the initial matter forms a galaxy in later times with a presumed galaxy ( the matter is referred to as the presumed galaxy matter ) and assign the central position of the region as the position of that presumed galaxy .
these presumed galaxies are expected to be randomly scattered in space . in terms of statistics , they are considered as un - cluttering sources . in our analysis ,
@xmath145 distributions of unclustered galaxies are those displayed in panels ( b ) , ( d ) and ( f ) of figure 5 .
the @xmath149 distribution of sample 1 is obviously different from them ( see panels a , c and e ) . revealed in the figure , a large amount of sources in sample 1 possess very large values of @xmath149 ( say , @xmath156 ) which the majority of unclustered galaxies do not have , suggesting that there is clustering in this sample . .
the values of @xmath40 directly estimated from sample 2 are adopted to determine @xmath33 and to calculate @xmath157 .
the cyan color stands for un - crowded sources ( sources with relatively low number densities in their neighborhoods ; @xmath158 ) ( a lot of cyan color objects are not observed in this figure due to the overlap of other color objects ) , the yellow color represents crowded sources ( sources with relatively large number densities in their neighborhoods ; @xmath159 ) , and the orange color symbolizes very crowded sources ( sources with relatively very large number densities in their neighborhoods ; @xmath160 ) .
the co - moving coordinates are determined by @xmath161 , @xmath162 , @xmath163 ( here @xmath41 is a coordinate , as generally adopted ; when it represents a redshift in the text below , we will present a note ) , where @xmath164 is the distance of the object to the observer , in units of @xmath165 . ] .
the values of @xmath40 figured out from the empirical polynomial function are adopted to determine @xmath33 and to calculate @xmath157 .
the symbols are the same as they are in figure 6 . for the definition of the coordinates ,
see figure 6 . ] .
the mean density @xmath152 evaluated from sample 1 , which is a constant with respect to redshft , is adopted to determine @xmath33 and to calculate @xmath157 .
the symbols are the same as they are in figure 6 . for the definition of the coordinates ,
see figure 6 . ] .
this plot is a copy of panel ( a ) in figure 6 , where different crowded sources are presented in different panels . ] .
panel ( a ) in this figure is a copy of panel ( b ) in figure 6 , while panel ( b ) in this figure is a copy of panel ( b ) in figure 7 , where the un - crowded sources are omitted . ]
according to figure 5 ( compare the right- and left - hand - side panels ) , we define those galaxies with @xmath158 , @xmath159 and @xmath160 as un - crowded , crowded and very crowded sources , respectively . in other words , sources with @xmath158 , @xmath166 and @xmath160
are identified as those locally located in regions with relatively small , large and very large number densities of galaxies , respectively .
they have different relative number density environments .
the spatial distributions of different crowded sources for samples 1 , 3 , 4 , and 5 are displayed in figures 6 - 8 .
as expected , one can find from panel ( a ) of figures 6 and 7 that very crowded galaxies seem to form the core of large - scale structures , where surrounding the core are those less crowded sources ( for identifying large - scale structures , see the analysis below ) .
although in the case of adopting the mean number density , the number of very crowded sources identified from sample 1 ( panel a in figure 8) is very large ( 45753 sources , 77% of the total number of the sample ) due to the smaller value of the density , @xmath152 , the region these sources occupy is smaller than that in other cases .
( in panel a of figures 6 and 7 , the numbers of very crowded galaxies are 27312 and 27476 , respectively . )
also as expected , all the un - crowded , crowded and very crowded sources in the background samples are seen to be homogenously ( relative to the expected observed density @xmath40 ) distributed over the whole sample volume ( see panel b in the three figures ) ( a similar issue will be investigated below ; see figure 16 ) .
we find that un - crowded sources in sample 1 show a homogenous distribution over the whole sample volume , which could be clearly seen in figure 9 ( panel a ) ( see also figures 15 and 16 presented below ) ( note that many distant sources are missed due to the flux limited effect ) .
it suggests that some isolated resided matter regions remain to be isolated from the very beginning to late times ( during this period , they form galaxies themselves ) .
crowded and very crowded galaxies in sample 1 are much less homogenously distributed ( see panels b and c in figure 9 ) .
this might be due to the continuous action of gravity that pulls these galaxies together , as expected in many models .
governato et al .
( 1998 ) have already pointed out that large concentrations are common in a universe dominated by cold dark matter and they are the progenitors of the rich galaxy clusters seen today .
it is interesting that filamentary features are observed in the state of random distribution , formed particularly by crowded sources ( see panel b in figures 6 and 7 ) .
this is plainly shown in figure 10 .
in addition , seeds of knots seem to be presented early in the random state as well ( see figure 10 ) . according to the view of governato et al .
( 1998 ) , we suspect that it might be these features that grow to the obvious clustering structure observed in later times , where many un - crowded presumed galaxies have joined .
it seems that , in this process , very dense regions move towards each other , and some of them form large - scale structures themselves in later times .
this is just the case expected in the concentration process of dark matter ( see figure 1 of governato et al .
revealed by recent observations , filamentary large - scale structures do exist at redshift as large as @xmath167 ( ouchi et al . 2005 ) .
observed in the two dimensional map of very crowded sources in sample 1 are some voids ( see figure 9 panel c ) .
these voids are also detected in the spatial distribution of crowded sources in the sample , but they are less obvious ( see figure 9 panel b ) . on the contrary , we find no obvious voids in the spatial distribution of un - crowded galaxies in the same sample .
meanwhile , some embryonic voids are seen in the two dimensional plot of crowded sources in sample 3 ( see figure 10 panel a ) .
it suggests that voids are likely the volumes within which no or very few very crowded sources are present and they are likely formed in embryo by fluctuation in the very early epoch of the universe
. it might be the continuous gravitation in later times that pulls more crowded galaxies closer and at the same time leaves behind adult voids .
shown in figure 11 are the finer resolution maps showing the details of the spatial distributions of different crowded sources in sample 1 in two local regions of the volume concerned ( see figure 6 panel a ) . indeed , from the figure we find that , if voids are identified according to the spatial distribution of very crowded sources , then they are seen to be filled mainly with un - crowded sources ( for a more detail , see figure 21 presented below ) .
the spatial density distributions of sample 1 calculated in the first and second approaches are almost the same ( see panel a in figures 6 and 7 ) .
meanwhile , one might observe that the distributions of @xmath145 for sample 1 calculated with the two approaches are not distinguishable ( see panels a and c in figure 5 ) .
this suggests that the adopted values of @xmath40 estimated directly from sample 2 and evaluated by the empirical polynomial function do not provide a significant difference in the analysis of density distribution . as discussed above , in examining homogeneity properties of background samples , where the values of the correlation dimension @xmath142 are checked ,
the same conclusion is obtained . in the following ,
we discuss only the case of adopting the @xmath40 directly estimated from sample 2 ( the first approach ) , where radial selection effects ( known and unknown ) are accounted for .
as long as the mean density @xmath40 is available , we are now able to apply the fof algorithm to identify clustering sources with the scale of @xmath168 .
to calculate the probability of creating a cluster with a given number of galaxies ( in terms of statistics , the total number of members of a sample is always referred to as the size of the sample ) and with a certain number density by chance , we perform a number of simulations and for each time of simulation we sort out coherent clusters and calculate their neighborhood functions .
let the number of simulations when clusters with numbers of galaxies larger than @xmath169 and neighborhood functions larger than @xmath170 are observed be @xmath171 , and the total number of trial be @xmath58
. then the probability of creating a cluster with its number of galaxies larger than @xmath169 and its neighborhood function larger than @xmath172 by chance is estimated by dividing @xmath171 with @xmath58 : @xmath173 .
the details of our simulation analysis are summed below ( some of them could be found in the previous sections ) .
a ) we randomly select 59497 redshifts from sample 2 . these form a simulation sample which has the same number of sample 1 . in doing so ,
226302 redshifts which correspond to 226302 sources contained in sample 2 have the same chance for being selected .
( in this way , the redshift distribution of the simulation sample would be the same as that of sample 2 .
see figure 3 panel b. ) b ) for each redshift that is selected , we assume that the digital number that is un - measurable is uniformly distributed within the range of that digital number and then we modify the redshift according to this assumption .
for instance , when selecting a redshift of 0.2981 from sample 2 , we assume that this observed redshift arises from a real redshift ranging within @xmath174 , and then we modify the original redshift 0.2981 by adding a random value @xmath175 to it , where @xmath175 is assumed to be uniformly distributed within the range of @xmath176 .
we then get a new redshift set for the simulation sample .
( as shown in figure 3 panel b , the distribution of redshifts of this kind of sample is much less scattered than that of sample 2 . )
c ) corresponding to each redshift of the simulation sample , we randomly select a sky position confined in the area of sample 1 and assign the redshift and the position to a source . in this way
, we get 59497 sources for the simulation sample . in doing so
, the universe is assumed to be isotropic and then the same area in the sky within the sample region ( the area of sample 1 ; see figure 1 ) has the same probability for being selected .
d ) we calculate @xmath33 for each source in the simulation sample by applying the relation @xmath177^{-1/3}$ ] , where the adopted @xmath132 is that directly measured from sample 2 .
e ) for this simulation sample , we apply the fof algorithm with the @xmath33 obtained above to sort out clusters . for each pair of sources we always have two values of @xmath33 .
the two sources are considered to be within a same cluster when the distance between them is smaller than the smaller value of the two @xmath33 .
f ) for each cluster , we calculate their @xmath178 , using estimator ( 7 ) , where the sum is taken over the whole number of sources of the cluster concerned .
g ) we perform a number of simulation and then , according to these simulations , estimate the probability for forming a cluster with a certain number of galaxies and certain value of @xmath178 by chance .
besides the reason proposed above , there is one more reason for choosing @xmath176 as the range of @xmath175 to be added to the provided redshift of sample 2 .
that is , the treatment itself makes the redshift distribution of the selected sample much less scattered than that of sample 2 ( see figure 3 panel b ) . in this way
, the chaos caused by the un - measurable digital number of the redshifts is eased . in doing so , we do not use the real redshift errors since they are not taken into account when we deal with sample 1 .
in fact , as shown by panel ( b ) of figs . 6 and 7 , spatial distributions of samples 3 and 4 are not distinguishable , although their redshift distributions are quite different in finer redshift intervals ( see figure 3 ) .
one can check that replacing @xmath176 with the real redshift errors in the simulation analysis would not provide an observable difference .
, @xmath179 , and @xmath180 , respectively . ]
displayed in figure 12 are the probability contours of coherent clusters formed by chance ( here , @xmath181 ) .
it shows that clusters with larger numbers of galaxies and larger number densities ( larger @xmath182 ) are harder to be formed .
the probability of forming coherent clusters with their numbers of galaxies larger than 350 is less than that of @xmath0 , @xmath183 .
our analysis shows that , if a cluster with its number of galaxies @xmath184 larger than 10 and its @xmath122 larger than 5.41 could be formed by chance , the probability must be less than that of @xmath180 , @xmath183 . under the same requirement of probability ,
other typical pairs of @xmath185 are @xmath186 , @xmath187 and @xmath188 .
it shows that , for a cluster with very large number of galaxies ( say , @xmath189 ) , the requirement of @xmath122 is very weak .
it could be a negative one .
this suggests that clusters with very large number of galaxies are very hard to be formed by chance .
even when their mean densities are smaller than the average ( i.e. , @xmath190 ) , they are still very hard to be formed .
thus , when a cluster containing the number of galaxies larger than 300 is identified in an area as large as that of sample 1 , it is unlikely that this cluster is formed by chance .
applying our method to samples 1 and 3 , we get entirely different sets of coherent clusters . in sample 1 ,
we detect 67 coherent clusters under the condition that if any of them is formed by chance then the probability will be less than that of @xmath0 .
listed in table 2 are the parameters of these coherent clusters . here , the scale of a coherent cluster is defined as the largest value of the distance measured for each pair of the sources of the cluster .
the smallest neighborhood function is @xmath191 .
for other 66 clusters , @xmath192 .
the largest number of galaxies is 12966 and the largest scale is @xmath1 , which belong to different coherent clusters .
for all these clusters , @xmath193 , suggesting that the number density , measured for each source within the scale of @xmath63 , for any of these clusters is larger than twice of the mean of the background sample . for sample 3 ,
the largest number of galaxies of the detected coherent clusters is 135 and the neighborhood function of this cluster is @xmath194 . the number of galaxies is larger than 134 , but its neighborhood function is slightly less than that coupling with 134 , @xmath195 ( see figure 12 ) .
thus , the probability of forming this cluster by chance is larger than that of @xmath0 .
the scattering distribution in the plane of @xmath196 of the coherent clusters detected from sample 3 is shown in figure 13 . demonstrated in the figure
, there are no coherent clusters detected within the @xmath0 probability region for sample 3 . displayed in figure 13
is also the scattering distribution of the coherent clusters detected from sample 1 .
one finds from this figure that almost all of coherent clusters detected from sample 3 are distributed beyond the @xmath197 probability region in the plane of @xmath198 .
while a sufficient number ( 67 ) of coherent clusters sorted out from sample 1 are situated within the @xmath0 probability region , there are some others detected from the sample existing within the region confined by the @xmath199 and @xmath0 probability contours . vs. number of galaxy members of coherent clusters identified in samples 1 ( crosses ) and 3 ( pluses ) , where the probability contours in figure 12 are also presented .
] we define coherent clusters with probabilities being less than that of @xmath200 as prominent clusters , and define those with probabilities being less than that of @xmath199 but larger than that of @xmath201 as less prominent clusters , and call those with probabilities being larger than that of @xmath199 as weak clusters .
spatial ( two dimensional ) distributions of the three kinds of cluster as well as the unclustered sources for the two samples are shown in figures 14 and 15 .
we find that the unclustered sources of both the observed and background samples are homogenously distributed over the whole space concerned ( see also figure 16 panels a and b ) .
they act like what the low relative density ( @xmath202 ) galaxies do ( see figure 9 panel a ) , and indeed , almost all of them are low relative density galaxies ( see figure 15 panels a and b ) . as suggested in panels ( a ) and ( b ) of figure 15 , the number of unclustered sources detected from sample 3 is larger than that of sample 1 .
this is expectable since many original unclustered regions ( the presumed galaxies ) might be pulled towards a nearby relatively dense region ( a presumed cluster ) in later times due to the continuous action of gravity .
figure 15 shows that the majority of mid and high relative density sources ( those of @xmath203 or @xmath204 ) are members of clusters ( weak , less prominent , or prominent clusters ) .
compared with the corresponding panels in figure 16 we find that sources of weak clusters in the background sample are homogenously distributed over the whole volume of the sample as well , and the spatial distribution of sources of this kind in the observational sample is a quasi - homogenous one ( see panels c and d in figures 15 and 16 ) .
in addition , we find that the number of weak clusters in the background sample is also larger than that in the observational sample .
this indicates that many unclustered galaxies and weak clusters become members of large - scale structures formed in later times .
as expected , shown in panels ( e ) , ( f ) , and ( g ) of figure 15 , we find that the core of prominent and less prominent clusters is filled mainly by sources with high number density neighborhoods ( @xmath205 ) , and surrounding them , galaxies with lower number density environments are observed .
the numbers of unclustered , weak clustering , less prominent clustering , and prominent clustering sources of sample 3 are 9033 , 50126 , 338 and 0 , respectively .
they occupy 15.2% , 84.2% , 0.568% and 0% of the total number of the sample .
the numbers of the corresponding sources in sample 1 are 3557 , 12700 , 3017 and 40223 respectively , and they are 5.98% , 21.3% , 5.07% and 67.6% of the total .
this reveals that , in terms of statistics ( i.e. , under the @xmath0 confidence level ) , the majority of galaxies observed in the 2df survey are members of prominent clusters ( in about 67.6% ) .
this indicates that the phenomenon of clustering is prominent in the present universe .
only a few galaxies ( about 5.98% ) are identified to be sources lonely staying in the space . comparing the corresponding numbers in the two samples we find that about 2/3 original unclustered sources and 3/4 original weak clustering sources would change their identification in late times .
it indicates that original weak clustering sources have more chance to become members of less prominent and prominent clusters in later times and indeed they provide about 87% of members of the latter ( note that only about 13% members of the latter come from the original unclustered sources ) .
this is not surprising since the seeds of clusters formed in later times are expected to be among those originally resided in regions with higher number densities ( see panel d of figure 15 ) .
displayed in figure 16 are redshift distributions of unclustered , weak clustering , less prominent clustering , and prominent clustering sources of samples 1 and 3 .
as expected , redshift distributions of different kinds of source in the background sample are well in agreement with the distribution of the sample itself since no dynamics are at work in the stage of random .
( note that the redshift distribution of less prominent clustering sources is not in agreement with that of the whole sample , which is due to the very small number of this kind of source , 0.568% , of the sample ) .
while the redshift distribution of the unclustered sources of the observational sample is consistent with that of the background sample ( see panel a of figure 16 ) , there is a slight deviation from that of the weak clustering sources of sample 1 to that of the background sample , and an obvious deviation from that of the prominent clustering sources of sample 1 to that of the latter sample is observed .
in addition we find that , for the observational sample , there is a change from weak clusters to prominent clusters : the number of sources in between ( sources of less prominent clusters ) is relatively small . according to this analysis
, we suspect that there might exist three kinds of galaxy or galaxy group in terms of clustering .
one is the class made up of isolated galaxies which would be homogenously distributed over the universe and would never be included in any clusters .
positions of these galaxies would stick to the framework of the universe .
another class includes trivial coherent clusters which would act like the isolated galaxies but their number densities would evolve with time . the other class contains prominent coherent clusters whose densities and numbers of galaxies would glow all the time from the very early epoch to present . after the clustering process of all galaxies in the universe ceases , the redshift distribution of the sources of these clusters would be that of the homogenous spatial distribution , when measured in the volume much larger than that associated with their typical scales .
measured within the volume comparable to that associated with their typical scales , it would not be surprising if the redhsift distribution shows a character of in - homogenous spatial distribution .
before the clustering process stoping , some potential members of these clusters look like trivial coherent clusters in densities ( described by @xmath122 ) and numbers of galaxies , and this would make the redshift distribution of apparent trivial coherent clusters ( weak clusters defined in this paper ; see panel c in figure 15 ) deviates from that of the homogenously spatial distribution ( see panel c in figure 16 ) . in this scheme ,
less prominent clusters defined in this paper should be the potential members of the third class , and they would change their identification in late times . according to this interpretation , once a prominent cluster is formed it will maintain its identification in late times .
this explains why there is a turn over change from the number of weak clustering sources to that of prominent clusters observed above .
revealed in figure 13 , there are two approaches for a weak cluster becoming a member of prominent clusters : increasing either its density ( described by @xmath122 ) or its number of galaxy members . in the first approach , they can accomplish the identification changing themselves when their members become closer , while in the second approach , the change would happen in the situation that they are coherently connected with other clusters . for an isolated cluster , when its relative density stops glowing , the process of identification changing will be ended .
this is why we expect a homogenous spatial distribution of members of the second class .
the reasons that we propose the above scheme to interpret the clustering process of galaxies are : a ) we find it natural to explain what we observe in the above clustering plots ; b ) it looks simple ; c ) it can provide unambiguous predictions in the clustering process which can be checked later ; d ) we find no simpler schemes accounting for our results . in the above analysis , we first classify four kinds of source according to their clustering properties available from observation : unclustered sources ( which do not appear in figure 13 ) , weak clustering sources ( which are beneath the @xmath199 contour in figure 13 ) , less prominent clustering sources ( which are within the area confined by the @xmath199 and @xmath0 contours in figure 13 ) , and prominent clustering sources ( which are above the @xmath201 contour in figure 13 ) .
later we propose a scheme with three kinds of galaxy or galaxy group ( isolated galaxies , trivial coherent clusters , and prominent coherent clusters ) to explain what we observed in the statistical analysis .
the four kinds of source might change their identifications during the clustering process .
but when the clustering process ceases , each of them belongs to one of the three kinds of galaxy or galaxy group .
here we show the structure of the two largest prominent coherent clusters identified in sample 1 . according to table 2 , the coherent cluster with the largest number of galaxies is cluster 1 which contains 12966 galaxies and extends to @xmath206 ; the one with the second largest number of galaxies is cluster 2 which contains less galaxy members ( 7788 ) but has the largest scale , @xmath1 .
the scale of cluster 1 is the distance between the two sources tgn141z157 and tgn353z218 , and that of cluster 2 is the distance between tgn163z153 and tgn276z102 . displayed in figure 17 are the two - dimensional positions ( relative to that of other galaxies of sample 1 ) of the two coherent clusters . shown in figures 18 and 19
are their plane and solid structures , viewed from various angles in the latter case . and the blue color for @xmath207 ) ,
the cyan and yellow colors represent crowded sources ( the cyan color for @xmath208 and the yellow color for @xmath209 ) , and the orange and wine colors symbolize very crowded sources ( the orange color for @xmath210 and the wine color for @xmath211 ) .
( note that , here we use six colors instead of only three to code different crowded sources in order to show their spatial distributions in a more detail .
the colors used are somewhat different from those adopted in figure 6 . ) for the definition of the coordinates , see figure 6 . ]
[ fig : 17 ] as shown in figure 17 , the two clusters are quite close .
the scales of the two coherent clusters are comparable with the whole volume of sample 1 ( the upper limit of redshift in sample 1 is @xmath212 which corresponds to @xmath213 ; see also figure 17 ) .
this might interpret why there is a bias of the redshift distribution of prominent coherent clusters of sample 1 to the redhsift distribution of the whole background sample ( see figure 16 panel g , especially at @xmath214 which corresponds to @xmath215 ) .
revealed by figure 18 , very crowded sources form the core as well as the frame of the corresponding structures .
the two clusters possess coral type structures in a fine map where the details of the structure are visible .
it seems that larger coherent clusters are made up of smaller ones and in forming the former the latter are likely connected by their antennae .
this is in agreement with what was discovered recently by einasto et al .
( 2006b ) : superclusters are asymmetrical and have multi - branching filamentary structure . in the same volume studied above , erdogdu et al .
( 2004 ) detected a large cluster , scngp06 .
we find that scngp06 is just inside the structure of cluster 1 , which is shown in figure 20 .
eke et al .
( 2004 ) also identified large galaxy number groups in the same volume . among their 9 largest groups identified within the volume ,
7 are inside the structure of cluster 1 ( see figure 20 ) . revealed in the figure , centers of scngp06 and the 7 large galaxy number groups of eke et al .
( 2004 ) are located at the knots of the structure . as suggested above ( see figure 9 panels b and c ) , voids are likely the volumes within which no crowded sources are present . in the structures of the two largest coherent clusters ,
voids are easily identified ( see figure 18 ) .
are they really the space where no crowded sources are found ?
figure 21 shows the structures of the two clusters together with that of other prominent clusters and the spatial distribution of different crowded galaxies of sample 1 within a smaller area enclosing the structures .
we find from panel ( a ) of the figure that , for the majority of voids detected from the structures of the two clusters , there indeed no or very few high density galaxies are present .
this can be clearly seen in panels ( d ) and ( e ) . within the apparent void of @xmath216 in panel ( a )
, there are some very crowded sources .
we suspect that this might not be a real void , but instead , it might come from the projected effect . in other words , the inner structures observed within the apparent void might not be close to the two clusters in the three - dimensional space , but they look like to be close to them in the projected two - dimensional space . indeed , as revealed in panels ( b ) and ( c ) , inside the apparent void , some structures of prominent clusters are present . shown in panel ( d ) , there are indeed few very crowded sources outside the structures of prominent clusters , and it is these galaxies that form the core and build the frame , of the structures .
crowded sources tend to be gathered around the structures , as seen in panel ( e ) .
just as pointed out above , un - crowded sources are seen to be homogenously distributed over the structure area , and they are indeed detected within the voids as well as within the frame of the structures ( see panel f ) .
in a recent investigation of voids , patiri et al .
( 2006 ) showed that faint galaxies populating the voids are clustered in small groups and filaments .
this is observed in panel ( f ) .
in addition , we find that the apparent filaments of un - crowded galaxies do not stretch along the frame of the structure of the prominent clusters , as what the very crowded and crowded sources do in panels ( d ) and ( e ) . recalled that filamentary features could appear in the state of random distribution ( see the discussion in section 4 ) , we thus suspect that the apparent filaments of low density galaxies might mainly be caused by fluctuation .
one might observe that , in the method adopted above , while the calculation of the relevant probability and the investigation of number density distribution depend on the exact volume of the adopted sample , the sorting out method ( i.e. , the fof algorithm ) is independent of the volume .
this enables us to identify the largest scale structure of a sample of the same survey , for which , the borders are not well defined and the volume is not available . although the probability of forming such structure by chance is unable to be evaluated , at least the following questions might be able to answer .
how large is the scale of the largest structure identified from the whole 2df survey by the fof algorithm ? could some unconnected structures identified in a smaller volume be coherently connected in a larger volume , which is strongly expected ? is the coral type structure detected in the space of sample 1 merely a local nature ? ( or , could it be found in other areas of the survey ? ) . here , we apply the sorting out method proposed above to the whole 2dfgrs data set . to match the basic criterions of sample 1 , we simply adopt sample 2 for our analysis .
the @xmath217 used to determine the @xmath33 is that directly measured from sample 2 ( i.e. , @xmath218 in table 1 ) .
[ fig : 21 ] the largest number of galaxies of the coherent clusters identified in sample 2 is 41813 , and the scale of this cluster is @xmath219 which is the largest one detected with the adopted method .
the second largest cluster contains 32188 galaxies and its scale is @xmath220 .
the two - dimensional positions ( relative to that of other galaxies of sample 2 ) of the two coherent clusters are illustrated in figure 22 which shows that the structures of the two groups of galaxies are situated in two distinct areas of the sky . comparing it with figure 17
we find that the two largest coherent clusters identified from sample 1 , clusters 1 and 2 , now become a single one , the second largest cluster identified from sample 2 ( the one shown in the left - hand side of figure 22 ) . besides those of clusters 1 and 2 , some other sources are also included in the second largest cluster .
this is expectable , since the area of sample 1 is well inside that of sample 2 in that direction of the sky ( see figure 1 ) , and some bridge sources outside the area of sample 1 might exist and then would connect clusters 1 and 2 together as well as connect other sources or clusters to them .
the right - hand side of figure 22 shows that large - scale structures could extend to the space as far as the adopted samples could reach ( the largest redshift detected in the right - hand side structure , the largest coherent cluster detected from sample 2 , is @xmath221 ) .
( for the left - hand side structure , the largest redshift detected is @xmath222 . )
this indicates that the large - scale structure is not a patent right of nearby galaxies .
instead , it could be formed in a quite early time , although space sampling is sparse at that epoch ( see figure 8 panel a ) .
in fact , it was found in recent observations that large structures are common at redshifts as large as @xmath12 ( steidel et al .
large - scale structures which are made up of protoclusters could be formed even at the epoch as early as @xmath223 ( ouchi et al .
2005 ) .
the structure of the largest cluster detected from sample 2 , that in the right - hand side of figure 22 , is displayed in figures 23 - 24 , where more details are visible ( shown in figure 23 is also the two dimensional structure of the second largest cluster ) .
figure 23 , which is a fine map , shows that the topology of the structure is coral - like as well .
although the probability for forming the two largest coherent clusters by chance is not available due to the ill - defined borders of sample 2 , we tend to believe that the two clusters are not formed by chance . according to figure 12 ,
large numbers of galaxies of coherent clusters are scarcely formed by chance , even their mean relative densities are small . in sample 3 ,
the largest number of galaxies is much less than 300 ( see figure 13 ) , but the largest one detected from sample 1 is 12966 which is 43 times of 300 .
the largest and the second largest numbers of galaxies of the coherent clusters identified from sample 2 are 139 and 107 times of 300 , respectively .
it is unlikely that the enhancement of the volume from sample 1 to sample 2 , where the number of galaxies of the latter is about 4 times of the former , would shift the @xmath0 probability contour in figure 13 to the number of galaxies as large as that of the second largest cluster , 32188 ( a detailed analysis on this issue will be performed later ) .
our analysis is performed in redshift space , where the peculiar velocity of individual sources is ignored . as is known , in measuring the distance of an object ,
a big problem is the joining of its dynamical velocity to the redshift observed ( see , e.g. , gellar & huchra 1989 ; guzzo 2002 ) . generally , the velocity distortion would shrink overdense regions and inflate underdense ones ( kaiser 1987 ; martinez & saar 2002a ) . as shown in figure 13 , correcting this effect would raise some weak clusters of sample 1 to be less prominent clusters since their number densities ( represented by @xmath122 ) would become larger .
in contrast , the correction would make some prominent clusters to be less prominent ones since their number densities would become smaller . when the redshift distortion is removed , the pattern of clustering observed above might be altered . however , as revealed by figure 18 , the possible change of the two largest structures of sample 1 would not be so severe , since the structures themselves are firmly built on the very dense frames . the measurement uncertainty of redshift for the 2df galaxy redshift survey is @xmath224 ( colless et al .
2001 ) , which corresponds to @xmath225
. the line of sight positions will certainly be affected by errors in redshifts .
would this play an important role in identifying clusters , where the fof algorithm is applied ?
the answer is yes when the redshift concerned is as small as @xmath226 .
accordingly , as revealed by table 2 , clusters 14 , 16 , 22 , 33 , 47 , 52 , 54 , 61 , and 64 will certainly be affected by the redshift measurement uncertainty .
when much smaller redshift uncertainty is available , some of these clusters might be disassembled while some of them might be merged into other clusters .
however , as illustrated in figure 3 , most sources in sample 1 have their redshifts much larger than the uncertainty and thus the effect of redshift errors must be very limited in identifying clusters among them .
in particular , the largest clusters discussed above are located in relatively high redshift regions ( see figure 22 and table 2 )
. they would be much less affected by this uncertainty .
colless et al .
( 2001 ) pointed out that the survey magnitude limit of 2dfgrs varies slightly with position on the sky . in the sgp
the median limiting magnitude is @xmath227 , with an rms of 0.05 mag ; in the ngp the median limiting magnitude is @xmath228 , with an rms of 0.11 mag . to show how this effect plays a role , we repeat the grouping analysis performed above by adopting a brighter magnitude cut so that the sample selected will be more complete in magnitude . according to colless et al .
( 2001 ) , we take the limiting magnitude as @xmath229 and consider only galaxies brighter than it . under this condition ,
there are 198647 sources left in sample 2 ( @xmath230 ) and 55846 galaxies remained in sample 1 ( @xmath231 ) . due to the cut in the total number of the samples ,
a straightforward result is that all clusters identified above have smaller members and smaller values of @xmath122 .
this is true .
cluster 1 now has only 12391 sources with @xmath232 . for cluster 2 ,
the number of members now becomes 6237 and the value of @xmath122 is 3.97 .
the scale of cluster 1 remains unchanged , while that of cluster 2 now becomes @xmath233 .
this is not surprising since faint sources are always far away sources and cluster 2 is identified in a larger distance than that of cluster 1 .
spatial distributions of the two clusters are shown in fig .
25 , where , for the sake of comparison , the original members of the two clusters are also presented .
one finds that while cluster 1 remains almost the same , cluster 2 is obviously affected .
some original substructures of cluster 2 are dismissed while some new substructures are attached .
the following previous conclusion holds : clusters could stretch to large distances .
( what would happen when continuously adopting brighter magnitude limits ?
currently we have no answers to this .
it deserves an intensive investigation . ) besides the number and density of clusters , the new magnitude limit might also affect the possibility contour of sample 1 . since the total number of the sample becomes smaller than the previous one , clusters identified with the same algorithm would have smaller numbers and would be less dense . in this way , the contours shown in figs . 12 and 13 would shift leftwards .
as revealed by fig .
13 , clusters 1 and 2 are far away from the @xmath180 contour , even though they have smaller numbers than before due to the brighter limit .
the conclusion that the probability of picking the two clusters by chance is much less than that of @xmath180 maintains .
some authors proposed that the three dimensional topology of large - scale structures is spongelike ( gott iii et al .
1986 ; vogeley et al . 1994 ; colley et al .
2000 ) . indeed , when plotting the structures of the two largest clusters identified from sample 1 in a lower resolution map , this type of feature emerges ( see figure 25 ) .
it suggests that the large - scale structure of the local universe is intrinsically coral - like , but it could also be spongelike if the resolution of the map is low enough . as already pointed out by gott iii et al .
( 2005 ) that some larger structures are found if larger samples are involved .
this is natural and is indeed observed in the analysis on sample 2 , where the two large - scale structures detected in a smaller sample ( sample 1 ) merge and a much larger structure is identified ( see figure 22 ) .
( accordingly , as the volume of sample 2 occupies only a small fraction of the nearby space , it is suspected that we might be able to observe a structure with its scale as large as @xmath234 in the local universe so long as the whole sky is surveyed . ) implied by figure 21 , some of the coherent clusters and unclustered sources seem to be able to connect with each other in a two - dimensional view .
this is indeed true . when one repeats the above analysis on sample 1 under the condition that the contribution of the vertical length ( i.e. , the contribution of the coordinate @xmath41 defined in figure 6 ) is ignored , one identifies a much larger coherent cluster than the largest one obtained above ( the details of the analysis are omitted ) .
this suggests that some large - scale structures observed in a two - dimensional map might not be a coherent one when the vertical length is taken into account .
as shown in table 2 , the mean redshift of cluster 1 is 0.0864 . at this redshift ,
the vertical length confined by the adopted window of sample 1 ( say , @xmath235 ) is @xmath236 , while the adopted criterion clustering scale at this redshift is @xmath237 .
the former is about 8 times of the latter . keeping this in mind , the phenomenon that some structures connected in a two - dimensional map are in fact separated is expectable .
note that @xmath33 might be different for different surveys .
it is expected that the number of galaxies we will eventually ever be able to see would be larger than that observable today by a factor of @xmath238 ( gott iii et al . 2005 ) . if sample 1 is enlarged by two times , @xmath33 might probably shrink to a smaller scale .
it is unnecessary that this would lead to a smaller scale of the largest structure identified with @xmath33 .
one reason is that the enlarging of the sample makes the sources more crowded and then separated sources are easier to be connected .
another reason is that the largest structure is constructed with the frames made up of very crowded sources ( with several times of number density than that of the background data ) .
the enlargement of the sample ( only two times larger ) would not be able to change this situation .
it should be pointed out that the structure scale as large as @xmath1 obtained from sample 1 and that as large as @xmath219 gained from sample 2 depend on the criterion clustering scale adopted in this paper , @xmath239^{-1/3}$ ] .
it is obvious that when a smaller @xmath33 is adopted , then a smaller value of the largest structure scale identified from the samples would be obtained , and when a larger one is taken , then one would get a larger value of the largest structure scale . thus , talking the largest structure scale one should refer to the corresponding criterion clustering scale . due to this and due to the fact that the criterion clustering scale adopted in this paper varies with redshift , the comparison of the largest structure scale between our analysis and other previous ones has less meaning .
it is desired that we can construct a standard criterion clustering scale one day and then can compare the largest structure scale obtained from various samples or from different regions of the universe .
although we are unable to compare our results with others , the algorithm adopted in this paper tends to pick out larger structures than what identified previously , when similar criterion clustering scales are adopted .
it is because that we are identifying large - scale structures in the co - moving frame of the universe rather than in the observer frame .
it is natural that those galaxies excluded in the previously identified large - scale structures due to their large redshifts are possible becoming members of some large - scale structures identified in this paper when they are located in a relatively dense region ( see fig .
martinez et al . ( 1998 ) searched for the scale of homogeneity by applying the @xmath30 function to galaxy catalogues .
they detected the fingerprint of the transition to homogeneity in all the cases considered .
amendola & palladino ( 1999 ) analyzed volume - limited subsamples of a redshift survey to search for the scale of homogeneity and found that the survey shows a trend of homogeneity at large scales .
as shown in figure 4 , one finds a trend toward homogeneity as well .
we argue that , at least what revealed by figure 4 could not be interpreted as the existence of a transition to homogeneity at a large scale since that scale is comparable to the maximum scale of the adopted sample ( see the caption of figure 4 ) . as pointed out in section 3 , we suspect that the flux limit as well as the possible evolutionary effect might be the main factors that block the detection of the real scale of homogeneity .
the previous methods can not solve this problem , neither can the neighborhood function introduced in this paper .
only when this effect is removed , one will be able to determined the homogeneity scale with galaxy redshift surveys .
it may be an intrinsic property that this method naturally produces larger structures than other approaches do . as suggested above
, filamentary features are observed in the state of random distribution , formed by crowded sources ( see figure 10 ) .
this is in agreement with the previous prediction that galaxies preferentially formed in large - scale filamentary or sheet - like mass overdensities in the early universe , which was detected at as far as @xmath14 ( matsuda et al . 2005 ) . when sorting out all sources of the background sample with @xmath33 we get many weak clusters and few less prominent clusters ( see figure 15 panels d and f ) .
according to definition , these clusters are likely the so - called protoclusters ( venemans 2005 ) .
the latter were identified at @xmath15 in recent observations ( venemans et al . 2002
; miley et al .
2004 ; intema et al .
2006 ) .
according to the above analysis we conclude that : 1 ) the probability of forming the large - scale structures , detected from sample 1 with our method , by chance is very small ; 2 ) the phenomenon of clustering is dominant in the local universe ; 3 ) coherent clusters with the scale as large as @xmath240 and the number of galaxies as large as 12966 are identified from sample 1 well within the @xmath201 confidence level ; 4 ) there exist some galaxies which are not affected by the gravitation of clusters and therefore are likely to rest on the co - moving frame of the universe ; 5 ) filamentary features could appear in the state of random distribution ; 6 ) voids are likely the volumes within which no very crowded sources are present and they are likely formed in embryo by fluctuation in the very early epoch of the universe and it might be the continuous gravitation in later time that pulls crowded galaxies closer and in the same time leaves behind adult voids ; 7 ) large - scale structures are coral - like and they are likely made up of smaller ones and in forming the former the latter are likely connected by their antennae ; 8) very crowded sources are mainly distributed within the structure of prominent clusters and it is them who form the frame of the large - scale structure .
this work was supported by the national science fund for distinguished young scholars ( 10125313 ) , the national natural science foundation of china ( no . 10573005 ) , and the fund for top scholars of guangdong provience ( q02114 ) .
we also thank the financial support from the guangzhou education bureau and guangzhou science and technology bureau . | in this paper , a relative number density parameter , called the neighborhood function , is introduced so that the crowded nature of the neighborhood of individual sources could be described . with this parameter
one can determine the probability of forming a cluster with a given number of galaxy members and a certain number density by chance .
a method is proposed to identify large - scale structure on the cosmological co - moving frame . to avoid those effects arising from the distance of objects
when applying the method , we compare the real number density concerned with the mean number density measured in the corresponding local area rather than the mean of the whole sample .
the scale used to sort out clustering sources is determined by that mean local number density , and thus it is a redshift - dependent scale . in applying this method ,
the sample adopted is required to have regular borders so that a simulation over the sample area could be performed and the volume confined by the sorting out scale within the sample area could be determined and then the probability referring to number density could be evaluated .
the method is applied to a sample drawn from the 2df survey , analyzed in redshift space .
we find from the analysis that the probability of forming the resulting large - scale structures by chance is very small and the phenomenon of clustering is dominant in the local universe . within the @xmath0 confidence level ,
a coherent cluster with its scale as large as @xmath1 and another with its number of galaxy members as large as 12966 are identified from the sample .
there exist some galaxies which are not affected by the gravitation of clusters and hence are suspected to rest on the co - moving frame of the universe .
voids are likely the volumes within which no very crowded sources are present and they are likely formed in embryo by fluctuation in the very early epoch of the universe . in addition , we find that large - scale structures are coral - like and they are likely made up of smaller ones ; sources with large values of the neighborhood function are mainly distributed within the structure of prominent clusters and it is them who form the frame of the large - scale structure . * * the neighborhood function and its application to identifying large - scale structure in the comoving universe frame * * * y .- p .
qin@xmath2 , l .- z .
l@xmath3 , f .- w .
zhang@xmath4 , b .- b . zhang@xmath4 and j. zhang@xmath4 * * @xmath5center for astrophysics , guangzhou university , guangzhou 510006 , p. r. china * * @xmath6national astronomical observatories / yunnan observatory , chinese academy of sciences , kunming 650011 , p. r. china * * @xmath7physics department , guangxi university , nanning 530004 , p. r. china * * @xmath8 the graduate school of the chinese academy of sciences , beijing 100049 , p. r. china * * key words * : cosmology : observations galaxies : distances and redshifts galaxies : statistics large - scale structure of universe methods : data analysis |
organic charge - transfer salts show a wide variety of quantum phases and represent prominent examples to study correlation effects in low - dimensional systems .
the most celebrated case is given by the ttf - tcnq salt that has been primarily regarded as a prototype for testing theories of one - dimensional conductors .
@xcite organic salts may also form crystals in two and three dimensions , and , in this respect , an increasing attention has been devoted to a particular family denoted by @xmath2-(et)@xmath3x , whose building block is the so - called bedt - ttf ( or et ) molecule and x is a monovalent anion .
@xcite here , strongly dimerized et molecules are arranged in a two - dimensional triangular lattice .
each dimer has a charge state with one hole and therefore the conducting band is half filled .
a sizable effective coulomb repulsion is felt by two holes on the same dimer .
a huge variety of phases have been found ( by varying temperature , pressure or the nature of the anion x ) , ranging from correlated ( bad ) metals with superconductivity at low temperatures , to mott insulators with magnetic order .
@xcite interestingly , by acting with hydrostatic pressure , metal - insulator transitions have been observed , @xcite with the remarkable possibility to stabilize a non - magnetic mott insulating phase in @xmath2-(et)@xmath3cu@xmath3(cn)@xmath6 .
@xcite in this material , there is no evidence of magnetic order down to @xmath9 , which is four orders of magnitude lower than the estimate of the super - exchange coupling @xmath10 .
it has been argued that @xmath2-(et)@xmath3x compounds can be described by a single - band hubbard model on the anisotropic triangular lattice , @xcite where chains described by an hopping @xmath1 are coupled together with zig - zag hoppings @xmath0 , see fig .
[ fig : lattice ] .
an on - site repulsive term @xmath4 is also present in the hamiltonian .
however , a realistic estimate of these microscopic parameters is not exempt from complications .
indeed , the values obtained some time ago by extended hckel band structure calculations @xcite have been put in doubt by two recent ab - initio calculations , based upon local - density approximation ( lda ) and generalized gradient approximation ( gga ) .
@xcite interestingly , the new results suggest that these organic salts are less frustrated than previously assumed , and that @xmath11 is smaller than one .
indeed , the frustrating ratio is @xmath12 for @xmath2-(et)@xmath3cu@xmath3(cn)@xmath6 and @xmath13 for @xmath2-(et)@xmath3cu@xmath3(scn)@xmath3 .
@xcite other materials , with x = cu[n(cn)@xmath3]cl or cu[n(cn)@xmath3]br , have a substantially smaller frustrating ratio , i.e. , @xmath14 .
@xcite unfortunately , an accurate determination of the correlation energy is rather difficult and these two calculations give a considerably different estimation of the coulomb repulsion , namely @xmath15 ( ref . ) and @xmath16 ( ref . ) . here
, we apply our improved monte carlo calculations , based upon the recently introduced backflow wave function @xcite in order to analyze the possibility of having a non - magnetic insulator for large enough frustration and interaction strength .
the paper is organized as follow : in section [ sec : model ] , we introduce the hamiltonian ; in section [ sec : approach ] , we describe our variational wave function ; in section [ sec : results ] , we present our numerical results and , finally , in section [ sec : conc ] we draw the conclusions .
we consider the hubbard model described by @xmath17 where @xmath18 creates ( destroys ) an electron with spin @xmath19 on site @xmath20 , @xmath21 , @xmath22 is the hopping amplitude and @xmath4 is the on - site coulomb repulsion . in this work ,
we focus our attention on the half - filled case with @xmath23 electrons on @xmath23 sites and consider a square lattice with a nearest - neighbor hopping @xmath0 , along the @xmath24 and @xmath25 directions , and a further next - nearest - neighbor hopping @xmath1 along @xmath26 ; this choice of the hopping amplitudes is topologically equivalent to the anisotropic triangular lattice , @xcite see fig .
[ fig : lattice ] . in the last years
, an intense effort has been devoted to this problem by use of a large variety of methods , including exact diagonalization , @xcite path - integral renormalization group , @xcite variational monte carlo , @xcite cluster dynamical mean field theory , @xcite and dual fermions .
@xcite all these methods give rather different outcomes and there are huge discrepancies on the phase boundaries and , most importantly , on the expected nature of the non - magnetic insulator .
the aim of this work is to clarify the ground - state properties for two values of @xmath27 and @xmath28 , relevant for materials with x = cu@xmath3(scn)@xmath3 and cu@xmath3(cn)@xmath6 , respectively .
illustration of the lattice in the square topology ( a ) used in this work and in the equivalent triangular one ( b ) .
solid and dashed lines indicate hopping amplitudes @xmath0 and @xmath1 , respectively . ]
( color online ) results for 18 electrons on 18 sites as a function of @xmath29 .
upper panels : accuracy of energy @xmath30 , @xmath31 and @xmath32 being the variational and the exact values , respectively .
lower panels : overlap between the exact ground state and the variational bcs wave functions .
the state without ( with ) backflow correlations is denoted by diamonds ( squares ) . ]
a variational wave function for an insulator with antiferromagnetic ( af ) order can be constructed by considering the ground state @xmath33 of a mean - field hamiltonian containing a band contribution and a magnetic term : @xmath34 where @xmath35 is a unitary vector that depends upon the lattice site @xmath20 and @xmath36 is the spin operator .
moreover , @xmath37 is a variational band term : @xmath0 gives the energy scale and @xmath38 can be optimized to minimize the variational energy . in order to correctly describe spin fluctuations orthogonal to the plane where the magnetic order lies
, we take @xmath35 in the @xmath39 plane and we include a spin jastrow factor @xmath40 $ ] in the wave function .
@xcite another density jastrow factor @xmath41 $ ] ( that includes the on - site gutzwiller term ) is considered to adjust electron correlations . in summary ,
the correlated wave function is defined by @xmath42 notice that , in this case , the variational state has not a definite total spin , which is suitable for a magnetically ordered phase .
in fact , both @xmath33 and the spin jastrow factor @xmath43 break the su(2 ) symmetry . on the other hand , superconducting or metallic phases can be constructed by considering the ground state @xmath44 of a superconducting bardeen - cooper - schrieffer ( bcs ) hamiltonian with both band and pairing contributions , @xcite @xmath45 here the band term may also contain a variational chemical potential @xmath46 , since the bcs hamiltonian does not conserve the particle number , i.e. , @xmath47 . in this case ,
@xmath38 and @xmath46 can be varied to optimize the variational wave function .
the full correlated state is given by @xmath48 in this case , no spin jastrow is considered , in order to have a perfect singlet state , suitable for a non - magnetic phase .
notably , within this kind of wave function , it is possible to obtain a pure ( i.e. , non - magnetic ) mott insulator just by considering a sufficiently strong jastrow factor , i.e. , @xmath49 ( @xmath50 being the fourier transform of @xmath51 ) .
@xcite as we recently demonstrated , @xcite the projected bcs state is not sufficiently accurate for hubbard - type models , especially in the important strong - coupling regime , i.e. , for @xmath52 , where the super - exchange energy scale @xmath53 is not correctly reproduced .
one efficient way to overcome this problem is to consider backflow correlations , @xcite that modify the single - particle orbitals , @xcite in the same spirit of what was put forward long - time ago by feynman and cohen .
@xcite ( color online ) variational energies per site ( in unit of @xmath53 ) for the bcs state with a density jastrow factor ( diamonds ) and for the af wave function with both density and spin jastrow terms ( circles ) ; the correlated fermi gas with jastrow factor is also reported for @xmath54 ( squares ) .
all states have backflow correlations and results are for 100 sites . ] following ref . , we consider a general definition of the new `` orbitals '' by taking all the possible virtual hoppings of the electrons : @xmath55 where we used the notation @xmath56 , being @xmath57 the eigenstates of the mean - field hamiltonian , @xmath58 , @xmath59 , with @xmath60 .
@xmath61 , @xmath62 , @xmath63 , and @xmath64 are variational parameters . as a consequence , already
the determinant part of the wave function includes correlation effects .
the backflow corrections of eq .
( [ backlattice2 ] ) ( in particular the @xmath62 term ) make it possible to mimic the effect of the virtual hopping , which leads to the super - exchange mechanism .
all the parameters of the wave function can be optimized by using the method of ref . .
finally , the accuracy of the variational calculations can be assessed by using lanczos diagonalizations on small lattices and green s function monte carlo within the so - called fixed - node approximation , @xcite which gives accurate ( but approximate ) results on large systems .
a detailed description of the fixed - node approximation can be found in ref . .
in brief , starting from the original hamiltonian @xmath65 , we define an effective hamiltonian by adding a perturbation @xmath66 : @xmath67 the operator @xmath66 is defined through its matrix elements and depends upon a given _ guiding function _
@xmath68 , that is for instance the variational state itself @xmath69 where @xmath70 and @xmath71 .
notice that the above operator annihilates the guiding function , namely @xmath72 .
therefore , whenever the guiding function is close to the exact ground state of @xmath65 , the perturbation @xmath66 is expected to be small and the effective hamiltonian becomes very close to the original one .
by allowing the most general singlet and complex bcs pairing in the state without backflow terms , we find that this quantity has @xmath73 symmetry up to @xmath74 , namely the best ( nearest - neighbor ) pairing function is @xmath75 , in agreement with previous results .
@xcite therefore , within our improved backflow wave function , we only considered a real bcs pairing .
we mention that @xmath76 is very small ( especially in the presence of backflow correlations ) in the conducting phase , see table [ tab : delta ] , and it becomes sizable only in the regime where the magnetic solution prevails over the bcs state . in this regard
, we do not find a clear signature of superconductivity close to the metal - insulator transition , as suggested in ref . .
we also stress that , once the backflow correlations are considered , there is no energy gain by allowing a translational symmetry breaking ( e.g. , by considering a @xmath77 unit cell in the bcs hamiltonian , suitable for dimerized states ) and the @xmath73 solution has always a lower energy than dimerized states . finally , we find that the variational band term of the bcs hamiltonian @xmath78 has @xmath79 for most of the cases considered , except for small @xmath29 , inside the conducting phase , where a finite @xmath38 can be stabilized .
.[tab : delta ] bcs pairing @xmath76 for various @xmath29 in the metallic region for two sizes of the lattice : @xmath80 ( third column ) and @xmath81 ( fourth column ) .
notice that for @xmath82 and @xmath54 and for @xmath83 and @xmath27 the bcs wave function is still metallic but the af one ( insulating ) has a lower variational energy .
[ cols="^,^,^,^",options="header " , ] the metal - insulator transition can be detected by a direct inspection of the static density - density correlations @xmath84 in fact , this quantity makes it possible to discriminate between gapless ( conducting ) and gapped ( insulating ) phases : a linear behavior @xmath85 for @xmath86 is typical of a conducting phase , whereas a quadratic behavior @xmath87 can be associated to an insulating character .
@xcite the results presented in fig .
[ fig : nq ] indicate that a metal - insulator transition takes place by increasing @xmath29 and it can be placed at @xmath88 and @xmath89 for @xmath27 and @xmath28 , respectively .
the transition is first order , with a small jump in the linear coefficient of @xmath90 for small momenta .
in fact , for small @xmath29 , the best wave function is the bcs one ( with small superconducting pairing ) , whereas , by increasing the interaction , the af one prevails , thus inducing a metal - insulator transition , see fig . [
fig : energy ] .
( color online ) size scaling of the spin - spin correlations @xmath91 for @xmath92 .
data are for @xmath27 with @xmath93 ( squares ) and @xmath94 ( circles ) , and @xmath54 with @xmath93 ( triangles ) and @xmath94 ( diamonds ) .
variational and fixed - node results are denoted by empty and full symbols , respectively . the variational results do not depend substantially upon @xmath4 and @xmath1 .
the fixed - node results indicate long - range order for @xmath27 but not for @xmath54 . ]
( color online ) proposed phase diagram for the two discussed hopping ratios , @xmath27 and @xmath54 . in the first case , variational ( vmc ) and fixed - node results ( fn )
indicate both a direct transition between a metal and an insulator with af nel order at a critical value of the electron - electron repulsion @xmath95 . for @xmath54 ,
the variational results predict the existence of three different phases at increasing @xmath29 : a metal , an af insulator with nel order and a spin liquid , while , within the fixed - node approximation , the non - magnetic ground state extends down to the metal - insulator transition . ] in the insulating regime and for small frustrating ratios , the af wave function has always a lower energy than the spin - liquid state , and this fact is particularly evident close to the transition , see fig .
[ fig : energy ] . on the contrary ,
for the case with @xmath54 , the bcs state competes with the af one and it becomes better in energy for @xmath96 , indicating an insulating spin - liquid behavior at large @xmath4 ( notice that in this region @xmath87 ) . in this regime , the bcs pairing is relevant , since the simple projected fermi sea has a much higher energy , see fig .
[ fig : energy ] .
remarkably , the bcs and af variational energies are always quite close for @xmath54 , suggesting that the actual ground state might be non - magnetic for all @xmath97 , or at least down to values lower than expected on the basis of the variational estimate .
this fact is supported by the fixed - node calculations for the spin - spin correlations @xmath98 in fig .
[ fig : sq ] , we report the size scaling of the variational and the fixed - node results by considering the bcs state as the _
guiding function_. we stress the fact that , in the insulating regime , @xmath99 has a peak at the commensurate momentum @xmath92 .
remarkably , the fixed - node approach is able to recover a finite value of @xmath91 for @xmath92 ( i.e. , the square of the magnetic order parameter ) in the thermodynamic limit for @xmath27 , even though the bcs wave function is non magnetic . by contrast , @xmath91 tends to zero for @xmath54 ( both for @xmath93 and @xmath100 ) , supporting the fact that the ground state is non magnetic for this frustrating ratio , even close to the metal - insulator transition .
the resulting phase diagram is summarized in fig .
[ fig : phase_diagram ] .
we have studied the anisotropic triangular lattice at half filling away from the isotropic point @xmath101 , with @xmath102 , using both a gutzwiller - jastrow variational ansatz including backflow correlations as well as a green s function monte carlo approach within the fixed node approximation .
we find that the square lattice states persist up to large values of @xmath103 , both in terms of the d - wave superconducting order parameter as well as for the af nel ordering .
the main outcome of this work is that , thanks to the improvement given by backflow correlations , a spin - liquid wave function can be stabilized over magnetic states , for large but still moderate coulomb repulsions and close to the isotropic limit .
these variational results are corroborated by fixed - node calculations .
we find , in particular , that for @xmath54 , which is relevant for @xmath2-(et)@xmath3cu@xmath3(cn)@xmath6 , @xcite the insulating phase has a pure mott character , without magnetic order . on the other hand , for @xmath27 , suitable for @xmath2-(et)@xmath3cu@xmath3(scn)@xmath3 , ( or even smaller @xmath11 values ) the insulating phase always shows nel order with @xmath92 .
let us finish by discussing our results also in comparison to other calculations and experimental findings .
first of all , in various papers , it has been suggested that the spin - liquid phase can be stabilized by charge fluctuations that may destabilize a magnetically ordered state .
this claim has been corroborated by calculations on heisenberg models in presence of a ring - exchange term @xmath104 ( that appears in the strong - coupling expansion in @xmath105 ) .
@xcite however , it turns out that the actual value of @xmath104 for stabilizing a disordered phase is rather large and , probably , beyond the validity of a perturbative expansion
. the existence of a direct transition from a magnetic phase to a disordered one has been also found in the original hubbard model , by decreasing the on - site repulsion @xmath4 .
@xcite we do not find any evidence of such a possibility and , in our approach , the magnetic phase is stable in presence of charge fluctuations , even close to the metal - insulator transition : this is the case of @xmath27 .
instead , the spin - liquid phase turns out to be directly connected with the one at strong coupling , while antiferromagnetic correlations become stronger when decreasing @xmath29 .
for example , for @xmath54 , the variational state with magnetic order has a slightly lower energy close to @xmath106 and we need to apply the green s function monte carlo approach to extend the spin - liquid region down to the metal - insulator transition , see fig .
[ fig : phase_diagram ] . at this stage
, we would also like to mention that the metallic phase is likely to be not superconducting .
in fact , the bcs pairing @xmath76 in the metallic region is slightly suppressed when improving the accuracy of the variational wave function by considering backflow correlations and , moreover , it is reduced by a factor @xmath107 when the lattice size is increased from @xmath108 to @xmath109 , see table [ tab : delta ] .
this fact contrasts the previous claim of a possible superconducting phase close to the metal - insulator transition by liu and collaborators .
@xcite another very important point is to clarify the nature of the low - energy excitations .
very recently , thermodynamic measurements of the specific heat suggested the possible existence of a fermi surface of neutral , @xmath110 fermionic spinons .
@xcite however , it should be noticed that such a measurement involves a difficult subtraction of a divergent nuclear specific heat , and instead the thermal conductivity ( which is not affected by a nuclear contribution ) shows an activated behavior with a tiny gap of @xmath111 .
@xcite this fact has been associated with the existence of spinless `` vison '' excitations .
@xcite from our calculations , it appears that the disordered insulating phase can not sustain a true spinon fermi surface , as previously suggested both on variational calculations @xcite and field - theory approaches , @xcite but it has dirac points at @xmath112 .
in fact , the projected fermi - sea has a much higher energy than our best variational ansatz with bcs pairing , see fig .
[ fig : energy ] .
should our results be correct , either a deeper investigation of the minimal microscopic model for describing organic charge - transfer salts is needed or a reinterpretation of the experimental data is required .
99 j. solyom , adv . phys . * 28 * , 201 ( 1979 ) .
k. kanoda , j. phys .
. jpn . * 75 * , 051007 ( 2006 ) .
k. miyagawa , a. kawamoto , y. nakazawa , and k. kanoda , , 1174 ( 1995 ) .
h. elsinger , j. wosnitza , s. wanka , j. hagel , d. schweitzer , and w. strunz , , 6098 ( 2000 ) .
s. lefebvre , p. wzietek , s. brown , c. bourbonnais , d. jerome , c. meziere , m. fourmigue , and p. batail , , 5420 ( 2000 ) .
p. limelette , p. wzietek , s. florens , a. georges , t.a .
costi , c. pasquier , d. jerome , c. meziere , and p. batail , , 016401 ( 2003 ) .
f. kagawa , t. itou , k. miyagawa , and k. kanoda , , 064511 ( 2004 ) .
y. kurosaki , y. shimizu , k. miyagawa , k. kanoda , and g. saito , , 177001 ( 2005 ) .
y. shimizu , k. miyagawa , k. kanoda , m. maesato , and g. saito , , 107001 ( 2003 ) .
h. kino and h. fukuyama , j. phys .
. jpn . * 65 * , 2158 ( 1996 ) .
mckenzie , comments condens .
matter phys . * 18 * , 309 ( 1998 ) ; science * 278 * , 820 ( 1997 ) .
k. nakamura , y. yoshimoto , t. kosugi , r. arita , and m. imada , arxiv:0903.5409 .
kandpal , i. opahle , y .- z .
zhang , h.o .
jeschke , and r. valenti , arxiv:0904.0302 .
tocchio , f. becca , a. parola , and s. sorella , , 041101(r ) ( 2008 ) . at half
filling , the sign of @xmath1 is irrelevant . r.t .
clay , h. li , and s. mazumdar , , 166403 ( 2008 ) .
h. morita , s. watanabe , and m. imada , j. phys .
jpn . * 71 * , 2109 ( 2002 ) .
t. watanabe , h. yokoyama , y. tanaka , and j. inoue , j. phys .
jpn . * 75 * , 074707 ( 2006 ) .
t. watanabe , h. yokoyama , y. tanaka , and j. inoue , , 214505 ( 2008 ) .
s. guertler , q .- h .
wang , and f.c .
zhang , , 144526 ( 2009 ) .
b. kyung , a .- m .
s. tremblay , , 046402 ( 2006 ) .
t. ohashi , t. momoi , h. tsunetsugu , and n. kawakami , , 076402 ( 2008 ) .
h. lee , g. li , and h. monien , , 205117 ( 2008 ) .
f. becca , m. capone , and s. sorella , , 12700 ( 2000 ) . c. gros , , 931(r ) ( 1988 ) .
zhang , c. gros , t.m .
rice , and h. shiba , supercond .
* 1 * , 36 ( 1988 ) . m. capello , f. becca , m. fabrizio , s. sorella , and e. tosatti , , 026406 ( 2005 ) .
single - particle states are defined in the bcs hamiltonian , after the transformation @xmath113 .
r.p . feynman and m. cohen ,
phys . rev . *
102 * , 1189 ( 1956 ) .
s. yunoki and s. sorella , , 014408 ( 2006 ) .
ten haaf , h.j.m .
van bemmel , j.m.j .
van leeuwen , w. van saarloos , and d.m .
ceperley , , 13039 ( 1995 ) .
m. lugas , l. spanu , f. becca , and s. sorella , , 165122 ( 2006 ) .
powell and r.h .
mckenzie , , 047004 ( 2005 ) .
gan , y. chen , z.b .
su , and f.c .
zhang , , 067005 ( 2005 ) .
j. liu , j. schmalian , and n. trivedi , , 127003 ( 2005 ) .
powell and r.h .
mckenzie , , 027005 ( 2007 ) .
motrunich , , 045105 ( 2005 ) .
s. yamashita , y. nakazawa , m. oguni , y. oshima , h. nojiri , y. shimizu , k. miyagawa , and k. kanoda , nature phys .
* 4 * , 459 ( 2008 ) .
m. yamashita , n. nakata , y. kasahara , t. sasaki , n. yoneyama , n. kobayashi , s. fujimoto , t. shibauchi , and y. matsuda , nature phys .
* 5 * , 44 ( 2009 ) .
y. qi , c. xu , and s. sachdev , , 176401 ( 2009 ) .
lee , p.a .
lee , and t. senthil , , 067006 ( 2007 ) . | the two - dimensional hubbard model on the anisotropic triangular lattice , with two different hopping amplitudes @xmath0 and @xmath1 , is relevant to describe the low - energy physics of @xmath2-(et)@xmath3x , a family of organic salts .
the ground - state properties of this model are studied by using monte carlo techniques , on the basis of a recent definition of backflow correlations for strongly - correlated lattice systems .
the results show that there is no magnetic order for reasonably large values of the electron - electron interaction @xmath4 and frustrating ratio @xmath5 , suitable to describe the non - magnetic compound with x = cu@xmath3(cn)@xmath6 . on the contrary
, nel order takes place for weaker frustrations , i.e. , @xmath7 , suitable for materials with x = cu@xmath3(scn)@xmath3 , cu[n(cn)@xmath3]cl , or cu[n(cn)@xmath3]br . |
despite being discovered 75 years ago@xcite , it still not well known what dark matter ( dm ) is composed of , and its properties .
the current paradigm is that dm is composed of slowly moving very massive particles , thus being called cold dark matter@xcite ( cdm ) . despite its high predictive power and being in good agreement with large scale structure observations ,
there is still some controversy respecting the small scale structure formation .
there are a lot of different dm profiles , some empirical@xcite and some based on n - body simulations@xcite , with different properties .
although cdm is the most accepted model , some problems still remain giving it the place to another type of dm , the warm dark matter ( wdm ) , with masses in an intermediate range between some kevs to some mevs .
it can solve some of the problems of cdm , but also has some of its own .
so , despite more than 50 years of research in dm , there is still a lot to explore . on a complete unrelated field ,
it is believed today that there is a supermassive black hole on the center of almost every galaxy .
recent observations@xcite of a microwave emission from the center of the milky way suggest that there is a considerable amount of dm in the center .
given these recent results , it could be possible that there is no black hole , but instead a distribution of dm particles .
some authors@xcite suggested that an extended distribution of fermions , a `` fermion ball '' , supported by degeneracy pressure , is present on the center of galaxies instead of a supermassive black hole .
we propose here a preliminary model for dm that tries to explain both the dm halos and the objects at the centers of galaxies by a self - gravitating system of semidegenerated fermions .
the equilibrium configurations of a self - gravitating semi - degenerate system of fermions were already studied by others@xcite .
we are considering spherical symmetry , so the spacetime is described by the schwarzchild metric @xmath0 solving the einstein equation we get the equilibrium equations : @xmath1 where @xmath2 the mass within a radius @xmath3 , @xmath4 the gravitational constant , @xmath5 the speed of light and @xmath6 and @xmath7 the mass - energy density and the pressure respectively , given by @xmath8 where @xmath9 is the multiplicity of states , @xmath10 is the planck constant , @xmath11 is the rest mass of the particle , @xmath12 the boltzmann constant , @xmath13 the temperature , @xmath14 the chemical potential and we write the integrals in terms of the kinetic energy of a single particle @xmath15 . to integrate numerically the set of equations ,
it is useful to change our variables to dimensionless ones : @xmath16 where @xmath17 has dimension of a length and @xmath18 is the planck mass . also , we write eq . [ dpdr ] in terms of the degeneracy parameter @xmath19 ( see refs . for details ) .
the final set of dimensionless equations is shown in ref . , which we integrate with initial conditions @xmath20 and @xmath21 .
it is important to notice that we can solve the whole system completely in a dimensionless way without specifying a mass for the particle , which appears only in the characteristic length @xmath22 and in the definition of the temperature parameter @xmath23 ; however , this parameter can be uniquely determined by the surface velocity , as will be shown later .
thus , the mass remains as only a scaling parameter and we are left with two free parameters , @xmath23 and @xmath24 .
galactic halos are thought to be composed of cold dark matter ( cdm ) , so that @xmath25.also , since we are interested in semidegenerate configurations , @xmath26 must be a non negative parameter@xcite . in fig .
[ fig1 ] we plot the degeneracy parameter and the density profile of the system as a function of the rescaled coordinate @xmath27 .
since the tail of the distribution scales like that , the system is unbounded and with infinite mass . in reality
, the system will be bound by tidal interactions , and a cutoff in the distribution function must be used@xcite .
the astrophysically relevant quantity is the rotation velocity , given by ( non - relativistic regime , since @xmath30 ) : @xmath31 * i : the degenerate core , where @xmath32 ; * ii : the first part of the inner halo , where the mass of the core prevails over the halo and @xmath33 ; * iii : second part of the inner halo , where the mass of the inner halo prevails over the halo , and @xmath32 ; * iv : the outer halo , where the velocity tends to constant after some oscillations . to confront the model with observations , it is useful to see how the physical quantities scale with the parameters of our model . our aim is to describe the galactic center and the dark matter halo , so we need to define both masses and radius for these regions .
the radius of the core ( @xmath34 ) is defined as the radius for the maximum velocity in region i ; the mass is then @xmath35 ; for the halo ( @xmath36 ) , we take the radius where the velocity is maximum in region iii , with corresponding mass @xmath37 ; finally , the last observable is the surface velocity ( @xmath38 ) , i.e. , the velocity in the outer halo where it is constant .
the scalings for these parameters are @xmath39 as we can see , the temperature parameter at the center , @xmath23 , can be uniquely determined by the observed surface velocity of the galaxy .
another interesting feature is that , while for the core the mass and radius scale with @xmath40 , in the halo they scale with @xmath41 .
these scaling laws are valid for @xmath42 $ ] and @xmath43 $ ]
. we can apply this model , for example , to the milky way .
the observed halo velocity is @xmath44 km / s , which gives @xmath45 .
taking the observations of the milky way , @xmath46 , @xmath47 and @xmath48 kpc ; and chosing eqs . [ vinf],[mc ] and [ rh ] for the three unknown parameters we have : @xmath49 which gives us @xmath50 pc .
this is unfortunately larger than the apocenter of the closest star to the center , ie .
the star s2 .
recently , some authors@xcite found that , by fitting velocity curves with a burkert profile , the dark matter surface density is constant within one dark matter halo scale - length for a wide span of galaxy magnitudes , @xmath51 .
this leads to a maximal acceleration due to dm at @xmath52 , @xmath53 . using this value in our model , taking into account that @xmath54 , we get from eqs . [ mh ] and [ rh ] @xmath55 common halo velocities range from tens to hundreds km / s , ie .
@xmath56 $ ] from eq .
[ vinf ] , and using the dm fermion mass obtained in the earlier application , we get that @xmath57 $ ] .
so , for a wide range of the parameters with a fixed dm mass , our model is in good agreement with this universality law . despite not being able to describe the core and the dm halo of our galaxy ,
our model is in good agreement with some observations . of course
, a cutoff in the distribution function is needed to avoid having an infinite mass , but the aim of this preliminary work is to show the scope of the fully relativistic treatment of the model in the application to galactic haloes and also to our milky way , concluding that some new physics must be added to the model in order to have a unified approach for the later case . | we propose a unified model for dark matter haloes and central galactic objects as a self - gravitating system of semidegenerated fermions in thermal equilibrium .
we consider spherical symmetry and then we solve the equations of gravitational equilibrium using the fermi integrals in a dimensionless manner , obtaining the density profile and velocity curve .
we also obtain scaling laws for the observables of the system and show that , for a wide range of our parameters , our model is consistent with the so called universality of the surface density of dark matter . |
a functional regression model with functional response variable can be defined by @xmath0 where @xmath1 ( @xmath2 ) stands for @xmath3 batches ( or curves ) of functional data , @xmath4 is an unknown nonlinear function , depending on a set of functional covariates @xmath5 and a set of scalar covariates @xmath6 , and @xmath7 is the random error .
a special case of such model is the following concurrent regression model with functional covariates @xmath5 ( see e.g. * ? ? ?
* ) @xmath8 however , when the relationship between the response and the covariates can not be justified as linear , it is intractable to model the function @xmath4 nonparametrically for multi - dimensional @xmath5 since most nonparametric regression models suffer from the _ curse of dimensionality_. a variety of alternative approaches with special model structures have been proposed to overcome the problem ; examples include dimension reduction methods , the additive model ( see e.g. * ? ? ?
* ) , varying - coefficient model ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and the neural network model ( see e.g. * ? ? ?
@xcite proposed a gaussian process functional regression ( gpfr ) model , which is defined by @xmath9 where @xmath10 is the mean structure of the functional data and @xmath11 represents a gaussian process regression ( gpr ) model having zero mean and covariance kernel @xmath12 ( for the detailed definition of gaussian process regression models , see * ? ? ?
* ; * ? ? ?
this nonparametric concurrent functional regression model can address the regression problem with multi - dimensional functional covariates and model the mean structure and covariance structure simultaneously ; see the detailed discussion in @xcite .
the aim of this paper is to extend the concurrent gpfr model to situations where the response variable , denoted by @xmath13 , is known to be non - gaussian .
the work is motivated by the following example , concerning data collected during standing - up manoeuvres of paraplegic patients .
the outputs are the human body s standing - up phases during rising from sitting position to standing position .
specifically , @xmath13 takes value of either 0 , 1 or 2 , corresponding to the phases of ` sitting ' , ` seat unloading and ascending ' or ` stablising ' respectively , required for feeding back to a simulator control system .
since it is usually difficult to measure the body position in practice , the aim of the example is to develop a model for reconstructing the position of the human body by using some easily measured quantities such as motion kinematic , reaction forces and torques , which are functional covariates denoted by @xmath14 .
this is to investigate the regression relationship between the non - gaussian functional response variable @xmath13 and a set of functional covariates @xmath14 .
since the standing - up phases are irreversible , @xmath13 is an ordinal response variable , taking value from three ordered categories .
if we assume that there exists an unobservable latent process @xmath15 associated with @xmath14 and the response variable @xmath13 depends on this latent process , then by using a probit link function , we can define a model as follows : @xmath16 where @xmath17 , @xmath18 , and @xmath19 are the thresholds .
now the problem becomes how to model @xmath20 by the functional covariates @xmath14 , or how to find a function @xmath21 such that @xmath22 .
more discussion of this example is given in section [ para_data ] and appendix g of the supplementary materials .
generally , letting @xmath23 be a given link function , a generalized linear regression model is defined as @xmath24 .
@xcite proposed a generalized linear mixed model to deal with heterogeneity : @xmath25 where @xmath26 is the coefficient for the fixed effect and @xmath27 is a random vector representing random effect .
however , if we have little practical knowledge on the relationship between the response variable and the covariates ( such as the case in the above paraplegia example ) , it is more sensible to use a nonparametric model . in this paper
, we propose to use a gaussian process regression model to define such a nonparametric model , namely a concurrent generalized gaussian process functional regression ( ggpfr ) model .
similar to gpfr model @xcite , the advantages of this model include : ( 1 ) it offers a nonparametric generalized concurrent regression model for functional data with functional response and multi - dimensional functional covariates ; ( 2 ) it provides a natural framework on modeling mean structure and covariance structure simultaneously and the latter can be used to model the individual characteristic for each batch ; and ( 3 ) the prior specification of covariance kernel enables us to accommodate a wide class of nonlinear functions .
this paper is organized as follows .
section 2 proposes the ggpfr model and describes how to estimate the hyper - parameters and how to calculate prediction , for which the implementation is mainly based on laplace approximation . the asymptotic properties , focusing on the information consistency ,
are discussed in section 3 .
several numerical examples are reported in section 4 .
discussion and further development are given in section 5 .
some technical details and more numerical examples are provided as the supplementary materials .
let @xmath28 be a functional or longitudinal response variable for the @xmath29-th subject , namely the @xmath29-th batch .
we assume that @xmath30 s are independent for different batches @xmath2 , but within the batch , @xmath31 and @xmath32 are dependent at different points .
we suppose that @xmath30 has a distribution from an exponential family with the following density function @xmath33 where @xmath34 and @xmath35 are canonical parameter and dispersion parameter respectively , both functional .
we have @xmath36 and @xmath37 , where @xmath38 and @xmath39 are the first two derivatives of @xmath40 with respect to @xmath41 .
suppose that @xmath5 is a @xmath42-dimensional vector of functional covariates .
nonparametric concurrent generalized gaussian process functional regression ( ggpfr ) models are defined by and the following @xmath43 here , the unobserved latent variable @xmath44 is modeled by a nonparametric gpr model via a gaussian process prior , depending on the functional covariates @xmath5 .
the gpr model is specified by a covariance kernel @xmath45 , and by the karhunen - love expansion @xmath46 where @xmath47 , @xmath48 are the eigenvalues and @xmath49 are the associated eigenfunctions of the covariance kernel .
one example of @xmath45 is the following squared exponential covariance function with a nonstationary linear term : @xmath50 where @xmath51 is a set of hyper - parameters involved in the gaussian process prior .
the hyper - parameter @xmath52 corresponds to the smoothing parameters in spline and other nonparametric models .
more specifically , @xmath53 is called the length - scale .
the decrease in length - scale produces more rapidly fluctuating functions and a very large length - scale means that the underlying curve is expected to be essentially flat .
more information on the relationship between smoothing splines and gaussian processes can be found in @xcite .
we can use generalized cross - validation ( gcv ) or empirical bayesian method to choose the value of @xmath54 .
when @xmath42 is large , gcv approach is usually inefficient .
we will use the empirical bayesian method in this paper ; the details are given in the next subsection .
some other covariance kernels such as powered exponential and matrn covariance functions can also be used ; see more discussion on the choice of covariance function in @xcite and @xcite
. in the model given by the response variable @xmath30 depends on @xmath5 at the current time only , therefore the proposed model can be regarded as a generalization of the concurrent functional linear model discussed in @xcite . in this model
the common mean structure across @xmath3 batches is given by @xmath10 .
if we use a linear mean function which depends on a set of @xmath55 scalar covariates @xmath6 only , can be expressed as @xmath56 in this case the regression relationship between the functional response @xmath30 and the functional covariates @xmath5 is modeled by the covariance structure @xmath57 .
other mean structures , including concurrent form of functional covariates , can also be used .
the proposed model has some features worth noting .
in addition to those discussed in section 1 , we highlight that the ggpfr model is actually very flexible .
it can model the regression relationship between the non - gaussian functional response and the multi - dimensional functional covariates nonparametrically .
moreover , if we had known some prior information between @xmath30 ( or @xmath58 ) and some of the functional covariates , we could easily integrate it by adding a parametric mean part .
for example we may define @xmath59 i.e. including a term in the ggpfr similar to the generalized linear mixed model @xcite ; an example of such models is provided in appendix g. the nonparametric part can still be modeled by @xmath44 via a gpr model .
other nonparametric covariance structure can also be considered ; some examples can be found in @xcite , @xcite and @xcite .
however , most of these methods are limited to small ( usually one ) dimensional @xmath14 or the covariance matrix with a special structure . as an example of the ggpfr model
, we consider a special case of binary data ( e.g. for classification problem with two classes ) . in this case , @xmath60 .
if we use the logit link function , the density function is given by @xmath61 z_m(t ) \ } } { 1 + \exp \{{\mbox{{\boldmath $ { u}$}$_{m}^{t}$ } } { \mbox{\boldmath $ { \beta}$}}(t)+\tau_m(t ) \ } } .
\label{binden}\ ] ] the marginal density function of @xmath30 is therefore given by @xmath62 where @xmath63 is the density function of @xmath44 , which is a multivariate normal distribution for any given points @xmath64 and depends on the functional covariates @xmath5 and the unknown hyper - parameter @xmath54 .
the density functions for other distributions from the exponential families can be obtained similarly .
now suppose that we have @xmath3 batches of data from @xmath3 subjects or experimental units . in the @xmath29-th batch ,
@xmath65 observations are collected at @xmath66 .
we denote @xmath67 , @xmath68 and @xmath69 by @xmath70 , @xmath71 and @xmath72 , respectively , for @xmath73 and @xmath74 .
collectively , we denote @xmath75 and @xmath76 , and denote @xmath77 , @xmath78 , @xmath79 and @xmath80 in the same way .
they are the realizations of @xmath30 , @xmath44 and @xmath81 at @xmath82 . a discrete ggpfr model is therefore given by @xmath83 for @xmath2 , where @xmath84 is a distribution from the exponential family and @xmath85 .
@xmath86 has an @xmath65-variate normal distribution with zero mean and covariance matrix @xmath87 for @xmath88 .
here we assume a fixed dispersion parameter @xmath89 , but the method developed in this paper can be applied to more general cases .
we consider the estimation of @xmath90 first . to estimate the functional coefficient @xmath90 , we expand it by a set of basis functions ( see e.g. * ?
* ) . in this paper
, we use b - spline approximation .
let @xmath91 be the b - spline basis functions , then the functional coefficient @xmath90 can be represented as @xmath92 , where the @xmath93-th column of @xmath94 , @xmath95 , is the b - spline coefficients for @xmath96 .
thus , at the observation point @xmath82 , we have @xmath97 , where @xmath98 is an @xmath99 matrix with the @xmath100-th element @xmath101 . in practice ,
the performance of the b - spline approximation may strongly depend on the choice of the knots locations and the number of basis functions .
there are three widely - used methods for locating the knots : equally spaced method , equally spaced sample quantiles method and model selection based method .
the guidance on which method is to use in different situations can be found in @xcite .
the first method is used in our numerical examples in section 4 which all have equally - spaced time points and the second is adopted in the pbc data in the supplementary materials .
the number of basis functions can be determined by generalized cross - validation or aic ( bic ) methods .
more details on this issue can be found in @xcite .
the covariance matrix @xmath87 of @xmath86 depends on @xmath79 and the unknown hyper - parameter @xmath54 .
if we use covariance kernel ( [ covfun0 ] ) , its element @xmath102 is given by @xmath103 the covariance matrix involves the hyper - parameter @xmath104 , whose value is given based on the prior knowledge in conventional bayesian analysis .
as discussed in @xcite , empirical bayesian learning method is preferable for gpr models when the dimension of @xmath54 is large .
the idea of empirical bayesian learning is to choose the value of the hyper - parameter @xmath54 by maximizing the marginal density function .
thus , @xmath54 as well as the unknown parameter @xmath94 can be estimated at the same time by maximizing the following marginal density @xmath105 or the marginal log - likelihood @xmath106 where @xmath107 is derived from the exponential family as defined in . for binomial distribution
, it is given in .
obviously the integral involved in the above marginal density is analytically intractable unless @xmath107 has a special form such as the density function of normal distribution .
one method to address this problem is to use laplace approximation .
we denote @xmath108 then the log - likelihood can be rewritten as @xmath109 let @xmath110 be the maximiser of @xmath111 , then by laplace approximation we have @xmath112 where @xmath113 is the second order derivative of @xmath114 with respect to @xmath115 and evaluated at @xmath110 ( see , for example , * ? ? ?
* ; * ? ? ?
the procedure of finding the maximiser @xmath110 can be carried out by the newton - raphson iterative method and is given in appendix a of the supplementary materials . however , as pointed out in section 4.1 in @xcite , the error rate of the approximation may be @xmath116 since the dimension of @xmath115 increases with the sample size @xmath65 .
a better method is to approximate @xmath117 in ( here and in the rest of the section the conditioning on @xmath79 is omitted for simplicity ) by @xmath118 where @xmath119 , @xmath120 is the gaussian approximation to the full conditional density @xmath121 , and is the mode of the full conditional density of @xmath86 for a given @xmath122 . here ,
@xmath123 we approximate @xmath124 by taylor expansion to the second order @xmath125 where @xmath126 and @xmath127 depend on the first two derivatives of @xmath128 respectively and evaluated at @xmath129 .
thus , @xmath130 where @xmath131 and @xmath132 .
we can then use the following fisher scoring algorithm @xcite to find the gaussian approximation .
starting with @xmath129 , the @xmath133-th iteration is given by * find the solution @xmath134 from @xmath135 * update @xmath136 and @xmath137 using @xmath134 and repeat ( i ) .
after the process converges , say at @xmath138 , we get the gaussian approximation @xmath120 which is the density function of the normal distribution @xmath139 we can then calculate @xmath140 by maximizing using the approximation .
now we consider two types of prediction problems .
first suppose that we have already observed some data for a subject , say @xmath141 observations in the @xmath133-th batch , and want to obtain prediction at other points .
this can be for one of the batches @xmath142 or a completely new one .
the observations are denoted by @xmath143 which are collected at @xmath144 .
the corresponding input vectors are @xmath145 , and we also know the subject - based covariate @xmath146 .
it is of interest to predict @xmath147 at a new point @xmath148 for the @xmath133-th subject given the test input @xmath149 .
secondly we will assume there are no data observed from the subject of interest except the subject - based covariate and want to predict @xmath147 at a new point @xmath148 with the input @xmath150 .
we use @xmath151 to denote all the training data and assume that the model itself has been trained ( i.e. all the unknown parameters have been estimated ) by the method discussed in the previous section .
the main purpose in this section is to calculate @xmath152 and @xmath153 , which are used as the prediction and the predictive variance of @xmath147 .
we now consider the first type of prediction .
let @xmath154 be the underlying latent variable at @xmath148 , then @xmath155 ( for convenience we ignore the subscript ) and @xmath156 satisfy , and the expectation of @xmath147 conditional on @xmath155 is given by ( [ hmi ] ) : @xmath157 it follows that @xmath158 = \int h({\mbox{{\boldmath $ { u}$}$_{k}^{t}$}}{\hat{{\mbox{\boldmath $ { b}$}}}}^t{\mbox{\boldmath $ { \phi}$}}(t^ * ) + \tau^ * ) p(\tau^*|{\mbox{${\cal d}$ } } ) d\tau^*. \label{pred_mean0}\ ] ] a simple method to calculate the above expectation is to approximate @xmath159 using a gaussian approximation @xmath160 as discussed around equation , that is , @xmath161 since it is assumed that both @xmath162 and @xmath155 come from the same gaussian process with covariance kernel @xmath163 , we have @xmath164}\ ] ] where @xmath165 is the covariance matrix of @xmath162 , and @xmath166 is a vector of the covariances between @xmath162 and @xmath155 .
thus , @xmath167 where @xmath168 and @xmath169 . from the discussion given in the last paragraph in section 2.2
, we have @xmath170 the integrand in is therefore the product of two normal density functions . it is not difficult to prove ( see the details in appendix b of the supplementary materials ) that @xmath159 is still a normal density function @xmath171 then can be evaluated by numerical integration . to calculate @xmath153 , we use the formula : @xmath172 + { \mbox{var}}[{\mbox{e}}(z^*|\tau^*,{\mbox{${\cal d}$ } } ) ] .
\label{var1}\ ] ] from the model definition , we have @xmath173 = { \mbox{e}}\big[{\mbox{e}}(z^*|\tau^*,{\mbox{${\cal d}$}})\big]^2 - \big({\mbox{e}}[{\mbox{e}}(z^*|\tau^*,{\mbox{${\cal d}$}})]\big)^2 \nonumber
\\ = & \int \big [ h({\mbox{{\boldmath $ { u}$}$_{k}^{t}$}}\hat{b}^t { \mbox{\boldmath $ { \phi}$}}(t^ * ) + \tau^*)\big]^2 p(\tau^*|{\mbox{${\cal d}$ } } ) d\tau^ * - \big[{\mbox{e}}(z^*|{\mbox{${\cal d}$}})\big]^2 , \label{var2 } \end{aligned}\ ] ] and @xmath174 = \int { \mbox{var}}(z^*|\tau^*,{\mbox{${\cal d}$ } } ) p(\tau^*|{\mbox{${\cal d}$ } } ) d\tau^ * = \int b''(\hat{\alpha}^ * ) a(\phi ) p(\tau^*|{\mbox{${\cal d}$ } } ) d\tau^ * , \label{var3}\ ] ] where @xmath175 is a function of @xmath176 , and @xmath159 is given by .
thus and can also be evaluated by numerical integration .
the posterior density @xmath159 in is obtained based on the gaussian approximation @xmath160 to @xmath177 .
it usually gives quite accurate results .
the methods to improve gaussian approximation were discussed in @xcite .
they can also be used to calculate @xmath159 from .
an alternative way is to use the first integral in to replace @xmath159 in and perform a multi - dimensional integration using , for example , laplace approximation ; see appendix c of the supplementary materials for the details .
the second type of prediction is to predict a completely new batch with subject - based covariate @xmath178 .
we want to predict @xmath147 at @xmath179 . in this case , the training data @xmath180 are the data collected from the batches @xmath181 . since we have not observed any data for this new batch , we can not directly use the predictive mean and variance discussed above .
a simple way is to predict @xmath147 by using @xmath182 , i.e. ignoring @xmath155 in .
this approach however does not use the information of @xmath150 , the observed functional covariates .
alternatively as argued in @xcite , batches @xmath183 to @xmath184 actually provide an empirical distribution of the set of all possible subjects .
a similar idea is used here .
we assume that , for @xmath2 , @xmath185 if we assume that the new batch or @xmath147 belongs to the @xmath29-th batch , we can calculate the conditional predictive mean by , formulated by @xmath186 the predictive mean @xmath187 in and the predictive variance @xmath188 in can be calculated as if the test data belong to the @xmath29-th batch . here
both @xmath178 and @xmath150 are used .
based on the above empirical assumption , the prediction of the response for the test input @xmath150 at @xmath148 in a completely new subject is @xmath189 and the predictive variance is @xmath190^{2 } - \big[{\mbox{e}}(z^*|{\mbox{${\cal d}$}})\big]^2 .
\label{newpvar}\ ] ] we usually use the equal weights , i.e. @xmath191 for @xmath2 . in general these @xmath3 batches may not provide equal information to the new batch . in this case
varying weights may be considered ; see more discussion in @xcite .
the consistency of gaussian process functional regression method involves two issues .
one is related to the common mean @xmath192 in and the other is related to the curve @xmath193 itself ( @xmath194 or a new one ) .
the common mean structure is estimated from the data collected from all @xmath3 subjects , and has been proved to be consistent in many functional linear models under suitable regularity conditions ( see * ? ? ?
* ; * ? ? ?
this paper focuses on the second issue , the consistency of @xmath195 to @xmath196 , one of the key features in nonparametric regression .
this kind of problems for gpr related models have received increasing attention in recent years , see for example @xcite , @xcite and @xcite .
@xcite considered the posterior consistency of gaussian process prior for normal response , @xcite proved the posterior consistency of gaussian process prior for nonparametric binary regression no matter what the mean function of gaussian process prior is set to , and @xcite extended the result to poisson distribution .
but the consistency for general exponential family distributions is yet to be investigated .
meanwhile , @xcite proved the information consistency via a regret bound on cumulative log loss . generally speaking ,
if the sample size of the data collected from a certain curve is sufficiently large and the covariance function satisfies certain regularity conditions , the prediction based on a gpr model is consistent to the real curve , and the consistency does not depend on the common mean structure or the choice of the values of hyper - parameters involved in covariance function ; see more detailed discussion in @xcite . in this section
, we discuss the information consistency and extend the result of @xcite to a more general context such as @xmath197 following poisson distribution which has not been covered in the literature .
similar to other gpr related models , the consistency of @xmath195 to @xmath196 depends on the observations collected from the @xmath133-th curve only .
we assume that the underlying mean function for the @xmath133-th curve , denoted by @xmath198 , is known .
the case where the mean function is estimated from data is discussed in the supplementary materials .
for ease of presentation we omit the subscript @xmath133 in the rest of the section and denote the data by @xmath199 at the points @xmath200 , and the corresponding covariate values @xmath201 where @xmath202 are independently drawn from a distribution @xmath203 .
let @xmath204 .
we assume that @xmath205 is a set of samples taking values in @xmath206 and follows a distribution in exponential family , @xmath207 for an inverse link function @xmath208 , and the underlying process @xmath209 .
therefore , the stochastic process @xmath210 induces a measure on space @xmath211 .
suppose that the hyper - parameter @xmath212 in the covariance function of @xmath210 is estimated by empirical bayesian method and the estimator is denoted by @xmath213 .
let @xmath214 be the true underlying function , i.e. the true mean of @xmath215 is given by @xmath216 .
denote @xmath217 then @xmath218 is the bayesian predictive distribution of @xmath205 based on the gpr model .
note that @xmath219 depends on @xmath220 since the hyper - parameter of @xmath210 is estimated from the data .
it is said that @xmath218 achieves _ information consistency _
if @xmath221 \big ) \rightarrow 0 \quad\text { as } n\rightarrow \infty , \label{inf : con}\ ] ] where @xmath222 denotes the expectation under the distribution of @xmath223 and @xmath224 $ ] is the kullback - leibler divergence between @xmath225 and @xmath226 , i.e. @xmath227=\int p_0(z)\log \frac{p_0(z)}{p_{gp}(z)}dz.\ ] ] * theorem 1 : * under the ggpfr models and and the conditions given in lemma 1 of the supplementary materials , the prediction @xmath228 is information consistent to the true curve @xmath229 if the rkhs norm @xmath230 is bounded and the expected regret term @xmath231 .
the error bound is specified in . * remark 1 . *
the condition @xmath232 in lemma 1 can be satisfied by a wide range of distributions , such as normal distribution where @xmath233 , binomial distribution ( with the number of trials @xmath29 ) where @xmath234 and poisson distribution where @xmath235 . * remark 2 . *
the regret term @xmath236 depends on the covariance function @xmath12 and the covariate distribution @xmath203 .
it can be shown that for some widely used covariance functions , such as linear , squared exponential and matrn class , the expected regret terms are of order @xmath237 ; see @xcite for the detailed discussion . * remark 3 .
* lemma 1 requires that the estimator of the hyperparameter @xmath212 is consistent . in appendix d of the supplementary materials we proved that the estimator by maximizing the marginal likelihood based on laplace approximation satisfies this condition when the number of curves and the number of observations on each curve are sufficiently large
this implies that the information consistency in ggpfr models is achieved for the covariance functions listed in remark 2 .
a more general asymptotic analysis is to study the convergence rates of both the mean function estimation and the individual curves prediction when the number of curves and/or the number of observations on each curve tend to infinity , as discussed in @xcite for the maximum likelihood estimators of the parameters in mixed - effects models .
the research along this direction is worth further development .
in this section we demonstrate the proposed method with serveral examples .
we first use simulated data and then consider the paraplegia data discussed in section 1 .
more simulated and real examples are provided in the supplementary materials . *
( i ) simulation study . * the true model used to generate the latent process is @xmath238 , where , for each @xmath29
, @xmath239 s are equally spaced points in @xmath240 and @xmath241 is a gaussian process with zero mean and the squared exponential covariance function defined in with @xmath242 , @xmath243 and @xmath244 . in this example , the covariate @xmath245 is the same as @xmath246 .
the observations @xmath70 follow a binomial distribution @xmath247 with @xmath248 sixty curves , each containing @xmath65 data points , are generated and used as training data .
we use a ggpfr model with binomial distribution and logit link function : @xmath249 where @xmath44 follows a gpr model .
cubic b - spline approximation is used to estimate the mean curve @xmath250 , where the knots are placed at equally spaced points in the range and the number of basis functions is determined by bic which is given by bic@xmath251 with @xmath252 being the total number of parameters . a gaussian approximation method as specified around is used to calculate the empirical bayesian estimates of @xmath94 and @xmath54 .
table_parameter ] lists the average estimates of the hyper - parameters @xmath253 for @xmath254 , 40 and 60 for ten replications .
the empirical bayesian estimates are closer to the true values with @xmath65 increasing .
the estimates of mean curve @xmath255 for different @xmath65 along with the true mean curves are presented in the left panels of figure [ simu_mean_curves ] . as discussed in section 3 ,
the consistency of @xmath255 to @xmath250 depends on the observations obtained from all training curves .
the figures show that the estimated mean curves by the ggpfr method are very close to the true one even for the case of @xmath254 .
[ [ para_fig11]interpolation].,title="fig : " ] [ [ para_fig12]interpolation].,title="fig : " ] + [ [ para_fig13]extrapolation].,title="fig : " ] [ [ para_fig14]extrapolation].,title="fig : " ] + [ [ figbetat]].,title="fig : " ] | in this paper we propose a generalized gaussian process concurrent regression model for functional data where the functional response variable has a binomial , poisson or other non - gaussian distribution from an exponential family while the covariates are mixed functional and scalar variables .
the proposed model offers a nonparametric generalized concurrent regression method for functional data with multi - dimensional covariates , and provides a natural framework on modeling common mean structure and covariance structure simultaneously for repeatedly observed functional data .
the mean structure provides an overall information about the observations , while the covariance structure can be used to catch up the characteristic of each individual batch .
the prior specification of covariance kernel enables us to accommodate a wide class of nonlinear models .
the definition of the model , the inference and the implementation as well as its asymptotic properties are discussed .
several numerical examples with different non - gaussian response variables are presented . some technical details and more numerical examples as well as an extension of the model
are provided as supplementary materials .
key words : covariance kernel , exponential family , concurrent regression models , nonparametric regression . * author s footnote * b. wang is lecturer in statistics , department of mathematics , university of leicester , leicester le1 7rh , uk ( e - mail : ` [email protected] ` ) .
j. q. shi is reader in statistics , school of mathematics and statistics , newcastle university , newcastle ne1 7ru , uk ( e - mail : ` [email protected] ` ) .
the authors thank the associate editor and the reviewers for their constructive suggestions and helpful comments . |
using the representation for @xmath214 in eq .
[ qcpropogator ] , eilenberger s equation can be expressed as coupled equations for the quasiparticle and pair propagators in a three - dimensional vector space , [ vector_eilenberger ] _ = , with _ 1 _ 2 _ 3 , = 0 & ^r & _ 2 - ^r & 0 & -_1 _ 2 & -_1 & 0 for a uniform order parameter defined by trajectory @xmath9 we express @xmath215 in terms of the eigenvectors of @xmath216 , @xmath217 .
the eigenvector with @xmath218 , [ egienvector0 ] = -_1 ( ) -_2 ( ) + , generates the bulk equilibrium propagator , _ 0 = -(- ( ) ) , where @xmath219 and @xmath220 .
this solution satisfies eilenberger s normalization condition in eq .
[ normalization_condition ] .
there is also a pair of eigenvectors with eigenvalues @xmath221 [ egienvector+- ] = -_1_2 _ 1 ^ 2 _
2_1 , with @xmath222 .
these eigenvectors generate `` exploding solutions '' to eq .
[ vector_eilenberger ] for energies within the gap of the bulk quasiparticle spectrum , @xmath223 , and thus are physical solutions only in the vicinity of a boundary , or near a localized defect such as a vortex or domain wall.@xcite for the same value of momentum , @xmath9 , the eigenvectors are orthonormal , @xmath224 .
are obtained from the adjoint of @xmath225 and the replacement @xmath226 since @xmath227 . ]
the nambu propagators corresponding to the eigenvectors @xmath228 are _ ( , ) = & ( & ( _ 2_1 ) + & & i_x(_1_2 ) + & + & i_x_1 ^ 2 ) . these matrices are non - normalizable and anti - commute with the bulk propagator , [ nambu_algebra ] ( _ ) ^2 = 0 , _ + = 0 . for a boundary
far from other boundaries or defects we must exclude solutions that explode into the bulk of the superfluid .
in particular , for a pair of specular or retro - reflected trajectories the solutions for the incident and reflected trajectories are & = & + c_()e^-2(,)x / v_x , + & = & + c_()e^-2(,)x / v_x , where @xmath83 for @xmath84 and @xmath85 is the coordinate normal to the boundary as shown in fig .
[ fig - time - reversed_trajectories ] .
the corresponding nambu propagator for the incident trajectory in the vicinity of the boundary is constructed from these solutions with eqs .
[ normalization_condition ] and [ nambu_algebra ] to fix the normalization , [ nambu_incident ] _ = - ( _ 0 ( , ) + c_(,)_+(,)e^-2()x / v_x ) . for an incident trajectory @xmath229 ,
the specularly reflected trajectory is @xmath230 .
thus , the eigenvectors for the specularly reflected trajectory @xmath231 are obtained from eqs .
[ egienvector0 ] and [ egienvector+- ] by the replacement , @xmath232 .
the specular boundary condition requires continuity of the incoming and outgoing propagators at @xmath108 , which fixes the amplitudes , @xmath233 and @xmath234 . for the incident trajectory , [ c_amplitude_specular ] c^ _ ( ) = = .
similarly for the specularly reflected trajectory : @xmath235 .
the resulting propagator from eqs .
[ nambu_incident ] and [ c_amplitude_specular ] gives the results for the pair propagators , @xmath236 , and quasiparticle propagator @xmath237 in eqs .
[ fr1]-[gr3 ] .
for retro - reflection we have @xmath238 , and in this case the eigenvectors are obtained from eqs . [ egienvector0 ] and [ egienvector+- ] by the replacements , @xmath232 and @xmath239 .
this boundary condition dramatically alters the propagator near the boundary with c^ _ ( ) = , which gives the propagator for retro - reflection in eq .
[ gr_retro - reflection ] , with a spectrum of zero - modes replacing the branch of chiral edge states for specular reflection .
the second integral in eq . [ lz_c2 ] , derived from the branch cut in fig .
[ fig - complex_plane ] , evaluates to _ 0^ = ^-1 ( ) . setting @xmath240 , @xmath241 reduces eq .
[ lz_c2 ] to l_z^ _
= n & & _ 0 ^ 1dtt^-1 ( ) .
integration by parts reduces to a beta function,@xcite _ 0 ^ 1dtt^-1 ( ) & = & ( , ) = , which yields the mt result , @xmath3 , given in eq .
[ lz_mt - result ] . | the spectra of fermionic excitations , pairing correlations and edge currents confined near the boundary of a chiral p - wave superfluid are calculated to leading order in @xmath0 .
results for the energy- and momentum - resolved spectral functions , including the spectral current density , of a chiral p - wave superfluid near a confining boundary are reported .
the spectral functions reveal the subtle role of the chiral edge states in relation to the edge current and the angular momentum of a chiral p - wave superfluid , including the rapid suppression of @xmath1 for @xmath2 in the fully gapped 2d chiral superfluid .
the edge current and ground - state angular momentum are shown to be sensitive to boundary conditions , and as a consequence the topology and geometry of the confining boundaries . for perfect specular boundaries the edge current accounts for the ground - state angular momentum , @xmath3 , of a cylindrical disk of chiral superfluid with @xmath4 fermion pairs .
non - specular scattering can dramatically suppress the edge current . in the limit of perfect retro - reflection
the edge states form a flat band of zero modes that are non - chiral and generate no edge current . for a chiral superfluid film confined in a cylindrical toroidal geometry
the ground - state angular momentum is , in general , non - extensive , and can have a value ranging from @xmath5 to @xmath6 depending on the ratio of the inner and outer radii and the degree of back scattering on the inner and outer surfaces .
non - extensive scaling of @xmath7 , and the reversal of the ground - state angular momentum for a toroidal geometry , would provide a signature of broken time - reversal symmetry of the ground state of superfluid , as well as direct observation of chiral edge currents . among the remarkable phases of liquid
is the a - phase .
in addition to being a superfluid which supports persistent currents , this fluid is believed to possess a spontaneous mass current in its ground state .
ground state currents are associated with the chirality of cooper pairs that condense to form the a - phase and conspire to produce a macroscopic angular momentum .
chirality is encoded in the p - wave orbital order parameter , @xmath8 , where @xmath9 is the relative momentum of a cooper pair , @xmath10 is an orthonormal triad of unit vectors that define the orbital coordinates of the cooper pair wave function , and @xmath11 is the pairing energy.@xcite this order parameter is an eigenfunction of the orbital angular momentum along @xmath12 with eigenvalue @xmath13 .
such broken symmetries in bulk condensed matter systems have implications for the spectrum of excitations bound to surfaces and topological defects.@xcite this phase breaks time - reversal symmetry as well as parity , and is realized at all pressures below melting in thin superfluid @xmath14he - a films . in the 2d limit the fermi surface is fully gapped , and belongs to the topological class of integer quantum hall systems.@xcite the 2d a - phase is also representative of layered p - wave superconductors with broken time - reversal symmetry , e.g. the proposed order parameter for superconducting .@xcite the macroscopic manifestation of chiral order in is the ground - state angular momentum , @xmath15 , where @xmath16 is the mass current density . for 2d chiral p - wave superfluids in the bcs limit ,
where the size of cooper pairs is large compared to fermi wavelength , @xmath17 , the ground state currents are predominantly confined to boundaries .
i discuss effects of surface scattering on the pairing correlations , the fermionic spectrum and ground state currents in the vicinity of boundaries confining a chiral p - wave superfluid .
results for the spectral current density highlight the fermionic spectrum that is responsible for the edge current and the ground state angular momentum .
the theory is extended to finite temperatures , non - specular boundaries and multiply connected geometries .
the results reported here are discussed in context with the results of kita,@xcite and stone and roy.@xcite starting from bogoliubov s equations in sec .
[ bogoliubov - andreev - eilenberger ] , i introduce eileberger s quasiclassical equation for the nambu propagator that is the basis for investigating the pairing correlations , spectrum of surface states and edge currents for chiral p - wave superfluids . the bound - state spectrum and results for the spectral current density are discussed in secs .
[ sec : chiral_edge_state ] and [ sec : spectral_current_density ] .
analysis of the continuum spectrum , edge current and the spectral analysis of the ground state angular momentum are reported in sec .
[ sec : edge_current ] , which is followed by results and a discussion of the temperature dependence of the edge current and angular momentum in sec .
[ sec : lz_vs_t ] . in sec .
[ sec : boundary_conditions ] i discuss the sensitivity of the edge current and ground - state angular momentum to boundary scattering and geometry , and in sec .
[ sec : toroidal_geometry ] discuss the non - extensive behavior of the ground - state angular momentum that develops for multiply connected geometries in which there is an asymmetry in the specularity on different surfaces .
i start with some background on the ground state current and angular momentum of superfluid .
the magnitude of the ground - state angular momentum , @xmath7 , has been the subject of considerable theoretical investigation .
predictions for @xmath7 of in a cylindrically symmetric vessel vary over many orders of magnitude,@xcite from @xmath18 to @xmath3 , where @xmath4 is the total number of fermion pairs in the volume @xmath19 .
this latter result is what one intuitively expects for bose - einstein condensate ( bec ) of tightly bound molecules , each carrying one unit of angular momentum , and whose molecular size @xmath20 , is small compared to the mean distance between molecules , @xmath21{\mbox{\small $ 2v / n$}}\gg\xi$ ] .
however , this result is also obtained in the opposite limit , @xmath22 , appropriate to bcs condensation of cooper pairs , each with angular momentum @xmath23 and radial size @xmath24 , where one expects almost exact cancellation of the internal currents from overlapping cooper pairs.@xcite in particular , mcclure and takagi @xcite showed that an n - particle ground state of the form , [ bcs_wave - function ] = ^n/2 , with an equal - spin , odd - parity chiral pairing amplitude , _ & & _ ( + /2)_(-/2 ) + & = & ( || ) ( i_y ) _ ( ( ) + i ( ) ) , of the am form that preserves cylindrical symmetry is an eigenstate of the total angular momentum with @xmath25 .
varies slowly on the scale of @xmath20 , such as the anderson - brinkman @xcite and mermin - ho @xcite textures for in a long cylinder . ]
thus , the ground - state angular momentum of a chiral condensate is the same for @xmath4 bose molecules or @xmath4 cooper pairs .
however , the magnitude and spatial distribution of the mass currents that give rise to the total angular momentum differ in the bec and bcs limits .
this somewhat non - intuitive result is intimately connected with the symmetry of the ground state and its implications for the surface fermionic spectrum and associated currents.@xcite numerous authors have addressed the question of the current distribution responsible for the ground state angular momentum.@xcite@xmath26 starting from the n - particle bcs wavefunction in eq .
[ bcs_wave - function ] , ishikawa@xcite and mermin and muzikar@xcite calculated the current density in the long wavelength limit , @xmath27 , for the am state at @xmath28 .
for spatially uniform @xmath29 and no center of mass supercurrent , [ hydro_bound_current ] = . in the bcs limit the density , @xmath30 , is spatially uniform _ except _ near the boundary , @xmath31 , where @xmath32 .
the current is then confined at the boundary , @xmath33 , from which one recovers the result for the ground - state angular momentum @xmath34 .
this highlights a limitation of the gradient expansion and hydrodynamic limit .
the order parameter is assumed to be the local equilibrium the am state , and spatial variations are assumed to be long wavelength on the scale of @xmath22 .
however , the density varies on atomic length scales near the surface , whereas the order parameter is , in general , strongly deformed on on length scales of order @xmath20 near a boundary .
thus , eq . [ hydro_bound_current ] , and the gradient expansion in particular , do not accurately describe the current density near the boundary , nor do they account for the source of the surface current .
this requires a theory valid for spatial variations of the condensate on length scales comparable to or smaller than the correlation length @xmath20 .
superfluid ( `` 2d '' ) confined in a cylindrical geometry with thickness @xmath35 , radius @xmath36 bounded by specular surfaces which reflect excitations , @xmath37 .
[ fig - cylindrical_film_geometry ] ] for a thin film of , as shown in fig .
[ fig - cylindrical_film_geometry ] , the orbital quantization axis is locked normal to the surface of the film , @xmath38.@xcite the a - phase also belongs to the class of equal spin pairing ( esp ) states with spin structure of the order parameter given by a linear combination of the symmetric pauli matrices , [ esp_a - phase ] _ ( ) = ( i_y ) _ ( ) , where @xmath39 label the projections of fermion spins of the cooper pair and @xmath40 is the direction in spin space along which cooper pairs have zero spin projection .
thus , for @xmath41 the spin state of the cooper pairs is given by @xmath42 , which is the triplet state with equal amplitude for the cooper pairs to be spin polarized along @xmath43 or @xmath44 : @xmath45 .
spin textures described by spatial variations of the @xmath40 vector are possible ; however , in what follows i assume the spin state is fixed by the nuclear dipolar energy which locks the @xmath46.@xcite the bulk a - phase of in 3d has gapless excitations for momenta along the nodal directions , @xmath47 . here
i consider 2d with a cylindrical fermi surface , ( or a set of cylindrical fermi surfaces generated by dimensional quantization ) , and an orbital order parameter given by [ am_order_parameter ] ( ) = ( p_x + ip_y)/p_f , which generates a bulk excitation spectrum that is fully gapped on the fermi surface . near a boundary , or domain wall ,
the orbital order parameter can deviate from the pure a - phase form .
thus , a more general form of the orbital p - wave order parameter is parametrized by two real amplitudes , ( , ) = ( _ 1()p_x + i_2()p_y)/p_f , with @xmath48 far from a boundary .
inhomogeneous states are described by the bogoliubov s equations@xcite ( -^2 - ) u _ ( ) + _ ( , ) v _ ( ) = u _ ( ) , + -(-^2 - ) v _ ( ) + ^_(,)u _ ( ) = v _ ( ) . for the particle ( @xmath49 ) and hole ( @xmath50 ) wavefunctions . for @xmath51
the bogoliubov equations reduce to @xmath52 equations for bogoliubov spinors , @xmath53 , in nambu ( particle - hole ) space , _
= , where @xmath54 is the bogoliubov hamiltonian expressed in terms of nambu matrices , @xmath55 , _
= ( ) + _ 1(,)+ _ 2 ( , ) , with @xmath56 , and the off - diagonal pair potentials interpreted as symmetrized operators , _ 1,2 ( , ) = ( _ 1,2()_x , y + _ x , y _ 1,2 ( ) ) .
the large difference between the fermi wavelength , @xmath57 , and the size of cooper pairs , @xmath20 , is the basis for andreev s quasiclassical approximation to the bogoliubov equations.@xcite the expansion is achieved by factoring the fast- and slow spatial variations of the bogoliubov spinor , = e^i_f/ , and retaining leading order terms in @xmath58 , which yields andreev s equation , [ andreev_equation ] _
+ i _
= 0 . with operator @xmath59 defined by _
= - ( , ) , and the nambu matrix order parameter given is by ( , ) = i_x(_2 ( , ) + _ 1 ( , ) ) , where @xmath60 is the fermi momentum , @xmath61 is the fermi velocity .
the latter defines classical straight - line trajectories for the propagation of wavepackets of bogoliubov excitations , which are coherent superpositions of particles and holes with amplitudes given by the andreev - nambu spinor , [ andreev - nambu_spinor ] = u _ _ .
andreev s equation expressed in terms of a row spinor is since @xmath62 is not hermitian . ] _ - i _
= 0 , with the normalization of the andreev - nambu spinor given by @xmath63 . there are two solutions ( branches ) to andreev s equation for a single trajectory defined by @xmath9 . for @xmath64 , the two branches are propagating solutions ; a particle - like solution , @xmath65 , with group velocity @xmath66 and hole - like solution , @xmath67 , with reversed group velocity , @xmath68 . for energies within the bulk gap the solutions are exploding and decaying amplitudes along the trajectory , and thus relevant only in the vicinity of boundaries , domain walls , etc . the product of the particle- and hole amplitudes in eq .
[ andreev - nambu_spinor ] , _ ( , ; ) = u_(,;)v _ ( , ; ) , is the _ pair propagator _ , which determines the spectral composition of the cooper pair amplitude , _
( , ) = df ( ) _ ( , ; ) , where @xmath69 is the fermi distribution .
the pair propagator is one component of the nambu matrix ( , ; ) = _ , g _ , which satisfies eilenberger s transport equation,@xcite [ eilenberger_equation ] + i _ ( , ; ) = 0 . physical solutions to eq .
[ eilenberger_equation ] must also satisfy eilenberger s
normalization condition,@xcite [ normalization_condition ] ( ( , ; ) ) ^2 = -^2 .
an advantage of eilenberger s formulation is that the spectral functions for both quasiparticle and pair excitations are obtained as components of the quasiclassical propagator . for a fixed spin quantization axis , @xmath70
, the off - diagonal components of the propagator describe pure equal - spin pairing correlations . as a result
the nambu propagator can be expressed in the form , [ qcpropogator ] = _ 3 + i_x(_2- _ 1 ) .
the superscript refers to the causal ( retarded in time ) propagator , obtained from eq .
[ eilenberger_equation ] with the shift , @xmath71 .
the diagonal propagator in nambu space , @xmath72 , determines the spectral function , or local density of states , for the fermionic excitations with momentum @xmath60 , ( , ; ) = -_3 ( , ; ) , while the off - diagonal propagators , @xmath73 and @xmath74 , determine the spectral function for the correlated pairs , _ 1,2 ( , ; ) = -_1,2 ( , ; ) .
these functions determine the mean pair potentials , @xmath75 and @xmath76 , through the bcs self - consistency condition , _ 1,2 ( , ) = v(, ) ^+__-_d ( ) _ 1,2(, ; ) _ , where @xmath77 is the bandwidth of attraction for the spin - triplet , p - wave pairing interaction , @xmath78 , which is integrated over the occupied states defining the pair spectrum and averaged over the fermi surface , @xmath79 . for a boundary
far from other boundaries only single reflections , @xmath80 , couple the propagators for the incoming ( @xmath9 ) and outgoing ( @xmath81 ) trajectories . in particular , for the pair of specularly reflected trajectories on the boundary shown in fig .
[ fig - cylindrical_film_geometry ] , with radius of curvature large compared to the correlation length , @xmath82 , the solutions for the components of the propagator are ( see appendix [ appendix - quasiclassical ] ) _ 1(x , ; ) & = & ( 1 - e^-2()x / v_x ) , [ fr1 ] + _ 2(x , ; ) & = & - e^-2()x / v_x , [ fr2 ] + _ 3(x , ; ) & = & - + e^-2()x / v_x .
[ gr3 ] where @xmath83 for @xmath84 and @xmath85 is the coordinate normal to the boundary as shown in fig .
[ fig - cylindrical_film_geometry ] .
note that the propagator corresponding to cooper pairs with relative momentum normal to the boundary vanishes at the boundary , @xmath86 .
de - pairing of the normal amplitude is partially compensated by an increase in the pairing correlations for pairs with relative momenta parallel to the boundary .
the origin of this enhancement is the fermionic state , bound to the surface , which appears as a pole in the propagators of eqs .
[ fr2 ] and [ gr3 ] at the energies , [ chiral_bound - state ] _
( ) = -_2 ( ) = - cp_|| .
the surface state disperses with momentum @xmath87 parallel to the surface , @xmath88 , and @xmath89 .
, illustrating the asymmetry in the occupation of pairs of time - reversed states .
[ fig - chiral_edge - state_dispersion ] ] the important feature of the spectrum of surface fermions , shown in fig .
[ fig - chiral_edge - state_dispersion ] , is that there is _ no _ branch with the opposite phase velocity .
the spectrum describes weyl or chiral fermions.@xcite@xmath26 . ] for each pair of time - reversed fermions the state with @xmath90 is _ occupied _ while its time - reversed partner with momentum @xmath91 is _ empty_. as a result the pairs of surface states generate a net mass or charge current .
this asymmetry in the occupation of the surface spectrum is a reflection of the chirality of the ground state order parameter and specular reflection at the boundary which preserves translation symmetry locally along the boundary .
the absence of a branch of fermions with energy @xmath92 is demonstrated by evaluating the residue of @xmath93 at the apparent pole ,
@xmath94 : @xmath95 . for energies in the vicinity of the bound - state pole , @xmath96
, the quasiparticle propagator reduces to [ gr_bound - state ] _ ( x , ; ) = e^-2x / v_f , where i include the line - width , @xmath97 , of the surface state due to weak disorder . for @xmath98
the states are sharp and the spectral function consists of delta functions at @xmath99 , [ dos_bound - state ] _ ( x , ; ) = |_1()|e^-2x / v_f ( - _ ( ) ) .
the spectral weight is maximum for trajectories at normal incidence and vanishes for grazing incidence .
note that every edge state is confined to the surface on the length scale _
= v_f/2 , independent of momentum @xmath100 , and of order the cooper pair size , @xmath101 .
the spectral current density is defined as the local density of current carrying states in the energy interval @xmath102 for states with momentum @xmath9 , [ jdos ] ( x , ; ) = 2 n_f _ , where @xmath103 is the spectral function calculated for the incident trajectory with momentum @xmath9 , @xmath104 is the normal - state density of states at the fermi level for one spin , and @xmath9 and @xmath105 define the pair of time - reversed incident trajectories shown in fig .
[ fig - time - reversed_trajectories]a , for which @xmath106 . , for specular reflections .
b ) retro - reflections are time - reversed partners for any incident angle .
the chirality of the bulk order parameter is indicated by the direction of the arc .
[ fig - time - reversed_trajectories ] ] the resulting local current density is obtained by thermally occupying the spectrum and integrating over all incoming trajectories , [ particle_current_density ] ( x ) = _ _ -^+df ( ) ( x , ; ) , where @xmath107 is the fermi distribution . for @xmath108 as a function of @xmath87 for linewidth @xmath109 .
states with @xmath110 are slightly offset to show the contributions to the current from time - reversed pairs .
[ fig - jydos_bound - state ] ] the spectral current density for the bound - state spectrum obtained from eqs .
[ gr_bound - state ] and [ jdos ] is shown in fig .
[ fig - jydos_bound - state ] for the full range of incident trajectories .
note that time - reversed states , incident angles @xmath111 and @xmath112 , add coherently to the spectral current density .
thus , the net current density parallel to the boundary carried by the surface bound states is given by , vanishes identically , as required by particle conservation . ] j_y^(x ) & = & i(/t ) n_f v_f e^-x/ _ , [ bound - state_current - density ] where the integration over the spectrum reduces to i = _ -1^+1duu(u/2 t ) = \ { ll 1 & , t0 & , tt_c .
note that near the transition the magnitude of the current decreases as @xmath113 , but also penetrates deeper into the bulk as @xmath114 .
mass currents confined near the boundary ( `` edge currents '' ) generate macroscopic angular momentum . for a galilean invariant system such as liquid the mass current density
is obtained from the spectral current density in eq .
[ jdos ] by the replacement @xmath115 , where @xmath116 is the quasiparticle effective mass , @xmath117 is the fermi momentum . in addition ,
@xmath118 , @xmath119 and the normal - state density of states , @xmath104 , determine the particle number density , which for a 2d fermi surface gives @xmath120 . for a chiral p - wave superfluid confined within a thin cylindrical vessel of radius @xmath121 and height @xmath122 in the 2d limit , @xmath123 , the angular momentum relative to the @xmath124-axis is determined by the radial moment of the azimuthal component of the mass current density , @xmath125 , [ angular - momentum_cylinder ] l_z = _ d^3r . for @xmath126 we can neglect the curvature of the surface , in which case the azimuthal mass current
is given by the tangential component of the boundary current calculated from eq .
[ particle_current_density ] .
thus , the bound - state contribution to @xmath7 at @xmath28 obtained from eqs .
[ angular - momentum_cylinder ] and [ bound - state_current - density ] , with @xmath127 becomes , [ lz_bound - states ] l_z^ & = & n_fp_f2h_0^r r^2dre^-(r - r)/ _
= n , which is a factor of _ two _ larger than that predicted by ishikawa @xcite and mcclure and takagi@xcite based on the real - space n - particle wave function of eq . [ bcs_wave - function ] . finite size corrections from eq .
[ lz_bound - states ] are negligible - of order @xmath128 . as stone and
roy pointed out the discrepancy is resolved by including the contribution to @xmath7 from the states comprising the continuum spectrum.@xcite below i analyze the continuum contributions to the edge current and ground state angular momentum . in particular ,
i show that there are two contributions to the continuum spectral current density : ( i ) an isolated scattering resonance that exactly cancels the bound - state contribution to the edge current for each value of @xmath9 and ( ii ) a non - resonant response of the bound continuum that accounts exactly for the mt result of @xmath25 .
the energy range @xmath129 constitutes the bound continuum , while the range @xmath130 represents excitations above the gap . at finite temperatures
sub - gap surface excited states @xmath131 also play an important role . the spectral weight associated with the continuum spectrum
is modified near the boundary . for @xmath132 , @xmath133 and
the spectral function becomes , [ dos_continuum ] _ ( x , ; ) & = & + & - & ( 2x / v_x ) + & - & ( ) ( ) ( 2x / v_x ) . the first term is the bulk continuum spectrum , while the corrections to the continuum spectrum are given by second and third lines in eq .
[ dos_continuum ] .
the third term is _ odd _ under either @xmath134 or @xmath135 , and thus gives a non - vanishing contribution to the spectral current density , _ ( x , ; ) & = & -2n_f _ ( ) ( ) + & & ( 2x / v_x ) note that for fixed energy @xmath136 and momentum @xmath9 the effect of surface scattering on the continuum spectrum is large and _ propagates _ into the bulk . thus , it is not a priori clear that the current is confined to the surface .
however , the wavelength of the disturbance is given by the tomasch wavelength for a specific trajectory , _ ( , ) = .
the net current parallel to the boundary is given by the sum over all incident trajectories , [ jy - continuuum ] j_y^(x ) = 2n_f v_f _ -/2^+/2_y_1()_2 ( ) j ( ) , where @xmath137 is given by [ jy - continuuum - kernel ] j ( ) = _ ^ d ( 2x / v_x ) .
the integration over the spectral current density leads to phase cancellation away from the boundary , and a _ net _ current that is confined to the edge .
although the integration is over the continuum spectrum , the chiral edge state nevertheless modifies the current carried by the continuum states .
trajectories near grazing incidence give a large enhancement to the kernel @xmath137 coming from the off - resonant bound state .
the kernel is weighted by the product @xmath138 , which is peaked near @xmath139 . at zero temperature
the kernel is evaluated by transforming to an integration over the radial momentum @xmath140 , or equivalently @xmath141 with @xmath142 , j ( ) = _ _ d e^2ix / v_x .
the singularities shown in the upper half of the complex @xmath20-plane determine the continuum current response .
in particular , the integral along the real axis is transformed to an integral around the pole at @xmath143 and the branch cut from @xmath144 to @xmath145 : @xmath146
. plane .
integration along the real axis ( @xmath147 ) is transformed into the sum of an integral around the isolated pole at @xmath143 and the branch cut along the imaginary @xmath20-axis .
[ fig - complex_plane ] ] the pole at @xmath148 is an isolated resonance that gives a contribution to the continuum current that is confined to the boundary on the length scale @xmath149 , j _ _
= e^-x/ _ .
the current generated by this resonance exactly cancels the bound - state edge current and bound - state contribution to the angular momentum , l_z^ _ = _ d^2r = - n. thus , the ground - state current and angular momentum come entirely from the non - resonant contribution to the continuum spectrum defined by the branch cut @xmath150 which evaluates to j _ _
= -_0^ e^-2x / v_x . the current density
is then given by j_y^_(x ) & = & 2n_f v_f _ -/2^+/2_y|_1()|_2 ( ) + & & _ 0^ e^-2x / v_x , [ chiral_edge_current ] which is confined to the edge , but in contrast to the bound - state and resonance terms , there is not a single confinement length , but rather a weighted average of exponential confinement on length scales @xmath151 .
for this reason an analytic expression for the net current density analogous to eq .
[ bound - state_current - density ] does not appear possible .
however , the total edge current and ground state angular momentum can be computed by first carrying out the integration over the region of the edge current . in the limit
@xmath126 the resulting ground - state angular momentum reduces to the following integration over the continuum spectrum , [ lz_c2 ] l_z^ _ = n & & _ -/2^+/2dp_xp_y_1()_2 ( ) + & & _ 0^ , which evaluates to ( see appendix [ appendix - angular_momentum_integration ] ) [ lz_mt - result ] l_z^ _ = .
thus , ishikawa@xcite , mcclure and takagi@xcite and stone and roys@xcite results are recovered from the continuum response to the formation of the chiral edge state . for @xmath152 thermal excitations out of the ground state lead to a reduction of the order parameter @xmath153 , the edge current and angular momentum .
the latter can be expressed as l_z(t ) = _ ( t ) , where @xmath154 for @xmath155 , vanishes for @xmath156 , and can be calculated from the edge current at finite temperature . calculations of the temperature dependence of the angular momentum for were carried out by t. kita on the basis of numerical solutions to the bogoliubov equations for mesoscopic cylindrical ( 3d ) geometries with dimensions @xmath157 .
kita showed that @xmath158 decreases rapidly for @xmath159 , indicating that there are low - lying excitations that are thermally populated even at low temperatures which reduce the ground - state angular momentum . based on his numerical results ( fig .
2a of ref . ) , kita conjectured that the temperature dependence of @xmath158 resulted from the excitations responsible for the suppression of the superfluid density @xmath160 of bulk corresponding to superflow along the _ nodal direction _ for the 3d chiral p - wave superfluid .
for 3d bulk superfluid the stiffness for @xmath161 is strongly suppressed at finite temperature compared to the stiffness for superflow perpendicular to the nodal direction , i.e. @xmath162 . however , as i discuss below , the softness of the angular momentum response function @xmath158 that kita found numerically , including its near equality with @xmath160 for the 3d a - phase , is also present in the 2d limit in which the chiral p - wave superfluid is fully gapped . for @xmath152 the edge current
is determined by the continuum contribution to the spectrum defined in eq .
[ jy - continuuum ] with , j ( ) = _ _ d e^2ix / v_x .
as is the case for @xmath28 , the resonant contribution to the continuum current density coming from the isolated pole at @xmath163 exactly cancels the bound state contribution .
however , the total edge current and angular momentum , which at @xmath28 is calculated from the branch cut in fig .
[ fig - complex_plane ] , now results from the sum of contributions from a discrete set of poles at the complex momenta defined by _
n = i , where @xmath164 , @xmath165 , are the fermion matsubara frequencies .. ] the resulting edge current density is given by [ continuum - current - finite_t ] j_y^_(x ) & = & 2n_f v_f _ -/2^+/2p_y|_1()|_2 ( ) + & & t__n e^-2x / v_x .
multiple confinement scales are manifest in eq .
[ continuum - current - finite_t ] .
the total surface current obtained by integrating over the boundary region determines the equilibrium angular momentum generated by these edge currents , [ lz_vs_t ] _
( t ) = _ 0 ^ 1dx(1-x^2)^ t__n , where @xmath166 .
figure [ fig - lz_vs_t ] shows the temperature dependence of the equilibrium angular momentum , @xmath158 , calculated from eq .
[ lz_vs_t ] . also shown for comparison is the bulk excitation gap and superfluid stiffness for both 2d and 3d chiral p - wave states .
note that the temperature dependence of the angular momentum is much softer than the bulk superfluid stiffness for the gapless 2d phase .
, for the 2d chiral p - wave superfluid ( ) . also shown is the superfluid stiffness ( ) and the bulk gap ( ) for the fully gapped 2d chiral p - wave state . shown for comparison are the two components of the superfluid stiffness for the 3d chiral , p - wave superfluid phase ( ) - @xmath167 for @xmath168 ( ) and @xmath169 for @xmath170 ( ) .
[ fig - lz_vs_t ] ] just as kita found for his 3d mesoscopic geometry , the temperature dependence of @xmath1 for the fully gapped 2d phase is nearly identical to the superfluid stiffness for superflow parallel to the nodal direction for the 3d phase .
however , the reasons for the rapid suppression of @xmath1 and @xmath160 are of different physical origin . for the bulk 3d phase @xmath160
is strongly reduced compared to @xmath171 due to the backflow current carried by the nodal excitations when @xmath161.@xcite by contrast , for the fully gapped 2d chiral phase there are low - energy backflow surface currents for @xmath131 that reduce the edge current when thermally populated ( _ cf _ fig .
[ fig - jydos_bound - state ] ) .
the presence of low - energy surface excitations is also evident in the spectral sums that define the edge current and angular momentum in eqs .
[ continuum - current - finite_t]-[lz_vs_t ] .
the result of ishikawa,@xcite and mcclure and takagi,@xcite , for the ground - state angular momentum is based a geometry with cylindrical symmetry , a chiral p - wave order parameter and many - body wave function that is an eigenfunction of the angular momentum operator and two - particle wave functions that vanish at the boundary .
the analysis presented above relies on the formation of edge states by boundary scattering in the presence of a chiral order parameter .
the resulting chiral edge states , and their dispersion relation shown in fig . [ fig - chiral_edge - state_dispersion ] , play a key role in generating the edge current carried by the continuum states and the resulting ground - state angular momentum of @xmath172 .
one can ask `` how robust are these results to boundary conditions , geometry and topology ? '' ) of @xmath173 superfluid ( `` 2d '' ) confined in a non - cylindrical geometry with area @xmath174 bounded in the @xmath175 plane by specular surfaces .
double reflections are important in determining the surface spectrum and edge current near a corner .
[ fig - film_geometry ] ] for example , consider the spectrum , edge currents and ground - state angular momentum for a geometry such as that shown in fig .
[ fig - film_geometry ] .
there are two classes of trajectories that determine the local spectral current density .
far from a corner ( @xmath176 ) trajectories with a _ single reflection _ determine the local surface spectrum , and for specular reflections we obtain the chiral edge states and the local edge currents of eqs .
[ chiral_bound - state ] and [ chiral_edge_current ] . however , near a corner the sharp change in curvature leads to _
double reflections _ as shown in fig .
[ fig - film_geometry ] .
these double reflections dramatically alter the local excitation spectrum .
they are also essential for enforcing current conservation near the corner , and they provide the mechanism for the edge currents to `` turn the corner '' and maintain continuity of the current circulating near the boundary .
furthermore , since the double reflections are relevant only for incident trajectories within a few coherence lengths of a corner the ground - state angular momentum measured from the center of mass of the film is given by @xmath172 for a finite number of corners , with corrections that are of order @xmath177 , where @xmath178 is the minimum linear dimension of the film .
are retro - reflectors of quasiparticles .
[ fig - retro - reflector ] ] this example also indicates how non - specular scattering can dramatically alter the surface spectrum , reduce or even eliminate the edge currents . to illustrate the effect non - specular scattering consider a surface that is facetted with mesoscopic mirror surfaces that are large compared to the fermi wavelength , but small compared to the coherence length , @xmath179 , and oriented at right angles to one another as shown in fig .
[ fig - retro - reflector ] .
such a surface is a retro - reflector analogous to optical retro - reflectors constructed from dense packing of corner reflectors.@xcite note that a retro - reflecting surface does not break time - inversion symmetry or reflection symmetry in a plane containing the normal to the surface , but translational invariance is broken on short - wavelength scales , @xmath180 . as a result
, retro - reflection can dramatically modify the spectrum of edge states .. ] in the limit of _ perfect retro - reflection _
- i.e. retro - reflection of all incident trajectories - the spectrum of edge states is obtained by an analogous calculation to that of perfect specular reflection since every incident trajectory is paired with a single reflected trajectory .
in particular , the quasiparticle propagator , and the corresponding bound - state spectral function , are given by ( see appendix [ appendix - quasiclassical ] ) [ gr_retro - reflection ] _
3(x , ; ) & = & - + e^-2()x / v_x , + _ ( x , ; ) & = & ||e^-2x / v_x ( ) .
[ dos_bound - state_retro - reflection ] in place of the chiral branch of edge states for perfect specular reflection ( fig .
[ fig - chiral_edge - state_dispersion ] ) , perfect retro - reflection leads to an edge state at the fermi level , @xmath181 , i.e. a _ zero - mode _ for every incident trajectory , @xmath9 .
these modes do not carry current , nor do they generate continuum currents . indeed the spectral current density ( eq . [ jdos ] ) vanishes identically , and thus the ground - state angular momentum resulting from the edge states vanishes as well .
the spectrum of zero - modes is also inferred from the observation that @xmath182 for any pair ( @xmath183 of retro - reflected trajectories .
thus , andreev s equation for a pair of retro - reflected trajectories is equivalent to dirac fermions in 1d coupled to a scalar field @xmath184 ( @xmath124 being the coordinate measured along the classical trajectory ) , which has the well - known jackiw - rebbi zero - mode bound to the domain wall at @xmath185.@xcite however , the zero modes generated by retro - reflection and a chiral p - wave order parameter are fragile and unprotected from small perturbations . for an _ imperfect _ retro - reflecting surface some incident trajectories will be reflected _ forward _ and generate edge currents and a ground - state angular momentum with a magnitude in proportion to the probability for forward reflection .
thus , depending on the distribution of trajectories with forward vs. retro - reflection the resulting ground - state angular momentum will generally be less than @xmath172 , and may take on any value in the range , @xmath186 , with the lower limit set by the _ intrinsic _ angular momentum,@xcite l_z^=(n/2)(/e_f)^2(e_f/ ) . for confined by the walls of an experimental cell
a realistic estimate for @xmath7 is likely below @xmath172 , but much larger than the intrinsic limit , and determined by the mean fraction @xmath187 of _ forward _ reflections by the boundary , i.e. @xmath188 ) , l_z = f(n/2 ) , with @xmath189 .
the sensitivity of the ground - state angular momentum to retro - reflection is at first sight in conflict with the result of mcclure and takagi ( mt ) .
however , the mt boundary condition does not account for retro - reflection on mesoscopic scales because it assumes perfect cylindrical symmetry on the atomic scale .
this result highlights the fact that spectrum of edge states , currents and the ground - state angular momentum is sensitive to surface scattering on all scales from several coherence lengths down to the atomic scale . ,
inner and outer radii @xmath190 , @xmath191 and area @xmath192 bounded by specular surfaces which reflect quasiparticles @xmath193 . for @xmath194
only single reflections are relevant to determining the surface spectrum and edgecurrents on the inner and outer boundaries .
[ fig - toroidal_geometry ] ] the combination of geometry and surface boundary conditions can lead to dramatically different results for the ground - state angular momentum of a chiral p - wave superfluid . consider the toroidal geometry shown in fig .
[ fig - toroidal_geometry ] in which the superfluid is confined between inner and outer baoundaries with radii @xmath195 and @xmath196 , respectively .
i assume both radii are large compared to the confinement scale of the chiral edge currents , and that the edge states on the inner and outer boundaries are well separated ; i.e. @xmath197 .
the ground state angular momentum is given the radial moment of the mass current density in eq .
[ angular - momentum_cylinder ] , which in these limits is determined by the mass _ sheet current _ on the inner and outer boundaries , @xmath198 and @xmath199 , respectively , l_z = 2h(k_1r_1 ^ 2 + k_2 r_2 ^ 2 ) . at @xmath28 the magnitude of the mass sheet current ( with units of `` action / volume '' ) for a specular boundary
is obtained from eq .
[ chiral_edge_current ] with @xmath127 , and evaluates to k = _
0^dxg_(x ) = n_fv_fp_f = n. for perfect specular reflection on both boundaries we obtain edge currents of equal magnitude flowing in opposite directions , @xmath200 , as indicated in fig .
[ fig - toroidal_geometry ] , and thus once again the mt result for the ground - state angular momentum , [ lz_toroidal_specular ] l_z = 2h(r_1 ^ 2 - r_2 ^ 2)n= .
note that the counter - propagating edge currents conspire to give a ground - state angular momentum , in units of @xmath201 , that is extensive and proportional to the volume , or total number of particles .
if the boundary is not perfectly specular then the corresponding sheet current is reduced by the suppression of the edge currents by retro - reflection : @xmath202 , with suppression factor @xmath203 . for the toroidal geometry the inner and outer boundaries may have different degrees of specularity , i.e. @xmath204 and @xmath205 with @xmath206 .
the generalization of eq .
[ lz_toroidal_specular ] is [ lz_toroidal_non - specular ] l_z = ( ) , where @xmath207 is the ratio of the radii , @xmath208 .
the asymmetry in the counter - propagating edge currents now leads to a ground - state angular momentum that no longer scales with the volume .
two cases highlight the non - extensive property of @xmath7 , and its sensitivity to the asymmetry in the edge currents on different boundaries . for perfect specular reflection on the outer boundary , @xmath209 , and perfect retro - reflection on the inner boundary , @xmath210 , the resulting ground - state angular momentum l_z = ( )
, can be much larger than the mt result @xmath172 for @xmath211
. equally dramatic would be to engineer the outer boundary to be retro - reflecting , @xmath212 , and the inner boundary to be specular reflecting , @xmath213 . in this limit only the counter - circulating current on the inner boundary survives , which leads to a ground - state angular momentum that is to the chirality of the cooper pairs , l_z = ( ) .
this reversal of the ground - state angular momentum for a toroidal geometry would provide both be a signature of the broken time - reversal symmetry of the ground state of superfluid , and also establish its origin as the edge current from the inner boundary .
this research is supported by the national science foundation ( grant dmr-0805277 ) .
i also acknowledge the hospitality and support of the aspen center for physics where part of this work was carried out . |
in the eighties several authors@xcite suggested that the universe could have been spontaneously created though a tunneling event . what the nature of such a tunneling event could be is still unsettled .
today s theories suggest however that shortly after the planck era the universe was in an arbitrary state . a homogeneous and isotropic universe is believed to be a rather special configuration , so an arbitrary state suggests the universe was inhomogeneous .
recently the question of topology has been added into the discussion . what is the most probable topology of the universe ?
we know that there exist a lot more hyperbolic manifolds than elliptic and flat ones @xcite .
do we therefore live in a hyperbolic universe ? in this paper we shall study a class of cosmological models which has 2-dimensional locally homogeneous compact spacelike surfaces .
the 2-dimensional compact oriented surfaces can be classified in term of an integer @xmath2 , the _ genus _ of the surface .
there is one elliptic surface @xmath3 , one flat @xmath4 and a infinite of hyperbolic surfaces @xmath5 .
we write the metric in a similar form to that of kucha@xmath6 @xcite and kenmoku and collaborators ( kkty ) @xcite ] : @xmath7 @xmath8 and @xmath9 are all functions of @xmath10 and @xmath11 , and @xmath12 is the metric the @xmath2-holed torus is _ a _ metric on the @xmath2-holed torus , @xmath13 .
since the teichmller space for @xmath14 have dimension greater than 1 , there are an uncountable number of non - isometric metrics on @xmath15 .
] , @xmath13 .
these metrics cover a great variety of different spatial topologies . in terms of the thurston classification@xcite ,
their covering spaces are for instance : @xmath16 and @xmath17 .
thus a great variety of cosmological models are incorporated in the metric ( [ ltmetric ] ) .
our main concern will however be the lematre - tolman models ( or lt for short ) . in the lt models
this case includes the first three of the thurston geometric models .
the case where @xmath18 are the kantowski - sachs ( @xmath3 ) metric and their hyperbolic versions @xmath19 studied by fugandes@xcite . throughout this paper
we will denote all the cases @xmath20 as ks .
this model will link the models studied in the paper by nambu and sasaki@xcite with for instance the homogeneous frw models or the einstein desitter models .
this model can therefore among others describe a quantum tunneling of a universe from a schwarzschild black hole into a desitter universe .
we will choose the curvature of @xmath13 to be @xmath21 for @xmath22 and @xmath5 respectively .
the gauss - bonnet theorem then implies _
g = 4(1-g ) where @xmath23 is the area of the @xmath24holed torus @xmath13 .
if we choose @xmath25 we can write _
g = 4n_g where n_g= 1 , & g=0 , 1 + ( g-1 ) , & g2 from the action of pure gravity .
] , @xmath26 , we obtain for these models @xmath27 } \end{split}\end{aligned}\ ] ] where the prime and dot denote the derivative with respect to @xmath10 and @xmath11 respectively . a cosmological constant
is also included .
we will also consider a matter term or source term of the form : s_d=-4n_gdt dr n where @xmath28 an independent function .
we introduce this source function because it is one of the simplest ways to generate inhomogeneities . in the presence of classical dust
this function @xmath28 could be related to the dust density @xmath29 by the equation @xmath30 .
thus in this description @xmath28 is not a dynamical matter field , we have basically solved the matter equations and reinserted the solutions into the lagrangian . the total action is now assumed to be @xmath31 .
the canonical momenta are : @xmath32\label{canmomp}\end{aligned}\ ] ] through a legendre transform we can bring the action into a canonical form : @xmath33\ ] ] where @xmath34\label{enconstraint } \end{split}\\ { \mathfrak h}_r= & r'p_r - qp_q ' \label{momconstraint}\end{aligned}\ ] ]
from the canonical form of the action we can readily write down the energy constraint and the momentum constraint : @xmath35 it is also useful to study a special combination of these two constraints , @xmath36 where @xmath37 is called the mass function and is defined by @xmath38\ ] ] and @xmath39 is defined by @xmath40 inserting the relation @xmath41 we see that in the case @xmath42 we can interpret the function @xmath39 as the mass of the dust inside a spherical shell of radius @xmath9 .
if we calculate the equal - time poisson algebra we obtain : \ { h_r ( r),h_r(r^*)}&=&h_r(r)(r - r^*)-h_r(r^ * ) (r^*-r ) + \ { h ( r),h_r(r^*)}&=&h(r)(r - r^*)+h(r ) (r - r^ * ) + \ { h_r ( r),h_r(r^*)}&=&h_r(r)(r - r^*)-h_r(r^ * ) (r^*-r ) + \ { h ( r),h(r^*)}&=&q(r)^-2h_r(r)(r - r^ * ) -q(r^*)^-2h_r(r^*)(r^*-r ) + \ { m(r),h_r(r^*)}&=&m(r)(r - r^ * ) + \ { m(r),h(r^*)}&=&-q(r)^-3r(r)h_r(r)(r - r^ * ) + \{m(r),m(r^*)}&=&0 thus according to the above equations the mass function is a constant of motion in a weak sense .
for simplicity , let us work in the gauge @xmath43 .
but for the time being @xmath44 is still arbitrary .
the constraint @xmath45 can easily be solved in quadrature . using the expressions for the canonical momenta , eq .
[ canmomq ] and [ canmomp ] , combined with the constraint equation [ momconstraint ] , and simplifying yields the differential equation : @xmath46 which has the solution @xmath47 .
the function @xmath48 satisfies @xmath49 for the present time we have to be a bit careful when we solve this equation . if @xmath50 which happens in the ks case , the momentum constraint yields @xmath51 . for @xmath52 we can solve the equation for @xmath48 : @xmath53\ ] ] and @xmath54 is an arbitrary function of @xmath10 .
we can add these results together and keep the above expression for @xmath55 but with the additional restriction @xmath56 whenever @xmath50 .
we will ignore the possibility that the term @xmath57 in the exponential function causing trouble''-expression for @xmath57 . ] .
thus if there exists a @xmath58 and a @xmath59 such that @xmath60 , then @xmath61 for all @xmath11 .
the mass function is always well defined and turns into : [ meq ] m=+(-f^2)r - r^3=m where @xmath62 . if @xmath63 the mass equation ( [ meq ] ) is separable and can be solved in quadrature .
let us assume @xmath64 .
this implies @xmath63 . solving the mass equation with respect to @xmath48 yields : @xmath65 in order to find the lapse @xmath44 we have to use the equation for @xmath66 : -_q & = & \{p_q , drnh } + & = & + n_g ( -(-r^2)+ ) inserting @xmath67 into the equation for @xmath68 yields the following equation for the lapse @xmath44 : @xmath69 as we see , we might as well ( at least locally ) consider @xmath9 as our new radial variable , i.e. @xmath70 . if we do so , the solution to the above equation can be written as : @xmath71 where @xmath72 and @xmath73 is an arbitrary function of @xmath11 .
thus [ timeinmetric ] ds^2=-^2(t)e^idt^2++r^2 d_g^2 wherever @xmath74 if for instance we have constant dust density , i.e .. we define @xmath75 so that it is not a funtion of @xmath11 , while @xmath29 in general could be a function of @xmath11 . ]
@xmath76 the integral may be evaluated in terms of elementary functions @xcite .
for all the desitter spaces ( @xmath77 ) the integral @xmath78 is simply : i=(r - r^3 ) for physical spacetimes .
we notice that @xmath79 is not compatible with a @xmath80 .
but for @xmath81 we may have any @xmath82 .
all of these spaces will have spatial sections which have constant negative curvature , i.e. they belong to the thurston classification @xmath83 .
the cases @xmath84 are not compatible with compact spatial sections , but @xmath42 may have compact spatial sections if we demand that the metric [ timeinmetric ] is defined locally in @xmath10 .
if we demand that the metric should be globally defined then we obtain the space @xmath83 itself .
unfortuneatly a classification of the possible compact quotients of @xmath83 is still far from being established .
let us for simplicity s sake set @xmath85 . 1 .
this is the non - compact multiply connected space investigated by sokoloff and starobinskii @xcite .
this space is best illustrated in the poincar half space model . ] .
if we do the coordinate transformation @xmath86 the metric turns into @xmath87 which is manifest on poincar half space form , we obtain the `` euclidean '' section of ads@xmath0 space . here
we see that the euclidean section is the 4-dimensional hyperbolic space @xmath88 . ] . we identify points along two pairs of planes parallel to the @xmath89-axis .
it is easily seen that this space does not allow compactification in the @xmath89 ( or @xmath10 ) direction .
this section of the hyperbolic space can be illustrated as follows .
draw two @xmath90-gons on the boundary of the poincare ball model .
these polygons can be placed so that they are symmetric to each other through the origin .
connecting the polygons using geodetic planes we obtain a space that look like an apple core(see figure [ apple ] ) .
also this case will not allow a compactification in the @xmath10 direction .
actually we have cheated a little bit , because the space contains a horizon with topology as a @xmath2-torus , much like that of a black hole @xcite .
the area of this horizon is @xmath91 thus apparently we can relate to it the bekenstein - hawking entropy : @xmath92 this metric is very similar to the btz black hole in ads@xmath1@xcite .
including a mass term in this case is straight forward .
note also that in the above case the `` black hole '' entropy is a feature of the negatively curved compact spaces @xmath13 . to obtain the ads@xmath0 apple we have actually taken two such spaces and glued them along their horizons to obtain a topological complete space
. note also that the horizon is the only @xmath93 surface which is totally geodesic . to ensure regularity of the total space after we have performed a gluing we have to glue along totally geodesic surfaces .
if we had demanded that all of these cases should be simply connected , they would have yielded the same space @xmath83 .
but since we have assumed that our locally homogeneous 2-dimensional surfaces are compact all of the above cases yield topologically different spaces . again @xcite we see that assuming _ compact _
2-dimensional locally homogeneous spaces ensure that the different cases will yield different solutions . in the kantowski - sachs case we have @xmath50 .
this implies @xmath94 .
the mass equation will now yield : = + r - r^3 where @xmath95 is a constant . inserting this into the energy constraint
we obtain an equation similar to the equation [ lapseeq ] : [ qeq ] = ( -+ ) again we see that we can equally well consider @xmath9 to be @xmath96 .
this equation is very similar to the equation [ lapseeq ] .
this is not surprising since if we smoothly continue the schwarzschild metric to the region inside the black hole horizon we obtain the ks metric inside the horizon@xcite .
smooth continuing the desitter solution to the region outside the desitter horizon gives the ks metric@xcite .
we also note that the @xmath97 symmetry is manifestly broken with the inclusion of classical dust . due to the conservation of the rest mass of the dust ,
we would obtain @xmath98 where @xmath99 is the coordinate volume if the spatial sections @xmath100 are compact .
we re - write these equations for a general dust distribution .
we assume that the topology in the @xmath10-direction is that of a circle @xmath101 .
if @xmath102 then we can write = m+_n=1^we use the same trick to write @xmath103 where g(r , t)=_n=0^ since the cosines and the sines form a complete and orthogonal set of functions on the circle the equations for @xmath104 are _ n(t)= the equation for @xmath105 is _ 0(t)= can be integrated at once by elementary calculus . in the absence of dust @xmath106
this reduces to the ordinary inner / outer solutions of the schwarzschild - desitter spacetime .
this is just the homogeneous solution of the ks metric . if there are inhomogeneities we have to solve the equations for @xmath107 .
unfortunately we then have to know the evolution of the matter content .
if we on the other hand knew the metric , i.e. the functions @xmath108 we could uniquely determine the matter distribution .
so far we have only investigated time - independent and ks - like solutions .
we saw that they represented different sections of a more general spacetime .
let us now look at the lt solutions , which we can obtain if we choose the universal time gauge @xmath109 .
the coordinates now follow the collapsing / expanding dust .
if we choose @xmath109 , @xmath110 will follow .
we can now write the mass equation as an `` energy '' equation : @xmath111 where the `` potential '' @xmath112 and the `` energy '' @xmath113 are given by @xmath114 @xmath115 this energy equation may be integrated and solved exactly .
a summary of the results and a qualitative description of the physical meaning of the solutions is given in @xcite .
the solutions may be written in terms of the weierstrass elliptic functions @xcite .
the actual expressions are not very informative unless the reader has massive knowledge of these elliptic functions . however a lot of qualitative information can be extracted from simple classical considerations .
we will however solve the energy equation in the big and small @xmath9 limit .
if we neglect the @xmath39 term in the energy equation , we get a simple and easily solvable equation .
the equation becomes : @xmath116 where @xmath117 ( assuming @xmath118 ) this equation can be solved in quadrature , and the solutions are desitter - like solutions : @xmath119 if the @xmath120 term is neglected , we end up with a energy equation which looks like : @xmath121 introducing a parameter @xmath122 we can also solve this equation in quadrature : @xmath123 from these solutions we notice that all the small @xmath9 solutions goes as + @xmath124 as @xmath125 . in the special case @xmath126 and
@xmath127 the equations [ storreennull ] , [ liknull ] and [ mindreennull ] reduces to the well known matter dominated homogeneous frw solutions .
the classical solutions will move on level curves of the energy function @xmath128 , since the total energy @xmath113 is independent of @xmath11 . in figure
[ levelcurves ] the level curves of a typical energy function are drawn . if @xmath118 there will exist a saddle - point of the energy function .
t his saddle point will be at @xmath129 where the energy function will have the value @xmath130 .
the saddle point solution is static , and is as a matter of fact the einstein static universe .
this can be done since we have to assume that @xmath131 on physical grounds . in the case
@xmath132 , the metric become degenerate . ] .
if @xmath133 , the solutions fall into two distinct classes : 1 .
schwarzschild - like solutions : these solutions expands , but they do not possess enough energy to escape the gravitational collapse , so they end as black holes . if @xmath39 is constant these solutions are those of a schwarzschild black hole in lematre coordinates .
desitter - like solutions : solutions where the universe evolves approximately as that of desitter solutions with positively curved hypersurfaces . if @xmath134 the ( test ) matter has enough energy to escape the gravitational collapse ( expanding solutions ) or enough energy to prevent the gravitational repulsion from the cosmological constant ( contracting solutions ) .
all of these solutions are also valid for the case @xmath94 the ks case . but to obtain the ks solutions it is not enough to solve just the mass equation .
from here on we will only consider the case where the cosmological constant is non - negative .
as we saw from the classical case , if the `` energy '' is small enough the solutions are confined to be either schwarzschild - like or desitter - like .
these two regions are separated by a classically forbidden region . from classical quantum mechanics
we have learned that this does not have to be the case for a quantum system .
a quantum mechanical system may tunnel though classical barriers . in our case
this means that a small schwarzscild - like universe may tunnel though the potential barrier to become a desitter - like universe !
this has striking consequences .
if we imagine that a `` mother '' universe could spontaneously create schwarzschild - like black holes of less than a planck - length , there will be a non - zero probability that one of these will tunnel through the potential barrier , and end up in an ever - expanding desitter state .
these considerations are of course rough and based on classical quantum mechanics .
whether this scenario is probable in the yet unknown theory of quantum gravity remains to be proven .
however we will use a simplified version of the wd equation and estimate the form of the wave function in various limits . we will from now on consider only the spherically symmetric case , i.e. @xmath42 . the dirac quantization procedure is to let the constraint equations become operator equations of the wave function : @xmath135 the first of these equations is the wheeler - dewitt equation .
in addition we have the momentum constraint equation .
these two equations are functional differential equations and thus in general very difficult to solve .
kkty managed in the case @xmath136 to find a couple of special cases where the wave function was exactly derived and interpreted @xcite . since the mass equation is a special combination of the momentum and energy constraint , the mass equation itself can be considered as a constraint .
the wave function has to obey : @xmath137 the wavefunction has to be an eigenfunctional of the mass - operator @xmath138 with eigenvalue @xmath139 , and from the classical interpretation we have to assume that the eigenvalue @xmath139 has to obey @xmath140 . in the wkb approximation
we can integrate the hamilton - jacobi equation if we choose a particular and simple path @xcite .
we will not use these results .
neither will we give any attempt to solve all three constraint equations simultaneously in the `` correct '' quantum - mechanical manner . to simplify the equations we will do the following : 1 .
solve the momentum constraint classically .
when we solve it we will assume that @xmath52 , thus the ks case is excluded .
solving the momentum constraint will reduce the number of canonical variables , so we end up with only @xmath9 and its canonical conjugated momentum @xmath141 .
2 . choose the gauge @xmath51 so that we will not have any problems with the function @xmath48 being dependent of @xmath11 .
3 . find the new hamiltonian which reproduces the classical equations .
the reduced action which reproduces the classical equations is : @xmath142'\ ] ] calculating the canonical momentum @xmath143 doing a legrendre transform to bring the action onto a canonical form , and removing boundary terms , we arrive at the hamiltonian which reproduces the classical equations : @xmath144'\ ] ] the constraint equation will now yield the correct classical equation ; the mass equation . calculating the equation for @xmath141
, we find the identity , @xmath145 , which tells us that the mass function is independent of time .
the wave function has to satisfy the constraint equation : @xmath146 where @xmath147 a possible constant , due to the derivative with respect to @xmath10 is assumed to be absorbed into the function @xmath39 .
this simplified version of the lt model is not only simpler but the action has the great advantage that it reduces to the desitter models studied by vilenkin and hartle - hawking if we demand that @xmath77 , @xmath148 and @xmath127 .
, @xmath148 and @xmath127 where @xmath75 and @xmath149 are constants into the action , the @xmath10 dependent part can be factored out , and the integral may be performed .
the actual integral obtained is : @xmath150 in the case of @xmath151 . ]
the solution most likely to describe the tunneling behaviour emphasized in the beginning of this section , is the tunneling wavefunction suggested by vilenkin . in the wkb approximation
we assume that the wave function has the form @xmath152 , to the lowest order we get the hamilton - jacobi equation : @xmath153=0\ ] ] if we assume that @xmath154 the resulting equation will be the hamilton - jacobi equation for a point particle with action @xmath155 ( @xmath10 is only a parameter ) . in the hamilton - jacobi equation
the functional @xmath156 turn out to be the action at the classical level . since the classical action can be written as an integral over @xmath10 the assumption @xmath154 is therefore reasonable at the lowest order wkb level .
this separation of the action is also a feature of the dust , in the sense that dust does not self - interact .
because the matter does not self - interact the matter equations do not depend on neighbouring points .
we can interpret the action @xmath157 as the action of a point particle moving in a potential @xmath158 $ ] with zero energy .
the wkb wavefunction @xmath159 for the point particle can then be written @xmath160 .
the two wkb wavefunctions can therefore be related by @xmath161 .
finding first the wavefunction @xmath162 we can then relate its wkb approximation with @xmath163 through @xmath161 .
we imagine that the universe point starts off at @xmath164 and the behavior of the wave function of such a particle is dependent on the form of the potential .
the zeros of the potential indicate classical turning points , separating classically allowed / forbidden regions .
the potential @xmath165 has one zero at @xmath164 and possibly two more for positive values of @xmath9 .
whether there are two , one or zero , depends on the entity : @xmath166 iff @xmath167 then there will be two positive roots , @xmath168 and @xmath169 .
the region @xmath170 ( assuming @xmath171 ) is a classical forbidden region , thus the wavefunction is exponential in some way .
iff @xmath172 these two roots have emerged into one , and the case @xmath173 and @xmath174 has no positive roots ; the classical particle rolls down the slope to infinity .
one case which is very instructive to look at is the case where @xmath127 and the dust density is constant : @xmath175 .
the entity @xmath176 will then be a constant with respect to @xmath10 : @xmath177 we can interpret this quantum mechanically as if we imagine a spontaneously created universe `` bubble '' with high curvature , and a relatively small cosmological constant and dust density .
it is most likely to vanish before it has the opportunity to enter a desitter phase .
let us look at this from another point of view .
imagine that we can only observe `` big '' universes , universes at scales much larger than the planck length , or essentially just universes that are in the expanding desitter stage .
if we could in some way sit outside and see the flow of universes in expanding desitter stages , we would find that most of the universes with low @xmath120 and @xmath75 also have low hypersurface curvature .
let these thought experiments be for the moment because we do not actually know how to interpret the wave function of the universe or how to calculate it .
let us however try to estimate the behavior of the wavefunction is this model .
the most interesting case seems to be @xmath167 , so let us just look at this case .
there are essentially three different regions with different behavior .
1 . region i : small @xmath9 .
we can neglect the @xmath178-term .
region ii : intermediate region .
this is dominated by the part where the wave function has an exponential behavior .
region iii : large @xmath9 .
we can neglect the @xmath9-term .
we neglect the @xmath178-term in eq .
[ energidiff ] and complete the square with a new variable @xmath179 . then the schrdinger equation for @xmath162 reduces to a harmonic oscillator - like equation : @xmath180 where @xmath181 .
the solution of this equation can be written in terms of the _ parabolic cylinder functions _ @xmath182 : @xmath183 in the case of @xmath184 an integer , the @xmath182 can be written in terms of the hermite polynomials ( as one would expect ) : @xmath185 the index @xmath186 can be any number , in contrast to the 1-dimensional harmonic oscillator .
we neglect the @xmath9-term in eq .
[ energidiff ] .
this is equivalent to setting the mass function @xmath39 equal to zero .
we are then left with a desitter space - time .
the wavefunction for such space - times has been calculated using the wkb approximation .
most interesting to us is the tunneling wave function of vilenkin . through the rescaling : @xmath187 we can bring the equation to the same form as that of vilenkin @xcite , with a scalar field potential @xmath188 .
we use this result to obtain : @xmath189\right ) , & \frac{r^2\lambda}{3(1 -f^2)}<1 \\ \exp\left(-\frac{1}{3v}\left[1+i(\frac{r^2\lambda}{3(1-f^2)}-1)^ { \frac{3}{2}}\right]\right ) , & \frac{r^2\lambda}{3(1-f^2)}>1 \end{cases}\ ] ] let us assume that @xmath190 and @xmath127 where @xmath149 is assumed to be positive .
we will now use the tunneling solution eq .
[ viltun ] to find out what is most likely , the creation of a highly curved or a less curved space .
if we calculate the ratio @xmath191 where @xmath192 , we get : @xmath193\ ] ] where @xmath194 is a real function independent of @xmath149 which satisfies @xmath195 .
we thus have to conclude that to lowest order in the wkb approximation it is equally probable for a space with small spatial curvature to tunnel through the classical barrier than for a highly curved space . comparing with our previous statement
, we se that larger curvarture itself increases the probability amplitude .
but smaller curvature increases the volume by exactly the amount so that it cancels the contribution from the curvature
. here the wave function will be of exponential behavior .
but the actual shape is strongly dependent on the boundary condition . to find the wave function in this region to lowest order ,
the best is perhaps to use the function from region i and iii and cut and glue them together .
nordbury @xcite studied the frw model with different types of perfect fluids .
the tunneling picture which we have described here seems to agree with his results for a closed universe @xmath196 .
however in the case of open ( @xmath197 ) frw - universes he concludes that these models can not have been created through the tunneling picture . from our considerations however , this can not be ruled out , on the contrary , open universes do seem to tunnel from a @xmath164 state a lot easier than closed ones .
also an article by atkatz and pagels @xcite seems to indicate that open universes can not be created by a tunneling event .
a key argument used in both of these articles is that the transition amplitude is suppressed by the infinite integral over the three - dimensional hypersurfaces .
however , a more recent article by coule and martin @xcite used the fact that there exists @xmath197 compact spacelike hypersurfaces .
since they are compact they will have finite volume .
their conclusion is opposite of that of the previous two articles : the creation of a open @xmath197 frw universe is more likely than a closed @xmath198 universe through a tunneling event .
our case is however more subtle , since the hypersurfaces are no longer homogeneous .
as we have investigated the wkb wavefunctions in this model , we might wonder whether an inhomogeneous universe is more likely to be created through a tunneling event than a homogeneous one . as an inhomogeneity measure in this model
it is very useful to use the weyl curvature tensor @xmath199 .
all of the frw models are conformally flat , thus for the all of the frw model the weyl tensor will vanish . in our case
the opposite is also true , if @xmath200 and @xmath51 in the lt model then the spacetime is spatially homogeneous . if we define a mean dust density function @xmath201 by : m(r)=|r^3 we can write the weyl curvature invariant in the lt model as : c^c _ = ^2(|-)^2 thus this space - time is homogeneous if and only if @xmath202 .
the inhomogeneity in this model is therefore almost completely characterized by the mass function @xmath139 .
the homogeneous mass function will go as @xmath203 , where @xmath204 is a constant . investigating only closed universes ,
we fix the constant @xmath204 so that @xmath205 .
the most physical interesting universes that these lt models describe will have a larger amount of dust consentrated in the inner regions than the outher regions .
the most extreme inhomogeneous mass function is the schwarzschild mass function , @xmath206 .
thus it is very reasonable to assume that we have a mass distribution that obeys : m_sch(r)m(r)m_h(r ) if we compare the wkb tunneling potential @xmath207 $ ] for these different mass distributions we see that we get v_sch(r)v(r ) v_h(r ) keeping all other parameters fixed .
thus the more inhomogeneous the space is the smaller the tunneling barrier will be . in the wkb approximation
this means that the probability of the creation of an inhomogeneous universe is larger than for a homogeneous universe .
more specific the spontaneous creation of a schwarzschild - desitter like universe appears to be more probable than a homogeneous frw universe .
we also have be aware of that the matter lagrangian describes dust .
dust is an idealised form of matter that has no internal pressure . as the dust density becomes high , either in an early stage of the universe or in large local inhomogeneities this assumption of a pressure - free matter langrangian becomes somewhat artificial and unphysical . in these situations
we know that the internal pressure becomes very important and even quantum effects from the matter fields are essential in the description . in this sence we may say that these calculations more or less are `` valid '' only with small local inhomogeneities or they describe the effect from _ gravity itself _ to produce inhomogeneities .
this point is also emphasized by conradi@xcite .
conradi discusses the wave function of an anisotropic ks space with a dust lagrangian .
he argues that since dust interacts only locally , in the sence that it does not self - interact , this local `` tunneling '' process can not be interpreted as a tunneling effect .
our results should therefore perhaps be seen from a more phenomological point of view .
the matter lagrangian used in this paper is one of the most simple source field that give rise to inhomogeneities .
the source field used is also interesting because it connects the schwarzschild spacetime with the frw spacetime in a smooth way .
in a hamiltonian description of general relativity there has been a consensus among physicists and cosmologists that the issue of topology is essential and necessary in the description @xcite .
this has also been emphasized in this paper . at the classical level we saw
that apparently different cases would lead to the same spacetime if the symmetry spaces were infinite and simply connected . assuming compactness and multiply connectedness this pathology
was removed and yielded some interesting spacetimes . in the quantum mechanical case we provided a simpler model which was in general inhomogeneous , but removed many of the algebraic difficulties in the full kslt model .
it was emphasized how this model could describe a quantum creation of a inhomogeneous universe through quantum tunneling .
we also showed that under some reasonable assumptions the creation of an inhomogeneous universe was more probable than a homogeneous frw universe . in the absence of interacting matter
this is very reasonable because there are a lot more inhomogeneous configurations than homogeneous ones .
even though the full theory of quantum gravity is far from being established , it seems like quantum cosmology _ could _
describe many of the features of the initial universe . in the recent decade
the interest for topology in cosmology has expanded the application for quantum cosmology . even in classical gravity
the inclusion of non - trivial topologies has yielded new and surprising results .
i deeply appreciate the help and comments which yvind grn has given me during the work of this paper .
99 a. vilenkin , _ phys .
* b117 * , 25 ( 1982 ) a. vilenkin , _ phys .
rev . _ * d27 * , 2848 ( 1983 ) a. vilenkin , _ phys .
rev . _ * d30 * , 509 ( 1984 ) a. vilenkin , _ phys
. rev . _ * d33 * , 3560 ( 1986 ) t. vachaspati and a. vilenkin , _ phys .
rev . _ * d37 * , 898 ( 1988 ) d. atkatz and h. pagels , _ phys .
_ , * d25 * , 2065 ( 1982 ) m. lachize - rey and j - p .
luminet , _ phys .
rep . _ * 254 * , 135 ( 1995 ) w.p .
thurston , _ three - dimensional geometry and topology _ , * volume 1 * , princeton university press ( 1997 ) k. kucha@xmath6 , _ phys .
rev . _ * d50 * , 3961 ( 1994 ) m. kenmoku , h. kubotani , e. takasugi , y. yamazaki , _ phys . rev . _ * d59 * , ( 1999 ) w.p .
thurston , _ bull . am .
_ * 6 * , 357 ( 1982 ) h. fagundes , _ lett .
phys . _ * 6 * , 417 ( 1982 ) y. nambu and m. sasaki , _ prog .
phys . _ * 79 * , 96 ( 1988 ) k. kucha and c.g .
torre , _ phys .
rev . _ * d43 * , 419 ( 1991 ) d.s .
salopek , _ phys .
rev . _ * d46 * , 4373 ( 1992 ) s. hervik , _ cand .
thesis _ , university of oslo ( 1999 ) d. brill , j. louko and p. peldn , _ phys .
rev . _ * d56 * , 3600 ( 1997 ) m. baados , c. teitelboim and j. zanelli , _ phys .
* 69 * ( 1992 ) d.d .
sokoloff and a.a .
starobinskii , sov . astron . * 19 * , 629 ( 1975 ) s. hervik , _ class .
quantum grav .
_ * 17 * , 2765 ( 2000 ) i. bengtsson and s. holst , _ class . quantum grav . _
* 16 * , 3735 ( 1999 ) a. krasiski , _ inhomogeneous cosmological models _ , cambridge university press ( 1997 ) a. zecca , _
il nuovo cimento _ , * 106b * , 413 ( 1990 ) e. eriksen and .
grn , _ int .
phys . _ * 4 * , 115 ( 1995 ) m. kenmoku , h. kubotani , e. takasugi , y. yamazaki , gr - qc/9906056 ( 1999 ) w. fischler , d. morgan and j. polchinski , _ phys .
* d42 * , 4042 ( 1990 ) a. vilenkin , _ phys . rev . _ * d37 * , 888 ( 1988 ) j.w .
norbury , _ eur . j. phys .
_ * b433 * , 263 ( 1998 ) j.w .
norbury , _ phys .
* 19 * , 143 ( 1997 ) d.h .
coule and j. martin , _ phys .
rev . _ * d61 * , 063501 ( 2000 ) h - d .
conradi , _ class .
quantum grav . _
* 12 * , 2423 ( 1995 ) h - d .
conradi , _ int . j. mod
_ * d7 * , 189 ( 1998 ) a. ashtekar and j. samuel , _ class .
quantum grav . _ * 8 * , 2191 ( 1991 ) h. kodama , _ prog .
phys . _ * 99 * , 173 ( 1998 ) | in this paper we study a class of inhomogeneous cosmological models which is a modified version of what is usually called the lematre - tolman model .
we assume that we have a space with 2-dimensional locally homogeneous spacelike surfaces .
in addition we assume they are compact .
classically we investigate both homogeneous and inhomogeneous spacetimes which this model describe .
for instance one is a quotient of the ads@xmath0 space which resembles the btz black hole in ads@xmath1 . due to the complexity of the model
we indicate a simpler model which can be quantized easily .
this model still has the feature that it is in general inhomogeneous . how this model could describe a spontaneous creation of a universe through a tunneling event
is emphasized . |
in this paper only finite , non - oriented graphs without loops and multiple edges are considered . the vertex set and the edge set of a graph @xmath17
are denoted by @xmath18 and @xmath19 respectively .
let @xmath20 and @xmath21 be two graphs .
then , we denote by @xmath22 the graph @xmath17 for which @xmath23 and @xmath24 , where @xmath25 : x \in \operatorname{v}(g_1 ) , y \in \operatorname{v}(g_2)\right\}}$ ] , i.e. @xmath17 is obtained by making every vertex of @xmath20 adjacent to every vertex of @xmath21 .
the ramsey number is denoted by @xmath26 .
more detailed information on the ramsey numbers , which we use in this paper , can be found in @xcite .
all undefined terms and notations can be found @xcite .
+ the expression @xmath27 means that in every coloring of the edges of the graph @xmath17 in two colors there is a monochromatic triangle .
it is well known that @xmath28 .
therefore , if @xmath29 , then @xmath27 . in @xcite erds and hajnal posed the following problem : _ is there a graph @xmath27 with @xmath30 _ ?
a positive answer was given by several authors , and folkman @xcite was the first to construct such a graph with clique number 3 .
later , we will use the following fact : @xmath31 which is a special case of a more general result from @xcite . denote : @xmath32 .
@xmath33 . the edge folkman number @xmath4 is defined with @xmath34 according to folkmans result @xcite , @xmath4 exists if and only if @xmath5 . since @xmath28 and @xmath35 , @xmath36 . in 1968
graham @xcite obtained the equality @xmath9 by proving that @xmath37 .
nenov @xcite constructed a 15-vertex graph in @xmath38 in 1981 , thus proving @xmath39 . in 1999
piwakowski , radziszowski and urbaski @xcite completed the computation of the number @xmath40 by proving with the help of a computer that @xmath41 .
they also obtained all graphs in @xmath42 .
the graph @xmath17 obtained by the construction of folkman , for which @xmath27 and @xmath43 , has a very large number of vertices . because of this , in 1975 erds @xcite posed the problem to prove the inequality @xmath12 . in 1986
frankl and rdl @xcite almost solved this problem by showing that @xmath44 . in 1988
spencer @xcite proved the inequality @xmath45 , after an erratum by hovey , by using probabilistic methods . in 2008
lu @xcite constructed a 9697-vertex graph in @xmath46 , thus considerably improving the upper bound on @xmath11 .
soon after that , lus result was improved by dudek and rdl @xcite , who proved @xmath47 .
the best upper bound known on @xmath11 was obtained in 2012 by lange , radziszowski and xu @xcite , who constructed a 786-vertex graph in @xmath46 , thus showing that @xmath13 . in 1972
lin @xcite proved that @xmath48 .
the lower bound was improved by nenov @xcite , who showed in 1981 that @xmath49 . in 1984
nenov @xcite proved that every 5-chromatic @xmath50-free graph has at least 11 vertices , from which it is easy to derive that @xmath51 . from @xmath10 @xcite
it follows easily , that @xmath52 .
the best lower bound known on @xmath11 was obtained in 2007 by radziszowski and xu @xcite , who proved with the help of a computer that @xmath53 . according to radziszowski and xu @xcite , any method to improve the bound @xmath53 would likely be of significant interest .
a more detailed view on results related to the numbers @xmath4 is given in the book @xcite , and also in the papers @xcite , @xcite , @xcite and @xcite .
+ in this paper , we improve the lower bound on the number@xmath11 by proving with the help of a computer the following [ maintheorem : f_e(3 , 3 ; 4 ) geq 20 ] @xmath54 at the end of the paper , we consider the vertex folkman number @xmath15 and prove @xmath16 ( theorem [ theorem : 20 leq f_v(2 , 3 , 3 ; 4 ) leq 24 ] ) .
this is an improvement over the known bounds @xmath55 from @xcite and @xcite .
+ this paper is organized in 4 sections . in the first section ,
the necessary notations are introduced , an overview of the related results is given , and the main theorem is formulated . in the second section ,
we present a computer algorithm ( algorithm [ algorithm : finding ml_(max)(n ; p ; s ) ] ) , with the help of which in the third section we prove the main theorem . in the last fourth section ,
new bounds on the number @xmath15 are obtained ( theorem [ theorem : 20 leq f_v(2 , 3 , 3 ; 4 ) leq 24 ] ) .
let @xmath56 , @xmath57 be an independent set of vertices of @xmath17 , @xmath58 and @xmath59 . then
, obviously , @xmath17 is a subgraph of @xmath60 , and therefore @xmath61 , from which it is easy to see that @xmath62 . by this reasoning , in @xcite it is proved , but not explicitly formulated ( see the proofs of theorem 2 and theorem 3 ) , the following [ proposition : k_1 + h in mh_e(3 , 3 ; 5 ; n - abs(a ) + 1 ) ] let @xmath56 , @xmath57 be an independent set of vertices of @xmath17 and @xmath59 .
then , @xmath63 . with the help of proposition [ proposition : k_1 + h in mh_e(3 , 3 ; 5 ; n - abs(a ) + 1 ) ] in @xcite the authors prove that every graph in @xmath64 can be obtained by adding 4 independent vertices to a 14-vertex graph @xmath65 such that @xmath66 .
all 659 graphs in @xmath42 were found in @xcite . with the help of a computer
it was proved that , by extending such 14-vertex graphs @xmath65 with 4 independent vertices , it is not possible to obtain a graph in @xmath64 , i.e. @xmath67 and @xmath53 .
this method is not suitable for proving the main theorem , because not all graphs in @xmath68 are known and their number is too large . because of this , we first prove that if @xmath69 , then @xmath17 can be obtained by adding 4 independent vertices to some of 1 139 033 appropriately selected 15-vertex graphs , or by adding 5 independent vertices to some of 113 14-vertex graphs known from @xcite . with the help of a computer we check that these extensions do not lead to the construction of a graph in @xmath70 and
we derive that @xmath71 and @xmath14 .
the computer algorithm which we use ( algorithm [ algorithm : finding ml_(max)(n ; p ; s ) ] ) , just as the algorithm used in @xcite to prove that @xmath53 , is based on some ideas of algorithm @xmath72 from @xcite . for convenience
, we will use the following notations : @xmath73 @xmath74 from @xmath75 it follows that @xmath76 and from @xmath77 it follows that @xmath78 in @xcite it is proved that @xmath79 if and only if @xmath80 , and all 153 graphs in @xmath81 are found . later , we will use the following fact : @xmath82 from ( [ equation : g arrowse ( 3 , 3 ) rightarrow chi(g ) geq 6 ] ) it follows @xmath83 obviously , @xmath84 .
let @xmath85 .
from the equality @xmath10 @xcite and proposition [ proposition : k_1 + h in mh_e(3 , 3 ; 5 ; n - abs(a ) + 1 ) ] it follows that either @xmath86 or @xmath87 , and from @xmath77 it follows that @xmath88 .
thus , we obtain @xmath89 we will need the following proposition , which follows easily from proposition [ proposition : k_1 + h in mh_e(3 , 3 ; 5 ; n - abs(a ) + 1 ) ] : [ proposition : h in ml(n - abs(a ) , p + 1 ) ] let @xmath90 , @xmath57 be an independent set of vertices of @xmath17 and @xmath59 . then @xmath91 .
we denote by @xmath92 the set of all maximal @xmath50-free graphs in @xmath93 , i.e. the graphs @xmath94 for which @xmath95 for every @xmath96 . since every graph in @xmath97 is contained in a maximal @xmath50-free graph ,
according to ( [ equation : mh(3 , 3 ; 4 ; 19 ) = ml(19 ; 0 ) = ml(19 ; 0 ; 4 ) cup ml(19 ; 0 ; 5 ) ] ) to prove the main theorem it is enough to prove that @xmath98 and @xmath99 . in the proofs of these inequalities
we will use algorithm [ algorithm : finding ml_(max)(n ; p ; s ) ] , formulated below .
the graph @xmath17 is called a @xmath100-graph if @xmath101 contains a new @xmath102-clique for every @xmath96 .
we denote by @xmath103 the set of all @xmath100-graphs in @xmath93 .
let @xmath104 .
let @xmath57 be an independent set of vertices of @xmath17 , @xmath58 and @xmath59 . according to proposition [ proposition : h in ml(n - abs(a ) ,
p + 1 ) ] , @xmath105 . from @xmath106
it follows that @xmath65 is @xmath100-graph , and from @xmath107 it follows that @xmath108 .
therefore , @xmath109 for some @xmath110 .
thus , we proved the following [ proposition : h in bigcup(s leq s)ml_(+k_3)(n - s ; p + 1 ; s ) ] let @xmath104 .
let @xmath57 be an independent set of vertices of @xmath17 , @xmath58 and @xmath59 .
then , @xmath111 .
the graph @xmath17 is called a sperner graph if @xmath112 for some pair of vertices @xmath113 . if @xmath94 and @xmath112 , then @xmath114 , @xmath115 .
therefore , every sperner graph @xmath94 is obtained by adding one vertex to some graph @xmath116 , @xmath115 . in the special case ,
when @xmath17 is a sperner graph and @xmath104 , from @xmath112 it follows that @xmath117 , and therefore @xmath118 , @xmath115 , i.e. @xmath17 is obtained by duplicating a vertex in some graph @xmath119 .
all non - sperner graphs in @xmath92 are obtained with the help of the following algorithm , which is based on proposition [ proposition : h in bigcup(s leq s)ml_(+k_3)(n - s ; p + 1 ; s ) ] : [ algorithm : finding ml_(max)(n ; p ; s ) ] finding all non - sperner graphs in @xmath92 for fixed @xmath120 , @xmath121 and @xmath122 . _ 1 .
_ find the set of graphs @xmath123 .
the obtained graphs in @xmath92 will be output in @xmath124 , let @xmath125 .
_ for each graph @xmath126 : _ 2.1 . _
find the family @xmath127 of all maximal @xmath128-free subsets of @xmath129 .
_ find all @xmath122-element subsets @xmath130 of @xmath131 , which fulfill the conditions : \(a ) @xmath132 for every @xmath133 and for every @xmath134 .
\(b ) @xmath135 $ ] for every @xmath136 .
\(c ) if @xmath137 , then @xmath138 .
_ for each @xmath122-element subset @xmath130 of @xmath131 found in step 2.2 construct the graph @xmath139 by adding new independent vertices @xmath140 to @xmath129 such that @xmath141 .
if @xmath17 is not a sperner graph and @xmath106 , then add @xmath17 to @xmath124 .
_ remove the isomorph copies of graphs from @xmath124 .
_ remove from @xmath124 all graphs with chromatic number less than @xmath142 .
_ remove from @xmath124 all graphs @xmath17 for which @xmath143 . to avoid repetition in the proofs of theorem [ theorem : finding ml_(max)(n ; p ; s ) ] and theorem [ theorem : 20 leq f_v(2 , 3 , 3 ; 4 ) leq 24 ] we formulate the following [ lemma : finding max supgraphs ] after the execution of step 2.3 of algorithm [ algorithm : finding ml_(max)(n ; p ; s ) ] , the obtained set @xmath124 coincides with the set of all maximal @xmath50-free non - sperner graphs with independence number @xmath122 which have an independent set of vertices @xmath144 such that @xmath145 .
suppose that in step 2.3 of algorithm [ algorithm : finding ml_(max)(n ; p ; s ) ] the graph @xmath17 is added to @xmath124 .
let @xmath139 where @xmath146 and the following notations are the same as in step 2.3 .
by @xmath126 , we have @xmath147 . since @xmath148 , are @xmath128-free sets , it follows that @xmath149 . by @xmath126
, we have @xmath108 . from this fact and the condition ( c ) in step 2.2
it follows that @xmath150 .
since @xmath151 is an independent set of vertices of @xmath17 , we have @xmath107 and @xmath152 . the two checks at the end of step 2.3 guarantee that @xmath17 is a maximal @xmath50-free non - sperner graph .
let @xmath17 be a maximal @xmath50-free non - sperner graph with independence number @xmath122 and @xmath153 is an independent set of vertices of @xmath17 such that @xmath154 .
we will prove that , after the execution of step 2.3 of algorithm [ algorithm : finding ml_(max)(n ; p ; s ) ] , @xmath155 . since @xmath17 is a maximal @xmath50-free graph , @xmath156 ,
are maximal @xmath128-free subsets of @xmath157 , and therefore @xmath158 ( see step 2.1 ) .
let @xmath159 . since @xmath17 is not a sperner graph
, @xmath146 is an @xmath122-element subset of @xmath131 and @xmath146 fulfills the condition ( a ) in step 2.2 . from the maximality of @xmath17
it follows that @xmath146 fulfills the condition ( b ) , and from @xmath107 it follows that @xmath146 also fulfills ( c ) .
thus , we showed that @xmath146 fulfills all conditions in step 2.2 , and since @xmath139 is a maximal @xmath50-free non - sperner graph , in step 2.3 @xmath17 is added to @xmath124 .
[ theorem : finding ml_(max)(n ; p ; s ) ] after the execution of algorithm [ algorithm : finding ml_(max)(n ; p ; s ) ] , the obtained set @xmath124 coincides with the set of all non - sperner graphs in @xmath92 .
suppose that , after the execution of algorithm [ algorithm : finding ml_(max)(n ; p ; s ) ] , @xmath155 . according to lemma [ lemma : finding max supgraphs ] , @xmath17 is a maximal @xmath50-free non - sperner graph and @xmath107 .
now , from step 5 it follows that @xmath104 .
conversely , let @xmath17 be an arbitrary non - sperner graph in @xmath92 .
let @xmath57 be an independent set of vertices of @xmath17 , @xmath58 and @xmath59 . according to [ proposition
: h in bigcup(s leq s)ml_(+k_3)(n - s ; p + 1 ; s ) ] , @xmath126 .
now , from lemma [ lemma : finding max supgraphs ] we obtain that , after the execution of step 2.3 , the graph @xmath17 is included in the set @xmath124 . by ( [ equation : g in ml(n ; p ) rightarrow chi(g ) geq 6 - p ] ) , after the execution of step 4 , @xmath17 remains in @xmath124 .
it is clear that after step 5 , @xmath17 also remains in @xmath124 .
[ remark : delta(g ) geq 8 ] in @xcite it is proved that if @xmath85 and @xmath160 , then @xmath161 . since @xmath162 , it follows easily that for each graph @xmath163 we have @xmath161 . using this result
we can improve algorithm [ algorithm : finding ml_(max)(n ; p ; s ) ] in the case @xmath164 in the following way : \1 . in step 1
we remove from the set @xmath165 the graphs with minimum degree less than 8 - s. \2 . in step 2.2
we add the following conditions for the subset @xmath146 : \(d ) @xmath166 for every @xmath134 .
\(e ) if @xmath137 , then @xmath167 for every @xmath168 .
this way it is guaranteed that in step 2.3 only graphs @xmath17 for which @xmath161 are added to the set @xmath124 .
let us note , however , that in the proof of theorem [ theorem : 20 leq f_v(2 , 3 , 3 ; 4 ) leq 24 ] , algorithm [ algorithm : finding ml_(max)(n ; p ; s ) ] must definitely be used without these changes .
the _ nauty _ programs @xcite have an important role in this paper .
we use them for fast generation of non - isomorphic graphs and for graph isomorph rejection .
according to ( [ equation : mh(3 , 3 ; 4 ; 19 ) = ml(19 ; 0 ) = ml(19 ; 0 ; 4 ) cup ml(19 ; 0 ; 5 ) ] ) it is enough to prove that @xmath98 and @xmath99 .
_ 1 . proof of @xmath98 .
_ we generate all 11-vertex graphs with a computer and among them we find all 362 439 graphs in @xmath169 and all 7 949 015 graphs in @xmath170 .
let us denote @xmath171 .
by ( [ equation : ml(n ; p ; s ) = emptyset , if n geq 9 and s leq 2 ] ) , @xmath172 .
we execute algorithm [ algorithm : finding ml_(max)(n ; p ; s ) ] with @xmath173 , @xmath174 , @xmath175 , having @xmath176 on the first step . according to [ theorem :
finding ml_(max)(n ; p ; s ) ] , we find all 5750 non - sperner graphs in @xmath177 . among the graphs in @xmath81 , which are known from @xcite ,
there are 8 maximal @xmath50-free graphs and they all have independence number 4 . by adding a new vertex to each of these 8 graphs which duplicates some of their vertices , we obtain all 20 non - isomorphic sperner graphs in @xmath177 ( see the text before algorithm [ algorithm : finding ml_(max)(n ; p ; s ) ] ) . thus , all 5770 graphs in @xmath177 are obtained . all graphs @xmath17 for which @xmath149 and @xmath178 are known and can be found in @xcite .
there are 640 such 15-vertex graphs , among which we find the only 2 graphs in @xmath179 . from ( [ equation : ml(n ; p ; s ) = emptyset , if n geq 9 and s leq 2 ] ) it follows that every graph from the set @xmath180 is a subgraph of some graph in @xmath179 or in @xmath177 . by removing edges from the graphs in @xmath181
we obtain all graphs in @xmath182 ( 1 139 028 graphs in @xmath183 and 5 graphs in @xmath184 ) . by ( [ equation : ml(n ;
p ; s ) = emptyset , if n geq 9 and s leq 2 ] ) , @xmath185 .
we execute algorithm [ algorithm : finding ml_(max)(n ; p ; s ) ] with @xmath186 , @xmath187 , @xmath175 , having @xmath188 on the first step . in step 2.3 ,
2 551 314 graphs are added to the set @xmath124 , 2 480 352 of which remain in @xmath124 after the isomorph rejection in step 3 .
after step 4 , 2 597 graphs with chromatic number 6 remain in @xmath124 . in the end , after executing step 5 we obtain @xmath125 . from @xmath53
it follows that in @xmath189 there are no sperner graphs and by theorem [ theorem : finding ml_(max)(n ; p ; s ) ] we obtain @xmath98 .
proof of @xmath99 .
_ from the graphs in @xmath81 , which are known from @xcite , with the help of a computer we find all 85 graphs in @xmath190 and all 28 graphs in @xmath191 .
let us denote @xmath192 . by ( [ equation : ml(14 ; 1 ; s ) neq emptyset leftrightarrow s in set(4 , 5 , 6 , 7 ) ] ) , @xmath193 .
we execute algorithm [ algorithm : finding ml_(max)(n ; p ; s ) ] with @xmath186 , @xmath187 , @xmath194 , having @xmath195 on the first step . in step 2.3 ,
502 901 graphs are added to the set @xmath124 , 251 244 of which remain in @xmath124 after the isomorph rejection in step 3 .
after step 4 , 31 graphs with chromatic number 6 remain in @xmath124 . in the end , after executing step 5 we obtain @xmath125 .
since there are no sperner graphs in @xmath97 , by theorem [ theorem : finding ml_(max)(n ; p ; s ) ] we obtain @xmath99 .
+ all computations were done on a personal computer .
using one processing core , the time needed to execute algorithm [ algorithm : finding ml_(max)(n ; p ; s ) ] in the case @xmath186 , @xmath187 , @xmath175 is about one hour , and in the case @xmath186 , @xmath187 , @xmath194 it is about half a minute .
note that in the first case among the 2597 6-chromatic graphs obtained after step 4 of algorithm [ algorithm : finding ml_(max)(n ; p ; s ) ] , 794 have a minimum degree of 8 or more . in the second case ,
the total number of 6-chromatic graphs is 31 , 11 of which have a minimum degree of 8 or more .
using the improvements of algorithm [ algorithm : finding ml_(max)(n ; p ; s ) ] described in remark [ remark : delta(g ) geq 8 ] , the time needed for computations is reduced almost 2 times in the first case and more than 10 times in the second case . [
remark : ml_(max)(19 ; 0 ; 5 ) = emptyset ] let us note that the result @xmath99 can also be obtained with the algorithm from @xcite , but more computations have to be performed by the computer .
let @xmath17 be a graph and @xmath196 be positive integers .
the expression @xmath197 means that in every coloring of @xmath18 in @xmath122 colors there exist @xmath198 such that there is a monochromatic @xmath199-clique of color @xmath200 .
folkman @xcite proved that @xmath203 exists if and only if @xmath205 . in this section
, we consider the numbers @xmath206 and @xmath207 , which are related to the edge folkman number @xmath11 .
according to folkmans result , these two numbers exist if and only if @xmath5 .
[ remark : ml(15 ; 1 ) and mh_v(3 , 3 ; 4 ; 15 ) ] let us note , that @xmath223 @xcite , but @xmath224 .
we found all 2 081 234 graphs @xmath225 . in the proof of the main theorem we already found the only 2 graphs in @xmath179 and all 5770 graphs in @xmath177 . with the help of algorithm [
algorithm : finding ml_(max)(n ; p ; s ) ] , similarly to the case @xmath175 , we determine @xmath226 .
we obtain all 826 graphs @xmath227 , all 12 graphs in @xmath228 and @xmath229 .
thus , we obtain all maximal @xmath50-free graphs in @xmath225 . by removing edges from them ,
we find all 2 081 234 graphs in @xmath225 . some properties of these graphs are listed in table [ table : ml(15 ; 1 ) statistics ] . among the graphs in @xmath225
there are exactly 20 graphs , which are not in @xmath230 .
properties of these 20 graphs are given in table [ table : ml(15 ; 1 ) statistics ] , and one of these graphs ( which has 51 edges ) is shown in figure [ figure : l_15_1_-_fv33_example ] .
it is enough to prove that @xmath232 .
suppose the opposite is true and let @xmath233 be a maximal @xmath50-free graph .
since @xmath234 , @xmath17 is not a sperner graph . from proposition [ proposition : h in mh_v(3 , 3 ; 4 ; n - abs(a ) ) ] and @xmath235 it follows that @xmath87 , and from @xmath77 it follows that @xmath88 .
therefore , the following two cases are possible : _ case 1 : @xmath236 .
_ let @xmath57 be an independent set of vertices of @xmath17 , @xmath237 and @xmath59 . from proposition [ proposition : h in mh_v(3 , 3 ; 4 ; n - abs(a ) ) ]
it follows that @xmath238 and from proposition [ proposition : mh_v(3 , 3 ; 4 ; n ) subseteq ml(n ; 1 ) ] we obtain @xmath239 .
since @xmath17 is maximal @xmath50-free graph , it follows that @xmath240 we execute algorithm [ algorithm : finding ml_(max)(n ; p ; s ) ] ( @xmath241 ) . according to ( [ equation : h in ml_(+k_3)(15 ; 1 ; s ) , s leq 4 ] ) , @xmath126 , and therefore , from lemma [ lemma : finding max supgraphs ]
we obtain that , after the execution of step 2.3 of the algorithm , @xmath155 . by ( [ equation : g arrowsv ( 2 , 3 , 3 ) rightarrow chi(g ) geq 6 ] ) , @xmath155 after step 4 . on the other hand , in the first part of the proof of the main theorem we found all 2 597 graphs which are in @xmath124 after step 4 .
none of these graphs belongs to @xmath242 , which is a contradiction .
_ case 2 : @xmath243 .
_ in the same way as in the proof of case 1 , we see that after the execution of step 4 of algorithm [ algorithm : finding ml_(max)(n ; p ; s ) ] @xmath155 . in the second part of the proof of the main theorem we found all 31 graphs in @xmath124 .
none of these graphs belongs to @xmath242 , which is a contradiction .
+ all vertex transitive graphs with up to 31 are known and can be found in @xcite . with the help of a computer we check which of these graphs belong to @xmath242 . in this way
, we find one 24-vertex graph , one 28-vertex graph and 6 30-vertex graphs in @xmath242 .
the only 24-vertex transitive graph in @xmath242 is given on figure [ figure : fv233_trans_24 ] .
it does not have proper subgraphs in @xmath242 , but by adding edges to this graph we obtain 18 more graphs in @xmath245 , two of which are maximal @xmath50-free graphs . | the set of the graphs which do not contain the complete graph on @xmath0 vertices @xmath1 and have the property that in every coloring of their edges in two colors there exist a monochromatic triangle is denoted by @xmath2 .
the edge folkman numbers @xmath3 are considered .
folkman proved in 1970 that @xmath4 exists if and only if @xmath5 . from the ramsey number
@xmath6 it becomes clear that @xmath7 if @xmath8 .
it is also known that @xmath9 and @xmath10 .
the upper bounds on the number @xmath11 which follow from the construction of folkman and from the constructions of some other authors are not good . in 1975
erdos posed the problem to prove the inequality @xmath12 .
this erdos problem was solved by spencer in 1978 .
the last upper bound on @xmath11 was obtained in 2012 by lange , radziszowski and xu , who proved that @xmath13 .
the best lower bound on this number is 19 and was obtained 10 years ago by radziszowski and xu . in this paper
, we improve this result by proving @xmath14 . at the end of the paper , we improve the known bounds on the vertex folkman number @xmath15 by proving @xmath16 . _ keywords : _ folkman number , clique number , independence number , chromatic number |
the energy produced by a biological system or by an artificial device must often be converted into a more stable form @xcite . the thermodynamic efficiency and the quantum yield of this process is of prime importance for the overall performance of the energy transducer .
this transducer consumes the energy of input particles , which move energetically downhill or just disappear in the process , and transfer this energy to another kind of particles moving energetically uphill . here
, the quantum yield ( @xmath1 ) is defined as the number of particles at the output of the energy transducer divided by the number of input particles .
the efficiency of such device can be higher when the quantum yield is more than one , i.e. , when a single input particle creates many output carriers .
the generation of two or more electron - hole pairs ( excitons ) by a single high - energy photon @xcite was observed in semiconductor nanocrystals @xcite .
a similar situation takes place in the @xmath2 complex embedded into the inner mitochondrial membrane as well as in the related complex @xmath0 , which mediates the electron transfer between the photosystem ii ( ps ii ) and the photosystem i ( ps i ) in the thylakoid membranes of plants and cyanobacteria @xcite .
according to the generally accepted q - cycle mechanism @xcite , the transfer of _ two _ electrons from a plastoquinol molecule pqh@xmath3 to plastocyanin ( in @xmath0 complexes ) is accompanied by an energetically - uphill translocation of _ four _ protons from the negative ( n ) to the positive ( p ) side of the membrane , resulting in a quantum yield @xmath4 .
we note that within the standard redox loop mechanism ( see @xcite and references therein ) , only two protons are transferred in parallel with the transfer of two electrons , implying that the quantum yield is equal to one .
hereafter , we primarily concentrate on the @xmath0 complex as a biological counterpart of our artificial system . despite numerous studies @xcite , the physical mechanism of the q - cycle in @xmath2 and @xmath0 complexes is not completely understood . in this work
we analyze a simple model ( see fig .
1 ) mimicking the main features of the q - cycle in the @xmath0 complex in the regime of ferredoxin - dependent cyclic electron flow @xcite . in this regime , electrons cycle between the ps i and the @xmath0 complex , which are electronically connected by a pool of ferredoxin molecules ( on the n - side of the membrane ) and by a pool of plastocyanin molecules on the lumenal ( p ) side of the membrane .
we treat these two pools as a source ( s ) and drain ( d ) electron reservoirs coupled to the electron - binding sites a and b , respectively . besides the sites a and b , the membrane - embedded central complex is comprised of sites l and h , which correspond to hemes @xmath5 and @xmath6 of the complex @xmath0 . the sites a and h
are assumed to be electronically decoupled as well as the sites b and l. a mobile shuttle @xmath7 ( an analog of a plastoquinone molecule ) diffuses inside the membrane , between the sites a and h ( on the p - side ) and the sites b and l ( on the n - side ) . as its biological counterpart , the shuttle has two electron sites and two proton - binding sites . at its n - position ,
the shuttle takes one electron from the site a and another electron from the site h and transfers these electrons to the sites b and l. at the n - side , the shuttle also accepts up to two protons from the stromal ( electrically - negative ) proton reservoir and donate these protons to the lumenal proton reservoir at the p - side of the membrane .
when the fully populated q - molecule arrives at the p - side , one electron from the shuttle goes strictly energetically downhill , to the site b , whereas another one returns to the l - h chain to be loaded again on the shuttle .
the origin of this bifurcated reaction @xcite , which occurs at the p - side catalytic center , remains unknown . here ,
we explore physicochemical conditions wherein our artificial complex is able to translocate twice as protons as the number of electrons transferred energetically downhill , from the source s to the drain d. protons are translocated energetically uphill , from the n- to p - side of the membrane .
we aim at the explanation of the q - cycle operation in this artificial complex .
we examine a wide range of parameters allowing the efficient performance of the q - cycle scheme .
the functional principles of the q - cycle in artificial systems can provide a better understanding of the q - cycle mechanism in the natural @xmath2 and @xmath0 complexes .
to simplify the problem , we divide the whole system of six electron and two proton - binding sites into four weakly - interacting subsystems : ( i ) the lh subsystem consisting of l and h sites ( with energies @xmath8 ) ; ( ii ) the shuttle q having two electron sites @xmath9 ( with energies @xmath10 ) and two proton sites @xmath11 ( with energies @xmath12 ) , ( iii ) the site a ( with energy @xmath13 ) , and ( iv ) the site b ( with energy @xmath14 ) . the lh - subsystem is characterized by 4 microscopic states starting with the empty ( vacuum ) state , @xmath15 and ending with the doubly - occupied system : @xmath16 the electron and proton populations of the shuttle are described by 16 states , where @xmath17 is the vacuum state and @xmath18 is the state of the completely - loaded shuttle . here , we use the notation @xmath19 for an empty ( occupied ) site @xmath20 , where @xmath21 for the electron sites on the shuttle , and @xmath22 for the proton sites .
the average populations of the sites a and b are denoted by @xmath23 and @xmath24 . here
@xmath25 means double - averaging over an initial wave function @xmath26 and over a thermal distribution @xmath27 of reservoirs and environment characterized by the common temperature @xmath28 .
we take into account strong coulomb interactions between sites from the same subsystem . a coulomb repulsion between electrons located on the sites l and h ( lh - subsystem )
is characterized by the energy @xmath29 , whereas for the electrons and protons on the shuttle ( q - subsystem ) we introduce the following parameters : @xmath30 ( coulomb repulsion between two electrons occupying the shuttle sites @xmath31 ) ; @xmath32 ( electrostatic repulsion between two protons located on the sites @xmath11 ) ; and @xmath33 ( electron - proton coulomb attraction on the shuttle ) . here the indices @xmath34 and @xmath35 run over the electron and proton - binding sites of the shuttle , respectively .
electrons can tunnel between the l and h sites as well as between sites belonging to different subsystems .
these tunnelings are described by relatively small tunneling amplitudes : @xmath36 ( l - h tunneling ) ; @xmath37 ( tunneling between the h site and the @xmath31 sites ; @xmath38 ( for electron transitions between the @xmath31 and l sites .
the amplitudes @xmath39 and @xmath40 describe the tunneling between the bridge sites a , b and the electron sites @xmath31 on the shuttle .
the amplitudes @xmath41 and @xmath42 depend on the position @xmath43 of the mobile molecule q. the proton rates @xmath44 ( for transitions between the sites @xmath45 and the n - side proton reservoir ) and @xmath46 ( for transitions between the proton sites on the shuttle and the p - reservoir ) are also functions of @xmath43
. the time evolution of @xmath47 is determined by an overdamped langevin equation with terms describing walls of the membrane and a potential barrier , which impedes the charged shuttle to cross the intermembrane space @xcite .
the walls of the membrane are located at @xmath48 ( n - side ) and at @xmath49 ( p - side ) , so that the width of the membrane is equal to @xmath50 . using the formalism outlined in ref .
@xcite and based on marcus transitions rates @xcite , we derive and numerically solve a set of master equations ( see an appendix ) for an electron distribution @xmath51 over states of the lh - subsystem coupled to a system of equations for the probabilities @xmath52 to find the q - subsystem in the state @xmath53 .
these equations are complemented by equations for the populations @xmath54 of the a and b - sites and by an overdamped langevin equation for the shuttle position @xmath47 .
it is known @xcite that charged ions , such as mg@xmath55 and cl@xmath56 , can easily cross a thylakoid membrane , which results in the equilibration of electrical potentials on both sides of a membrane . therefore
, the difference between the proton electrochemical potential @xmath57 of the p - side , and the potential @xmath58 of the n - side of the membrane @xcite is mainly determined by the proton concentration gradient @xmath59 : @xmath60 where @xmath28 is the temperature of the reservoirs , and @xmath61 k is the room temperature
. it should be noted , however , that a surface potential @xmath62 , which is positive on the n - side , @xmath63 , and negative , @xmath64 , on the p - side of the membrane , was calculated for the @xmath0 complex of _ m. laminosus _
@xcite with @xmath65 and @xmath66 ( the boltzmann constant @xmath67 ) .
this model assumes that there is a similar transmembrane potential ( see fig . 1 ) ,
@xmath68 with @xmath69 mev and @xmath70 mev , which correspond to the above - mentioned values of @xmath71 and @xmath72 at room temperature .
all energies are measured in mev . in the presence of the surface potential ,
the energy levels of electrons and protons on the shuttle are shifted from their initial values @xmath73 and @xmath74 depending on the shuttle s position @xmath43 : @xmath75 with @xmath76 and @xmath77 correspondingly , the energy levels of the electron sites a and h , located near the n - side , are shifted down from their initial values : @xmath78 ; whereas the energies of the sites b and l , located near the p - side , are shifted up : @xmath79
instead of searching over a multidimensional space of system parameters , we consider a reasonable sequence of events , which provides an optimal performance of the energy transducer .
in particular , we analyze a scenario where an electron transferred from the high - energy source reservoir @xmath80 along the chain : @xmath81 to the drain @xmath82 performs the main energetic function in the transfer of two protons from the n to the p - side of the membrane .
another electron traveling on the shuttle and recycled by the lh - system along the chain @xmath83 plays a more passive role of a passenger , which is necessary to compensate a shuttle charge . according to the marcus formula @xcite
( see also an appendix ) , @xmath84 , \label{marcusrate}\ ] ] the rate @xmath85 for an electron transition from the site @xmath86 , with an energy @xmath87 , to the site @xmath88 , with an energy @xmath89 , has a maximum at @xmath90 here @xmath91 is the tunneling amplitude between the sites @xmath86 and @xmath88 , @xmath92 is the corresponding reorganization energy , which is due to electron coupling to an environment with temperature @xmath28 .
the shuttle can accept protons from the n - reservoir provided that the electrochemical potential of the n - side , @xmath58 , is higher than the proton energy level on the shuttle .
protons move from the shuttle q to the p - side reservoir if the energy of the q - proton exceeds the p - side potential @xmath57 .
we start with a situation when the empty shuttle ( quinone ) is near the n - side catalytic center , @xmath93 , and the @xmath94-system ( analog of cytochrome @xmath95 in the @xmath0 complex ) is preloaded with one electron located presumably at the site @xmath96 , which has a lower energy than the site @xmath97 : @xmath98 . the site a is also occupied with an electron taken from the electron source s. we have the following sequence of electron ( @xmath99 ) and proton ( @xmath100 ) transfer from and to the shuttle located near the n - side of the membrane : a ) @xmath101 .
b ) @xmath102 .
c ) @xmath103 .
d ) @xmath104 .
here we have written relations between energy levels of electrons and protons which make possible these transfers .
the shuttle loaded with two electrons and two protons travels to the p - side of the membrane , where the following sequence of electron and proton transitions occurs : e ) @xmath105 .
f ) @xmath106 .
g ) @xmath107 .
h ) @xmath108 . finally , an electron tunnels from the l to the h site : i ) @xmath109 .
the empty shuttle diffuses to the n - side of the membrane and the process repeats .
we expect that two protons will be translocated from the n - side to the p - side of the membrane per one electron transferred from the source to the drain electron reservoir with a quantum yield @xmath110 here we assume that , as in the case of the quinone molecule q @xcite , the shuttle populated with one electron ( after step 1 ) does not bind a proton but accepts another electron ( step 2 ) .
the doubly - reduced quinol is known to have a much stronger ability for binding two protons ( see steps 3 and 4 ) . at the p - side of the membrane the process presumably evolves in the opposite direction when the transfer of one electron from q to the site b is accompanied by the unloading of two protons . in the absence of an attraction to two positive charges ,
the energy of the electron remaining on the shuttle goes up ; thus , allowing its tunneling to the l - site . here
, an electron recycled by the lh - system plays a passive role of a shuttle s passenger " since its transitions to and from the q - molecule are not immediately accompanied by a proton transfer .
transitions of another electron , which is loaded to the shuttle from the source s ( via the site a ) and unloaded to the drain d ( via the site b ) , are more closely coupled to the energetically - uphill proton translocation .
it follows , from the relations a),h),i ) in this section , that the recycling of one electron by the lh - chain , h @xmath111 q @xmath111 l @xmath111 h , which lies at the heart of the q - cycle , takes place if the difference of surface potentials , @xmath112 is of the order of the total reorganization energy along the recycling path : @xmath113 we see from the relations d ) and g ) in this section that the energetically - uphill proton transfer from n- to the p - side of the membrane is possible if the original energy of the proton on the shuttle , @xmath114 , obeys the following inequality : @xmath115 which can be true only for a sufficiently strong attraction potential , @xmath116 between electrons and protons on the shuttle @xmath117 the relations a),b),e),h),i ) in this section allow to estimate the original energies of the electron - binding sites counted , e.g , , from the level @xmath118 : @xmath119 we assume that the potentials of the electron source , @xmath120 , and electron drain , @xmath121 , are of the order of the energies of the a and b sites , respectively : @xmath122 . taking into account eqs .
( [ energy0 ] ) we obtain a relation for the source - drain energy drop , @xmath123 with eqs .
( [ deltav],[uep ] ) we obtain the following requirement for the energy difference between the source and drain electron reservoirs : @xmath124 where the combined reorganization energy , @xmath125 accumulates all losses along both electron transport chains : @xmath126 and @xmath127 . the thermodynamic efficiency @xmath128 of proton translocation can be defined as @xmath129 where @xmath130 is the number of protons translocated from the n- to the p - side of the membrane , and @xmath131 is the number of electrons transferred from the source s to the electron drain d. the efficiency @xmath128 is proportional to the quantum yield @xmath132 it follows from eq .
( [ musd ] ) that , within the passenger " scenario , the efficiency @xmath133 of the electron - to - proton energy conversion can be estimated as @xmath134 this means that for a high electrochemical proton gradient , @xmath135 the efficiency @xmath128 has the maximum : @xmath136 thus , in the ideal case , when @xmath4 , the thermodynamic efficiency can reach the perfect mark : @xmath137 when almost all electron energy is converted to the transmembrane proton - motive force .
in the model presented here an electron transport chain begins at the source reservoir s , corresponding to a pool of ferredoxin ( fd ) molecules , which carry electrons from photosystem i to the @xmath0 complex ( see fig . 1 ) .
the electron drain d is related to the high - potential chain of the @xmath0 complex comprised of the ion - sulfur protein ( isp ) , cytochrome @xmath138 , and soluble plastocyanin ( pc ) molecules . taking into account a redox potential of ferredoxin , @xmath139 v , and the fact that a redox potential of the isp/@xmath138/pc chain , @xmath140 is in the range from 0.3 to 0.45 v @xcite , we estimate that the total energy drop between the source and the drain , @xmath141 , takes values from 710 mev up to 860 mev .
we also assume that , as for the @xmath0 complex @xcite , the surface voltage gradient , @xmath142 mev , has been applied to the membrane , with a positive potential , @xmath143 mev , at the n - side and a negative potential , @xmath144 mev , at the p - side of the membrane .
the system reaches its optimal performance when @xmath145 mev .
hereafter , we assume that @xmath146 mev .
we use the following values for the electron transfer rates , @xmath147ev , proton transition rates , @xmath148ev , and for peak values of the electron tunneling amplitudes , @xmath149 mev , @xmath150 mev .
the coefficients @xmath151 and @xmath152 determine the rates of the electron transfer between site a and source s and between site b and drain d , respectively ( see the definitions of @xmath153 and @xmath152 in the appendix ) .
the rates @xmath154 and @xmath155 , describing proton transitions between the n - side reservoir and the shuttle as well as between the shuttle and the p - side of the membrane , respectively , are defined in the appendix and in the subsection ie of the supporting information . for the diffusion coefficient , @xmath156
, we have the value , @xmath157 m@xmath158/s , which is close to experimental data for plastoquinone molecules in a lipid membrane @xcite .
hereafter , we assume that the reorganization energies , corresponding to h - q and l - q transitions , are equal : @xmath159 , and the same relation is true for the a - q and b - q transitions : @xmath160 to satisfy eq .
( [ deltav ] ) we start with small values of the reorganization energies along the recycling pathway , @xmath161 mev , and @xmath162 mev , which add up to the value of the surface gradient , @xmath163 mev . here
, we also have @xmath164 mev . for the driving force , @xmath165 mev , related to the electron energy drop in the @xmath0 complex ,
the system is able to translocate almost two protons , @xmath166 , against the electrochemical gradient , @xmath167 mev , per each electron transferred from the source to the drain .
this proton electrochemical gradient corresponds to the value @xmath168 at room temperature , @xmath169 k. the thermodynamic efficiency of the energetically - uphill proton translocation is about 33 % ( @xmath170 ) . a proton translocation process is shown in fig . 2 , where we plot the time dependence of the total electron , @xmath171 , and proton , @xmath172 , populations of the shuttle ( fig .
2b ) , together with the time - evolving position of the shuttle @xmath47 ( fig .
here , we also show ( see fig .
2c ) the populations of the l - site , @xmath173 , and h - site , @xmath174 , as well as the average number of electrons @xmath175 transferred from the source to the drain , as well as the average number of protons @xmath176 translocated from the n - side to the p - side proton reservoir . the brackets @xmath177 are dropped in figs . 2c and 2d for the notations of the populations , and throughout the paper , except in the appendix .
data for fig .
2 are calculated at higher values of the reorganization energies : @xmath178 mev , @xmath179 mev , and @xmath180 mev .
we assume that @xmath145 mev , @xmath163 mev , @xmath165 mev , and @xmath167 mev .
it can be seen from fig . 2 that during 30 @xmath181s the shuttle performs about 8 trips from the n - side ( @xmath93 ) to the p - side ( @xmath182 nm ) of the membrane and back , translocating in the process about 7 electrons ( @xmath183 = 6.7 ) and 12 protons ( @xmath184 ) , with quantum yield @xmath185 , and thermodynamic efficiency @xmath186 .
at the n - side ( see fig .
2b ) the shuttle accepts an electron from the initially populated site h and another electron from the source s ( via site a ) as well as two protons from the n - side proton reservoir .
we note ( see fig .
2c ) that site h is not completely depopulated .
this means that there are events when both electrons occupying the shuttle arrive from the site a and the source s , shorting out the q - cycle pathway .
this leakage process increases the number of electrons transferred from the source to the drain ( with the same number of protons ) , thus decreasing the quantum yield @xmath1 .
here we do not impose any additional restrictions , except a proper choice of energy levels , which are close to values given by eqs .
( [ eq0],[energy0],[musd ] ) , with @xmath187 and @xmath188 at the n - side catalytic center ( all energies are measured in mev ) . for the p - side center we use the following energies : @xmath189 for electrons and protons on the shuttle , @xmath190 for the high - potential redox chain , and @xmath191 mev for the recycling pathway .
this choice of energy levels makes the h - to - q electron transition ( at the n - side ) much easier than the a - to - q electron transfer , since @xmath192 , whereas @xmath193 .
moreover , the s - to - a electron transition is also hampered since the energy level of the a - site , @xmath13 , is higher than the potential of the source , @xmath194 on arrival at the p - side of the membrane the shuttle donates an electron to the b - site and , finally , to the drain d. two protons move to the p - side proton reservoir ( figs .
2b and 2d ) .
it is evident from fig .
2c that another electron from the shuttle q goes to the l - site ( see the small spike at the bottom of fig .
this electron is rapidly transferred to the h - site , and the empty shuttle returns to the n - side .
figures 2b,2c,2d illustrate the bifurcated reaction which occurs at the p - side . here
, one electron from the shuttle q goes to the high - potential ( and low - energy ) chain , @xmath195 while another electron ( a passenger ) returns to the lh - system for recycling ( along the pathway @xmath196 ) .
no additional gate mechanisms are required for this reaction .
an escape of the first electron from q to b , followed by the transition of two protons to the p - reservoir , increases the energy of the remaining electron to the level @xmath197 mev , which is of order of the l - site energy , @xmath191 mev , but is much higher than the energy of the b - site , @xmath198 mev .
furthermore , the site b is probably occupied with an electron taken from the drain since @xmath199 these two factors strongly suppress the leakage of the second electron from the shuttle to the high - potential chain . in figure 3
we show the numbers of protons , @xmath130 , and electrons , @xmath131 , transferred across the membrane , as well as the quantum yield , @xmath1 , and the power - conversion efficiency , @xmath128 , as functions of the proton electrochemical gradient , @xmath200 ( measured in mev ) .
the graphs are plotted for two sets of reorganization energies : ( i ) @xmath201 mev , @xmath179 mev , @xmath202 mev ( blue curves ) ( ii ) @xmath203 mev , @xmath204 mev , @xmath205 mev ( green dashed curves ) .
other parameters , such as @xmath206 ( in mev ) , are the same as in fig . 2 .
we numerically calculate the output of the system ( @xmath207 etc . ) at the end of the stochastic trajectory @xmath47 ( with the duration @xmath208
@xmath181s ) and average results over 10 trajectories . for each value of the shuttle s position
@xmath47 we solve a set of master equations ( [ rmn],[rhomunu],[na ] ) , which have been averaged over electron and proton reservoirs as well as over fluctuations of the environment coupled to the electronic degrees of freedom .
this can be done since the electron and proton transitions are much faster than the mechanical motion of the shuttle .
it follows from fig . 3 that more protons , @xmath209 , and electrons , @xmath210 , are transferred across the membrane at lower proton gradients , @xmath211 mev with a higher quantum yield , @xmath212 .
however , the thermodynamic efficiency is higher , @xmath213 , at larger proton gradients , @xmath214 mev , where @xmath215 and @xmath216 .
these numbers are for the set ( i ) , with smaller values of the reorganization energies ( see blue continuous curves in fig . 3 ) .
a stronger electron - environment interaction , described by the set ( ii ) of reorganization energies , significantly reduces an energetically - uphill proton flow with almost no impact on the electron current ( see green dashed curves in fig .
in this case the quantum yield drops to almost one , which means that the recycling pathway ( via the lh - system ) is practically closed .
figure 4 demonstrates the performance of the system as a function of the surface potential gradient , @xmath217 at a fixed difference between the n - side and p - side potentials , @xmath218 mev .
here we choose two sets of system parameters .
the first set ( see the blue continuous curves in fig . 4 ) , with @xmath219 and @xmath220 mev , was considered before .
the second set ( green dashed curves in fig .
4 ) is characterized by a higher electron - proton attraction potential , @xmath221 mev , and a higher source - drain difference , @xmath222 mev . at these parameters
the system can translocate protons against the electrochemical difference @xmath223 mev , which is related to the proton concentration gradient @xmath224 at room temperature .
the reorganization energies correspond to the set ( i ) described before .
it follows from fig . 4 that the system performs very well in a wide range of surface voltage gradients , from @xmath225 mev up to @xmath226 mev , translocating more than 40 protons and 20 electrons ( in 100 @xmath181s ) .
these numbers are almost the same for both sets of parameters ( see figs . 4a and 4b ) .
the quantum yield @xmath1 monotonically goes down , from @xmath227 at @xmath228 mev to @xmath229 at @xmath226 mev , no matter what set is chosen .
the power - conversion efficiency @xmath128 also decreases with increasing @xmath230 .
however , @xmath133 is higher for the second set of parameters .
for example , at @xmath163 mev , the efficiency is about 31% for the first set , and @xmath231 for the second set
( with @xmath232 ) where both electron and proton electrochemical gradients are higher .
this trend is consistent with eq .
( [ eqeta ] ) derived for the passenger scenario .
we have theoretically examined a model of the protonmotive force generation by the q - cycle mechanism mimicking the operation of the @xmath0 complex in the thylakoid membranes of plants and cyanobacteria .
we concentrate on a simulation of the regime of ferredoxin - dependent cyclic electron flow , where the @xmath0 complex translocates up to two protons ( a quantum yield @xmath233 ) across the membrane per one electron transferred from the electron source ( ferredoxin pool ) to the electron drain ( pool of plastocyanin molecules ) .
this model includes two electron and two proton sites on the shuttle q ( an analog of a plastoquinone molecule ) , diffusing inside the membrane , as well as two electron sites , a and b , connecting the electron sites on the shuttle to the source and drain reservoirs .
the recycling of an electron by the cytochrome @xmath95 , which forms the basis of the q - cycle , can be described by adding two electron sites l and h corresponding to hemes @xmath5 and @xmath6 of the @xmath0 complex .
we have derived and numerically solved a set of master equations for the populations of the electron and proton - binding sites together with a langevin equation for the position of the shuttle . within a reasonable scenario and in the presence of the surface potential ,
we have determined the conditions which are necessary for the efficient translocation of protons across the membrane .
we have found that the system is able to transfer , on average , about 1.8 protons per one electron ( @xmath234 ) with a thermodynamic efficiency of the order of 32% against the transmembrane proton gradient @xmath167 mev at the source - drain difference of electron potentials @xmath235 mev .
these values of the electron and proton gradients are closely related to experimental values for the @xmath0 complex .
no conformational gating is necessary for the bifurcation of the electron transfer reaction at the p - side catalytic center , where one electron goes to the drain and another electron returns to the lh - chain , to be loaded on the shuttle again .
we have studied the performance of the model as a function of the proton electrochemical gradient and the surface potential .
it is shown that the system demonstrates even better results , with a quantum yield of the order of 1.9 and an efficiency of the order of 47% , when both the source - drain difference and the proton gradient are higher .
* acknowledgements .
* fn acknowledges partial support from the laboratory of physical sciences , national security agency , army research office , darpa , air force office of scientific research , national science foundation grant no .
0726909 , jsps - rfbr contract no .
09 - 02 - 92114 , grant - in - aid for scientific research ( s ) , mext kakenhi on quantum cybernetics , and jsps via its first program .
here we briefly outline a set of master equations describing the process of electron - driven proton translocation across a membrane .
as we mentioned before , the total system is characterized by the average populations of the a and b - sites , @xmath236 , as well as by four states of the lh - subsystem with electron distributions @xmath51 ( @xmath237 ) and 16 states of the q - subsystem ( electrons and protons on the shuttle ) with distributions @xmath52 ( @xmath238 ) .
the time evolution of the lh - system is governed by the equation @xmath239 with the following relaxation matrix @xmath240 here the rate @xmath241\end{aligned}\ ] ] describes the l - to - h electron transitions .
hereafter , @xmath242 refer to the creation / annihilation operators for an electron on the site @xmath20 . for protons on the site @xmath243 the creation / annihilation operators
are denoted by @xmath244 the rate @xmath245 ( and a similar rate @xmath246 ) is related to the electron transfer between the l ( or h ) sites and the sites @xmath247 on the shuttle , @xmath248\ , \langle \rho_{\mu}\rangle + \right .
\nonumber\\ \left .
|\langle n|a_l|m\rangle |^2\ , \exp \left[\ , -\ ,
\frac{(\omega_{mn } + \omega_{\mu\nu } + \lambda_{lq})^2}{4\lambda_{lh}t}\,\right ] \ , \langle \rho_{\nu}\rangle \right\},\end{aligned}\ ] ] where @xmath249 are operators of the electron - binding sites on the shuttle , @xmath250 are frequencies of the lh - system ( @xmath251 ) , and @xmath252 is the frequency spectrum of the coupled electron - proton states on the shuttle q. for the distributions @xmath52 of the 16 states of the q - system we derive the following equation @xmath253 where @xmath254 components of this relaxation matrix can be written as @xmath255 \langle 1 - n_a\rangle \right .
\nonumber\\ \left .
+ |\langle \nu|a_1+a_2|\mu\rangle |^2 \ , \exp\left[\ , -\ , \frac { ( \omega_{\mu\nu } - \varepsilon_a + \lambda_{aq})^2}{4 \lambda_{aq}t } \ , \right ] \langle n_a\rangle \right\},\end{aligned}\ ] ] with a similar matrix @xmath256 , and @xmath257 \right . \nonumber\\ \left .
+ \ , |\langle \mu|a_1+a_2|\nu\rangle |^2 \ , \langle r_m\rangle \ , \exp\left[\ , - \ , \frac { ( \omega_{\mu\nu } - \omega_{mn } + \lambda_{aq})^2}{4 \lambda_{lq}t } \ , \right]\ , \right\},\end{aligned}\ ] ] with a matrix @xmath258 , which is similar to @xmath259 .
the proton transitions to and from the shuttle are described by the rate @xmath260 + \nonumber\\ \gamma_n |\langle \nu|a_1 + a_2|\mu\rangle|^2
f_n(\omega_{\mu\nu}),\end{aligned}\ ] ] and by a similar rate @xmath261 . here @xmath262^{-1}\ ] ] is the fermi distribution of the protons in @xmath263reservoir ( @xmath264 ) .
the average population of the a - site is governed by the equation @xmath265 + \nonumber\\ |\delta_{aq}|^2 \sqrt{\frac{\pi}{\lambda_{aq}t } } \ ; \sum_{\mu\nu } |\langle \mu|a_1+a_2|\nu\rangle |^2 \times \nonumber\\ \left\ { \langle 1 - n_a \rangle \exp\left [ -\frac{(\omega_{\mu\nu } + \varepsilon_a + \lambda_{aq})^2}{4\lambda_{aq}t}\right ] \langle \rho_{\nu}\rangle \right .
\nonumber\\ \left .
- \langle n_a \rangle \exp\left [ -\frac{(\omega_{\mu\nu } + \varepsilon_a - \lambda_{aq})^2}{4\lambda_{aq}t}\right ] \langle \rho_{\mu}\rangle \right\},\end{aligned}\ ] ] where a coefficient @xmath153 ( or @xmath152 ) describes the electron transitions from the a ( or b ) site to the source s ( drain d ) electron reservoir characterized by a fermi distribution with electrochemical potentials @xmath120 or @xmath121 ( @xmath266 ) , @xmath267^{-1}.\ ] ] a similar equation takes place for the population @xmath268 .
we solve the rate equations ( [ rmn],[rhomunu],[na ] ) for both the distributions @xmath269 and for the populations @xmath270 and @xmath271 , together with an overdamped langevin equation for the mechanical motion of the shuttle , @xmath272 where @xmath273 is the drag coefficient , @xmath274 is a gaussian fluctuation source with zero mean value , @xmath275 , and with a correlator @xmath276 the potential @xmath277 confines the shuttle between the membrane walls , and the potential @xmath278 prevents the charged molecule q from crossing the lipid core of the membrane ( for details see ref .
we note that the tunneling amplitudes @xmath41 and the proton rate @xmath154 depend on the distance between the shuttle ( with a coordinate @xmath43 ) and the n - side catalytic center located at @xmath279 whereas the tunneling amplitudes @xmath42 and the proton rate @xmath155 depend on the distance between the shuttle and the p - side catalytic center located at @xmath49 . of the shuttle diffusing between the membrane walls located at @xmath280 nm ; ( b)-(d ) a sequence of electron and proton transfer reactions at the following set of parameters : @xmath281 mev , and @xmath163 mev .
the total proton population @xmath282 of the shuttle [ blue continuous line in ( b ) ] almost coincides with the electron population @xmath283 marked by the dashed green line .
the bifurcated electron transfer reaction takes place at the p - side of the membrane ( at @xmath284 nm ) where one electron moves from q to the site b and the drain d [ see , e.g. , a step down for @xmath283 in ( b ) and a step up for @xmath131 in ( d ) at the moment @xmath285s ] . at almost the same time another electron moves to the site l [ see a barely visible dashed red line in ( c ) ] and rapidly proceeds to the site h [ blue continuous spike in ( c ) ] .
two protons are unloaded from the shuttle to the p - side of the membrane as follows from the step for @xmath130 [ blue continuous curve in ( d ) ] .
it can be seen from ( d ) that the number of protons @xmath130 translocated to the p - side ( blue continuous line ) is nearly twice as large as the number of electrons @xmath131 moved to the drain ( dashed green line).,width=755,height=453 ] transferred to the p - side of the membrane ; ( b ) number of electrons @xmath131 moved to the drain ; ( c ) their ratio ( quantum yield @xmath1 ) ; ( d ) the thermodynamic efficiency @xmath128 as functions of the transmembrane proton gradient @xmath200 ( measured in mev ) for different couplings to the environment ( in mev ) : weaker couplings ( i ) @xmath286 ( blue continuous curves ) ; and stronger couplings ( ii ) @xmath287 ( green dashed curves ) .
other parameters are the same as in fig . 2 .
the data are averaged over 10 stochastic realizations .
each realization lasts for 100 @xmath181s .
these graphs demonstrate the ability of the system to translocate protons against the gradient up to 200 mev with an efficiency @xmath133 up to 40%.,width=755,height=453 ] ( a ) , the number of electrons moved to the drain , @xmath131 ( b ) , the quantum yield @xmath288 and the efficiency @xmath128 ( d ) , on the surface potential gradient @xmath289 across the membrane
. two kinds of curves correspond to two sets of energy values : ( 1 ) @xmath290 mev ( blue continuous curves ) , and ( 2 ) @xmath291 mev ( green dashed curves ) and for @xmath292 mev .
it can be seen from this figure that the system is operational in the range of the surface potentials from @xmath293 mev up to @xmath294 mev . in agreement with eq .
( [ eqeta ] ) the complex works more efficiently ( with @xmath295 and @xmath227 ) for the second set of parameters where both electron and proton gradients are higher . , | the q - cycle mechanism plays an important role in the conversion of the redox energy into the energy of the proton electrochemical gradient across the biomembrane .
the bifurcated electron transfer reaction , which is built into this mechanism , recycles one electron , thus , allowing to translocate two protons per one electron moving to the high - potential redox chain .
we study a kinetic model of the q - cycle mechanism in an artificial system which mimics the @xmath0 complex of plants and cyanobacteria in the regime of ferredoxin - dependent cyclic electron flow . using methods of condensed matter physics
, we derive a set of master equations and describe a time sequence of electron and proton transfer reactions in the complex .
we find energetic conditions when the bifurcation of the electron pathways at the positive side of the membrane occurs naturally , without any additional gates . for reasonable parameter values ,
we show that this system is able to translocate more than 1.8 protons , on average , per one electron , with a thermodynamic efficiency of the order of 32% or higher . |
the dynamical system of the dts-@xmath0is described by the magnetic induction and navier - stokes equations for an incompressible fluid .
equations are nondimensionalized by taking as length - scale the radius of the outer sphere @xmath16 and as characteristic time @xmath17 is the angular velocity difference between the inner and outer sphere . without loss of generality , @xmath18 , while @xmath19 can be either positive ( corotation ) or negative ( counter - rotation ) .
after choosing characteristic velocity @xmath20 , the nondimensionalized magnetic induction equation reads : @xmath21 where @xmath22 is the incompressible velocity field , @xmath23 is the magnetic field , @xmath11 is the ( signed ) rossby number @xmath24 , and @xmath25 is the magnetic ekman number @xmath26 , with @xmath27 the magnetic diffusivity ) .
the navier - stokes equation , in the frame rotating with the outer sphere , is nondimensionalized as @xmath28 where @xmath29 is the ekman number @xmath30 with @xmath31 the viscosity ) , @xmath32 is the elsasser number at the outer sphere @xmath33 , with @xmath34 the vertical magnetic field at @xmath16 , @xmath35 the magnetic constant , and @xmath36 the density of the medium ) , and @xmath37 a reduced pressure absorbing all potential forces .
the magnetic field is represented in the code as an alfvn velocity @xmath38 , and the nondimensional magnetic field is this @xmath39 .
the values of these parameters in the dts-@xmath0are given in tab .
[ tab : dtsnondim ] .
the simulations presented here were designed to match the @xmath11 , @xmath25 , and @xmath40 values of the machine as closely as possible .
the @xmath29 is treated as a nuisance parameter ; how and why we treat it that way is handled below .
.definitions of useful nondimensional parameters for dts-@xmath0 , and their values / ranges at various outer sphere rotation rates .
the two values given for @xmath11 are , respectively , the maximum and minimum values attainable for a given @xmath4 .
the values given for @xmath32 and @xmath41 are , respectively , the elsasser number at the outer sphere , where the applied field is weakest , and at the inner sphere , where the applied field is strongest . the @xmath25 and @xmath40 of the simulations are chosen to match those of the machine .
the @xmath29 of the simulations are set to 10 ^ -6^. [ tab : dtsnondim ] [ cols=">,^,^,^,^,^ " , ]
because the flows discussed in section [ sec : pca ] seem to primarily be saturated instabilities , they re amenable to a principal component analysis ( pca ) , wherein the flow is decomposed into individual spatial modes , each with its own timeseries @xcite .
the modes are chosen to maximize the cross correlation in time between different measurements ( here the kinetic and magnetic spectra at each radial gridpoint ( _ i.e. _ the @xmath42 and @xmath43 of eqn .
[ eqn : specdef ] ) .
the analysis is carried in the frame of the inner sphere according to here , @xmath45 is a matrix representing spatio - temporal data ; @xmath46 is a matrix of so - called right eigenvectors , each representing a single spatial configuration that evolves together ( @xmath47 is the complex conjugate and transpose of the matrix @xmath46 ) ; @xmath48 is a matrix of so - called left eigenvectors , each representing the collective time evolution of its corresponding right eigenvector ; and @xmath49 is a diagonal matrix of singular values , each representing the relative strengths of the modes .
the matrix @xmath45 is computed by removing the means of the magnetic and velocity fluctuations @xmath50 , and the singular value decomposition ( svd ) is carried out using ` numpy ` s linear algebra module @xcite .
the time series of fig .
[ fig : surfpca]c and d are generated by converting individual principle components @xmath51 back into a collection of snapshots @xmath52 by @xmath53 where @xmath54 is an individual column of @xmath46 , @xmath55 is an individual entry in @xmath48 , and @xmath56 is the singular value of @xmath51 .
g. dumas , _ study of spherical couette flow via 3-d spectral simulations : large and narrow - gap flows and their transitions_. phd thesis , california institute of technology , pasadena , california ( usa ) , 1991 .
k. nakabayashi , z. zheng , and y. tsuchida , `` characteristics of disturbances in the laminar turbulent transition of spherical couette flow .
2 . new disturbances observed for a medium gap , '' _ physics of fluids ( 1994-present ) _ , vol .
14 , no . 11 , pp . 39733982 , 2002 .
n. kleeorin , i. rogachevskii , a. ruzmaikin , a. m. soward , and s. starchenko , `` axisymmetric flow between differentially rotating spheres in a dipole magnetic field , '' _ journal of fluid mechanics _ , vol .
344 , pp . 213244 , 8 1997 .
d. brito , t. alboussire , p. cardin , n. gagnire , d. jault , p. la rizza , j .-
masson , h .- c .
nataf , and d. schmitt , `` zonal shear and super - rotation in a magnetized spherical couette - flow experiment , '' _ phys .
e _ , vol .
83 , p. 066310
, jun 2011 .
nataf , t. alboussire , d. brito , p. cardin , n. gagnire , d. jault , and d. schmitt , `` rapidly rotating spherical couette flow in a dipolar magnetic field : an experimental study of the mean axisymmetric flow , '' _ physics of the earth and planetary interiors _ , vol .
170 , pp . 6072 , sept .
d. schmitt , t. alboussire , d. brito , p. cardin , n. gagnire , d. jault , and h .- c .
nataf , `` rotating spherical couette flow in a dipolar magnetic field : experimental study of magneto - inertial waves , '' _ j. fluid mech _ , vol . 604 , pp . 175197 , 2008 .
g. a. glatzmaier and p. h. roberts , `` a three - dimensional convective dynamo solution with rotating and finitely conducting inner core and mantle , '' _ physics of the earth and planetary interiors _ , vol .
91 , no . 1 - 3 , pp . 63 75 , 1995 .
study of the earth s deep interior . | the derviche tourneur sodium experiment , a spherical couette magnetohydrodynamics experiment with liquid sodium as the medium and a dipole magnetic field imposed from the inner sphere , recently underwent upgrades to its diagnostics to better characterize the flow and induced magnetic fields with global rotation . in tandem with the upgrades , a set of direct numerical simulations
were run with the ` xshells ` code @xcite to give a more complete view of the fluid and magnetic dynamics at various rotation rates of the inner and outer spheres .
these simulations reveal several dynamic regimes , determined by the rossby number . at positive differential rotation
there is a regime of quasigeostrophic flow , with low levels of fluctuations near the outer sphere .
negative differential rotation shows a regime of what appear to be saturated hydrodynamic instabilities at low negative differential rotation , followed by a regime where filamentary structures develop at low latitudes and persist over five to ten differential rotation periods as they drift poleward .
the simulated flows can produce the same measurements as those that the physical experiment can take , with signatures akin to those found in the experiment .
this paper discusses the relation between the internal velocity structures of the flow and their magnetic signatures at the surface .
spherical couette flow , a fluid medium sandwiched between two spheres rotating about a common axis , has a long history in the fluid dynamics community .
transitions to chaos were investigated in spherical couette flow with a stationary outer sphere , with many groups reporting axisymmetric instabilities , akin to taylor grtler vortices , that sometimes broke the equatorial symmetry of the system @xcite .
planet- and star - like systems were studied by rapidly rotating the outer sphere . at low differential rotation these flows
are organized quasigeostrophically @xcite with a large shear across the tangent cylinder , the immaterial cylinder aligned with the rotation axis and touching the inner sphere s equator .
the first instability of this stewartson shear layer is non - axisymmetric @xcite . at moderate counter - rotation ,
experiments in maryland reported inertial modes @xcite excited by an over - critical shear across the stewartson layer .
wicht @xcite provides a thorough bestiary of the different regimes in numerical realizations of spherical couette flow .
magnetized spherical couette ( msc ) flow , where the fluid is electrically conducting and a magnetic field is applied from outside the fluid domain , introduces new dynamics . with an applied dipole field and a conductive inner sphere , magnetic entrainment yields a fluid domain rotating faster than the inner sphere . this linear
, theoretical result @xcite at low differential rotation has been extended to nonlinear systems in both experiments @xcite and simulations @xcite .
experiments reveal that fluctuations show up as a succession of broad peaks in the magnetic frequency spectra @xcite which correspond to increasing azimuthal wave numbers .
they have been related to linear modes of the non - linear magnetic base flow , but numerical simulations suggest that non - linear instabilities could show similar signatures @xcite .
the experimental model for the msc flow simulated here is the derviche tourneur sodium experiment ( dts-@xmath0 ) , a @xmath1 mm radius inner sphere containing a strong permanent dipole magnet inside a @xmath2 mm radius outer sphere @xcite . as the name implies , the fluid medium is liquid sodium .
the shell of the inner sphere is made of copper ( with four times the electrical conductivity of liquid sodium ) , while the outer sphere is made of stainless steel ( with one ninth the electrical conductivity of liquid sodium ) .
the rotation frequencies of the inner and outer sphere are respectively @xmath3 and @xmath4 , in the lab frame .
a recent set of upgrades to the diagnostics allows for better diagnosing ( both in terms of quality and quantity ) of flows when the outer sphere rotates ( @xmath5 ) . in tandem with these upgrades and new run campaigns , a set of direct numerical simulations ( dns )
have been run to predict and interpret the kinetic and magnetic measurements of the system .
the set of dns experiments shows three distinct dynamic regimes , though the lines between them are less than sharp .
figure [ fig : dynregime ] shows the ( scaled ) energies in the kinetic and magnetic fluctuations for the full set of dnss , with indications of where the dynamics , discussed herein , change from one regime to another .
the rossby number ( @xmath6 , where @xmath7 is the differential rotation rate ) appears as the prime determinant of which regime the system is in .
for @xmath8 the bulk flows appear quasigeostrophic with weak fluctuations .
for @xmath9 the flow is in a turbulent state , characterized by long lived filamentary structures . for @xmath10 fluctuations
are dominated by large scale , persistant modes . the determining factor for the kinetic fluctuation levels changes between regimes .
`` modal '' fluctuation levels are determined by the @xmath11 ; `` filamentary '' fluctuation levels are determined by the @xmath4 , which mostly likely relates to the diminished influence of the background magnetic field as the outer sphere spins faster .
the magnetic fluctuation levels are determined by both @xmath11 and @xmath4 , but this only demonstrates that the velocity term in the magnetic induction equation ( eqn .
[ eqn : magneticinduction ] in sec .
[ sec : numericalmodel ] ) has increased .
the changes in both fluctuation levels are small over the range of @xmath11s and @xmath4s shown in `` quasigeostrophic '' regime , so the exact relationship between the parameters and the fluctuation levels is out of the scope of this paper .
, in @xmath12 and @xmath13
. the velocities and magnetic fields are both in nondimensional units ( @xmath14 for velocity , @xmath15 for magnetic fields ) .
[ fig : dynregime ] ] this paper will proceed as follows .
section [ sec : numericalmodel ] introduces the numerical code used to model the dts-@xmath0device .
section [ sec : space ] shows the analysis of the computed velocity and magnetic fields when they are averaged over time and azimuth .
section [ sec : spattemp ] considers the full spatio - temporal data of the simulations and applies several techniques for analyzing them .
the persistent modes are subject to a principal component analysics ( pca ) .
the filaments are identified with their magnetic signatures at the surface .
section [ sec : time ] is a discussion of magnetic measurements at a single meridian on the outer surface of the sphere .
this is the measurement that is available on the physical dts-@xmath0device .
section [ sec : conclusions ] are conclusions for this work and perspectives for the future . |
one of the author , ken - ichi okubo would like to express his sincere gratitude to mr .
atsushi iwasaki at kyoto university for his wholesome advice .
_ proof of theorem 2 _ : we obtain the relation of eq .
( [ modified generalized boole ] ) by substituting @xmath17 by @xmath121 as , @xmath122
. \label{cot map}\end{aligned}\ ] ] the transformations @xmath123 have the topological conjugacy relation with eq .
( [ modified generalized boole ] ) .
consider the following cylinder sets of @xmath124 on which @xmath125 is surjective to @xmath126 , satisfying @xmath127 for any measurable set @xmath128 , and @xmath129-invariant measure @xmath130 , consider , for arbitrary @xmath131 @xmath132 where it is enough to assume @xmath133 is @xmath134 without loss of generality .
here , @xmath135 and the number of the intervals of @xmath136 which are included by @xmath133 is @xmath137 .
thus , the measure of @xmath136 on each interval @xmath138 whose measure @xmath139 is _ equidistributed _ , is @xmath140 .
then , for arbitrary @xmath131 , @xmath141 thus , the dynamical system @xmath142 has mixing property according to the ref .
therefore , the generalized boole transformations for @xmath0 have mixing property .
99 k. umeno , _ _ physical review e__*55 * ( 1997 ) , 5280 - 5284
. b. a. huberman and j. rudnick , _ phycisal review letter _ * 45 * ( 1980 ) , 154 .
y. pomeau and p. manneville , _ communications in mathematical physics _ * 74 * ( 1980 ) , 189 - 197 .
g. boole , _ philos .
* 147 * ( 1857 ) , 745 - 803 .
r. adler and b. weiss , _ israel journal of mathematics _ * 16 * ( 1973 ) , 263 .
j. aaronson , _ an introduction to infinite ergodic theory _ , ( american mathematical society , province , 1997 ) .
k. umeno , _ physical review e _ * 58 * ( 1998 ) , 26442647
. j. h. b. kemperman , _ stochastic processes and related topics i ( proc .
summer research inst . on statistical inference for stochastic processes , indiana univ . ,
bloomington , 1974 ) _ ( 1975 ) , 249 - 258 .
g. letac , _ proc .
* 67 * ( 1977 ) , 277 - 286 .
b. pesin , _ russian mathematical surveys _ * 32 * ( 1977 ) , 55114 .
eckmann and d. ruelle , _ reviews of modern physics _
* 57 * ( 1985 ) , 617 .
t. akimoto , m. nakagawa , s. shinkai and y. aizawa , _ physical review e _ * 91 * ( 2015 ) , 012926 .
t. akimoto and y. aizawa , _ chaos : an interdisciplinary journal of nonlinear science _ * 20 * ( 2010 ) , 033110 . m. nakagawa and y. aizawa , _ journal of the physical society of japan _ * 83 * ( 2014 ) , 104004 .
k. umeno , to appear in _ nolta _ * 7 * ( 2016 ) .
v. i. arnold and a. avez , _ ergodic problems of classical mechanics _ , benjamin , 1968 . | the generalized boole transformations have rich behavior ranging from the _ mixing _ phase with the cauchy invariant measure to the _ dissipative _ phase through the _ infinite ergodic _ phase with the lebesgue measure . in this letter , by giving the proof of mixing property for @xmath0 we show an _ analytic _ formula of the lyapunov exponents @xmath1 which are explicitly parameterized in terms of the parameter @xmath2 of the generalized boole transformations for the whole region @xmath3 and bridge those three phase _ continuously_. we found the different scale behavior of the lyapunov exponent near @xmath4 using analytic formula with the parameter @xmath2 . in particular ,
for @xmath0 , we then prove an existence of extremely sensitive dependency of lyapunov exponents , where the absolute values of the derivative of lyapunov exponents with respect to the parameter @xmath2 diverge to infinity in the limit of @xmath5 , and @xmath6 .
this result shows the computational complexity on the numerical simulations of the lyapunov exponents near @xmath7 0 , 1 .
[ [ introduction - generalized - boole - transformations ] ] 1 .
introduction - generalized boole transformations + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + lyapunov exponent has extensively studied for characterizing chaotic systems and plays an important and essential role as a chaos indicator characterizing the degree of chaoticity quantitatively .
however , one can find it difficult to obtain an analytic formula of lyapunov exponents for given chaotic systems in general .
there are few analytic formulas of ergodic invariant measures @xcite .
the analytic parametric density functions of ergodic invariant measure can be used for analytically estimating lyapunov exponents for such specific systems . however , for most cases , that analytic result can not be extended for the whole region of parameter space .
consequently , we are forced to use of numerical simulations to research a parameter dependence of lyapunov exponents . in the case of the logistic map , huberman and
rudnick @xcite found a scaling behavior of lyapunov exponent from zero to positive and pomeau and manneville @xcite found a similar behavior of lyapunov exponent .
both of them use numerical simulations . here
, we consider the boole transformation @xcite given by @xmath8 which is known to preserves the lebesgue measure @xmath9 as an invariant measure @xcite .
the lebesgue measure @xmath9 is an infinite measure .
thus , we can not normalize the invariant measure as a probability measure .
one can generalize the boole transformation as two - parameter map family as aaronson @xcite treats the following generalized boole transformations which are defined as @xmath10 preserving the cauchy measure @xmath11 for @xmath0 given by @xmath12}{dx}. \label{cauchy measure}\end{aligned}\ ] ] as a subset of the generalized boole transformations , the map at @xmath13 corresponds to the exactly solvable chaos map @xcite given by ref .
@xcite @xmath14 one of the authors showed that this map can be expressed by the double - angle formula of cot function having lyapunov exponent @xmath15 and one can use eq.([umeno map ] ) in order to generate general lvy stable distributions via their superposition @xcite .
the generalized boole transformations have the standard forms which do not depend on @xmath16 .
that is , by substituting @xmath17 by @xmath18 , eq .
( [ generalized boole ] ) is replaced by @xmath19 which is nothing but @xmath20 the transformation @xmath21 is a one - to - one diffeomorphism . then the transformations of eq .
( [ modified generalized boole ] ) have the same lyapunov exponent for fixed @xmath2 with this topological conjugacy relation .
consequently , the generalized boole transformations have the same lyapunov exponents which are independent of @xmath16 . [ [ analytic - formula - of - lyapunov - exponents - for - the - generalized - boole - transformations ] ] 2
. analytic formula of lyapunov exponents for the generalized boole transformations + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + here , we prove the following theorem .
[ t0.1 ] the generalized boole transformations for @xmath0 have the lyapunov exponents given by @xmath22 note that the lyapunov exponents do not depend on the parameter @xmath16 but purely depend on the parameter @xmath2 .
the ergodicity of similar kinds of this generalized boole transformations was shown by kempermann @xcite and letac @xcite .
we prove the theorem 2 concerning the mixing property of the generalized boole transformations for @xmath0 in the appendix .
the generalized boole transformations in @xmath0 have mixing property .
thus , we can prove the ergodicity with respect to the cauchy measure in eq .
( [ cauchy measure ] ) for @xmath0 . by the ergodicity
, we can calculate the lyapunov exponents of the generalized boole transformations as @xmath23 } dx .
\end{aligned}\ ] ] here , we substitute @xmath24 by @xmath25 , then we obtain : @xmath26 let us substitute @xmath27 for @xmath28 and put @xmath29 and @xmath30 as , @xmath31 evaluate @xmath32 .
we consider a function @xmath33 defined by @xmath34 which is regular on upper half - plane .
thus we can integrate along the path c1 in figure [ integral1 ] .
then we have the relations , width=245 ] @xmath35 as @xmath36 . next ,
as for @xmath37 , we consider a function @xmath38 defined by @xmath39 which is regular on upper half - plane .
thus we can integrate @xmath38 along the path c1 in figure [ integral1 ] .
we get : @xmath40 as @xmath36 . in terms of evaluating @xmath41 , by applying a transformation as @xmath42 we have the relation @xmath43 .
, width=245 ] as for @xmath30 , we consider a function @xmath44 defined by @xmath45 which is regular along the path c2 in figure [ integral2 ] and on the inner part of c2 .
then by integrating of @xmath44 along c2 , we obtain @xmath46 as @xmath47 . as for @xmath48
, we substitute @xmath49 by @xmath50 and we have the relation : @xmath51 then we set @xmath52 and we obtain : @xmath53 using these facts , we have @xmath54 therefore , we obtain analytic formula of the lyapunov exponents for the generalized boole transformations @xmath55 as @xmath56 the lyapunov exponents do not depend on @xmath16 , so that we can change the notation from @xmath57 to @xmath58 .
the invariant measure @xmath59 is cauchy and smooth with respect to the lebesgue measure .
thus , according to the pesin identity @xcite , we can calculate ks - entropy @xmath60 analytically as @xmath61 figure [ lyapunovb ] shows the comparison results of our numerical estimate of the lyapunov exponents for the generalized boole transformations with @xmath62 and @xmath63 using the long double precision and our analytic formula of the lyapunov exponents .
the initial condition of computation is @xmath64 .
the coincidence is remarkable as shown in figure [ lyapunovb ] .
equation ( [ lyapnov ] ) converges to zero in the limit of @xmath65 and @xmath6 .
however , different phenomena occur for @xmath66 and @xmath4 respectively . in the case of @xmath66 ,
every orbits are periodic because @xmath67 on the other hand , at @xmath4 , _ subexponential chaos _ @xcite occurs .
then , in calculation of finite lyapunov exponents , the value does not converge to a constant but converges in distribution .
[ [ lyapunov - exponents - for - alpha1 ] ] 3 .
lyapunov exponents for @xmath68 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + here , we prove another theorem .
[ t0.2 ] the generalized boole transformations for @xmath68 have the lyapunov exponents given by @xmath69 in the range of @xmath68 , the generalized boole transformations have two equilibrium points at @xmath70 .
two of them are unstable since the slope of the liner inclinations are both @xmath71 .
now by substituting @xmath17 by @xmath72 , we obtain another form of the generalized boole transformations as @xmath73 the equilibrium points of eq .
( [ medified gb2 ] ) are @xmath74 .
two points @xmath75 are unstable .
the equilibrium point at @xmath76 are stable and attractive because @xmath77 thus , for almost all initial points it is hold that @xmath78 hence , the lyapunov exponents of the generalized boole transformations for @xmath68 are given by taking logarithmic function of inclination in the limit of @xmath79 as @xmath80 the complete behavior of lyapunov exponent as a function of parameter @xmath2 is shown in figure [ fig : lyapunov exponent ] . via the analytic lyapunov exponents . ] according to ref .
@xcite , the generalized boole transformations for @xmath68 are _ dissipative _ in the sense that , for almost all initial points , @xmath81 as @xmath82 .
table [ table : classification ] summarizes a classification of the phase of systems by the nature of invariant measure .
we can say the system is in _ mixing phase _ for @xmath0 , _ infinite ergodic phase _ for @xmath4 and _ dissipative phase _ at @xmath68 . .
the classification of phase of dynamical systems , invariant measure and lyapunov exponents in terms of parameter @xmath2 . [ cols="^,^,^,^",options="header " , ] the standard cauchy density function @xmath83 is defined by a scale parameter @xmath84 @xmath85 according to the probability preservation relation @xcite , the density functions @xmath83 and @xmath86 satisfy the relation @xmath87 in the case of @xmath88 @xcite , the scale parameter @xmath84 of cauchy density function of invariant measure is updated in accordance with @xmath89 then , @xmath84 converges to unity .
similarly , by the probability preservation relation , the scale parameter of the density function of the generalized boole transformation is updated in accordance with @xmath90 and it converges to the fixed point @xmath91 . +
as we make @xmath2 close to unity from below , the scale parameter @xmath84 converges to infinity and the density function of invariant measure becomes the lebesgue measure @xmath9 .
+ for @xmath68 , the fixed point of scale parameter @xmath92 must be an imaginary number .
this fact shows for @xmath68 , the density function of the invariant measure can _ never _ be a cauchy distribution .
compare with the al map @xcite whose lyapunov exponent diverges to infinity at the fixed point , the generalized boole transformations for @xmath68 have the positive lyapunov exponent @xmath93 at the attractive fixed point @xmath76 .
[ [ scaling - behavior - at - alpha0 - 1 ] ] 4 .
scaling behavior at @xmath94 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + with parametric analytic formula of lyapunov exponents , we can investigate a parametric dependency .
the first derivative of the lyapunov exponents with respect to @xmath2 is given by @xmath95 remarkably , it diverges in the limit of @xmath96 as @xmath97 this is similar to the well - known critical phenomena that the derivative of magnetization diverges at the critical temperature .
this means that a _ slight _ modification of parameter @xmath2 toward unity ( the boole transformations ) from below causes a _ large _ effect on the value of lyapunov exponent at the edge of @xmath0 .
furthermore , we can say that it is _ extremely _ difficult to numerically obtain the lyapunov exponents near @xmath4 and @xmath0 because for @xmath4 _ subexponential chaos _ @xcite occurs and we can not determine a lyapunov exponent for finite calculation . according to eq .
( [ modified generalized boole ] ) , the generalized boole transformation at the limit of @xmath98 corresponds to the boole transformation , such that a transition behavior between chaotic behavior and _ subexponential _ behavior @xcite can be observed at @xmath99 . consider the scaling behavior at @xmath94 . when @xmath100 or `` @xmath101 and @xmath0 '' , the relation @xmath102 holds .
thus , we obtain a taylor expansion of eq .
( [ lyapnov ] ) as @xmath103 thus , the first approximation is given by the first term of the taylor expansion @xmath104 then , the scaling behavior of lyapunov exponents near @xmath105 is shown as @xmath106 in this paper , we define a critical exponent as a power exponent of lyapunov exponent .
a critical exponent @xmath107 is defined by @xmath108 where @xmath109 is a parameter , @xmath110 is a critical point and @xmath111 is a constant .
therefore , the both critical exponents @xmath107 at the egde of @xmath0 of the scaling behavior of lyapunov exponents are @xmath112 . the floquet multiplier for the fixed point for @xmath66 is -1 and for @xmath4 is 1 .
thus , our analytic results agree with the prediction by pomeau and manneville @xcite that critical exponents of lyapunov exponents of _ type 1 _ ( floquet multiplier is unity ) and _ type 3 _ ( floquet multiplier is @xmath113 ) intermittency are @xmath114 respectively .
the critical exponent in the limit of @xmath115 is different from that of @xmath116 .
this remarkable difference in critical exponents of lyapunov exponents is caused by the difference between the mixing ( ergodic ) phase for @xmath117 and the dissipative ( non - ergodic ) phase for @xmath68 .
[ [ conclusion ] ] 5 .
conclusion + + + + + + + + + + + + + we obtain the analytic formula of lyapunov exponents for the generalized boole transformations for the full range of parameters and prove the mixing of the map for @xmath0 .
we also obtain the ks - entropy by using the pesin s formula . as a result
, we can explicitly modulate lyapunov exponent @xmath1 in the range of @xmath118 by changing parameter @xmath3 .
in addition , the absolute values of the derivative @xmath119 at @xmath66 to @xmath120 are proven to be _
infinite_. then
, it is analytically shown to be extremely difficult to distinguish values of lyapunov exponent near @xmath66 or @xmath4 because of such strong parameter dependency .
the scaling behavior of our analytic lyapunov exponents of the boole transformations is found to be consistent with pomeau and manneville s numerical results about intermittency of _ type 1 _ and _ type 3_. thus , we expect that our analytic analysis can be useful for investigating physically chaotic systems further . |
a reciprocal lattice vector ( rlv ) has the form @xmath1 , where @xmath2 are primitive translations of the reciprocal lattice .
the integer `` indices '' @xmath3 were originally introduced as the miller indices , which are reciprocals of the intersection points ( in units of @xmath4 , etc . ) , of a crystal lattice plane with the axes along the primitive translation vectors @xmath5 of the direct lattice .
the `` 4-index '' notation @xmath0 in a hexagonal crystal denotes an rlv with @xmath6 .
the extra , or third , index @xmath7 in @xmath0 is redundant .
this index must equal the negative sum of the first two indices .
the notation restores symmetry between equivalent directions which is lost if the third index is omitted .
this notation has long been used by crystallographers , dating back at least to fedorov in 1890@xcite .
an extra miller index is natural in hexagonal symmetry , because , rather than two @xmath8-plane axes @xmath9 at 120@xmath10 , the three symmetrical axes @xmath11 , shown in fig .
1 , are natural .
the plane of atoms intersects all three axes .
the reciprocal intersection points @xmath12 are forced by trigonometry to obey the rule @xmath13 .
the proof@xcite is hinted in fig . 1 .
even before bragg scattering was observed and explained ( 1912 - 13 ) , the mathematical concept of a dual or reciprocal lattice was used@xcite .
after 1912 , physicists recognized the rlv as an x - ray momentum transfer .
the `` indices '' which label planes seem secondary unless we are specifically studying an atomic plane .
the redundant index may seem only a nuisance .
since the advantages of the 4-index notation are not always understood , it is common to omit the third index , as would be done in crystal systems which lack @xmath14 rotations .
this note is written in the belief that , once the underlying idea is clearly understood , the four index notation is natural .
it can be used to some advantage to label rlv s @xmath15 and also to label directions @xmath16 $ ] ( called `` zones '' when the direction perceived as the common axis of a family of planes ) in the direct lattice . however
, the 4-index labeling of directions @xmath17 that emerges in my analysis makes a subtle improvement over the one in use in electron microscopy@xcite .
otte and crocker@xcite discuss notation carefully , but with a different aim . here is the main idea . in an hexagonal crystal ,
define four unit vectors , with the fourth ( @xmath18 ) pointing along the @xmath19-axis .
the other three lie in the @xmath8-plane , at @xmath14 to each other , as shown in fig.[fig : hex ] .
it is conventional to have them point in the directions of primitive translations of the lattice .
then an arbitrary 3-vector @xmath20 is written with four components , as @xmath21 if @xmath20 is a position vector @xmath22 in `` real '' space , then it can be written as @xmath23 , \label{eq:4r}\ ] ] where the indices @xmath24 are dimensionless , and squared brackets conform to the crystallographic convention for directions in coordinate space
. if @xmath22 is a translation vector of the lattice , then the indices are integers . if @xmath20 is a vector of the reciprocal lattice @xmath25 , then it can be written @xmath26 where again the indices are dimensionless , and rounded brackets conform to convention . if @xmath25 is a translation vector @xmath27 of the reciprocal lattice , then the dimensionless indices are integers , and the shorthand @xmath28 is common . in this paper
, i will stick to a convention that vectors written in column form as in eqs.([eq:4r],[eq:4 g ] ) , or in row form with components separated by commas , contain the full dimensioned components , @xmath29 .
dimensionless abbreviations are never implied except in the index notations @xmath30 $ ] and @xmath31 without commas .
the term `` index '' will always refer to a dimensionless version of a component of a vector .
-plane of a hexagonal crystal , showing @xmath32 and @xmath33 axes , the third axis ( @xmath34 ) . in red
is shown the line of intersection of a plane of atoms with the @xmath8 plane .
this red plane intersects the @xmath35 axis at @xmath36 , the @xmath37 axis at @xmath38 , and the @xmath34 axis at @xmath39 .
it is always possible to choose the origin such that the intersection is at @xmath40 .
then the @xmath41 geometry guarantees that @xmath42 .
( b ) plaquette of 2-d hexagonal crystal , showing primitive translations @xmath9 of the direct lattice and @xmath43 of the reciprocal lattice.,width=453 ] in dirac notation , @xmath44 indicates a column vector , while @xmath45 indicates a row vector .
however , in this paper notation switches between two and three dimensions and between 2-vector , 3-vector and 4-vector systems . the dirac notation is mostly avoided to reduce ambiguity .
when a vector is written alone , it can be assumed a column vector . when written with another vector in a dot product @xmath46 ,
the left vector is a row vector and the right vector a column vector .
when written as a dyad @xmath47 , the left vector is a column vector and the right vector a row vector . when written in a line of text with commas , the vector @xmath48 is probably a column vector written sideways to save space .
the 4-vector `` basis '' is overcomplete .
it has both conventional and unconventional features .
the basis vectors are , in 4-component notation @xmath49 note that the sum of the first three indices is always zero , which follows from the fact that the first three unit vectors add to zero , @xmath50 .
it is perhaps disconcerting that the unit vectors @xmath51 do not `` look like '' unit vectors in the 4-vector notation , but this is a consequence of overcompleteness .
for example , the second component of the 4-vector that represents @xmath52 , is , according to eq.([eq:4v ] ) , @xmath53 .
the notation makes symmetry explicit .
the six translations have their first three indices chosen by the rule , organize three integers chosen from @xmath54 and @xmath55 , in all ways such that they add to zero .
because of the symmetrical mathematics , the scalar product ( _ e.g. _ @xmath56 ) in 4-vector form is simple but unconventional : @xmath57 this can easily be verified , and will be explained in sec .
when one drops the redundant third index from the 4-index notation , the remaining indices express vectors in a non - orthogonal basis .
there is perfectly good mathematics behind this , but it hides symmetry and simplicity . in a hexagonal system , these disadvantages are removed by the 4-index system . in systems of lower symmetry , where translation vectors are not at @xmath58 and @xmath41 to each other , such simplification is not available . in this section
, the mathematics of non - orthogonal basis sets is reviewed .
to simplify , the @xmath19-direction is now omitted .
the discussion thus refers to crystallography of hexagonal crystals in two dimensions .
the third dimension returns in sec .
iv . in the 2-d space of the @xmath8 plane ,
any two vectors that are not parallel or antiparallel can be chosen for a basis .
the most obvious choices are the primitive translations @xmath59 and @xmath60 , or alternately , the primitive translations @xmath61 and @xmath62 of the reciprocal lattice , defined as @xmath63 , and similarly for @xmath62 .
@xmath64 is the volume of the three - dimensional unit cell , @xmath65 .
then we have the usual vector relations @xmath66 and @xmath67 .
these relations indicate that the basis sets @xmath68 and @xmath69 are bi - othogonal .
it not always mentioned in texts that the primitive direct lattice vectors and the primitive rlv s are examples of the mathematical notion of bi - orthogonal basis vectors .
the important property is completeness : any vector can be expressed as a unique linear combination @xmath70 .
note that @xmath71 , where @xmath72 equals @xmath73 in hexagonal systems .
the coefficients @xmath74 and @xmath75 are found by solving a @xmath76 linear system in @xmath77 and @xmath78 .
a nice aspect of bi - orthogonality is that it diagonalizes the @xmath76 system and gives simple formulas for the coefficients , namely @xmath79 , @xmath80 .
it is arbitrary which basis ( direct space or reciprocal ) is taken to be primary and which to be dual .
thus , an arbitrary vector @xmath20 has an alternate representation , @xmath81 the coefficients are @xmath82 , @xmath83 .
the inner product @xmath46 of two vectors is not given simply by @xmath84 , but involves also the cross term @xmath85 .
the simple formula is @xmath86 , where the row vector is expressed in the basis dual to the one chosen for the column vector .
an equivalent formula is @xmath87 .
a compact mathematical representation of these relations is given by the equation @xmath88 , or by the alternate equation @xmath89 . here
the notation @xmath44 means a 2-vector , and @xmath90 is the @xmath76 unit matrix .
if written as a 2-component column vector , the basis should either be orthonormal , or if non - orthonormal , one has to be careful to use the direct and dual basis for the column vector @xmath44 and the row vector @xmath45 . in dyad notation ,
the relations are @xmath91 this formula is called `` the completeness relation '' or , equivalently , `` the decomposition of unity . ''
although this gives elegant formulas for inner products in non - orthogonal basis sets , these formulas are not likely to be used unless the vectors @xmath92 and @xmath20 belong separately , one to direct , and the other to reciprocal space .
then the formulas are obvious .
one has no trouble realizing that @xmath93 is equal to @xmath94 .
for two dimensional vectors , or the @xmath8-plane components of 3-d vectors of an hexagonal crystal , the overcomplete symmetrical basis is the three vectors @xmath95 which lie at @xmath41 to each other , as shown in fig.[fig : hex ] .
the key relationship is the dyad formula @xmath96 this decomposition of unity is a very nice alternative to eq.([eq : unityaa ] ) .
it is no longer necessary to have dual sets of direct and reciprocal lattice vectors .
the three vectors @xmath97 are self - dual . the scalar product of any two 2-vectors can be written as @xmath98 this formula can be written in an overcomplete 3-vector notation as @xmath99 where the usual array multiplication rule is obeyed .
unlike the case of the biorthogonal basis , here there is no need for care about whether @xmath92 or @xmath20 is in direct or reciprocal space , or whether the primary or the dual basis is implied .
the formula works for any two vectors .
it is worth emphasizing one subtlety . in the bi - orthogonal basis ,
when a column vector has indices @xmath100 $ ] , this means that @xmath70 , even though @xmath101 .
similarly , if the indices are written @xmath102 , this indicates a vector @xmath103 .
very different relations hold true in the symmetric overcomplete basis .
if the vector has components @xmath104 , this means that @xmath105 , _ etc .
_ , even though @xmath106 .
the actual answer is @xmath107 .
this follows immediately from eq.([eq : unity3e ] ) . it does not matter whether the vector @xmath20 is in direct space and indexed as @xmath108 $ ] or in reciprocal space and indexed as @xmath109
now let us express the reciprocal lattice vectors in the overcomplete symmetric notation .
first , recall that since @xmath110 , we have @xmath111 .
then we find @xmath112 , @xmath113 , and @xmath114 .
therefore we have @xmath115 the set of six symmetry - related smallest rlv s @xmath12 are simply all permutations of the three indices @xmath116 .
the general reciprocal lattice vector is written in various ways , as @xmath117 @xmath118 = \frac{2\pi}{a}\left(\begin{array}{c}n_a\\n_b\\-n_a - n_b\end{array}\right ) \rightarrow ( n_a n_b \
\overline{n_a+n_b } ) .
\label{eq : g33}\ ] ] it is amazing how similar the bi - orthogonal representation ( eq.[eq : g2 ] ) is to the overcomplete symmetric representation ( eq.[eq : g33 ] ) for rlv s .
they involve different coordinate systems and rules , yet the former derives from the latter by just dropping the extra third index , and the latter from the former by adding an index which is the negative sum of the first two indices .
the additional advantages of the extra index representation are that it is completely explicit ( if @xmath119 is included , the vector is completely specified ) , and there is no ambiguity about the direct versus dual parts of the bi - orthogonal representation .
it is best to think of eq.([eq:4v ] ) as a way of representing the vector @xmath20 but not to think of the components of the 4-vector as if they had meaning similar to the ordinary 3-component notation . in ordinary vector notation , if the vector @xmath20 is denoted @xmath120 $ ] , then it has the formula @xmath121 . or , if it is denoted by @xmath122 , then it has the formula @xmath123 . in 4-index notation , vectors indexed as @xmath31 or @xmath30 $ ] have the formulas @xmath124 + \frac{1}{c}n_4 \hat{e}_4 \right ) .
\label{eq : vyes}\ ] ] @xmath125 = \left(\begin{array}{r } n_1 a \\
n_3 a \\ n_4
c\end{array}\right ) \rightarrow \left(\frac{2a}{3}[n_1\hat{e}_1 + n_2 \hat{e}_2 + n_3\hat{e}_3 ] + c
n_4 \hat{e}_4 \right ) .
\label{eq : vyes2}\ ] ] note the unconventional factor 2/3 .
when written as an additive sum of primitive vectors proportional to @xmath97 , an arbitrary additive constant can be added to @xmath126 or @xmath104 with no algebraic or notational error .
for example , @xmath127 + v_4 c
\hat{e}_4 \ne \left(\begin{array}{c } ( v_1+c)a \\ ( v_2+c)a \\ ( v_3+c)a \\ v_4 c\end{array}\right ) \label{eq : vno}\ ] ] when written in overcomplete 4-vector notation , @xmath128 are fixed numbers , necessarily adding to zero , and therefore with no arbitrariness
. the constant @xmath129 can not be added , even though the vector relation @xmath130 is true . the inner product of two vectors , in 4-vector notation ,
is @xmath131 this is just an alternative way of writing eq.([eq : dotprod ] ) , that emphasizes the unconventional metric .
a factor @xmath132 occurs in the first three diagonal entries .
the metric is positive , so the inner product is safely defined .
eq.([eq:3dotprod ] ) holds for all vectors provided the indices @xmath133 and @xmath134 are written in full rather than abbreviated index form . for the special case where one of @xmath135 is in direct and the other in reciprocal space , the inner product ,
_ modulo _ @xmath136 , also has this form , _
i.e. _ @xmath137=2\pi[(2/3)(hs+kt+ir)+lu]$ ] , in the index form . if both are direct or reciprocal space vectors , an extra factor @xmath138 or @xmath139 appears in the fourth term .
the unconventional metric with the factor @xmath132 does not appear in previous literature . if the 4-index notation is confined to indexes of planes and reciprocal lattice vectors , then the current definition eq.([eq:4v ] ) agrees with the one in the literature .
it is less common for a 4-index notation to be used for coordinate space directions .
inner products are seldom discussed , and the factor @xmath132 does nt arise . for real space directions directions , the current definition eq.([eq:4v ] ) differs from used in the literature@xcite .
consider for example the lattice point @xmath140 .
using eqs.([eq:4v],[eq:4r ] ) , this is indexed as @xmath141 $ ] .
the literature definition is different .
it requires first writing @xmath142 .
the extra term is zero but is added to make the sums of the coefficients of @xmath143 , @xmath144 , and @xmath145 add to zero . by the literature definition ,
this vector is indexed as @xmath16 $ ] where the indices are the coefficients of @xmath146 , etc . , and @xmath147 .
thus we get @xmath148 $ ] .
the first three indices have been reduced by @xmath132 .
it never seems to be mentioned that the definition was changed between direct and reciprocal space .
the factor @xmath132 is now incorporated into the definition of the first three indices @xmath149 of @xmath16 $ ] of direct space vectors , but not into the first three indices @xmath150 of reciprocal space vectors
. then the inner product of a direct space vector @xmath16 $ ] and a reciprocal space vector @xmath15 is @xmath151 , without the extra @xmath132 .
the penalty is that inner products of two direct space vectors or two reciprocal space vectors must be computed with awkwardly different rules , whereas the unified definition offered here gives a simple unified ( but unconventional ) rule . in order to retain the usual definitions
, one could accept as a compromise the _ ad hoc _ rule that whenever dimensionless real space indices @xmath16 $ ] are written , they incorporate an extra factor @xmath132 beyond eq.([eq:4v ] ) , in the first three indices . however , when written out in full column vector notation , there is no need for index notation , and such a compromise would be unwise .
eq.([eq:4v ] ) should be used , and the inner product rule eq.([eq : dotprod ] ) applies .
overcomplete basis sets are not abnormal in physics .
they are used frequently for infinite - dimensional problems .
a quantum harmonic oscillator , for example , has an infinite complete orthonormal basis of eigenstates @xmath152 , but the `` coherent state '' representation , which is overcomplete , is often preferable@xcite . in finite - dimensional vector problems ,
hexagonal crystallography is not unique ; symmetry and orthonormality may compete and suggest an overcomplete symmetric representation .
an example is electronic @xmath153-states , where cubic symmetry lifts 5-fold degeneracy into a 3-fold degenerate t@xmath154 manifold ( spanned by orthonormal functions @xmath155 , @xmath156 , and @xmath157 ) and the 2-fold degenerate e@xmath158 manifold ( spanned by orthogonal functions @xmath159 and @xmath160 . ) the t@xmath154 orthogonal basis is nicely adapted to the 4-fold rotations of cubic symmetry , whereas these same rotations mix the orthogonal e@xmath158 functions in an ugly way .
a cure@xcite is an overcomplete non - orthogonal basis such as @xmath161 , @xmath162 , and @xmath163 .
the mathematics of this representation is exactly the same as the symmetric overcomplete representation described here for hexagonal crystals .
i thank my collaborators in the solar water splitting simulation team ( swassit ) who helped me understand the @xmath0 surface of wurtzite materials . m. blume and a.
g. abanov made valuable comments .
i thank the cfn of brookhaven national laboratory for hospitality .
this work was supported by doe grant defg0208er46550 .
99 e. fedorov , zeits . kristall . *
17 * , 615 ( 1890 ) .
j. d. h. donnay , am .
mineralogist * 32 * , 52 ( 1947 ) .
j. w. gibbs and e. b. wilson , _ vector analysis _ , yale university press , new haven , 1901 ) .
j. w. edington , _ practical electron microscopy in materials science _ , ( van nostrand reinhold , new york , 1976 ) . h. m.
otte and a. g. crocker , phys .
* 9 * , 441 ( 1965 ) .
j. w. negele and h. orland , _ quantum many - particle systems _ ( addison - wesley , redwood city , 1988 ) pp . 20 - 25 .
v. perebeinos and p. b. allen , phys .
* 85 * , 5178 ( 2000 ) . | a four index notation ( _ e.g. _ @xmath0 ) is often used to denote reciprocal lattice vectors or crystal faces of hexagonal crystals . the purposes of this notation have never been fully explained .
this note clarifies the underlying mathematics of a symmetric overcomplete basis .
this simplifies and improves the usefulness of the notation . |
since their discovery , grbs turned out to be incredibly powerful sources , with detected fluences up to @xmath2 erg/@xmath3 in the @xmath4ray band , above few tens of kev .
the first spectroscopic measurements of their redshifts ( e.g.@xcite ) , besides confirming their cosmological nature , indicated that these events release , in the @xmath4ray band , a huge amount of energy , up to @xmath5 erg .
this extraordinary energetic content became a challenge for the proposed grb models .
one implicit hypothesis in deriving the grb energy , from the observed fluence and measured redshift , consisted in assuming that grbs emit isotropically . however , it was suggested @xcite that grbs might be collimated into a cone of semiaperture @xmath6 .
the jet opening angle could be directly estimated , under some simplifying assumptions on few other parameters , from the measure of the achromatic break time @xmath7 in the afterglow light curve @xcite .
the presence of a jet in grb outflows , supported by observations in most events @xcite , allowed to correct the isotropic equivalent energy @xmath8 for the collimation factor , therefore obtaining the collimation corrected energy @xmath9 @xcite .
these results suggested that grbs might have a unique energy @xmath10 erg .
the study of the rest frame spectral properties of a sample of @xmath11sax grbs @xcite ( see also @xcite ) led to the discovery of a correlation between the isotropic equivalent energy @xmath8 and the @xmath12 spectral peak energy , @xmath13 ( the amati correlation ) . with the largest sample of bursts with spectroscopically measured redshifts , published spectra and well determined jet break time , we estimated the jet opening angles and derived @xmath14 .
we discovered a very tight correlation between @xmath14 and the spectral peak energy @xmath15 @xcite ( the ghirlanda correlation ) .
this correlation relates the grb prompt emission energy
properly corrected for the burst geometry to its peak frequency .
it might be the key to understand some still obscure aspects of the physics and origin of grbs . besides , its small scatter and good powerlaw fit allowed to use , for the first time , grbs as standard candles to constrain the cosmological parameters ( see @xcite ) .
we collected all the grbs with redshift measurements and published spectral parameters . in most cases the spectrum is represented by the empirical band function .
the spectrum allows to compute the burst bolometric fluence @xmath17 ( i.e. the time integrated flux ) and , hence , the isotropic equivalent energy , i.e. @xmath18 . the jet break time , typically observed between 0.1 and 10 days since the burst trigger , is due to the deceleration , by the external ism , of the grb relativistic fireball .
when the lorentz factor of the fireball is @xmath19 a change in the time decay slope of the afterglow flux is observed .
this characteristic time depends also on the ism density @xmath20 and on the kinetic energy which is still in the fireball after the prompt emission phase ( parametrized through the efficiency @xmath21 ) .
therefore , the measure of @xmath22 allows to estimate @xmath23 .
after the publication of the original work of ggl04 , the redshifts and spectral parameters of four more bursts were published .
we present here the updated correlations ( either @xmath13@xmath8 and @xmath13@xmath14 ) .
a continuously updated version of the correlations and the relative tables can be found at http://www.merate.mi.astro.it/@xmath24ghirla/deep/blink.htm ( with the complete reference list ) .
we use @xmath25 , @xmath26 .
the grbs added to the original sample of 23 ( see also @xcite , @xcite ) are : * grb 021004 @xcite , with a rest frame peak energy @xmath27 kev and @xmath28 erg .
the jet break time is @xmath29 days , the jet opening angle @xmath30 and , therefore , @xmath31 erg .
* grb 030323 @xcite , with a rest frame peak energy @xmath32 kev and @xmath33 erg .
the afterglow light curve is relatively flat i.e. indicating a jet break time @xmath34 days , implying @xmath35 and , therefore , @xmath36 erg .
* grb 040924 @xcite , with a rest frame peak energy of @xmath37 kev and @xmath38 erg .
the afterglow light curve is relatively flat [ @xmath39 up to 1 day , indicating a jet break time @xmath40 day , implying @xmath41 and , therefore , @xmath42 erg .
* grb 041006 @xcite , with a rest frame peak energy @xmath43 kev and @xmath44 erg .
the jet break time is @xmath45 days , implying @xmath46 and , therefore , @xmath47 erg .
we added the four grbs to the 23 grbs of tab.1 and 2 of ggl04 . in fig .
1 we report the updated correlations . with 27 grbs ( black symbols is fig .
1 ) the best fit powerlaw ( weighting for the errors on both coordinates ) of the @xmath13@xmath8 correlation is : @xmath48 with a reduced @xmath49 ( long dashed line in fig .
the least square fit to the same data points ( i.e. ignoring the errors on the two coordinates ) gives a slope of [email protected] ( dashed line in fig .
the ghirlanda @xmath13@xmath14 correlation , updated with two of the four grbs with known @xmath6 ( green filled symbols in fig .
1 ) , is : @xmath51 with a reduced @xmath52 ( solid line in fig . 1 , shaded region is the uncertainty of this correlation ) .
least square fit gives a slightly flatter slope of 0.6 .
the present gaussian fit of the distribution of the perpendicular scatter of the 27 grb around their best fit ( eq .
1 ) is @xmath53 .
the scatter of the ghirlanda correlation ( eq .
2 ) , instead , is only @xmath54 ( i.e. consistent with what found with the 15 grbs in ggl04 ) .
we stress that the ghirlanda correlation is well fitted ( also with the 2 new grbs ) by a single powerlaw and that its slope , 0.69 , is consistent with what found with the 15 grbs in ggl04 .
moreover , the reduced @xmath55 allows its use in cosmology ( gglf04 , ggf05 and @xcite ) .
this is in net contrast with what claimed by @xcite who find @xmath56 for the @xmath0@xmath14 fit .
although the authors do not investigate the reason of their statistically unacceptable results , we note that their severe underestimation of the uncertainty on two relevant parameters ( i.e. the ism density and the jet break time ) is driving their conclusions .
they assume , when unknown , an ism density with an uncertainty of 50% .
this parameter is highly uncertain due to the few measured values ( and also in these few cases highly debated ) . until precise measurements of this parameter
will not be available , it is preferable to let it vary within a relatively large range .
this is indeed what ggl04 did in their original work ( where @xmath57 ) .
moreover , in @xcite a few jet break times are reported with unreliably small errors ( down to 1% ) .
this means that for some grbs we could determine a break in their light curves at say 1.5 days with an uncertainty of 28 minutes ( e.g. grb 011211 ) , which may be a challenge for the future but which was extremely hard in the past afterglow observational campaigns .
we have presented the updated ghirlanda correlation with 17 grbs with firm redshift measurements and published spectral parameters .
these 2 more events perfectly fit the ghirlanda correlation as found by ggl04 . while waiting for future events , even the present small sample of grbs have important and intriguing implications for the use of grbs as standard candles to measure our universe .
we thank annalisa celotti & fabrizio tavecchio for useful discussions .
gg and gg thank the miur for the cofin grant 2003020775_002 . | the recently discovered correlation between the rest frame grb peak spectral energy @xmath0 and the collimation corrected energy @xmath1 in long grbs is potentially very important , yet awaits confirmation from an independent sample .
it may help to shed light on the radiation mechanism of the prompt grb phase and on the way and in which form the energy is released from the central engine .
we here present some additional evidence for the correlation ( two new bursts ) and re - derive the best - fit parameters .
the tightness of the correlation is confirmed ( sigma=0.1 dex ) .
we show that this correlation allows us , for the first time , to use grbs as cosmological probes to constrain the expansion history of the universe .
[ 1999/12/01 v1.4c il nuovo cimento ] |
the opal collaboration presented measurements of the fractional energy spectra of inclusive @xmath2 @xcite , @xmath0 , @xmath1 , @xmath3 , and @xmath4 @xcite production in @xmath11-boson decays based on their entire lep1 data sample .
apart from the full cross sections , they also determined the contributions arising from @xmath12 decays .
this enabled us , partly in collaboration with binnewies , to determine lowest - order ( lo ) and next - to - leading - order ( nlo ) sets of fragmentation functions ( ff s ) for these charmed ( @xmath13 ) hadrons @xcite .
we took the charm - quark ff to be of the form proposed by peterson _
et al . _
@xcite and thus obtained new values of the @xmath14 parameter , which are specific for our choice of factorization scheme .
we worked in the qcd - improved parton model implemented in the pure modified minimal - subtraction ( @xmath5 ) renormalization and factorization scheme with @xmath15 massless quark flavors ( zero - mass variable - flavor - number scheme ) .
this scheme is particularly appropriate if the characteristic energy scale of the considered production process , _ i.e. _ , the center - of - mass energy @xmath16 in the case of @xmath6 annihilation and the transverse momentum @xmath17 of the @xmath13 hadron in other scattering processes , is large compared to the bottom - quark mass @xmath18 .
owing to the factorization theorem @xcite , the ff s defined in this scheme satisfy two desirable properties : ( i ) their scaling violations are ruled by the time - like dokshitzer - gribov - lipatov - altarelli - parisi ( dglap ) @xcite evolution equations ; and ( ii ) they are universal .
thus , this formalism is predictive and suitable for global data analyses .
we verified that the values of the branching and average momentum fractions of the various @xmath19 transitions evaluated at lo and nlo using our ff s @xcite are in reasonable agreement with the corresponding results from opal @xcite and other experiments @xcite .
we tested the scaling violations of our @xmath0 , @xmath1 , @xmath3 , and @xmath4 ff s @xcite by comparing the fractional energy spectra of these hadrons measured in non - resonant @xmath6 annihilation at @xmath20 gev @xcite , 29 gev @xcite , and 34.7 @xcite with our lo and nlo predictions to find reasonable agreement .
since events of @xmath13-hadron production from @xmath21-hadron decay were excluded from the data samples at @xmath20 gev , we obtained a clean test of our charm - quark ff s . in refs .
@xcite , the starting points @xmath22 for the dglap evolution in the factorization scale @xmath7 were taken to be @xmath8 , where @xmath9 .
this choice is phenomenologically motivated by the observation that , in @xmath6 annhilation , which has been providing the most constraining input for the determinations of ff s , these values of @xmath22 represent the very production thresholds of the respective flavors .
unfortunately , this choice is inconsistent with the convention underlying modern sets of parton density functions ( pdf s ) @xcite , which prefer to place the flavor thresholds at @xmath10 . for the reader
s convenience , in this addendum to refs .
@xcite , we thus repeat the analysis of that papers for the choice @xmath10 , so as to provide alternative lo and nlo sets of @xmath13 ff s that can be conveniently utilized together with those pdf s .
the ff s presented below were already used as input for a nlo study @xcite of charmed - meson hadroproduction in @xmath23 collisions , which yielded agreement within errors with data collected by the cdf collaboration in run ii at the fermilab tevatron @xcite .
we note in passing that , in the case of perturbatively induced ff s , which is quite different from the case of non - perturbative ff s ( involving substantial intrinsic components ) under consideration here , the choice @xmath10 is more natural , since , at nlo , it avoids finite matching conditions at the flavor thresholds @xcite .
in the following , we concentrate on the most important results of refs . @xcite that are affected by the shift in @xmath22 . these include the fit parameters @xmath24 , @xmath25 , @xmath26 , and @xmath14 defining the @xmath27 distributions of the @xmath28 ff s @xmath29 at @xmath30 , @xmath31 ^ 2 } , \label{eq : cff}\\ d_b(x,\mu_0 ^ 2)&=&nx^{\alpha}(1-x)^{\beta } , \label{eq : bff}\end{aligned}\ ] ] the @xmath32 values per degree of freedom ( @xmath33 ) achieved in the fits , and the branching fractions @xmath34 and average momentum fractions @xmath35 , @xmath36 where @xmath37 , at @xmath38 and @xmath39 .
in the present analysis , we adopt the up - to - date input information from our 2005 paper @xcite .
our new results are presented in tables [ tab : par][tab : xav ] .
comparing tables [ tab : br ] and [ tab : xav ] with the corresponding tables in refs .
@xcite , we observe that the branching and average momentum fractions are changed very little by the reduction in @xmath22 . for a comparison of these observables with experimental data ,
we refer to refs .
@xcite .
.fit parameters of the charm- and bottom - quark ff s in eqs .
( [ eq : cff ] ) and ( [ eq : bff ] ) , respectively , for the various @xmath13 hadrons at lo and nlo .
the corresponding starting scales are @xmath40 gev and @xmath41 gev , respectively .
all other ff s are taken to be zero at @xmath42 . [
cols="^,^,^,^,^,^,^",options="header " , ] for lack of space , we refrain from presenting here any updated versions of figures included in refs .
@xcite ; they would not exhibit any qualitatively new features . however , as already mentioned in ref .
@xcite , the reduction in @xmath22 has an appreciable effect on the gluon ff s , which are only feebly constrained by @xmath6 data .
this effect is visualized for @xmath43 in fig .
[ fig : one ] , where the @xmath10 to @xmath8 ratios of @xmath44 at @xmath45 , 10 , and 20 gev are shown as functions of @xmath27 .
we observe that the reduction in @xmath22 leads to a significant enhancement of the gluon ff , especially at low values of @xmath27 .
the results for @xmath46 , @xmath1 , @xmath3 , and @xmath4 are very similar and , therefore , not shown here .
in this addendum to refs .
@xcite , we repeated the fits of non - perturbative @xmath0 , @xmath1 , @xmath2 , @xmath3 , and @xmath4 ff s , both at lo and nlo in the @xmath5 factorization scheme , to opal data from lep1 @xcite for the reduced choice @xmath10 ( @xmath9 ) of starting point for the dglap evolution in the factorization scale @xmath7 .
these ff s are appropriate for use in connection with modern sets of pdf s @xcite , which are implemented with the same convention for the heavy - flavor thresholds .
a fortran routine that evaluates the values of these ff s as functions of the input variables @xmath27 and @xmath7 may be obtained by electronic mail upon request from the authors .
this reduction in @xmath22 is inconsequential for the theoretical interpretation of experimental @xmath6 data because it is compensated by corresponding shifts in the fit parameters @xmath24 , @xmath25 , @xmath26 , and @xmath14 .
however , the gluon ff s , which are only feebly constrained by @xmath6 data , play a significant role in hadroproduction .
in fact , detailed analysis @xcite revealed that the increase in the gluon ff s due to the extension of the evolution length leads to a rise in cross section and thus improves the agreement with the cdf data of charmed - meson production in run ii at the tevatron @xcite .
v. n. gribov and l. n. lipatov , yad .
* 15 * , 781 ( 1972 ) [ sov . j. nucl
. phys . * 15 * , 438 ( 1972 ) ] ; g. altarelli and g. parisi , nucl .
* b126 * , 298 ( 1977 ) ; yu .
l. dokshitzer , zh . eksp .
* 73 * , 1216 ( 1977 ) [ sov .
jetp * 46 * , 641 ( 1977 ) ] .
l. gladilin , report no .
hep - ex/9912064 ( unpublished ) ; delphi collaboration , p. abreu _ et al .
_ , eur .
j. c * 12 * , 225 ( 2000 ) ; aleph collaboration , r. barate _
_ , _ ibid . _ * 16 * , 597 ( 2000 ) ; s. padhi , in _ proceedings of the ringberg workshop on new trends in hera physics 2003 _ , edited by g. grindhammer , b. a. kniehl , g. kramer , and w. ochs , ( world scientific , singapore , 2004 ) , p. 183 ; h1 collaboration , a. aktas _ et al . _ ,
j. c * 38 * , 447 ( 2005 ) .
cleo collaboration , d. bortoletto _ et al .
_ , phys .
d * 37 * , 1719 ( 1988 ) ; cleo collaboration , r. a. briere _ et al .
. _ * 62 * , 072003 ( 2000 ) ; cleo collaboration , m. artuso _ et al .
_ , _ ibid .
_ * 70 * , 112001 ( 2004 ) . | in phys . rev .
d * 58 * , 014014 ( 1998 ) and * 71 * , 094013 ( 2005 ) , we determined non - perturbative @xmath0 , @xmath1 , @xmath2 , @xmath3 , and @xmath4 fragmentation functions , both at leading and next - to - leading order in the @xmath5 factorization scheme , by fitting @xmath6 data taken by the opal collaboration at cern lep1 .
the starting points for the evolution in the factorization scale @xmath7 were taken to be @xmath8 , where @xmath9 .
for the reader s convenience , in this addendum , we repeat this analysis for @xmath10 , where the flavor thresholds of modern sets of parton density functions are located . |
let @xmath17 be the language of arithmetic with symbols @xmath18 and @xmath19 the ordinary axiomatization of arithmetic in first order logic .
let @xmath20 be the standard model of pa and @xmath9 be its domain , i.e. , the set of natural numbers .
if @xmath21 let @xmath22 ( or @xmath23 if @xmath4 is understood from the context ) be the initial segment @xmath24 of @xmath4 .
@xmath25 means that @xmath26 is an initial segment of @xmath4 , i.e. , @xmath27 for all @xmath28 .
@xmath25 is a cut if it is closed under taking successors .
let @xmath29 be a @xmath30-formula in the language of arithmetic used for coding sets in @xmath31 ( see @xcite ) .
is pa with the induction axioms restricted to @xmath32 formulas . ] if @xmath26 is a cut in @xmath4 and @xmath33 then @xmath34 .
the set of all coded subsets of @xmath26 will be denoted @xmath35 .
the standard system of @xmath4 is @xmath36 .
if @xmath37 is a cut closed under multiplication and @xmath38 we can regard the structure @xmath39 as a second order structure in the language @xmath17 .
the subtheories @xmath40 , @xmath41 and @xmath42 of second order arithmetic are as defined in @xcite .
a cut @xmath43 is semiregular iff @xmath44 , and strong iff @xmath45 .
when we write @xmath46 it will be understood that @xmath46 is a second order structure in the language @xmath17 , i.e. , that @xmath4 is equipped with addition and multiplication and @xmath47 .
@xmath46 is an @xmath9-model if @xmath48 is the standard model of pa . in this case
@xmath49 iff @xmath2 is a scott set .
we will often specify an @xmath9-model by only specifying @xmath50 .
given a theory @xmath3 , in the language of arithmetic , we say that a set @xmath51 is represented in @xmath3 if there is a formula @xmath52 such that @xmath53 for all @xmath54 and @xmath55 for all @xmath56 . by @xmath57
we denote the collection of all sets represented in @xmath3 . in @xcite scott proved a variant of the following theorem : [ thm : scott ] for any countable @xmath9-model @xmath2 and complete consistent theory @xmath58 the following are equivalent : * @xmath59 and @xmath60 , and * there is a countable nonstandard model @xmath61 with @xmath62 .
this theorem could be taken a bit further : given a scott set @xmath2 of cardinality @xmath10 there is a nonstandard model of @xmath19 with standard system @xmath2 : [ union ] for any @xmath9-model @xmath2 of cardinality at most @xmath10 and complete consistent theory @xmath63 the following are equivalent : * @xmath59 and @xmath60 , and * there is a nonstandard model @xmath61 with @xmath62 . the proof is based on a union of chains argument .
it should be noted that for presburger arithmetic , @xmath64 , this argument could be extended to any cardinality @xmath65 as proved in @xcite .
for pa the argument only applies for cardinalities less than or equal @xmath10 , since models of pa does not admit the amalgamation property needed , as pointed out in @xcite . for recursively saturated models
the situation is almost the same as wilmers proved in @xcite : [ wilmers ] for any @xmath9-model @xmath2 and complete consistent theory @xmath66 the following are equivalent : * @xmath67 and there is a model @xmath61 with @xmath62 , and * there is a recursively saturated model @xmath61 with @xmath62 .
the proof is a rather straightforward application of the arithmetized completeness theorem .
observe that if @xmath2 is a scott set and @xmath67 then @xmath68 , however it may well happen that @xmath60 and @xmath69 . given @xmath46 and
@xmath37 a cut , we say that @xmath70 is coded in @xmath2 if there is @xmath71 such that @xmath72 .
also , if @xmath73 where @xmath74 also is a cut we say that @xmath75 is coded in @xmath2 if there is @xmath71 such that @xmath76 .
given a theory @xmath3 let @xmath77 be the first order theory consisting of the sentences @xmath78 , where @xmath79 ranges over all finite subtheories of @xmath3 and @xmath78 is the sentence saying that @xmath80 is not provable in first order logic , where @xmath81 is the conjunction of all sentences in @xmath82 .
the next theorem is a special version of the arithmetized completeness theorem tailormade for our purposes .
[ act ] let @xmath83 , where @xmath3 is some first order theory extending @xmath19 coded in @xmath2 .
then there is a model @xmath84 with domain @xmath4 and a set @xmath85 such that @xmath86 iff @xmath87 .
furthermore , the canonical embedding @xmath88 is coded in @xmath2 and such that @xmath89 is an initial segment of @xmath6 .
let @xmath12 code the theory @xmath3 , i.e. , @xmath90 . by the assumption on @xmath46 we have @xmath91 for every @xmath92 , where @xmath93 is the sentence saying that there is no proof of absurdity from sentences satisfying the formula @xmath94 . if @xmath95 then clearly @xmath96 by the compactness theorem
. assuming @xmath4 is nonstandard we can use overspill ( this is where we need @xmath46 to satisfy @xmath42 ) to find @xmath97 such that @xmath98 .
using the fact that the completeness theorem is provable in @xmath41 we get the desired set @xmath99 ( see @xcite ) .
let @xmath100 be a partial order .
a filter @xmath101 on @xmath102 is an upwards closed non - empty subset of @xmath102 such that if @xmath103 then there is @xmath104 satisfying @xmath105 and @xmath106 .
a filter @xmath11 is an ultrafilter if it is a maximal filter .
given @xmath46 let @xmath1 be the partial order of all unbounded sets in @xmath2 ordered by @xmath107 .
a filter @xmath11 on @xmath1 is complete if for all @xmath108 coded in @xmath2 with a bounded range there is @xmath109 such that @xmath110 is constant on @xmath75 .
any complete filter on @xmath1 is a nonprincipal ultrafilter on the boolean algebra @xmath2 . also , if @xmath2 is an @xmath9-model any nonprincipal ultrafilter on @xmath2 is a complete filter on @xmath1 .
recall that a filter @xmath11 on @xmath1 is definable if for all @xmath12 @xmath111 , where @xmath112 and @xmath15 is some canonical function coding pairs .
we will , somewhat sloppily , say that a filter @xmath113 on @xmath1 is an ultrafilter if @xmath113 is an ultrafilter on @xmath2 .
[ lem : def.ar ] if @xmath114 and there is a definable ultrafilter @xmath113 on @xmath1 then @xmath0 . to see this let @xmath115 , where @xmath12 and @xmath116 is @xmath117 . define @xmath118 , then @xmath119 and thus @xmath120 since @xmath113 is definable .
we are now in a position to start proving the following theorem .
theorem [ maintheorem ] will follow from it .
[ mainlemma ] if @xmath0 , @xmath1 has a definable complete filter @xmath11 and @xmath63 is coded in @xmath2 such that @xmath121 , then there is an end - extension @xmath84 of @xmath4 satisfying @xmath122 .
first we construct the ultrapower we will use to build the model @xmath6 .
given the setup of theorem [ mainlemma ] let @xmath123 be the model of @xmath3 given by theorem [ act ] and let @xmath124 be such that @xmath125 and @xmath126 is the canonical embedding of @xmath4 into @xmath123 .
let @xmath127 be the set of all functions @xmath128 such that the function @xmath129 is coded in @xmath2 . for any ultrafilter @xmath113 on @xmath1 define @xmath130 to be the set of equivalence classes of the equivalence relation @xmath131 defined on @xmath127 by @xmath132 iff the set where @xmath110 and @xmath133 are equal , @xmath134 , is in @xmath113 .
the collection @xmath130 of equivalence classes can be interpreted as a structure in the language of arithmetic by the ordinary definitions of functions and relations .
let @xmath6 denote some model of the form @xmath135 , where @xmath113 will be understood to be an ultrafilter on @xmath1 .
let @xmath81 be a sentence in the language @xmath136 , i.e. the language of arithmetic extended with the set @xmath127 as parameters .
the @xmath137-sentence we get by replacing all occurrences of functions @xmath110 by the value @xmath138 will be denoted by @xmath139 $ ] . by @xmath140 $ ] we mean the @xmath141-sentence we get by replacing all functions @xmath110 by the equivalence class @xmath142 $ ] .
the o theorem in this setting follows : for any sentence @xmath81 of @xmath136 we have @xmath143 $ ] iff @xmath144 } } \in u.\ ] ] let us embed @xmath145 in @xmath6 via the canonical embedding @xmath146 , @xmath147 $ ] , where @xmath148 for all @xmath149 .
we will identify @xmath150 with @xmath151 making @xmath145 a substructure of @xmath6 .
in fact o theorem gives us that that @xmath152 and thus that @xmath84 .
let us summarize .
we have @xmath153 , where @xmath154 and there is a model @xmath123 with the same domain as @xmath4 such that @xmath155 is coded in @xmath2 and @xmath156 is an isomorphism .
[ complete ] if @xmath113 is complete then @xmath157 .
let @xmath142 \in n$ ] be such that @xmath158 < a$ ] , @xmath33 .
take @xmath159 $ ] such that @xmath160 for all @xmath149 .
observe that the function @xmath161 is coded in @xmath2 and that @xmath162 . since @xmath163 is bounded in @xmath4
so is @xmath164 .
now @xmath165 has bounded range and thus , by the completeness of @xmath113 , there is @xmath166 such that @xmath167 , @xmath149 . therefore @xmath158=[g ] = \varrho^{-1}(b)$ ] and thus @xmath142 \in m$ ] .
[ lem2 ] for any complete ultrafilter @xmath113 on @xmath1 , @xmath168 . given @xmath12
we will find @xmath142 \in n$ ] such that @xmath142 $ ] codes @xmath75 , i.e. , @xmath169 $ ] iff @xmath170 , for all @xmath33 .
let @xmath171 be the least code in @xmath4 of the bounded set @xmath172 .
since @xmath173 we have @xmath174 and thus @xmath171 also codes the set @xmath172 in @xmath145 .
it should be clear that @xmath110 is coded in @xmath2 .
for any @xmath175 we have @xmath176 iff @xmath177 and @xmath170 , and so @xmath178
thus , @xmath170 iff @xmath179 , i.e. , iff @xmath180 $ ] .
[ lem3 ] if @xmath113 is complete and definable then @xmath181 .
we show that @xmath182 ) \in { \mathscr{x}}$ ] for any @xmath142 \in n$ ] . by o s theorem @xmath183 ) = { \set{a \in m | { \set{b \in m | k \models a \in f(b)}}\in u } } .\ ] ] thus , if @xmath184 then @xmath182 ) = { \set{a \in m | ( a)_a \in u } } \in { \mathscr{x}}$ ] by the definability of @xmath113 .
given @xmath0 and @xmath113 a definable and complete ultrafilter on @xmath1 and a theory @xmath63 coded in @xmath2 such that @xmath121 , let @xmath6 be @xmath130 where @xmath145 is as above . by the preceding lemmas @xmath84 , @xmath157 and @xmath122 .
now we are ready to prove the central theorem of this paper , theorem [ maintheorem ] .
let us recall it .
let @xmath2 be a scott set that carries a definable ultrafilter and let @xmath185 be a consistent completion of pa .
then there is a recursively saturated model of @xmath3 with standard system @xmath2 . if @xmath2 is a scott set admitting a definable ultrafilter then @xmath186 .
any ultrafilter on @xmath2 is complete so theorem [ mainlemma ] gives us a model @xmath84 with @xmath187 .
an application of theorem [ wilmers ] gives a recursively saturated model of @xmath3 with standard system @xmath2 .
a natural question to ask is whatever every @xmath188 where @xmath26 is a strong cut in @xmath4 admits a complete definable nonprincipal ultrafilter .
this is not the case ; enayat , in @xcite , constructed a scott set @xmath189 of cardinality @xmath10 with no nonprincipal definable ultrafilter .
however , by theorem [ union ] and [ wilmers ] there is a recursively saturated @xmath190 such that @xmath62 . that @xmath9 is strong in @xmath4 follows from the fact that @xmath191
theorem [ maintheorem ] partly reduces the question of constructing models of arithmetic with standard system @xmath2 to the question when we can find definable ultrafilters on @xmath1 .
one way of constructing these ultrafilters is by finding generic ultrafilters . *
a set @xmath193 is dense in @xmath102 if for every @xmath194 there is @xmath195 such that @xmath196 . *
if @xmath197 is a set of dense sets in @xmath102 a filter @xmath11 is _ @xmath197-generic _ if @xmath198 for every @xmath199 . *
a filter on @xmath1 is generic if it is @xmath200-generic , where @xmath200 is the collection of all dense sets ( parameter ) definable in @xmath46 .
a partial order @xmath100 is said to have the countable chain condition ( c.c.c . for short )
if for every uncountable set @xmath202 there are @xmath203 such that @xmath204 and @xmath205 are compatible , i.e. , there is @xmath206 such that @xmath207 and @xmath208 .
martin s axiom , ma for short , says that for any partial order @xmath102 with the c.c.c . and any collection @xmath197 of dense sets of cardinality @xmath209 there is a @xmath197-generic filter on @xmath102 .
clearly , @xmath210 , but it is also the case that if @xmath211 is consistent then so is @xmath212 .
in fact , if @xmath211 is consistent and @xmath213 is regular such that @xmath214 , then @xmath215 is consistent .
if @xmath216 holds , @xmath217 is an arithmetically closed scott set , i.e. , @xmath191 , such that @xmath1 has the c.c.c .
and @xmath63 is a consistent theory coded in @xmath2 then there is a recursively saturated @xmath84 such that @xmath187 . for every @xmath218 ,
let @xmath219 be the set of codes of initial segments of @xmath75 , i.e. , @xmath220 , where @xmath221 is the least code in @xmath4 of the bounded set @xmath222 .
it should be clear that if @xmath223 then @xmath224 .
also if @xmath225 then @xmath226 is bounded and thus @xmath227 and @xmath228 are incompatible elements in @xmath201 . therefore , the set @xmath229 is an anti - chain in @xmath201 of the same cardinality as @xmath2 .
there might still be some hope of at least proving the consistency of @xmath230 , where ` @xmath231 ' denotes the sentence saying that every arithmetically closed scott set is the standard system of some model of pa .
this might be done by forcing generic ultrafilters into a model of @xmath232 . however , for this mission to succeed we have to control the cardinals while doing the forcing .
the following result of enayat @xcite is somewhat discouraging in that direction .
recently gitman @xcite used similar ideas to show that assuming the proper forcing axiom ( pfa ) any proper arithmetically closed scott set is a standard system .
gitman also proved the consistency of @xmath233 together with the existence of an arithmetically closed proper scott sets of size @xmath10 .
however it is , to our knowledge , open if pfa is consistent with the existence of such a scott set . | let @xmath0 be such that @xmath1 , the collection of all unbounded sets in @xmath2 , admits a definable complete ultrafilter and let @xmath3 be a theory extending first order arithmetic coded in @xmath2 such that @xmath4 thinks @xmath3 is consistent .
we prove that there is an end - extension @xmath5 of @xmath4 such that the subsets of @xmath4 coded in @xmath6 are precisely those in @xmath2 . as a special case
we get that any scott set with a definable ultrafilter coding a consistent theory @xmath3 extending first order arithmetic is the standard system of a recursively saturated model of @xmath3 .
the standard system of a model @xmath4 of pa ( the first order formulation of peano arithmetic ) is the collection of standard parts of the parameter definable subsets of @xmath4 , i.e. , sets of the form @xmath7 , where @xmath8 is a parameter definable set of @xmath4 , and @xmath9 is the set of natural numbers .
it turns out that the standard system tells you a lot about the model ; for example , any two countable recursively saturated models of the same completion of pa with the same standard system are isomorphic .
a natural question to ask is then which collections of subsets of the natural numbers are standard systems .
this problem has become known as the _ scott set problem_. for countable models the standard systems are exactly the countable scott sets , i.e. , countable boolean algebras of sets of natural numbers closed under relative recursion and a weak form of knig s lemma @xcite .
it follows , by a union of chains argument ( see @xcite ) , that for models of cardinality at most @xmath10 the standard systems are exactly the scott sets of cardinality at most @xmath10 .
if the continuum hypothesis holds this settles the scott set problem .
however , if ch fails then very little is known about standard systems of models of cardinality strictly greater than @xmath10 , although it is easy to see that any standard system of any model is a scott set ( see @xcite ) . the key notion of a definable
ultrafilter plays a central role in this paper .
suppose @xmath2 is a collection of sets of natural numbers , and @xmath1 is the collection of infinite members of @xmath2 .
a filter @xmath11 on @xmath1 is said to be definable if for all @xmath12 @xmath13 , where @xmath14 and @xmath15 is some canonical pairing function .
the central theorem of this paper is theorem [ maintheorem ] below ( whose proof will be presented in section [ sec : con ] ) .
[ maintheorem ] let @xmath2 be a scott set that carries a definable ultrafilter and let @xmath16 be a consistent completion of pa . then there is a recursively saturated model of @xmath3 with standard system @xmath2 .
it should be noted that all scott sets that carry a definable ultrafilter are arithmetically closed ( see lemma [ lem : def.ar ] ) and that any countable arithmetically closed scott set carries a definable ultrafilter , thus our result partly extends scott s theorem ( theorem [ thm : scott ] below ) .
our hope is that this paper may shed some light on the scott set problem for uncountable standard systems .
we will end the paper by offering some hints as to how this might be done . some of the results in this paper can also be found in the author s doctoral thesis @xcite . |
some extensions of the standard model provide dark matter ( dm ) candidate particles with sub - gev mass ; in the following these candidates will be indicated as light dark matter ( ldm ) .
light dark matter particles have been considered for example in warm dark matter scenarios such as e.g. kev - scale sterile neutrino @xcite , axino or gravitino @xcite . in addition , mev - scale particles ( e.g. axino @xcite , gravitino @xcite , heavy neutrinos @xcite , moduli fields from string theories @xcite , _ elko _ fermions @xcite ) have been proposed as dark matter and as source of 511 kev gamma s from the galactic center , due either to dm annihilation @xcite or to decay @xcite in the bulge . moreover , also supersymmetric models exist where the lsp naturally has a mev - scale mass and the other phenomenological properties , required to generate the 511 kev gammas in the galactic bulge @xcite . in this paper the direct detection of ldm candidate particles is investigated considering the possible inelastic scattering channels either on the electron or on the nucleus target .
we note , in fact , that since the kinetic energy for ldm particles in the galactic halo does not exceed hundreds ev the elastic scattering of such ldm particles both on electrons and on nuclei yields energy releases well below the energy thresholds of the detectors used in the field ; this prevents the exploitation of the elastic scattering as detection approach for these candidates .
thus , the inelastic process is the only possible exploitable one for the direct detection of ldm .
the following process is , therefore , considered for detection ( see fig.[fg : diagram ] ) : the ldm candidate ( hereafter named @xmath12 with mass @xmath13 and 4-momentum @xmath14 ) interacts with the ordinary matter target , @xmath15 , with mass @xmath16 and 4-momentum @xmath17 .
the target @xmath15 can be either an atomic nucleus or an atomic electron depending on the nature of the @xmath12 particle interaction . as result of the interaction a lighter particle is produced ( hereafter @xmath18 with mass @xmath19 and 4-momentum @xmath20 ) and the target recoils with an energy @xmath21 , which can be detectable by suitable detectors . [ !
ht ] the lighter particle @xmath18 is neutral and it is required that it interacts very weakly with ordinary matter ; thus , the @xmath18 particle escapes the detector . in particular
, the @xmath18 particle can also be another dm halo component ( dominant or subdominant with respect to the @xmath12 one ) , or it can simply be a standard model particle ( e.g. @xmath18 can be identified with an active neutrino in the scenario of the mev axino of ref .
@xcite , where the diagrams @xmath22 have been considered ) .
the production mechanism of the @xmath12 ( and eventually of the @xmath18 ) in the early universe are beyond the scope of the present paper and can be investigated in the future ; however , since @xmath23 , it is important to require that the lifetime of the possible decay of the @xmath12 particle is longer than the age of the universe . in particular , we note that the detection process of fig.[fg : diagram ] also offers a possible ( unavoidable ) decay channel of @xmath12 into @xmath18 and @xmath24 pair if @xmath25 .
finally , we remind that a ldm particle detection has already been discussed in ref .
@xcite , where the possible light bosonic ( kev mass axion - like particles ) dark matter candidate has been studied , considering the total conversion of its mass into electromagnetic radiation . in the present case , because of the presence of the @xmath18 particle in the final state , the ldm can be either a boson or a fermion . in the following ,
the direct detection of ldm candidate particles is investigated , some theoretical arguments are developed and related phenomenological aspects are discussed .
in particular , the impact of these dm candidates will also be discussed in a phenomenological framework on the basis of the 6.3 @xmath26 c.l .
dama / nai model independent evidence for particle dark matter in the galactic halo @xcite .
we remind that various corollary analyses , considering some of the many possible astrophysical , nuclear and particle physics scenarios , have been analysed by dama itself @xcite , while several others are also available in literature , such as e.g. refs . @xcite .
many other scenarios can be considered as well . at present the new second generation dama / libra set - up is running at the gran sasso national laboratory of the i.n.f.n ..
in this section the recoil energy , @xmath21 , of the target @xmath15 for the process of fig.[fg : diagram ] is evaluated . considering a lorentz transformation with @xmath27 ( hereafter @xmath28 , @xmath29 , @xmath30 and @xmath31
are the time components of the respective 4-momenta in the laboratory frame , see fig.[fg : diagram ] ) , it is possible to evaluate the energy conservation relation in the center of mass ( cm ) frame : @xmath32 knowing that @xmath33 , one obtains : @xmath34 defining the lorentz boost factor : @xmath35 one can write the total energy of the target in the laboratory frame after the scattering by means of a lorentz transformation : @xmath36 assuming the target at rest before the scattering , i.e. @xmath37 and @xmath38 , and the non - relativistic nature of the ldm particle , i.e. @xmath39 and @xmath40 , with @xmath41 , the energy released to the target @xmath15 : ( that is , the target recoil energy ) can be written as : @xmath42 where : i ) @xmath43 is the scattering angle in the cm frame ; ii ) @xmath44 is the average energy ; iii ) @xmath45 is the minimum recoil energy ; iv ) @xmath46 is the maximum recoil energy . considering the non - relativistic nature of the ldm , one gets : @xmath47 , and for the @xmath48 average energy : @xmath49 there @xmath50 is the mass splitting of the @xmath12-@xmath18 particle system and @xmath51 is the average mass of the @xmath12-@xmath18 particle system .
the spread of the recoil energy : @xmath52 can be appreciable only for hundreds mev ldm interacting on heavy nucleus targets . as an example , assuming @xmath53 mev and @xmath54 mev , the spread in energy is few @xmath55 both for electrons and for every nuclear targets . therefore , since the ldm kinetic energy is expected to be negligible with respect to @xmath56 , one has : @xmath57 , and the recoil energy is approximatively fixed at the @xmath58 value . [ !
t ] vs @xmath13 ( shaded areas ) , corresponding to released energies in nai(tl ) within the energy interval 1 - 6 kev electron equivalent .
the upper area is due to ldm interaction on na and i nuclear targets , while the lower area to ldm interaction on electron target .
the dashed line ( @xmath59 ) marks the case where @xmath18 is a massless particle .
the configurations characterized by @xmath60 ( dark area ) are of interest for the positron annihilation line from the galactic center through the decay : @xmath61 .
the thresholds of the possible annihilation processes : @xmath62 ( solid vertical line at @xmath13 = 511 kev ) ; @xmath63 ( solid curve ) ; @xmath64 ( solid curve ) ; @xmath65 ( dotted curve ) are shown . , title="fig : " ] in order to offer to the reader just a view of the detectability of ldm particles , fig .
[ fg : reg ] shows examples of the configurations in the plane @xmath56 vs @xmath13 ( shaded areas ) , corresponding to released energies in nai(tl ) within the energy interval 1 - 6 kev electron equivalent . the lower area in fig .
[ fg : reg ] refers to ldm interacting on electron targets , while the upper area refers to ldm interacting on the na and i targets ( the effect of the quenching factor @xcite and of the channeling in the nai lattice @xcite have been considered ) .
the dashed line @xmath59 marks the configurations where @xmath18 is a massless particle ( or also a very light particle , such as e.g. an active neutrino or a nearly massless sterile one or the light axion , etc . ) .
the configurations characterized by @xmath60 ( dark area ; hereafter @xmath66 is the electron mass ) are of interest for the positron annihilation line from the galactic center due to possible ( energetically allowed ) decay : @xmath61
. finally , the configurations on the right of the vertical line at @xmath13 = 511 kev are of interest for the possible annihilation process : @xmath62 in the galactic center , while the configurations of @xmath13 on the right of the solid or dotted curves are of interest for the possible annihilation processes : @xmath63 ( solid curve ) , @xmath64 ( solid curve ) , and @xmath65 ( dotted curve ) in the galactic center .
the interaction rate of the ldm particle with the target @xmath15 and for the studied process can be written as : @xmath67 where : i ) @xmath68 , with @xmath69 local halo density and @xmath70 fractional amount of @xmath12 density in the halo ; ii ) @xmath71 is the target number density in the detector ; iii ) @xmath72 is the differential cross section of the considered ldm particle inelastic scattering with the target @xmath15 ; iv ) @xmath73 is the velocity ( @xmath74 ) distribution of the @xmath12 particles in the earth ( laboratory ) frame .
since the sub - gev ldm wavelength ( @xmath75 fm ) is much larger than the nucleus size , the targets can be considered as point - like and the form factors of the targets can be approximated by unity .
in such a reasonable hypothesis and assuming , for simplicity , the isotropy of the differential cross section value . ] , one gets : @xmath76 where : i ) @xmath77 is the target cross section ; ii ) the heaviside functions , @xmath78 , define the domain of the differential cross section .
we note that @xmath46 and @xmath45 depend on the target mass and slightly on the ldm velocity , @xmath79 .
generally , @xmath77 can be a function of @xmath79 , depending on the peculiarity of the particle interaction . in the following ,
we adopt a widely considered approximation for the non - relativistic case @xcite ( the subscript @xmath15 drops ) : @xmath80 there @xmath81 and @xmath82 are constant and depend on the peculiarity of the particle interaction with the target @xmath15 ; they will be considered as free parameters in the description of the process . in the following , in order to deal the direct detection process with more usual parameters , the cross sections @xmath83 and @xmath84 will be used as free parameters ; they are related to the @xmath81 and @xmath82 parameters rescaled with the dark matter local velocity , @xmath85 , @xcite . thus , eq . ( [ eq : rate ] ) ,
becomes : @xmath86 since : @xmath87 , the total rate @xmath88 can be written as : @xmath89 similarly as the light bosonic case discussed in ref .
@xcite , the @xmath90 term provides an annual modulation of the expected counting rate for ldm interactions .
in fact , the velocity of the ldm particle in the galactic frame can be defined as : @xmath91 .
thus , one obtains for non rotating halo : @xmath92 .
since the earth velocity in the galactic frame ( @xmath93 ) is given by the sum of the sun velocity ( @xmath94 ) and of the earth s orbital time - dependent velocity around the sun ( @xmath95 ) , neglecting the @xmath96 term one gets : @xmath97 in the last equation the angle of the terrestrial orbit with respect to the galactic plane has been considered to be @xmath98 .
moreover , we consider @xmath99 with @xmath100 year , and the phase @xmath101 ( which corresponds to @xmath102 2nd june , that is , when the earth s speed in the galactic frame is at the maximum ) . the sun velocity can be written as @xmath103 km / s , where @xmath85 is the local velocity , whose value is in the range 170 - 270 km / s @xcite .
the earth s orbital velocity is @xmath104 km / s .
finally , @xmath105 depends on the halo model and on the @xmath85 value ( just for reference , in the particular simplified case of isothermal halo model : @xmath106 ) . in conclusion ,
the expected signal is given by the sum of two contributions : one independent on the time and the other one dependent on the time through , with relative amplitude depending on the adopted scenario .
for atomic electron targets , the energy interval @xmath107 is always very narrow @xmath108 .
therefore , using the approximation @xmath109 , the interaction rate with atomic electrons can be simply written as : @xmath110 where @xmath111 is the electron number density in the target detector and @xmath112 and @xmath113 are the average ( over all the atomic electrons ) cross section parameters for electrons in the target material .
after the interaction the final state can have beyond the @xmath18 particle either a prompt electron and an ionized atom or an excited atom plus possible x - rays / auger electrons .
therefore , the process produces x - rays and electrons of relatively low energy , which are mostly contained with efficiency @xmath114 in a detector of a suitable size .
therefore , the differential counting rate can simply be written accounting for the detector energy resolution , by means of the @xmath115 kernel , which generally has a gaussian behaviour : @xmath116 there @xmath117 is the energy resolution of the detector and @xmath118 is the area of the gaussian peak centered at the @xmath44 value .
finally , the expected differential rate is given by the sum of two contributions : @xmath119 , where @xmath120 and @xmath121 are the unmodulated and the modulated part of the expected differential counting rate , respectively . as regards the interaction with target nuclei ,
when considering the cases of heavy target and for relatively `` high '' ldm mass the energy interval @xmath107 can not be neglected .
therefore , the interaction rate with the nucleus @xmath15 can be approximated as : @xmath122 where @xmath71 is the nucleus number density in the material and @xmath123 are the @xmath124 values for the nucleus @xmath15 calculated at @xmath125 .
considering the case of a crystal detector target ( as e.g. the nai(tl ) ) , the possible channeling effect @xcite has also to be taken into account .
the detection of a nuclear recoil of kinetic energy , @xmath21 , is , in particular , related to the response function @xmath126 @xcite , where @xmath127 is the released energy in kev electron equivalent .
thus , the differential @xmath127 distribution for the nucleus @xmath15 when accounting for the channeling effect can be written as : @xmath128 hence , the expected differential counting rate for ldm interaction with nucleus @xmath15 can easily be derived accounting for the detector energy resolution kernel @xmath129 : @xmath130 finally , as for the electron interacting ldm , the expected differential rate is given by the sum of two contributions : @xmath131 , where @xmath120 and @xmath121 are the unmodulated and the modulated part of the expected differential counting rate , respectively .
in this section as template purpose , some examples on interaction rate of ldm particles in nai(tl ) detectors are given .
in particular , since the expected energy distribution for electron kg@xmath132 in nai(tl ) . ] interacting ldm has a simple gaussian behaviour ( see eq .
( [ eq : rate333 ] ) ) , only some examples for the nucleus interacting ldm will be shown in the following . in nai(tl )
detectors , both na and i nuclei ( @xmath133 kg@xmath132 ) can contribute to the detection of ldm .
therefore , the total expected counting rate is given by the sum of two contributions similar as eq .
( [ eq : labelmul ] ) , arising from the two nuclei species ( t = na , i ) . the cross section parameters @xmath134 and @xmath135 for na and i
are related with some model dependent scaling laws ; in particular , in the following , the two illustrative cases of coherent and incoherent ( @xmath136 ) nuclear scaling laws are investigated .
obviously , other different nuclear scaling laws are in principle possible . [ !
ht ] mev on the target nuclei .
_ left : _ case of coherent cross section scaling laws ; here @xmath137 mev and @xmath138 pb . _ right : _ case of incoherent cross section scaling laws ; here @xmath139 mev and @xmath140 pb . the a5 halo model ( a nfw halo model with local velocity equal to 220 km / s and density equal to the maximum value , see ref .
@xcite ) has been considered .
the quenching factors have been assumed as the case a of ref .
the channeling effect has been included ; see text .
the vertical dotted lines correspond to the energy threshold of the nai(tl ) detectors used in dama / nai set - up.,title="fig : " ] mev on the target nuclei .
_ left : _ case of coherent cross section scaling laws ; here @xmath137 mev and @xmath138 pb . _ right : _ case of incoherent cross section scaling laws ; here @xmath139 mev and @xmath140 pb . the a5 halo model ( a nfw halo model with local velocity equal to 220 km / s and density equal to the maximum value , see ref .
@xcite ) has been considered .
the quenching factors have been assumed as the case a of ref .
the channeling effect has been included ; see text .
the vertical dotted lines correspond to the energy threshold of the nai(tl ) detectors used in dama / nai set - up.,title="fig : " ] as template purpose , in fig .
[ fg : spectra_dm ] , the expected energy distributions of the unmodulated / modulated part of the differential rate in nai(tl ) are shown for interactions of ldm with @xmath141 mev on the target nuclei .
in particular , there are shown the cases of coherent cross section scaling laws ( fig .
[ fg : spectra_dm ] _ left _ ) with @xmath137 mev and @xmath138 pb and the case of incoherent cross section scaling laws ( fig .
[ fg : spectra_dm ] _
right _ ) with @xmath139 mev and @xmath140 pb . the a5 halo model ( a nfw halo model with local velocity equal to 220 km / s and density equal to the maximum value , see ref .
@xcite ) has been considered .
the quenching factors have been assumed as the case a of ref . @xcite and
the channeling effect has been included , as mentioned .
it is worthwhile to note that similar behaviors can also be obtained by using other choices of the halo model , quenching factors and values of the masses of the involved particles in the interaction .
the 6.3 @xmath26 c.l . model independent evidence for dark matter particles in the galactic halo achieved over seven annual cycles by dama / nai @xcite ( total exposure @xmath142 kg @xmath143 days ) can also be investigated for the case of a ldm candidate ( in addition to the other corollary quests analyzed by dama itself @xcite and available in literature , e.g. refs .
@xcite ) . in the following , the same dark halo models and related parameters given in table vi of ref .
@xcite are considered .
the related dm density is given in table vii of the same reference .
note that , although a large number of self - consistent galactic halo models have been considered , still many other possibilities exist . as regards the case of ldm interacting on nuclei ,
also the na and i quenching factors @xcite and the na and i channeling response functions given in @xcite have been considered . the uncertainties on the na and i quenching factors have been taken into account as done in ref .
in addition , the presence of the existing migdal effect and of the possible sagdeg contributions we discussed in @xcite , respectively will be not included here for simplicity .
as regards the cross section parameters , the @xmath144 terms do not contribute to the annual modulation of the signal .
in particular , the measured modulation amplitude easily allows to derive @xmath145 . in the following , for simplicity
, the contributions of the @xmath146 terms are assumed to be dominant with respect to the contributions of the @xmath144 ones .
the results are calculated by taking into account the time and energy behaviours of the _ single - hit _ particle is no more present after the inelastic interaction and , therefore , can only provide _ single - hit _ events in a multi - detectors set - up .
] experimental data through the standard maximum likelihood method , where @xmath147 is the number of events collected in the @xmath148-th time interval , by the @xmath149-th detector and in the @xmath150-th energy bin .
@xmath147 follows a poissonian distribution with expectation value @xmath151 m_j \delta t_i \delta e \epsilon_{jk}$ ] .
the unmodulated and modulated parts of the signal , @xmath152 and @xmath153 , respectively , are here functions of the ldm mass @xmath13 , the splitting @xmath56 and the cross section @xmath146 .
the b@xmath154 is the background contribution ; @xmath155 is the detector running time during the @xmath148-th time interval ; @xmath156 is the energy bin ; @xmath157 is the overall efficiency and @xmath158 is the detector mass . ] . in particular , they are presented in terms of slices of the three - dimensional allowed volume ( @xmath13 , @xmath56 , @xmath159 ) where @xmath146 is alternatively @xmath160 , @xmath161 and @xmath162 obtained as superposition of the configurations corresponding to likelihood function values _
distant _ more than @xmath163 from the null hypothesis ( absence of modulation ) in each one of the several ( but still a very limited number ) of the considered model frameworks . in this way one
accounts for at least some of the existing theoretical and experimental uncertainties .
it is worth to note that the inclusion of other existing uncertainties would further extend the allowed volumes / regions and increase the sets of obtained best fit values .
the projection of the whole 4@xmath26 allowed volume on the plane ( @xmath13 , @xmath56 ) gives typical patterns similar as those reported in fig.[fg : reg ] .
moreover , since the two regions are disconnected , the ldm detection is always dominated by only one of the different target contributions .
therefore , in the following , slices of the allowed volume either for electron interacting ldm or nucleus interacting ldm are presented separately . in case of electron interacting ldm ,
the projection of the 4@xmath26 allowed volume on the plane ( @xmath13 , @xmath56 ) for the same dark halo models and parameters described in ref .
@xcite is reported in fig.[fg : elett1 ] . [ !
ht ] allowed 3-dimensional volume on the plane ( @xmath13 , @xmath56 ) for the same dark halo models and parameters described in ref .
@xcite;see text .
the dashed line ( @xmath59 ) marks the case where @xmath18 is a massless particle .
the decay through the detection channel , @xmath164 , is energetically not allowed for the selected configurations .
the configurations with @xmath165 ( dark area ) are interesting for the possible annihilation processes : , @xmath63 , @xmath64 , and @xmath65 in the galactic center .
the three nearly vertical curves are the thresholds of these latter processes as mentioned in the caption of fig.[fg : reg ] and in the text.,title="fig : " ] the allowed @xmath13 values and the splitting @xmath56 are in the intervals for values of @xmath13 greater than o(gev ) , the definition of light dark matter is no more appropriate .
moreover , the kinetic energy of the particle would be enough for the detection also through the elastic scattering process , as demonstrated in ref .
and , respectively .
it is worth to note that in such a case the decay through the detection channel : @xmath164 , is energetically forbidden for the given @xmath56 range .
the configurations with @xmath166 kev ( dark area in fig.[fg : elett1 ] ) are instead of interest for the possible annihilation processes : , @xmath63 , @xmath64 , and @xmath65 , in the galactic center . as examples ,
some slices of the 3-dimensional allowed volume for various @xmath13 values in the ( @xmath167 vs @xmath56 ) plane are depicted in fig.[fg : elett]_left _ in the scenario given above . [ !
ht ] allowed 3-dimensional volume for various @xmath13 depicted in the ( @xmath167 vs @xmath56 ) plane .
_ right : _ slice of the 4@xmath26 allowed 3-dimensional volume for @xmath59 , that is for a massless or a very light @xmath18 particle , as e.g. either an active neutrino or a nearly massless sterile one or the light axion , etc . the same dark halo models and parameters described in ref .
@xcite have been used ; see text . , title="fig : " ] the slice of the 4@xmath26 allowed 3-dimensional volume for @xmath59 is shown in fig.[fg : elett]_right_. this slice has been taken along the dotted line of fig.[fg : elett1 ] , restricting @xmath168 , that is for a massless or a very light @xmath18 particle , such as e.g. either an active neutrino or a nearly massless sterile one or the light axion , etc .
finally , it is worth to summarize that electron interacting ldm candidates in the few - tens - kev / sub - mev range are allowed ( see figs.[fg : elett1 ] and [ fg : elett ] ) .
this can be of interest for example in the models of warm dark matter particles such as e.g. weakly sterile neutrino @xcite , axino or gravitino @xcite or in the models where the dm is made of moduli fields @xcite .
moreover , also configurations with @xmath13 in the mev / sub - gev range are allowed ; similar ldm candidates ( such as e.g. axino @xcite , sterile neutrino @xcite , moduli fields from string theories @xcite , and even mev - scale lsp in supersymmetric theories @xcite ) can also be of interest for the production mechanism of the 511 kev gammas from the galactic bulge @xcite . in case of nucleus interacting ldm ,
the projections of the 4@xmath26 allowed volumes on the plane ( @xmath13 , @xmath56 ) are reported in fig.[fg : regn ] for the two above - mentioned illustrative cases of coherent ( _ left _ ) and incoherent ( _ right _ ) nuclear scaling laws .
they are evaluated for the same dark halo models and parameters described in ref .
@xcite .
ht ] allowed 3-dimensional volumes on the plane ( @xmath13 , @xmath56 ) for the two above - mentioned illustrative cases of coherent ( _ left _ ) and incoherent ( _ right _ ) nuclear scaling laws .
they are evaluated for the same dark halo models and parameters described in ref .
the regions enclose configurations corresponding to likelihood function values _ distant _ more than @xmath163 from the null hypothesis ( absence of modulation ) .
the dashed lines ( @xmath59 ) mark the case where @xmath18 is a massless particle .
the decays through the diagram involved in the detection channel are energetically forbidden .
see text . , title="fig:",width=226 ] allowed 3-dimensional volumes on the plane ( @xmath13 , @xmath56 ) for the two above - mentioned illustrative cases of coherent ( _ left _ ) and incoherent ( _ right _ ) nuclear scaling laws .
they are evaluated for the same dark halo models and parameters described in ref .
the regions enclose configurations corresponding to likelihood function values _ distant _ more than @xmath163 from the null hypothesis ( absence of modulation ) .
the dashed lines ( @xmath59 ) mark the case where @xmath18 is a massless particle .
the decays through the diagram involved in the detection channel are energetically forbidden .
, title="fig:",width=226 ] the allowed @xmath13 values and the splitting @xmath56 are in the intervals values greater than o(gev ) the detection would also be possible through the elastic scattering process @xcite . ] and , respectively ( see fig.[fg : regn ] ) .
it is worth to note that in such a case the decays through the diagram involved in the detection channel ( e.g. in nucleon anti - nucleon pairs or in meson(s ) , as @xmath169 ) are obviously energetically forbidden .
moreover , there are allowed configurations that could contribute in principle , if suitable couplings exist to the positron generation in the galactic center ; in fact , the decay @xmath164 is energetically allowed for @xmath170 ( dark area in fig.[fg : regn ] ) , while the annihilation processes into @xmath171 pairs are energetically allowed for almost all the allowed configurations .
it is worth to note that for nuclear interacting ldm the 3-dimensional allowed configurations are contained in two disconnected volumes , as seen e.g. in their projections in fig .
[ fg : regn ] .
the one at larger @xmath56 at @xmath13 fixed is mostly due to interaction on iodine target , while the other one is mostly due to interaction on sodium target . as examples ,
some slices of the 3-dimensional allowed volumes for various @xmath13 values in the ( @xmath159 vs @xmath56 ) plane are depicted in fig.[fg : panel_n ] for the two above - mentioned illustrative cases of coherent ( _ left _ panel ) and incoherent ( _ right _ panel ) nuclear scaling laws . [ !
ht ] values in the ( @xmath159 vs @xmath56 ) plane for the two above - mentioned illustrative cases of coherent ( _ left _ ) and incoherent ( _ right _ ) nuclear scaling laws .
the 3-dimensional volumes enclose configurations corresponding to likelihood function values _ distant _ more than @xmath163 from the null hypothesis ( absence of modulation ) . the same dark halo models and parameters described in ref . @xcite
have been used.,title="fig : " ] values in the ( @xmath159 vs @xmath56 ) plane for the two above - mentioned illustrative cases of coherent ( _ left _ ) and incoherent ( _ right _ ) nuclear scaling laws .
the 3-dimensional volumes enclose configurations corresponding to likelihood function values _ distant _ more than @xmath163 from the null hypothesis ( absence of modulation ) . the same dark halo models and parameters described in ref .
@xcite have been used.,title="fig : " ] [ !
ht ] allowed 3-dimensional volumes for @xmath59 , that is for a massless or a very light @xmath18 particle , as e.g. either an active neutrino or a nearly massless sterile one or the light axion , etc .
they are evaluated for the two illustrative cases of coherent ( _ left _ panel ) and incoherent ( _ right _ panel ) nuclear scaling laws , using the same dark halo models and parameters described in ref .
, title="fig : " ] allowed 3-dimensional volumes for @xmath59 , that is for a massless or a very light @xmath18 particle , as e.g. either an active neutrino or a nearly massless sterile one or the light axion , etc .
they are evaluated for the two illustrative cases of coherent ( _ left _ panel ) and incoherent ( _ right _ panel ) nuclear scaling laws , using the same dark halo models and parameters described in ref .
, title="fig : " ] the slices of the 4@xmath26 allowed 3-dimensional volumes for @xmath59 are shown in fig.[fg : ster_n ] for the two illustrative cases of coherent ( _ left _ panel ) and incoherent ( _ right _ panel ) nuclear scaling laws .
these slices have been taken along the dotted lines of fig.[fg : regn ] , restricting @xmath168 , that is for a massless or a very light @xmath18 particle , as e.g. either an active neutrino or a nearly massless sterile one or the light axion , etc .
finally , it is worth to summarize that ldm candidates in the mev / sub - gev range are allowed ( see figs.[fg : regn ] , [ fg : panel_n ] and [ fg : ster_n ] ) .
also these candidates , such as e.g. axino @xcite , sterile neutrino @xcite , moduli fields from string theories @xcite , and even mev - scale lsp in supersymmetric theories @xcite , can be of interest for the positron production in the galactic center @xcite .
in this paper , the possibility of direct detection of a light dark matter candidate particle has been investigated .
the inelastic scattering processes on the electron or on the nucleus targets have been considered ; these are the only possible processes useful for the direct detection of such ldm candidates .
some theoretical arguments have been developed and related phenomenological aspects have been discussed . in particular ,
the impact of the ldm candidate has also been analyzed in a phenomenological framework on the basis of dama / nai annual modulation data . allowed volumes and regions for the characteristic phenomenological parameters
have been derived in the considered model .
the allowed phenomenological parameters ( masses and cross sections ) of ldm can be of interest for various ldm candidates proposed in theories beyond the standard model and for the production mechanism of the 511 kev gammas from the galactic bulge .
in conclusion this paper has shown that
in addition to other candidates , as wimp / wimp - like particles and axion - like bosons , already discussed by dama collaboration elsewhere @xcite
there is also possibility for a ldm candidate interacting either with the electrons or with the nuclei to account for the 6.3@xmath26 c.l .
model independent evidence for the presence of a particle dm component in the galactic halo .
see e.g. : a. kusenko , aip conf .
( 2007 ) 58 ; + a. palazzo et al . , phys .
d 76 ( 2007 ) 103511 ; + m. shaposhnikov , nucl . phys .
b 763 ( 2007 ) 49 ; + r. volkas , prog . part .
phys . 48 ( 2002 ) 161 . c. boehm and y. ascasibar , phys
d 70 ( 2004 ) 115013 ; + y. ascasibar et al . , mon . not .
r. astron .
368 ( 2006 ) 1695 ; + c. jacoby and s. nussinov jhep 05 ( 2007 ) 17 ; + p. fayet phys . rev .
d 70 ( 2004 ) 023514 .
r. bernabei et al . , int . j. mod
. phys . a 21 ( 2006 ) 1445 .
r. bernabei el al . , la rivista del nuovo cimento 26 n.1 ( 2003 ) 1 - 73 .
r. bernabei el al . , int . j. mod
d 13 ( 2004 ) 2127 .
r. bernabei et al .
j. c. 47 ( 2006 ) 263 .
r. bernabei el al . , int . j. mod
a 22 ( 2007 ) 3155 - 3168 .
r. bernabei et al .
j. c 53 ( 2008 ) 205 .
r. bernabei el al . , phys .
d 77 ( 2008 ) 023506 .
a. bottino et al .
d 67 ( 2003 ) 063519 ; + a. bottino et al . , phys . rev .
d 68 ( 2003 ) 043506 .
a. bottino et al . , phys .
d 69 ( 2004 ) 037302 .
a. bottino et al . , phys .
b 402 ( 1997 ) 113 ; + phys . lett .
b 423 ( 1998 ) 109 ; + phys . rev .
d 59 ( 1999 ) 095004 ; + phys . rev .
d 59 ( 1999 ) 095003 ; + astrop . phys . 10
( 1999 ) 203 ; + astrop .
( 2000 ) 215 ; + phys . rev .
d 62 ( 2000 ) 056006 ; + phys . rev .
d 63 ( 2001 ) 125003 ; + nucl
b 608 ( 2001 ) 461 .
k. belotsky , d. fargion , m. khlopov and r.v .
konoplich , hep - ph/0411093 .
d. smith and n. weiner , phys .
d 64 ( 2001 ) 043502 ; + d. tucker - smith and n. weiner , phys . rev .
d 72 ( 2005 ) 063509 .
r. foot phys .
d 69 ( 2004 ) 036001 .
s. mitra , phys .
d 71 ( 2005 ) 121302(r ) .
drobyshevski et al . ,
astron . & astroph . trans .
26:4 ( 2007 ) 289 ; + e.m .
drobyshevski , arxiv:0706.3095 . c. arina and n. fornengo , arxiv:0709.4477 .
a. bottino et al . , arxiv:0710.0553 .
p. belli et al . , phys . rev .
d61 ( 2000 ) 023512 .
leonard and s. tremaine , _ astrophys .
j. _ 353 ( 1990 ) 486
kochanek , _ astrophys .
j. _ 457 ( 1996 ) 228 ; + k.m .
cudworth , _ astron .
j. _ 99 ( 1990 ) 590 . | some extensions of the standard model provide dark matter candidate particles with sub - gev mass .
these light dark matter particles have been considered for example in warm dark matter scenarios ( e.g. the kev scale sterile neutrino , axino or gravitino ) .
moreover mev scale dm candidates have been proposed in supersymmetric models and as source of the 511 kev line from the galactic center . in this paper
the possibility of direct detection of a light dark matter candidate is investigated considering the inelastic scattering processes on the electron or on the nucleus targets .
some theoretical arguments are developed and related phenomenological aspects are discussed .
allowed volumes and regions for the characteristic phenomenological parameters of the considered scenarios are derived from the dama / nai annual modulation data .
# 1#2#3#4#1 # 2 ( # 3 ) # 4 * investigation on light dark matter + * r.bernabei@xmath0 , p.belli@xmath1 , f.cappella@xmath2 , r.cerulli@xmath3 , c.j.dai@xmath4 , h.l.he@xmath4 , a.incicchitti@xmath5 , h.h.kuang@xmath4 , j.m.ma@xmath4 , x.h.ma@xmath4 , f.montecchia@xmath0 , f.nozzoli@xmath0 , d.prosperi@xmath2 , x.d.sheng@xmath4 , z.p.ye@xmath6 , r.g.wang@xmath4 , y.j.zhang@xmath4 @xmath7_dip . di fisica , universit di roma `` tor vergata '' , i-00133 rome , italy _ @xmath1_infn , sez .
roma `` tor vergata '' , i-00133 rome , italy _ @xmath8_dip .
di fisica , universit di roma `` la sapienza '' , i-00185 rome , italy _
@xmath9_infn , sez .
roma , i-00185 rome , italy _
@xmath10_laboratori nazionali del gran sasso , i.n.f.n .
, assergi , italy _
@xmath4_ihep , chinese academy , p.o .
box 918/3 , beijing 100039 , china _
@xmath11_university of jing gangshan , jiangxi , china _ _ keywords : _ light dark matter ; underground physics _ pacs numbers : _ 95.35.+d |
in a previous paper @xcite we have discussed how functions of position , defined on a discrete lattice of time points , may be analytically interpolated in time by functions of both @xmath0 and @xmath1 , defined on a continuum of time points , framed by the lattice , through the use of solutions to schrder s nonlinear functional equation @xcite . for a more pedagogical discussion to augment this introduction , we encourage readers to look at that companion paper . if the effect of the first discrete time step is given as the map @xmath2 , for some parameter @xmath3 , schrder s functional equation is@xmath4 with @xmath5 to be determined .
so , @xmath6 where the inverse function @xmath7 obeys poincar s equation,@xmath8 the @xmath9th iterate of ( [ se ] ) gives @xmath10 with @xmath11 acting @xmath9 times , and thus the @xmath9th order functional composition the so - called splinter of the functional equation @xmath12 a continuous interpolation between the integer lattice of time points is then , for _ any _ @xmath1,@xmath13 this can be a well - behaved and single - valued function of @xmath0 and @xmath1 provided that @xmath14 is a well - behaved , single - valued function of @xmath0 , even though @xmath15 might be , and typically is , multi - valued . as discussed in @xcite
, the interpolation can be envisioned as the trajectory of a particle , @xmath16 where the particle is moving under the influence of a potential according to hamiltonian dynamics .
the velocity of the particle is then found by differentiating ( [ interpolation ] ) with respect to @xmath1,@xmath17 where any dependence of @xmath5 on @xmath3 is implicitly understood .
therefore , the velocity will inherit and exhibit any multi - valuedness possessed by @xmath18 . indeed , suppose that hamiltonian dynamics have been specified , trajectories have been computed , and @xmath19 has emerged in terms of initial velocity and initial position @xmath20 .
a solution of the functional equation ( [ se ] ) can then be constructed , and expressed in very physical terms , as just an exponential of _ the time elapsed along any such particle trajectory _ that passes through position @xmath0 .
that solution is@xmath21 here @xmath22 is the velocity as a function of the position along the trajectory .
( hence @xmath23 , if you wish . of course
, for a hamiltonian system with time translational invariance , we may supplant @xmath24 in all these expressions , as well as those to follow . ) for a particle moving in a potential @xmath25 at fixed energy , @xmath26 , the velocity can be expressed in the usual way as @xmath27 , with suitably chosen mass units . when written in a form more closely related to schrder s functional equation ,
the solution ( [ generalpsi ] ) is simply@xmath28 at @xmath29 , @xmath30 , and schrder s equation ( [ se ] ) re - emerges .
but , as this construction clearly shows , one must be careful at turning points where @xmath31 , especially if these are encountered at finite times along the trajectory .
typically these turning points produce branch points in @xmath5 so that it is multi - valued .
an interesting physical effect of such branch points , when they are encountered in finite times , is the possibility to switch from one branch of the underlying analytic potential function to another , thereby changing the functional form of the hamiltonian governing the motion along seemingly identical intervals of a real - valued trajectory .
we will therefore call these _ quasi_-hamiltonian systems , and we stress that when such switches occur this is _ not _ standard textbook hamiltonian dynamics . further explanation will be given below in the context of the logistic map .
the interpolation ( [ interpolation ] ) can also be viewed as a holographic specification on the @xmath32 plane @xcite , determining @xmath33 in the surrounded bulk from the total data given at the bounding times , @xmath34 .
in this point of view , _ fixed points _ in @xmath0 complete the boundary data and facilitate the solution of schrder s equation through power series in @xmath0 , hence leading to @xmath25 _ which was not known a priori _ @xcite . in this holographic approach
, the potential first appears as a quadratic in @xmath35,@xmath36 up to an additive constant , where any dependence of @xmath5 on @xmath3 is again implicit .
the @xmath0 dependence of the potential , @xmath37 follows from that of the velocity profile of the interpolation , defined and given by @xcite@xmath38 monotonic motions between two fixed points , such as occur for the ricker model @xcite , provide the most elementary examples @xcite .
however , situations where one or both of the fixed points are absent were not fully addressed in our previous work .
this is precisely the situation that occurs when turning points are encountered at finite times in the particle dynamics interpretation , and leads to an intriguing modification in the physical picture involving the potential and its effect on the particle trajectories .
we consider here a specific example of such a situation involving the well - studied logistic map of chaotic dynamics , namely _
@xcite_,@xmath39 applying our functional methods to this example , the resulting interpolations from a discrete lattice of time points to a time continuum then allow us to appreciate analytic features of the logistic map , and to derive the governing differential evolution laws indeed , subtly time - translation - invariant hamiltonian dynamical laws of the underlying physical system , hence to obtain potentials that were _ not _ previously known .
for the all - familiar logistic map illustrating transition to chaos , the functional interpolation reveals that there are well - defined expressions for the continuous time evolution of this map for _ all _ parametric values of the map , whether chaotic or not .
we obtain agreement with the explicit closed - form solutions of ( [ se ] ) for the special values of the parameters : @xmath40 and @xmath41 .
( these explicit solutions have been known for almost a century and a half @xcite . )
moreover , as indicated , from ( [ generalv ] ) we can now find the potentials needed to produce these explicit , continuously evolving trajectories in the language of hamiltonian dynamics .
a new feature for the potentials so obtained for the @xmath42 case in particular , and also for other values of @xmath43 , is that they must change at discrete intervals of the envisioned hamiltonian particle s motion , to be consistent with the evolution trajectories : _ every time the particle hits a turning point , the potential changes _
we therefore call the corresponding @xmath44s switchback potentials and to echo our previous general remarks , we will call the dynamical system as a whole quasi - hamiltonian . for @xmath42 the switchback potentials
deepen successively and thus lead to more rapid , higher frequency motion , resulting in the familiar chaotic behavior of the discrete @xmath42 logistic map that they interpolate .
the familiar frequency - cascade - and - folding behavior of the chaotic discrete map is thus understood from indeed , _ explained _ by the subtle properties of the switchback potentials .
consider in detail the logistic map ( [ logisticmap ] ) on the unit interval , @xmath45 $ ] .
schrder s equation for this map is @xmath46 the inverse function satisfies the corresponding poincar equation,@xmath47 as originally obtained by schrder , there are three closed - form solutions known , for @xmath48 @xmath49 and @xmath41:@xmath50 note that , while the @xmath5 are multi - valued , the inverse functions are all single - valued
. more generally , consider a power series for any @xmath3.@xmath51 the poincar equation then leads to a recursion relation for the @xmath3-dependent coefficients.@xmath52 with @xmath53 , @xmath54 , @xmath55 , etc .
the explicit coefficients are easily recognized for @xmath48 @xmath49 and @xmath41 , and immediately yield the three closed - form cases .
similarly let @xmath56 then , as a consequence of schrder s equation , @xmath57 , and for @xmath58,@xmath59 where @xmath60 is the ( integer - valued ) floor function . in principle , these series solve ( [ pfe ] ) and ( [ sfe ] ) for any @xmath3 , within their radii of convergence . from extensive numerical studies we believe the radius of convergence for ( [ psiseries ] ) depends on @xmath3 as follows.@xmath61{ccc}% \left\vert 1-\dfrac{1}{s}\right\vert & \text{if } & 0\leq s\leq2\ , \\ & & \\ \dfrac{s}{4 } & \text{if } & 2\leq s\leq4\ .
\right . \label{psiseriesradius}%\ ] ] preliminary work further suggests that these conjectural results can actually be established analytically by comparison with known convergent series , at least in some cases .
the radius of convergence for the inverse function series ( [ psiinverseseries ] ) is intriguing for @xmath62 . a fit by mathematica to radii obtained numerically from explicit series coefficients
@xmath63 up to @xmath64 , for 20 different values of @xmath3 in this range , using the functional form @xmath65 , gives @xmath66 and @xmath67 .
thus @xmath68 falls from @xmath69 to zero as @xmath3 goes from @xmath70 to @xmath71 .
however , for the parameters of most interest , namely @xmath72 , numerical studies suggest the series all have _ infinite _ radius of convergence , indicating the @xmath7 are entire functions for these parameters .
this is true for @xmath73 and @xmath42 , of course , as is evident from ( [ exactcases ] ) .
the numerics provide compelling evidence that this is also true for all intermediate @xmath3 .
given convergent series , the functional equations can then be used to continue the series solutions outside their radii of convergence and thereby obtain accurate determinations of the various branches of @xmath5 , and of the large - but - finite argument behavior of @xmath7 , in a manner familiar from , say , the @xmath74 and @xmath75 functions .
( see @xcite for details , and the appendix for a numerical example . )
comparison of the results from these series - plus - functional - equation methods with the known exact solutions ( [ exactcases ] ) shows perfect agreement .
the first few terms for @xmath5 and @xmath7 for generic @xmath3 are given explicitly by @xcite@xmath76 from which we infer that @xmath77 and @xmath78 are actually series in @xmath79 with @xmath3-dependent coefficients that are analytic near @xmath80 .
the trajectories interpolating the splinter ( integer @xmath1 ) of the logistic map are then,@xmath81 the trajectories are single - valued functions of the time so long as @xmath82 is single - valued , and in fact , they exist even for @xmath83 as formal series solutions .
@xmath84explicitly , @xmath85 for the three special cases , @xmath40 and @xmath41 , there are closed - form results for various quantities of interest .
for example , for @xmath42 , the trajectory and velocity are given by@xmath86 ( as originally presented in @xcite , p 306 ) and by @xmath87 the last expression evinces a continuous time - translational invariance .
however , this velocity function has branch points ( i.e. turning points ) at @xmath88 and @xmath89 , so some care is needed to determine which branch of the function is involved , particularly when the turning points are encountered at finite times .
the system is therefore a quasi - hamiltonian one , as we have previously indicated , and as we shall discuss in more detail . with this caveat in mind
, we may thus deduce the velocity profile @xmath90 , the effective potential , and the force for the underlying quasi - hamiltonian system : @xmath91 note @xmath92 is an unstable fixed point for the system .
also , the system is time - translationally invariant , so these same expressions hold at all times everywhere along a trajectory , _ provided _ due care is taken to determine which branch of the various functions is in effect .
thus @xmath93 , etc .
similar closed - form results hold for the other two solvable cases , @xmath94 and @xmath73 . in principle , there are also potentials underlying the trajectories for other @xmath3 by dint of the series constructions .
( we illustrate one other case , @xmath95 , in the appendix . ) from the general expression for @xmath96 in terms of @xmath5 and @xmath7 , and with use of the chain rule , the potential may be expressed entirely in terms of @xmath97 , as given in the introduction by ( [ generalv ] ) . the schrder auxiliary function
is then recognized as just an exponential of the time function @xmath98 computed along a zero - energy trajectory , as given by ( [ generalpsi ] ) .
the effective potentials for all three closed - form cases are somewhat unusual : @xmath99 another way to express the potential for @xmath42 is similar in form to that for @xmath94 , namely,@xmath100 indeed , it is well - known that the logistic maps for @xmath42 and @xmath94 are intimately related through the functional conjugacy of the underlying schrder equations . but note the potential for @xmath94 is in fact complex , since @xmath101 , as are the trajectories under real time evolution for this case . of course , since the complexity of @xmath102 is solely a multiplicative factor , if we switch to complex time , @xmath103 , then@xmath104 again yields real trajectories @xmath105 . as trivial as this observation is
, nevertheless it does raise several questions about complex @xmath0 , and about the behavior of @xmath44 in the complex plane .
we discuss only one aspect of this behavior here , and consider non - principal values for the multi - valued functions @xmath106 and @xmath107 .
in particular , _ all _ branches of the @xmath108 function are important to understand the behavior of the explicit trajectories ( [ exacttrajectory4 ] ) . *
* switchbacks on the road to chaos . *
* an interesting _ new _ feature , which we shall call the _ switchback effect _ , appears for a particle moving in the @xmath42 effective potential .
this is a distinguishing feature that we encounter for the @xmath42 chaotic map , but not , say , for the non - chaotic @xmath73 map .
the effect is traceable to the fact that , while the trajectory is single - valued as a function of the time , the velocity and hence the effective potential are _ not _ single - valued as functions of _ position_.
moreover , the branch points of the multi - valued functions are encountered by zero - energy trajectories at _ finite _ times for the @xmath42 case , unlike , say , for the @xmath73 case . for the latter case
, it takes an infinite time for a particle with zero energy to reach a turning point .
switchbacks are essentially transitions from one branch of the position - dependent velocity function to another , and occur at the turning points encountered at finite times by the interpolating particle trajectories , ( [ exacttrajectory4 ] ) . the effect is easily seen upon viewing animations of these trajectories @xcite .
consider a particle with zero total energy that starts in the @xmath109 potential given above , in the region @xmath110 .
if the particle is initially moving to the left , it will continue towards @xmath92 , taking an infinite amount of time to reach that turning point .
but if the particle is initially moving to the right , it will reach the @xmath111 turning point in a _
finite _ amount of time that depends on its initial @xmath0 , as given by @xmath112 so , for example , starting from the midpoint @xmath113 , it takes unit time to reach the turning point at @xmath111 . upon reaching @xmath111 ,
the explicit form of the time - dependent solution ( [ exacttrajectory4 ] ) exhibits the classical counterpart of _ a sudden transition _ that keeps @xmath114 , but _ changes the potential _ for the return trip towards @xmath92 , exactly as follows from the particle moving on a different branch of the @xmath108 function .
explicitly , the potential deepens to@xmath115 with the particle s speed changing accordingly as a function of @xmath0 .
@xmath84the return velocity profile is now negative , and given by@xmath116 the @xmath108 in this last expression , as well as in @xmath117 , is understood to be the principal value . moving in this modified negative potential
, it now takes the zero - energy particle a _ finite _ amount of time to travel from @xmath111 down to @xmath92 , as given by @xmath118 . upon reaching @xmath92 ,
the exact solution ( [ exacttrajectory4 ] ) exhibits another sudden transition that keeps @xmath114 , but again alters the potential for the return trip towards @xmath111 . the potential becomes@xmath119 with corresponding changes in the velocity profile .
again , a finite amount of time is needed for the particle to go from @xmath92 to @xmath111 in the potential @xmath120 , as given by @xmath121 . upon reaching @xmath111 again
, the switchback process continues .
the total energy remains at zero , but the potential for the second return trip further deepens,@xmath122 etc . in order for the particle to follow the interpolating trajectory ( [ exacttrajectory4 ] ) specified by the schrder equation for the @xmath42 case
, it has to move under the influence of successively deepening potentials . at later times
, the effective potential seen by the particle on its zigzag path between the turning points will depend on the total number of previous encounters with those turning points . in this sense
, the particle remembers its past .
let @xmath123 be the total number of turning points previously encountered on the particle s trajectory .
the motion of the particle before encountering the next , @xmath124st , turning point is then completely determined , for @xmath42 , by the effective potential ( again , @xmath108 here is understood to be the principal value)@xmath125 where @xmath60 is again the floor function .
that is to say , the potential deepens as @xmath123 increases .
the corresponding velocity profile speeds up : @xmath126 the particle will either be traveling to the left , with @xmath127 for odd @xmath123 , or traveling to the right , with @xmath128 for even @xmath123 , with its speed increasing with @xmath123 .
as mentioned previously , this effect is clearly seen upon viewing numerical animations of the @xmath42 trajectories @xcite .
it is also instructive to plot @xmath129 versus @xmath1 for various @xmath123 , with @xmath130 , to check @xmath114 , as well as to see how the energy would _ not _ be conserved if the potentials were _ not _ switched .
see figure 1 .
1.eps the time for the zero - energy particle to traverse the complete unit interval in @xmath0 while moving through the @xmath131 potential ( the @xmath123th pseudo - half - cycle ) is always finite , for @xmath132:@xmath133 this _ transit time _
decreases monotonically as @xmath123 increases , with @xmath134 . starting from an initial @xmath0 , with initial @xmath135 , the times
at which changes in the potential occur , i.e. the times at which the particle encounters turning points , are obtained by summing these transit times and then adding @xmath136 .
the transit time sums are simple enough : @xmath137 thus , a particle beginning at @xmath0 , with initial @xmath135 , will encounter its @xmath123th turning point , and the potential @xmath131 will switch on , at time @xmath138 .
the potential @xmath131 will remain in effect for a time span equal to the particle s transit time , @xmath139 , i.e. until the next potential @xmath140 switches on .
the potential index @xmath123 is in fact related to the action for the pseudo - half - cycle @xcite . for even @xmath123 , we have @xmath141 , while for odd @xmath123 , we have @xmath142 .
thus ( in the units chosen , the mass is @xmath143)@xmath144 where for the first pass through the complete unit interval the action is @xmath145 moreover , @xmath146 for integer @xmath147 , so after the initial crossing , the @xmath148st zigzag @xmath149 ( pseudo - full - cycle ) contributes to the action : @xmath84@xmath150 the progressive deepening of the potential is also evident in the expanding phase - space trajectories for the evolving particle , where the switchbacks enhance the velocity after each encounter with a turning point ( @xmath31 ) . for example ,
see figure 2 .
2.eps this figure provides a good opportunity to clarify the differences between conventional hamiltonian and quasi - hamiltonian systems .
the apparent failure to be a conventional hamiltonian system is due entirely to the fact that a given phase - space trajectory seems to repeatedly cross itself at two points ( @xmath88 , and @xmath89 , as shown in the figure ) and at various times .
thus a vector field on the phase space needed to describe the motion in the conventional hamiltonian formalism would appear to be singular at those two points .
these phase - space singularities are tied to the branch point and riemann surface sheet structure of the relevant analytic functions .
in fact , an alternate way to visualize the motion is as a trajectory on the sheets of a riemann surface .
consider the particle moving on the complex @xmath0 plane , and not just the real line segment @xmath151 $ ] , the endpoints of which are now branchpoints .
there are cuts from @xmath152 to @xmath153 and from @xmath70 to @xmath154 . the particle first moves along the real axis , approaches the cut at @xmath152 , and then goes around the branchpoint such that @xmath155 , @xmath156 , and @xmath157 .
the particle returns along the real axis to the origin and encircles the branchpoint at @xmath70 , such that @xmath158 , @xmath159 , and @xmath160 .
the particle then goes back to @xmath152 and goes around it once more such that again @xmath161 , @xmath156 , and @xmath162 .
the trajectory continues in this way , flipping signs and adding @xmath163s according to the formulae ( [ indexed v ] ) and ( [ indexed v ] ) @xcite . for particles with initial @xmath135 , a third way to picture the dynamics
is in terms of the total distance traveled by the particle . in this point of view , the successive @xmath131 patch together to form a continuous potential @xmath164 on the real half - line , @xmath165 , as shown in figure 3 .
indeed , the half - line may be thought of as a _ covering manifold _ of the unit interval in @xmath0 , with the previous multi - valued functions of @xmath0 now single - valued functions of @xmath166 . to avoid stagnation at any of the cusps of @xmath167 , either physical _ or _ conceptual , in this picture the @xmath114 right - moving trajectories
@xmath168 may be considered as limits of positive energy right - movers , @xmath169 .
this limiting process corresponds to the encirclement of the branch points in the riemann surface picture .
3.eps also , an extended phase space trajectory involving @xmath168 and @xmath170 gives some improvement over the situation depicted in figure 2 , with the trajectory no longer self - intersecting .
however , the corresponding phase - space vector field would still be singular with cusps appearing in @xmath171 periodically in @xmath168 , similar to the cusps in the potential @xmath167 .
it is again possible to avoid the cusps by considering the extended phase space trajectories as limits of positive energy trajectories , but we leave that for the interested reader to pursue . finally , in terms of the coordinate transformation introduced in @xcite , namely , the angle @xmath172 the motion is completely unraveled , at least for @xmath42 . in that case
@xmath114 implies @xmath173 , and the trajectory for all @xmath1 is just that of a particle moving in a repulsive quadratic potential ( inverted sho ) .
this is the easiest example solvable in closed - form by the schrder method @xcite .
the exponentially growing solution , @xmath174 , whose values in the real - line cover of the circle are all physically distinct , immediately leads to @xmath96 for the particle moving through the various @xmath175 , as is evident from ( [ exacttrajectory4 ] ) . to amplify our discussion of the physics encoded in the switchback potentials
, we note that the chaotic behavior of the @xmath42 motion ( i.e. the sensitive dependence on initial conditions ) follows entirely from the behavior of the inverse schrder function .
the sine function in ( [ exacttrajectory4 ] ) discards any integer multiple of @xmath176 .
so , for large and increasing times the trajectory @xmath177 only depends on digits farther to the right of the decimal in the numerical value of the initial @xmath178 , hence it is sensitive to those initial conditions . of course , this latter fact is well - known in the context of the discrete @xmath42 map , and has been discussed time and again in the literature for the case where @xmath1 is an integer ( for example , see @xcite ) .
as is evident from ( [ exacttrajectory4 ] ) the discussion carries over _ mutatis mutandis _ to the continuous time case .
the only point we add here is to emphasize that it is @xmath7 which is responsible for the effect .
insofar as @xmath82 follows from @xmath5 and the latter follows directly from the potential ( as well as the other way round , as discussed in our introduction ) the chaotic behavior follows directly from the specific multi - valued form of the potential . for other values of @xmath3 , from ( [ trajectory ] ) and ( [ interpolation ] )
we again have @xmath179 , so the distinction between regular and chaotic behavior is again determined entirely by the asymptotic behavior of @xmath7 , hence the asymptotic branch structure of @xmath5 , and hence the form of the switchback potentials .
the link between the analytic potential function and chaotic behavior is arguably somewhat subtle , inasmuch as it requires some computation to determine @xmath5 and @xmath7 , but the link is a direct one .
having said that , there remains a case - by - case need for inspection of @xmath7 to determine chaos or its absence . at this stage and in this paper ,
we do _ not _ purport to produce a universal analysis of all models we only present methods that may facilitate such an analysis .
in summary , in this paper we use a functional equation to convert nonlinear map equations into dynamical systems in the continuous time domain , thereby facilitating the study of such systems . illustrate this _ holographic _ continuum interpolation technique with the well - known model of the logistic map , explicitly for the chaotic case @xmath42 , an example which foreshadows an analysis for generic @xmath3 . construct an intricate sequence of switchback hamiltonians which drive the system to just those phase - space trajectories resulting from such an analytic interpolation of the logistic map , and produce the familiar chaotic behavior for @xmath42 . for nontrivial systems such as the chaotic @xmath42 logistic map illustrated here , in order to produce the global features of the trajectories the potential of the underlying hamiltonian system _
must _ change at each switchback of the motion , as occurs when turning points are reached in finite time .
therefore we have called these systems _ quasi - hamiltonian_. the trajectories dictate and completely determine the succession of switchback potentials , in a richly extended analogy of inverse scattering techniques . for values of @xmath3 between @xmath180 and @xmath41 , a _
_ numerical analysis pursuant to the techniques in this paper leads to quasi - hamiltonian systems involving switchbacks among sequences of potentials , eventually producing trajectories that move between elaborate sets of turning points , in general .
this analysis is reported in full in a companion paper @xcite . however , to make the present paper more self - contained , and perhaps more convincing as to the generality of the switchback effect
, we have included some details for one other case ( @xmath95 ) in an appendix .
to conclude , we encourage experimenters to look in various settings for the particular type of continuous interpolates that we have described , including fluid dynamical experiments , ecological systems , and , more generally , any situation where an inherently continuous parameter ( time , scale , etc . ) has been routinely sampled only at predetermined discrete values .
_ we thank d. callaway for incisive questions and encouragement , d. sinclair for his emphasis on an appropriate title , and an anonymous reviewer for suggestions to improve our presentation of the material .
this work was supported in part by nsf award 0855386 , and in part by the u.s .
department of energy , division of high energy physics , under contract de - ac02 - 06ch11357 .
the coefficients in the series expansion of @xmath5 may be computed recursively for any @xmath3 , from ( [ psiseriescoefs ] ) . while not immediately recognizeable as the series expansion of any well - known function , nonetheless , the coefficients may be used to estimate the radius of convergence of the series ( [ psiseries ] ) by the simple ratio test : @xmath181 , with @xmath182 . to illustrate this consider the case @xmath95 . for this case numerical application of the ratio test
leads to the data in figure 4 . from the numerical data we infer that @xmath183 , as indicated earlier in ( [ psiseriesradius ] ) . while this numerical demonstration is not a proof of convergence , of course , we believe it provides compelling evidence for convergence .
other values of @xmath3 produce numerical data qualitatively the same as the @xmath95 case , with only one noteworthy difference : while for @xmath184 , the convergence of the numerical data to the expressions in ( [ psiseriesradius ] ) is from above , for @xmath185 the convergence is from below .
a similar numerical calculation for the coefficients in the expansion of the inverse function ( [ psiinverseseries ] ) suggests @xmath186 is an entire function of @xmath0 , for @xmath95 , much like the closed - form results in ( [ exactcases ] ) .
that is to say , @xmath187 for @xmath188 combining the series data for @xmath5 , @xmath7 , and @xmath44 with the functional equations obeyed by them leads to an accurate numerical determination of the functions and the switchback potentials , for any @xmath3 .
the methods are described in detail in @xcite .
numerical results for @xmath189 , with @xmath190 , are shown in figure 5 along with the expected asymptote , @xmath191 .
this interval in @xmath1 corresponds to plotting @xmath192 for @xmath193 . for this particular case , the sequence of potentials does not deepen progressively , but
rather the sequence converges , albeit very slowly , onto the stable fixed point of the map , namely , @xmath194 , shown in the figure as a small circle on the @xmath0-axis .
hence the zero - energy particle trajectory will also converge onto this fixed point , rigorously conserving @xmath114 in the process .
although the particle trajectory is not chaotic for @xmath95 , nevertheless the potentials are somewhat more exotic than the @xmath42 case in the sense that there are an _ infinite _ number of branch points for the @xmath95 case . these may be obtained by iterating the action of the map on @xmath113 . that is to say , @xmath195 .
these branch points are shown as the turning points of the potential sequence , @xmath196{ccccc}% 0.75 , & 0.562\,5 , & 0.738\,281\ , , & 0.579\,666\ , , & \cdots \end{array } $ ] , represented by colored dots on the @xmath0-axis in figure 6 .
the branch points may also be seen by flipping the graph of @xmath7 in figure 5 about the sw - ne diagonal , in the usual way , to exhibit the various branches of @xmath5 .
this complicated branch structure for @xmath95 extends to other cases with @xmath197 and most likely accounts not only for the failure to render them in terms of simple , known functions , but also for some of the associated wide - spread mystique that surrounds @xmath5 and @xmath7 in these cases .
of course , @xmath73 and @xmath42 only have branch points at @xmath113 , and at @xmath92 and @xmath71 , respectively , as evident in the closed - form results ( [ exactcases ] ) and as discussed in the text , so these cases are relatively bland by comparison .
the results ( [ interpolation ] ) ( [ generalv ] ) establish strong ties between the @xmath5-functional method and the standard logical sequence conventionally used for time evolution , wherein one begins with a given potential and then computes the trajectory .
perhaps this is not without precedent . consider a small but macroscopic object made of weakly conducting dielectric material moving in the electric field produced by some charged surfaces .
occasionally the object might come in contact with one of the surfaces ( i.e. hit a turning point ) and exchange some significant amount of electric charge in the process .
this would abruptly change the forces on the object , and correspondingly its potential , affecting the object s motion and altering its subsequent trajectory . the potential could change abruptly again if the object were to come in contact with another charged surface . of course , this process is more difficult to model than that of the logistic map .
the @xmath198 case gives a simple , closed - form result , namely , @xmath199 and @xmath200 .
however , it is also pathological in the sense it does not yield a particle trajectory from ( [ interpolation ] ) as written . of course , a scaling limit might be considered , with @xmath201 fixed as @xmath202 .
we leave this as an exercise for the reader . for odd @xmath123 , we may think of @xmath203 as a morse index for the @xmath114 phase space pseudo - cycle : @xmath204 .
14 in m. c. gutzwiller , _ chaos in classical and quantum mechanics _ ,
springer , new york ( 1990 ) . by comparison to the electrically charged object described in a previous footnote , in the case of the particle interpolating the logistic map the charge acquired at the turning point is actually _ topological _ , as is clear in the riemann surface point of view :
the charge is just the winding number for the path of the particle around the branch points . | holographic functional methods are introduced as probes of discrete time - stepped maps that lead to chaotic behavior .
the methods provide continuous time interpolation between the time steps , thereby revealing the maps to be quasi - hamiltonian systems underlain by novel potentials that govern the motion of a perceived point particle . between turning points
, the particle is strictly driven by hamiltonian dynamics , but at each encounter with a turning point the potential changes abruptly , loosely analogous to the switchbacks on a mountain road .
a sequence of successively deepening switchback potentials explains , in physical terms , the frequency cascade and trajectory folding that occur on the particular route to chaos revealed by the logistic map . |
magnetograms provide the radial magnetic field on the visible surface of the sun .
the actual measurement is for the line - of - sight component of the magnetic field , which is then transformed into the radial component assuming an ( approximately ) radial field near the solar surface .
as the sun rotates , the individual magnetograms can be combined into a synoptic magnetogram that covers the whole spherical surface .
synoptic magnetograms are provided by many observatories , including wilcox solar observatory ( wso ) , the michelson doppler imager ( mdi ) instrument on the solar and heliospheric observatory ( soho ) , the global oscillation network group ( gong ) , solar dynamic observatory ( sdo ) and the synoptic optical long - term investigations of the sun ( solis ) observatory . today
s magnetograms contain hundreds to thousands of pixels along each coordinate direction .
these magnetograms can be used to extrapolate the magnetic field into the solar corona .
the simplest model @xcite assumes a current - free , in other words potential , magnetic field that matches the radial field of the magnetogram on the surface , while it satisfies a simple boundary condition at the outer boundary at some radial distance @xmath0 .
the outer boundary condition is usually taken at @xmath1r@xmath2 ( solar radii ) , and a purely radial field is assumed at this `` source surface '' .
mathematically the problem is the following : given the magnetogram data that defines the radial component of the magnetic field as @xmath3 at @xmath4r@xmath2 , find the scalar potential @xmath5 so that @xmath6 here @xmath7 $ ] and @xmath8 $ ] are the co - latitude and longitude coordinates , repectively .
once the solution is found , the potential field solution is obtained as @xmath9 and it will trivially satisfy both the divergence - free and the current - free properties @xmath10 we note that the current is only zero inside the domain . if the solution is continued out to @xmath11 with a purely radial magnetic field @xmath12 , there will be a finite current at @xmath13 , on the other hand , the divergence will be zero for all @xmath14 .
the potential field solution is often obtained with a spherical harmonics expansion @xcite . here
we briefly summarize the procedure in its simplest possible form .
the base functions @xmath15 are the spherical harmonic functions @xmath16 multiplied with an appropriate linear combination of the corresponding radial functions @xmath17 and @xmath18 so that the boundary condition @xmath19 is satisfied : @xmath20 the indexes @xmath21 and @xmath22 ( @xmath23 ) are the integer degree and order of the spherical harmonic function , respectively .
the @xmath15 functions are solutions of the laplace equation ( [ eq : laplace ] ) , satisfy the boundary condition at @xmath24 , and they form an orthogonal base in the @xmath25 coordinates .
the magnetic potential solution can be approximated as a linear combination of the base functions @xmath26 where @xmath27 is the highest degree considered in the expansion and the @xmath28 harmonics is not included , as it corresponds to the monopole term .
the coefficients @xmath29 can be determined by taking the radial derivative of equation ( [ eq : phi ] ) and equating it with the magnetogram radial field at @xmath30 : @xmath31 \sum_{m =- n}^{n } f_{nm } y_{nm}(\theta,\phi ) \label{eq : m}\ ] ] exploiting the orthogonality of the base functions , we can take the inner product with @xmath16 to determine @xmath29 as @xmath32 } \int_0^\pi d\theta \sin \theta \int_0^{2\pi } d\phi m(\theta,\phi ) y_{nm}(\theta,\phi ) \label{eq : integral}\ ] ] where the @xmath33 coefficient results from the normalization of the spherical harmonics .
an alternative approach of obtaining the harmonic coefficients @xmath29 is to employ a ( least - squares ) fitting procedure in equation ( [ eq : m ] ) .
this is much more expensive than evaluating the integral in ( [ eq : integral ] ) , but it can be more robust if the magnetogram does not cover ( well ) the whole surface of the sun . using the spherical harmonic coefficients the potential can be determined on an arbitrary grid using ( [ eq : phi ] ) and the magnetic field can be obtained with finite differences .
alternatively , one can calculate the gradient of the base functions analytically and obtain the magnetic field as @xmath34 for @xmath35 .
spherical harmonics provide a computationally efficient and very elegant way of solving the laplace equation on a spherical shell .
however , one needs to be cautious of how the integral in equation ( [ eq : integral ] ) is evaluated , especially when a large number of harmonics are used in the series expansion .
we will use the gong synoptic magnetogram for carrington rotation 2077 ( cr2077 , from november 20 to december 17 , 2008 ) as an example to demonstrate the problem .
the magnetogram contains the radial field on a @xmath36 latitude - longitude grid on the solar surface .
the grid spacing is uniform in @xmath37 ( or sine of the latitude ) and in longitude @xmath38 .
figure [ fig : magnetogram ] shows the radial field .
section 2 discusses the naive and more sophisticated ways of obtaining the potential field solution with spherical harmonics .
section 3 describes an alternative approach using an iterative finite difference .
the various methods are compared in the final section 4 , where we also demonstrate the ringing effect that can arise in the spherical harmonics solution , and we draw our conclusions .
to turn the analytic prescription given in the introduction into a scheme that works with real magnetograms , one has to pick the maximum degree @xmath27 , and evaluate the integrals in equation ( [ eq : integral ] ) for each pair of @xmath39 and @xmath22 up to the highest order .
the resulting @xmath29 coefficients can be used to construct the 3d potential magnetic field solution at any given point using equation ( [ eq : bpot ] ) .
the simplest approximation to equation ( [ eq : integral ] ) is a discrete integral using the original magnetogram data : @xmath32 } \sum_{i=1}^{n_{\theta } } \sum_{j=1}^{n_{\phi } } ( \delta\cos\theta)_i ( \delta\phi)_j m_{i , j } y_{nm}(\theta_i,\phi_j ) \label{eq : sum}\ ] ] where the @xmath40 is the radial field in a pixel of the @xmath41 by @xmath42 sized magnetogram .
the pixel is centered at the @xmath43 coordinates , and the area of the pixel is given by @xmath44 .
unfortunately , the uniform @xmath37 mesh used by most of the magnetograms is not at all optimal to evaluate the integral in equation ( [ eq : integral ] ) .
in fact this procedure will only work with maximum order @xmath27 that is much less than @xmath41 .
figure [ fig : legendre ] shows the @xmath45 associated legendre polynomial discretized in different ways .
the red curve shows the discretization on 180 grid points that are uniform in @xmath37 . clearly the red curve is a very poor representation near the poles , where @xmath46
this is important , because the amplitude of the legendre polynomial is actually largest near the poles .
this means that the orthonormal property is not satisfied in the discrete sense , and the coefficients obtained with equation ( [ eq : sum ] ) are very inaccurate .
the legendre polynomial can be represented much better on a uniform @xmath47 grid ( shown by the green curve in figure [ fig : legendre ] ) , as we will discuss below .
a clear signal of this problem is that the amplitudes of the higher order spherical harmonics are not getting smaller with increasing indexes @xmath39 and @xmath22 , i.e. , the harmonic expansion is not converging .
the black line in figure [ fig : sn ] shows the amplitudes @xmath48 which is oscillating wildly for @xmath49 for this @xmath50 magnetogram .
the oscillations are almost exclusively due to the @xmath51 coefficients , the @xmath52 coefficients are well behaved .
this plot can be directly compared with figure 15 in @xcite , where the power spectrum is more - or - less exponentially decaying .
we believe that the reason is that these authors used a least - square fitting to the line - of - sight ( los ) magnetic field instead of calculating the spherical harmonics from the radial field as shown above .
while the two methods are identical analytically ( assuming that the los and radial fields correspond to the same solution ) , the use of least - square fitting mitigates the lack of orthogonality among the discretized legendre funtions , while the naive approach described above heavily relies on the orthogonality property .
given the non - converging series expansion , the resulting potential field will be very inaccurate in the polar regions and will have essentially random values depending on the number of spherical harmonics used .
this is demonstrated by figure [ fig : naive_fft ] that shows the radial magnetic field reconstructed with various number of harmonics using the original magnetogram grid .
one would expect the radial component of the potential field to reproduce the magnetogram shown in figure [ fig : magnetogram ] .
instead , we find that the solution deviates strongly in the polar region if the harmonics expansion is continued above @xmath53 . for @xmath53 ( or lower )
the solution looks reasonable , but strongly smoothed due to the insufficient number of harmonics .
this is most obvious around the active regions in the top panel of figure [ fig : naive_fft ] .
we note that these numerical errors are not related or comparable to the observational uncertainties of the magnetograms , which are usually also quite large in the polar regions .
the observational uncertainties are essentially unavoidable but are within some well - understood range . on the other hand ,
these numerical artifacts are definitely avoidable while the errors are essentialy unbounded if one uses too many harmonics .
one can get much more accurate results if the magnetogram is remeshed to a grid that is uniform in the co - latitude @xmath47 , has an odd number of nodes , and contains the two poles @xmath54 and @xmath55 .
in fact this is the standard grid used in spherical harmonics transforms ( e.g. @xcite ) and it is often referred to as using the chebyshev nodes , since the uniform @xmath47 grid points correspond to the chebyshev nodes in the original @xmath37 coordinate which is the argument of the legendre polynomials .
figure [ fig : legendre ] shows that the legendre polynomial is much better represented on the uniform @xmath47 grid than on the uniform @xmath37 grid . remeshing
the magnetogram introduces some new adjustable parameters into the procedure : the number of grid cells @xmath56 on the new mesh , and the interpolation procedure .
if the remeshing is done with the same number of grid points as is in the original magnetogram grid , the latitudinal cell size at the equator will be a factor of @xmath57 larger than in the uniform-@xmath37 grid .
on the other hand , the uniform-@xmath47 grid will contain many more points than the original in the polar regions , so the interpolation procedure may create some unwanted artifacts . to maintain the resolution of the original data around the equator
, we set @xmath56 to @xmath58 rounded to an odd integer . for the remeshing we chose a simple linear interpolation procedure , and it works satisfactorily , but one could certainly use higher order interpolation procedures , such as splines . before doing the interpolation
, we add extra grid cells corresponding to the north and south poles of the magnetogram grid , and the values at these two extra cells are set as the average of the pixels around the poles : @xmath59 the co - latitude coordinates of the uniform-@xmath47 mesh are @xmath60 for @xmath61 .
we use simple linear interpolation from the extended magnetogram mesh to the uniform @xmath47 mesh : @xmath62 where the @xmath63 index is determined so that @xmath64 and @xmath65 finally the spherical harmonics coefficients are determined with the integral approximated as @xmath32 } \sum_{i=1}^{n'_{\theta } } \sum_{j=1}^{n_{\phi } } \epsilon_i w_i(\delta\phi)_j m'_{i , j } y_{nm}(\theta'_i,\phi_j ) \label{eq : sum2}\ ] ] where @xmath66 and @xmath67 for all other indexes .
the @xmath68 coefficients are the clenshaw - curtis weights @xcite defined as @xmath69 where @xmath70 and @xmath71 and @xmath72 for all other indexes .
we note that for @xmath56 order of 10 or more @xmath73 where @xmath74 and @xmath75 are the indexes of the neighboring cells , or the cell itself at the poles . using the proper grid allows us to use larger number of spherical harmonics .
@xmath27 is limited only by the @xmath76 and @xmath77 alias - free conditions @xcite . for our example
@xmath36 magnetogram , we remesh it a to @xmath78 uniform-@xmath47 grid , and we can obtain accurate solutions up to @xmath79 degree spherical harmonics .
figure [ fig : remeshed_fft ] shows the potential field solution obtained with the remeshed magnetogram grid with @xmath79 . compared with the naive method ,
the solution is much more reasonable in the polar regions .
there is some smoothing when compared to the magnetogram shown in figure [ fig : magnetogram ] , most obvious near the active regions .
while the remeshing is definitely a big improvement over using the original magnetogram , it would be nice to be able to use the original magnetogram data without remeshing and interpolation .
the next section shows that this can be easily achieved with a finite difference solver .
the laplace equation ( [ eq : laplace ] ) with the boundary conditions ( [ eq : innerbc ] ) and ( [ eq : outerbc ] ) can be solved quite easily with an iterative finite difference method .
the advantage of finite differences compared with spherical harmonics is that the boundary data given by the magnetogram directly effects the solution only locally , while the spherical harmonics are global functions , and their amplitudes depend on all of the magnetogram data .
if the magnetogram contains large discontinuities , we expect the finite difference scheme to be better behaved .
the finite difference method has advantages if the solution is to be used in a finite difference code on the same grid , because one can guarantee zero divergence and curl for the magnetic field in the finite difference sense .
the solution obtained with the spherical harmonics has zero divergence and curl analytically , but not on the finite difference grid , which may severly underresolve the high order spherical harmonic functions in some regions ( see figure [ fig : legendre ] ) .
the finite difference method was applied to the solar potential magnetic field problem as early as 1976 ( @xcite ) , but the method was limited by the computational resources available at the time . solving a 3d laplace equation on today s computers is an almost trivial problem .
we implemented the new finite difference iterative potential - field solver ( fdips ) code in fortran 90 .
the serial version does not require any external libraries , while the parallel version uses the message passing interface ( mpi ) library for communication .
fdips can solve the laplace equation on a @xmath80 spherical grid to high accuracy on a single processor in less than an hour .
the parallel code can solve the same problem in less than 5 minutes on 16 processors .
we briefly describe the algorithm in fdips .
we use a staggered spherical grid : the magnetic field is discretized on cell faces while the potential is discretized at the cell centers .
we use one layer of ghost cells to apply the boundary conditions so the cell centers are located at @xmath81 with @xmath82 , @xmath83 and @xmath84 .
the @xmath85 and @xmath86 coordinates of the real cells are given by the magnetogram , while the ghost cell coordinates are given by @xmath87 , @xmath88 , @xmath89 , and @xmath90 .
we allow for a non - uniform radial grid extending from @xmath30 to @xmath0 , but for sake of simplicity in this paper a uniform radial grid is used with @xmath91 with @xmath92 .
the radial magnetic field components are located at the radial cell interfaces at @xmath93 where @xmath94 for @xmath95 , @xmath96 and @xmath97 .
similarly the latitudinal components are at @xmath98 with @xmath99 , and @xmath100 .
note that the interface is taken half - way in the @xmath37 coordinate and not in @xmath47 , because this makes the cells equal area when the magnetogram is given on uniform @xmath37 grid .
finally the longitudinal field components are located at @xmath101 where @xmath102 for @xmath103 .
the staggered discretization keeps the stencil of the laplace operator compact and it makes the boundary condtions relatively simple .
the magnetic field is obtained as a discrete gradient of @xmath5 : @xmath104 note the @xmath105 factor in the @xmath47 derivative for the uniform @xmath37 grid . for uniform @xmath47 grid this
is replaced with @xmath106 .
the divergence of the magnetic field , i.e. the laplace of @xmath5 is obtained as @xmath107 again @xmath108 is used for the uniform @xmath37 grid , while on the uniform @xmath47 grid this is replaced with @xmath109 .
the magnetogram boundary condition is applied by setting the inner ghost cell as @xmath110 where @xmath111 is the magnetogram with the average field ( i.e. the monopole due to observation errors ) @xmath112 removed .
the zero potential at @xmath113 is enforced by setting the ghost cell as @xmath114 the boundary conditions at the poles are a bit tricky .
cells @xmath115 and @xmath116 are on opposite sides of the north pole if @xmath117 .
therefore the ghost cells in the @xmath47 direction are set as @xmath118 and @xmath119 .
we note here that @xmath42 is assumed to be an even number .
the periodic boundaries in the @xmath38 direction are simple : @xmath120 and @xmath121 .
we need to find @xmath122 that satisfies the discrete laplace equation ( [ eq : laplacenum ] ) with the boundary conditions applied via the ghost cells .
the initial guess is @xmath123 which provides a non - zero residual because of the inhomogeneous boundary condition at the inner boundary applied by equation ( [ eq : innerbcnum ] ) .
we use this residual with a negative sign as the righ - hand - side of the poisson equation @xmath124 , and use @xmath125 as the new homogeneous inner boundary condition instead of ( [ eq : innerbcnum ] ) .
we use the krylov - type iterative method bicgstab @xcite to find the solution .
the linear system is preconditioned with an incomplete lower - upper decomposition ( ilu ) preconditioner to speed up the convergence .
we use ilu(0 ) with no fill - in compared to the original matrix structure , so the preconditioner is a diagonal matrix , but its elements depend on all elements of the original matrix . we have implemented a serial as well as a parallel version of the algorithm . in the parallel version
the preconditioner is applied separately in each subdomain .
fdips finds an accurate ( down to @xmath126 relative error ) solution on a @xmath127 grid in less than 1000 iterations . even running serially ,
this takes less than an hour on today s computers .
once the solution is found in terms of the discrete potential @xmath122 , we apply the original boundary conditions including ( [ eq : innerbcnum ] ) and calculate the magnetic field with equation([eq : gradient ] ) .
the divergence of the magnetic field will be zero to the accuracy of the poisson solver .
the curl of the magnetic field will be zero in a finite difference sense simply because it is constructed as the discrete gradient of the potential .
the boundary condition at the inner boundary is also satisfied exactly : @xmath128 .
averaging @xmath129 and @xmath130 to the @xmath131 position at the outer boundary also gives exactly zero tangential fields due to the equations ( [ eq : outerbcnum ] ) and ( [ eq : gradient ] ) .
depending on the application , we may interpolate the potential magnetic field onto a collocated grid , or use it on the original staggered grid .
figure [ fig : fdips ] shows the solution of the magnetic field obtained with the finite difference solver fdips on a @xmath132 grid .
since we use the same uniform-@xmath37 grid as the magnetogram , the obtained radial field is identical with the magnetogram at @xmath30 .
the tangential components agree well with the remeshed spherical harmonics solution shown in figure [ fig : remeshed_fft ] .
it took 1166 iterations to get a solution with a relative accuracy of @xmath126 .
the run time was almost exactly one hour on a 2.66 ghz intel cpu .
we have discussed various ways to obtain the potential field solution based on solar magnetograms .
while spherical harmonics provide an efficient and elegant method , there are some subtle restrictions that require attention .
if one wants to use many spherical harmonics ( the same order as the number of magnetogram pixels in the colatitude direction ) , the magnetogram data on the @xmath133 grid has to be remeshed onto a uniform-@xmath47 grid with @xmath134 points , @xmath56 must be an odd number , and the new grid must include both poles . after the remeshing the maximum degree of harmonics @xmath27 is only limited by the anti - alias limit to @xmath135 , @xmath136 .
we used a simple linear interpolation for the remeshing .
the remeshing can be avoided by the use of a 3d finite difference scheme .
one can use the original magnetogram grid , and the only freedom is in choosing the radial discretization .
the finite difference scheme provides a solution that is fully compatible with the boundary conditions , and the solution has zero divergence and curl in the finite difference sense .
figure [ fig : radial ] compares the solutions obtained with the three methods along the radial direction for a fixed latitude @xmath137 and longitude @xmath138 .
the spherical harmonics series were truncated at @xmath79 for both the naive and remeshed methods .
the naive spherical harmonics algorithm gives incorrect results close to the solar surface where the high order harmonics dominate .
this is most obvious for the radial component , which is given at @xmath30 by the magnetogram , and it is exactly reproduced by the finite difference scheme .
the latitudinal component at @xmath30 is also very different from the values given by the remeshed harmonics and the finite differences .
the latter two methods agree reasonably well with each other . for radial distances above @xmath139 all three methods
agree quite well .
so far we restricted our example to a gong magnetogram taken at the solar minimum .
if one uses an mdi magnetogram during solar maximum , the largest magnetic fields are much stronger ( order of 1000 g ) and the resolution of the magnetogram is much finer ( order of 1000 pixels ) .
the finer magnetogram resolution allows going to larger number of harmonics , even when using the original magnetogram grid ( naive approach ) .
but the strong and sharp gradients in the magnetogram will bring out another problem with the spherical harmonics approach , the ringing effect .
the ringing is due to the so - called gibbs phenomenon : the step - function like magnetogram data results in high amplitude high order harmonics in fourier space .
the ringing effect and other artifacts are discussed in great detail by @xcite .
figure [ fig : ringing ] demonstrates this effect on the @xmath140 resolution mdi magnetogram for carrington rotation 2029 ( from april 21 to may 18 , 2005 ) , with the maximum radial field strength around @xmath141 g .
the remeshed harmonics method with @xmath142 is compared with the finite difference method on a @xmath132 grid ( the magnetogram data is coarsened to a @xmath50 grid ) . in the spherical harmonics solution
the ringing is very clearly visible around the active regions , both in the radial and latitudinal components .
the finite difference scheme , on the other hand , shows no sign of ringing in either components .
this is obvious for the radial component , which simply coincides with the coarsened magnetogram , but for the latitudinal component it is due to the fact that the finite difference solution of the laplace equation does not suffer from ringing artifacts even for discontinuous boundary data . for the spherical harmonics
approach the ringing becomes weaker with increased number of harmonics , but it is quite apparent even for @xmath143 ( not shown ) .
the results of the remeshed and naive harmonics methods are essentially the same up to @xmath143 , i.e. the ringing is not due to the remeshing of the magnetogram . in terms of computational efficiency ,
a good implementation of the spherical harmonics scheme is much faster than the finite difference scheme .
in fact , it may be more costly to construct the potential field solution on a 3d grid from the spherical harmonics coefficients than obtaining the coefficients themselves .
our fortran 90 code can obtain the spherical coefficients up to @xmath53 , 90 and 120 degrees in 1 , 1.8 and 3.3 seconds , respectively , while the reconstruction of the solution on the @xmath144 grid takes 5 , 12 , and 20 minutes , respectively .
all timings were done on a single 2.66 ghz intel cpu .
the reconstruction cost can be improved by running the code in parallel , and/or truncating the series in parts of the grid where the higher order harmonics have a negligible contribution .
we also note that going beyond about @xmath145 harmonics becomes fairly complicated @xcite .
the computational cost of the finite difference scheme scales with the number of grid cells and the number of iterations .
the number of iterations is fairly constant for multigrid type methods , but for the krylov sub - space schemes it grows with the problem size , although slower than linearly .
the finite difference scheme can be sped up by parallelizing the code , which is fairly straightforward for the krylov subspace schemes .
since we limit the ilu preconditioning to operate independently on the subdomains of each processor , the preconditioner becomes less efficient as the number of processors increases , which results in an increase in the number of iterations . to minimize this effect , the parallel fdips code splits the grid in the @xmath47 and @xmath38 directions only , so the subdomains in each processor contain the full radial extent of the grid .
our experiments confirmed that using this domain decomposition , the number of iterations indeed does not depend much on the number of processors .
our largest test so far involves a @xmath146 grid with 30 times more cells than the @xmath147 grids discussed in most of this paper .
for the large problem we need about 8,500 iterations to reach the @xmath126 relative accuracy , a factor of 9 increase relative to the smaller problem . using 108 cpu - s , the solution is obtained in about 5.3 hours . despite the various limitations , for some applications the spherical harmonics approach may still be preferred .
for example if the solution is needed to obtain a spherical power spectrum of the solar magnetic field .
if the solution is to be used in a finite difference code , the finite difference solution is probably preferable .
we are using the fdips code to generate the potential field solution as the initial field for our solar corona model @xcite .
this paper attempts to call the attention of astrophysicists and solar physicists to the limitations and potential pitfalls of using the spherical harmonics approach to obtain a potential field solution .
the spherical harmonics representation of the potential field solutions are available from several synoptic magnetogram providers , although the details of the method used to obtain the spherical harmonics is not always clear .
a spherical harmonics based pfss package implemented in idl is available as part of the solar - soft library ( http://www.lmsal.com/solarsoft ) .
this package uses the magnetogram remeshing technique either onto the chebyshev ( uniform-@xmath47 ) or the legendre collocation points .
we are not aware of any publically available code that uses finite differences to solve this particular problem . to allow other researchers to use and compare the two approaches , we make our finite difference code fdips publically available at the http://csem.engin.umich.edu/fdips/ website .
schatten , k. j. , wilcox , j. m. & ness , n. f. 1969 , , 6 , 442 altschuler , m. d. , levine , r. h. , stix , m. , & harvey , j. 1977 , , 51 , 345 suda , r. & takami , m. 2001 , math . of comp . , 71 , 703 clenshaw , c. w. & curtis , a. r. 1960 , numerishe mathematik , 2 , 197 potts , d. , steidl , g. , & tasche , m. 1998 , linear algebra appl . , 275 , 433 adams , j. & pneuman , g. w. 1976 , , 46 , 185 van der vorst , h. 1992 , siam j. sci . statist .
, 13 , 631 tran , t. 2009 , `` improving the predictions of solar wind speed and interplanetary magnetic field at the earth '' , ph.d .
thesis , ucla van der holst , b. , manchester iv , w. b. , frazin , r. a. , vsquez , a. m. , tth , g. , & gombosi , t. i. 2010 , , 725 , 1373 | potential magnetic field solutions can be obtained based on the synoptic magnetograms of the sun . traditionally , a spherical harmonics decomposition of the magnetogram is used to construct the current and divergence free magnetic field solution .
this method works reasonably well when the order of spherical harmonics is limited to be small relative to the resolution of the magnetogram , although some artifacts , such as ringing , can arise around sharp features .
when the number of spherical harmonics is increased , however , using the raw magnetogram data given on a grid that is uniform in the sine of the latitude coordinate can result in inaccurate and unreliable results , especially in the polar regions close to the sun .
we discuss here two approaches that can mitigate or completely avoid these problems : i ) remeshing the magnetogram onto a grid with uniform resolution in latitude , and limiting the highest order of the spherical harmonics to the anti - alias limit ; ii ) using an iterative finite difference algorithm to solve for the potential field .
the naive and the improved numerical solutions are compared for actual magnetograms , and the differences are found to be rather dramatic .
we made our new finite difference iterative potential - field solver ( fdips ) a publically available code , so that other researchers can also use it as an alternative to the spherical harmonics approach . |
su(2 ) yang - mills - higgs ( ymh ) theory possesses monopole @xcite , multimonopole @xcite , and monopole - antimonopole pair solutions @xcite
. the magnetic charge of these solutions is proportional to their topological charge . while monopole and multimonopole solutions reside in topologically non - trivial sectors , the monopole
antimonopole pair solution is topologically trivial .
when gravity is coupled to ymh theory , a branch of gravitating monopole solutions emerges smoothly from the monopole solution of flat space @xcite .
the coupling constant @xmath0 , entering the einstein - yang - mills - higgs ( eymh ) equations , is proportional to the gravitational constant @xmath1 and to the square of the higgs vacuum expectation value @xmath2 .
the monopole branch ends at a critical value @xmath3 , beyond which gravity becomes too strong for regular monopole solutions to persist , and collapse to charged black holes is expected @xcite . indeed ,
when the critical value @xmath3 is reached , the gravitating monopole solutions develop a degenerate horizon @xcite , and the exterior space time of the solution corresponds to the one of an extremal reissner - nordstrm ( rn ) black hole with unit magnetic charge @xcite . beside the fundamental gravitating monopole solution , eymh theory possesses radially excited monopole solutions , not present in flat space @xcite .
these excited solutions also develop a degenerate horizon at some critical value of the coupling constant , but they shrink to zero size in the limit @xmath4 .
rescaling of the solutions reveals , that in this limit the bartnik - mckinnon ( bm ) solutions @xcite of einstein - yang - mills ( eym ) theory are recovered . for the excited solutions
the limit @xmath4 therefore corresponds to the limit of vanishing higgs expectation value , @xmath5 . in this letter
we investigate how gravity affects the static axially symmetric monopole
antimonopole pair ( map ) solution of flat space @xcite , and we elucidate , that curved space supports a rich spectrum of map solutions , not present in flat space . in particular , we show that , with increasing @xmath0 , a branch of gravitating map solutions emerges smoothly from the flat space map solution , and ends at a critical value @xmath6 , when gravity becomes too strong for regular map solutions to persist .
but while the branch of monopole solutions can merge into an extremal rn black hole solution at the critical @xmath0 , there seems to be no neutral black hole solution with degenerate horizon available for the map solutions to merge into .
indeed we find that at @xmath6 a second branch of map solutions emerges , extending back to @xmath7 . along this upper branch
the map solutions shrink to zero size , in the limit @xmath4 , and approach the bm solution with one node ( after rescaling ) .
since the bm solution with one node is related to a branch of map solutions , it immediately suggests itself that the excited bm solutions with @xmath8 nodes are related to branches of excited map solutions .
indeed , constructing the first excited map solution by starting from the bm solution with two nodes , we find , that it represents a map solution , possessing two monopole - antimonopole pairs .
the static axially symmetric map solutions of su(2 ) eymh theory with action @xmath9 ( with yang - mills coupling constant @xmath10 , and vanishing higgs self - coupling ) , are obtained in isotropic coordinates with metric @xcite @xmath11 where @xmath12 , @xmath13 and @xmath14 are only functions of @xmath15 and @xmath16 .
the map ansatz reads for the purely magnetic gauge field ( @xmath17 ) @xcite @xmath18 and for the higgs field @xmath19 with @xmath20 matrices ( composed of the standard pauli matrices @xmath21 ) @xmath22 the four gauge field functions @xmath23 and the two higgs field functions @xmath24 depend only on @xmath15 and @xmath16 .
we fix the residual gauge degree of freedom @xcite by choosing the gauge condition @xmath25 @xcite . to obtain regular asymptotically flat solutions with finite energy density we impose at the origin ( @xmath26 ) the boundary conditions @xmath27 on the @xmath28-axis the functions @xmath29 and the derivatives @xmath30 have to vanish , while on the @xmath31-axis the functions @xmath32 and the derivatives @xmath33 have to vanish .
for solutions with vanishing net magnetic charge the gauge potential approaches a pure gauge at infinity .
the corresponding boundary conditions for the fundamental map solution are given by @xcite @xmath34 introducing the dimensionless coordinate @xmath35 and the higgs field @xmath36 , the equations depend only on the coupling constant @xmath0 , @xmath37 .
the mass @xmath38 of the map solutions can be obtained directly from the total energy - momentum `` tensor '' @xmath39 of matter and gravitation , @xmath40 @xcite , or equivalently from @xmath41 , yielding the dimensionless mass @xmath42 .
subject to the above boundary conditions , we solve the equations numerically @xcite . in the limit @xmath4 , the lower branch of gravitating map solutions emerges smoothly from the flat space solution @xcite .
the modulus of the higgs field of these map solutions possesses two zeros , @xmath43 , on the @xmath28-axis , corresponding to the location of the monopole and antimonopole , respectively . with increasing @xmath0 the monopole and antimonopole move closer to the origin , and the mass @xmath44 of the solutions decreases .
the lower branch of map solutions ends at the critical value @xmath45 . in fig .
[ eps ] we show the energy density @xmath46 of the map solution at @xmath6 .
it possesses maxima on the positive and negative @xmath28-axis close to the locations of the monopole and antimonopole and a saddle point at the origin .
forming a second branch , the map solutions evolve smoothly backwards from @xmath6 to @xmath7 . in the limit @xmath4
the mass @xmath44 diverges on this upper branch , and the locations of the monopole and antimonopole approach the origin , @xmath47 , as seen in fig .
[ z02 ] . at the same time the map solution shrinks to zero . rescaling the coordinate @xmath48 and the higgs field @xmath49 reveals that the axially symmetric map solutions approach the spherically symmetric @xmath50 bm solution on the upper branch as @xmath4 .
consequently , also the scaled mass @xmath51 of the map solutions tends to the mass of the @xmath50 bm solution , as seen in fig .
[ e ] . on the upper branch the limit @xmath4 thus corresponds to the limit @xmath5 ( with fixed @xmath1 ) .
we note that the ansatz ( [ ansatz ] ) for the gauge potential includes the spherically symmetric bm ansatz , @xmath52 where @xmath53 denotes the gauge field function of the bm solution .
anticipating the existence of excited map solutions , linked to the bm solutions with @xmath8 nodes on their upper branches , we construct the first excited map solution , starting from the @xmath54 bm solution .
since the boundary conditions of the @xmath54 bm solution differ from those of the @xmath50 bm solution at infinity , the boundary conditions of the first excited map solution at infinity must be modified accordingly , @xmath55 the upper branch of the first excited map solutions ends at the critical value @xmath56 , from where the lower branch of the excited map solutions evolves smoothly backwards to @xmath7 . as seen in fig.[e ] , in the limit @xmath4 the scaled mass @xmath57 approaches the mass of the @xmath54 bm solution on the upper branch , and the mass of the @xmath50 bm solution on the lower branch .
the modulus of the higgs field of the first excited map solution possesses four zeros , @xmath58 and @xmath59 , located on the @xmath28-axis , representing two monopole - antimonopole pairs .
the locations of the monopole and antimonopole on the positive @xmath28-axis , @xmath60 resp .
@xmath61 , are shown in fig .
[ z02 ] as functions of @xmath0 , together with the node @xmath62 of the fundamental map solution . as @xmath4
, @xmath61 tends to zero on both branches ; in contrast , @xmath60 tends to zero only on the upper branch . on
the lower branch @xmath60 tends to @xmath62 , the location of the monopole of the fundamental map solution . inspecting the limit @xmath4 for the first excited map solution on the lower branch reveals , that in terms of the radial coordinate @xmath35 , the solution differs from the fundamental map solution on its lower branch only near the origin , where the excited map solution develops a discontinuity . in terms of the coordinate @xmath63 ,
on the other hand , the first excited map solution approaches the @xmath50 bm solution for all values of @xmath64 , except at infinity .
hence , the first excited map solution does not possess a counterpart in flat space .
having constructed the fundamental and the first excited map solutions , we expect , that eymh theory possesses a whole sequence of map solutions , labeled by the number of monopole - antimonopole pairs @xmath8 .
each map solution forms two branches , merging and ending at @xmath65 . in the limit @xmath4 , the upper branch of the @xmath8th map solution always reaches the bartnik - mckinnon solution with @xmath8 nodes , while the lower branch of the @xmath8th map solution always reaches the bartnik - mckinnon solution with @xmath66 nodes , except for @xmath50 , where the flat space map solution is reached in the limit @xmath4 .
we conjecture , that the critical values @xmath65 decrease with @xmath8 , such that , as a function of @xmath0 , the scaled mass @xmath67 assumes a characteristic `` christmas tree '' shape .
thus instead of the single map solution present in flat space , in curved space a whole tower of map solutions appears .
an analogous pattern is encountered for gravitating skyrmions , which are likewise linked to the bm solutions @xcite .
we expect the graviating map solutions to be unstable like the flat space map solution @xcite . for
the gravitating monopole solutions a regular event horizon can be imposed @xcite , yielding magnetically charged black hole solutions with hair .
likewise for the map solutions of eymh theory a regular event horizon can be imposed , yielding static axially symmetric and neutral black hole solutions with hair @xcite . within the framework of distorted isolated horizons the masses of these black hole solutions
may possibly be simply related to the masses of the corresponding regular solutions @xcite .
it is interesting , that the spherically symmetric bm solutions of eym theory appear in the limit @xmath4 of the axially symmetric map solutions .
but eym theory also possesses static axially symmetric regular solutions , which are not spherically symmetric @xcite .
could these solutions also appear in the @xmath4 limit of more general @xcite gravitating map solutions ?
we conjecture , that eymh theory allows for the existence of map solutions , consisting of pairs of static axially symmetric multimonopoles , where each multimonopole has winding number @xmath68 @xcite .
it is then conceivable that such multimonopole - antimultimonopole solutions will form an analogous set of solutions as the ones encountered above , but with their upper branches reaching axially symmetric eym solutions with winding number @xmath68 in the @xmath4 limit . but
also flat space should contain further interesting solutions , for instance an antimonopole - monopole - antimonopole system , with the poles located symmetrically with respect to the origin on the @xmath28-axis .
000 g. t hooft , nucl .
* b79 * ( 1974 ) 276 ; + a. m. polyakov , jetp lett . *
20 * ( 1974 ) 194 .
r. s. ward , commun .
* 79 * ( 1981 ) 317 ; + p. forgacs , z. horvarth and l. palla , phys . lett . * 99b * ( 1981 ) 232 ; + p. forgacs , z. horvarth and l. palla , nucl
* b192 * ( 1981 ) 141 ; + m. k. prasad , commun .
phys . * 80 * ( 1981 ) 137 ; + m. k. prasad and p. rossi , phys . rev .
* d24 * ( 1981 ) 2182 .
c. rebbi and p. rossi , phys
* d22 * ( 1980 ) 2010 ; + b. kleihaus , j. kunz and d. h. tchrakian , mod . phys
a13 * ( 1998 ) 2523 .
see also p. m. sutcliffe , int .
a12 * ( 1997 ) 4663 . c. h. taubes , commun .
* 86 * ( 1982 ) 257 ; + c. h. taubes , commun
* 86 * ( 1982 ) 299 .
bernhard rber , thesis , university of bonn 1985 ; + b. kleihaus and j. kunz , phys . rev .
* d61 * ( 2000 ) 025003 .
k. lee , v.p . nair and e.j .
weinberg , phys .
* d45 * ( 1992 ) 2751 .
p. breitenlohner , p. forgacs and d. maison , nucl . phys .
* b383 * ( 1992 ) 357 ; + p. breitenlohner , p. forgacs and d. maison , nucl . phys . *
b442 * ( 1995 ) 126 .
a. lue and e.j .
weinberg , phys .
rev . * d60 * ( 1999 ) 084025 . for zero and very small
higgs self - coupling at @xmath69 a short second branch extends backwards . for large higgs self - coupling
the exterior solution is not of rn - type @xcite .
r. bartnik and j. mckinnon , phys .
lett . * 61 * ( 1988 ) 141 .
b. kleihaus and j. kunz , phys .
* 78 * , 2527 ( 1997 ) ; + b. kleihaus and j. kunz , phys .
lett . * 79 * , 1595 ( 1997 ) ; + b. kleihaus and j. kunz , phys . rev .
* d57 * , 834 ( 1998 ) ; + b. kleihaus and j. kunz , phys . rev . * d57 * , 6138 ( 1998 ) .
s. weinberg , gravitation and cosmology ( wiley , new york , 1972 ) b. kleihaus and j. kunz , in preparation .
p. bizon and t. chmaj , phys . lett . *
b297 * ( 1995 ) 55 .
b. kleihaus and j. kunz , in preparation .
a. ashtekar , s. fairhust and b. krishnan , isolated horizons : hamiltonian evolution and the first law , gr - qc/0005083 .
y. brihaye and j. kunz , phys . rev .
* d50 * , 4175 ( 1994 ) . | we construct static axially symmetric solutions of su(2 ) einstein - yang - mills - higgs theory in the topologically trivial sector , representing gravitating monopole
antimonopole pairs , linked to the bartnik - mckinnon solutions .
= 6.125truein = 8.125truein preprint hep - th/0006148 |
microquasars are well known miniatures of quasars , with a central black hole ( bh ) , an accretion disk and two relativistic jets very similar to those found in the centers of active galaxies , only on much smaller scales ( mirabel @xmath3 rodrguez 1999 ) .
the typical timescales in these systems are also @xmath4 times shorter than those in quasars , thus evolutions of microquasar jets can be studied in details .
therefore , microquasar systems have been considered ideal laboratories for understanding accretion process in black hole systems and might provide us a good alternative to study agn phenomena instead of observing them directly ( massi & kaufman bernad 2008 ) . since discovered in 1992 , radio jets
have been observed in a series of bh binary systems and several of them showed apparent superluminal features . in the two well known microquasars , grs 1915 + 105 ( mirabel @xmath3 rodrguez 1999 ) and gro j1655 - 40 ( tingay et al .
1995 ; hjellming @xmath3 rupen 1995 ) , relativistic jets with actual velocities greater than 0.9@xmath5 were observed . in some other systems ,
small - size compact jets " , e.g. cyg x-1 ( stirling et al . 2001 ) , and large scale diffuse emission , e.g. ss433 ( dubner et al . 1998 ) , were also detected . among these microquasars ,
xte j1550 - 564 is especially interesting since it was the first galactic accretion systems that a fast moving x - ray jet was detected .
the large scale and long existing time of the jets have also made this source unique and valuable in jet studies .
xte j1550 - 564 was discovered with rxte in 1998 during its strong x - ray outburst on september 7 ( smith 1998 ) .
it is believed to be an x - ray binary system at a distance of @xmath15.3 kpc , containing a black hole of [email protected] solar masses and a low mass companion star ( orosz et al .
2002 ) . soon after the discovery of the source , a jet ejection with an apparent velocity greater than 2@xmath5 was reported by vlbi team ( hannikainen et al .
2001 ) . in the period between 1998 and 2002 , the source also exhibited strong x - ray activities but no similar radio and x - ray flares were detected again during these activities ( tomsick et al .
2003 ) . with the @xmath0 satellite , corbel et al .
( 2002 ) found two large scale x - ray jets lying to the east and west of the central source , which were also in good alignment with the central source .
the eastern jet has been detected first in 2000 at a projected distance of @xmath121 arcsec and a position angle e of n of [email protected]@xmath8 from the central black hole .
its apparent proper motion velocity has dropped from an average of @xmath132.9 mas / day between 1998 and 2000 to @xmath121.2 mas / day during its 2000 observations ; its x - ray flux also decayed quite rapidly .
two years later , it could only be seen marginally in the x - ray image , while a western counterpart became visible at @xmath122 arcsec on the other side .
the corresponding radio maps are consistent with the x - ray observations ( corbel et al .
2002 ) . from the year 2002 to 2003
, there were altogether five observations of this source in the @xmath0 archive .
the eastern jet has disappeared from the x - ray images in late 2002 , while the western jet could be seen clearly in all the five observations , although the flux decayed quickly as well . among these observations ,
only the first two have been reported and discussed previously ( kaaret et al .
2003 ; wang et al .
we therefore use these additional data points to better constrain the kinematic and spectral analysis for the eastern jet , and to extend the analysis to the western jet . in this paper
, we first describe the observational data obtained by @xmath0 x - ray satellite in section 2 and 3 .
then we propose a model for the system and discuss the kinematic and emission process in section 4 . finally in section 5 and 6
, we apply this model to another source h 1743 - 322 and make some further discussions based on these analysis .
rrrrrrrr 1 & 679 & 2000 jun 9 & 3816 & 01 . & 36.7 & & + 2 & 1845 & 2000 aug 21 & 5160 & 01.4@xmath62@xmath910@xmath10 & 36.7@xmath61@xmath910@xmath10 & & + 3 & 1846 & 2000 sep 11 & 4630 & 01.5@xmath62@xmath910@xmath10 & 36.6@xmath61@xmath910@xmath10 & & + 4 & 3448 & 2002 mar 11 & 26118 & 02.1@xmath61@xmath910@xmath10 & 37.4@xmath61@xmath910@xmath10 & 56.0@xmath64@xmath910@xmath11 & 33.5@xmath62@xmath910@xmath11 + 5 & 3672 & 2002 jun 19 & 18025 & & & 55.9@xmath64@xmath910@xmath11 & 33.7@xmath62@xmath910@xmath11 + 6 & 3807 & 2002 sep 24 & 24442 & 02.17 & 37.6 & 55.86@xmath66@xmath910@xmath11 & 33.9@xmath63@xmath910@xmath11 + 7 & 4368 & 2003 jan 28 & 23680 & & & 55.82@xmath66@xmath910@xmath11 & 33.6@xmath64@xmath910@xmath11 + 8 & 5190 & 2003 oct 23 & 47831 & & & 55.74@xmath66@xmath910@xmath11 & 33.6@xmath63@xmath910@xmath11
there are altogether eight 2-dimensional imaging observations of xte j1550 - 564 in the @xmath0 archive during june 2000 and october 2003 ( henceforth observations 1@xmath18 ) .
these observations were all made by the advanced ccd imaging spectrometer ( acis ) .
there are also grating observations providing 1-dimensional imaging in the archive ; but since our interests are in the jet kinematics analysis , these are not our focus at this stage .
we have down - loaded all data of these eight observations from the archive and examined the level 2 event files produced by standard data processing procedures to make our analysis .
exposure time for these observations ranges from a few kiloseconds ( observations 1@xmath13 ) to a few tens of kiloseconds ( observations 4@xmath18 ) ( table 1 ) .
background light curves are extracted for each of the observations and only in observation 5 a high background flare is detected . except for this period ,
almost all valid exposure time intervals of these observations are included in our work . in the section 2 and 3
, we limit most of our analysis to the kinematic and spectral properties of the western jet .
the central accreting source has already attracted much attention in recent years and there were already plenty of studies in details in literature ( corbel et al .
2001 ; jain et al . 2001
; kubota @xmath3 makishima 2004 ; yuan et al .
the eastern jet has also been studied fully in 2003 when it was discovered ( tomsick et al .
2003 ; kaaret et al .
2003 ) . as a result
, we do not discuss much of our results for observations 1@xmath13 here ; only comparisons with the previous works will be noted for consistency . a complete study of the kinematic and light curve evolution of the western jet is our main goal in these two sections .
we list the basic information of observations 1@xmath18 in table 1 , including the observation i d , date , the positions of the central point source and the eastern and western jets respectively ( in j2000 coordinate ) . the positions are obtained by the @xmath0 interactive analysis of observations ( ciao ) routine @xmath12 ( freeman et al . 2002 ) , a commonly used program in determining x - ray sources in @xmath0 images . from this table , we could see clearly that an x - ray emission source is detected to the east of the central source in the first four observations and another source is detected to the west in the last five observations . calculations also show that these two sources , when presented in a single combined image , are in good alignment with the central compact object with a position angle e of n of [email protected] . in observations 5 and 6 ,
no x - ray source is detected by @xmath12 at the position of the eastern jet .
however , from the gaussian smoothed images ( fig .
2 ) , a weak source could be recognized in observation 6 .
we thus select the strongest emission region of this source manually with a small circle of the radius of 0.7 arcsec and use the position of its center as one data point in our further analysis . in fig.1 ,
the raw , unsmoothed images of the last five observations are shown .
an extended and moving source can be seen easily in these images .
its elongation points towards the central source , and has thus indicated clearly the relation between them . in most cases
the flux from this western source is comparable to the central source ; however in observation 5 , the observed surface brightness from the western source is even higher than the central source ( also see section 3 ) .
0.3 - 8 kev raw image showing xte j1550 - 564 and the western jet .
the images are in linear scale and no count saturation has been set .
each image is normalized to its own color scale .
the color bar below the images indicates the relative strength .
the numbers under the color bar indicate the number of photons in each pixel for the first image ( obs 4 , id=3448 ) , and the peak count values in the other four images ( obs 5 - 8 ) are 27 , 60 , 58 , 36 , respectively .
the pixel size is @xmath13 .
the green circle in each observation shows the spectral extraction region for each jet , which is of 8 arcsec in diameter and includes almost all of the jet counts ( albeit that there is no guarantee that all jet photons are included , due to the wing of the point spread function of the _ @xmath0 _ mirror assembly . ) .
the images are aligned by ra ( x - axis).,width=377 ] x - ray images of the eight observations of xte j1550 - 564 and the two jets together . except for the first observation ,
in which the central bh is too bright so a log scale is set to reveal the eastern jet , all the images are in linear scale and no count saturation has been set .
the appearance and disappearance of the eastern and western jets can be seen clearly in the image .
each image is normalized to its own color scale .
the color bar below the images indicates the relative strength .
the numbers under the color bar indicate the number of photons in each pixel for the fourth image ( obs 4 , id=3448 ) ( to be consistent with fig .
1 ) , and the peak count values in the other seven images ( obs 1 - 3 , 5 - 8 ) are 18.8 , 3.8 , 6.3 , 10.3 , 13.4 , 15.2 , 9.0 , respectively .
the green elliptical regions are source emission regions detected by _
wavdetect_. observation 4 shows the good alignment of the two jets and the central source .
the images are aligned by ra ( x - axis).,width=377 ] with the data in table 1 , we can calculate the angular separations between the jets and the central source as well as the average proper motions . in this paper , we have adopted the assumption that both the eastern and western jets are related to the same outburst in september 1998 .
except for the outburst occurred in 1998 , there were some other outbursts detected afterwards , including a major outburst in 2000 and some mini - outbursts in 2001 to 2003 ( tomsick et al .
2003 and references therein ) .
however , the long term rxte - asm light curve shows that the x - ray flux of the 1998 outburst is four times higher than that of the 2000 outburst and more than twenty times higher than that of all the rest ( sturner @xmath3 shrader 2005 ) .
the complex behaviors found in the 1998 outburst , such as irregular light curves and multiple state transitions , were not detected in other outbursts .
the two x - ray jets are also found to be aligned with the 1998 radio jets ( corbel et al .
2002 ) , which is also consistent with the assumption of a common origin of the two jets , but by itself does not provide definite evidence for the assumption unless the source is precessing strongly .
based on the above arguments , we conclude that it is reasonable to assume that both the eastern and western jets originated from the same 1998 outburst . our calculation results are listed in table 2 .
a distance of approximately 23 arcsec is obtained for the western jet with a slowly increasing value ; errors are estimated following the method described by tomsick et al .
we calculate the source centroid for the central source and the x - ray jet respectively and for all the five observations , the changes of the newly calculated position with the previous results are less than 0.5 . therefore , an upper limit of 0.5is set for the error of the jet distance ( except the eastern jet in observation 6 , where a larger error of 1 arcsec was set because of the manual selection method we applied ) .
we divide the net angular separation by the time interval between the neighboring observations to estimate the average proper motion for the jets , and from the results , an approximate estimate of deceleration could be seen for both jets .
the absolute astrometry of chandra is calibrated by fixing the x - ray position of the central source as its radio position reported by corbel et al ( 2001 ) .
rrrrrrr 1 & 679 & 628 & [email protected] & & [email protected] & + 2 & 1845 & 700 & [email protected] & & [email protected] & + 3 & 1846 & 720 & [email protected] & & [email protected] & + 4 & 3448 & 1265 & [email protected] & [email protected] & [email protected] & [email protected] + 5 & 3672 & 1365 & & [email protected] & & [email protected] + 6 & 3807 & 1462 & [email protected] & [email protected] & [email protected] & [email protected] + 7 & 4368 & 1588 & & [email protected] & & [email protected] + 8 & 5190 & 1856 & & [email protected] & & [email protected]
we extract the x - ray spectrum in 0.3 - 8 kev energy band for each observation . to be consistent with previous analysis of the eastern jet , we follow some of those previous procedures . because of the much longer exposure time , the total counts collected from each of the last five observations are much more than those from the first three .
we use a circular source region with a radius of 4 , an annular background region with an inner radius of 5and an outer radius of 15 , for each observation .
the photon number inside each selected region is about a few hundreds , possible for spectral analysis , compared to only @xmath120 counts in the jet region in observations 1 , 2 and 3 .
we perform background subtraction but did not make independent background spectra .
instrument response matrices ( rmf ) and weighted auxiliary response files ( warf ) are created using ciao programs _ mkacisrmf _ and _ mkwarf _ and used in the fitting in _
xspec_. we re - bin the spectra with 10 counts per bin and fit them in _
xspec_. several spectral models are tested , but the statistical quality is not high enough to distinguish between them completely . taking observation 4 for example ( for the reason that the photon number of this observation is the largest ) , synchrotron model ( powerlaw ) and self compton scattering ( compsl ) seem to work equally well with the corresponding reduced @xmath14 of 0.924 and 0.925 .
their combination ( powerlaw+compsl ) does not improve the fitting ( @xmath15 ) , but creates a large uncertainty for each parameter ( the best fitting photon index is @xmath16 in the combined model ) . in table 3 ,
the results of spectra fitting to the western jet with an absorbed power - law model are shown ( spectra shown in fig .
the absorption column density is fixed to the galactic value in the direction of xte j1550 - 564 obtained by the radio observations ( @xmath17cm@xmath10 ) ( dickey & lockman 1990 ) .
the first two observations ( obs .
4 @xmath3 5 ) have been analyzed by kaaret et al .
( 2003 ) , and our results are quite consistent with their work .
the mean photon index of the five observations is around 1.8 , also consistent with the previous work on the eastern jet ( tomsick et al .
a slightly different photon index is obtained for observation 6 and 8 ; however , the value of 1.8 is still within the 1-@xmath19 error range .
the calculated absorbed energy flux in 0.3 - 8 kev band is comparable to the value of the eastern jet . the observed flux decayed from @xmath20 erg cm@xmath10 s@xmath21 in march 2002 to only one sixth of this value in october 2003 ( see section 4.2 ) . for observation 6
, we also try to fit the spectrum of the eastern jet .
however , the data points are just too few to yield a satisfactory result . with only 17 counts between 0.3 to 10 kev inside a 8-arcsec - diameter circle
, we obtained a reduced @xmath14 of only 0.13 when using the chi - squared statistic in fitting .
when changing to the cash statistic , a more suitable method for low counts cases , the resulting photon index is 1.11 and the estimated flux is 1.6@xmath910@xmath22 erg cm@xmath10 s@xmath21 , even higher than the value in observation 4 .
we consider this result not very convincing , or the re - heating process is more complicated than a simple shock .
complete understanding of this problem is beyond the scope of the current work . as a result
, we drop this spectral data point in the light curve fitting .
8 respectively .
an absorbed power - law model with the neutral hydrogen column density fixed to the galactic value , @xmath23cm@xmath10 , is used in the fitting .
values of all fitted parameters are listed in table 3.,height=680 ] rrrrrrr 4 & 1265 & 419 & 0.016 & [email protected] & 0.91 & @xmath24 + 5 & 1355 & 248 & 0.014 & [email protected] & 0.91 & @xmath25 + 6 & 1462 & 197 & 0.008 & [email protected] & 0.90 & @xmath26 + 7 & 1588 & 114 & 0.005 & [email protected] & 0.80 & @xmath27 + 8 & 1856 & 137 & 0.003 & [email protected] & 1.76 & @xmath28 +
we now describe the model of wang et al .
( 2003 ) , and apply it to the data , demonstrating how it needs to be revised to fit the full set of observations of the eastern and western jets . the kinematic and light curve models are constructed according to the external shock model in the standard gamma - ray burst ( grb ) afterglow theory . in this model
, the kinematic evolution and light curve of the jets can be interpreted by the interactions between the jet materials and the interstellar medium ( ism ) .
the relativistic ejecta losses its energy to the ism and is therefore slowed down to non - relativistic phase . in wang et al .
( 2003 ) , they used the reverse shock emission model to fit the light curve of the eastern jet successfully .
we therefore apply this model to the analysis of the western jet as well . as we know , in the external shock model for afterglows of grbs ,
the kinematic and light curve evolution could be understood as the interaction between the outburst ejecta and the surrounding ism ( rees & mszros 1992 ) .
since microquasar systems are also galactic sources embedded inside the ism , we may infer that the jets of microquasars should also encounter such interactions and thus be decelerated during their expansion .
the large scale and long existing time of xte j1550 - 564 jets have provided us a good opportunity to test this scenario . because of the good alignment of the two jets and the central accreting object , we adopt the model of a collimated conical beam with a half opening angle @xmath29 expanding into the ambient medium with the number density @xmath30 .
the initial kinematic energy and lorentz factor of the outflow material are @xmath31 and @xmath32 , respectively .
shocks should arise as the outflow moves on and heat the ism , and its kinematic energy will turn into the internal energy of the medium gradually .
neglect the radiation loss , the energy conservation equation writes ( huang , dai , & lu 1999 ) : @xmath33 the first term on the left of the equation represents the kinematic energy of the ejecta , where @xmath34 is the lorentz factor and @xmath35 is the mass of the original ejecta .
the second term represents the internal energy of the swept - up ism , where @xmath36 and @xmath37 are the corresponding lorentz factor and mass of the shocked ism respectively , and @xmath38 with @xmath39 as the shocked gas volume .
coefficient @xmath19 differs from 6/17 to 0.73 for ultra - relativistic and non - relativistic jets ( blandford & mckee 1976 ) .
wang et al .
( 2003 ) took @xmath400.7 and we adopt their approximation in our work . equation ( 1 ) and the relativistic kinematic equations @xmath41 can be solved and give the relation between the projected angular separation @xmath42 and time @xmath43 . in equations ( 2 ) , the subscript ` a ' and ` r ' represent the approaching and receding jets in a pair of relativistic jets respectively .
@xmath44 is the distance between the jet and the source , which can be transformed into the angular separation by @xmath45 kpc , and @xmath46 is the jet inclination angle to the line of sight .
we can get the @xmath47 curve numerically with the above equations . to be consistent with the work done to the eastern jet ,
we test first the same initial conditions that @xmath48 , @xmath49 erg , @xmath50 and @xmath51 as those found by wang et al .
( 2003 ) and would like to find how these parameters affect the fitting of the jet motions . in the case of the eastern jet
, the number density of the ism was assumed by wang et al .
( 2003 ) as a constant in the whole region outside the central source , thus the interaction was supposed to have taken place all the way along the jet s path .
this assumption does not work well in the case of its western counterpart , because the western jet has traveled by 2002 as far as the eastern jet had traveled by 2000 , but was decelerated much more significantly .
this can be seen in fig .
the western jet moved comparatively little after its appearance and acted like hitting into a wall " .
this requires much stronger interactions and thus a much denser environment .
we tried to fit the western jet itself but were not very successful .
if we increase the number density , the gas will simply block the jet and the jet will never travel to such a far distance ( a 10@xmath52 erg jet can only move to @xmath17 - 8 arcsec in 1 @xmath2 ism ) .
on the other hand , if the initial energy is too large ( @xmath110@xmath53 erg ) , the jet could penetrate ism and travel far but would be very difficult to slow down so quickly ( for such high energy , even a density of 100 @xmath2 can not bend the curve down ) . changing
the initial lorentz factor has similar results with changing initial energy and the opening angle of the jet affects the results quite little .
this tendencies are plotted in fig.4 ( a ) , where the fitting to the eastern jet shows the model of wang et al .
( 2003 ) and the corresponding solid - line fitting to the western jet uses the same model and same parameters .
the dashed line and dot dashed line there represent the results of changing the number density or the initial kinetic energy to ten times of the eastern results .
all three curves do not match the data points well . as a result , we consider keeping the initial lorentz factor and energy , while changing the geometric assumptions to describe the jet motions . for the inclination angle , we first examined the paper of orosz et al .
( 2002 ) , where they find a value of 72.6 @xmath6 4.8@xmath8 from the optical observations .
we took this value , but varying it within their 1-sigma error bars , and find @xmath54 gives out the best fit . to simplify the problem , we just fix the inclination angle to this value in our work . after setting the inclination angle
, we test a model that the ism density varies as the distance changes , i.e. , the density is lower in the center and higher in the outside region . for simplicity
, we test the case that the jet traveled first through a cavity " with a constant velocity and then through a dense region and was decelerated there . for this modification
, we introduce another parameter @xmath55 , the outer radius of the cavity .
the ism number density is set to be a constant @xmath30 outside this region and zero inside .
the results are shown in panel ( b ) and ( c ) in fig.4 .
we first apply this model to the eastern and the western jet separately .
the results are quite satisfactory but not well constrained . because of only 10 data points in constraining 4 free parameters ( @xmath56 , @xmath57 , @xmath58 , @xmath59 ) , there are more than one group of parameters that seem work well .
fitting the two jets simultaneously with a fixed cavity radius helps a bit .
it constrains the fitting tighter and @xmath60 arcsec seems an acceptable result for the both sides ( fig.4 ( b ) ) .
however , this is not a sound assumption .
the inferred ism number density corresponding to this value is 0.015 @xmath2 in the east and 0.06 @xmath2 in the west respectively , indicating a clear asymmetry in the ism density on the two side .
thus it is not reasonable to assume a perfect symmetric geometry .
we still need other analysis to help determining consistent parameters . in the standard grb scenario
, the afterglow emission is produced by the synchrotron radiation or inverse compton emission of the accelerated electrons in the shock front of the jets .
the forward shock emission is from the heated electrons in the swept - up ambient medium , whereas the reverse shock emission is from the electrons of the jet itself when a shock moves back through the ejecta .
the same processes should also happen in the case of microquasars , with only the lorentz factor much lower . in wang et al .
( 2003 ) , they tested these two possibilities and proved that the forward shock emission would decay too slowly to fit for the observed decay index @xmath61 of the eastern jet , because of the continuously heating effect .
however , since only operating once , the reverse shock emission could decay rather fast as the electrons cooled adiabatically and this model indeed performed quite well in describing the data . in the case of the western jet ,
the energy flux decayed even more rapidly ( decay index @xmath62 , see fig .
5 ) and as a result , we take the reverse shock mechanism in our work .
assuming the distribution of the electrons obeys a power - law form , @xmath63 , for @xmath64 , the volume emissivity at frequency @xmath65 in the comoving frame is given by ( rybicki & lightman 1979 ) @xmath66 where @xmath67 with @xmath68 and @xmath69 is the bessel function .
the physical quantities in these equations include @xmath70 and @xmath71 , the charge and mass of the electron , @xmath72 , the magnetic field strength perpendicular to the electron velocity , and @xmath73 and @xmath74 , the characteristic frequencies for electrons with @xmath75 and @xmath76 . assuming the reverse shock heats the ejecta at time @xmath77 at the radius @xmath78 ( with the assumptions of no synchrotron cooling , conservation of the total number of electrons and the magnetic field being frozen into the plasma ) , the physical quantities in the adiabatically expanding ejecta with radius @xmath44 evolve as ( van der laan 1966 ) @xmath79 @xmath80 where the initial values of these quantities are free parameters to be fitted in the calculation . with these assumptions
, we can then calculate the predicted flux evolution of the jets .
the comoving frequency @xmath65 relates to our observer frequency @xmath81 by @xmath82 , where @xmath83 is the doppler factor and we have @xmath84 and @xmath85 for the approaching and receding jets respectively . considering the geometry of the emission region , the observed x - ray flux in 0.3 - 8 kev band
could be estimated by @xmath86d\nu,\ ] ] where @xmath87 is the width of the shock region and is assumed to be @xmath88 , after wang et al .
( 2003 ) in the calculation . to reduce the number of free parameters
, we set @xmath89 in our calculation because the results are quite insensitive to this value . according to our kinematic model in section 4.1
, we choose the time that the reverse shock takes place to be the time that the lorentz factor reduced to @xmath90 of its original value , which is called deceleration timescale @xmath91 in the grb external shock model , which is supposed to be the strongest point of the external shock ( private communication with x. y. wang ) .
then we fit the data to find out the initial values of @xmath92 and @xmath72 . just like the fitting to the kinematics of the jets
, we could not decide the best fitting result only using the flux data since we could always find one group of parameters that fit the flux data approximately well for each parameter @xmath55 .
thus , we combine the kinematic and light curve fitting together to find some more useful hints . we know that the energy and the number density of the gas in the pre - shock and post - shock regions are connected by the jump conditions @xmath93 and @xmath94 , where @xmath95 and @xmath96 are coefficients related to the jet velocity ( wang et al .
therefore if we assume the shocked electrons and the magnetic field acquire constant fractions ( @xmath97 and @xmath98 ) of the total shock energy , we have @xmath99 and @xmath100 for p@xmath1012 .
since we have assumed that @xmath102 and @xmath32 ( so that @xmath103 ) are equal for the two jets , if we further assume that factor @xmath104 of the eastern and the western jets is also the same , we may infer that @xmath105 for the two jets .
we therefore search for the combination of parameters that could satisfy the kinematic and light curve fitting , as well as the relationship @xmath106 . to search for the best parameters fitting the data , we follow the following procedure . with numerical calculations , for the 19 data points ( including 10 kinematic data and 9 light curve data )
we build a large 8-dimensional database for the eight parameters to be estimated , with one constraining relationship , @xmath106 .
intuitively it looks very difficult to determine so many parameters from so few data points .
however , one approach we take is that we divide the fittings into two stages : we first fit the kinematic data regardless of the lightcurve information and get a series of @xmath107 all of which describe the data almost equally well .
then , for each set of @xmath107 , we fit the lightcurve and calculate the joint @xmath14 value to choose the group of parameters yielding the least @xmath14 value .
the advantage of this approach is that , since @xmath107 is fixed every time in the fitting of the lightcurve , we consider the kinematic fittings to be independent of the lightcurve fittings in calculating @xmath14 . at each search step ,
the total @xmath14 is calculated corresponding to the 19 data points .
a set of parameters that yields the minimum total @xmath14 is taken as our best fitting parameters .
the resulting total @xmath14 is 7.05 and the errors are estimated by searching for the range for one parameter that can change the @xmath14 value by a given delta while keeping all the other parameters fixed ( press et al .
1992 , constant chi - squared boundaries as confidence limits " ) .
the result corresponding to the lightcurve fitting is shown in fig.5 and the corresponding kinematic fitting is shown in panel ( c ) in fig.4 .
the best fitting parameters are listed in table 4 . from the parameters ,
we conclude that the boundary of the cavity lies at @xmath10812 arcsec to the east and @xmath117 arcsec to the west of the central source .
the corresponding number density of the ism outside this boundary is @xmath10.0034 @xmath2 and @xmath10.12 @xmath2 , respectively .
these values are both lower than the canonical ism value of @xmath11 @xmath2 , although the value in the western region is much higher than in the eastern region .
the asymmetry of the density on the both sides probably involves the generation history of the cavity , which should be explored with further studies .
the electron energy fraction relationship is satisfied as @xmath109 .
however , we should mention here that the other relation concerning the magnetic field strength ( if @xmath110 are equal on the two sides , then @xmath111 ) could not be satisfied simultaneously by these parameters . although the cavity radius and the number density are allowed to vary significantly , the best fitted magnetic field strength remains quite stable ( @xmath10.35 - 0.55 mg ) .
one possible interpretation for this is that the equipartition parameter varies as the physical conditions of the jet varies or equipartition assumptions do not hold here for the magnetic fields since this is not a steady system ; an alternative explanation may involve the _ in situ _ generation ( or amplification ) of the magnetic field ( e.g. , originated from external ism instead of from the jet itself ) . in the fitting , the assumed initial lorentz factor is @xmath48 , indicating an initial velocity of @xmath112 .
the fixed spectral index , @xmath113 , is consistent with the value obtained by xue et al .
( 2008 ) in their broadband spectral analysis ( @xmath114 ) .
the fitted magnetic field strengths is lower than the value given by xue et al .
( 2008 ) ( @xmath115 mg ) . however , they claimed a large uncertainty in their fittings .
we look forward to further analysis to explore this more clearly and take our fitting as a reasonable one at this stage .
rrrrrr @xmath55 ( arcsec ) & @xmath116 & @xmath117 & & & cavity radius + @xmath30 ( @xmath2 ) & @xmath118 & @xmath119 & & & ism density outside the cavity + @xmath46 ( ) & 68 & 68 & & & line of sight angle + @xmath120 ( arcsec ) & 15.0 & 17.9 & 0.03 & 0.03 & reverse shock hitting place + @xmath77 ( days ) & 343 & 842 & & & reverse shock hitting time + @xmath121 ( mg ) & @xmath122 & @xmath123 & & & initial value at @xmath77 + @xmath124 ( @xmath125 & @xmath126 & @xmath127 & & & initial value at @xmath77 + @xmath14 for fitting & kinematics : 1.46 & kinematics : 1.55 & & & total @xmath14=7.05 reaches the minimum + & light curve : 1.68 & light curve : 1.68 & & & +
h 1743 - 322 is an x - ray transient first discovered in 1977 ( kaluzienski & holt 1977 ) . it was then classified as a black hole candidate in 1984 based on its x - ray spectral characteristics ( white & marshall 1984 ) .
new activity was found by integral in 2003 ( revnivtsev et al .
2003 ) , and a bright radio flare was observed by vla on 2003 april 8 ( rupen et al . 2003 ) . @xmath0 x - ray and atca radio observations from 2003 november to 2004 june revealed the presence of large - scale jets on both sides ( rupen et al . 2004
; corbel et al .
the source is then labeled as a microquasar " .
we take the data from the work of corbel et al .
( 2005 ) and fit the proper motion with the external shock model . following their approach ,
the ejection date is set to the time of the major radio flare ( 2003 april 8) .
the distance to h 1743 - 322 is basically unknown now .
however , its location toward the galactic bulge could possibly imply a galactic center location . assuming a source distance of 8 kpc ( distance to the galactic center ) ,
the first radio detection of the jets gives out an intrinsic velocity of the ejection of @xmath128 and an angle of @xmath129 for the axis of the jets ( corbel et al . 2005 ) .
however , the evolution of the jets shows that the velocity is not constant all the way through .
deceleration is required since the linear extrapolation of the proper motion data yields an ejection date earlier than the chosen zero point time , as shown in panel ( a ) in fig .
the linear extrapolation of the zero point and the first data point also gives out lines with steeper slope than the real data . as a result
, we also apply the external shock model to this source .
the line of sight angle is set to be @xmath129 in all fittings .
it is worth mentioning that the 73 degree angle is only valid if the jets have both been propagating outwards with the same speed and the same angle to the line of sight and the source is assumed to be located at the galactic center with a distance of 8 kpc .
however , since there are no better constraints on the jet orientation , we simply take this value as a reasonable approximation .
we first test the model that assumes the source located in a continual gas medium . in this case
, the jets would decelerate gradually after ejection .
the model fits the data quite well this time ( panel ( b ) in fig.6 , with @xmath14=1.03 for the approaching jet and @xmath14=0.98 for the receding jet , with degree of freedom=2 for each case ) , indicating that a cavity is not required in this source .
however , we also notice that the best fitting density of the environment ism is @xmath130 @xmath2 , much lower than the canonical galactic ism value .
we also apply the cavity model to the fitting , which describes the data equally well ( panel ( c ) in fig .
6 , with @xmath14=0.95 for the approaching and @xmath14=0.97 for the receding jet , with degree of freedom=2 for each case ) , with a cavity region with radius of 3 arcsec on both sides and the gas density of @xmath131 @xmath2 .
this value is one magnitude higher than the previous attempt , but still quite low . due to the limited data
, we can not test if the environment outside the central source is symmetric or not .
the flux is not fitted for this source because of the limited number of x - ray data points .
erg , @xmath130 @xmath2 .
panel ( c ) : deceleration in medium outside a cavity region .
results : @xmath132 erg , @xmath133 arcsec , @xmath131 @xmath2 .
, width=340,height=680 ]
the above analyses of xte j1550 - 564 and h 1743 - 322 led us to the conclusion that in microquasars the interactions between the ejecta and the environmental gas play major roles in the jet evolution .
we could further argue that the low density of the environment is a necessary requirement for the jet to develop to a long distance .
heinz ( 2002 ) derived the scale relations for the jets from accreting black holes , using a simple analytic model .
he adopted the jet - ism interaction scenario to estimate the slow - down distance of the ejecta : @xmath134 where @xmath135 erg is the kinetic energy , @xmath136 is the lorentz factor of the jet , @xmath137 is the external gas density in the unit of @xmath2 and @xmath138 is the opening angle of the jet .
he applied this estimation to microquasars grs 1915 + 105 and gro j1655 - 40 . in both cases
, he found that the upper limit on the gas density is roughly @xmath139 @xmath2 , given @xmath140 of 0.05pc , quite consistent with the ism density found for xte j1550 - 564 and h1743 - 322 in this work .
for a larger @xmath140 , a lower @xmath137 should be required .
heinz s approach led to an interesting comparison between the microquasar jets and the radio quasar jets .
he claimed that the jets length in grs 1915 + 105 ( @xmath1410.04 pc ) would correspond to a jet length of 4 mpc when scaled by @xmath142 ( mass of the central black hole ) to agn conditions ( such as m87 or cyg a ) ( 2002 ) .
we could follow this way and infer that the jet in xte j1550 - 564 ( @xmath1410.5 pc ) will correspond to a 50 mpc long agn jet , which has never been detected .
this again confirms that the environment of microquasars should be comparatively vacuous , in a dynamical sense even ( although not in the absolute sense ) less dense than the agn environment ( the intergalactic medium of densities between 10@xmath143 @xmath2 to 10@xmath10 @xmath2 ) , when compared to the thrust of the jets in microquasars and agns . when presented in unit of pc with the information of source distance and scaled by the central black hole mass ,
the proper motions of different microquasar jets can be plotted in one figure ( fig .
7 ) ( grs 1915 + 105 : 14@xmath64 @xmath144 ( greiner et al . 2001 ) , 12.5 kpc ( rodrguez et al . 1995 ) ; gx 339 - 4 : [email protected] @xmath144 ( hynes et al . 2003 ) , lower limit estimation of distance of 4 kpc ( zdziarski et al . 1998 ) ; xte j1550 - 564 : [email protected] @xmath144 and 5.3 kpc ( orosz et al .
2002 ) ; h 1743 - 322 : since no good mass and distance observation up to date , we follow the assumption of 10 @xmath144 ( miller et al . 2006 ) and galactic center origin of 8 kpc ( corbel et al . 2005 ) ) .
it is interesting to notice from this figure that although the detailed properties of the binary systems ( e.g. , the compact objects and the companion stars ) differ a lot from source to source , the normalized jet proper motions are quite consistent .
this is not surprising since the jet properties are set by the compact object ( and its immediate neighborhood ) and their evolutions are mostly influenced by the surrounding environment .
although the companion star provides the material which is eventually launched into the jets , and thus affects the mass transfer rate and the mass flux in the jets , it nevertheless does not seem to play significant roles for the structure and evolution of the large scale jets and cavities surrounding the microquasar .
therefore the nature of accretion disk physics dominates over the nurture of the companion star , as far as the jets are concerned .
thus it leads us to the attempt to generalize the microquasar jets in one consistent picture .
this figure , together with other observational data concerning the motionless large scale jets ( e.g. , cyg x-1 , ss433 ) , has led us to the suggestion that all microquasar jets can be classified into roughly three groups : small scale moving jets , large scale moving jets and large scale jet relics . for the first type , the small jets " , only radio emissions are detected .
the jets are always relatively close to the central source and dissipate very quickly .
examples of this class include grs 1915 + 105 ( rodrguez & mirabel 1999 ; miller - jones et al . 2007 ) , gro j1655 - 40 ( hjellming & rupen 1995 ) , and cyg x-3 ( mart et al . 2001 ) .
the typical spatial scale is [email protected] pc and the time scale is several tens of days . in this class ,
no obvious deceleration is observed before the jets become too faint .
for the second type , the large jets " , both x - ray and radio detections are obtained , at a place far from the central source several years after the outburst . examples are xte j1550 - 564 , h1743 - 322 , and gx 339 - 4 ( gallo et al . 2004 ) .
the typical jet traveling distance for this type is [email protected] pc from the central engine .
deceleration is clearly observed in this type of sources .
the last type , the large relics " , is a kind of diffuse structures observed in radio , optical and x - ray band , often ring or round nebula shaped that are not moving at all . in this class , some well studied sources , cygnus x-1 ( gallo et al .
2005 ) , ss433 ( dubner el al .
1998 ) , circinus x-1 ( stewart et al .
1993 , tudose et al . 2006 ) and grs 1915 + 105 ( kaiser et al .
2004 ) are included .
the typical scale for this kind is 1@xmath130 pc , an order of magnitude larger than the second type .
the estimated lifetime often exceeds one million years , indicating that they are related to previous outbursts . from these properties
, it is reasonable to further suggest a consistent picture involving all the sources together .
we make a conjecture that extreme low density regions , or namely , large scale cavities , exist in all microquasar systems .
the small jets " observed right after the ejection are just traveling through these cavities . since there are few or none interactions between the jets and the surrounding gas in this region , the jets travel without obvious deceleration .
the emission mechanism is synchrotron radiation by particles accelerated in the initial outburst .
the emissions of jets decay very quickly and are not detectable after several tens of days . in some cases ( e.g. xte j1550 - 564 ) , the cavity has a dense ( compared to the cavity ) boundary at some radius and the interactions between the jets and the boundary gas heat the particles again and thus make the jets detectable again .
those are the large jets " .
the emission mechanism then is synchrotron radiation by the re - heated particles in the external shocks .
then , after these interactions , the jets lost most of their kinematic energy into the ism gradually , causing the latter to expand to large scale structures , the large relics " , in a comparatively long time ( thousands to several millions of years ) and create the observed nebula - like jet - inflated bubbles .
different systems are allowed to fall into more than one of the three postulated groups .
for example , cygnus x-1 has small - scale ejecta ( fender et al . , 2006 ) and also has a large - scale relic ( gallo et al . , 2005 ) .
also xte j1550 - 564 has small - scale ejecta ( hannikainen et al . , 2001 ) and also large - scale jets ( corbel et al . , 2002
as we proposed in the above paragraph , the three observational groups are perhaps just three stages in a complete ejection evolution process : the small jets " observed in the outburst events will travel in dark and manifest themselves as the large jets " when they hit the boundary of the cavity and eventually develop into the large relics " gradually by the continuous interactions and energy dissipation .
if this conjecture holds , then the three groups may not be actually three " but just three phenomena in one " system . the three groups identified here are probably just observational manifestations of one generic mechanism , and different sources can belong to multiple groups .
if this general model of microquasars is true , properties of different jet systems could be dealt with separately in one consistent scheme .
one implication is that the electrons from the large jet " and the small jets " belong to different groups of electrons , from the ejecta itself and the ism respectively , which means we could not simply use the properties of electron distributions in the large jet " blobs to determine the emission properties of the central source in the ejection ( xue et al .
2008 ) .
one important remaining question would be why there are such vacuous regions and how to generate them .
there are several possibilities , involving previous outbursts , jets or winds activities .
before we go further into these possibilities , we should first distinguish one class from the others .
we have used the term cavity " all through our discussion , but actually there are two geometric possibilities for a cavity .
one is a cavity " , the other is a tunnel " . instead of a spherical cavity , it is possible that there is only a vacuum conical path outside the central engine .
it may be created by previous jet ejection events and got fully developed through a series of ejections .
this could explain the collimation of the jets naturally , but it requires almost continuous jet ejections ( contrary to the episodic jet ejections discussed here ) to prevent the surrounding gas to fall back and fill up the thin path between ejections .
this assumption is of great interest because it provides a possible connection between the large - scale structures and the persistent , low - power , steady jets reported in several important observations ( etc .
cyg x-1 ( stirling et al .
2001 ) and grs 1915 + 105 ( dhawan et al .
kaiser and alexander ( 1997 , hereafter ka97 ) has developed a model for radio galaxies in which continuous and collimated jet may clear its way out and create a hot spot at the end of its path , and the hot electrons escaped from the shocks would form a cocoon " outside the central compact object . if this is also the case for galactic accretion systems , as postulated for cygnus x-1 ( gallo et al . , 2005 ) , circinus x-1 ( tudose et al . , 2006 ) , and grs 1915 + 105 ( kaiser et al . , 2004
) , then the vacuous regions ( tunnel and cocoon ) and the large relics " are perhaps both the results of the continuous jets traveling in dark .
the two galactic center sources 1e1740 - 294 and grs 1958 - 258 may also belong to this case . in this frame of assumptions , the episodic outbursts are not included here : the outbursts will increase the instantaneous mass flux down the jet , however since low - mass x - ray binaries spend the majority of their time in a low - luminosity quiescent state , the outbursts will have little effect on the time - averaged jet luminosity , and thus on the long - term evolution of the relics .
however , we may infer that if there are dark continuous jets and dark relics in a system like xte j1550 - 564 , a strong episodic outburst would just hit and brighten the boundary of the relic and may provide a bridge between the source and the relic .
if there are multiple outbursts , we will expect repeated large jets " in the same direction with larger and larger lengths .
future observations would then be able to testify this possibility .
rrrrrrrr gro j1655 - 40 & @xmath0 & 7 & 10.5 & 300 - 1000 & 300 - 500 & @xmath145 & 1900 + & & & & & & @xmath146 & 600 + & xmm - newton & & & 2600 - 4500 & & @xmath147 & 6000 + & & & & & & @xmath148 & 3000 + h 1743 - 322 & @xmath0 & rxte & 10 & 15 & @xmath149 & @xmath150 & @xmath151 & 4240 + grs 1915 + 105 & asca & @xmath152 & 21 & 1000 & & @xmath153 & 1340 + & chandra & & & @xmath154 & @xmath155 & @xmath156 & 770 + apart from this case , several other possibilities may take place .
one possible way to create the observed cavities is through recent supernovae outbursts that produced the central black holes .
however , only hmxbs should be close to their parent supernova remnants ( being young ) ; accumulated velocity dispersion ( scattering from spiral arms or giant molecular clouds over time ) will carry lmxbs far from their natal supernova remnant in a few galactic orbits , e.g. for gro j1655 - 40 ( israelian et al .
furthermore , some sources such as cygnus x-1 most likely never had supernovae , because of their low peculiar velocity ( mirabel & rodrgues 2003 ) . in the case of recent supernovae ( @xmath157 years ) and
low kick - out velocity for the black hole , snrs surrounding them should have been detected , and for much older supernovae and very low kick - out velocity for the black hole , the microquasars may still resides in the snrs but regular interstellar medium should have filled those very old snrs now . winds from progenitors of the central compact objects at their last stages may also be an interpretation .
cavities in hydrogen maps have been detected around neutron stars and were supposed to be driven by wind from the progenitors of the stars ( gaensler et al .
however , the lifetime of such bubbles were estimated to be of order of @xmath11 myrs . for a region still as vacuous as we have estimated in our studies
, the compact object is expected to be much younger than that .
winds of the companion star in the binary are also a possible candidate .
wind - blown bubbles and shells are observed around evolved massive stars , with typical radii of @xmath12 - 10 pc for wolf - rayet stars ( gruendl et al .
2000 ) and @xmath10.1 - 2.3 pc for luminous blue supergiants ( smith et al .
. however , the estimated density at the bubble shells is of order of @xmath1100 @xmath2 , simply too high for our case , and among all the well established microquasar systems , only the companion star in cyg x-1 is o type ( o9.7iab ) ( remillard & mcclintock 2006 ) , whereas most of the other systems contain only m , k or g type companion stars , which are not powerful enough to generate such large scale bubbles .
a plausible possibility involves the accretion disk winds . in this scenario
, it is the disk winds with mild or non - relativistic velocities that have pushed away the surrounding ism gas and created the low density regions surrounding the central source .
the wind should go as shells that travel in a ballistic way while interacting with the ism cold gas .
indeed , in several microquasars , fast ionized winds have already been observed ( miller et al . 2006 ; kotani et al . 2000 ; netzer , 2006 ; sala et al .
2006 ; fuchs et al .
strong absorption metal lines are detected with obvious blueshift .
the speed of the outflow could be calculated from the blueshift of these lines .
the ejection place could also be estimated using the measured ionization parameters .
we list some of the results in table 5 , including the calculated escape velocity at the inferred ejection radius ( @xmath158 and @xmath159 ) . also , in another interesting work , fuchs et al .
( 2006 ) showed that there are indeed strong subrelativistic winds coming out of the ss 433 system , very much alike the winds from a wr star or from a thick torus or envelope outside the accretion disk and the central object .
the estimated mass lost rate is @xmath160 yr@xmath21 for discontinuous winds and 3 times higher for continuous winds , much higher than the observed jet mass loss of @xmath110@xmath161 yr@xmath21 in this system , and a double - cone structure may be formed by these winds .
these inferences are also supported by blundell et al .
( 2001 ) , where the equatorial radio emission seen in ss433 was taken as evidence for a disc wind .
these observations support the idea of co - existence of jets and winds in the microquasars systems .
we also notice that the wind velocity in h1743 - 322 is lower than those in grs 1915 + 105 and gro j1655 - 40 , suggesting larger cavities are formed for the latter two sources than for the former one .
it is consistent with the fact that no decelerating jets have been observed in these two sources up to now .
however , whether the winds are powerful enough to create such large cavities in all cases for all sources is unclear at this stage .
the terminal places of the wind shells ( which determine the size of the cavity ) are determined by the wind energy and geometry , specifically , the wind velocity at the ejection place and the opening angle , and the interaction conditions between winds and the gas .
these parameters are still very uncertain nowadays because of the lack of enough studies and the densities of the winds are only estimated at the foot of the ejection place .
we would roughly estimate that , in order for the wind to move to large radii far from the black hole , the wind velocity should be quite close to the escape velocity at the ejection radius . from table 5 , we can see that most of the observed wind velocities are comparable to , although not as large as , the local escape velocity , with the exception of the recently discovered disk wind of about its escape velocity in grs 1915 + 105 in its soft state , with mass loss rate sufficient to suppress jet ejection ( neilsen & lee 2009 ) . therefore in the last case
the accretion disk wind alone ( because no jet is expected in the soft state ) may play a significant role for forming a large scale cavity .
however for other cases the winds alone may not be enough to power the cavities , due to insufficient velocities .
therefore some kinds of acceleration or velocity - maintaining mechanisms may be required , in order for the winds to expand far enough ( @xmath10.1 pc to 10 pc ) .
magnetic fields may play an important role here .
it has been known that rotating wind - up magnetic fields can accelerate and launch jets ( see spruit 2008 and references therein ) ; therefore continuous high velocity winds can set up the favorable condition for producing continuous outflows or jets with velocities around or even exceeding the escaping velocities of these systems .
once continuous high velocity outflows or jets are produced , they should be able to inflate their surrounding interstellar medium to form bubbles , as we have discussed earlier within the framework of the ka97 model .
finally , we comment that it seems counter intuitive that the boundary of the cavity lies further in the western side , yet the ism density is also higher in the west . it would be a problem if the cavity is produced by the observed large scale two - sided jets with equal power .
therefore this asymmetry may be used to argue against such a model . however , the asymmetric cavity may be naturally produced if the system has a non - negligible space velocity with respect to the local ism along the eastern direction . in this case
additional energy is injected into the ism on the head - on side along the direction of motion of the system , in analogy to process producing the head - on bow - shock of young spin - down powered pulsars moving in ism ( e.g. , caraveo et al 2003 ) .
therefore the cavity on the approaching side should be emptier ( with smaller density ) than the opposite wind or jet , unless the system moves perpendicular to the wind or jet .
the system itself is also catching up with the cavity on the approaching side , placing itself closer to the boundary of the cavity on the approaching side .
for example , a space velocity of 100 km / s with respect to the local ism would displace the source by 0.1 pc in less than 1000 years . in the absence of data for this system on its proper motion and the power of wind or continuous ( dark ) jets
, we will not speculate further on what combination of parameters may eventually produce the observed asymmetry of the cavity .
this should be tested in the future .
heinz et al .
( 2008 ) have proposed that the microquasar s space velocity should be important and leave trails behind their moving path in the galaxy .
they estimated the volume of radio plasma released by these microquasars and claimed that they should be observable at low frequencies .
future observational confirmation of their prediction for xte j1550 - 564 would provide evidence for our conjecture .
two large - scale x - ray jets have been observed in xte j1550 - 564 . we have analyzed the @xmath0 x - ray data for these two jets and fitted their kinematics and light curves with the external shock model in the grb afterglow model . in this model ,
the interactions between the jet material and the surrounding ism slow the jets down and accelerate the jet particles to radiate the observed radio and x - ray synchrotron radiations . under such assumptions ,
the number density of the ism in the surrounding regions of the central black hole is a key parameter .
we found in the fitting that this number density has to be extremely low , or the strong interactions would block the jet s expansion . however on the other hand , the number density at @xmath120 arcsec has to be comparatively high ( but still lower than the typical ism ) in order to allow the jet to be decelerated quickly , as the observational data indicated .
thus , a model consisting of two different regions , a cavity and its gas boundaries , is tested and shown to work well .
the cavity is found to be large scale ( @xmath10.4 pc in size ) and asymmetric ( @xmath162 ) .
the reason of the asymmetry of the environment on the two sides is not clear from current observational data .
the deceleration described in the model is quite obvious ( from @xmath48 to @xmath163 at @xmath164 days ) , thus it is the first time that we observe the whole process of the deceleration of a pair of relativistic jets .
the similar analysis of h1743 - 322 also supports this jet - ism interaction scenario .
large scale decelerating jets are also observed in this source and the interaction model describes the data consistently .
whether a cavity exists is not clear in this case ; nevertheless the density is also found to be very low , compared to the canonical galactic value . with these studies , a generic model of the environment of microquasars is proposed .
we suggest that microquasars are located in large scale and very low density cavities .
the cavity provides the space for the jets to go through freely and the interactions between the jets and the material outside the cavity provide the deceleration mechanism of the jets .
the cavity is likely created by continuous high velocity outflows or jets , which are seeded by accretion disk winds that have been observed in several microquasar systems producing strong absorption lines .
microquasars are powerful probes of both the central engine and their surrounding environment
. more studies of the jets behaviors may give us information on the ism gas properties , as well as the ejecta components .
the link between jets and wind is a key issue in current studies of accreting systems .
if the general scenario we have proposed here is further confirmed , we then need to understand the mechanism producing these outflows ( winds and jets ) and the connections between them .
detailed analysis of the geometry and timing properties of the ejecta in the future would be of great significance ; it will not only provide insights of the jet formation process , but also offer another approach into black hole physics and accretion flow dynamics .
we are grateful to x.y .
wang for many helps in this work .
we also thank z.g .
dai and r. soria for useful discussions .
we appreciate very much the insightful comments and helpful suggestions by the anonymous referee .
snz acknowledges partial funding support by the yangtze endowment from the ministry of education at tsinghua university , directional research project of the chinese academy of sciences under project no .
kjcx2-yw - t03 and by the national natural science foundation of china under grant nos . 10821061 , 10733010 , 10725313 , and by 973 program of china under grant 2009cb824800 .
blandford , r. d. , & mckee , c. f. 1976 , phys .
fluids , 19 , 1130 blundell , k. m. , mioduszewski , a. j. , muxlow , t. w. b. , et al .
2001 , , 562:l79 caraveo p. a. et al .
2003 , science , 301 , 1345 corbel s. , kaaret p. , jain r. k. , et al .
2001 , apj , 554 , 43 corbel , s. , fender , p. r. , et al .
2002 , science , 298 , 196 corbel , s. , kaaret , p. , et al .
2005 , , 632 , 504 dhawan , v. , mirabel , i. f. , rodrgues , l. f. 2000 , , 543 , 373 dickey , j. m. , & lockman , f. j. 1990 , , 28 , 215 dubner , g. m. , holdaway , m. , goss , w. m. , & mirabel , i. f. 1998 , aj , 116 , 1842 freeman , p. e. , kashyap , v. , rosner , r. , & lamb , d. q. 2002 , apjs , 138 , 185 fuchs , y. , miramond , l. k. , brahm , p. 2006
, , 445 , 1041 gaensler , b. m. et al . 2005 , 620 , l95 gallo , e. , corbel , s. , fender , r. p. , et al .
2004 , mnras , 347 , l52 gallo , e. , fender , r. p. , kaiser , c. , et al .
2005 , nature , 436 , 819 greiner , j. , cuby , j. g. , & mccaughrean , j. 2001 , nature , 414 , 522 gruendl , r. a. , chu , y. h. , dunne , b. c. , & points , s. d. 2000 , , 120 , 2670 hannikainen , d. , campbell - wilson , d. , hunstead , r. , et al .
2001 , apss supp . , 276 , 45 heinz , s. , 2002 , aa , 388 , l40 heinz , s. , grimm , h. j. , sunyaev , & r. a. , fender , r. p. , 2008 ,
apj , 686 , 1145 hjellming , r.m . , &
rupen , m. p. 1995 , nature , 375 , 464 huang , y. f. , dai , z.g . , & lu , t. 1999 , mnras , 309 , 513 hynes , r. i. , steeghs , d. , casares , j. , charles , p. a. , obrien k. , 2003 , apj , l583 israelian , g. , rebolo , r. , basri , g. , et al .
1999 , nature , 401 , 142 jain , r. k. , bailyn , c. d. , orosz , j. a. , et al .
2001 , apj , 554 , l181 kaluzienski , l. j. , & holt , s. s. 1977 , iau circ .
3099 kaaret , p. , corbel , s. , & tomsick , j.a .
2003 , apj , 582 , 945 kotani , t. , ebisawa , k. , dotani , t. , et al .
2000 , apj , 539 , 413 kubota , a. , & makishima , k. 2004 , apj , 601 , 428 mart , j. , paredes , j. , m. , & peracaula , m. , 2001 , , 375 , 476 massi , m. , & kaufman bernad , 2008 , , 477 , 1 miller , j. m , raymond j. , homan , j. , et al . 2006 ,
apj , 646 , 394 miller - jones , j. c. a. , rupen , m. p. , fender , r. p. , et al .
2007 , mnras , 375,1087 mirabel , i. f. , & rodrguez , l. f. 1999 , , 37 , 409 mirabel , i. f. , & rodrigues , i. , 2003 , science , 300 , 1119 neilsen , j. , & lee , j. c. , 2009 , nature , 458 , 481 netzer , h. 2006 , apj , 652 , l117 orosz , j. a. , groot , p. j. , van der klis , m. , et al .
2002 , apj , 568 , 845 press , w. h. , flannery , b. p. , et al .
1992 , numerical recipes in c : the art of scientific computing rybicki , g. b. , & lightman , a. p. 1979
, radiative process in astrophysics ( new york : wiley ) rees , m. j. , and mszros , p. 1992 , mnras , 258 , 41 remillard r. a. , mcclintock , j. e. 2006 , , 44 , 49 revnivtsev , m. , chernyakova , m. , capitani , f. 2003 , atel , 132 rodrguez , l. f. , mirabel , i. f. , & mart , j. 1992 , , 401 , l15 rodrguez , l. f. , gerard , e. , mirabel , i. f. et al .
1995 , , 101 , 173 rodrguez , l. f. , & mirabel , i. f. 1999 , apj , 511 , 398 rupen , m. p. , mioduszewski , a. j. , & dhawan , v. 2003 , atel , 142 , 1 - 2004 , aas , 204 , 516 sala , g. , greiner , j. , vink , j. , et al .
2007 , , 461 , 1049 smith , d. a. , 1998 , int .
union circ .
7008 smith , n. , bally , j. , & walawender , j. 2007 , , 134 , 846 stewart , r. t. , caswell , j. l. , haynes , r. f. , & nelson , g. j. 1993 , mnras , 261 , 593 stirling , a. m. , spencer , r. e. , de la force , c. j. , et al .
2001 , mnras , 327 , 1273 sturner s. j. , & shrader , c. r. 2005 , apj , 625,923 tingay , s. j. , jauncey , d. l. , prestonet , r. a. , et al .
1995 , nature , 374 , 141 tomsick , j. a. , corbel , s. , & fender , r. 2003 , apj , 582 , 933 tudose , v. , fender , r. p. , kaiser , c. r. , tzioumis a. k. et al .
2006 , mnras , 372 , 417 van der laan , h. 1966 , nature , 211 , 1131 wang , x. y. , dai , z. g. , & lu , t. 2003 , apj , 592 , 347 white , n. e. , & marshall , f. e. 1984 , apj , 281 , 354 xue , y. q. , wu , x. b. , & cui , w. 2008 , mnras , 384 , 440 yuan , f. , zdziarski , a. a. , xue , y. , & wu x .- b .
2007 , apj , 659 , 541 zdziarski , a. a. , poutanen , j. , mikolajewska j. , et al .
1998 , mnras , 301 , 435 | the black hole x - ray transient xte j1550 - 564 has undergone a strong outburst in 1998 and two relativistic x - ray jets have been detected years later with the @xmath0 x - ray observatory ; the eastern jet was found previously to have decelerated after its first detection .
here we report a full analysis of the evolution of the western jet ; significant deceleration is also detected in the western side .
our analysis indicates that there is a cavity outside the central source and the jets first traveled with constant velocity and then were slowed down by the interactions between the jets and the interstellar medium ( ism ) .
the best fitted radius of the cavity is @xmath10.31 pc on the eastern side and @xmath10.44 pc on the western side , and the densities also show asymmetry , of @xmath10.034 @xmath2 on the east to @xmath10.12 @xmath2 on the west .
the best fitted magnetic fields on both sides are @xmath10.5 mg .
similar analysis is also applied to another microquasar system , h 1743 - 322 , and a large scale low density region is also found .
based on these results and the comparison with other microquasar systems , we suggest a generic scenario for microquasar jets , classifying the observed jets into three main categories , with different jet morphologies ( and sizes ) corresponding to different scales of vacuous environments surrounding them .
we also suggest that either continuous jets or accretion disk winds , or both may be responsible for creating these cavities .
therefore x - ray jets from microquasars provide us with a promising method of probing the environment of accreting black holes . |
topological insulators ( tis ) are a new class of materials characterized by a nontrivial topology of the hilbert space spanned by the wave functions of the occupied electronic states @xcite .
they are expected to be useful for various applications including high - frequency electronics , transparent electrodes , spintronics , and quantum computations @xcite .
already a number of bulk materials have been found to be three - dimensional ( 3d ) tis @xcite , and they are all narrow - gap semiconductors composed of heavy elements which cause band inversions @xcite .
when the band inversion occurs at an odd number of the high - symmetry points in the brillouin zone , the topological principle @xcite dictates that charge - conducting gapless states show up on the surface .
hence , 3d tis are peculiar in that they consist of gapped bulk states accompanied by gapless surface states of topological origin .
syntheses of new 3d ti materials are of great current interest for expanding our knowledge about the relationship between chemistry and quantum - mechanical functionalities @xcite . in this context
, there is an intriguing class of materials whose crystal structures realize a naturally - formed heterostructure consisting of alternating layers of topological insulators and ordinary insulators @xcite ; due to their constructions , such materials can be considered to lie at the boundary between topologically trivial and nontrivial materials .
specifically , in the homologous series of pb - bi - se ternary compounds expressed in the formula ( pbse)@xmath0(bi@xmath3se@xmath4)@xmath5 @xcite , which we call psbs here , layers of pbse ( which is an ordinary insulator ) alternate with layers of bi@xmath3se@xmath4 ( which is a prototypical topological insulator @xcite ) . in these series of compounds , it was inferred from the band structure data obtained from angle - resolved photoemission spectroscopy ( arpes ) that each internal interface between pbse and bi@xmath3se@xmath4 layers harbor topological interface states , leading to the situation that the whole bulk is filled with quasi - two - dimensional states of topological origin @xcite .
interestingly , since the topological interface states appear on both sides of each bi@xmath3se@xmath4 layer sandwiched by pbse layers , the two topological states at the top and bottom interfaces can hybridize _ within _ each bi@xmath3se@xmath4 layer , leading to the opening of a gap in the topological states @xcite .
the size of this hybridization gap becomes smaller as the bi@xmath3se@xmath4 layer becomes thicker and the hybridization becomes weaker .
based on the electronic structure data obtained from ultrathin films of bi@xmath3se@xmath4 @xcite , it is expected that the hybridization is gone when the bi@xmath3se@xmath4 layer becomes thicker than 5 nm .
note that the crystal structure of bi@xmath3se@xmath4 consists of covalently - bonded se - bi - se - bi - se quintuple layers ( qls ) that are interconnected by weak van der waals force @xcite .
the thickness of each ql is 0.9545 nm .
thus , the hybridization of topological states in psbs would disappear when each bi@xmath3se@xmath4 layer contains more than 5 qls . in the ( pbse)@xmath0(bi@xmath3se@xmath4)@xmath5 homologous series ,
the parameter @xmath6 gives the number of qls included in each bi@xmath3se@xmath4 layer , and the topological states becomes more robust for larger @xmath6 .
so far , syntheses of compounds with @xmath6 = 1 , 2 , and 3 have been reported @xcite , and it is naively expected that , based on the phase diagram of the pbse - bi@xmath3se@xmath4 pseudo - binary system reported by shelimova _
@xcite , compounds with larger @xmath6 would not be naturally synthesized because they are not thermodynamically stable . despite this naive expectation ,
we have successfully synthesized @xmath6 = 4 compound based on a melt - growth strategy in which the composition of the melt changes continuously as the growth proceeds . the expansion of the available compounds in this interesting homologous series is useful , because one can tune the hybridization of the topological states by changing @xmath6 , which is a unique tunability among topological materials . in this paper , we report the synthesis and analyses of the pb@xmath0bi@xmath1se@xmath2 [ = ( pbse)@xmath0(bi@xmath3se@xmath4)@xmath7 , called psbs @xmath6 = 4 ] compound , as well as its anisotropic transport properties which support the quasi - two - dimensional nature of its electronic states .
the x - ray diffraction ( xrd ) analysis using 2@xmath8@xmath8 scan is performed with rigaku ultima - iv x - ray apparatus equipped with a 1d - detector .
for single - crystal x - ray analysis , rigaku mercury ccd system with graphite - monochromated mok@xmath9 radiation is used .
the in - plane transport properties of psbs single crystals are measured by standard ac six - probe method , which allows measurements of the in - plane resistivity @xmath10 and the hall resistivity @xmath11 at the same time . by sweeping the magnetic field up to @xmath122 t
, the linear slope of @xmath11 vs. @xmath13 is obtained , from which the hall coefficient @xmath14 is calculated .
the out - of - plane resistivity @xmath15 is measured by using a four - probe method , in which the current and voltage contacts are made near the edges of the top and bottom surfaces of a rectangular sample . in the present work , @xmath10 and @xmath15
are measured on the same samples to avoid unnecessary complications , and the results from several samples are compared to assess the reproducibility of the anisotropy measurements .
single crystalline samples of psbs with a series of @xmath6 values are grown by a combination of modified bridgman and self - flux methods .
the growth process is complicated , because these compounds do not exhibit congruent melting @xcite . to grow high - quality single crystals , the following steps are taken : high - purity raw material shots of pb ( 99.998% ) , bi ( 99.9999% ) , and se ( 99.999% ) are sealed in a quartz tube that was heat - treated beforehand . for removing oxidization layers formed in air on the raw shots of pb and bi ,
surface cleaning procedures are performed ; pb shots are annealed in hydrogen atmosphere , and bi shots are washed with diluted hno@xmath4 .
based on the phase diagram reported by shelimova _
@xcite , we choose the starting composition of pbse : bi@xmath3se@xmath4 = 45:55 in molar ratio .
the raw materials are reacted and homogenized in a melt held at 1173 k for 6 h , and the crystal growth was fostered by slowly sweeping the temperature from 1023 k to 923 k at a rate of 2 k / hour in a temperature gradient of roughly 1 k / cm . at the beginning of the growth , @xmath6 = 1 crystals are formed at the low - temperature end of the melt , which causes the composition of the melt to become pbse - poor . as the temperature is lowered and the crystals growth proceeds , the melt becomes more and more pbse - poor and the composition of the grown crystals changes from @xmath6 = 1 to higher @xmath6 values .
after cutting the grown boule [ fig .
1a ] with a wire saw , cleavable single crystals can be separated .
the phase of each piece is determined by the 2@xmath8@xmath8 xrd analysis on a cleaved surface .
even when the appearance of the surface is clean and shiny , the xrd profile can sometimes contain peaks from multiple phases ; we discard such multi - phase crystals and only use those crystals that show single - phase peaks on both top and bottom surfaces for further characterizations .
the largest size of the single - phase crystal we obtained was roughly @xmath160.2 mm@xmath17 .
the xrd pattern confirmed that the cleavage plane is along the @xmath18-plane .
note that , because the crystal structures of the psbs homologous series are monoclinic @xcite , the @xmath19-axis is _ not _ perpendicular to the @xmath18-plane , and hence the direction normal to the @xmath18 plane is called @xmath20-axis .
this means that the 2@xmath8@xmath8 xrd profile on a cleaved surface of a psbs single crystal reflects the periodicity along the @xmath20-axis , which corresponds to @xmath21 with @xmath19 and @xmath22 the @xmath19-axis lattice constant and the angle between the @xmath19-axis and the @xmath18-plane , respectively .
the periodicities ( i.e. the @xmath23 values ) for @xmath6 = 1 , 2 , and 3 are obtained in the present experiments to be 1.5795(19 ) , 2.5287(11 ) , and 3.4753(15 ) nm , respectively [ fig .
1b - d and table 1 ] @xcite ; these results are consistent with those previously reported @xcite . however , in addition to these three phases , we found a new phase of unknown xrd profile , as shown in fig .
1e . the crystals of this new phase presents a shiny cleavage plane , and the appearance is indistinguishable from other psbs phases nor the bi@xmath3se@xmath4 compound .
nevertheless , from the xrd profile , the periodicity perpendicular to the cleavage plane is calculated to be 4.4274(15 ) nm , which is very long and is obviously different from any known phase in this pb - bi - se ternary system .
the detailed xrd data for all four phases are presented in the supplementary materials .
se@xmath4 = 45:55 .
( b - e ) 2@xmath8@xmath8 xrd profiles taken on cleaved surfaces of single - crystal samples of the psbs homologous series with various @xmath6 values : ( b ) @xmath6 = 1 ; ( c ) @xmath6 = 2 ; ( d ) @xmath6 = 3 ; ( e ) new phase , @xmath6 = 4 .
( f ) plot of @xmath23 vs. @xmath6 , where the data point from the new phase is assigned to @xmath6 = 4 .
( g ) a laue picture taken on the cleaved surface of a single crystal of the @xmath6 = 4 phase .
( h , i ) schematic pictures of the crystal structures drawn @xcite for the @xmath6 = 3 and 4 phases of the psbs homologous series .
because of the monoclinic structure , the direction normal to the cleavage plane ( @xmath18-plane ) is denoted as @xmath24-axis.,width=434 ] .@xmath25 is the increment in @xmath26 (= @xmath23 ) when @xmath6 increases by one . [
cols="^,^,^ " , ] the above icp - aes results indicate that the actual chemical formula for psbs had better be expressed as ( pb@xmath27bi@xmath28se)@xmath0(bi@xmath3se@xmath4)@xmath5 ( note that the solubility of pb into bi@xmath3se@xmath4 is reported to be negligible @xcite ) .
we show in table 3 the @xmath29 and @xmath30 values in this formula for @xmath6 = 1 4 calculated from the icp - aes results .
the bi@xmath31 antisites are donors to provide @xmath32-type carriers , and thus one can calculate the expected @xmath32-type carrier density from the charge imbalance @xmath33 per mol ; this formula is based on the valence states pb@xmath34 , bi@xmath35 , and se@xmath36 .
positive @xmath37 corresponds to @xmath32-type .
the expected carrier density @xmath38 is obtained by dividing @xmath37 by the unit - cell volume , and for @xmath6 = 4 , one obtains @xmath39 .
figure 2a shows the temperature dependences of @xmath10 for five samples of psbs @xmath6 = 4 single crystals obtained from the same boule .
the temperature dependence is metallic , and the absolute value of @xmath10 varies by about 40% between samples .
the temperature dependences of @xmath40 for the same set of samples are shown in fig .
2b ; the variation of @xmath40 between samples is again about 40% at room temperature .
it is useful to note that the data in fig . 2b
tell us that the sample - to - sample variation in @xmath10 is caused essentially by a change in the carrier density , which is presumably related to different levels of bi antisite defects in different pieces of crystals .
in fact , as shown in fig . 2c , the hall mobility @xmath41 calculated from @xmath10 and @xmath40 for those samples converges at room temperature , indicating that the level of electron scattering is unchanged among samples at room temperature . the carrier density @xmath42 calculated from @xmath40 varies from @xmath43 to @xmath44 , which is smaller than that expected from the density of bi antisites by roughly a factor of two .
a major part of this discrepancy is probably due to the hall factor , which accounts for the difference between the actual carrier density and that calculated from @xmath45 , because the latter can vary due to the multiband effects and the relaxation - time anisotropy .
bi@xmath1se@xmath2 ( psbs @xmath6 = 4 ) , measured on five different pieces of single crystals ( a e ) .
panels ( a)-(c ) present the temperature dependences of ( a ) @xmath10 , ( b ) @xmath40 , and ( c ) @xmath41 .
the inset of panel ( a ) shows a photograph of sample c with four voltage contacts.,width=472 ] the out - of - plane resistivity @xmath15 is useful for evaluating the anisotropy and the dimensionality of the charge transport . in fig .
3a , the temperature dependences of @xmath15 for the same set of samples as those in fig .
2 are shown .
the magnitude of @xmath15 is of the order of 10 m@xmath46 cm , which is large for a degenerate semiconductor with @xmath47 .
the temperature dependence is either weakly metallic or almost temperature independent , from which it is difficult to judge whether the transport along the @xmath24-axis is coherent . nevertheless , as shown in fig .
3b , the anisotropy ratio @xmath48 grows with lowering temperature and reaches 75170 at low temperature , which points to the quasi - two - dimensional nature of the transport @xcite . for comparison
, we show in fig .
3c - e the data of @xmath10 , @xmath15 , and @xmath48 for two samples of psbs @xmath6 = 2 single crystals .
one can see that @xmath10 is comparable to that of the @xmath6 = 4 phase , but @xmath15 ( and hence @xmath48 ) is much smaller in the @xmath6 = 2 phase .
thus , the present results suggest that the transport anisotropy increases with increasing @xmath6 in the psbs homologous series .
( we note that it has so far been difficult to obtain sufficiently large single crystals of the @xmath6 = 3 phase , and hence we do not have reliable data for the resistivity anisotropy for @xmath6 = 3 . ) and ( b ) @xmath48 measured on the same @xmath6 = 4 samples ( a e ) as those shown in fig .
2 . ( c - e ) temperature dependences of ( c ) @xmath10 , ( d ) @xmath15 , and ( e ) @xmath48 measured on two @xmath6 = 2 samples ( a and b).,width=472 ]
to understand the implications of the difference in the resistivity anisotropy between @xmath6 = 2 and 4 , it is useful to remember that in bulk bi@xmath3se@xmath4 , which can be viewed as the @xmath49 limit of the psbs homologous series , the anisotropy is as small as @xmath503 @xcite , and this is because bi@xmath3se@xmath4 has a 3d ellipsoidal fermi surface . therefore , the _ increase _ in the anisotropy for larger @xmath6 can not be an indication that the system is approaching the @xmath49 limit .
in other words , the present result on the resistivity anisotropy strongly supports the conjecture that pbse layer in this homologous series works as a block layer for the charge transport .
nevertheless , the volume density of the pbse layer becomes _ smaller _ for larger @xmath6 , and hence the larger @xmath15 value at @xmath6 = 4 can not be due simply to the presence of pbse layers as series resistors .
one possibility to understand the trend in the resistivity anisotropy is to consider that the level of hybridization between the top and bottom interface states of topological origin within each bi@xmath3se@xmath4 layer plays an important role in determining the out - of - plane transport .
in such a case , a thicker bi@xmath3se@xmath4 layer ( i.e. a larger @xmath6 ) means weaker hybridization , which would lead to smaller transfer integral between top and bottom interface states within the bi@xmath3se@xmath4 layer , resulting in a larger @xmath24-axis resistivity .
although speculative , this possibility suggests the importance of the topological interface state in the out - of - plane transport in the psbs homologous series .
we now discuss the chemical reason why it was possible to synthesize the @xmath6 = 4 phase in the present work . according to the phase diagram of the pbse - bi@xmath3se@xmath4 pseudo - binary system reported by shelimova _
@xcite , there are only three stable phases , pb@xmath0bi@xmath51se@xmath52 ( @xmath6 = 1 ) , pb@xmath0bi@xmath7se@xmath53 ( @xmath6 = 2 ) , and pb@xmath0bi@xmath54se@xmath55 ( @xmath6 = 3 ) , which respectively form below 993 , 973 , and 948 k. the phase diagram also suggests that if the bi@xmath3se@xmath4:pbse ratio is higher than that corresponds to @xmath6 = 3 , a mixed phase solidifies below 930 k. however , this phase diagram was constructed from only a finite number of compositional data points , and it is possible that there is a narrow compositional window in which the @xmath6 = 4 phase preferentially grows below some temperature that lies between 930 and 948 k. in our growth method for this homologous series , the composition of the melt changes continuously toward a higher bi@xmath3se@xmath4:pbse ratio as the temperature is lowered and the crystal growth proceeds , which must have brought the system to hit the right composition / temperature window for the @xmath6 = 4 growth . hence , our method is suitable for the growth of rare compounds in a homologous series , although it is difficult to obtain large single crystals for each phase . finally , it would be useful to elaborate on the difference in the crystal structures of the homologous series obtained for pb - bi - se and pb - bi - te ternary systems .
the pb - bi - te ternary system also gives rise to a homologous series , but its construction is fundamentally different from that obtained for the pb - bi - se ternary system .
namely , in the case of tellurides , although the composition formula is superficially similar to the selenides and is written as ( pbte)@xmath56(bi@xmath3te@xmath4)@xmath57 ( @xmath6 and @xmath32 are integers ) , in this homologous series `` pbte '' is incorporated into the original rhombohedral structure of bi@xmath3te@xmath4 , forming a covalently - bonded te - bi - te - pb - te - bi - te septuple - layer unit @xcite .
hence , in the telluride homologous series , the rhombohedral symmetry of the crystal structure which stems from a cubic structure distorted along ( 111 ) direction is essentially unchanged from that of bi@xmath3te@xmath4 .
importantly , the @xmath58 phase of this telluride homologous series , pbbi@xmath3te@xmath59 , which simply consists of a stack of the te - bi - te - pb - te - bi - te septuple - layers , is known to be a topological insulator @xcite ; hence , the telluride homologous series is a heterostructure of two topological insulators ( i.e. pbbi@xmath3te@xmath59 and bi@xmath3te@xmath4 ) and the topological states show up only on the surrounding surface .
in contrast , in the selenide homologous series @xcite , pbse is not incorporated into the rhombohedral structure of bi@xmath3se@xmath4 and the rock - salt structure of bulk pbse partly remains , forming the pbse layer of square symmetry ; this layer alternates with the bi@xmath3se@xmath4 layer of hexagonal symmetry , and the resulting structure is no longer rhombohedral . as is already emphasized ,
this selenide homologous series is a heterostructure of a topological insulator and an ordinary insulator , and hence its electronic nature is fundamentally different from that of the telluride homologous series .
we have synthesized a new compound in the ( pbse)@xmath0(bi@xmath3se@xmath4)@xmath5 homologous series with @xmath6 = 4 , which became possible by using a melt - growth method in which the composition of the melt changes continuously to allow the whole series of compounds to grow in a sequential manner . this new compound , pb@xmath0bi@xmath1se@xmath2 , is a member of the unique family of topological materials in which alternating layers of the topological insulator bi@xmath3se@xmath4 and the ordinary insulator pbse form natural heterostructures , leading to the appearance of topological interface states throughout the bulk .
the trend in the @xmath24-axis resistivity and the resistivity anisotropy with increasing @xmath6 values suggests that the hybridization of the topological interface states plays an important role in the out - of - plane transport in this compound .
* acknowledgment * + + we acknowledge the comprehensive analysis center , isir , osaka university , for technical supports in the ccd x - ray diffraction analysis .
we also thank t. toba for technical assistances . this work was supported by jsps ( kakenhi 25220708 , 24540320 , and 25400328 ) , mext ( innovative area
topological quantum phenomena " kakenhi 22103004 ) , and afosr ( aoard 124038 ) .
eremeev , g. landolt , t.v .
menshchikova , b. slomski , y.m .
koroteev , z.s .
aliev , m.b .
babanly , j. henk , a. ernst , l. patthey , a. eich , a.a .
khajetoorians , j. hagemeister , o. pietzsch , j. wiebe , r. wiesendanger , p.m. echenique , s.s .
tsirkin , i.r .
amiraslanov , j. hugo dil , e.v .
chulkov , nat .
commun . 3 ( 2012 ) 635 . | we have synthesized pb@xmath0bi@xmath1se@xmath2 , which is a new member of the ( pbse)@xmath0(bi@xmath3se@xmath4)@xmath5 homologous series with @xmath6 = 4 .
this series of compounds consist of alternating layers of the topological insulator bi@xmath3se@xmath4 and the ordinary insulator pbse .
such a naturally - formed heterostructure has recently been elucidated to give rise to peculiar quasi - two - dimensional topological states throughout the bulk , and the discovery of pb@xmath0bi@xmath1se@xmath2 expands the tunability of the topological states in this interesting homologous series .
the trend in the resistivity anisotropy in this homologous series suggests an important role of hybridization of the topological states in the out - of - plane transport .
topological insulator , bismuth selenide , crystal growth , quasi - two - dimensional transport |
circular arc graphs are graphs whose vertices can be represented as arcs on a circle such that any two vertices are adjacent if and only if their corresponding arcs intersect . circular arc graphs form useful mathematical objects with many practical applications such as in _ genetic research _
@xcite , _ traffic control _
@xcite , _ compiler design _ @xcite and _ statistics _ @xcite .
circular arcs were first discussed in @xcite as a natural generalization of interval graphs ( defined analogously , but using intervals on a line instead of arcs on a circle ) and they have since been studied extensively @xcite .
see golumbic @xcite for a brief introduction on circular arc graphs .
circular arc coloring problem _ consists of finding a minimum coloring of a set of arcs of a circle such that no two intersecting arcs have the same color .
tucker @xcite gave a simple 2-approximation algorithm for coloring circular arc graphs and conjectured that @xmath2 colors are sufficient to color any family @xmath3 of arcs , where @xmath4 represents the size of the maximum set of pairwise intersecting arcs .
karapetyan @xcite proved tucker s conjecture .
further analysis of tucker s algorithm by pabon @xcite gave a tighter upper bound for coloring the arcs based on the _ circular cover _ of the family of arcs .
circular cover of a family @xmath3 of arcs is the minimum number of arcs required to cover the entire circle . a graph @xmath0 is a _ proper _ circular - arc graph if there is a circular - arc representation of @xmath0 such that no arc is properly contained in any other .
the set of arcs in a proper circular arc representation is called a _
family of proper arcs_. because of this additional restriction , some of the difficult problems for circular arc graphs in general becomes easier for proper circular arc graphs .
for example , even though coloring of circular arc graphs in general was shown to be np hard by garey et al .
@xcite , proper circular arc graphs can be colored in polynomial time as shown by orlin , bonuccelli and bovet @xcite .
proper circular arc graphs are a strict subclass of circular arc graphs .
see tucker @xcite for characterization of proper circular arc graphs . in this paper
, we prove hadwiger s conjecture for proper circular arc graphs . a * vertex coloring * of a graph
@xmath5 is a map @xmath6 such that @xmath7 whenever @xmath8 and @xmath9 are adjacent .
the smallest integer @xmath1 such that @xmath0 has a vertex coloring @xmath10 is called the * chromatic number * of @xmath0 .
chromatic number of a graph @xmath0 is denoted by @xmath11 .
* contraction of an edge @xmath12 * is the replacement of vertices @xmath13 and @xmath14 with a new vertex @xmath15 whose incident edges are exactly those edges other than @xmath16 that were incident to at least one of @xmath13 or @xmath14
. a * minor * of a graph @xmath0 is obtained by a sequence of applications of the following three elementary operations on graph @xmath0 1 .
deletion of a vertex 2 .
deletion of an edge 3 .
contraction of an edge * complete graph * of order @xmath17 is a graph on @xmath17 vertices where all the vertices are pairwise adjacent and is denoted by @xmath18 .
see diestel @xcite for more information on minors . in 1943 , hadwiger made the famous conjecture linking the chromatic number of a graph with its clique minor : if a graph @xmath0 has chromatic number @xmath19 , then @xmath20 is a minor of @xmath0 .
the conjecture is easy to see for @xmath21 .
the case @xmath22 was proved by dirac @xcite . the hadwiger s conjecture for any fixed @xmath23 is equivalent to the assertion that every graph without a @xmath20 minor has a @xmath24 coloring .
therefore , the case @xmath25 implies four color theorem because any planar graph has no @xmath26 minor . on the other hand ,
hadwiger s conjecture for the case @xmath25 follows from the four color theorem and a structure theorem of wagner @xcite . for @xmath27 , dirac
@xcite proved that if the chromatic number of a graph @xmath0 is 6 , then @xmath0 can be contracted into @xmath28 , a complete graph on 6 nodes with one edge missing .
robertson , seymour and thomas @xcite have obtained a proof for @xmath27 .
they showed that every minimum contraction - critical graph ( a graph is said to be contraction - critical if for every proper minor @xmath29 of @xmath0 , @xmath30 ) , different from the complete graph @xmath31 , is an _ apex _ graph which has a special vertex @xmath8 the removal of which results in a planar graph . as a result , hadwiger s conjecture for @xmath27 reduces to the four color theorem . the case for @xmath32
onwards is still open and the best known results for @xmath33 are due to jakobsen @xcite .
he proved that a @xmath1-chromatic graph can be contracted to @xmath34 , @xmath35 and @xmath36 respectively for @xmath37 and @xmath38 .
recently , kawarabayashi and toft @xcite proved that any @xmath39-chromatic graph has @xmath36 or @xmath40 as a minor . since hadwiger s conjecture in the general case seems to be too difficult , it is interesting to prove it for special classes of graphs .
reed and seymour @xcite studied the hadwiger s conjecture in the case of _ line graphs_. a line graph @xmath41 of @xmath0 is the graph on the edge set @xmath42 in which @xmath43 are adjacent as vertices in @xmath41 , if and only if they are adjacent as edges in @xmath0 .
they showed that for every loop less graph @xmath0(possibly with parallel edges ) , its line graph @xmath41 satisfies the hadwiger s conjecture . for a detailed history of hadwiger s conjecture as well as an account of recent developments in that area
the reader is referred to the survey by toft @xcite .
a graph is called perfect if for every induced subgraph @xmath29 of @xmath0 , @xmath44 where @xmath45 is the order ( i.e. , the number of vertices ) of the largest complete subgraph of @xmath29 .
@xmath45 is also known as the _ clique number _ of @xmath29 .
interval graphs are _ perfect _ graphs .
see golumbic @xcite for more information on interval graphs .
hence hadwiger s conjecture trivially holds true for interval graphs . for the class of circular arc graphs , which is a generalization of interval graphs , hadwiger
s conjecture is still open .
therefore , it is natural to study hadwiger s conjecture for circular arc graphs . in this paper
, we prove hadwiger s conjecture for proper circular arc graphs .
hajos conjecture for proper circular arc graphs : it may be a matter of curiosity to note that the class of proper circular arc graphs does nt satisfy the closely related hajos conjecture .
this is because the counterexample to hajos conjecture given by catlin @xcite is a proper circular arc graph .
this gives us one more reason to verify hadwiger s conjecture for proper circular arc graphs .
for a circular arc graph @xmath0 , without loss of generality we can assume that the family of arcs @xmath3 representing @xmath0 has all its arc endpoints distinct .
also , without loss of generality we can assume that no arc in the circular arc representation of a circular arc graph spans the whole circle similarly , we can assume that no arc is formed of a single point .
these assumptions also hold true for proper circular arc graphs . given a family @xmath3 of arcs ,
the * overlap set * of point @xmath46 on the circle is the set of all arcs that contain the given point @xmath46 and is denoted by @xmath47 .
the overlap set with the largest number of arcs in it is called a * maximum overlap set * and its cardinality is denoted by @xmath48 .
an overlap set with the smallest number of arcs in it is called a * minimum overlap set * and its cardinality is denoted by @xmath49 .
it is easy to see that the arcs in an overlap set induce a clique in the corresponding circular arc graph . from now on , when there is no ambiguity we use the term `` arcs '' and `` vertices '' interchangeably .
for example , we use the same labels to refer to the vertices in a circular arc graph as well as the corresponding arcs in its circular arc representation that is being considered .
each arc in @xmath3 has two endpoints .
the * left endpoint * @xmath50 @xmath51respectively * * right endpoint @xmath52**@xmath53 of arc @xmath54 is the first endpoint of @xmath54 encountered in an _ anticlockwise _ @xmath51respectively _ clockwise_@xmath53 traversal from any interior point of @xmath54 . @xmath51 recall that the circle itself is not considered as an arc .
a single point is also not considered as an arc .
thus the definition makes sense , and each arc has a left end point as well as a right end point.@xmath53 an arc @xmath8 is said to be * clockwise adjacent * to an arc @xmath54 if @xmath8 belongs to the overlap set of @xmath52 , i.e. @xmath55 .
an arc @xmath8 is said to be * anticlockwise adjacent * to an arc @xmath54 if @xmath8 belongs to the overlap set of @xmath50 , i.e. @xmath56 . for a family of arcs , an arc @xmath54 can be adjacent to an arc @xmath8 without being _ clockwise adjacent _ or _ anticlockwise adjacent _ to @xmath8 by being properly contained in @xmath8 . for a family of _ proper _
arcs , if an arc @xmath54 intersects arc @xmath8 then it must be either _ clockwise _ or _
anticlockwise adjacent _ to @xmath8 .
+ in a family of proper arcs , if an arc @xmath54 is clockwise adjacent to an arc @xmath8 , then arc @xmath8 is anticlockwise adjacent to arc @xmath54 .
[ lem : lem24 ] an arc in a family of proper arcs has at most @xmath57 arcs and at least @xmath49 arcs clockwise adjacent to it .
similarly , an arc has at most @xmath57 and at least @xmath49 arcs anticlockwise adjacent to it .
there can be at most @xmath57 arcs that are clockwise adjacent to an arc @xmath54 since cardinality of overlap set at the right endpoint of @xmath54 , @xmath58 and @xmath54 is also part of this overlap set .
similarly , we can prove for the anticlockwise direction using @xmath50 .
for an arc @xmath54 , consider the point @xmath46 just after right endpoint @xmath52 of arc @xmath54 in clockwise direction . as the cardinality of the minimum overlap set is @xmath49
, we have @xmath59 . clearly , @xmath60 and hence each arc in @xmath47
is clockwise adjacent to @xmath54 .
therefore , there are at least @xmath49 arcs that are clockwise adjacent to @xmath54 .
similarly we can argue for the anticlockwise direction by looking at a point @xmath46 just after @xmath50 in anticlockwise direction .
the minimum number of colors needed to color a family @xmath3 of arcs such that no two intersecting arcs have the same color is its _ chromatic number _ @xmath61 .
as the arcs correspond to vertices in the corresponding circular arc graph @xmath0 , we have @xmath62 . a straightforward upper bound on
the chromatic number of @xmath3 is [ lem : lem25 ] for a family @xmath3 of arcs , @xmath63 .
this can be easily seen from the fact that removing all the arcs in any minimum overlap set will result in a family of arcs that correspond to an interval graph which can then be colored using at most @xmath48 colors .
( recall that interval graph is a perfect graph and the cardinality of maximum overlap set corresponds to the clique number in an interval graph . )
[ def : def26 ] for a family @xmath3 of arcs , * circular cover @xmath64 * is the smallest cardinality of any subset of arcs of @xmath3 needed to cover the circle .
note that circular cover is defined for a family of arcs only if a finite number of arcs in the family can cover the circle .
if no number of arcs can cover the entire circle then the corresponding graph is an interval graph and we know the chromatic number equals clique number for interval graphs and thus hadwiger s conjecture trivially holds . hence in the following sections we will assume that the circular cover is defined and finite . [ thm : thm26 ] if the circular cover of a family @xmath3 of arcs @xmath65 , then @xmath66
suppose , for contradiction , assume that the class of proper circular arc graphs does not satisfy hadwiger s conjecture .
then , consider a proper circular arc graph on the minimum possible number of vertices that is a counterexample to hadwiger s conjecture .
we will use the notations given in table [ table : table1 ] throughout the remaining part of this section . if @xmath67 , then hadwiger s conjecture trivially holds true for @xmath68 as all the arcs of the maximum overlap set form a clique .
therefore , we have @xmath69 .
also , if @xmath70 , then @xmath68 will be a complete graph and hence not a counterexample to hadwiger s conjecture .
hence , @xmath71 . we fix a proper circular arc representation for graph @xmath68 which will also be referred using the same notation @xmath68 .
any induced subgraph of a proper circular arc graph is also a proper circular arc graph .
this is because removing a vertex in the graph is equivalent to removing an arc in the corresponding circular arc representation . as @xmath68 is a proper circular arc graph on the smallest possible number of vertices that is a counterexample to hadwiger s conjecture , it is the only counterexample to hadwiger s conjecture among its induced subgraphs .
therefore by theorem [ thm : thm31 ] , a graph @xmath74 is said to be color critical if @xmath75 for every @xmath76 .
every graph has an induced subgraph that is color critical . for a color critical graph @xmath0
, we have the minimum degree of the graph , @xmath77(see west @xcite for the proof ) .
it is easy to see that the minimum counterexample to hadwiger s conjecture @xmath68 must be _ color critical_. therefore we have , we will first show that @xmath80 .
if the circular cover ( see definition [ def : def26 ] ) @xmath81 , then by theorem [ thm : thm26 ] , we have @xmath80 . in a family of proper arcs ,
the set of arcs that are clockwise adjacent to a given arc is exactly the set of arcs that have their left endpoints inside that arc .
therefore , by lemma [ lem : lem24 ] , every arc in @xmath68 can have left endpoints of at most @xmath82 arcs in it .
if @xmath83 , then there exists three arcs say @xmath13 , @xmath14 and @xmath15 such that the union of these arcs cover the entire circle . since each arc including @xmath13,@xmath14 and @xmath15 should have its left endpoint in the interior of at least one of these three arcs , it follows that @xmath84 . as @xmath68 is a minimum counterexample by lemma [ lem : lem32 ]
, we have @xmath85 .
hence even if @xmath83 , we have @xmath80 . recalling that @xmath86
, we get @xmath79 .
[ lem : fr ] for a family @xmath3 of proper arcs , if we are traversing in the clockwise direction from a point @xmath46 on the circle , the right endpoints of all the arcs in the overlap set of @xmath46 would be encountered before the right endpoints of any other arc in @xmath3 .
when we traverse along the circle in clockwise direction from a point @xmath46 , if we encounter the right endpoint of an arc @xmath54 before encountering its left endpoint , then @xmath54 is in the overlap set of @xmath46 , i.e. @xmath90 .
instead , as we traverse along the circle in clockwise direction from @xmath46 , if we encounter the left endpoint of arc @xmath8 before its right endpoint , then @xmath91 . if the right endpoint of @xmath8 ( @xmath92 ) also occurs before the right endpoint of at least one arc @xmath54 ( @xmath90 ) , then moving in clockwise direction from @xmath50 we encounter the point @xmath46 and then both the left and right endpoints of @xmath8 before we encounter @xmath52 which implies that @xmath8 is entirely contained in arc @xmath54 which is in contradiction to proper circular arc property .
hence the lemma follows .
[ cor : fr1 ] ( a ) if an arc @xmath54 is clockwise adjacent to an arc @xmath8 , then all the arcs whose right endpoints are encountered after the right endpoint of @xmath8 and before the right endpoint of @xmath54 in clockwise direction are also clockwise adjacent to arc @xmath8 .
\(b ) similarly , if an arc @xmath54 is anticlockwise adjacent to an arc @xmath8 , then all the arcs whose right endpoints are encountered after the right endpoint of @xmath54 and before the right endpoint of @xmath8 in clockwise direction are also anticlockwise adjacent to arc @xmath8 . in the circular arc representation of @xmath68 ,
identify a point @xmath46 such that @xmath93 , the maximum overlap set .
traverse the circle in the clockwise direction starting from the point @xmath46 labeling the arcs in the order in which their right endpoints are encountered .
note that the first @xmath23 arcs to be labeled are from the maximum overlap set @xmath94 by lemma [ lem : fr ] .
let the first @xmath23 arcs(i.e .
the arcs in @xmath94 ) be labeled as @xmath95 and the remaining @xmath1 arcs be labeled @xmath96 .
see fig [ fig : fig1 ] for the labeling scheme . by corollary [ cor : cor34 ]
, we have @xmath89 and therefore @xmath97 . 1 .
if an arc @xmath98 is clockwise adjacent to an arc @xmath99 , then there are at least @xmath100 arcs , namely @xmath101 clockwise adjacent to arc @xmath99 .
2 . if an arc @xmath98 is clockwise adjacent to an arc @xmath102 , then @xmath103 and there are at least @xmath104 arcs , namely @xmath105 clockwise adjacent to arc @xmath102 .
if an arc @xmath98 is anticlockwise adjacent to an arc @xmath102 , then @xmath106 and there are at least @xmath107 arcs , namely @xmath108 anticlockwise adjacent to arc @xmath102 .
if an arc @xmath98 is anticlockwise adjacent to an arc @xmath99 , then there are at least @xmath109 arcs , namely @xmath110 anticlockwise adjacent to arc @xmath99 .
as @xmath68 is a family of proper circular arcs , any arc @xmath54 in @xmath68 has at least @xmath49 arcs clockwise adjacent to it by lemma [ lem : lem24 ] . in other words , @xmath114 .
therefore by lemma [ lem : fr ] , the @xmath49 arcs whose right endpoints are encountered after the right endpoint of @xmath54 in clockwise direction are in the overlap set of @xmath52 and hence clockwise adjacent to @xmath54 . by lemma [ lem : lem25 ] and the fact that @xmath86 we have @xmath115 .
therefore , the @xmath13 arcs whose right endpoints are encountered after the right endpoint of an arc @xmath54 in clockwise direction are clockwise adjacent to @xmath54 . by the labeling scheme described above , for any arc @xmath102 where @xmath113 , these @xmath13 arcs whose right endpoints are encountered after the right endpoint of @xmath102 would be labeled @xmath116
. therefore @xmath116 are clockwise adjacent to @xmath102 .
if @xmath68 contains a good path set then we will show a @xmath121 ( recall that @xmath86 ) clique minor leading to a contradiction .
for this , we will first show that every arc in @xmath94 is adjacent to either @xmath102 or @xmath119 for @xmath122 .
arcs in @xmath94 are labeled @xmath95 .
suppose we have an arc @xmath123 and an integer @xmath124 , where @xmath122 such that @xmath123 is adjacent to neither @xmath102 nor @xmath119 .
what is the degree of @xmath123 ?
clearly @xmath123 intersects with all the remaining @xmath82 arcs in @xmath94 , at most @xmath125 arcs from @xmath126 are clockwise adjacent to @xmath123 , namely @xmath127 and at most @xmath128 arcs from @xmath126 are anticlockwise adjacent to @xmath99 , namely @xmath129 .
therefore , @xmath130 which contradicts lemma [ lem : lem33 ] by which @xmath131 .
hence , every arc in @xmath94 must be adjacent to either @xmath102 or @xmath119 for @xmath132
. now if we contract each vertex disjoint path @xmath133 to a single vertex , then such a vertex would be adjacent to all the arcs in @xmath94 .
also each of the contracted vertices would be adjacent to each other as @xmath134 are clockwise adjacent to @xmath135 by lemma [ lem : lem35 ] .
also , all the arcs in @xmath94 are pairwise adjacent to each other as they all belong to the maximum overlap set .
hence , we have an @xmath121 clique minor if @xmath68 has a good path set contradicting the assumption that @xmath68 is a counterexample to hadwiger s conjecture .
if @xmath1 is a multiple of @xmath13 then we show a good path set which contradicts lemma [ lem : lem37 ] .
we define @xmath13 vertex disjoint paths where each path @xmath136 for @xmath137 is of the form @xmath138 and @xmath139 . by lemma [ lem : lem35 ]
, @xmath140 must be clockwise adjacent to @xmath141 for @xmath142 and hence each of @xmath136 is a path .
it is easy to see that each vertex @xmath141 for @xmath143 belongs to a unique path , namely @xmath136 where @xmath144 .
moreover , vertex @xmath145 is in @xmath136 .
thus , we have a good path set contradicting lemma [ lem : lem37 ] . from lemma [ lem : lemk ]
, @xmath1 is not a multiple of @xmath13 .
let @xmath148 and @xmath149 .
suppose we have an @xmath102 which is clockwise adjacent to @xmath146 where @xmath147 .
we will demonstrate a good path set which contradicts lemma [ lem : lem37 ] .
a _ successor vertex _ of a vertex is the vertex obtained by applying the successor function on the given vertex .
see fig [ fig : fig2 ] for a demonstration of the successor function where the arrow points from a vertex to its successor vertex .
we will now demonstrate @xmath13 vertex disjoint paths of a good path set using the successor function defined above , contradicting lemma [ lem : lem37 ] .
the required result would follow immediately .
define @xmath13 sequences of vertices @xmath156 as follows : each sequence @xmath136 for @xmath157 begins at @xmath98 .
any other vertex in the sequence is determined by applying the successor function on the previous vertex in the sequence .
for example , @xmath158 .
the sequence ends when we encounter a vertex for which the successor function is not defined .
note that the successor function is defined for every vertex @xmath159 in the range @xmath160 .
moreover , by applying the successor function on a vertex @xmath159 ( where @xmath160 ) we always get another vertex @xmath161 whee @xmath162 .
therefore the sequences defined above are of finite length , and the last vertex of each of these sequences belong to the set @xmath163 , for which the successor function is not defined . in order to demonstrate that the sequences @xmath156 indeed form a good path set , we prove the following claims : + * claim 1 .
* @xmath164 is clockwise adjacent to @xmath141 for @xmath165 .
thus for @xmath166 , the sequence @xmath136 forms path . + * proof . * 1 .
this is easy to see in the range @xmath167 and in the range @xmath168 by lemma [ lem : lem35 ] .
2 . in the range @xmath169
, we have @xmath170 .
if @xmath146 is anticlockwise adjacent to @xmath102 then since @xmath171 , @xmath141 is also anticlockwise adjacent to @xmath102 by corollary [ cor : fr2].3 .
we have @xmath172 .
therefore , by corollary [ cor : fr2].2 , as @xmath102 is clockwise adjacent to @xmath141 , @xmath173 is also clockwise adjacent to @xmath141 .
thus in the range @xmath169 we have @xmath164 clockwise adjacent to @xmath141
. * claim 2 . *
the paths @xmath156 are vertex disjoint + * proof . *
every vertex @xmath141 in the range @xmath174 belongs to exactly one path .
we can easily determine the unique path to which the vertex belongs by studying the successor function and it is given below 1 .
in the range @xmath175 we have @xmath141 in path @xmath136 where @xmath144 .
this is because if @xmath141 is in @xmath136 for @xmath176 then @xmath177 is also in @xmath136 and @xmath98 is in @xmath136 , @xmath178 .
for example , vertices @xmath179 where @xmath180 , belong to path @xmath181 .
2 . in the range @xmath182 , the vertex @xmath141 is a successor of the vertex @xmath183 .
hence it belongs to the same path to which @xmath183 belongs . as @xmath184 , @xmath183 belongs to @xmath136 where @xmath185 .
3 . in the range @xmath186
, the vertex @xmath141 is a successor of the vertex @xmath187 and hence belongs to the same path to which @xmath187 belongs.as @xmath188 , @xmath187 belongs to path @xmath136 where @xmath189 .
+ the reader may observe that the @xmath13 vertices @xmath190 have their successor vertices rearranged in the consecutive @xmath13 vertices @xmath191 .
see fig [ fig : fig2 ] for the rearrangement of successor vertices in the above range .
4 . from ( 2 ) and ( 3 ) , it is clear that @xmath141 for @xmath192 belongs to @xmath136 where @xmath185 . in the range @xmath193 , @xmath141 is in the same path to which @xmath194 belongs and @xmath194 belongs to @xmath136 where @xmath195 . * claim 3 . *
every path @xmath196 , for @xmath197 ends at @xmath198 .
+ * proof . * for @xmath199 in the range @xmath200
, vertex @xmath198 belongs to path @xmath136 where @xmath201 .
then , clearly @xmath198 should be the last vertex of @xmath196 since the successor function is not defined for @xmath198 . from claims 1,2 and 3 ,
it is clear that @xmath156 form a good path set .
this contradicts lemma [ lem : lem37 ] and hence @xmath102 is not clockwise adjacent to @xmath146 in @xmath68 for @xmath124 in the range @xmath147 .
let @xmath203 .
by lemma [ lem : lemk ] , we have @xmath204 .
if @xmath205 , we will demonstrate a good path set which contradicts lemma [ lem : lem37 ] .
let us consider @xmath13 vertex disjoint paths @xmath206 where @xmath207 .
note that @xmath136 is a path because @xmath140 is clockwise adjacent to @xmath141 for @xmath208 by lemma [ lem : lem35 ] .
it is easy to see that , for @xmath209 the vertex @xmath141 is in @xmath136 if @xmath210 and in particular , @xmath211 is in @xmath136 as @xmath212 .
now , we will show that , for @xmath166 , @xmath211 is anticlockwise adjacent to @xmath213 .
then we get a good path set by attaching the vertex @xmath213 to the path @xmath136 after @xmath211 . from lemma [ lem :
lem38 ] , @xmath211 is not clockwise adjacent to @xmath214 ( we can apply lemma [ lem : lem38 ] to @xmath211 because from lemma [ lem : lemr2x ] , @xmath215 and as @xmath216 it is clear that @xmath217 ) . hence , there are at most @xmath218 arcs that are anticlockwise adjacent to @xmath211 .
it follows from lemma [ lem : lem33 ] that there are at least @xmath219 arcs clockwise adjacent to @xmath211 . if @xmath213 is not clockwise adjacent to @xmath211 then there are at most @xmath220 ( the last inequality following from the assumption @xmath221 ) arcs that are clockwise adjacent to @xmath211 which is a contradiction .
hence @xmath213 is clockwise adjacent to @xmath211 if @xmath221 resulting in a good path set as explained above .
suppose hadwiger s conjecture is false for proper circular arc graphs .
then let @xmath68 be a proper circular arc graph on the smallest number of vertices that is a counterexample to hadwiger s conjecture .
we continue to use the notations provided in table [ table : table1 ] and the labeling scheme shown in fig [ fig : fig1 ] for the proof . since @xmath68 is a minimum counterexample to hadwiger s conjecture it satisfies lemma [ lem : lem37 ] , lemma [ lem : lemk ] , lemma [ lem : lem38 ] and lemma [ lem : lem39 ] .
we will show that these properties will allow us to color @xmath0 using @xmath222 colors leading to a contradiction .
let @xmath223 .
let @xmath149 . as @xmath1 can not be a multiple of @xmath13 ( by lemma [ lem : lemk ] ) , we have @xmath224 .
now we demonstrate a vertex coloring of @xmath68 , @xmath225 .
we need slightly different strategies for coloring depending on whether @xmath226 is even or @xmath226 is odd .
[ note : indset ] after proposing a vertex coloring @xmath227 , to prove that it is a valid vertex coloring , our strategy would be to show that for each @xmath124 , @xmath228 , the set of vertices that are colored @xmath124 form an independent set in @xmath229 . to make the discussion easy , we first observe the following simple fact : let @xmath230 be a set of vertices of @xmath229 , where @xmath231 .
let next@xmath232 , next@xmath233 and next@xmath234 , for @xmath235 . then to prove that @xmath236 is an independent set in @xmath0 , it is sufficient to show that for each @xmath237 , @xmath54 is not anticlockwise adjacent to next@xmath238 .
this in fact is an easy consequence of corollary [ cor : fr2 ] . 1 .
t is even .
a. @xmath239 for @xmath240 b. @xmath241 for @xmath242 c. @xmath243 for @xmath244 + in order to show that @xmath245 is a valid coloring , we will show for each @xmath246 , @xmath247 , the subset of vertices that are given color @xmath246 induces an independent set .
the number of vertices that get color @xmath246 vary with the different ranges of @xmath246 as seen below . 1 . for the range @xmath248 , the arcs that get color @xmath246 are @xmath249 .
note that @xmath250 and hence @xmath251 gets color @xmath252 .
if @xmath159 is clockwise adjacent to @xmath253 then by corollary [ cor : fr2].1 , at least @xmath23 arcs are clockwise adjacent to @xmath253 which contradicts lemma [ lem : lem24 ] .
hence , @xmath159 is not clockwise adjacent to @xmath253 .
also @xmath254 is not clockwise adjacent to @xmath98 for @xmath255 as @xmath256 is not clockwise adjacent to @xmath98 by lemma [ lem : lem38 ] .
arc @xmath257 is not anticlockwise adjacent to @xmath253 because otherwise we have by corollary [ cor : fr2].4 , @xmath258 ( since by lemma [ lem : lem39 ] , @xmath259 ) arcs anticlockwise adjacent to @xmath253 which contradicts lemma [ lem : lem24 ] .
hence , by note [ note : indset ] vertices that are given color @xmath246 form an independent set .
2 . for @xmath260 ,
the arcs that get color @xmath246 are @xmath261 .
the arcs in this set are also not adjacent to each other for the same reasons as discussed above .
3 . for @xmath262
, only arc @xmath253 gets color @xmath246 and hence it is an independent set .
4 . only one arc gets color @xmath246 for @xmath246 in the range @xmath263 and hence each is an independent set .
+ thus , for the case when @xmath226 is even , we have demonstrated a valid vertex coloring using @xmath222 colors .
+ we have two sub - cases here based on the values of @xmath13 and @xmath264 . we know from lemma [ lem : lem2x ] that @xmath265
. also we have @xmath266 .
therefore , we have @xmath267 with the equality holding when @xmath268 and @xmath269 . + * sub - case 1 : @xmath270 * a. @xmath239 for @xmath240 b. @xmath271 for @xmath272 c. @xmath273 for @xmath274 d. @xmath275 for @xmath276 + for @xmath277 , let @xmath278 denote the set of vertices of @xmath229 that are given the color @xmath246 .
we will show that @xmath278 is an independent set . as before we consider the different ranges of @xmath246 and study @xmath278 .
the reader may find it useful to note now itself that no two arcs that belong to the range @xmath279 ( i.e. the range defined in ( iii ) ) are given the same color , since there are only at most @xmath280(last inequality true as @xmath281 ) arcs that belong to this range
. 1 . for @xmath282 : recall that @xmath283 .
if @xmath284 and @xmath269 , then the range defined by @xmath285 is empty .
thus in this case , @xmath286 .
otherwise we consider two cases : if @xmath287 then @xmath288 else @xmath289 .
+ now we verify that @xmath278 is an independent set in @xmath229 : @xmath159 is not clockwise adjacent to @xmath253 and @xmath254 is not clockwise adjacent to @xmath98 for @xmath290 for reasons discussed in 1(a ) of @xmath226 is even case above .
@xmath291 is not anticlockwise adjacent to @xmath292 since the number of arcs anticlockwise adjacent to @xmath292 would then be @xmath293 which contradicts lemma [ lem : lem24 ] .
arc @xmath292 is not anticlockwise adjacent to @xmath253 as the number of arcs that are anticlockwise adjacent to @xmath253 by corollary [ cor : fr2].4 would then be @xmath294 ( since by lemma [ lem : lem39 ] @xmath295 ) which contradicts lemma [ lem : lem24 ] .
arc @xmath253 is also not anticlockwise adjacent to @xmath292 because then by corollary [ cor : fr2].1 number of arcs that will be clockwise adjacent to @xmath253 would be @xmath296(using lemma [ lem : lem2x ] and lemma [ lem : lemr2x ] ) which contradicts lemma [ lem : lem24 ] .
now by note [ note : indset ] we can easily deduce that the above set is an independent set .
2 . for @xmath262 , arcs that get color @xmath246
are @xmath253 and @xmath292 ( if @xmath297 ) .
@xmath292 is neither anticlockwise adjacent nor clockwise adjacent to @xmath253 as discussed above .
therefore , arcs that get color @xmath246 form an independent set .
arcs that get color @xmath298 for @xmath299 are @xmath300 .
arc @xmath301 is not anticlockwise adjacent to @xmath302 as then the number of arcs that are anticlockwise adjacent to @xmath302 by corollary [ cor : fr2].3 would be @xmath303 which is a contradiction by lemma [ lem : lem24 ] .
it is easy to see that @xmath301 is also not clockwise adjacent to @xmath302 .
+ we have seen that the arcs that get color @xmath246 for @xmath304 form an independent set and hence the coloring is a valid coloring .
+ * sub - case 2 : @xmath305 * + as mentioned before , if @xmath305 we should have @xmath268 and @xmath269 .
we first make the following claim : + * claim * if @xmath306 , then for any @xmath307 in the range @xmath308 , the arc @xmath211 is not anticlockwise adjacent to @xmath213 .
+ * proof of claim : * suppose there exist a @xmath307 , where @xmath166 such that @xmath211 is anticlockwise adjacent to @xmath213 .
then , we demonstrate a good path set , contradicting lemma [ lem : lem37 ] . for @xmath166 ,
the @xmath13 sequences defined by @xmath309 indeed form vertex disjoint paths ( by lemma [ lem : lem35 ] ) .
we would show how to extend each path @xmath136 ( @xmath166 ) to a path @xmath310 such that @xmath310 ends at the vertex @xmath213 .
if we assume that the vertex @xmath311 is anticlockwise adjacent to @xmath213 then the path @xmath136 can be readily extended to a path @xmath310 , by adjoining the vertex @xmath213 at the end of @xmath136 , i.e. just after @xmath211 .
now we extend the remaining @xmath312 paths @xmath118 ( @xmath313 and @xmath314 ) to get @xmath315 as follows : @xmath316 by applying lemma [ lem : lem35 ] , it can be easily verified that @xmath317 form paths .
moreover they are vertex disjoint paths : notice that the last but one vertex @xmath141 in each path @xmath315 ( @xmath318 , @xmath314 ) belong to the range @xmath319 . since @xmath320 , there are sufficient number of `` intermediate '' vertices to connect the last vertex of @xmath118 ( i.e. @xmath321 ) to @xmath119 .
thus we have demonstrated a good path set .
the claim follows .
+ now , we demonstrate a vertex coloring of @xmath229 using @xmath222 colors : a. @xmath239 for @xmath240 b. @xmath322 for @xmath323 c. @xmath324 for @xmath308 d. @xmath325 for @xmath326 + as before we examine the set @xmath278 , the set of arcs that get color @xmath246 .
for @xmath327 , @xmath328 .
this is an independent set for reasons discussed in 1(a ) of @xmath226 is even case above . for @xmath329 , @xmath330 . letting @xmath331
, we have @xmath332 which is not anticlockwise adjacent to @xmath333 by the claim we proved above . also , @xmath334 is not anticlockwise adjacent to @xmath253 because otherwise the number of arcs that are anticlockwise adjacent to @xmath253 would be @xmath335 by corollary [ cor : fr2].4 which contradicts lemma [ lem : lem24 ] .
thus @xmath278 is an independent set for @xmath329 . finally , it is easy to see that @xmath278 is a singleton set for @xmath336 and hence forms an independent set
. therefore the above coloring is a valid coloring that uses only @xmath222 colors .
therefore , we can see that irrespective of whether @xmath226 is even or odd , we can show that we either have a good path set(and thus an @xmath121 clique minor ) or we can color using @xmath222 colors which is a contradiction . hence hadwiger s conjecture is true for proper circular arc graphs . | circular arc graphs are graphs whose vertices can be represented as arcs on a circle such that any two vertices are adjacent if and only if their corresponding arcs intersect .
proper circular arc graphs are graphs which have a circular arc representation where no arc is completely contained in any other arc .
hadwiger s conjecture states that if a graph @xmath0 has chromatic number @xmath1 , then a complete graph on @xmath1 vertices is a minor of @xmath0 .
we prove hadwiger s conjecture for proper circular arc graphs . , circular arc , proper circular arc , hadwiger s conjecture , minor , graph coloring |
yakhshiev acknowledge international centre for theoretical physics ( ictp ) for the hospitality during his stay at the centre . the work of u.t .
yakhshiev was supported in part by intas ( ysf 00 - 51 ) .
khanna would like to thank tony thomas for the hospitality at cssm where the main ideas were considered .
the research of f.c.k . is partly supported by nserc .
finally , this work was supported in part by the australian research council .
99 a.m.rakhimov , u.t.yakhshiev , f.c.khanna .
phys . rev . * c61 * , 024907 ( 2000 ) .
r.machleidt , adv .
. phys . * 19 * , 189 ( 1989 ) .
a.w.thomas , k.holinde .
* 63 * , 2025 ( 1989 ) ; p. a. guichon , g. a. miller and a. w. thomas , phys .
b * 124 * , 109 ( 1983 ) .
k.holinde , a.w.thomas , phys . rev . *
c42 * , r1195 ( 1990 ) . | the temperature and density dependence of the meson - nucleon vertex form factors is studied in the framework of thermofield dynamics .
results are obtained for two rather different nucleon - nucleon potentials : the usual bonn potential and the variation with a softer @xmath0 form factor , due to holinde and thomas . in general , the results show only a modest degree of sensitivity to the choice of interaction .
recently @xcite the temperature dependence of the meson nucleon form factors has been calculated in thermofield dynamics ( tfd ) using the bonn @xmath1 interaction @xcite that includes exchanges of @xmath2 and @xmath3 mesons , with the @xmath0 vertex form factor having a monopole form @xmath4 where @xmath5 gev for the @xmath0 vertex .
it is anticipated that in low energy @xmath1 scattering @xmath6 so that the 4-momentum transfer squared is usually minus the 3-momentum transfer squared .
however it has been argued @xcite that there are several factors that point to a significantly softer form factor with @xmath7 mev .
it was concluded that the @xmath0 form factor for a free nucleon is close to the measured axial - vector form factor , thus giving @xmath8 mev .
it has been suggested that a soft @xmath0 form factor would make it difficult to fit the @xmath1 scattering and , in particular the properties of the deuteron .
however holinde and thomas @xcite established that a good fit can be obtained for @xmath9 mev .
the study of the temperature and density dependence of the coupling constants , form factors and the critical temperature , where the coupling constant goes to zero , is needed for this softer @xmath0 form factor .
this would clarify any dependence on the choice of the interaction and give us a better understanding of the equation of state that may be relevant for heavy - ion collisions .
the behaviour of nuclear matter at finite temperature and density and the phase transition from the hadronic to the quark - gluon phase with subsequent hadronisation can provide valuable information about the nature of confinement in qcd . in this brief communication we present results for the temperature and density dependence of the meson - nucleon couplings , using the obepti @xmath1 interaction @xcite , and compare them to the results obtained earlier with the bonn potential , obep @xcite .
the main calculational procedure has been described earlier @xcite and we concentrate on discussing the results in the present note .
the temperature dependence is calculated within thermofield dynamics , a finite temperature field theory .
the finite temperature modifications of the vertex functions are calculated from the one - meson - exchange feynmann diagrams shown in fig .
[ fig1 ] .
the mathematical representation was explained earlier @xcite . to initiate the calculations , zero temperature coupling constants and vertex form factors
are chosen to be either i)the bonn potential , obep , or ii ) the modified bonn potential with a soft @xmath0 form factor , obepti , due to holinde and thomas ( see table [ potpar ] ) .
cccccc meson & @xmath10 & mass ( gev ) & @xmath11 ( gev ) & @xmath12 & @xmath13 + + @xmath14 & @xmath15 & 0.138 & 1.05 & 14.9 & @xmath16 + @xmath17 & @xmath18 & 0.769 & 1.3 & 0.99 & 6.1 + @xmath3 & @xmath19 & 0.783 & 1.5 & 20.0 & @xmath16 + @xmath20 & @xmath21 & 0.550 & 2.0 & 8.383 & @xmath16 + + @xmath14 & @xmath15 & 0.138 & 0.8 & 14.6 & @xmath16 + @xmath22 & @xmath15 & 1.200 & 2.0 & 100 & @xmath16 + @xmath23 & @xmath24 & 0.549 & 1.5 & 5.0 & @xmath16 + @xmath17 & @xmath18 & 0.769 & 1.3 & 0.92 & 6.6 + @xmath3 & @xmath19 & 0.783 & 1.5 & 20.0 & @xmath16 + @xmath25 & @xmath26 & 0.983 & 2.0 & 2.881 & @xmath16 + @xmath20 & @xmath21 & 0.550 & 2.0 & 8.383 & @xmath16 + it is important to note that results for @xmath27 , appearing in obepti , can be obtained from those of the @xmath14-meson by the following procedure @xmath28 the ratios of the coupling constants , @xmath29 , are plotted in fig .
[ fig2 ] .
note that the @xmath30 and @xmath31 mesons are present only for the potential of holinde and thomas .
we observe that for the @xmath32 and @xmath23-mesons this ratio goes through zero at different temperature depending on the density .
this temperature is called the critical temperature , @xmath33 . for @xmath34 and @xmath35 mesons
the behaviour of the ratio is opposite
i.e. it increases instead of going through zero . for @xmath36 mesons
the difference between the two potentials is smallest at a density of order @xmath37 , where @xmath38 is normal nuclear matter density .
the changes with temperature for the two sets of parameters are usually quite similar , though the actual numbers differ . in fig .
[ fig3 ] we show the critical temperatures as a function of the density of nuclear matter .
the behaviour of @xmath39 for various mesons using the two potentials is quite different . for the @xmath36 and @xmath40 mesons , @xmath39 changes more rapidly for the softer pion vertex , while for the @xmath41meson @xmath39 is practically constant in contrast with the obep potential , where it changes rapidly .
the changes in @xmath42 show quite a significant dependence on the choice of potential .
we have also investigated the @xmath43-dependence of the meson - nucleon form factors with temperature .
the results of these calculations presented in fig .
[ fig4 ] . in this case
the pion form factor shows a large change and a stronger dependence on the choice of @xmath44 parameter .
for other mesons the form - factors show very little dependence on a hard or soft pion vertex function .
it is clear that at density @xmath45 , the behaviour of the critical temperature for the @xmath17 meson is dramatically changed when the parameters of the obe potential are changed . in other words
, @xmath46 is sensitive to parameter changes . for the obepti potential @xmath47
is almost constant , while for the obep potential @xmath47 decreases rapidly . for the @xmath14 meson
@xmath48 decreases even more rapidly with increasing density . for other mesons
@xmath48 has the same behaviour , increasing with increasing density . for all mesons except the pion , the critical temperature increases when one uses the obepti parametrization
nevertheless , at moderate densities the critical temperature for the @xmath14 is also somewhat higher . in fig .
[ fig4 ] we present the form factors at several temperatures at zero density .
the form factors of the @xmath0 and @xmath49 vertices are more sensitive to parameter changes , while @xmath50 and @xmath51 are not .
this is also seen from fig .
[ fig3 ] , where the critical temperature for the @xmath3 and @xmath20 mesons is almost the same for both parameter sets at zero density . at high densities
@xmath52 , and @xmath53 fall off more rapidly at high momentum transfer , constituting quenching of the cut off parameters for corresponding meson form factors . to summarise our results we present a practical parametrization of the temperature and density dependence of the form factors . this may be useful for other applications . at small momentum transfers
we can parametrize them by a monopole form @xmath54 where we still use the mass of the corresponding meson at zero temperature and density . in general the effective mass of a meson , @xmath55 ,
is also temperature dependent , but that has not been considered here . from fig .
[ fig3 ] we see that @xmath56 has an almost linear dependence on density , except for the case of @xmath20-meson .
however , even for this case , at moderate densities there is an almost linear dependence .
consequently , we can write the following density dependence of the critical temperature @xmath57 furthermore , using a polynomial form for the ratios @xmath58 we find the values of the parameters given in table [ paramset ] .
again it is clear that @xmath59 can be restored from @xmath60 as given in eq .
( 2 ) .
ccccccccc meson & @xmath61 & @xmath62 & @xmath63 & @xmath64 & @xmath65 & @xmath66 & @xmath67 & @xmath68 + + @xmath14 & 1.329 & -2.390 & 0.033 & -0.230 & -0.400 & -0.008 & -0.008 & 406 + @xmath17 & 0.066 & -1.058 & -0.002 & -0.070 & -1.976 & -0.007 & -0.008 & 285 + @xmath3 & 0.138 & 0.808 & -0.008 & 0.008 & -0.524 & 0.006 & 0.031 & 170 + @xmath20 & 0.559 & -1.547 & 0.010 & -0.018 & -0.024 & -0.003 & 0.005 & 93 + + @xmath14 & 1.018 & -2.029 & 0.031 & -0.142 & -0.369 & -0.007 & -0.016 & 429 + @xmath17 & 0.108 & -1.079 & -0.001 & -0.081 & -2.012 & -0.006 & 0.001 & 281 + @xmath3 & 0.062 & 0.904 & -0.013 & 0.005 & -0.547 & 0.008 & 0.058 & 171 + @xmath20 & 0.554 & -1.528 & 0.016 & -0.008 & -0.043 & -0.002 & 0.007 & 93 + @xmath23 & 1.449 & -2.442 & -0.020 & -0.963 & 14.468 & -0.003 & 0.012 & 478 + @xmath25 & 0.211 & 0.724 & -0.014 & 0.233 & -0.364 & 0.019 & 0.045 & 152 + in conclusion , the choice of hard and soft @xmath0 form factors does change the behaviour of the critical temperature with density . while the changes are not , in general , dramatic , there is a qualitative change for the @xmath17 meson . when it comes to the @xmath43-dependence of the form - factors , the changes for all mesons are similar .
the parametrization of the changes with temperature and density of the boson coupling constants to the nucleon , as well as the cut - off parameters in the form factors , provides a useful way to summarise the variation of these parameters .
finally it is important to realize that while the couplings of the mesons to nucleons decrease to zero at a critical temperature , it is possible that before we reach that temperature , the system may have already undergone a phase transition , leading to a deconfined quark - gluon state . |
in 2006 , babar collaboration first reported a broad structure @xmath10 with mass @xmath11 mev and full width @xmath12 mev @xcite , almost at the same time , belle collaboration also reported a new strange charmed @xmath5 state @xmath0 , with mass @xmath13 mev and width @xmath14 mev @xcite , later modified to @xmath15 mev and @xmath16 mev @xcite . in 2009 , babar showed their new observation of @xmath0 @xcite : @xmath17 mev and a broad full width @xmath18 mev . recently , lhcb collaboration confirmed the existence of the @xmath0 and measured its mass and width to be @xmath19 mev , @xmath20 mev @xcite . based on different models , interested theorists gave different assignments of @xmath0 .
in the works of refs .
@xcite , authors consider the possibility of it to be a multiquark configuration , but their conclusions are negative .
most authors believe that it is a conventional @xmath4 state .
for example , in refs .
@xcite , the first radial excited @xmath21 ( @xmath22 ) state is favored , while in ref .
@xcite , the orbitally excited @xmath23 ( @xmath6 ) state is expected , and others @xcite believe it is most likely a mixture of @xmath24 . although its existence has been confirmed by three different experiments through different production mechanism , and the data of its mass and full width agree with each other s within their error bars , @xmath0 is not currently reported in the summary table of the particle data group .
this shows that more careful experimental researches and theoretical studies are still needed . in this paper , we give another assignment of @xmath0 .
we claim that the current detected @xmath0 is not a single state , but could be two overlapping states .
one of them is the @xmath22 dominant state with a little @xmath25 wave mixed in it . throughout the whole paper ,
we use the symbol @xmath26 to label this @xmath22 dominant state , and since there may be confusion , it will be also called low - mass state .
the other is @xmath6 dominant state with a little @xmath27 wave mixed in it , we use the symbol @xmath28 for it , or high - mass state if there is confusion . because the invariant mass distributions describing these two resonances overlap together , it is equal to say the overlapping states are mixtures of pure @xmath22 state and pure @xmath6 state .
there are several reasons resulting in this overlapping assignment of two states , we will show these in different aspects . in section 2 ,
we give the masses and wave functions of @xmath5 @xmath4 states ; in section 3 , we show the decay modes and relative branching ratios ; the production rates of @xmath22 and @xmath6 states in @xmath29 decays is given in section 4 ; finally we do some discussions and make a conclusion in the last section .
theoretically , the @xmath4 @xmath22 and @xmath6 states have similar masses , and the same quantum number @xmath30 , so they mix together , and become other two different states .
there are many examples , such as @xmath31 and @xmath32 .
@xmath32 is a @xmath6 dominant state but mixed with some contribution from @xmath22 wave , famous as the @xmath33 mixing state ( there may be some contributions from @xmath34 and @xmath35 states , see ref .
@xcite for example ) , while @xmath31 is the orthogonal partner of @xmath32 , and it is @xmath22 dominant state with a little @xmath25 wave mixed in it . after mixing ,
two pure @xmath22 and @xmath6 states change into another two mixing states : @xmath36 @xmath37 where the mixing angle @xmath38 , and because of @xmath27 wave dominant , @xmath39 is the low - mass mixing state , while @xmath40 is @xmath25 wave dominant high - mass state .
the method above of mixing seems sort of artificial , so we do not choose this method . instead
, we believe that the reason of mixing is @xmath22 and @xmath6 states have the same quantum number of @xmath30 , and the forms of their wave functions are similar for the same reason . when we choose a relativistic method based on quantum field theory to deal with the bound state problem , the @xmath22 and @xmath6 states will be obtained , and the mixing should be exist automatically , not like man - made by hand .
the bethe - salpeter equation @xcite is such a relativistic method to describe a bound state , by solving it , the eigenvalue and relativistic wave function for a bound state will be obtained .
so we choose the ordinary cornell potential , and solve the exact instantaneous bethe - salpeter equations ( or the salpeter equations @xcite ) for @xmath41 states .
as expected , @xmath22 dominant state mixed with a little @xmath25 wave component and @xmath6 dominant state mixed with a little @xmath27 wave component are obtained .
we will outline the key point in this section .
the general form of a relativistic wave function for a vector meson ( @xmath22 or @xmath6 , or their mixing states ) can be written as 16 terms constructed by @xmath42 , @xmath43 , @xmath44 and the gamma matrices . because we make the instantaneous approximation of the bs method , the 8 terms with @xmath45 vanished .
so the general form for the relativistic salpeter wave function which has the quantum number of @xmath46 for a vector state can be written as @xcite : @xmath47 + m{\not\!\epsilon}^{\lambda}_{\perp}f_5(q_{\perp})\ ] ] @xmath48 where the @xmath42 , @xmath43 and @xmath49 are the momentum , relative inner momentum and polarization vector of the vector meson , respectively ; @xmath50 is a function of @xmath51 , and we have used the notation @xmath52 ( which is @xmath53 in the center of mass system ) . with this form of wave functions , we solved the full salpeter equation for @xmath5 states numerically . for strange quark
, we choose the constitute quark mass @xmath58 mev ; for the parameter @xmath59 , we give it by fitting the ground state mass @xmath60 mev ; for other parameters , we choose the same values as given in ref .
( @xcite ) .
the mass spectra of @xmath4 and @xmath61 systems are shown in table i , where we also show the experimental data of @xmath61 system from particle data group @xcite .
one can see that our predictions for the mass of @xmath61 @xmath5 system fit the data very well . for @xmath62 vector states
, we found there are two states around @xmath7 mev , their masses are @xmath63 mev and @xmath64 mev . to see the nature of these states ,
we draw the wave functions of the first three states in figure 1 - 3 .
the results show that , each state has four different wave functions , @xmath65 , @xmath66 , @xmath67 and @xmath68 .
for the first two states , the numerical values of @xmath27 wave functions @xmath67 and @xmath68 are dominant , while the @xmath25 wave components , @xmath69 and @xmath70 ( @xmath65 and @xmath66 always show up followed by @xmath71 ) are neglected .
so we conclude that the first one with mass @xmath72 mev is the ground state @xmath73 , and the second one with a node structure is the @xmath22 dominant state . for the third state , see figure 3 , @xmath69 , @xmath70 and @xmath67 , @xmath68 are all sizable , and they look all like @xmath25 waves on seemingly .
but if we write them in spherical polar coordinates , there are sizable @xmath27 wave components mixed in @xmath25 wave ( the @xmath74 term of @xmath75 in eq .
( 21 ) in ref .
@xcite ) , so the third state is a @xmath25 wave dominant state .
as for the @xmath27 wave dominant states , @xmath34 and @xmath22 states ( see figures 1 - 2 ) , we have the rough relations @xmath76 and @xmath77 . if we delete the negligible terms @xmath65 and @xmath66 in eq .
( [ eq13 ] ) , then the vector wave function is changed to the non - relativistic case @xmath78 .
but if we solve the salpeter equation with this non - relativistic wave function form as input , as expected , we only obtain the @xmath34 , @xmath22 , @xmath35 states , _ et al _ , and no @xmath25 wave states appear in this non - relativistic case . in this case
, the @xmath25 wave state has to be dealt with another wave function form as input and the mixing should be treated as the method shown in eq .
( [ eq1 ] ) . &
@xmath35 & @xmath79 + @xmath80 & 3096.9 ( input ) & 3688.1 & 3778.9 & 4056.8 & 4110.7 + @xmath81 & 3096.916 & 3686.093 & 3772.92 & 4040 & 4159 + @xmath82 & 2112.0 ( input ) & 2669.0 & 2737.3 & 2994.3 & 3033.2 + wave functions for the ground state @xmath83.,scaledwidth=65.0% ] wave functions for the first excited state @xmath84.,scaledwidth=65.0% ] wave functions for the second excited state @xmath85.,scaledwidth=65.0% ] [ mass ] pierro @xcite & close @xcite & li @xcite&matsuki @xcite&nowak @xcite + @xmath1 & @xmath63 & 2730 & 2730 & 2722 & 2806 & 2711 & 2653 & 2755 & 2632 + @xmath2 & 2737 & 2900 & 2820 & 2845 & 2913 & 2784 & 2775 & 2817 & 2720 + @xmath86 & 68 & 170 & 90 & 123 & 107 & 73 & 122 & 62 & 88 + many authors have studied the mass spectra of @xmath4 @xmath5 states using different models .
we list some of them which have both the masses of @xmath22 and @xmath6 in table [ mass ] . except our results ,
all of the predictions are pure @xmath22 and @xmath6 states , and the mass shift @xmath87 between @xmath22 and @xmath6 is also shown in table [ mass ] , the values of mass shifts lie in the region from @xmath88 mev to @xmath89 mev .
we notice that in the mass region of @xmath90 mev which has been scanned by experiments , currently there is no @xmath5 candidate except @xmath91 .
the @xmath5 candidate @xmath92 observed by selex collaboration @xcite has not been confirmed by babar @xcite , cleo and focus @xcite , and its mass of @xmath93 mev seems a little small as a @xmath22 or @xmath6 state ( see table [ mass ] ) .
both @xmath22 and @xmath6 @xmath94 @xmath5 vector states around @xmath7 mev ( or higher ) can decay to @xmath95 , @xmath96 , @xmath97 , @xmath98 , @xmath99 and @xmath100 . comparing with other possible decay channels ,
these six have dominant branching ratios because they are all ozi allowed strong decays .
so one can use them to estimate the full widths of @xmath22 and @xmath6 . in a previous letter @xcite , we have already calculated the pure @xmath101 or @xmath102 @xmath94 @xmath103 state decaying to @xmath95 , @xmath96 and @xmath99 @xcite , where we used the reduction formula , pcac relation and low energy theorem , thus the transition @xmath27-matrix is given as a formula involving the light meson decay constant and the corresponding transition matrix element between two heavy mesons .
the transition matrix element is written as an overlapping integral of the relevant wave functions , which are obtained numerically by solving the salpeter equation with further non - relativistic approximation . in this paper , we re - calculate the ozi allowed decay modes but with full wave functions as input , and use the new predicted masses of @xmath22 and @xmath6 in table i as input , so three more channels are opened , and the results are shown in table iii , where the errors in our results are obtained by varying all the input parameters simultaneously within @xmath104 .
we do not repeat the calculation here , interested reader can find the details in ref .
@xcite .
one can see that , both @xmath105 and @xmath106 have broad full widths , since they have at least 6 ozi allowed strong decay channels .
the full width of @xmath106 is much broader than that of @xmath105 , which is caused by two reasons .
one is the node structure of @xmath22 wave function ( see figure 2 ) : the @xmath22 wave function before the node provides positive contribution to the width , while after the node it gives negative contribution
. the @xmath106 state does not suffer from the node structure since it s wave function has no node structure as a @xmath6 state .
the other reason is that the higher mass state @xmath106 has larger phase space .
thus the full width of @xmath106 is broader than that of @xmath105 .
[ ourwid ] & @xmath98 & @xmath99 & @xmath100 & total widths + @xmath105 & @xmath107 & @xmath108 & @xmath109 & @xmath110&@xmath111 & @xmath112 & @xmath113 + @xmath106 & @xmath114 & @xmath115 & @xmath116 & @xmath117&@xmath118 & @xmath119 & @xmath120 + [ widths ] d^{*}k)}{br(d^{+}_{sj}\rightarrow dk)}$ ] & full width + ours & @xmath22 & 17.6 & 23.8 & 1.35&46.4 + close @xcite & @xmath22 & 22 & 78 & 3.55&103 + zhang @xcite & @xmath22 & 3.2 & 27.2 & 8.5&32 + colangelo @xcite & @xmath22 & & & 0.91 & + zhong @xcite & @xmath22 & 11 & 18.1 & 1.65&31 + li @xcite & @xmath22 & 4.4 & 34.9 & 7.9&41.4 + ours & @xmath6&44.8 & 23.8 & 0.53&73 + zhang @xcite & @xmath6&49.4 & 13.2 & 0.27&73 + colangelo @xcite&@xmath6 & & & 0.043 & + zhong @xcite&@xmath6 & 148.6 & 36.3&0.24&200 + li @xcite&@xmath6 & 86.8 & 37.2&0.43&138.2 + babar recently measured the ratios of branching fractions @xcite : @xmath121 we listed theoretical predictions of these two decay widths and their ratios with @xmath22 and @xmath6 assignments of @xmath0 in table iv .
one can see that in the current existing theoretical predictions , except the colangelo s result , all the ratios of @xmath22 assignments are larger than @xmath122 , and all the ratios of @xmath6 assignments are smaller than @xmath123 , so neither the pure @xmath22 nor pure @xmath6 assignments of @xmath0 consist with babar s data .
this data favor sizable @xmath22 component and sizable @xmath6 component in @xmath0 , so only the assignments of two overlapping states or mixing state can match the experimental data .
the belle collaboration @xcite have detected the production of @xmath0 which is produced in @xmath124 exclusive decay , and the branching fraction is : @xmath125 in ref .
@xcite , we calculated the production rate of @xmath0 in @xmath124 strong decay with two assignments , @xmath0 is a @xmath22 dominant state @xmath105 which is mixed with a little bit of @xmath25 wave component , or it is a @xmath6 dominant state @xmath106 with small @xmath27 wave mixed in it . with these assignments , we obtained @xmath126 @xmath127 these results are exactly proportional to the square of decay constant of @xmath105 or @xmath106 , so the results are very sensitive to the values of decay constants . we know that usually the decay constant of @xmath22 state is much larger than that of @xmath6 state . with further calculations we gave the product of ratios @xcite @xmath128 and
@xmath129 the former is consistent with belle s data , and the later is about one order smaller .
these results show that the production of @xmath22 state @xmath105 is favored , while @xmath6 state @xmath106 is suppressed in the experiment of belle .
the new state @xmath0 detected by belle is based on an analysis of @xmath130 events collected at the @xmath131 resonance .
while in the experiment of babar , their analysis of @xmath0 is based on data sample recorded at the @xmath131 resonance and @xmath132 mev below the resonance , and the background from @xmath133 events is removed by requiring the center of mass momentum @xmath134 of the @xmath8 or @xmath9 system to be greater than @xmath135 gev . unlike belle or babar ,
lhcb collaboration try to made sure that the candidates of @xmath0 are produced in the primary @xmath136 interaction , and reduces the contribution from particles originating from @xmath137hadron decays .
it is not clear to us whether the @xmath6 state @xmath106 is suppressed or not in the experiments of babar and lhcb , but according to the broad full width @xmath138 mev of @xmath0 and the ratio shown in eq .
( [ eq4 ] ) detected by babar , it seems that there is no suppression of @xmath6 state @xmath106 .
we also point out that it is crucial to detect the ratio @xmath139 in belle and lhcb to see if the @xmath6 state is suppressed or not .
it is known that there exist the @xmath22 and @xmath6 vector states , and their masses are usually close to each other . in the case of @xmath4 bound states ,
the mass predictions by different theoretical models are listed in table [ mass ] , and we did find that most of the predicted mass shifts @xmath140 are small . in table
[ ourwid ] and table [ widths ] , theoretical models also show that both @xmath22 and @xmath6 are broad states .
two states whose full widths are all broad with closed masses means that we will find overlapping in @xmath8 or @xmath9 invariant mass distributions in the mass regions of @xmath22 and @xmath6 @xmath4 states .
some models , for example , godfrey @xcite estimated a higher mass of @xmath6 state with large mass shift @xmath140 , but we argue that higher @xmath6 mass will result in broader full width . and currently , the scanned results in experiment show that there is no other @xmath5 state candidate in the mass region from @xmath141 mev to @xmath142 mev
. one may argue that the production of @xmath6 state may be suppressed like in belle , but the @xmath143 state of @xmath144 mev is already found , there is no reason to believe that it is more difficult to produce a @xmath6 state than to produce a @xmath143 state . for the pure @xmath22 state assignment of @xmath0 , theoretically , except for ref .
@xcite , which gives a large @xmath22 full width of @xmath145 mev , all the predictions show that the pure @xmath22 whose full width ranges from @xmath146 to @xmath147 mev is too narrow to fit data , and the ratios of @xmath0 decays to @xmath9 over those to @xmath8 can not match data except the result of colangelo @xcite . for a pure @xmath6 assignment , though the predicted full widths can match the data , we still can not explain the ratio of @xmath148 detected by babar , and we can not explain why we found the @xmath6 state , but the @xmath22 state is not found since the later is favored while the former is suppressed in belle . so the pure @xmath22 and pure @xmath6 assignment of @xmath0 can be ruled out .
the other possibility is that @xmath0 is a @xmath33 mixing state .
as shown in the bs relativistic method , the reason of mixing is that they have the same quantum number of @xmath5 , and the way of mixing may not follow the the method shown in eq .
( [ eq1 ] ) .
even we believe that it is correct , _
i.e _ , two states @xmath22 and @xmath6 mix together , and after mixing , they turn into another two mixing states . experiment find one of them , the open question is , where is the orthogonal partner of this found mixing state ?
one may argue that the missing one may be far away from the region of @xmath7 mev .
but if it is true , then there is no reason to mix together .
so the most reasonable assignment is that @xmath0 is not one single state , but two overlapping states , @xmath26 and @xmath28 , both of them are broad states and there is a narrow mass gap between them .
so the @xmath8 or @xmath9 invariant mass distributions from their decays overlap together , and then one single state @xmath0 with a broader full width is detected and reported in experiments .
we point out that , with same quantum number @xmath30 and close masses , the two broad overlapping states can not be easily distinguished by angular analysis in experiments . by this two overlapping states assignment ,
the experimental data can be explained easily .
if we set the possibilities of production of @xmath26 and @xmath28 are same , we roughly estimate that a state with mass @xmath149 mev and full width @xmath150 mev will be detected by experiments , and the predicted ratio is @xmath151 , all these values consist with the experimental data . in summary , from a study of relativistic bs method , we find around mass region @xmath7 mev , there are two @xmath4 states , @xmath26 and @xmath28 , which have the same quantum number @xmath30 and similar masses .
theoretical calculations show that both of them have broad full width , so they overlap together .
this two states assignment of @xmath0 can fit data very well .
babar collaboration , b. aubert _ et al .
* 97 * ( 2006 ) 222001 .
belle collaboration , k. abe _
et al . _ , e - print : hep - ex/0608031 .
belle collaboration , j. brodzicka _
et al . _ , phys .
* 100 * ( 2008 ) 092001 .
babar collaboration , b. aubert _ et al .
_ , phys rev . *
d 80 * ( 2009 ) 092003 [ arxiv : 0908.0806 ] .
lhcb collaboration , r. aaij _ et al .
_ , jhep * 1210 * ( 2012 ) 151 .
wang , chin .
c * 32 * 797 ( 2008 ) [ arxiv : 0708.0155 ] .
j. vijande , a. valcarce , f. fernandez , phys . rev .
* d79 * ( 2009 ) 037501 [ arxiv : 0810.4988 ] .
f. e. close , c. e. thomas , olga lakhina and eric s. swanson , phys .
* b 647 * ( 2007 ) 159 ; f. e. close and e. s. swanson , phys . rev .
* d 72 * ( 2005 ) 094004 .
p. colangelo , f. de fazio , s. nicotri and m. rizzi , phys . rev .
* d 77 * ( 2008 ) 014012 .
f. de fazio , [ arxiv : 0910.0412 ] . b. chen , d. x. wang , a. l. zhang , phys rev . * d 80 * ( 2009 ) 071502 [ arxiv : 0908.3261 ] .
wang , j .-
zhang and z .- h .
wang , phys .
* b 681 * ( 2009 ) 326 [ arxiv : 1001.2035 ] .
bo zhang , xiang liu , w .- z .
deng and s .- l .
zhu , eur .
j. * c 50 * ( 2007 ) 617 ; x .- q .
li , xiang liu , z .- t .
wei , front .
china 4 ( 2009 ) 49 [ arxiv : 0808.2587 ] .
li , bing ma , y .- h .
liu , eur .
j. * c 51 * ( 2007 ) 359 [ arxiv : hep - ph/0703278 ] .
li , bing ma , phys rev . * d 81 * ( 2010 ) 014021 .
zhong , qiang zhao , phys rev . *
d 81 * ( 2010 ) 014031 .
j. m. richard , z. phys .
* c 4 * ( 1980 ) 211 .
e. e. salpeter and h. a. bethe , phys . rev .
* 84 * ( 1951 ) 1232 .
e. e. salpeter , phys . rev .
* 87 * ( 1952 ) 328 .
chao - hsi chang , jiao - kai chen , xue - qian li and guo - li wang , commun .
* 43 * ( 2005 ) 113 .
j. zeng , j. w. van orden and w. roberts , phys rev . *
d 52 * ( 1995 ) 5229 [ arxiv : hep - ph/9412269 ] .
t. a. lahde , c. j. nyfalt and d. o. riska , nucl . phys . * a 674 * ( 2000 ) 141 [ arxiv : hep - ph/9908485 ] . m. di pierro and e. eichten , phys rev .
* d 64 * ( 2001 ) 114004 [ arxiv : hep - ph/0104208 ] . t. matsuki , toshiyuki , kazutaka sudoh , eur . phys .
j. * a 31 * ( 2007 ) 701 [ arxiv : hep - ph/0610186 ] ; t. matsuki , t. morii and k. sudoh , prog .
* 117 * ( 2007 ) 1077 [ arxiv : hep - ph/0605019 ] .
m. a. nowak , m. rho and i. zahed , acta phys .
polon . * 35 * 2377 ( 2004 ) .
selex collaboration , phys .
lett . * 93 * 242001 ( 2004 ) .
babar collaboration , b. aubert _ et al .
_ , hep - ex/0408087 .
cleo and focus collaborations reported their negative results at 32nd international conference on high - energy physics ( ichep 04 ) , beijing , china , 16 - 22 agu 2004 . | the new particle @xmath0 has stimulated many attentions .
there are different assignments of its inherent properties
. it may be a @xmath1 , @xmath2 or the mixture of @xmath3 @xmath4 @xmath5 state . by considering its mass , decay modes , full width , production rate , and comparing with current experimental data ,
we point out that there is another more reasonable assignment : @xmath0 could be identified as two resonances , one of which is a @xmath1 state , another is a @xmath6 state , and both are @xmath4 @xmath5 states .
the two states have very close masses , which are around @xmath7 mev , and both have broad decay widths .
so in experiments , the overlapping of @xmath8 or @xmath9 invariant mass distribution coming from their decays is found , but the current experiments could not distinguish these two resonances and reported one particle . |
symmetry is ubiquitous in nature and in artificial man - made objects .
the detection and characterization of shape symmetry has attracted much attention in recent times , especially within the computer graphics community @xcite .
although most of the existing literature has focused on the detection of _ extrinsic _ symmetries ; a popular approach being transformation space voting @xcite ; there has been steadily growing interest in detection and characterization of _ intrinsic _ symmetries .
most recent efforts in intrinsic symmetry detection have focused on detection of global symmetries @xcite ; @xcite .
it is generally recognized that detection of overlapping intrinsic symmetry is a more challenging problem due to the larger search spaces involved in the detection of symmetric regions ( in comparison to global symmetry analysis ) and the determination of symmetry revealing transforms ( in comparison to extrinsic symmetry detection ) .
also , overlapping intrinsic symmetry detection and characterization is more important because it is a more generalized problem in nature than extrinsic symmetry detection problem which can be considered as a special case of overlapping intrinsic symmetry detection .
we present a formal definition of overlapping intrinsic symmetry .
an intrinsic symmetry over a shape is a subregion with associated self - homeomorphisms that preserve all pairwise intrinsic distances @xcite . in this paper
, we address the problem of intrinsic symmetry detection and characterization of shapes based on their intrinsic symmetries .
complex shapes often exhibit multiple symmetries that overlap spatially and vary in form as depicted in figure [ fig : teaser ] . overlapping symmetry analysis enables the construction of high - level representations that enhance the understanding of the underlying shape and facilitate solutions to such problems as shape correspondence , shape editing , and shape synthesis @xcite . however ,
analysis of overlapping symmetry poses additional challenges as described below .
existing approaches to intrinsic symmetry detection , including those based on region growing @xcite , partial matching @xcite , and symmetry correspondence @xcite are not able to extract physically overlapping symmetries .
lipman et al .
@xcite cluster sample surface points from an input shape and use a symmetry correspondence matrix ( scm ) to identify intrinsic symmetry properties of groups of surface points . in their approach , each scm entry measures how symmetric two surface points are based on some measure of intrinsic geometric similarity between the local neighborhoods of the points .
xu et al .
@xcite let surface point pairs vote for their partial intrinsic symmetry and perform intrinsic symmetry grouping using a 2-step spectral clustering procedure .
however , their approach lacks the ability to retrieve the final symmetry map which makes characterization of the specific intrinsic symmetry a difficult problem .
our key idea behind our approach to intrinsic symmetry detection and characterization is to approach the problem from a shape correspondence perspective and generate the transformation map which can be further used to describe the symmetry space . to this end
, we perform two stages of processing where in the first stage , representative symmetric point pairs are identified based on their local geometry and a global distance representation and in the second stage the original transformation is retrieved as a _ map _ to facilitate further characterization of the underlying symmetry .
the detected intrinsic symmetries are quite general , as shown in figure [ fig : teaser ] .
a fundamental question in symmetry detection is quantifying the extent of symmetry between a pair of points .
the primary challenge in identifying potentially symmetric point pairs is to come up with conditions strong enough to adequately constrain the symmetry search space such that the symmetry detection procedure is computationally tractable .
we rely on a local criterion , such as geometric similarity , and a global criterion , such as the distance - based symmetry support received by a point pair @xmath0 , to detect and quantify the extent of symmetry between points @xmath1 and @xmath2 .
we refer to a point pair @xmath0 which satisfies the local geometric similarity criterion as a _
good voter_. the point pairs within the population of _ good voters _ , that enjoy sufficiently strong global distance - based symmetry support are deemed to be _ symmetric _ point pairs .
the global symmetry support for point pair @xmath0 is quantified by the number of other point pairs which potentially share the intrinsic symmetry properties of @xmath0 .
the global symmetry support is computed using a simple , distance - based symmetry criterion defined over two point pairs within the population of _ good voters _ as shown in figure [ fig : overview ] .
the input to the proposed symmetry detection and characterization algorithm is a 3d shape that is approximated by a 2-manifold triangular mesh .
the local intrinsic geometry is quantified using the _ wave kernel signature _ ( wks ) @xcite whereas the global intrinsic geometric distance measure chosen is the _ biharmonic distance measure _ ( bdm ) @xcite .
the proposed algorithm consists of a voting procedure on the correspondence space defined over a set of locally symmetric surface point pairs sampled from the input 3d shape , followed by the generation of a functional map ( see figure [ fig : overview ] ) . [ [ correspondence - space - voting ] ] correspondence space voting + + + + + + + + + + + + + + + + + + + + + + + + + + + the input to the correspondence space voting procedure comprises of point pairs that can be considered as candidates for symmetry .
prior to performing the transformation space mapping procedure , we sample a set of locally symmetric point pairs from the input shape based on the similarity of their wks s . to estimate the symmetry support received by a point pair , we perform a voting procedure by counting the number of _ good voters _ which potentially share the same intrinsic symmetry as the source point pair .
the voting procedure ensures that we have a set of good point pair initializations from which we can create an initial map of the symmetry transformation that can then be extrapolated to other surface point pairs on the 3d shape .
[ [ computation - of - the - functional - map ] ] computation of the functional map + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + a functional map provides an elegant representation for the maps between surfaces , allowing for efficient inference and manipulation . in the functional map approach ,
the concept of a map is generalized to incorporate correspondences between real valued functions rather than simply between surface points on the 3d shapes .
our choice of the multi - scale eigenbasis of the laplace - beltrami operator makes the functional map representation both , very compact and yet informative .
we show that our functional map formulation not only results in a compact description of the underlying map of the symmetry transformation , but also enables meaningful characterization of the symmetry transformation .
the primary contributions of our paper can be summarized as follows : 1 .
robust and meaningful characterization of the symmetry transformation via formulation of a symmetry space to quantitatively distinguish between instances of simple and complex symmetry .
leveraging of the functional map representation to succesfully represent the map of the underlying symmetry transformation regardless of its complexity .
3 . providing a solution to the symmetric flipping issue via interpolation of the correspondence from the base surface point pairs using functional maps .
4 . enabling recovery of the symmetry groups via clustering on the functional maps
the research literature on symmetry detection has grown substantially in recent years as shown in the excellent survey by mitra et .
al . @xcite . in this paper
, we do not attempt to provide an _ exhaustive _ exposition of the state of the art in symmetry detection ; rather we focus on discussing existing works that are most closely related to our proposed approach .
several recent approaches to detect approximate and partial extrinsic symmetries have focused on algorithms that cluster votes for symmetries in a parameterized `` transformation space '' @xcite ; @xcite ; @xcite ; @xcite ] .
mitra et al .
@xcite generate `` votes '' in a transformation space to align pairs of similar points and then cluster them in a space spanned by the parameters of the potential symmetry transformations .
regardless of how good the shape descriptors are , the aforementioned methods are not effective at finding correspondences between points in complex symmetry orbits that are spread across multiple distinct clusters in the transformation space . since the dimensionality of the transformation space increases with the complexity of the symmetry , the voting procedure in transformation space becomes increasing intractable when dealing with complex symmetries .
there exists a body of published research literature that characterizes shape representations based on the extent of symmetry displayed by an object with respect to multiple transformations .
kazdhan et al .
@xcite have proposed an extension of zabrodsky s symmetry distance to characteristic functions , resulting in a _ symmetry descriptor _ that measures the symmetries of an object with respect to all planes and rotations through its center of mass .
podolak et al .
@xcite have extended the symmetry descriptor to define a planar reflective symmetry transform ( prst ) that measures reflectional symmetries with respect to all planes through space .
rustamov et al .
@xcite have extended the prst to consider surface point - pair correlations at multiple radii .
although the above representations provide a measure of symmetry for a regularly sampled set of transformations within a group , they are practical only for transformation groups of low dimensionality ( for example , rigid body transformations would require one to store a six - dimensional matrix ) and break down when faced with groups of higher dimensionality .
the exists a class of techniques that exploits the redundancy in repeating structures to robustly detect symmetries @xcite ; @xcite ; @xcite ; @xcite ; @xcite . the transformation space voting method proposed by mitra et al .
@xcite is extended in @xcite by fitting parameters of a transformation generator to optimally register the clusters in transformation space .
berner et al .
@xcite and bokeloh et al .
@xcite have taken a similar approach using subgraph matching of feature points and feature lines , respectively , to establish potential correspondences between repeated structures .
this is followed by an iterative closest points ( icp ) algorithm to simultaneously grow corresponding regions and refine matches over all detected patterns , allowing the detection of repeated patterns even in noisy data @xcite , but at the cost of requiring a - priori knowledge of the commutative group expected in the data . also , the non - linear local optimization procedure within the icp algorithm could cause it to get trapped in a local minimum if the initialization is not good enough .
lipman et al .
@xcite have proposed an eigen - analysis technique for symmetry detection that relies on spectral clustering .
the top eigenvectors of their geometric similarity - based scm characterize the symmetry - defining orbits , where each orbit includes all points that are symmetric with one another .
however , their work is not suited for multi - scale partial symmetry detection .
first , expressing local point similarities as symmetry invariants is only appropriate for global intrinsic symmetry detection . in the case of partial symmetry detection
, it is not always possible to reliably judge if two surface points are symmetric by comparing only their point ( i.e. , local ) signatures , especially when one point lies on the boundary of symmetric regions .
moreover , their single - stage clustering procedure is unable to identify overlapping symmetries .
xu et al .
@xcite have extended the eigen - analysis approach of lipman et al .
@xcite by incorporating the concept of global intrinsic distance - based symmetry support accompanied by a 2-stage spectral clustering procedure to distinguish between scale detection and symmetry detection .
although they showed some interesting results , the 2-stage spectral clustering procedure made their method extremely slow .
furthermore , the absence of transformation map retrieval meant that further processing of the detected symmetries , which are represented as point pairs , was extremely inefficient .
the proposed scheme is decoupled into two steps of _ correspondence space voting _ and _ transformation space mapping_. the _ correspondence space voting _ technique is inspired from the work of xu et al .
@xcite , but in our technique , we bypassed the 2 particularly lengthy steps of spectral clustering and the all - pair geodesic distance calculation to improve the running time quite significantly .
moreover , our introduction of _ transformation space mapping _ in symmetry detection is quite novel in the sense that this not only provides a concise description of the underlying symmetry transformation , but according to our knowledge this is one of the first works which has the unique ability of characterizing the symmetric transformation .
in this section , we present a formal description of the problem being addressed in this paper .
in particular , we define the input to and output of the proposed algorithm .
we also define the type of intrinsic symmetries the proposed algorithm is designed to detect and the formal characterization of these symmetries .
there are two primary aspects to the theoretical framework for the proposed algorithm , i.e. , _ correspondence space voting _ ( csv ) and _ functional map retrieval _
( fmr ) . in the case of csv , a joint criterion , that combines local intrinsic surface geometry and global intrinsic distance - based symmetry ,
is proposed and shown to result in a provably necessary condition for intrinsic symmetry . in the case of fmr ,
a formal scheme for the characterization of the detected symmetries , based on their complexity is proposed .
we have mainly restricted our study of intrinsic symmetries to isometric involutions for two reasons ; first , having a provably necessary condition for intrinsic symmetry provides theoretical soundness and second , the proposed symmetry criterion bounds the search space sufficiently , ensuring that the solution is computationally tractable @xcite .
following the initial detection of good symmetric correspondences , we proceed to retrieve the map of the symmetry transformation by leveraging the functional map framework .
moreover , the intrinsic symmetry ensures that the retrieved maps exhibit a diagonality characteristic .
a cost matrix is designed to exploit this characteristic such that the inner product of functional map with the cost matrix results in a _ quantitative _ evaluation of the complexity of the detected symmetry .
our problem domain is a compact , connected 2-manifold , @xmath3 , with or without a boundary .
the manifold @xmath3 is a _
3d shape _ ,
i.e. , @xmath4 . distances on the manifold @xmath3 are expressed in terms of an intrinsic distance measure . in particular , the biharmonic distance measure ( bdm ) is used on account of its ease of calculation and greater robustness to local surface perturbation when compared to the geodesic distance measure @xmath5 @xcite . in the remainder of the paper ,
we use the term intrinsic distance and biharmonic distance interchangeably . in the proposed algorithm each symmetry transformation
is represented by a map .
consequently , the output of the proposed algorithm consists of maps represented as matrices .
the output can be regarded as a complete description of all the overlapping intrinsic symmetries represented in a compact and informative manner .
suppose we are given a compact manifold @xmath3 without boundary .
following @xcite we call @xmath3 intrinsically symmetric if there exists a homeomorphism @xmath6 on the manifold that preserves all geodesic distances .
that is : @xmath7 where @xmath8 is the intrinsic distance between two points on the manifold . in this case
, we call the mapping @xmath9 an intrinsic symmetry .
we propose two simple criteria to test whether two surface point pairs on the manifold potentially share the same intrinsic symmetry .
specifically , given two surface point pairs @xmath10 and @xmath11 on manifold @xmath3 , the first criterion , which is based on local intrinsic geometry , determines the symmetry potential of the two surface point pairs by comparing their corresponding wave kernel signatures ( wks s ) as follows : @xmath12 where @xmath13 is a scale parameter .
the second criterion is based on intrinsic distance as follows : @xmath14 the above two criteria are necessarily satisfied if the surface point pairs under consideration correspond to the same intrinsic symmetry .
the proposed framework fails to detect intrinsic symmetry in cases where the second symmetry criterion ( in equation ) is not satified .
for example , figure [ fig : failure ] , depicts two human figures that form what may be perceived as a translational symmetry , although they do not possess intrinsic symmetry . the second symmetry criterion ( equation ) fails to hold in this situation . generally speaking ,
the proposed algorithm is not designed to detect all forms symmetry resulting from repeated patterns , especially if the patterns are not connected .
in contrast to the approach of xu et al .
@xcite , the use of biharmonic distance instead of geodesic distance as the intrinsic distance measure ensures that the proposed algorithm is capable of detecting intrinsic symmetry even in the presence of small perturbations ( such as small bumpy regions ) on the 3d surface @xcite .
the first step in the proposed symmetry detection and characterization framework is the correspondence space voting ( csv ) procedure .
although a voting procedure has been previously incorporated in an earlier symmetry detection technique , it was carried out primarily in transformation space @xcite .
the importance of correspondence space for the detection of symmetry was explained more recently by lipman et al .
@xcite . in our case , although the detected symmetry is finally represented in functional space , it is critical to have good initialization to ensure the success of the final map generation .
we have designed and implemented a csv algorithm to facilitate good initial guesses . the csv algorithm
comprises of three stages described in the following subsections : a subset of points with adequate discriminative power needs to be selected prior to the generation of surface point pairs .
a subset @xmath15 consisting of @xmath16 sample points is chosen from the surface of the given input 3d shape using the _ farthest point sampling _ strategy @xcite .
although originally designed to operate on geodesic distances generated using the marching cubes algorithm , in our case we have employed the farthest point random sampling strategy in biharmonic distance space .
the results of the sampling procedure are depicted in figure [ fig : sampling ] . the _ biharmonic distance measure _ ( bdm ) is similar in form to the _ diffusion distance measure _ ( ddm ) @xcite .
the bdm kernel is based on the green s function of the biharmonic differential equation . in the continuous case
, the ( squared ) biharmonic distance between two points @xmath17 and @xmath18 can be defined using the eigenvectors ( @xmath19 ) and eigenvalues ( @xmath20 ) of the laplace- beltrami operator @xcite as follows : @xmath21 the above definition of the bdm is slightly different from that of the ddm where the denominator in the case of the ddm is @xmath22 .
however , this subtle change ensures greater control over the characterization of the global and local properties of the underlying manifold in the case of the bdm .
consequently , the bdm is fundamentally different from the ddm with significantly different properties .
the bdm , as expressed in equation , captures the rate of decay of the normalized eigenvalues @xmath23 of the laplace- beltrami operator ; if the decay is too slow , it produces a logarithmic singularity along the diagonal of the green s function @xcite .
alternatively , too fast a decay basically ignores eigenvectors associated with higher frequencies , resulting in the bdm being global in nature ( i.e. , the local surface details are ignored ) .
et al . _
@xcite , demonstrated that performing quadratic normalization provides a good balance , ensuring that the decay is slow enough to capture the local surface properties around the source point and yet rapid enough to encapsulate global shape information .
in particular , lipman _ et al . _
@xcite have theoretically proved two important properties of the bdm , i.e. , that it is ( i ) a metric , and ( ii ) smooth everywhere except at the source point where it is continuous .
the key observation is that for 3d surfaces , the eigenvalues @xmath20 , @xmath24 , of the laplacian are an increasing function of @xmath25 resulting in the continuity of the bdm everywhere and also smoothness of the bdm everywhere except at the source point , where it has only a derivative discontinuity . in our implementation of the farthest point
random sampling strategy , a single point is selected randomly at first and the remaining points are chosen iteratively from remainder set by selecting the farthest point in the biharmonic distance space at each iteration .
this strategy generates a set of points located mostly in the vicinity of the shape extrema which can then be used in the subsequent surface point pair generation procedure . from the chosen subset @xmath15 consisting of @xmath16 sample points , the surface point pairs
are generated by exploiting the similarity of their local intrinsic geometric structure .
the similarity of local intrinsic geometric structure of two surface points is determined by comparing their corresponding wave kernel signatures ( wks s ) . to determine the wks of a surface point @xmath17 ,
one evaluates the probability of a quantum particle with a certain energy distribution to be located at point @xmath17 .
the behavior of the quantum particle on the surface is governed by the schrodinger equation .
assuming that the quantum particle has an initial energy distributed around some nominal energy with a probability density function @xmath26 , the solution of the schrodinger equation can then be expressed in the spectral domain as : @xmath27 aubry et al .
@xcite considered a family of log - normal energy distributions centered around some mean log energy @xmath28 with variance @xmath29 .
this particular choice of distributions is motivated by a perturbation analysis of the laplacian spectrum @xcite .
having fixed the family of energy distributions , each point @xmath17 on the surface is associated with a wks of the form : @xmath30 where @xmath31 is the probability of measuring a quantum particle with the initial energy distribution @xmath32 at point @xmath17 .
aubry et al .
use logarithmic sampling to generate the values @xmath33 .
the wks can be shown to exhibit a band - pass characteristic .
this reduces the influence of low frequencies and allows better separation of frequency bands across the descriptor dimensions . as the result ,
the wave kernel descriptor exhibits superior feature localization compared to the heat kernel signature ( hks ) @xcite .
although the wks is invariant under isometric deformation , in most practical cases where the underlying surface is represented by a discrete triangular mesh , it is not possible to strictly satisfy the invariance criterion in equation .
consequently , we have considered the 2-norm ( simply squared eucledian distance in the wks space ) of the wks and chosen surface point pairs from the subset @xmath15 which satisfy the wks invariance criterion in equation to within a prespecified threshold @xmath34 instead of requiring strict equality .
the surface point pairs that satisfy the wks invariance criterion in equation are considered for the next step of global distance - based voting .
this relaxation of the wks invariance criterion ensures that even the overlapping intrinsic symmetries are considered for voting .
we have coined the term _ good voters _ to denote the subset of surface point pairs that satisfy the wks invariance criterion ( to within the prespecified threshold ) .
the global distance - based voting step in the proposed symmetry detection technique is inspired by the work of xu et .
al . @xcite .
a subset of symmetric point pairs is extracted from the set of _ good voters _ using the global distance - based voting procedure prior to functional map generation .
the goal of the voting procedure is to accumulate symmetry support for the _ good voters _ and extract the symmetric point pairs based on the level of symmetry support received .
a point pair in the set of _ good voters _ is deemed to be symmetric if it has a sufficiently large global symmetry support , which in @xcite is measured by the number of point pairs that satisfy the intrinsic distance criterion in equation .
xu et al .
@xcite have presented a voting technique based on two point pairs followed by spectral clustering on the symmetric point pairs which enables one to distinguish whether a point pair supports one particular symmetry or more than one type of symmetry .
however , since we are are not interested , at this stage , in making the above distinction between the point pairs , we have chosen to adopt a straightforward approach of choosing a point pair and letting the set of _ good voters _ vote and decide whether or not the particular point pair satisfies the intrinsic distance criterion in equation .
another modification to the previous voting procedure of xu et al .
@xcite is our use of the biharmonic distance as the intrinsic distance instead of the geodesic distance .
this modification provides two advantages over the previous voting procedure , first , since the computation of geodesic distances between all surface point pairs is more intensive than the computation of biharmonic distances , our procedure is much faster .
second , biharmonic distances are less sensitive to noise and surface perturbations compared to geodesic distances , making our procedure more robust .
before describing our functional map approach to symmetry extraction , we first provide an overview of the functional map framework proposed by ovsjanikov et al .
a functional map is a novel approach for inference and manipulation of maps between shapes that tries to resolve the issues of correspondences in a fundamentally different manner . rather than plotting the corresponding points on the shapes ,
the mappings between functions defined on the shapes are considered .
this notion of correspondence generalizes the standard point - to - point map since every pointwise correspondence induces a mapping between function spaces , while the opposite , in general , is not true .
the new framework described above provides an elegant way to avoid direct representation of correspondences as mappings between shapes using a functional representation .
ovsjanikov et al .
@xcite have noted that when two shapes @xmath15 and @xmath35 are related by a bijective correspondence @xmath36 , then for any real function @xmath37 , one can construct a corresponding function @xmath38 as @xmath39 . in other words ,
the correspondence @xmath13 uniquely defines a mapping between the two function spaces @xmath40 , where @xmath41 denotes the space of real functions on @xmath15 . equipping @xmath15 and @xmath35 with harmonic bases , @xmath42 and @xmath43 , respectively
, one can represent a function @xmath44 using the set of ( generalized ) fourier coefficients @xmath45 as @xmath46 . translating this representation into the other harmonic basis @xmath43 ,
one obtains a simple representation of the correspondence between the shapes given by @xmath47 where @xmath48 are fourier coefficients of the basis functions of @xmath15 expressed in the basis of @xmath35 , defined as @xmath49 .
the correspondence @xmath13 between the shapes can thus be approximated using @xmath25 basis functions and encoded using a @xmath50 matrix @xmath51 of these fourier coefficients , referred to as the functional matrix . in this representation
, the computation of the shape correspondence @xmath52 is translated into a simpler task of determining the functional matrix @xmath53 from a set of correspondence constraints .
the matrix @xmath53 has a diagonal structure if the harmonic bases @xmath42 and @xmath54 are compatible , which is a crucial property for the efficient computation of the correspondence . in our symmetry extraction algorithm , instead of comparing two different shapes , we propose to compare two symmetric regions within the same shape .
in particular , based on the previously detected set of symmetric point pairs , we leverage the functional map representation in the following manner : for each pair of symmetric points , we deem one point as the source and the other as the destination and choose a local region around each point .
the ordering of the source and destination points within the pair is the same as originally chosen during the voting procedure .
the corresponding eigenbases for the points in the source and destination regions are computed .
these eigenbases are ordered based on their similarity with each other and the final functional map for that particular symmetry is extracted .
the functional map representation ensures that ( a ) the problem of symmetry extraction is tractable and , ( b ) the resulting symmetry can be represented , not by a large matrix of point correspondences , but rather as a more compact map which can be further manipulated for other applications as well .
in general , the characterization of a specific transformation based on its functional map is a challenging task . however in our case , the proposed csv framework ensures that the point pairs used in the generation of the functional map are intrinsically symmetric to a reasonable extent .
this property of intrinsic symmetry ensures that the resulting functional map is diagonal or close to diagonal @xcite .
however , in reality , there are several cases where the actual transformation deviates substantially from @xmath55-isometric deformation resulting in a densely populated functional matrix @xmath53 .
the diagonality property of the @xmath53 matrix was first exploited succesfully in an intrinsic correspondence framework using a sparse modeling technique @xcite . in their paper , the authors introduced a weight matrix @xmath56 of same size as @xmath53 where lower weights are assigned to elements close to the diagonal , and larger weights are assigned to the farther off - diagonal elements by using an inverted gaussian model where the 0 weights are assigned at diagonal and the weights at off - diagonals are larger .
the element - wise multiplication of @xmath56 with @xmath53 is a determining factor in the assessment of the diagonality of @xmath53 since the matrix inner product results in higher values for the off - diagonal elements and lower values for the diagonal elements thereby resulting in a measure of the diagonality of the matrix @xmath53 . and
represented in the increasing order of the value of inner product of @xmath56 and @xmath53.,width=302 ] in the context of symmetry characterization , problem , we assume that the off - diagonality of @xmath53 corresponds to the complexity of the symmetry transformation and therefore is penalized during multiplication with elements of the weight matrix @xmath56 , i.e. , more non - zero off - diagonal elements in @xmath53 would represent a more complex symmetry transformation .
thus , we can , not only determine the complexity of the symmetry transformation but can also succesfully formulate a 1d semi - metric symmetry space , wherein each symmetry transformation is represented as a point in the symmetry space with a value given by the inner product beween the matrix @xmath53 and weight matrix @xmath56 .
the euclidian distance between the points in the 1d symmetry space represents the complexity distance between the transformations . in the symmetry space
, any perfectly isometric transformation will result a point at origin of the line . on the other hand
, more complex symmteric transformations would result in points farther away signifying more complex transformations .
this space is called a semi - metric space because of it follows two main properties of the distance definition but not the third one . for two points @xmath15 and @xmath35 in the symmetry space ,
the distance @xmath57 from @xmath15 to @xmath35 always follows 1 .
non - negativity : @xmath58 2 .
symmetry : @xmath59 + but not 3 .
triangle inequality .
it is also possible to cluster the ponts in the 1d symmetry space to identify intrinsic symmetries which are potentially similar in nature as shown in the experimental results section to follow .
in this section , we present and discuss the results obtained by the proposed intrinsic symmetry detection algorithm on 3d shapes .
we also provide comparisons of our results with those obtained from the most closely related approaches @xcite , @xcite .
we also shown some applications where the detected symmetries can be further analyzed for symmetry characterization and clustering , potentially revealing greater semantic information about the underlying 3d shape .
most of the 3d shape models used in our experiments are from the _ non - rigid world _
dataset @xcite unless mentioned otherwise .
several discrete schemes have been proposed in recent years to approximate the laplace - beltrami operator on triangular meshes . among these ,
the one mostly used most widely for computing the discrete laplace operator is the cotangent ( cot ) scheme , originally proposed by pinkall and polthier @xcite .
recently , belkin et al @xcite proposed a discrete scheme based on the heat equation , which has been proven to possess the point - wise convergence property for arbitrary meshes .
although it is well known that no discrete laplace operator can share all of the properties of its continuous counterpart , in our experiments the aforementioned discrete schemes produce eigenfunctions that approximately preserve the convergence property of the continuous laplace operator for a reasonably well sampled triangular mesh . in all of the experiments described below ,
we have used the cotangent scheme @xcite for the computation of the discrete laplace - beltrami operator .
the results of the proposed symmetry detection algorithm are depicted in figures [ fig : teaser ] , [ fig : overlap ] , [ fig : noise ] and [ fig : clustering ] .
several important properties of the proposed algorithm are highlighted in these results .
the ability of the proposed algorithm to identify multiple intrinsic symmetries is evident from the results shown in figures [ fig : overlap ] and [ fig : clustering ] .
the extracted symmetries are seen to cover the global symmetry of the underlying 3d shape which has undergone approximate isometric deformations .
additionally , the proposed algorithm is also observed to be capable of detecting symmetry transformations that cover individual components of a 3d object that possess various forms of self - symmetry .
one particularly important aspect of the proposed algorithm is its ability to detect instances of overlapping symmetry .
an instance of overlapping symmetry is deemed to occur when a specific region on the surface of the 3d shape is simultaneously subjected to more than one symmetry transformation and , as a result , is symmetric to more than one region on the 3d shape surface .
for example , figure [ fig : overlap ] shows that the overlapping symmetry between _ all _ paws of the _ cat _ shape model is succesfully detected by the proposed technique .
six different combinations of symmetry transformation between the four paws are depicted in figure [ fig : overlap ] .
although the csv procedure is statistical in nature , the proposed biharmonic distance - based voting scheme ensures its robustness to noise .
in particular , figure [ fig : noise ] demonstrates the robustness of the proposed symmetry detection technique for different levels of synthetic gaussian noise added to the _ human _ shape model .
all of the experiments reported in this paper were performed on an intel core@xmath60 3.4 ghz machine with 8 gb ram .
for all the example models , the number of sample points were in the range [ 20 , 100 ] .
table [ tab : timing ] reports the timing results for the various steps in the proposed symmetry detection algorithm .
in particular , unlike @xcite , wherein the most time consuming step of all - pairs geodesic distance computation is not reported , we also report the timing results for the equivalent step in our formulation i.e. , the all - pairs biharmonic distance computation .
the most time consuming step in the proposed algorithm , i.e. , the all - pairs biharmonic distance computation , accounts for around 80% of the execution time of the proposed algorithm .
more importantly , bypassing the two - step spectral clustering procedure described in @xcite reduces significantly the computation time of the proposed algorithm .
[ tab : timing ] we have compared the proposed symmetry detection algorithm primarily with methods that could be deemed sufficiently similar @xcite , @xcite . the symmetry detection technique of xu et al .
@xcite can detect overlapping partial intrinsic symmetries whereas that of lipman et al .
@xcite is designed to deal with partial extrinsic symmetries . however , if the symmetric sub - shapes do not undergo significant pose variations , the global alignment component of @xcite may allow it to detect certain partial intrinsic symmetries as well . however , whereas both methods @xcite , @xcite are capable of detecting instances of partial intrinsic symmetry , neither is able to characterize the underlying symmetry .
in contrast , the proposed algorithm , not only detects overlapping intrinsic symmetries , but it also has the ability to characterize and cluster the detected symmetries in symmetry space . since their entire formulation is based solely on global intrinsic distance - based voting , the technique of xu et al .
@xcite suffers from the shortcoming of not being able to reliably detect symmetry flips .
the symmetry flip phenomenon is deemed to occur when for the same symmetric transform , the point correspondences interchange their relative positions due to ill - posed symmetry detection criteria .
the proposed algorithm , on the other hand , interpolates the functional map of symmetry transformations from the chosen directed point pair , to the remaining point pairs , i.e. , once the source and destination within a point pair are identified , the remaining correspondences are obtained via interpolation using the functional map .
the functional map ensures that the interpolated point pairs observe the same direction of symmetry as the initial directed pair , i.e. , the functional map preserves the relative positions of points within a point pair that are subject to the same symmetry transformation .
the functional map thus provides a natural solution to the symmetry flip problem as evidenced in figure [ fig : flip ] .
the symmetry - factored embedding ( sfe ) technique of lipman et al .
@xcite , though designed for extrinsic symmetry detection , in simpler cases , is able to detect intrinsic symmetries as well . however , it fails completely in cases of overlapping symmetries whereas the proposed csv procedure ensures the detection of instances of overlapping intrinsic symmetry as shown in figure [ fig : sfe ] . in this section
, we present quantitative evaluation of the results obtained by the proposed approach . in order to evaluate the proposed algorithm , we used the shrec 2010 feature detection and description benchmark @xcite .
this dataset comprises of three shapes ( null shapes ) and a set of shapes obtained by applying a set of transformations on the null shapes .
shapes have approximately 10,000 to 50,000 vertices and their surfaces are represented by triangular meshes .
we have specifically considered shapes which have undergone changes characterized by isometry , topology , micro - holes , scale and noise . for each transformation , in the initial phase
, a total 50 sample points are generated using the farthest random sampling strategy as shown in figure [ fig : quant ] .
almost all transformations maintain a high repeatability ( greater than 80% ) at overlap values @xmath61 except for changes in topology .
the diagonality property of the functional map could be used to characterize the underlying symmetry transformation and classify it as a simple transformation or as one that is more complex in nature .
our assumption is that , greater the complexity of symmetry transformation , greater the deviation of the shape deformation from intrinsic isometry , resulting in a deformation characterized by a functional matrix @xmath53 with higher off - diagonal element values .
the resulting characteriztion of the isometric deformation is depicted in figure [ fig : charac ] . as discussed in @xcite
, symmetry has a group structure .
consequently , the retrieval of symmetry transformation can be cast as a clustering problem from an algorithmic perspective .
we exploit the functional maps generated previously to cluster the detected instances of symmetry in transformation space and retrieve the symmetry groups . due to the symmetric flip problem inherent in the csv procedure
, we adjusted the relative positions of source and destination points within each point pair by finding the minimum distance between the source and destination points in the biharmonic distance space .
this ensures that the functional maps generated from potentially similar transformations will have similar structure since the symmetry flip problem between the maps is resolved .
in particular , we have used a simple @xmath25-means clustering algorithm to cluster the functional maps based on their symmetry groups as depicted in figure [ fig : clustering ] .
we have presented an algorithm for detection and characterization of intrinsic symmetry in 3d shapes .
while the results obtained are encouraging , we regard our work as an initial attempt towards complete understanding of and a verifiable solution to the general problem of symmetry detection and characterization . we have identified some limitations of our approach and we hope to address these in our future work . to the best of our knowledge ,
this is one of the first attempts to formalize the symmetry analysis problem not only as one of symmetry detection , but as one that can be extended to include symmetry characterization and symmetry clustering in the transformation space .
in particular , the introduction of the functional map formalism in symmetry detection enables us to come up with a novel representation of the symmetry transformation as a map . in future , we aim to formulate operations , such as addition and subtraction , on these generated maps that would potentially provide a deeper and more comprehensive understanding of intrinsic symmetry in general . the incorporation of _ transformation space mapping _ in symmetry characterization is a completely new idea and the full potential of it can only be realized after more extensive experimentations .
in particular , we plan to study the possibility of map based exploration of similar symmetric transformations across shapes in near future .
another important direction that can be considered is the possibility of incorporating this technique in computer aided geometric design for urban architecture . in urban architecture ,
symmetric repetition of same pattern is a common thing and during the design phase , if the basic structure is stored only once and the symmetric repetitions are saved as _
functional maps _
, it can possibly solve both the space complexity problem and the design efficiency problem .
bronstein , a. m. , bronstein , m. m. , bustos , b. , castellani , u. , crisani , m. , falcidieno , b. , guibas , l. j. , kokkinos , i. , murino , v. , sipiran , i. , ovsjanikov , m. , patane , g. , spagnuolo , m. , and sun , j. ( 2010 ) shrec 2010 : robust feature detection and description benchmark .
workshop on 3d object retrieval ( 3dor10)_. bronstein , a. m. , bronstein , m. m. , and kimmel , r. ( 2007 ) calculus of non - rigid surfaces for geometry and texture manipulation .
_ ieee trans . visualization and computer graphics _ , vol 13(5 ) , september - october , pp . 902913 .
berner , a. , bokeloh , m. , wand , m. , schilling , a. , and seidel , h .-
( 2008 ) a graph - based approach to symmetry detection .
fifth eurographics / ieee vgtc conf .
point - based graphics _ , eurographics association , pp
. 18 .
li , w. , zhang , a. , and kleeman , l. ( 2005 ) fast global reflectional symmetry detection for robotic grasping and visual tracking .
_ proc . australasian conference on robotics and automation ( acra05 ) _ , december .
podolak , j. , shilane , p. , golovinskiy , a. , rusinkiewicz , s. , and funkhouser , t. ( 2006 ) a planar - reflective symmetry transform for 3d shapes .
_ acm transactions on graphics ( tog ) _
25(3 ) , july , pp . 549559 .
porkass , j. , bronstein , a. m. , bronstein , m. m. , sprechmann , p. , and sapiro , g. ( 2013 ) sparse modeling of intrinsic correspondences .
_ computer graphics forum _ , 32(2.4 ) , blackwell publishing ltd . , pp .
459468 .
wang , y. , xu , k. , li , j. , zhang , h. , shamir , a. , liu , l. , cheng , z. , and xiong , y. ( 2011 ) symmetry hierarchy of man - made objects .
_ computer graphics forum _ , 30(2 ) ,
blackwell publishing ltd . , pp .
287-296 yen , l. , fouss , f. , decaestecker , c. , francq , p. , and saerens , m. ( 2007 ) graph nodes clustering based on the commute - time kernel .
11th pacific - asia conference on knowledge discovery and data mining ( pakdd 2007 ) _ , springer lecture notes in computer science , lncs . | a comprehensive framework for detection and characterization of overlapping intrinsic symmetry over 3d shapes is proposed . to identify prominent symmetric regions which overlap in space and vary in form
, the proposed framework is decoupled into a _ correspondence space voting _ procedure followed by a _ transformation space mapping _ procedure . in the correspondence space
voting procedure , significant symmetries are first detected by identifying surface point pairs on the input shape that exhibit _ local _ similarity in terms of their intrinsic geometry while simultaneously maintaining an intrinsic distance structure at a _ global _ level . since different point pairs can share a common point , the detected symmetric shape regions can potentially overlap . to this end
, a _ global intrinsic distance - based voting _ technique is employed to ensure the inclusion of only those point pairs that exhibit significant symmetry . in the transformation space mapping procedure ,
the _ functional map _ framework is employed to generate the final map of symmetries between point pairs .
the transformation space mapping procedure ensures the retrieval of the underlying dense correspondence map throughout the 3d shape that follows a particular symmetry .
additionally , the formulation of a novel cost matrix enables the inner product to succesfully indicate the complexity of the underlying symmetry transformation .
the proposed transformation space mapping procedure is shown to result in the formulation of a semi - metric symmetry space where each point in the space represents a specific symmetry transformation and the distance between points represents the complexity between the corresponding transformations .
experimental results show that the proposed framework can successfully process complex 3d shapes that possess rich symmetries . |
from a long time it is believed that , due to the van der waals nature of the nucleon - nucleon interaction , nuclear matter is likely to exhibit a liquid - gas phase transition @xcite .
while such a phase transition was predicted within various models , from the experimental point of view so far things are rather inconclusive .
the reasons are related to both the inherent difficulties of backtracking the freeze - out stage and the lack of any rigorous mathematical apparatus permitting the unambiguous determination of a liquid - gas phase transition when it exists in experimental data .
some years ago , the general conviction was that the experimental measurement of the nuclear caloric curve could give a definite answer to this problem . neglecting the weaknesses of each of the `` thermometers '' tested with that occasion ,
there is one fundamental barrier preventing the inference of a first order phase transition from any experimentally obtained caloric curve even with back - bending shape : the experimentally obtained microcanonical systems follow an _ unknown _ path in the excitation energy ( @xmath0)- freeze - out volume ( @xmath1 ) plane @xcite . _
even _ if the experimental path intersects the coexistence region the phase transition may remain unrevealed in the caloric curve @xcite . for overcoming that situation
it was proposed that the heat capacity being a quantity _ independent _ on the experimental path but depending on the _ local _ freeze - out constrain could give precise information concerning the occurrence of the transition @xcite .
more exactly , _ if _ the experimental path is intersecting the spinodal region of the system s phase diagram , this would be signaled by a negative value of the heat capacity .
an analytical heat capacity formula was proposed with that occasion and was subsequently employed in investigating the experimental data obtained by the multics - miniball collaboration occasion with which negative branches of the heat capacity curves were identified @xcite .
the formula was further applied on the indra data with the same success @xcite .
however , it is worth noticing that though formally correctly deduced , this formula is based on unevaluated terms ( such as the microcanonical temperature ( @xmath2 ) or the heat capacity ( @xmath3 ) corresponding to a partial energy of the system ( @xmath4 ) ) which further imply a chain of assumptions and approximations is deduced from the @xmath5 dependence which results in even larger deviations . ] resulting in ( unpredicted ) deviations from the exact results .
the above matters motivated us to draw the attention in the present paper on an alternate way to calculate the microcanonical heat capacity of nuclear systems proposed in ref .
as will be further shown this method is _ exact _ , model independent , simply applicable and does not depend on any unevaluated term .
the present paper is structured as follows : section 2 gives a brief review of the model used for illustrating the paper s ideas .
section 3 investigates on the possible signals one can ( experimentally ) get about a possible liquid - gas phase transition .
microcanonical formulas for temperature , heat capacity and second - order derivative of the entropy versus energy are deduced for various conservation options in section 4 .
examples for the functioning of the above - mentioned formulas are given in section 5 .
the same quantities are deduced in section 6 for the case in which only a percent of the total number of fragments are available per event .
conclusions are drawn in section 7 .
for testing the paper s argumentation we use the sharp microcanonical multifragmentation model proposed in ref .
the model concerns the disassembly of a statistically equilibrated nuclear source @xmath6 ( i.e. the source is defined by the parameters : mass number , atomic number , excitation energy , and freeze - out volume respectively ) ; its basic hypothesis is equal probability between all configurations @xmath7 ( the mass number , the atomic number , the excitation energy , the position and the momentum of each fragment @xmath8 of the configuration @xmath9 , composed of @xmath10 fragments ) which are subject to standard microcanonical constraints : @xmath11 , @xmath12 , @xmath13 , @xmath14 , @xmath0 - constant ; integration over the fragments momenta can be analytically performed in the expression of the total number of states , then one works in the smaller configuration space : @xmath15 ; a metropolis - type simulation is employed for determining the average value of any system observable .
function of the desired conservation restriction ( @xmath16 ) level ( @xmath17 ) one can write : @xmath18 with @xmath19 ; @xmath20 , @xmath21 , @xmath22 ; @xmath23 and @xmath24 , where @xmath25 is the inertial tensor of the system . here
@xmath26 denotes the system s hamiltonian : @xmath27 and @xmath28 ( @xmath29 is the coulomb interaction between fragments @xmath8 and @xmath30 ; @xmath31 and @xmath32 are respectively the internal excitation and the binding energy of fragment @xmath8 ) . in the standard version of the model fragments
are idealized as nonoverlapping hard spheres placed into a spherical recipient ; intersection between fragments and the `` recipient s '' wall is also forbidden .
we call this freeze - out hypothesis ( i ) . in a simplified version ,
( ii ) , one can further perform an approximate integration over the position variables as in @xcite and get the extra factor : @xmath33 in the expression of the statistical weight of a configuration @xmath34 ( i.e. @xmath35 ) , where @xmath36 stands for the volume of the nuclear system at normal nuclear matter density . in this version of the model
the wigner seitz approximation is used for the interfragment coulomb interaction .
when angular momentum conservation is also considered , @xmath37 is approximated by the quantity corresponding of a sphere of volume @xmath1 uniformly filled with nuclear matter of density @xmath38 .
the latter version is computationally faster and provides a different freeze - out perspective ( i.e. spherical hard - core interaction is missing ) .
which are the echos one can experimentally get from a liquid - gas phase transition ? for answering this question one should first define the liquid - gas phase transition in _
_ extensive _ or _ nonextensive _ systems .
this matter was recently addressed in @xcite .
there , it was shown that the techniques usually employed for very large systems work for the case of finite extensive or nonextensive systems as well .
namely , it was shown that curves like @xmath39 ( here @xmath40 is an extensive variable , @xmath41 is the @xmath40 s conjugate and @xmath42 is another intensive variable ) have the property of revealing the phase transition ( even in small and nonextensive systems ) by bending backwards .
the microcanonical isobaric caloric curves or isothermal pressure versus volume curves are natural examples for the physical situation under study .
the probability distributions of an isobaric canonical ensemble proved to be a precious tool for identifying the phase transition when it exists ( by exhibiting a double peaked structure ) and for constructing the corresponding @xmath43 and @xmath44 curves .
reasons lay in the simple expression of the probability of a state with excitation energy @xmath0 and volume @xmath1 in a constant pressure canonical ensemble characterized by the parameters @xmath45 ( canonical pressure ) and @xmath46 ( inverse of the canonical temperature ) : @xmath47 leading to the expression of the microcanonical entropy ( @xmath48 ) : @xmath49 where @xmath50 represents the constant pressure canonical partition function . starting from their definitions ( @xmath51 , @xmath52 )
, one can easily obtain the microcanonical temperature and pressure : @xmath53 one can then follow the method proposed in @xcite and follow a @xmath54 or a @xmath55 path in the @xmath56 plane by using the @xmath57 pairs evaluated ( by means of eq .
( [ eq : tp ] ) ) in each @xmath56 point with sufficiently collected events from the probability distribution of a canonical ensemble at constant pressure ( one can simulate this ensemble within the present model by multiplying the statistical weight of a configuration by @xmath58 and letting both energy and freeze - out volume to fluctuate ) . the corresponding isobaric and isothermal paths are plotted in fig . 1 ( upper - right panel ) for a system with @xmath59 , @xmath60 for which coulomb interaction is switched to zero
. the corresponding constant pressure canonical probability distribution is represented by a contour plot ( in log scale ) in the same picture .
the double peaked structure points the presence of a first order phase transition .
both isothermal and isobaric paths are intersecting the two peaks of the distribution and also a third saddle - like point . the corresponding @xmath43 and @xmath44 curves are plotted in the lateral panels with thick lines .
they both present a backbending region indicating the phase transition . and
@xmath45 are labeled on the picture .
the thick lines correspond to microcanonical isobaric and isothermal paths corresponding to @xmath54 and @xmath55 .
the lateral panels correspond to projections ( dashed lines ) of the probability distributions on @xmath61 ( left ) and @xmath0 ( down ) axes .
canonical pressure ( left panel ) and constant-@xmath62 temperature are figured with thin lines .
the corresponding microcanonical curves are plotted with thick lines .
( here and in the rest of the figures @xmath0 stands for excitation energy . ) ,
height=377 ] while such constructions can be simply performed in theoretical models , experimentally one can usually only access microcanonical ensembles ( due to the sorting of events in bins of @xmath0 ) .
moreover , it is widely believed that the events corresponding to a microcanonical ensemble at a given @xmath0 have a fluctuating volume at freeze - out - somewhat similar to the case of a volume constraint @xmath63 proposed in ref .
@xcite ( where @xmath62 is the lagrange multiplier corresponding to the volume observable ) . as clearly resulting from fig .
1 , the constant @xmath62 microcanonical ensemble corresponds to the folding on the @xmath1 axis of the constant pressure canonical ensemble . indeed , replacing in eq .
( [ eq : p(e , v ) ] ) @xmath64 with @xmath62 , and making the integration over @xmath1 one gets : @xmath65 and then the expression of the microcanonical temperature : @xmath66 the dependence of the last two quantities on @xmath0 is represented in fig .
1 ( lower panel ) .
thus , @xmath67 is a projection on the @xmath0 axis of the @xmath68 distribution .
therefore , a double peaked structure in @xmath67 is _ necessarily _ related to a double peaked structure in @xmath69 , and thus , is _ reflecting _ a first order liquid - gas phase transition as defined in refs .
@xcite .
of course , the same reasoning may be done for the projection on the @xmath1 axis .
the corresponding curves are plotted in fig . 1 ( left panel ) .
however , the relevance of this latter projection for the experimental data is conditioned by a good event by event resolution in freeze - out volume which was not achieved so far .
let us further deduce the _ exact _
expressions for microcanonical temperature and heat capacity .
the microcanonical density of states corresponding to a total energy of the system @xmath0 writes : @xmath70 where @xmath71^{\frac32(n - n)-1 } \label{eq : w(c')}\end{aligned}\ ] ] and @xmath72 here @xmath73 is the factor in front of @xmath74 from eq .
( [ eq : int ] ) ( @xmath75 is a generic notation for fragment partition ) and @xmath76 , the expression ( [ eq : sum_c ] ) being formal .
then , eq . ( [ eq : w(e ) ] ) may be rewritten as : @xmath77 the microcanonical temperature writes @xmath78 with @xmath79 . which means : @xmath80 _ if _ the limits of the ( formal ) integral over @xmath81 are _ not _ depending on @xmath0 , then one can simply make the derivative versus @xmath0 _ inside _ the integral from eq .
( [ eq : w(e ) ] ) : @xmath82 where we used the implication : @xmath83 and the notation@xmath84 for the average over the ensemble s states . the heat capacity of the system is by definition : @xmath85 . using the same arguments as those from the deduction of @xmath2 one obtains : @xmath86 which implies : @xmath87 \left[\frac32(n - n)-2\right]}{k^2 } \right>. \label{eq : c}\end{aligned}\ ] ] alternatively , one can ( similarly ) evaluate the second order derivative of the system s entropy versus @xmath0 : @xmath88\left[\frac32(n - n)-2\right]}{k^2 } \right>- \left < \frac{\frac32(n - n)-1}k \right>^2 .
\label{eq : d2sde2}\end{aligned}\ ] ] in these terms , the system s spinodal region will be pointed by positive values of the above quantity .
one can easily check that for @xmath89 ( i.e. energy and total momentum are the conserved quantities ) , eq .
( [ eq : c ] ) is similar with the one deduced in ref .
the microcanonical formulas given by eqs .
( [ eq : t ] ) , ( [ eq : c ] ) and ( [ eq : d2sde2 ] ) are _ universally _ applicable to _ any _ system for which the `` external '' degrees of freedom can be treated classically irrespective to other specificities of the system . indeed , the integral over the momenta variables ( with the various possible conservation options ) given in eq .
( [ eq : int ] ) , from which the microcanonical @xmath2 , @xmath9 and @xmath90 formulas were further deduced is valid for any classical @xmath10 particle system .
this is the case of nuclear multifragmentation as well : it is widely accepted that , since at freeze - out the system is rather dilute , a classical treatment of the fragments motion is appropriate .
these features make eqs .
( [ eq : t ] ) , ( [ eq : c ] ) and ( [ eq : d2sde2 ] ) suitable to be applied for experimental data : they are _ independent _ on specific model assumptions ( i.e. freeze - out hypothesis , internal excitation treatment - level density , interfragment interactions , etc . ) , and moreover , they only depend on two parameters ( @xmath91 and @xmath10 ) which have to be estimated for the freeze - out stage . ) .
each column correspond to the case labeled on it .
, height=415 ]
while the deduction of the above formulas guarantees their accuracy , some examples of their functioning on concrete cases are further discussed . as an independent test
, the results will be compared with the temperature / heat capacity computed using the method synthesized by eq .
( [ eq : tp_lambda ] ) .
more exactly , the caloric test curves are computed using eq .
( [ eq : tp_lambda ] ) from the probability distributions of a canonical ensemble with a given freeze - out constraint and the heat capacity and @xmath90 curves are farther evaluated from the obtained caloric curves ( @xmath92 , @xmath93 ) .
calculations are performed with both versions of the model ( ( i ) and ( ii ) ) , for two different freeze - out constraints : constant @xmath62 and constant volume , for the case in which the coulomb interaction is present and the one in which it is switched off .
( this variety of examples constitute a further test for the universality of the deduced formulas . )
the considered conservation `` level '' is @xmath94 , i.e. the total energy , the momentum and the angular momentum are conserved quantities .
results corresponding to the cases : ( 50,23 ) nucleus with constant @xmath95 @xmath96 with coulomb interaction included , version ( ii ) of the model , ( 200,82 ) , @xmath97 @xmath96 , without coulomb interaction , version ( ii ) of the model and the same nucleus with coulomb interaction included , but a constant volume constraint : @xmath98 , version ( i ) of the model are presented in fig . 2 ( first , middle and third column respectively ) . in the first two cases , backbendings of the caloric curves reflected in negative branches of the heat capacity curves and positive regions in the second order derivative of the entropy versus
@xmath0 - all related with a first order phase transition can be observed . in the third case ( constant volume constraint )
the caloric curve has a monotonic increase with a `` plateau - like '' region ( where the curve slope is smaller , but positive ) which is reflected in a positive @xmath9 ( with a peak in the plateau - like region ) and a negative @xmath90 .
( as will be further seen , _ one can not _ draw any conclusion from the third case concerning the intersection of the coexistence region since a constant volume path is _ not necessarily _ the order parameter of the system . ) a very good agreement between the curves calculated using the probability distribution of a canonical ensemble and the points evaluated by means of the microcanonical formulas is to be noticed . the calculated microcanonical heat capacity and second derivative of the entropy versus excitation energy points are clearly revealing the spinodal region .
while the calculated heat capacity points have large error bars near the border of the spinodal region due to the asymptotic behavior of the heat capacity , @xmath90 has small error bars even in the border region - thus being an even more robust quantity for identifying the spinodal region
. with the coulomb interaction included , for three excitation energies ( 0.5 , 1 and 2 mev / u ) .
, height=340 ] some details concerning the applicability of the microcanonical formulas have to mentioned .
as mentioned earlier , the validity of these formulas is restricted by the condition that the integration domains in eq .
( [ eq : w(e ) ] ) over the quantity @xmath81 _ not to be restricted _ by the total energy of the system @xmath0 .
one should thus limit the cases on which this formula is applied to those obeying the above condition . but how to distinguish those `` forbidden '' points from the rest ?
a restriction by @xmath0 of the upper limit of the integral over @xmath81 has to be clearly pointed by a `` cut '' in the right part of @xmath81 distribution . since @xmath83 , a similar `` cut '' ( corresponding to @xmath99 ) has to be noticed in the left part of the @xmath91 probability distribution . in other words , when the probability that @xmath99 is significant ( i.e. a cut clearly affects the @xmath91 distribution ) then deviations from the microcanonical formulas should occur .
these `` cuts '' ( usually ) occur at very low energies , so in a region unimportant for the system s thermodynamics .
all the points represented in fig .
2 have been checked against the cut in the @xmath91 probability distribution . for illustrating this idea we take the case ( 200,82 ) , with coulomb , @xmath98 .
the probability distributions of @xmath91 corresponding to the @xmath1000.5 , 1 and 2 mev / u cases are illustrated in fig .
3 . while the @xmath91 probability distributions corresponding to @xmath1000.5 and 1 mev / u are intersecting the @xmath99 axis , ( giving thus a non - negligible probability for @xmath99 ) , the probability distribution corresponding to the @xmath101 mev / u case is not intersecting the @xmath99 vertical axis .
this fixes the lowest excitation energy limit for which the microcanonical formula works to some value between 1 and 2 mev / u . given by : @xmath102 . ,
height=453 ] the crossing of the spinodal region by a ( experimental ) path in the ( e , v ) plane _ may not _ be reflected as a backbending in a caloric curve .
however , if the volume constraint is of the @xmath63 type the spinodal region _ is likely _ to be revealed by a negative branch in the heat capacity .
this point was nicely argued in ref .
@xcite and can be evidenced by means of the present microcanonical formulas as a further test for their abilities . to this aim
, we chose @xmath62 as the microcanonical `` @xmath62 '' parameter ( @xmath103 ) corresponding to a constant volume path ( @xmath98 ) in the ( e , v ) plane .
i.e. : @xmath104 the microcanonical @xmath62 can be simply obtained from eq .
( [ eq : tp ] ) : @xmath105 as illustrated in fig .
4 this path appears to intersect the system s spinodal region without inducing any backbending in the caloric curve for the case of the system ( 200,82 ) without coulomb interaction , ( ii ) version of the model .
indeed , while the caloric curve has a monotonical increase , the corresponding heat capacity presents a negative branch and the second order derivative of the entropy a positive region .
the negative region in the heat capacity and the positive region in the second order derivative of the entropy versus energy are evidencing a `` spinodal line '' of the system corresponding to the volume @xmath98 in the system s phase diagram in @xmath56 representation .
the limits @xmath0 of this region are _ approximately _ those delimited by the above mentioned signals .
let s now assume that experimentally only a ( small ) number @xmath106 ( @xmath107 ) of fragments corresponding to a given event are detected .
can one still get the desired thermodynamical information about the system ?
this question is addressed in the present section .
let us turn back to eq .
( [ eq : int ] ) . for the highest conservation level ( @xmath94 ) we can write : @xmath108 where @xmath109 , @xmath110 and @xmath111 .
the integration over the momenta of the first @xmath106 fragments can be further performed : @xmath112 where @xmath113 is the inertial tensor corresponding to the first @xmath106 fragments at freeze - out , @xmath114 and @xmath115 stands for the step function .
one can further deduce the thermodynamical quantities obtained before similarly , starting from their statistical definitions . in the general case ( i.e a generic `` level '' of conservation @xmath17 ) one obtains : @xmath116\left<\frac1{k_2}\right>,\end{aligned}\ ] ]
@xmath117\left[\frac32(n_1-n)-2\right ] \left < \frac1{k_2 ^ 2 } \right>,\end{aligned}\ ] ] @xmath118\left[\frac32(n_1-n)-2\right ] \left < \frac1{k_2 ^ 2 } \right > - \left [ \frac32(n_1-n)-1 \right]^2 \left < \frac1{k_2 } \right>^2,\end{aligned}\ ] ] where @xmath119 with @xmath120 , @xmath121 and @xmath122 .
following the same argumentation as in section 4 one can deduce the criterion of validity of the above formulas : the probability distribution of @xmath123 must _ not _ intersect the @xmath124 vertical axis at a non - negligible value .
the advantages of the above formulas are obvious : they only depend on one parameter ( @xmath123 ) and , more importantly , they can be used for inferring information about the liquid - gas phase transition even when one has _
incomplete _ information about the fragmentation events ( i.e. only @xmath106 fragments from the @xmath10 fragments of one fragmentation event are detected ) . for a given microcanonical formula , depending on the desired conservation level ( @xmath17 ) ,
the minimum allowed value of @xmath106 is fixed by the condition that the factors containing @xmath106 be positive .
thus , for example , when only the total energy , @xmath0 , is conserved ( @xmath125 ) one only needs _
one _ freeze - out fragment per event in order to obtain the correct microcanonical temperature !
in summary , the possibilities of obtaining information about a possible liquid - gas phase transition from experimental heavy ion collision data have been investigated .
the connection between negative heat capacity ( or positive @xmath126 ) obtained in a fluctuating volume ensemble and the liquid gas phase transition has been discussed .
it was shown that a negative value of the heat capacity in a fluctuating volume ( constant @xmath62 ) ensemble is in connection with a first order phase transition as defined in ref .
microcanonical formulas for temperature , heat capacity and second order derivative of the entropy versus energy have been rigorously deduced for various conservation options ( conserved energy ; conserved energy and momentum ; conserved energy , momentum and angular momentum ) .
these formulas were tested with an independent method for obtaining the above mentioned quantities which extracts the microcanonical information from the probability distributions of a canonical ensemble with a given freeze - out volume constraint .
an excellent agreement between the two methods have been obtained for all the considered curves ( @xmath127 , @xmath128 , @xmath129 ) for various cases ( ( 50,23 ) , @xmath95 @xmath96 with coulomb , ( 200,82 ) , @xmath97 @xmath96 , without coulomb , ( 200,82 ) , @xmath98 with coulomb ) . in the first two cases , backbendings in the caloric curves reflected in negative branches of the heat capacity curves and positive regions in the @xmath130 curves have been obtained reflecting the occurrence of a first order phase transition . in the third case no backbending in the caloric curve or negative region in the heat capacity curve
was observed which _ however _ does not imply any conclusion about the occurrence of a first order phase transition simply because a constant volume path in the @xmath56 plane is _ not necessarily _ the system s order parameter . indeed , choosing for the nucleus ( 200,82 ) without coulomb a dependence of @xmath62 on @xmath0 corresponding to the microcanonical lambda with @xmath98 ( @xmath131 ) in an ensemble with fluctuating volume
one gets a well defined negative branch in the heat capacity curve and a positive region of the second order derivative of the entropy _ in spite of _ a monotonically increasing caloric curve .
the case of incomplete ( experimental ) information concerning the fragmentation event is further discussed .
microcanonical formulas corresponding to all conservation levels depending only on a given number of fragments ( smaller then the total number of fragments from the given event ) are deduced . while demanding some more kinematical information about the detected fragments ( only for the @xmath132 cases ) their utility is obvious .
thus , for example in the @xmath125 case one only needs the kinetic energy of _ one _ freeze - out fragment per event in order to deduce the correct microcanonical temperature .
the resulted microcanonical formulas are model independent thus being applicable on experimental data .
moreover they do not depend on any unevaluated term ( such as heat capacity of a partial energy or the level density parameters as is the case in ref.@xcite ) .
of course , the success of these formulas depends on the accuracy of backtracking the information about the primary fragments . here the microcanonical formulas based on partial break - up information may be very effective since some particles ( such as freeze - out neutrons ) carry at the asymptotic stage the untouched information about the freeze - out . | a general discussion is made concerning the ways in which one can get signatures about a possible liquid - gas phase transition in nuclear matter .
microcanonical temperature , heat capacity and second order derivative of the entropy versus energy formulas have been deduced in a general case .
these formulas are _ exact _ , simply applicable and do not depend on any model assumption .
therefore , they are suitable to be applied on experimental data . the formulas are tested in various situations .
it is evidenced that when the freeze - out constraint is of fluctuating volume type the deduced ( heat capacity and second order derivative of the entropy versus energy ) formulas will prompt the spinodal region through specific signals .
finally , the same microcanonical formulas are deduced for the case when an incomplete number of fragments per event are available .
these formulas could overcome the freeze - out backtracking deficiencies .
, multifragmentation , temperature , heat capacity , phase transition 24.10.pa , 25.70.pq |
nonclassicality of fields has been a topic of great interest in quantum optics and quantum information processing @xcite .
experimentally , the traditional quantum states , such as fock states and coherent states as well as squeezed states , have been generated but there are some limitations in using them for various tasks of quantum information process @xcite . alternately , it is possible to generate and manipulate various nonclassical optical fields by quantum superpositions and subtracting or adding photons from / to traditional quantum states @xcite . on the other hand , the single mode displaced , squeezed , mixed gaussian states have been paid enough attention by both experimentalists and theoreticians .
marian et .
al @xcite investegated the superposition of a squeezed thermal radiation and a coherent one .
they examined the squeezing properties of the field using the distribution functions of the quadratures .
for gaussian squeezed states of light , a scheme is also presented experimentally to measure its squeezing , purity and entanglement @xcite .
as is well known , dissipative quantum channels tend to deteriorate the degree of nonclassicality ( i.e. , render quantum features unobservable ) .
thus , it is usually necessary to investigate the decoherence properties in dissipative channels , such as dynamical behaviors of the partial negativity of wigner function ( wf ) and how long a nonclassical field preserves its partial negativity of wf . for instances ,
the nonclassicality of single photon - added thermal states in the thermal channel is investigated by exploring the volume of the negative part of the wf @xcite ; souza and nemes have derived an upper limit for the mixedness of the single bosonic mode gaussian states @xcite in a thermal channel . in this paper
, we shall introduce the photon - added squeezed thermo state ( pasts ) and investigate its statistical properties , such as mandel s q - parameter , number distribution , the wigner function .
we then study its decoherence in a photon - loss channel in term of the negativity of wf by deriving the analytical expression of wf for pasts .
it is found that the wf with single photon - added is always partial negative for the arbitrary values of @xmath0 and the squeezing parameter @xmath1 .
the work is arranged as follows : in section 2 , we introduce the state pasts and derive its normalized constant , and photon number distribution is discussed in section 3 .
section 4 is devoted to calculating the wf . in the last section ,
we explore the decoherence of pasts in a photon - loss channel by discussing the evolution of wf .
for a displaced - squeezed thermal field , the density operator is @xmath2 where @xmath3 and @xmath4i@xmath5,$ ] are the displacement operator and the squeezing operator @xcite , respectively , @xmath6 @xmath7 = 1,$ ] and@xmath8(@xmath9 is the boltzmann constant , @xmath10 denoting temperature ) , is qualified to be a density operator of thermal ( chaotic ) field , since tr@xmath11 . for a coherent state
@xmath12 @xcite , due to @xmath13 , matrix elements of any normally ordered operators @xmath14 ( the symbol @xmath15denotes normally ordering ) in the coherent state is easily obtained , i.e , @xmath16 so in ref.@xcite by using the weyl ordering invariance under similarity transformations and the technique of integration within an ordered product of operators ( iwop ) fan et al have converted @xmath17 to its normally ordered gaussian form @xmath18 where @xmath19@xmath20and @xmath0 is the average photon number for @xmath21 i.e. @xmath22 @xcite .
the form in eq.(3 ) is similar to the bivariate normal distribution in statistics , which is useful for us to further derive the marginal distributions of @xmath17 .
theoretically , the pasts can be obtained by repeatedly operating the photon creation operator @xmath23 on a displacement squeezed thermal state , so its density operator is defined as @xmath24 where @xmath25 is a non - negative integer , @xmath26tr@xmath27 is the normalization constant . using eq.([3 ] )
we known immediately the normally ordered gaussian form of @xmath28 , i.e. , @xmath29 next we shall determine the normalization constant @xmath30 . using the completness relation of coherent states
@xmath31 as well as eq.(2 ) , we have @xmath32 , \label{7}\end{aligned}\ ] ] where we have set @xmath33due to @xmath34 thus we know @xmath35further using the following integral formula @xcite @xmath36whose convergent condition is re@xmath37 and@xmath38 , eq.([8 ] ) can be rewritten as follows @xmath39 \right\ } , \label{13}\]]which is the normalization constant of pasts for photon - added number @xmath25 . in particular , when @xmath40 leading to @xmath41 eq.([13 ] ) reduces to the following form , @xmath42 especially when @xmath43 then the normalization constants are given by @xmath44 , @xmath45 and @xmath46 , respectively .
to see clearly the photon statistical properties of the pasts , we will examine the mandel s @xmath47-parameter defined as@xmath48which measures the deviation of the variance of the photon number distribution of the field state under consideration from the poissonian distribution of the coherent state .
if @xmath49 we say the field has poissonian photon statistics while for @xmath50(@xmath51 ) we say that the field has super-(sub- ) poissonian photon statistics.from eq.([13 ] ) and @xmath26tr@xmath52 , we can easily calculate @xmath53 and @xmath54thus we obtain the @xmath47-parameter of the pasts @xmath55 it is well known that the negativity of the @xmath56-parameter refers to sub - possonian statistics of the state .
but a state can be nonclassical even though @xmath56 is positive .
this case is true for the present state . from fig.1
, one can clearly see that for the cases of @xmath57 ( fig.1(a))@xmath58 @xmath56 is always positive ; while for @xmath59 ( for instance @xmath60 ) and a given @xmath61 value , @xmath56 becomes positive only when the squeezing parameter @xmath62 is more than a certain threshold value that increases as @xmath25 increases .
in addition , from fig.1(a ) and fig.1(b ) one can see that the threshold value of @xmath1 decreases as @xmath0 increases .
we emphasize that the wf has negative region for all @xmath1 and @xmath0 thus the pasts is nonclassical ( see next section below ) .
in this section , we study photon number distribution of pasts optical field . using the un - normalized coherent state @xmath63\left\vert 0\right\rangle $ ] , leading to @xmath64 @xmath65 ,
it is easy to see that the photon number distribution formula is given by @xmath66 employing the normal ordering form of @xmath28 in eq.([2 ] ) , eq.(16 ) can be put into the following form @xmath67 \right\ } \left\vert _ { \alpha = \alpha ^{\prime } = 0}\right . .
\label{19}\ ] ] further expanding the exponential term @xmath68 $ ] as series and using the generating function of single - variable hermite polynomials , @xmath69we can calculate the photon number distribution ( pnd ) of pasts , i.e. , @xmath70 \right\vert ^{2}. \label{20}\end{aligned}\]]after making the scale transformation and noticing the recurrence relation @xmath71 , we can easily obtain @xmath72 ^{2}}\left\vert h_{n - m - l}\left [ \mathtt{i}b/(2\sqrt{c})% \right ] \right\vert ^{2 } , \label{23}\ ] ] especially when @xmath40 , eq.([23 ] ) reduces to @xmath73 } \frac{n!(1-a)^{\left ( n - m-2j\right ) } \left\vert c\right\vert ^{2j}}{\left ( n - m-2j\right ) !
\left ( j!\right ) ^{2}}}{\frac{\partial ^{m}}{\partial a^{m}}% \left ( a^{2}-4c^{2}\right ) ^{-1/2 } } \notag \\ & = \frac{n!\sigma ^{n - m}p_{n - m}(\frac{1-a}{\sigma } ) } { \left ( n - m\right ) ! % \frac{\partial ^{m}}{\partial a^{m}}\left ( a^{2}-4c^{2}\right ) ^{-1/2 } } , \label{24}\end{aligned}\ ] ] where@xmath74and in the last step of ( [ 24 ] ) we have used the new expression of legendre polynomials @xcite@xmath75}\frac{m!(1-\frac{1}{x^{2}})^{l}}{% 2^{2l}(l!)^{2}\left ( m-2l\right ) ! } .\]]eq . ( [ 23 ] ) or ( [ 24 ] ) is just the the analytical expression of the pnd of pasts . in particular , when @xmath76 eq.([24 ] ) becomes ( with @xmath77 ) @xmath78 eq.([e26 ] ) is just the pnd of the squeezed thermo state which seems a new result .
the pnds of pasts for some given parameters ( @xmath79 ) and @xmath25 are plotted in fig.2 . from fig .
2 it is found that the pnd is constrained by @xmath80 . by adding photons , we have been able to move the peak from zero photons to nonzero photons ( see fig.2 ( a)-(c ) ) .
the position of peak depends on how many photons are created and how much the state is squeezed initially .
in addition , comparing fig.2(b ) and fig.2(d ) we see that , for a given @xmath25 , the tail `` of pnd becomes more wide '' with the increasing parameter @xmath1 .
the wigner function ( wf ) @xcite was first introduced by wigner in 1932 to calculate quantum correction to a classic distribution function of a quantum - mechanical system .
it now becomes a very popular tool to study the nonclassical properties of quantum states .
it is well known that wfs are quasiprobability distributions because it may be negative in phase space @xcite .
nevertheless , the partial negativity of the wf is indeed a good indication of the highly nonclassical character of the state .
thus , to study the dynamical behaviors of the partial negativity of wf and understand that a nonclassical field preserves its partial negativity , wigner distribution may be very desirable for experimentally quantifying the variation of nonclassicality @xcite .
the presence of negativity of the wf for an optical field is a signature of its nonclassicality . in this section , using the normally ordered form of pasts , we evaluate its wf . for a single - mode system ,
the wf in the coherent state representation @xmath81 is given by @xcite @xmath82 where @xmath83 .
then substituting eq .
( [ 6 ] ) into eq .
( [ 25 ] ) and using eq .
( [ 2 ] ) , we derive the wf of pasts@xmath84 \right\ } , \label{26}\]]where we have set @xmath85and used eq.([11 ] ) .
especially when @xmath40 , eq.([26 ] ) reduces to @xmath86 \right\ } } { \frac{\partial ^{m}}{\partial a^{m}}\left ( a^{2}-4c^{2}\right ) ^{-1/2 } } , \label{27}\]]and further when @xmath87 eq.([27 ] ) becomes to@xmath88which is the wf of the squeezed thermo state , and @xmath89respectively , where we have set @xmath90 eq.([28 ] ) just agrees with the result of eq.(48 ) in ref .
@xcite , whose form is normal distribution . from eq.([27 ] ) one can see that the wf of the pasts is always real , as expected . when the factor @xmath91 in eq.([29 ] ) , the wf of the pasts with @xmath60 has its negative distribution in phase space .
noticing @xmath92 always positive , this indicates that the wf of the pasts always has the negative values under the condition @xmath93 ( i.e. , @xmath94 ) at the phase space center @xmath95 .
in fact , by substituting eqs .
( [ 4 ] ) , ( [ 30a])-([30b ] ) into @xmath96 , we find that for the arbitrary values of @xmath0 and @xmath1 , the wf with @xmath60 is always partial negative .
using eq.([27 ] ) , the wfs of the pasts are dipicted in phase space for several different values of @xmath97 and @xmath1 in phase space .
3 exhibits the wfs of the pasts in phase space with @xmath60 for different @xmath98 , @xmath1 .
it is easy to see that the wfs of single - pasts always have the negative region .
the minimum in the negative region becomes larger with the increasing of @xmath0 ( see fig3.(a ) and ( c ) ) . in fig .
4 , we have presented the wfs with @xmath99 , @xmath100 for different @xmath101 which indicates that the peak ( absolute ) value of wf become smaller as the increasing parameter @xmath102 .
the partial negativity of wf indicates the nonclassical nature of the pasts field .
when the pasts evolves in the amplitude decay channel , the evolution of the density matrix can be described by the following master equation in the interaction picture @xcite , @xmath103where @xmath104 represents the rate of decay . by using the thermal field dynamics theory and thermal entangled state representation ,
the time evolution of wf at time @xmath105 to be given by the following form @xcite , i.e.,@xmath106eq.([33 ] ) is just the evolution formula of wf of single mode quantum state in photon - loss channel . by observing eq.([33 ] ) , we see that when @xmath107 @xmath108 so @xmath109 as expected .
thus the wf at any time can be obtained by performing the integration when the initial wf is known . for simplicity
, here we only discuss the special case @xmath40 . substituting eq.([27 ] ) into eq.([33 ] ) and using eq.([11 ] ) , we derive the time evolution of wf for pasts in photon - loss channel : @xmath110 where we have set@xmath111 ^{2}-64c^{2}\allowbreak \allowbreak } , \notag \\ % tcimacro{\u{211d } } % % beginexpansion \mathbb{r } % endexpansion & = \frac{4% % tcimacro{\u{2115 } } % % beginexpansion \mathbb{n } % endexpansion e^{-2\kappa t}}{\mathcal{t}^{2}}\left [ 4f+\left ( \frac{2e^{-2\kappa t}}{% \mathcal{t}}-2\right ) \left ( f^{2}-4c^{2}\right ) \right ] -\frac{2}{\mathcal{t% } } , \label{36 } \\
\bbbk & = \frac{16% % tcimacro{\u{2115 } } % % beginexpansion \mathbb{n } % endexpansion ce^{-2\kappa t}}{\mathcal{t}^{2}}. \notag\end{aligned}\]]in particular , when @xmath57 eq.([35 ] ) becomes @xmath112which is just the wf of the squeezed thermo state in photon - loss channel .
this result can also be checked by substituting eq.([28 ] ) into eq.(33 ) .
when @xmath113 exceeds a threshold value , the wf has no chance to be negative in the whole phase space . at long time
@xmath114leading to @xmath115 @xmath116 @xmath117 the wf in eq.([35 ] ) becomes @xmath118 which corresponds to the gaussian state . in fig.5 ,
the wfs of pasts are depicted in phase space with @xmath60 and @xmath100 for several different @xmath119 .
it is easily seen that the negative region of wf gradually disappears as @xmath113 increases .
this implies that the system state reduces to a gaussian state after a long time interaction in the channel .
thus the loss of channel causes the absence of the partial negativity of the wf if the decay time @xmath113 exceeds a threshold value . in figs .
6 , we have also presented the time - evolution of wf for different @xmath1 .
one can see clearly that the partial negativity of wf decreases gradually as @xmath1 increases .
the squeezing effect in one of the quadratures can be seen in fig.6 . in eq.([35 ] ) , for the pasts we have obtained the expression of the time evolution of wf . in principle , by differentiating as shown in eq.([35 ] ) we can derive the wf of other pasts ( @xmath120 ) .
but its form is too complicated .
however , we can draw the wigner distributions of pasts for @xmath121 by numerical simulation , as shown in fig.7 ( a)-(c ) , respectively , from which one can see that the absolute value of the negative minimum of the wf decreases as @xmath25 increases .
based on the normally ordered gaussian form of displaced - squeezed thermal field , we have introduced a kind state : the photon - added squeezed thermo state ( pasts ) .
then we have investigated the statistical properties of pasts ( such as mandel s q - parameter , number distribution , the wigner function ) and its decoherence in photon - loss channel with dissipative coefficient @xmath104 in term of the negativity of wf by deriving the analytical expression of wf for pasts .
it is found that the photon number distribution is just a legendre polynomial and that the wf with single photon - added is always partial negative for the arbitrary values of @xmath0 and the squeezing parameter @xmath1 .
the technique of integration within an ordered product of operators brings convenience in our derivation . | using the normally ordered gaussian form of displaced - squeezed thermal field characteristic of average photon number @xmath0 , we introduce the photon - added squeezed thermo state ( pasts ) and investigate its statistical properties , such as mandel s q - parameter , number distribution ( as a legendre polynomial ) , the wigner function .
we then study its decoherence in a photon - loss channel in term of the negativity of wf by deriving the analytical expression of wf for pasts .
it is found that the wf with single photon - added is always partial negative for the arbitrary values of @xmath0 and the squeezing parameter @xmath1 .
pacs : 03.65.yz , 42.50.dv , 03.67.-a , 03.65.wj keywords : open quantum systems ; decoherence ; photon - added squeezed thermal states ; photon - loss channel ; continuous variable systems ; iwop technique |
the last decade has witnessed the explosive growth of high - quality data on type ia supernovae ( sne ia ) both from the ground and from space with spectacular results and new perspectives .
one of the main advances was the confirmation of the long suspected correlation between the peak brightness and the decline rate ( phillips , 1993 ) which clearly established that sne ia span a wide range in brightness but , nevertheless , most are found in a narrow range of brightness ( branch & tammann 1992 ) .
this established sne ia as a tool for accurate distance determinations on cosmological scales .
sne ia have provided new estimates for the value of the hubble constant ( h@xmath7 ) based on a purely empirical procedure ( hamuy et al . 1996ab ; riess et al .
1996 ) and on a comparison of detailed theoretical models with observations ( hflich & khokhlov 1996 , hereafter hk96 ; nugent et al .
the values obtained ( @xmath10 ) are in good agreement with one another and with those from the hst key project on the hubble constant based on cepheid distances . more recently , routinely successful detections of supernovae at large redshifts ( perlmutter et al . 1997 , 1999 ; schmidt et al .
1998 ; riess et al .
1999 ) have provided a new tool to measure cosmological parameters such as @xmath11 and @xmath12 . this has led to refined estimates of h@xmath7 , indicated that we live in a low matter - density universe , and , most intriguing of all , gives hints that the cosmological constant is not zero .
although details of the explosion are still under discussion , there is general agreement that sne ia result from some process of combustion of a degenerate white dwarf ( hoyle & fowler 1960 ) . within this general picture ,
three classes of models have been considered : ( 1 ) explosion of a co white dwarf ( wd ) , with mass close to the chandrasekhar mass , which accretes mass through roche - lobe overflow from an evolved companion star ( nomoto & sugimoto 1977 ) ; ( 2 ) explosion of a rotating configuration formed from the merging of two low - mass wds , caused by the loss of angular momentum due to gravitational radiation ( webbink 1984 ; iben & tutukov 1984 ; paczyski 1985 ) ; and ( 3 ) explosion of a low mass co - wd triggered by the detonation of an outer helium layer ( nomoto 1980 ; woosley & weaver 1986 ; hk96 ) . only the first two scenarios seem to be in agreement with current observations .
delayed - detonation models ( khokhlov 1991 ; yamaoka et al .
1992 ; woosley & weaver 1994 ) of chandrasekhar mass co - wds and their variations can account for the spectral and light curve evolution of normally bright " and subluminous sne ia in the optical and ir ( hflich 1995a ; hflich et al .
1995a , hereafter hkw95 ; wheeler et al . 1998 ; lenz et al .
1999 ; hflich et al .
2000 ) . in this model ,
a burning front starts as a subsonic deflagration and then undergoes a transition to a supersonic detonation .
pure deflagration models like w7 ( nomoto et al .
1984 ) can also account for optical spectra of normal supernovae ( e.g. nugent et al . 1997 ) .
the merging scenario remains an interesting alternative which to a great extent has only been parameterized in theoretical studies but may produce results in agreement with some supernovae ( hk96 ) .
the sub - chandrasekhar model triggered by edge - lit helium ignition is disfavored on the basis of predicted light curves and spectra ( hflich et al .
1996 ; nugent et al .
1997 ) . despite the success of analyses of light curves and spectra
, these traditional methods do not provide any direct information on the geometrical structure of the envelope .
polarization does provide information about the geometry in the scattering - dominated atmospheres of supernovae ( van den hulst 1957 ; chandrasekhar 1960 ; shapiro & sutherland 1982 ; hflich 1991 , hereafter h91 ; jeffery 1991 , hereafter j91 ) .
this is because electron scattering causes the electric field vectors of scattered photons to take on a preferential orientation . for an unresolved spherical scattering atmosphere ,
the electric vectors cancel , yielding zero net polarization ; however , in an aspherical sn , there will be increased scattering from the wings which produces electric field vectors that are not completely cancelled by scattering from the rest of the disk of the sn . in this case
the observer would measure a net polarization .
polarization can thus provide unique information on the geometry of sne ia that will give essential hints to understand the physics of thermonuclear explosions and the progenitor systems . though historically the vast majority of computational models have assumed spherical symmetry , for some of the scenarios we may expect strong deviations from spherical geometry .
for instance , the deflagration front has been found to be highly aspherical due to large rising plumes ( khokhlov 1995 , 2000 ; niemeyer & hillebrandt 1995 ) .
for delayed detonations , the transition of the deflagration front into a detonation may occur at a point rather than in a shell ( livne 1999 ) .
the coalescence of two white dwarfs or an otherwise rapidly rotating progenitor could also produce an overall , global asphericity .
deviations from sphericity may result in a direction dependence of the emitted flux , and consequently may affect the use of sne ia as distance indicators . in this work , we report polarization measurement and interpretation of the subluminous sn 1999by .
the observations are presented in 2 and the results in 3 . in
4 we describe an empirical method to analyse the data and to derive the interstellar polarization . in 5 , theoretical models for delayed detonations are used to analyze the observations .
we close in 6 with a discussion of the results for sn 1999by , put our findings into context with other sne ia and different explosion scenarios , and discuss the implications for the use of sne ia as standard candles .
all observations were taken with the dual beam imaging grism polarimeter ( igp ) at the cassegrain focus of the mcdonald observatory 2.1 m telescope .
igp is the imaging grism instrument ( igi ) combined with spectropolarimetry optics ( goodrich 1991 ) .
igi is a simple focal reducer with a grism that can be moved into the collimated beam .
off - the - shelf optics are used which have been anti - reflection coated for @xmath13 .
polarization capability is provided by a modified glan - taylor polarizing beam splitter and a rotatable waveplate .
the glan - taylor prism is made up of two calcite blocks .
the ordinary ray ( o - ray ) is totally internally reflected , while the extraordinary ray ( e - ray ) passes through , due to different indices of refraction .
the waveplate acts as a halfwave retarder and rotates the plane of polarization of incoming light .
the halfwave plate is `` superachromatic '' with a retardance of 180 @xmath14 2 from 3200 to 11000 .
the intrinsic polarization of igp has been previously determined to be low , @xmath15 , see wang , wheeler , & ( 1997a ; hereafter wwh97 ) and wang et al .
( 2000a ) .
this was confirmed with the observation of null polarization standards .
more extensive information about the data reduction techniques , standard stars , and observational setup is available in howell ( 2000a ) .
a 2.1 slit was used with the 6000 grism , giving a resolution of about 14 . for all observations ,
the 85 mm lens was used in conjunction with the tk4 @xmath16 ccd , yielding a plate scale of 0.5 ( @xmath17 3.9 ) per pixel .
the spectra were wavelength calibrated with an argon lamp .
polarization standards were observed at least once per night . the observing log is shown in table [ log99by ] .
the data were reduced in iraf , using tasks written by the authors .
the reduction methods followed were those presented in miller ( 1988 ) , trammell ( 1994 ) , and howell ( 2000a ) .
for sn 1999by , two sets of data were taken each night with each `` set '' consisting of four 2030 minute integrations at waveplate position angles of 0 , 45 , 22.5 , and 67.5 degrees .
polarization and flux standards were observed each night . since spectropolarimetry measurements require a high signal - to - noise , the sn
was observed over the course of three nights , and the data were combined .
the data presented here have been corrected for redshift , though this correction is negligible ( @xmath18 ) .
sn 1999by was discovered independently by r. arbour , south wonston , hampshire , england ( arbour 1999 ) , and by the lick observatory supernova search ( loss ; cf .
treffers 1997 , li 1999 ) .
the sn was found on the outskirts ( 100 west and 91 north of the nucleus ) of the well - known sb galaxy ngc 2841 . according to the nasa / ipac extragalactic database ( ned )
the host galaxy is a liner and has a heliocentric radial velocity of @xmath19 .
according to the lyon / meudon extragalactic database ( leda ) , the velocity corrected for virgo infall is 811 . using @xmath20 ,
this puts the distance to ngc 2841 at 12.49 mpc , with a distance modulus of 30.48 mag .
this prodigious galaxy has also produced sn 1912a ( type uncertain ) and the `` peculiar type i s '' sn 1957a and sn 1972r . according to bonanos ( 1999 ) ,
sn 1999by reached a maximum light of @xmath21 on ut 1999 may 10.5 , and a maximum in the @xmath22 band of @xmath23 ( date not given ) . according to weidong li ( private communication ) the maxima were may 10.3 ( @xmath24 ) and may 12.3 ( @xmath22 ) . using a distance modulus of 30.48 , and the photometry of bonanos ,
the absolute magnitudes at peak are @xmath25 and @xmath26 . thus sn 1999by is underluminous by roughly 2.5 magnitudes compared to a typical sn ia .
sn 1999by also had a large @xmath27 of 1.87 .
this makes sn 1999by nearly as steeply declining as sn 1991bg , which had @xmath28 .
sn 1999by is thus one of the most underluminous and rapidly declining sne ia known .
sn 1999by is similar to sn 1991bg in other respects as well . the line at 5800 is strong relative to the line at 6150 , another signature of the subluminous subclass . according to bonanos et al .
( 1999 ) the flux near 4000 was depressed as in other subluminous sne ia , a feature thought to be due to absorption .
the feature at 7500 is also strong in both sn 1991bg and sn 1999by .
flux spectra of sn 1999by can be seen in figure [ flux ] .
maximum light in @xmath24 occurred on may 10 , therefore the spectra shown are two days before , one day before , and at maximum light .
several features are apparent upon inspection figure [ flux ] .
most obvious is the fact that sn 1999by is unquestionably a subluminous sn ia .
it is similar to the prototype , sn 1991bg .
beginning at the blue side of the spectrum , we see that the spectrum starts near 4300 in a huge absorption feature .
this is thought to be due to , a feature only observed in subluminous events ( filippenko et al .
1992 , hereafter f92 ) .
the fact that this ionization state of ti is seen indicates a relatively low temperature in the ejecta , since is ionized in typical sne ia ( nugent et al .
1995 , hereafter n95 ) . is also responsible for absorption features in the 45005000 region ( f92 ) . moving redward , another signature of subluminous events
is seen in the pair of lines on either side of 5400 .
as one moves down the sequence from overluminous to subluminous sne , the blue line is overtaken in strength by the red one ( n95 ) . here
the bluer line is scarcely seen at all , consistent with our finding that sn 1999by is a very subluminous sn ia . from the blueshift of the absorption features of si
, one can measure an approximate velocity of this element ( and thus its placement in the ejecta , since @xmath29 ) . for sn 1999by
we measure @xmath30 , comparable to speeds of @xmath31 and @xmath32 reported for sn 1991bg by f92 and leibundgut ( 1993 ) , respectively .
this is slightly lower than the average si velocity in a normally bright sn ia , @xmath33 to @xmath34 , though velocities derived from absorption minima of broad doublets as seen in the 6150 feature are only accurate to @xmath17 1000 . on the far red end of the spectrum ,
the absorption at 7500 is yet another hallmark of subluminosity .
this line is due largely to a blend of @xmath357774 and on the red wing , separated here by narrow atmospheric absorption .
this feature is not seen in the luminous sn 1991 t , but increases in strength as one goes down the luminosity sequence to subluminous sne ia ( n95 ) .
this feature is also absent in sn 1998de , a fast - declining sn ia that was peculiar in several ways ( modjaz 2000 ) .
the dominant sources of opacity in an sn ia , line scattering and electron scattering , can both change the polarization of incident flux .
electron scattering polarization is independent of wavelength . in the case of axisymmetric scattering ,
the polarization position angle can take on only one of two values : 0 or 90 with respect to the symmetry axis ( see cropper 1988 ; j91 , h91 ) . line scattering need not lead to complete angular redistribution . the degree to which line scattering can polarize depends on the lower- and upper - level total angular momenta of that line ( j91 ) .
despite the fact that lines can polarize , most line scattering is less polarizing than electron scattering by an order of magnitude . as a result , electrons are the chief polarization mechanism , while the lines generally depolarize the flux previously polarized by electron scattering .
lines can redirect photons by decay in the same transition , but they can also redistribute photons to both higher and lower frequencies by further excitation or ionization from the level at which the photon is absorbed or by decay into multiple discrete lower levels ( j91 , hflich 1995a ) .
dust can polarize by scattering through dichroic extinction and through emission of polarized photons .
dust scattering is usually wavelength dependent , though the results vary with scattering angle . for dust typical of the ism ,
the polarized flux of optical photons undergoing a single scattering can decrease with wavelength for small scattering angles , increase with wavelength for large scattering angles , or remain constant for @xmath36 ( white 1979 ; webb et al . 1993 ) .
nonspherical dust grains with aligned rotation axes ( from , e.g. a magnetic field ) can be a source of polarization through dichroic extinction . in this case
, the polarization position vector is perpendicular to the grain alignment vector .
table 2 gives all the sne ia for which some polarization data are available .
figure [ quftotrc ] shows the normalized stokes parameters @xmath37 and @xmath38 and the flux spectrum of sn 1999by .
the dashed line is the unsmoothed data .
a darker colored line on the @xmath37 and @xmath38 plots shows the data smoothed with a running boxcar smoothing function of 17 pixels ( 65 ) .
features from 48005100 are blue , from 51005270 light blue .
the features from 5270 to 5600 are green .
the features are marked by two different colors .
the 5800 feature is yellow , while the 6150 line is magenta .
black represents the area between the si features and the area slightly redward .
finally , the large , broad polarization feature to the red of 6500 is colored red . in figure [ tsstot ] we present @xmath39 for the unsmoothed data , but we note that this indicator is biased high for low signal to noise data . therefore , we generate the smoothed version , @xmath40 ( dark solid line ) from the smoothed @xmath37 and @xmath38 values . in other words , @xmath41}\ ] ] where the subscript @xmath42 denotes smoothed values .
similarly , the smoothed polarization position angle is : @xmath43}\ ] ] in addition to @xmath1 , figure [ tsstot ] also shows @xmath44 and the associated errors @xmath45 , and @xmath46 .
figure [ qutotrc ] shows the smoothed values for @xmath37 and @xmath38 in the plane for sn 1999by .
the colors are identical to the color scheme used in figure [ quftotrc ] .
in such a plot , each point defines a vector from the origin .
the length of this vector is @xmath40 and the orientation is @xmath47 .
figure [ qutotrc ] and its implications are discussed further in [ qudiscussion ] . as is often the case in spectropolarimetry measurements , the data for any single set of observations has a low signal to noise .
thus we have combined the data for all three nights to produce figures [ quftotrc ] and [ tsstot ] .
these figure show the sum of all observations , and consequently have the best signal to noise , with @xmath48 , and @xmath49 .
note that despite this low formal statistical error , systematic errors prevent us from actually discriminating small scale changes in @xmath40 of order 0.05 to 0.1% .
we use this `` lowest noise '' case for most subsequent analysis , but we also look to the individual data sets for confirmation that features are real .
figure [ 4p ] shows the results of combining the data in different ways .
the top panel shows @xmath1 derived from all observations .
the second panel shows @xmath1 constructed from only the best four data sets .
the third panel shows @xmath1 from the best single data set , taken on may 10 when the supernova was at maximum light .
finally , the lower panel of figure [ 4p ] shows the sum of the best data from may 8 and 9 combined with the second best data set from may 10 .
these are the next three best data sets after the best set from may 10 .
the dashed lines in figure [ 4p ] are 1 @xmath50 errors .
our indicator @xmath40 falls beneath the midpoint of the unsmoothed data because @xmath1 is biased high as an indicator , though @xmath40 is less biased .
the degree of the bias to @xmath1 increases with low signal to noise data .
we also tested wang s indicator , ( wwh97 ) and the rotated stokes parameter ( tran 1995 ) , and both agree with @xmath40 to within 0.02% across all wavelengths .
it is important to remember that there is no perfect indicator for the total degree of polarization ( leonard 2000b ; simmons & stewart 1985 ) .
three features stand out in every data set .
these are : the overall level of @xmath51 , the change in @xmath44 across the 6150 feature , and the rise in @xmath1 to 0.6% at 6900 .
we discuss each in turn . * the level of @xmath1 : * the first of these significant features is that throughout most of the spectrum @xmath51 , with some excursions to lower values .
not every bump in the smoothed data in figure [ 4p ] is significant because smoothing can introduce artifacts into low signal to noise data ; however , the overall level of @xmath52 shown in figure [ tsstot ] is significant .
the statistical error for @xmath1 is @xmath53 in the region of interest , and the statistical error for the smoothed spectrum is @xmath54 .
we should also keep in mind that the fact that we have smoothed the data means that we can not distinguish features narrower than 65 . * 6150 : * an overall level of polarization is interesting , but could be caused by interstellar or circumstellar sources .
perhaps the most striking feature of the spectropolarimetry of sn 1999by is then the change in position angle , @xmath44 , of the polarization through the 6150 feature as seen in figure [ tsstot ] .
the pa jumps from 100 to 150 over this region , peaking at 6200 . as shown in the second panel of figure [ tsstot ] , @xmath55 without smoothing . after smoothing , @xmath56 .
_ even without smoothing , the change in @xmath44 across si is a 4@xmath50 result .
_ this is clear evidence that the sn is intrinsically polarized .
it is also remarkable that the polarization change in si is roughly equal to the polarization change across the entire wavelength region . at 4800
@xmath57 but , by 7100 @xmath44 has dropped to @xmath58 .
this is significant because we will later show that there is a single choice of interstellar polarization that , when subtracted from the data , reduces the position angle across all wavelengths to nearly a single value .
* the 6900 rise : * the rise in polarization toward the red peaks at @xmath59 at 6900 .
this feature is wider ( 1000 ) than any particular line , and it occurs in a region normally blanketed by fe - peak lines . such a rise in polarization in this wavelength region was predicted theoretically by wwh97 .
we will revisit this feature when we discuss the theoretical interpretation of the data in [ theory ] .
the plane is a powerful tool for interpreting sn polarimetry data both with respect to testing the overall geometry and to determining the interstellar component of the polarization .
the plot for the smoothed data of sn 1999by is given in figure [ qutotrc ] .
each point corresponds to one pixel .
the points are color coded to reflect their distribution in wavelength .
the color coding is from figure [ quftotrc ] which shows the smoothed @xmath37 , @xmath38 , and flux spectra .
polarization arising from axisymmetric scattering will show up as a straight line on the plot ( see cropper 1988 ; j91 ) .
uncorrected for interstellar polarization , the data for sn 1999by clearly define an axis of symmetry in figure [ qutotrc ] .
the scatter is @xmath60 , comparable to the scatter expected from photon statistics and systematic effects .
the distribution of the 6150 feature is particularly interesting , as noted in the previous section . here
the si feature is colored magenta .
it was previously noted that the variation in position angle across the si feature matches the variation in position angle seen across the entire wavelength region of study .
stated another way , is not localized to one spot on the diagram , but is distributed throughout the `` line '' demarked by the rest of the points in the plot , except the points from the 6900 region , which are colored red . physically , this means that si does not have a special geometry it shares the same axis of symmetry as the rest of the sn .
there is a slight deviation from linearity in the red points on figure [ qutotrc ] .
if we take this jog to be real , noting that it is nearly perpendicular to the rest of the points , then the jog corresponds to a switching from an oblate to a prolate geometry in the corresponding wavelength region or vice versa .
this seems to be an unnecessary complication of the model .
the discrepant points can be explained by noise .
if a point deviates from the line by 3@xmath50 , smoothing can introduce correlated errors , causing several neighboring points to appear to deviate from the line as well . on the basis of these few points ,
it is not possible to conclude that there are deviations from the global geometry in this wavelength region .
interstellar polarization ( isp ) due to dust in the interstellar medium of both the host galaxy and the milky way can change the observed polarization spectrum of an object .
the difficult question of how much of the polarization signature measured for sn 1999by is intrinsic and how much is interstellar plagues polarization measurements of all objects .
the issue is particularly difficult in the case of extragalactic objects which can have two sources of isp one contribution from our galaxy , and one from the host galaxy .
fortunately , we can estimate the isp for sn 1999by via two essentially independent methods , the empirical plane method and theoretical considerations .
we will discuss these specific techniques below , but in this section we provide general constraints on interstellar polarization . the maximum interstellar polarization allowable , corresponding to the most favorable alignment of dust grains along the line of sight is @xmath61 ( serkowski , mathewson & ford 1975 ) .
the galactic longitude of sn 1999by is 166.91 and latitude 44.12 ( ned ) .
the galactic extinction towards ngc 2841 is estimated to be @xmath62 mag ( schlegel 1998 ) though burstein & heiles ( 1982 ) give @xmath63 mag .
schlegel also estimate @xmath64 mag . using schlegel s estimate of @xmath64 mag ,
the maximum galactic component of polarization is 0.144% .
this is indicated by a dashed circle in the plane diagram of figure [ qutotrc ] .
some limited information about the galactic component of isp can be gleaned from observations of stars near the line of sight to sn 1999by .
there are only three stars with polarization measurements in the literature within @xmath65 of sn 1999by , and they are presented in table [ polstars - table ] .
they are also plotted as light blue asterisks on the plane , figure [ qutotrc ] .
the closest star to sn 1999by is hd 82328 .
it is 2.16 away from the sn on the sky .
while hd 82328 shows a low level of polarization , @xmath66 , it is also the closest star at 19 pc , so it does not sample much of the isp between us and the supernova .
the next closest star , hd 82621 is 2.7 away with @xmath67 , but it is only 40 pc away . if we go out to hd 77770 , we can look through 1.3 kpc of the galaxy to see @xmath68 , but now we are looking @xmath69 away from the line of sight to the sn .
this is probably the reason this star is outside the allowed galactic isp region of figure [ qutotrc ] .
the three stars also have different polarization position angles , and it is likely that any galactic polarization toward sn 1999by would have still a different position angle .
in addition , if there is a contribution from isp in the spectrum of sn 1999by , it could arise from the host galaxy .
these three stars alone do not allow us to place meaningful limits on the isp .
fortunately , the plane does suggest a choice of interstellar polarization .
isp is a smoothly varying function of wavelength , described by the serkowski law : @xmath70\ ] ] where @xmath71 is the percent polarization at wavelength @xmath35 , @xmath72 is the maximum polarization at wavelength @xmath73 and @xmath74 ( @xmath35 is in microns ; whittet 1992 ) . to remove isp
, we subtract a serkowski law from the data .
this is essentially a vector subtraction in the plane . correcting for interstellar polarization can either increase or decrease the derived intrinsic polarization of the object , and it can turn absorption features into emission features and vice versa
this is easy to understand if we remember that we are `` undiluting '' the light from the sn by removing contaminating isp .
mathematically , this is because the combination of the intrinsic polarization and isp is a vector addition . to remove the interstellar signature from our data ,
we must first determine @xmath75 .
as remarked above , the data points fall on a line in the plane of figure [ qutotrc ] .
we can assume for sne ia that some parts of the spectra are depolarized ( wwh97 ) .
taken together , this allows two possible choices of interstellar polarization ( see wang et al .
if we place the isp at either end of the line delineated by the data points and subtract it , then the data will fall along a radial line in space with essentially one constant value of @xmath44 for the system ( note that @xmath44 is one - half the angle of a vector on the plane ) .
it is unlikely that the isp point could be placed well within the spread of unsubtracted sn 1999by points . in this case , after subtraction the position angle would change abruptly where there was no particular feature in the polarization spectrum .
we argue that this behavior is not likely to be physical .
the two choices for isp are the light blue circles marked as a and b in figure [ qutotrc ] .
the diameter of the circles corresponds to the approximate uncertainty in the placement of the isp .
isp a has a value of @xmath76 and @xmath77 .
isp b has a value of @xmath78 and @xmath79 . both of the choices for interstellar polarization lie outside of that allowed for the galaxy along the line of sight to sn 1999by . assuming that the @xmath80 of schegel ( 1998 ) is correct
, this implies that some of the isp arose in the host galaxy . note that we could place the isp any place on the line defined by the data in the plane that was not well within the spread of data points .
the further the isp from the data , the more tightly , but arbitrarily , the isp position angle would be defined .
we can not rule out more extreme choices , but argue that the isp is most plausibly at either a or b so that some portions of the data represent low polarization in the supernova .
we think this argument applies to sne ia , but note that the data on the type ii sn 1999em show that at some phases there is finite polarization at all observed wavelengths ( wang et al .
2000b ) .
* isp choice a : * first we consider isp a. figure [ quptaisp ] presents the resulting @xmath37 , @xmath38 , @xmath1 , and @xmath44 with this choice of @xmath75 subtracted from the data . note that in the lower panel , the effect of removing the interstellar polarization is that the position angle becomes nearly constant across the entire wavelength region .
there is some deviation at longer wavelengths .
this is due to the fact that when the isp - corrected points are relocated on the plane , the long wavelength points are closest to the origin , and deviations in position angle are magnified .
isp a is disfavored for three reasons : it produces a decrease in polarization at 7000 , it produces an increase in polarization in the 6150 feature , and it is less likely due to reddening considerations .
we consider each in turn .
subtracting isp a produces a high @xmath1 @xmath17 0.8% , at short wavelengths , and a decrease in polarization at wavelengths greater than 6500 .
this is contrary to theoretical expectations .
numerical calculations presented in wwh97 predicted a _ rise _ in polarization longward of 6500 due to decreased line blanketing .
this is discussed in further detail below in [ theory ] .
the removal of isp a from the data also causes the polarization to be enhanced in the si 6150 feature .
while this increase can not be ruled out completely , from a theoretical perspective , this feature is easier to understand if it is a depolarization .
p - cygni lines are generated above the photosphere , so polarization produced by electron scattering at the photosphere is expected to be depolarized by absorption and reemission in the layers above it ( see the appendix ) .
note that by using this qualitative theoretical prediction to discriminate between the two choices of isp , the `` empirical method '' is not completely independent of theory .
however , the approach is significantly different from the determination of isp from a quantitative comparison to a specific theoretical model as done in 5 below .
another consideration that slightly disfavors the choice of isp a is the observational constraint that the sn does not appear to be reddened .
as mentioned above , the maximum interstellar polarization allowable , corresponding to the most favorable alignment of dust grains along the line of sight is @xmath61 .
we can use this to place a lower limit on the reddening expected from our choice of isp .
inverting , we have : @xmath81}\ ] ] thus for model a , @xmath82 , and for a standard extinction law ( savage & mathis 1979 ) , @xmath83 . given that sn 1999by is already bluer than sn 1991bg , which was thought to have low extinction along the line of sight , it is likely that sn 1999by is not significantly reddened .
sn 1999by has a pseudocolor of @xmath84=0.440.47 ( bonnanos 1999 ; li , private communication ) .
this compares to a value of @xmath85 for sn 1991bg ( m97 ) .
note that @xmath84 and @xmath85 are not , strictly speaking , the same quantity , but this is the closest comparison we can make until the photometry data is published . both sn 1991bg and sn 1999by
are redder than typical sne ia , but this is thought to be intrinsic to subluminous sne ia ( f92 ; leibundgut et al .
1993 ; hkw95 ; turatto 1996 ; mazzali 1997 ) .
since neither the extinction towards sn 1991bg nor towards sn 1999by is known with certainty , and we can only place a lower limit on the reddening expected from our model , this is not a hard limit .
* isp choice b : * the subtraction of isp b from the data , like isp a , causes the sn data to lie essentially on a straight radial line in the plane as can be seen in figure [ qutotrcisp ] . for this choice of isp ,
the polarization deduced for the sn has a constant position angle of @xmath86 across the entire wavelength region . in the bottom panel of figure [ quptbestisp ] ,
the isp subtracted data is compared to the position angle from the non - isp - corrected data .
this figure also shows smoothed and unsmoothed versions of @xmath1 , @xmath37 , and @xmath38 after isp correction .
isp b is slightly favored due to reddening considerations . for this model @xmath87 , and @xmath88 , less than half that of isp a. from a theoretical perspective , this choice also gives the desired depolarization of the si feature .
isp b also gives a rise in polarization at red wavelengths as theoretically predicted .
for these reasons it is a better choice than isp a. with isp b subtracted , we find @xmath1 @xmath17 0.4% from 5600 6600 with the exception of the si feature .
the difference between the polarization maximum at 6900 and the overall level , ( e.g. at 5800 ) , is 0.4% .
this relative difference does not depend on the choice of isp given the assumption that intrinsic polarization defines a radial line on the plot .
the maximum degree of polarization after subtraction of isp b is @xmath89 . as noted above , subtraction of isp
a gives the same maximum value , but for isp b the maximum polarization occurs at wavelengths @xmath90 .
this is where the maximum polarization is expected from theory ( wwh97 ) .
the subtraction of isp b produces relative depolarization around 48005600 , where the level is 0.2% or lower .
this is where line blanketing opacity dominates electron scattering opacity and thus is where depolarization is expected from theory ( hflich , mller & khokhlov 1993 , hereafter hmk93 ; wwh97 ) .
we have estimated the degree of uncertainty associated with this choice of isp by choosing different values of @xmath75 close to isp b and observing the degree to which they cause the axis ratio of the system to deviate from a single value across all wavelengths .
these are shown in howell ( 2000a ) , but are omitted here for brevity .
after consideration of the uncertainties , our best choice of isp , case b , is @xmath91 at @xmath92 . regardless of whether we choose isp a or b , the intrinsic polarization of sn 1999by , given by the total length of the distribution of points in the plane , is @xmath93 .
as mentioned earlier , si is distributed throughout most of the line in the plane and hence samples this total intrinsic polarization .
one implication is that the si 6150 feature will always show a minimum or maximum in the polarization after subtraction of a reasonable choice of isp lying at one or the other extreme of the data in the plane .
the chemical and density structures for the initial models are based on calculations of the stellar evolution from the main sequence to white dwarf formation and the subsequent phase of accretion onto the white dwarf ( umeda et al .
1999 ; hflich , et al .
spherical dynamical explosions and corresponding light curves are first calculated .
aspherical structures are then constructed by mapping the spherical structures in the homologous expansion phase onto ellipsoidal density structures with an axis ratio @xmath94 . here
@xmath95 is the equatorial major axis and @xmath96 is the polar major axis .
subsequently , detailed non - lte spectra are calculated for the flux and polarization .
details are given in the appendix .
delayed detonation ( dd ) models and pulsating delayed detonation models have been found to reproduce the optical and infrared light curves and spectra of sne ia reasonably well ( hflich 1995b ; hkw95 ; hk96 ; nugent et al .
1997 ; wheeler 1998 ; lenz et al .
1999 ; 2000 ; gerardy 2000 ) .
in particular , the dd models with a variation in the parameter , @xmath97 , at which the transition is made from deflagration to detonation , give a range of ejecta structures and mass of 56 .
models with smaller transition density give less nickel and hence both lower peak luminosity and lower temperatures .
the latter gives lower opacities and hence a steeper decline in the light curve .
the dd models thus give a natural and physically well - motivated origin of the brightness - decline rate relation of sne ia within the paradigm of thermonuclear combustion of chandrasekhar - mass co white dwarfs ( hkw95 ; hflich et al .
1996 ) . here
we present the polarization spectra of a dd model with parameters which have been adjusted to provide a fit to the time evolution of the infrared spectra observed for sn 1999by ( gerardy et al .
2000 ) without further tuning . in the delayed detonation scenario ,
the transition density , @xmath97 , is the dominant factor that determines the amount of 56 produced and hence distinguishes the normally - bright from the subluminous models ( hkw95 ; hflich 1995a ; iwamoto et al .
we contrast the ejecta structure of a normally - bright sn ia model with that required to reproduce a subluminous event like sn 1999by in figure [ model1 ] . the same progenitor structure and central density at the time of the explosion have been taken as the normally - bright sne ia studied in hflich et al .
we assume a wd with a chandrasekhar mass and solar metallicity which originates from a main sequence star of 7 @xmath98 ( umeda et al .
1999 ) . at the time of the explosion ,
the central density is @xmath99 g @xmath100 . to produce a subluminous supernova consistent with the ir - spectra of sn 1999by
, @xmath97 has been chosen to be @xmath101 g @xmath100 compared to @xmath102 g @xmath100 for the normally - bright model .
the subluminous model produced @xmath103 of 56 compared to @xmath104 for the normally - bright model .
a typical feature of subluminous dd models is the greater production of o , s and si ( @xmath105 and @xmath106 , respectively ) compared to normally - bright sne ia ( @xmath107 and @xmath108 , respectively ) .
another feature of subluminous sne ia models is that the 56 is constrained to the inner layers with low expansion velocities which become visible a few weeks after maximum light .
in contrast , for normally - bright sne ia models , the outer edge of the 56 layers extends to about 10,000 to 13,000 km s@xmath109 so that the ni is already visible at maximum light ( figure [ model1 ] ) . as we will see below , this qualitative difference between normally - bright and subluminous sne ia has important implications for our understanding of the polarization spectra . for more details on the explosion models , light curves and spectral evolution ,
see hflich et al .
2000 and gerardy et al .
( 2000 ) . for this paper
, we calculated the polarization spectrum at day 15 after the explosion which corresponds to maximum light in b. at this epoch the model gives @xmath110 , and @xmath111 .
the temperature structures for the aspherical models were based on the spherical calculations .
the optical spectrum is formed in layers with expansion velocities between @xmath112 and @xmath113 which corresponds to about 1 @xmath114 in mass coordinate , in layers dominated by intermediate mass elements ( figure [ model1 ] ) .
we calculated the spectra for a variety of ellipsoidal structures at the photosphere with axis ratios @xmath94 between 1.15 and 1.5 and adjusted @xmath94 to fit the level of polarization observed in sn 1999by for various inclination angles .
currently , three - dimensional models are not a suitable tool to adjust parameters such as the main sequence mass , accretion rates , metallicity , rotation and combustion physics to find best fits .
therefore , we restrict our discussion of the fluxes to the main spectral features . in our model , the spectrum at day 15 is formed at the outer edge of the si - rich layers , i.e. well above the region containing 56 ( figure [ model1 ] ) .
the top panel of figure [ model2 ] shows reasonable agreement between theoretical and observed features , and thus confirms the validity of this model in terms of doppler shifts of the lines , their strengths , the ionization stages , and the overall slope of the spectrum .
most of the strong features can be attributed to the intermediate mass elements , and .
most of the weaker lines are due to iron - group elements which are singly ionized as expected for subluminous sne ia with a relatively cool photosphere ( hflich et al .
the feature at about 7450 is too weak in the models .
this is not due to a lack of oxygen ( figure [ model1 ] ) , but the strength of the line depends sensitively on the excitation .
the discrepancy may be an artifact of the approximations imposed by the numerical treatment , or it may hint at the need for a slightly higher excitation in the outer layers , due to non - thermal electrons or an increased @xmath115-ray flux .
we note that some fast - declining ( subluminous ) sne ia do not show strong in their spectra ( modjaz 2000 ) .
the flux spectrum depends on the inclination , @xmath116 ( here taken to be the angle between the equatorial plane and the line of sight ) . for oblate geometries ,
the flux increases with @xmath116 because of the increased escape probability of photons towards polar directions ( h91 ) .
although the asphericity is small in our example , the absolute flux varies by about 30% or @xmath117 from equator to pole . with increasing @xmath116 , weak absorption features ( e.g. due to , ) tend to be smeared out , and the overall spectrum becomes smoother as shown in the top panel of figure [ model2 ] .
these absorption features are formed in a narrow region close to the origin of the quasi - continuum which is produced by electron scattering , bound - free processes , and by a large number of very weak lines ( hflich 1995a ) . because @xmath118 , for an oblate geometry the photosphere seen by an observer spans a wider velocity range as @xmath116 increases .
in addition , the probability of multiple scattering of photons tangential to the photosphere into the absorption " feature increases with @xmath116 in the case of oblate geometries , thus diluting the depolarization of the line .
flux and polarization spectra of a model with 17% asymmetry observed at various inclination angles are shown in figure [ model2 ] in comparison with observations .
the model polarization spectra have been rebinned with @xmath119 , to provide a statistical error comparable to the observations which , for this section , have been rebinned with @xmath120 .
the model polarization spectra for various choices of the inclination , @xmath116 , and the raw and binned data are given in the second panel of figure [ model2 ] .
the statistical errors in the model and observations and the polarization angle before and after correcting for the model - determined isp are given in the third and fourth panels of figure [ model2 ] , respectively .
in the models , the polarization is produced by electron scattering . as discussed in the appendix ,
we assume complete redistribution for lines which causes depolarization in lines . as expected both from analytical and numerical studies , the polarization @xmath121 ( e.g. van den hulst 1957 ; h91 ) .
the frequency dependence of the polarization is governed by the absorption probability in lines relative to electron scattering .
most prominent is the strong depolarization by the line at 6150 and the , and si ii features at 7600 ( figure [ model2 ] , panel 2 ) . at shorter wavelengths ,
a large number of overlapping lines due to iron - group elements is responsible for the low polarization .
the wavelength region above 5400 shows an increasing degree of polarization due to the decreasing importance of line opacities .
shortwards of the line at 6150 , the polarization level in the models is generally lower and shows modulations due to moderately strong , , and fe ii lines , as expected from the opacity pattern for si - rich layers at temperatures between 5000 to 10000 k ( see figure 1 in hmk93 ) .
we note that the frequency pattern of the model polarization spectra for subluminous supernovae is rather different from normally - bright sne ia at the same phase ( e.g. sn 1996x , wwh97 ) . in the latter ,
the polarization spectra near maximum light are dominated by iron - group elements because the photosphere is between 10,000 and 12,000 km s@xmath109 , right at the interface between complete and incomplete si burning ( figure [ model1 ] ) .
in addition , the photospheric temperatures are higher than in subluminous events by several thousand degrees and , thus , iron - group elements are present both in the second and third ionization stage .
the presence of these ubiquitous iron lines gives a more pronounced rapid frequency dependence of the model polarization spectra for the normally - bright sne ia near maximum as contrasted with the relatively smooth variation predicted for the polarization spectra of the subluminous events .
we have iterated the component of the interstellar polarization to optimize the fits with respect to the line width and depth of the strong lines , in particular the feature at 6150 .
the best agreement is for @xmath122 and @xmath6 .
small changes of the isp - vector result in slightly broader components of the depolarization features . with increasing deviations from the optimized values ,
local maxima appear in the line centers which eventually dominate the spectra .
the vector for the interstellar polarization derived in this way is consistent with the contribution of the isp found with the empirical method described in 4.3 .
the agreement supports the soundness and relibility of both the empirical and theoretical approaches . as shown in figure [ model2 ] ( panel 2 ) , the observed polarization spectrum can be reproduced by an ellipsoid with an axis ratio of 1.17 seen equator on ( @xmath123 ) .
the overall level of polarization in different wavelength ranges and the velocity shift and strength of features produced by strong lines are consistent .
the theoretical polarization spectrum shows a distinct physical pattern between 5400 and 6000 at the 0.1% level which can not be discriminated in the data .
these small scale fluctuations may be valuable tools to analyze small scale structures caused by inhomogeous mixing if better data become available in the future .
an axis ratio of 1.17 is a lower limit to the amount of asphericity that is required , though we can also estimate an upper limit . as an alternative to a model with small asphericity seen equator on
, models with larger asphericity may be able to reproduce the observations if seen from larger inclinations , @xmath116 . to test this possibility
, we have calculated a set of models with larger axis ratios as presented in figure [ model3 ] . to reproduce the overall level of polarization for @xmath116 of @xmath124 and @xmath125
, the axis ratio @xmath94 must be boosted to 1.25 and 1.5 , respectively .
strong discrepancies at the @xmath126-level are present for @xmath127 which are well beyond the level of uncertainty .
most noticable are the strong peaks in @xmath1 at about @xmath128 and @xmath129 , and the change of @xmath130 between @xmath131 and @xmath132 .
such variations can not be eliminated by a different choice for isp . for @xmath133 and @xmath94=1.25 ,
the agreement with the observation of @xmath1 is better , but problems remain on the level of @xmath134 , and there are some problems with the location of the minima
. the model with @xmath133 may be marginally consistent with the observations .
we can conclude that sn 1999by has intrinsic asymmetries of the order of @xmath135 and that it is seen almost equator on .
whereas core collapse supernovae are commonly found to be polarized at 1% or greater , the degree of polarization is much smaller in sne ia ( wang et al . 1996 , 2000a ) .
spectropolarimetric and broadband measurements of polarization in sne ia are still exceedingly rare . for most measurements ,
the interstellar component can not be determined , making any intrinsic component impossible to determine .
table [ polsne - table ] shows all sne ia for which polarization data are available .
mccall et al . ( 1984b ) measured no significant polarization intrinsic to sn 1983 g , but did place ( large ) upper limits on spectropolarimetric signatures . the first slightly subluminous sn ia with
a polarization measurement was sn 1986 g .
broadband ubvrijh polarimetry with a maximum polarization of 5.2% is well fit by a serkowski law ( hough 1987 ) .
broadband polarimetry of sn 1998bu is presented in hernandez ( 2000 ) .
the polarization , ranging from 12% again follows the serkowski law .
spyromilio & bailey ( 1993 ) placed limits on the spectropolarimetry of sn 1992a which was observed about 15 days and 100 days past maximum light .
the overall level of @xmath1 is roughly 0.3% , comparable to the noise level .
while spyromilio & bailey make no claim for intrinsic polarization of the sn , polarization signatures of the type suggested for sn 1996x ( wwh97 ) can not be ruled out from the published data .
leonard ( 2000a ) presented spectropolarimetry of sn 1997dt , showing apparent changes in @xmath1 across spectral features .
the level of intrinsic versus interstellar polarization is not known .
wwh97 observed the normally - bright supernova sn 1996x about one week before maximum light in v. the data showed a small average polarization with modulations on the level of 0.2% .
wwh97 concluded that sn 1996x could have only a very small , if any , global asymmetry in the geometry , given the lack of detectable mean polarization , but that the modulations were consistent with inhomogeneities in distributions of the elements within the ejecta .
figure [ qucomp ] shows the data for sn 1996x and sn 1999by in the plane .
the lack of any preferred orientation to the data and hence a global asymmetry for sn 1996x contrasts distinctly with the linear distribution of points for sn 1999by .
the finer scale variations in the polarization spectra of sn 1996x are qualitatively similar to the expectations for the behavior of a photosphere significantly contaminated with iron - peak elements as a normally - bright sn ia should be ( figure [ model1 ] , wwh97 ) .
a comparison with detailed model calculations for a delayed detonation model confirmed that the data are consistent with little or no global asymmetry in the density structure of sn 1996x .
the continuum polarization in the model is modulated by ubiquitous weak lines of fe ii - iii , co ii - iii and ni ii - iii that serve to depolarize on scales of @xmath136 .
these lines are formed at the interface region between complete and incomplete si burning .
wwh97 suggest that there is an asymmetry in the _ chemical _ structure that may be a relict of the chemical plumes rising during the deflagration phase .
strong depolarization over the 6150 feature is predicted by the model , but not observed . as noted in wwh97 ,
another problem with the comparison to theory of sn 1996x is that the model shows a rise in the polarization longward of 6500 which is not apparent in the data .
sn 1999by shows both the depolarization at the feature and the rise in the polarization level to the red .
one of the open questions in sne ia research is the nature of the very subluminous subclass .
the prototypical example of this subclass is sn 1991bg .
light curves and spectra of sn 1991bg are presented in f92 , leibundgut ( 1993 ) , and turatto ( 1996 ) .
other subluminous sne include sn 1992k ( hamuy 1994 ) , sn 1997cn ( turatto 1998 ) and sn 1998de ( modjaz 2000 ) .
subluminous sne ia may be rare , but some analyses indicate that they could make up 16% ( li 2000 ) or more ( schaefer 1996 ) of all local sne ia . some defining characteristics of the subclass are rapidly - declining light curves ( @xmath137 ) , fainter than normal peak magnitudes by 23 mag , redder colors at maximum ( @xmath138 @xmath139 0.40.74 ) , a non - existent or weak secondary maximum in r and i , and an earlier transition to the nebular phase .
moreover , the 56 region is limited to expansion velocities below 4000 to 4500 km s@xmath109 , i.e. about a factor of 2 to 3 times smaller than in normally - bright sne ia . by comparison ,
the expansion velocity implied by si ii indicates a wider range of the si - dominated layers in subluminous sne ia ( @xmath140 to 12,000 km s@xmath109 ) compared to normally - bright sne ia ( @xmath141 to 13,000 km s@xmath109 ; turatto et al .
1996 ; gerardy et al .
subluminous events also show strong absorption features at 4200 ( ) , 5800 ( ) , 7500 ( & ) and 8300 ( ca ii ) . theoretical interpretations of subluminous sne ia given in ruiz - lapuente ( 1993 ) , woosley & weaver ( 1994 ) , hkw95 , n95 , mazzali ( 1997 ) , and milne ( 1999 ) include all three types of explosion mechanisms outlined in 1 .
sn 1999by is the first subluminous sn ia to show definitive evidence for intrinsic polarization . before isp correction
, the supernova shows an overall level of polarizaion , a rise in @xmath1 redward of 6500 , and a change in polarization position angle across the 6150 feature .
the strong wavelength dependence and the individual features are not expected from interstellar polarization and show the need for a component intrinsic to the supernova .
we employ a new method ( see wang et al .
2000b ) for the analyses the polarization data of supernovae by using the plane .
this representation provides important constraints on the overall geometry of the configuration , and it allows strong constraints to be placed on the isp if the data points are aligned .
the small spread of the data points in sn 1999by about a line in the plane reveals that the sn has a well - defined axis of symmetry .
though the spread around this line is consistent with rotational symmetry , we can not exclude small additional off - axis components . in the classical approach to determine the polarization component of the ism , the object needs to be observed at several epochs from which the polarization vector can be deduced by the reasonable assumption that it is time invariant . in the plane and for reasonable assumptions for the polarization mechanism
, we have shown that the component of the ism can be determined by the vector between the origin of the plane and the endpoint of the line of the polarization data points .
obviously , with this choice , the polarization position angle should not vary over lines .
the residuals can be used to judge the quality of the data and the assumptions . for sn 1999by
, this method suggests that the @xmath75 @xmath142 with @xmath79 which is consistent with the best fits for the ism based on detailed models ( @xmath75@xmath143 ) .
based on a delayed detonation model , synthetic nlte - spectra have been computed and compared to the observations .
the free parameters have been adopted from a spherical delayed detonation model that reproduces the infrared spectrum of sn 1999by between 1 and 2 @xmath144 and its evolution with time ( gerardy et al . 2000 ) .
no further tuning of the initial model is done to fit the polarization data , although three new parameters must be introduced , the axis ratio , the angle of the line of sight to an observer , and the isp . the isp could be taken from the observations , but the theoretical analysis independently confirms the empirical value .
the axis ratio and inclination angle are constrained by two aspects of the observations : the overall level of polarization and the shapes of spectral features .
the initial spherical model is remapped to an ellipsoidal geometry with an axis ratio of @xmath94 .
the axis ratio @xmath94 has been adjusted to provide the overall level of @xmath1 observed .
both the optical flux spectra and polarization spectra of sn 1999by can be reproduced by models with an asphericity of @xmath4 which is seen equator on .
higher inclinations , which require larger axis ratios to give the overall polarization , can be ruled out from the spurious frequency dependence of the resulting polarization spectra .
the relatively high polarization we observe for sn 1999by may be a significant clue to the nature of sne ia .
existing limits on polarization of other sne ia are very sparse ( 7 objects at the time of this paper , see table 2 ) .
the polarization due to asphericity decreases with inclination , further reducing the significance of any statistics . within these limitations
, we would have expected to see more polarized sne ia if @xmath145 to @xmath146 is common .
the lack of such detections supports the notion that the intrinsic polarization of the subluminous sn 1999by is unusually large .
another argument is that , although small , the size of the asymmetry we derive implies a change of the observed luminosity of about @xmath117 from equator to pole .
this spread is larger by a factor of @xmath147 than the mean dispersion in the brightness - decline relation @xmath148 ( hamuy et al .
1995 ; riess et al .
the dispersion in @xmath148 is dominated by the normally - bright sne ia and the orientation of the symmetry axis of a given supernova is arbitrary . taken together , these facts again suggest that the asphericity is unusually large for sn 1999by compared to most normally - bright sne ia .
the low mean polarization observed for the normal sn 1996x ( wwh97 ) is consistent with this .
if future observations confirm that the large asphericity of sn 1999by is characteristic of subluminous sne ia , this may provide important clues to the physical reasons for both normal and less - than - normal peak luminosity .
our quantitative analysis of the data of sn 1999by was based on delayed - detonation models with imposed asphericity without addressing the question of the physical mechanism which produces aspherical envelopes . within the delayed detonation model , possible mechanisms to induce asymmetry
include the following : 1 ) instabilities in the nuclear burning front during the deflagration phase , 2 ) rapid rotation of the progenitor white dwarf , 3 ) delayed detonation transition at a point rather than simultaneously on a sphere , and 4 ) impact of the supernova ejecta on the secondary star .
aspherical configurations may also be the result of 5 ) the merging process of two degenerate white dwarfs and 6 ) , in principle , edge - lit helium detonations in sub - chandrasekhar mass wds .
studies of the deflagration phase of dd models reveal large plumes of burning material ( e.g. khokhlov 1995 , 2000 ; niemeyer & hillebrandt 1995 ) .
these plumes may leave their imprint on the chemical structure of the ejecta .
the plumes do not significantly perturb the density distribution and hence will not explain global asymmetries in the overall density structure ( khokhlov 2000 ) , but could leave an imprint at composition interfaces .
wwh97 suggested this mechanism as a possible explanation for the polarization signatures seen in sn 1996x , but these plumes are an unlikely explanation for the global deviations from sphericity with a well - defined axis of symmetry as in sn 1999by .
livne ( 1999 ) pointed out that we can not expect a deflagration / detonation transition on a sphere as implicitly assumed in spherical models , but that the transition may begin at one point .
detailed 2d calculations showed large scale asymmetries , but livne only followed the explosion process .
he did not follow the calculation into the homologous expansion , which may or may not destroy any initial asphericities by the time of maximum light when the observations of sn 1999by were taken .
it is also not clear why such a mechanism should apply to subluminous sne ia and not , perhaps , to most sne ia . in any accretion model
, we expect an impact of the supernova ejecta with the secondary star .
marietta , burrows , & fryxell ( 2000 ) computed 2d numerical simulations of the impact of an sn ia explosion with hydrogen - rich main sequence , subgiant , and red giant companions .
they find that the blast strips main sequence and subgiant companions of 15% of their mass while red giants lose @xmath17 97% of their envelopes ( 0.5 ) .
the impact of the ejecta with the secondary star creates a hole in the debris of angular size @xmath149 , corresponding to @xmath17 10% of the ejecta surface .
the result was similar for all cases considered because the companion was close enough to be in roche lobe overflow .
this effect may be consistent with the polarization observed in sn 1999by , but we would expect the same effect in both normal and subluminous sne ia which may be in conflict with the low polarization observed in normally - bright sne ia .
mahaffey & hansen ( 1975 ) calculated a rotating detonation model , but no detailed models have been calculated for rapidly rotating , deflagrating white dwarfs .
rapid rotation will produce a global distortion of the density structure of the presupernova wd ( mller & eriguchi 1985 ) and may also affect the propagation of nuclear burning fronts in the supernova , resulting in small asphericities .
we regard this possibility as very attractive since the combined effect of rotational distortion and modification of the nuclear deflagration front has the potential to link the subluminosity of sn 1999by to the large , global asymmetry .
a comparison with model 7 of mller & eriguchi suggests that even with solid body rotation ( which produces the minimum distortion for a given angular momentum ) , sufficient rotation to distort the structure in the silicon layers of our subluminous model would need to have a rotation energy only about 5% of the gravitational energy .
thus a significant , but not extreme , rotation may be sufficient to explain the degree of polarization we observe in sn 1999by .
rotation could thus serve as a single parameter that distinguishes normal sne ia arising in slowly - rotating white dwarf progenitors from subluminous sne ia that occur in more rapidly - rotating white dwarfs .
one possible problem with the scenario of rotation as a single parameter is that sn properties appear to show a correlation with environment and by implication , progenitor age ( howell 2000a , b ) . in the merging scenario ,
two c / o white dwarfs merge to produce an sn ia ( webbink 1984 ; iben & tutukov 1984 ) .
this would provide a natural explanation for the lack of hydrogen seen in ia spectra .
there is evidence that such systems exist ( saffer et al .
khokhlov et al .
( 1993 ) and hk96 parameterized merger events as spherically symmetric c / o wds with thick envelopes .
they produced light curves that were in reasonable agreement with some sne ia .
doubts about the merger models have been raised .
three - dimensional hydrodynamical simulations of the process show that ( unless the system ignites during the merger process ) the less massive star is disrupted into a hot envelope and an accretion disk around the primary ( benz 1990 ; rasio & shapiro 1995 ; mochkovitch , guerrero , & segretain 1997 ) . in this case , off center carbon ignition produces a flame that propagates inward , converting the star to an o+ne+mg wd ( nomoto & iben 1985 ; saio & nomoto 1985 , kawai , saio , & nomoto 1987 ; saio & nomoto 1998 ) .
magnesium-24 then undergoes electron capture , inducing accretion - induced collapse , to produce a neutron star ( saio & nomoto 1985 , mochkovitch & livio 1990 , nomoto & kondo 1991 ) .
alternatively , it may be possible to ignite the system due to tidal heating during the merger process ( iben , tutukov , & fedorova 1998 ) which could produce a highly aspherical explosion .
this scenario has not been demonstrated in realistic simulations , but deserves further study .
we have to stress the limits of this study which must be seen only as a first step . clearly , more high quality data must be obtained to provide a statistical sample and to make use of small scale features to evaluate details of the geometry and , ultimatively , to probe details of the explosion mechanism ( e.g. the burning properties ) and the scenario ( e.g. rotation of a single rapidly rotating wd vs. merging of two wds ) .
future data should have a noise level well below 0.1% to address questions of deviations from axial symmetry and to utilize the small scale variations seen in the models .
this is a very feasible goal with the current and upcoming generation of 8 m class telescopes .
our models are based on aspherical envelopes which are distorted artificially .
although this is a reasonable approach to estimate the size of asphericity and general properties , it hardly provides the desired link to the physical mechanisms responsible .
detailed hydro calculations for the various mechanism are needed . within the delayed detonation scenario ,
consistent hydro calculations of rotating wds , deflagration fronts and the multi - dimensionality of the deflagration to detonation transition are feasible .
although any successful model must reproduce the overall pattern in the chemical composition , alternative scenarios with full multi - dimensional hydrodynamics may do this job as well .
in particular , the merging scenario needs more attention .
this research was supported in part by nsf grant 95 - 28110 , by nasa grant lsta-98 - 022 and a grant from the texas advanced research program .
the calculations were done on a cluster of workstations financed by the john w. cox fund of the department of astronomy at the university of texas , and processors donated by advanced micro devices ( amd ) .
we are grateful to a. filippenko , d. leonard , and an anonymous referee for discussions and to g. hill for providing the igp which made this work possible .
explosion models are calculated using a one - dimensional radiation - hydro code ( hk96 ) that solves the hydrodynamical equations explicitly by the piecewise parabolic method ( colella & woodward 1984 ) .
nuclear burning is taken into account using an extended network of 216 isotopes ( thielemann , nomoto & hashimoto 1996 and references therein ) .
the propagation of the nuclear burning front is given by the velocity of sound behind the burning front in the case of a detonation wave and in a parameterized form during the deflagration phase based on detailed 3-d calculations ( e.g. khokhlov 1995 , 2000 ; niemeyer & hillebrandt 1995 ) .
we use the parameterization as described in dominguez & hflich ( 2000 ) .
we assume that @xmath150 , where @xmath151 and @xmath152 are the laminar and turbulent velocities with @xmath153}\ ] ] with @xmath154 and with @xmath155}\ ] ] and with @xmath156}\ ] ] here @xmath157 is the atwood number , @xmath158 is the characteristic length scale , and @xmath159 and @xmath160 are the densities in front of and behind the burning front , respectively .
the quantities @xmath161 and @xmath158 are directly taken from the hydro at the location of the burning front and for this choice of @xmath162 we take @xmath163 . the transition density is treated as a free parameter .
the description of the deflagration front does not significantly influence the final structure of the explosion .
the amount of matter consumed during burning ( and the total @xmath8ni production ) is governed by the pre - expansion of the wd and , consequently , is determined by the transition density @xmath97 , at which the burning front switches from the deflagration to the detonation mode ( hflich 1995a ) .
the value @xmath97 can be adjusted to produce a given amount of 56 which determines the brightness ( hflich 1995a ) and through the temperature dependence of the opacity , the decline rate ( hkw95 ; hflich et al .
1996 ) .
the code also simultaneously solves for the energy and radiation transport with variable eddington factors .
the radiation transfer portion of the code consists of ( i ) an lte radiation transfer scheme based on the time - dependent moment equations which are solved implicitly , ( ii ) a frequency - dependent radiation transport to determine the eddington factors and the frequency averaged opacities , ( iii ) a detailed equation of state with an elaborate treatment of the ionization balance and the ionization energies , ( iv ) time - dependent expansion opacities which take into account the composition structure of the explosion model , ( v ) photon scattering and thermalization calibrated by nlte - calculations , and ( vi ) a monte carlo @xmath164-ray deposition scheme which takes into account all relevant @xmath164-ray transitions and interaction processes . for more details ,
see hflich et al .
( 1999 ) , and references therein .
aspherical density structures are constructed from the spherical density distribution by imposing a homologous expansion function which depends on the angle @xmath44 from the equatorial plane and which conserves the total energy and the mass fraction per steradian from the spherical model ( hflich et al .
the spherical density structure is given by @xmath165}\ ] ] with r the initial distance of a mass element from the center .
homologous expansion of this density distribution is assumed with a scaling that depends on angle , i.e. , @xmath166}\ ] ] so that @xmath167}\ ] ] where @xmath168 is the time since the explosion and @xmath169 is the distance of the mass element after time @xmath168 .
because little is known about the general geometry of the ejecta , we construct ellipsoidal isodensity contours with an axis ratio @xmath170 at the photosphere where @xmath95 is the distance of the photosphere in the @xmath171 ( symmetry ) plane and @xmath96 is the distance in the ( axial ) @xmath172-direction .
this contour is given by @xmath173}\ ] ] the homology scaling constant , @xmath174 , is determined to produce the desired axis ratio for an ellipsoid ( hflich et al .
1999 ) , and the total explosion energy is normalized to that of the spherical model . here , the asphericities are small and the temperature profiles in sne ia are shallow ( hmk93 , hflich 1995a ) .
therefore , we assume identical isodensity and isotemperature contours .
our radiation transport code works both for spherical geometry and in three dimensions using different modules for the radiation transport . for details and references to the atomic database ,
see hflich ( 1995a ) and references therein .
the density and abundance structure is taken from the hydro calculations .
excitation due to @xmath164-rays is included via the monte carlo code . based on the list of atomic data of kurucz ( 1995 )
, we have constructed detailed atomic models for a couple of ions similar to those used in wwh97 .
the use of 3-d geometries caused some restrictions on the complexity of the atomic models .
multilevel atoms have been restricted to the main ionization species of ( 23/57 ) , ( 28/75 ) , ( 25/85 ) , ( 28/83 ) , ( 26/90 ) of the intermediate mass elements . here
, the numbers in the parentheses denote the number of levels and of line transitions , respectively .
for those ions , the statistical equations are solved consistently including bound - bound and bound - free transitions .
the method of accelerated lambda iteration is applied to remove the global dependence of the level populations and the radiation field . in order to include the line blocking effect , about 1,000,000 additional lines are taken into account under the assumption of local thermodynamical equilibrium for the population number but including scattering terms calibrated by the nlte - elements .
the electron temperature structure is based on the depth - dependent luminosity from the monochromatic light curve calculation at a certain time .
the complete system of equations is given by the time - independent radiation transport and the statistical equations , i.e. , the rate equation , the particle conservation equation , and the charge conservation equation .
all allowed bound - bound and bound - free transitions between the nlte levels are taken into account .
complete redistribution over each individual line both in frequency and in angle are assumed ( mihalas 1978 ) in the comoving frame ( but not the observer s frame ) .
this means that the populations of sublevels of the upper and lower transition are described by a maxwell - boltzman distribution and that the light becomes unpolarized .
the current calculations use a modified version of the monte - carlo code previously applied to calculate the continuum in sn 1987a ( h91 ) and line polarization in sn 1993j and sn 1996x ( hflich et al .
1995b ; wwh97 ) .
the code is capable of handling arbitrary 3-dimensional geometries , both for the density and the distribution of the sources .
polarization and flux spectra for rapidly expanding envelopes can thus be computed .
polarization is treated within the stokes formalism ( see , e.g. , van de hulst , 1957 ) .
we include electron scattering as a polarization process but omit scattering at dust particles .
the modifications for the present work to the previous versions of our monte carlo scheme ( h91 ) are mainly related to the inclusion of nlte - effects . in scattering dominated atmospheres , deviations from local thermodynamic equilibrium are also relevant at large optical depths ; however , at large optical depths , monte - carlo methods become very costly and/or inaccurate ( hflich et al .
therefore , we use a hybrid scheme of monte - carlo and non - equilibrium diffusion methods . in the latter case
, we implicitly solve the time dependent radiation transport equation in a non - equilibrium , diffusion approximation for 3-d geometry including the scattering and thermalization terms for the source functions and include the frequency derivatives in the formulation for the opacities and emissivities ( lucy & solomon 1970 ; karp et al .
1977 ) . at large optical depths
, this provides the solution for the full radiation transport problem .
we note that the use of a non - equilibrium diffusion does _ not _ imply that the mean intensity j is given by the black body field . to obtain the correction solution for the radiation transport equation at small optical depths ,
the difference between the solution of the diffusion and full radiation transport equation is calculated by a monte carlo method .
we calculate the difference between the solutions for computational accuracy and efficiency .
consistency between the solution at the outer and inner region is obtained iteratively .
the same monte carlo solver is used which has been applied to compute @xmath164-ray and polarization spectra ( e.g. hflich et al .
the monte carlo method is appropriate for this problem because of its flexibility with respect to the geometrical and velocity structures . the solution of the radiation transport and the statistical equations for the level populations are coupled .
a perturbation method is used to obtain consistency between the solutions ( hflich 1995a ) . in the equations for the microphysical quantities we express the actual mean intensity j@xmath175 by the following equation : here ,
the indices in brackets denote the iteration step , and @xmath177 and @xmath178 are the source function and the radiation transport matrices , respectively . the matrix @xmath179 is a band - matrix with elements corresponding to the diagonal and first off - diagonal elements of the complete matrix @xmath180 . the elements of @xmath181 are computed in the narrow line limit ( sobolev et al .
1957 ; hflich 1995a ) .
the use of a monte - carlo scheme introduces an additional complication due to the photon " noise . to achieve numerical stability for the solution in the rate equations , the deviation from the solution for @xmath182 obtained by the monte - carlo calculations and the non - equilibrium diffusion
are set to zero if they are less than the statistical noise .
this avoids instabilities during the model iterations when using the accelerated lambda iteration technique ; however , this also implies an increase of the computational expense with increasing rate of convergence . to increase the photon statistics , solutions of the monte - carlo calculation from previous iterations , @xmath183 ,
are taken into account with individual weight factors .
currently , we use a @xmath184 weight where @xmath185 is the total number of model iterations .
clearly , this weight function needs further fine tuning .
tests for realistic structures of the envelope show that the resulting error in the population numbers inherent to our procedure is comparable to those introduced by discretization errors if we solve the radiation transport equation in a difference scheme .
99 arbour , r. , papenkova , m. , li , w. d. , filippenko , a. v. , & armstrong , m. 1999 , iauc 7156 benz w. , cameron , a. g. w. , press , w. h. , & bowers , r. l. 1990 , apj , 348 , 647 bonanos , a. , garnavich , p. , schlegel , e. , jha , s. , challis , p. , kirshner , r. , hatano , k. , & branch , d. 1999 , aas , 195 , 3806 branch d. , & tammann g.a , 1992 , araa , 30 , 359 burstein , d. , & heiles , c. 1982 , aj , 87 , 1165 chandrasekhar , s. 1960 , radiative transfer , ( new york : dover ) collela , p. , & woodward , p.r .
1984 , j.comp.phys . , 54 , 174 cropper , m. , 1988 , mnras , 231 , 695 dominguez i. , & hflich p. 2000 , apj , 528 , 854 filippenko , a. v. , 1992 , aj , 104 , 1543 ( f92 ) gerardy , c.l , hflich p. , fesen r. , & wheeler j.c . 2000 , apj , in preparation goodrich , r. g. 1991 , pasp , 103 , 1314 hamuy , m. , 1994 , aj , 108 , 2226 hamuy , m. , 1996a , aj , 112 , 2391 hamuy , m. , phillips , m. m. , maza , j. , suntzeff , n. b. , schommer , r. a. , & avils , r. 1995 , aj , 109 , 1 hamuy , m. , phillips , m. m. , schommer , r. a. , suntzeff , n. b. , maza , j. , & avils , r. 1996b , aj , 112 , 2398 hernandez , i. , 2000 , mnras , submitted ( astro - ph/0007022 ) , p. 1991 ,
a&a , 246 , 481 ( h91 ) hflich , p. 1995a
, apj , 440 , 821 hflich , p. 1995b
, apj , 443 , 89 hflich , p. , dominik , c. , khokhlov , a. , mller , e. , & wheeler , j.c .
1996 , @xmath186 _ texas symposium on relativistic astrophysics _ , annals of the new york academy of science , 759 , 348 hflich , p. , & khokhlov , a. 1996 , apj , 457 , 500 ( hk96 ) hflich , p. , khokhlov , a. , & wheeler , j. c. 1995a , apj , 444 , 831 ( hkw95 ) hflich , p. , khokhlov a. , wheeler , j. c. , phillips , m. m. , suntzeff , n. b. , & hamuy , m. 1996 , apj , 472 , l81 , p. , mller , e. , & khokhlov , a. m. 1993 , a&a , 268 , 570 ( hmk93 ) hflich , p. , nomoto , k. , umeda , h. , & wheeler , j.c . 2000 , apj , 528 , 854 hflich , p. , wang , l. , wheeler , j. c. 1999 , apj , 521 , 179 hflich , p. , wheeler , j. c. , hines , d. , & trammell , s. 1995b , apj , 459 , 307 howell , d. a. 2000a , phd thesis , university of texas howell , d. a. 2000b , in preparation hough , j. h. , bailey , j. a. , rouse , m. f. , & whittet , d. c. b. 1987 , mnras , 227 , 1p hoyle , f. , & fowler , w. a. 1960 , apj , 132 , 565 iben , i. , & tutukov , a. v. 1984 , apjs , 54 , 335 iben , i. , tutukov , a. v. , & fedorova , a. 1998 , apj , 503 , 344 iwamoto , k. , brachwitz , f. , nomoto , k. , kishimoto , n. , umeda , h. , hix , w. r. , & thielemann , f. k. 1999 , apjs 125 , 439 jeffery , d. 1991 , apj , 375 , 264 ( j91 ) karp , a. h. , lasher , g. , chan , k. l. , & salpeter , e. e. 1977 , apj , 214 , 161 khokhlov , a. 1991 , a&a , 245 , 114 khokhlov , a. , mller , e. , & , p. 1993 , a&a , 270 , 223 khokhlov , a. 1995 , apj , 449 , 695 khokhlov , a. 2000 , apj , submitted ( astro - ph/0008463 ) kurucz , r.l .
1995 , cd-23 , center for astrophysics kawai , y. , saio , h. , & nomoto , k. 1987 , apj , 315 , 229 leibundgut , b. , 1993 , aj , 105 , 301 lentz , e. j. , baron , e. , branch , d. , hauschildt , p. , & nugent ,
p. 1999 , apj , 527 , 746 leonard , d.c . , filippenko , a. v. , & matheson , t. 2000a in _ cosmic explosions _
, eds . s. s. holt & w. w. zhang , ( melville , ny : aip ) , p. 165
leonard , d.c . ,
filippenko , a. v. , ardila , d. r. , & brotherton , m. s. 2000b , apj , submitted , astro - ph/0009285 li , w .- d . ,
filippenko , a. v. , treffers , r. r. , riess , a. g. , hu , j. , & qiu , y. 2000 , apj , submitted ( astro - ph/0006292 ) li , w .- d . ,
modjaz , m. , king , j. y. , papenkova , m. , johnson , r. a. , friedman , a. , treffers , r. r. , & filippenko , a. v. 1999 , iauc 7126 livne , e. 1999 , apj , 527 , l97 lucy , l.b . , & solomon , p.m. 1970 , apj 159 , 879 mahaffy , j. h. , & hansen , c. j. 1975 , apj , 201 , 695 marietta , e. , burrows , a. , & fryxell , b. 2000 , apjs , 128 , 615 mathewson , d. s. , & ford , v. 1970 , mem .
ras , 74 , 139 mathewson , d. s. , ford , v. , klare g. , neckel , t. , & krautter , j. 1978 , bull .
cds , 14 , 115 mazzali , p. a. , 1997 , mnras , 284 , 151 mccall , m. l. , reid , n. , bessell , m. s. , & wickramasinghe , d. 1984a , mnras , 210 , 839 mccall , m. l. , reid , n. , bessell , m. s. , & wickramasinghe , d. 1984b , mnras , 211 , 991 mihalas , d. 1978 , _ stellar atmospheres _ , ( san francisco : freeman ) miller , j. s. , robinson , l. b. , & goodrich , r. w. 1988 , in _ instrumentation for ground - based optical astronomy , present and future , _ ed .
robinson , ( new york : springer - verlag ) milne p. , the , l.s .
, & leising , m.d . 1999 ,
apjs 124 , 503 mochkovitch , r. guerrero , j. , & segretain , l. 1997 , in _ thermonuclear supernovae _ , eds .
ruiz - lapuente , canal , & isern 1997 , ( dordrecht : kluwer ) , pp .
187 mochokovitch , r. , & livio , m. 1990 , a&a , 236 , 378 modjaz , m. , 2000 , pasp , submitted ( astro - ph/0008012 ) mller , e. , & eriguchi , y. 1985 , a&a , 152 , 325 niemeyer , j. c. , & hillebrandt , w. 1995 , apj , 452 , 779 nomoto , k. 1980 , apj , 248 , 798 nomoto , k. , & iben , i. j. 1985 , apj , 297 , 531 nomoto , k. , & kondo y. 1991 , apj , 367 , l19 nomoto , k. , & sugimoto , d. 1977 , pasj , 29 , 765 nomoto , k. , thielemann , f. k. , & yokoi , k. 1984 , apj , 286 , 644 nugent , p. , baron , e. , hauschildt , p. , & branch . , d. 1997 , apj , 485 , 812 nugent , p. , phillips , m. m. , baron , e. , branch , d. , & hauschildt , p. 1995
, apj , 455l , 147 ( n95 ) paczyski b. 1985 , in _ cataclysmic variables and low - mass x - ray binaries _ , eds .
lamb & j. patterson , ( dordrecht : reidel ) p. 1
perlmutter , s. , et al .
1997 , apj , 483 , 565 perlmutter , s. , et al . 1999 ,
apj , 517 , 565 phillips , m. 1993 , apj , 413 , l105 rasio , f. a. , & shapiro , s. l. 1995 , apj , 438 , 887 riess , a. g. , et al . 1999 ,
aj , 117 , 707 riess , a. g. , press , w. h. , & kirshner , r. p. 1996 , apj , 473 , 88 ruiz - lapuente , p. , jeffery , d. j. , challis , p. , filippenko , a. v. , kirshner , r. p. , ho , l. c. , schmidt , b. p. , sanchez , f. , & canal , r. 1993 , nature 365 , 728 saffer , r. a. , livio , m. , & yungelson , l. r. , 1998 , apj , 502 , 394 saio , h. , & nomoto , k. 1985 , a&a , 150 , l21 saio , h. , & nomoto , k. 1998 , apj , 500 , 388 savage , b. d. , & mathis , j. s. 1979 , araa , 17 , 73 schaefer , b. e. 1996 , apj , 464 , 404 schlegel , d. j. , finkbeiner , d. p. , & davis , m. 1998 , apj , 500 , 525 schmidt , b. p. , et al .
1998 , apj 507 , 46 serkowski , k. , mathewson , d. , & ford , v. l. 1975 , apj , 196 , 261 shakhovskoi , n. m. 1976 , soviet astron .
lett . , 2 , 107 shapiro , p. r. , & sutherland , p. g. 1982 , apj , 263 , 902 simmons , j. f. l. , & stewart , b. g. 1985 , a&a , 142 , 100 sobolev , v. v. 1957 , sov .
astron . 1 , 297 spyromilio , j. , & bailey , j. 1993 , pasau , 10 , 263s thielemann f .- k . , nomoto k. , & hashimoto m. 1996 , apj , 460 , 408 trammell , s. r. 1994 , phd thesis , university of texas , pp . 246 - 260 tran , h. d. 1995 , apj , 440 , 578 treffers , r. r. , peng , c. y. , filippenko , a. v. , & richmond , m. w. , 1997 , iauc 6627 turatto , m. , 1996 , mnras , 283 , 1 turatto , m. , piemonte , a. , benetti , s. , cappellaro , e. , mazzali , p. a. , danziger , i. j. , & patat , f. 1998 , aj , 116 , 2341 umeda , h. , nomoto , k. , yamaoka , h. , & wanajo s. 1999 , apj , 513 , 861 van de hulst , h.c .
1957 , _ light scattering by small particles _ ,
( new york : wiley ) wang , l. , baade , d. , fransson , c. , , p. , & wheeler , j. c. 2000b , apj , submitted wang , l. , howell , d. a. , , p. , & wheeler , j. c. 2000a , apj , submitted ( astro - ph/9912033 ) wang , l. , & wheeler , j. c. 1996 , apj , 462 , l27 wang , l. , wheeler , j. c. , & hflich , p. 1997a
, apj , 476 , l27 ( wwh97 ) wang , l. , wheeler , j. c. , & hflich , p. 1997b
, in sn1987a , ten years after , eds .
m. m. phillips & n. suntzeff , ( new york : kluwer ) , in press wang , l. , wheeler , j. c. , li , z. , & clocchiatti , a. 1996 , apj , 467 , 435 webb , w. , malkan , m. , schmidt , g. , & impey , c. 1993 , apj , 419 , 494 webbink , r. f. 1984 , apj , 277 , 355 wheeler , j. c. , , p. , harkness , r. p. , & spyromilio , j. 1998 , apj , 496 , 908 white , r. l. 1979 , apj , 229 , 954 whittet , d. c. b. , martin , p. g. , hough , j. h. , rouse , m. f. , bailey , j. a. , & axon , d. j. 1992 , apj , 386 , 562 wolstencroft , r. d. , & kemp , j. c. 1972 , nature , 238 , 452 woosley , s. e. , & weaver , t. a. 1994 , apj , 423 , 371 woosley s.e . , & weaver t.a .
1986 , araa , 24 , 205 yamaoka h. , nomoto k. , shigeyama t. , & thielemann f. , 1992 , apj , 393 , 55 of @xmath187 g @xmath100 and @xmath188 g @xmath100 have been used to produce the subluminous and normally - bright models , respectively .
note that in the subluminous model , 56 is confined to a lower region of velocity space , beneath the optical photosphere at maximum light.,width=359 ] ) .
the model sn has been mapped into an oblate ellipsoid with axis ratio 1.17 shown for various inclinations @xmath116 and compared to the observations of sn 1999by at about maximum light in b ( see 2 ) . for comparison with the models ,
an interstellar component with @xmath189 and @xmath6 has been subtracted from the observations .
the observed flux spectrum is the solid black line .
model flux spectra are shown by solid magenta , dotted red , and dashed blue lines for inclination angles , @xmath116 , of 9.1 , 28.3 , and 52.1 , respectively . in the second panel , the raw polarization data is shown by a light blue line , while the smoothed data ( with @xmath190 ) is the black line .
the model polarization spectra are the same as the first panel .
the third panel presents the statistical error in @xmath1 for the raw data ( blue ) , smoothed data ( dotted blue ) , and theoretical model ( red ; binned to @xmath119 ) .
the model seen nearly equator - on ( @xmath191 ) does the best job of reproducing the general features of the polarization spectrum : depolarization from 4900 to 5500 , moderate polarization from 5500 to 6100 , a depolarization in the si 6150 feature , and a rising polarization to the red . in the bottom panel , the observed polarization angle @xmath192 is shown as a function of wavelength .
the unbinned data are given by dashed lines and the binned data by solid lines .
the data are shown before and after correcting for the isp contribution determined from the models ( see text ) .
after correction for isp is clear that the position angle is essentially constant across all wavelengths .
there is an @xmath193 shift across the trough feature , though this is an artifact of low polarization close to the origin giving an uncertain position angle .
compare this choice of isp to the empirically determined choice , figure [ quptbestisp].,width=359 ] but with @xmath194 and @xmath195 ( upper two panels ) and with @xmath196 and @xmath197 ( lower two panels ) in comparison with the observations of sn 1999by .
these models , particularly the model with @xmath198 do not match the data as well as the nearly edge - on geometry of figure [ model2].,width=359 ] lrrrrlrr sn1999by&1200&4:25&5/8/99&0&o&11.8&1.31sn1999by&1200&4:47&5/8/99&45&o&6.9&1.38sn1999by&1200&5:10&5/8/99&22.5&o&2.1&1.47 sn1999by&1200&5:32&5/8/99&67.5&o&-2.2&1.58 sn1999by&1200&6:03&5/8/99&0&o&-7.9&1.77 sn1999by&1200&6:25&5/8/99&45&o&-7.3&1.94 sn1999by&1200&6:47&5/8/99&22.5&o&-15.6&2.16 sn1999by&1200&7:08&5/8/99&67.5&o&-19.2&2.42 hd 154445&3&9:55&5/8/99&0&p&68.1&1.21hd 154445&3&9:56&5/8/99&45&p&67.8&1.21hd 154445&3&9:57&5/8/99&22.5&p&67.4&1.21hd 154445&3&9:58&5/8/99&67.5&p&67.1&1.22hd 192281&25&10:26&5/8/99&0&f&17.7&1.08hd 245310&240&3:02&5/9/99&0&p&-27.4&3.70hd 245310&240&3:07&5/9/99&45&p&-27.7&3.96hd 245310&240&3:14&5/9/99&22.5&p&-28.1&4.39hd 245310&240&3:19&5/9/99&67.6&p&-28.4&4.76sn 1999by&1800&4:22&5/9/99&0&o&11.6&1.31sn 1999by&1800&4:58&5/9/99&45&o&3.7&1.44sn 1999by&1800&5:31&5/9/99&22.5&o&-2.7&1.60sn 1999by&1800&6:04&5/9/99&67.5&o&-8.8&1.81sn 1999by&1800&6:40&5/9/99&0&o&-15.1&2.13sn 1999by&1800&6:56&5/9/99&45&o&-17.9&2.31sn 1999by&1800&7:12&5/9/99&22.5&o&-20.6&2.53sn 1999by&1800&7:28&5/9/99&67.5&o&-23.3&2.80hd 155528&120&8:49 & 5/9/99&0&p&85.3&1.22hd 155528&120&8:52 & 5/9/99&22.5&p&86.5&1.22hd 155528&120&8:55 & 5/9/99&45&p&87.6&1.22hd 155528&120&8:59 & 5/9/99&67.5&p&89.1&1.22bd + 40 4032&240 & 9:42&5/9/99&0&f&10.3&1.14sn1999by&1800&3:07&5/10/99&0&o&32.6&1.15sn1999by&1800&3:39&5/10/99&45&o&22.6&1.20sn1999by&1800&4:12&5/10/99&22.5&o&13.9&1.28sn1999by&1800&4:46&5/10/99&67.5&o&6.2&1.39sn1999by&1200&5:21&5/10/99&0&o&-0.8&1.54sn1999by&1200&5:42&5/10/99&45&o&-4.8&1.66sn1999by&1200&6:03&5/10/99&22.5&o&-8.6&1.80sn1999by&1200&6:24&5/10/99&67.5&o&-12.3&1.97bd + 22 3782&180&8:30&5/10/99&0&p&20.5&1.31bd + 22 3782&180&8:35&5/10/99&45&p&20.5&1.29bd + 22 3782&180&8:39&5/10/99&22.5&p&20.5&1.27bd + 22 3782&180&8:43&5/10/99&67.5&p&20.6&1.26bd + 25 3941&240&9:01&5/10/99&0&f&26.6&1.19 llcrrlcc 1972e&ngc 5253&b&n&30:&[email protected]&undetermined&11975n&ngc 7723&b&n:&0 : , 34:&1.5&isp&21981b&ngc 4536&b&n&56&[email protected]&undetermined&31983g&ngc 4753&s&n:&-2:&2.0&upper limit&41986g&ngc 5128&b&s&-9,-8&5.2&isp&51992a&ngc
1308&s&n&15:,100:&[email protected]&undetermined&61994d&ngc 4526&b&n&-10&[email protected]&undetermined&71994ae&ngc 3370&b&n:&@xmath199&0.3&upper limit&71995d&ngc 2962&b&n:&14 , 41&0.2&upper limit&71996x&ngc 5061&s&n&-7,30&[email protected]&maybe?&81997dt&ngc 7448&s&n:&@xmath200:&?&yes?&91998bu&m 96&b&n&-4,-3,-2&[email protected]&isp&101999by&ngc 2841&s&s&-2,-1,0&[email protected]&yes&11 | we present polarization spectra near maximum light for the strongly subluminous type ia supernova 1999by that show that the supernova is intrinsically polarized .
sn 1999by has an observed , overall level of polarization of @xmath0 0.3 to 0.8% , a rise of the polarization @xmath1 redward of 6500 , and a change in polarization across the 6150 feature of about 0.4% .
the presentation of the polarization at different wavelengths in the plane is shown to be a powerful tool to determine the overall geometry and the interstellar component .
the distribution of points with wavelength using this empirical plane method reveals that sn 1999by has a well - defined axis of symmetry and suggests an interstellar polarization ( isp ) vector with @xmath2 and position angle @xmath3 with an error circle in the plane of radius about 0.1% .
synthetic nlte - spectra for axisymmetric configurations based on delayed detonation models have been computed assuming ellipsoidal geometry .
the input ejecta structure and composition are based on a chandrasekhar - mass delayed - detonation model .
the parameters of the explosion are chosen to reproduce the time evolution of ir - spectra of sn 1999by without further adjustments .
spherical models are then mapped onto ellipsoidal geometries and the axis ratio , viewing angle and interstellar polarization adjusted to provide the best agreement with the polarization spectra .
both flux and polarization spectra can be reasonably well reproduced by models with an asphericity of @xmath4 observed equator - on .
the general properties of the polarization can be understood as a consequence of the structure of subluminous models .
best fits are obtained for the theoretical models with @xmath5 and @xmath6 , consistent with the empirical method .
we discuss our results for this subluminous type ia in the context of normally bright " type ia supernovae . for normally - bright type ia , the photosphere is near the inner iron - rich layers at maximum light and the ubiquitous iron lines give a rapid variation to the model polarization spectra . in subluminous models ,
the photosphere near maximum is in the silicon layers with fewer lines and a smoother overall polarization spectrum , as observed for sn 1999by .
though data are sparse , the low upper limits for polarization determined for many normal events in contrast to the high polarization in sn 1999by may suggest a relation between the asymmetry we observed and the mechanism that produces a subluminous type ia . among various mechanisms , rapid rotation of the progenitor white dwarf , or an explosion during a binary white dwarf merger process are likely candidates to explain the asphericity in sn 1999by .
h@xmath7 # 1= # 1 56@xmath8ni 56@xmath8co 56ni@xmath9 |
visible and near - ir spectroscopy of transiting extrasolar planets has led to the detection of na i , h@xmath2o , ch@xmath8 , co , and co@xmath2 @xcite in the atmospheres of extrasolar giant planets ( egps ) .
discoveries such as these have helped to shift the emphasis in the study of extrasolar planets from mere detection towards characterization of the dynamics and composition of their atmospheres .
modeling and observations of a large variety of objects allow us to constrain theories of the formation and evolution of different planets and their atmospheres . in this respect ,
studies of mass loss from the atmospheres of close - in egps and terrestrial planets are of particular interest .
in addition to being interesting in their own right , these studies are valuable for constraining models of planetary atmospheres during the early history of the solar system .
approximately 25% of the currently known extrasolar planets orbit within 0.1 au from their host stars .
at such small orbital distances many of these planets are subject to intense stellar irradiation .
in particular , the outermost layers of the atmosphere are strongly ionized and heated by the absorption of fuv and euv radiation , which leads to the formation of the thermosphere ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) .
modeling indicates that the thermospheres of close - in extrasolar giant planets are hot , greatly extended , and their composition is dominated by atomic constituents .
the same should apply to close - in super - earths orbiting m dwarf stars .
these are the closest types of object to an earth - like habitable planet whose atmospheres we may be able to probe with space - born or even ground - based telescopes in the near future ( e.g. , * ? ? ?
* ; * ? ? ?
as a consequence , the uv transit depths of extrasolar planets should be much larger than the visible and ir transit depths because the observations probe the upper atmosphere that strongly absorbs uv radiation in electronic resonance lines .
transit observations of extrasolar planets in the fuv were pioneered by @xcite ( hereafter vm03 ) and @xcite ( hereafter vm04 ) who detected atomic hydrogen ( h i ) , oxygen ( o i ) , and ionized carbon ( c ii ) in the upper atmosphere of the transiting planet hd209458b .
all of these detections were obtained by using the space telescope imaging spectrograph ( stis ) onboard the hubble space telescope ( hst ) .
vm03 used the g140 m grating to observe three transits of hd209458b in the wavelength range covering the stellar h lyman @xmath0 ( h ly@xmath0 ) emission line .
the spectral resolution of these observations was @xmath90.08 , which allowed for the details of the line profile to be resolved .
vm04 used the g140l grating with a low spectral resolution of @xmath92.5 to observe four transits in the wavelength range of [ 1180,1710 ] and detected absorption in the o i and c ii stellar emission lines .
vm03 deduced a h i transit depth of 15 @xmath10 4 % from the ratio of the flux in two wavelength regions around the core of the h ly@xmath0 line to the flux in the wings of the line during transit . based on this observation
, they concluded that h i overflows the roche lobe with a mass loss rate of @xmath11 10@xmath12 kg s@xmath13 .
they also suggested that the planet is followed by a cometary tail that is shaped by stellar radiation pressure acting on the escaping hydrogen .
later , @xcite ( hereafter bj07 ) and @xcite ( hereafter bj08 ) presented a thorough and convincing reanalysis of the same data .
based on this analysis , bj08 argued that there is no evidence for a cometary tail in the transit light curve or the detailed transit depth measurements .
further , he showed that the observed absorption may arise from atomic hydrogen below the roche lobe and that the h i absorption line profile is broadened by thermal and natural broadening in the thermosphere of the planet .
however , his results also imply that hydrogen overflows the roche lobe , and thus the atmosphere is still evaporating @xcite .
vm04 used the low resolution g140l data to obtain o i and c ii transit depths of 13 @xmath10 4.5 and 7.5 @xmath10 3.5 % , respectively .
they also calculated a full - width h i transit depth of 5 @xmath10 2 % from the unresolved h ly@xmath0 line and claimed that this transit depth is consistent with the much stronger absorption observed within a limited section of the line profile by vm03 .
further , they argued that the large o i and c ii transit depths are possible because these species overflow the roche lobe and the absorption lines are broadened by the velocity dispersion of the escaping gas .
thus the observations were interpreted as proof that the atmosphere is undergoing fast hydrodynamic escape .
@xcite ( hereafter bj10 ) published a thorough reanalysis of the g140l data .
they obtained revised full - width h i , o i , and c ii transit depths of 6.6 @xmath10 2.3 % , 10.5 @xmath10 4.4 % , and 7.4 @xmath10 4.7 % , respectively .
we note that these depths are only 2 , 1.93 , and 1.15@xmath14 , respectively , away from the fuv continuum transit depth of @xmath92 % .
the more detailed h i transit depth measurements reported by bj08 provide stronger constraints for the h ly@xmath0 line but no such constraints are available for the o i and c ii transits . in order to explain the large transit depths in these lines
, bj10 argued that oxygen and ionized carbon are preferentially heated to a temperature more than ten times higher than the temperature of hydrogen within a layer in the atmosphere located between @xmath92.25 r@xmath6 and the boundary of the roche lobe at 2.9 r@xmath6 .
most analyses of the hd209458b uv absorption signatures to date have either been limited to first order deductions , such as the effective size of the absorbing obstacle ( vm03,vm04 ) or been based on complicated first principle models for the atmosphere @xcite .
we believe that there is an important role for an intermediate class of models that satisfy some basic physical constraints , but parameterize aspects of the atmosphere that are difficult or impossible to predict accurately .
for example , it is well established that the thermosphere of hd209458b should be composed primarily of h and h@xmath15 , but the location of the transition from h@xmath2 to h is uncertain with different physical models making vastly different predictions @xcite . moreover
, the boundary between the atmosphere and interplanetary space is dependent upon the unknown strength of the stellar wind and the planetary and interplanetary magnetic field and has yet to be modeled in a realistic fashion . finally ,
although it seems well established that the temperature is of the order 10,000 k , the precise value depends on the heating efficiencies , which have yet to be calculated , and radiative cooling by minor species , which is not included in any of the models . because of these uncertainties it is important to analyze the data in a way that makes clear
what aspects of the atmosphere are constrained by the observations and which are not . in this paper
we introduce a generic methodology that can be used to interpret uv transit light curves in stellar emission lines .
we demonstrate this methodology by using a simple empirical model of the thermosphere to analyze the h i and o i transit depths of hd209458b summarized by bj10 .
in section [ sc : methods ] we introduce a model for calculating transit light curves for planets with extended atmospheres and discuss the basic features of the model thermosphere . in section [ sc : results ]
we discuss the h i transit depth measurements in detail and confirm that they _ can _ be explained by absorption by atomic hydrogen below the roche lobe .
nevertheless , we also show that the core of the h ly@xmath0 absorption line is optically thick up to the roche lobe and that the atmosphere is evaporating .
further , we demonstrate that the disagreement between bj08 and vm03 is due to differences in the treatment of the data and different definitions of the transit depth . in the rest of section [ sc : results ]
, we discuss the transits in the o i triplet lines and show that the empirical model thermosphere with a solar abundance of oxygen can be used to obtain transit depths that are statistically consistent with the observations .
we also address the feasibility of the idea that energetic oxygen atoms are present in the thermosphere of hd209458b and present alternative ways to explain o i transit depths that are comparable to or larger than the full - width h i depth .
we conclude section [ sc : results ] by discussing a variety of different models for the thermosphere of hd209458b and use them to calculate transit depths . in section [ sc : discussion ]
we discuss the feasibility of our assumptions and in section [ sc : conclusion ] we summarize our findings and conclusions .
in order to compare the observed uv transit depths with models of the thermosphere , we need to calculate the flux decrement of the stellar emission lines observed at earth at different times @xmath16 during the transit .
the wavelength - integrated transit depth , @xmath17 , is given by : @xmath18 where @xmath19 is the distance to earth , @xmath20 is the instrument response function , @xmath21 corresponds to the spectral resolution of the instrument , @xmath22 is the specific intensity of the star expressed as a function of position @xmath23 on the stellar disk of area @xmath24 , @xmath25 is the optical depth due to the planet and its atmosphere along the line of sight ( los ) from the host star to the observer , @xmath26 is the optical depth of the interstellar medium ( ism ) along the sightline to the star , and @xmath27 is the out - of - transit flux within @xmath21 observed at earth .
we note that interstellar absorption does not affect the transit depth unless the ism is optically thick . in that case
the transmitted flux is so small that the observations can not be used to calculate transit depths .
this affects the analysis of transit depths calculated from unresolved or partly resolved stellar emission lines .
the optical depth @xmath28 is a complicated function of time and position , and in general equation ( [ eq : integ_trans ] ) can not be integrated analytically . in order to integrate the equation
numerically , we adopted a cartesian coordinate system with the origin at the center of the stellar disk . at time @xmath16 , the coordinates of the center of the planet in this system are : @xmath29 where @xmath30 and @xmath31 are the absolute values of the @xmath32 and @xmath33 coordinates of the planet , respectively , @xmath34 is the distance between the centers of the star and the planet , @xmath35 is the inclination , and @xmath0 is the angle between the los to earth and the line joining the centers of the planet and the star _ in the orbital plane_. the @xmath33 coordinate of the planet at the center of the transit , @xmath36 , expressed in terms of the stellar radius @xmath37 , is known as the _
impact parameter_. the angle @xmath0 is related to the true anomaly @xmath38 of a given orbital position by @xmath39 where @xmath40 is the true anomaly at the center of the transit . if the orbit is circular , obtaining @xmath0 is straightforward . for an eccentric orbit
the distance @xmath34 and true anomaly @xmath41 as a function of time can be calculated by solving kepler s equation ( e.g. * ? ? ?
* ) . assuming that the atmosphere is spherically symmetric , the optical depth @xmath42 at frequency @xmath43 along a los with a tangent altitude @xmath44 from the center of the planet is given by : @xmath45 where @xmath46 is the number density of the absorbing species @xmath47 , @xmath48 is the absorption cross section , @xmath49 is pressure , @xmath50 is temperature , and @xmath51 is the distance from the center of the planet .
we note that @xmath49 and @xmath50 depend on the altitude @xmath51 and thus @xmath48 is a function of altitude . for individual absorption lines
the cross section is given by : @xmath52 where @xmath53 is the oscillator strength and @xmath54 is the voigt function .
the arguments of the voigt function are given by : @xmath55 where @xmath56 is due to natural broadening , @xmath57 is the laboratory frequency of the absorption line , @xmath58 is the mass of the absorbing species , and @xmath59 is the doppler broadening parameter .
the relationship between the tangent altitude @xmath44 and the @xmath32 and @xmath33 coordinates of a position on the stellar disk is given by : @xmath60 this relationship can be used to calculate @xmath44 for any point on the stellar disk and the los optical depth @xmath42 can then be calculated by integrating equation ( [ eq : tau ] ) .
if the atmosphere is not spherically symmetric , the problem of obtaining the los optical depth as a function of position on the stellar disk is much more complicated . in that case ,
equation ( [ eq : tau ] ) must be integrated separately for each latitude point , taking into account the longitudinal variation of the atmosphere along the los .
this technique may be relevant if , for instance , future observations of uv transit light curves indicate that the upper atmosphere deviates significantly from a spherical shape . as stated above , we integrated equation ( [ eq : integ_trans ] ) numerically in cartesian coordinates centered on the stellar disk .
we used simpson s rule to set up a sequence of one - dimensional integrals in the @xmath33 direction that cover the disk of the star in the @xmath32 direction .
this technique is iterative and the algorithm increases the number of points on the stellar disk until a certain predetermined numerical precision is achieved [ see section 4.6 of @xcite ] .
the transit depths in the uv are of the order of 110 % , and thus we required numerical errors less than 0.1 % in all of our simulations .
given a model atmosphere , equation ( [ eq : tau ] ) was integrated by using the method described by @xcite ( see section [ subsc : model_atmosphere ] below ) .
the optical depths were calculated for a fixed altitude grid of the model atmosphere , and the values were interpolated to the appropriate stellar disk coordinates by the iterative algorithm that was used to integrate equation ( [ eq : integ_trans ] ) .
voigt functions were evaluated numerically by using the method of @xcite .
model atmospheres based on hydrodynamics and photochemistry are complicated and time - consuming to use .
in addition , they are based on many uncertain assumptions that can not always be changed easily to match with the observations . in order to fit the data without the bias due to any individual model
, we have constructed a simple empirical model of the upper atmosphere of hd209458b .
this model is based on only a few free parameters that can be constrained by the generic features of the more complex models .
figure [ fig : model_atmos ] shows the structure of the upper atmosphere and highlights some of the important transition altitudes of the model .
the lower boundary of the thermosphere is located at the pressure of @xmath61 0.1 @xmath4bar , at an altitude @xmath62 1.1 r@xmath6 .
this choice is motivated by the fact that the stellar euv radiation , which causes intense heating and ionization of the upper atmosphere , is almost fully absorbed above the 1 @xmath4bar level and as a result the volume heating rate peaks near the 0.1 @xmath4bar level @xcite . at higher altitudes h@xmath2
is dissociated thermally and by dissociative photoionization .
this is also true of the more complex molecules , and at @xmath63 1.1 r@xmath6 the atmospheric constituents appear as atoms or atomic ions @xcite .
we assume that the lower atmosphere below @xmath64 is opaque to the fuv radiation considered in this study , and that the mean effective temperature is @xmath1 1300 k up to the 0.1 @xmath4bar level .
this temperature , together with the assumption of hydrostatic equilibrium , is used to constrain the vertical extent of the lower atmosphere .
we note that x rays and mid - uv radiation penetrate to the mbar levels , while the bulk of the visible radiation penetrates down to the @xmath91 bar level or slightly above it .
thus these types of radiation do not have a significant effect on the thermosphere where the conditions are largely determined by euv heating and ionization .
several different models suggest that the temperature of the thermosphere is @xmath65 10,000 k ( e.g. , * ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
consequently , the pressure scale height is relatively large , reaching values of @xmath66 km .
this implies that the thermal escape parameter @xmath67 13 , and thus several authors have suggested that the upper atmosphere of hd209458b is undergoing fast hydrodynamic escape ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
despite this possibility we assume that the density profile is hydrostatic .
this assumption is justified because the density profile is approximately hydrostatic up to the critical level at @xmath68 3 r@xmath6 even when the atmosphere is escaping at supersonic speed at higher altitudes .
based on the ratio of the vertical advection term to the pressure gradient term in the momentum equation , the outflow velocity must be nearly equal to or faster than the speed of sound for the departure from hydrostatic equilibrium to be significant globally .
most models indicate that the mass loss rate from the atmosphere of hd209458b is limited to @xmath69 10@xmath1210@xmath70 kg s@xmath13 by the available euv energy .
the pressure at @xmath71 3 r@xmath6 in the empirical model shown in figure [ fig : model_atmos ] is @xmath72 0.04 nbar , which implies that the vertical velocity at the critical level is @xmath73 110 km s@xmath13 . due to the conservation of mass ,
the vertical velocity is inversely proportional to density and thus it decreases steeply with decreasing altitude .
for example , at @xmath3 1 nbar , which corresponds to the altitude of @xmath7 1.7 r@xmath6 , the vertical velocity required to support mass loss is only @xmath73 30300 m s@xmath13 .
such a slow vertical velocity does not cause a significant deviation from hydrostatic equilibrium in the thermosphere ( e.g. , * ? ? ?
. given a hydrostatic density profile , we integrated equation ( [ eq : tau ] ) numerically .
for the purposes of the numerical solution , we created a grid with an altitude spacing of 0.01 h@xmath6 for the thermosphere .
the number density @xmath46 of species @xmath47 at altitude @xmath74 corresponding to the grid point @xmath75 is simply given by : @xmath76 where @xmath77 is the volume mixing ratio of species @xmath47 , and @xmath78 is the total number density .
the thermal escape parameter , @xmath79 , is given by : @xmath80 where @xmath81 and @xmath82 are the mean molecular weight and temperature , respectively , at a lower grid point @xmath83 . in order to match the observations we treated the number density at the lower boundary , @xmath84 , the mean temperature of the thermosphere , @xmath50 , and the mixing ratios @xmath77 as free parameters . in practice , we used @xmath85 and @xmath50 to constrain @xmath84 , and assumed a multiple of the solar composition to fix @xmath77 . the resulting mixing ratios of h , he , c , o , and n
were used to calculate the mean molecular weight @xmath86 .
the upper boundary of the model is at an altitude @xmath87 above which the atmosphere is mostly ionized .
the altitude @xmath87 need not be the same for different species and generally it depends on the ionizing flux , photoabsorption cross sections , photochemical reaction rates , and the outflow velocity .
we assumed that the the column densities of the neutral species are negligible above @xmath87 and treated it as a free parameter for each species .
we also assumed uniform mixing throughout the model atmosphere and that diffusive separation does not take place in the thermosphere .
this assumption can be justified because the large pressure scale height in the atmosphere of hd209458b may lead to efficient turbulent mixing and implies that the exobase is located at a relatively high altitude .
in general , diffusive separation occurs when the coefficient of turbulent ( eddy ) diffusion , @xmath88 , is smaller than the coefficients of molecular diffusion , @xmath89 , for different species .
the coefficient of eddy diffusion can be estimated crudely as @xmath90 ( e.g. , * ? ? ?
* ) , where @xmath91 is the characteristic velocity of turbulence .
the pressure scale height in the extended thermosphere of hd209458b is @xmath92 10@xmath93 m and consequently it is possible that k@xmath94 10@xmath93 m@xmath95 s@xmath13 .
the coefficients of molecular diffusion reach similar values at @xmath96 20 nbar @xcite indicating that uniform mixing can take place up to the nbar levels .
drag forces between the escaping hydrogen and minor species can also lead to uniform mixing of the thermosphere . whether or not a heavier atom moves along with hydrogen
depends on its mass .
if the mass of the atom is smaller than the _ crossover mass _ @xmath97 , the drag forces are strong enough to prevent diffusive separation .
approximately , the crossover mass is given by @xcite : @xmath98 where @xmath99 is the mass of atomic hydrogen , @xmath100 is the escape flux , and @xmath31 is the binary diffusion coefficient for hydrogen and the heavier species . assuming a mass loss rate is @xmath69 10@xmath1210@xmath70 kg s@xmath13 and a binary diffusion coefficient of @xmath101 1.6 @xmath102 10@xmath103 m@xmath13 s@xmath13 , we obtain the crossover mass of @xmath104 11102 .
this implies that drag forces affect the mixing ratios of oxygen and carbon atoms , and could explain a significant abundance of these atoms in the thermosphere .
in this section we investigate the degree to which the existing uv transit observations of hd209458b are consistent with the simple atmospheric model described in section [ subsc : model_atmosphere ] .
we begin by describing the fitting of the model to the transit depth measurements covering the full width of the stellar h ly@xmath0 emission line profile in the wavelength range of [ 1212,1220 ] .
we note that vm04 obtained a line - integrated absorption depth of 5.3 @xmath10 1.8 % in this wavelength range while bj10 obtained a higher depth of 6.6 @xmath10 2.3 % . in order to calculate the depth of the transit
, we used equation ( [ eq : integ_trans ] ) together with the absorption line parameters listed in table [ table : spectra ] .
we simulated the stellar emissions by using the h ly@xmath0 line profile of hd209458b calculated by @xcite ( hereafter w05 ) .
this line profile is based on a high resolution spectrum obtained with the stis e140 m grating and it was generated by reversing the absorption of the line profile by the ism .
we fitted a sum of three gaussian distributions to the pre - ism line profile to represent the self - reversed core and the broad wings of the line .
we then scaled the line - integrated flux to match with the observations of hd209458b analyzed by bj07 after accounting for interstellar absorption .
lccccc hi ly@xmath0 & 1215.67 & 4.164 @xmath102 10@xmath13 & 6.265 @xmath102 10@xmath70 & 0.087 & 5.87 @xmath102 10@xmath105 +
di ly@xmath0 & 1215.34 & 4.165 @xmath102 10@xmath13 & 6.270 @xmath102 10@xmath70 & 0.062 & 6.23 @xmath102 10@xmath105 + oi & 1302.17 & 4.887 @xmath102 10@xmath106 & 3.204 @xmath102 10@xmath70 & 0.023 & 2.95 @xmath102 10@xmath105 + oi * & 1304.86 & 4.877 @xmath102 10@xmath106 & 1.911 @xmath102 10@xmath70 & 0.024 & 2.95 @xmath102 10@xmath105 + oi * * & 1306.03 & 4.873 @xmath102 10@xmath106 & 6.352 @xmath102 10@xmath12 & 0.024 & 2.95 @xmath102 10@xmath105 + [ table : spectra ] most of the flux contained within the line profile is absorbed by the ism , and thus it is important to make sure that the model fluxes are properly attenuated before the line - integrated transit depth is calculated .
w05 measured the column density of hydrogen towards hd209458 and it is @xmath107 2.3 @xmath102 10@xmath103 m@xmath106 .
they reported a doppler broadening parameter along this line of sight of @xmath108 12 km s@xmath13 .
this doppler parameter is typical of the local interstellar medium ( lism ) where the temperature is thought to vary between @xmath109 7,00012,000 k @xcite .
the column density , on the other hand , is relatively low and indicates that the sightline to hd209458 is in a low - density direction of the lism .
we calculated the extinction as a function of wavelength by using these parameters and adopting a voigt function to simulate thermal and natural broadening in the ism .
figure [ fig : ism ] illustrates the effect of interstellar absorption on our model line profile .
we obtained the observed full - width h i transit depth of 6.6 % by adopting a mean temperature of @xmath1 11,000 k and assuming that most of the hydrogen in the atmosphere of hd209458b is ionized above @xmath110 2.9 r@xmath6 . the rest of the free parameter values for this best - fitting empirical model , labeled m1 , and the other models discussed in this paper are listed in table [ table : models ] . in order to illustrate the appearance of the transit in real observations , we convolved the data with the line - spread function ( lsf ) appropriate for the stis g140 m grating
. figure [ fig : synthetic_spectrum ] shows the resulting synthetic stellar h ly@xmath0 line in and out of transit .
the location of the upper boundary in the m1 model agrees with models of the thermosphere , which suggest that the h / h@xmath15 ionization front is located between @xmath111 1.55 r@xmath6 @xcite .
it is also close to the boundary of the roche lobe in the direction perpendicular to the los at @xmath112 2.9 r@xmath6 .
further , the pressure at @xmath87 is @xmath3 0.04 nbar , which is comparable to the pressure of the solar wind scaled to @xmath113 0.047 au .
thus , if the planet lacks a significant magnetic field ( e.g. * ? ? ?
* ) , the upper boundary of the empirical model may be close to the ionopause .
the optical depth profiles for the m1 model are shown in figure [ fig : opdepth ] for the core of the line at @xmath114 1215.67 and for the wing of the line at @xmath114 1215.2 , which corresponds to @xmath115 116 km s@xmath13 in the stellar reference frame .
we agree with bj10 that due to thermal and natural broadening , and large column densities along the lines of sight through the atmosphere , the optical depth in the wings of the line profile is significant even in the absence of actual bulk flows towards or away from the observer .
the core of the line is optically thick up to and beyond the upper boundary while the wing is optically thick up to the altitude of @xmath7 2 r@xmath6 .
the latter altitude corresponds to a transit depth of @xmath95.4 % .
ccccccccccc m1 & 11,000 & 2.92 & 0.1 & 6.6 @xmath102 10@xmath116 & 1 & 0.0 & 5.8 @xmath102 10@xmath117 & 6.6 & 4.3 & 1.9 + m2 & 11,000 & 2.92 & 0.1 & 6.6 @xmath102 10@xmath116 & 1 & 10 & 5.8 @xmath102 10@xmath117 & 6.6 & 5.8 & 1.9 + m3 & 11,350 & 2.92 & 0.1 & 6.4 @xmath102 10@xmath116 & 5 & 0.0 & 5.7 @xmath102 10@xmath117 & 6.6 & 5.0 & 1.9 + m4 & 12,000 & 2.96 & 0.1 & 6.0 @xmath102 10@xmath116 & 10 & 0.0 & 5.4 @xmath102 10@xmath117 & 6.6 & 5.5 & 1.9 + m5 & 15,950 & 3.08 & 0.1 & 4.5 @xmath102 10@xmath116 & 40 & 0.0 & 4.0 @xmath102 10@xmath117 & 6.6 & 7.1 & 2.2 + m6 & 11,000 & 2.92 & 0.001 & 6.6 @xmath102 10@xmath118 & 1 & 0.0 & 6.1 @xmath102 10@xmath119 & 2.3 & 2.3 & 8.9 + m7 & 8250 & 2.72 & 1 & 8.8 @xmath102 10@xmath120 & 1 & 0.0 & 5.4 @xmath102 10@xmath121 & 6.6 & 3.9 & 1.8 + k10 & 10,280@xmath122 & 2.92 & 10 & 6.6 @xmath102 10@xmath123 & 1 & 0.0 & 6.8 @xmath102 10@xmath117 & 5.2 & 3.7 & 3.1 + y04 & 12,170@xmath122 & 3.0 & 200 & 1.9 @xmath102 10@xmath119 & 1 & 2.2@xmath124 & 9.9 @xmath102 10@xmath103 & 2.9 & 2.7 & 7.3 + [ table : models ] figure [ fig : transmission ] shows transmission of the stellar h ly@xmath0 emission during transit as a function of wavelength based on the m1 model .
it also shows the transmission data points from figure 6 of bj08 .
the m1 model is consistent with @xmath911 % absorption within the core of the line profile .
we calculated @xmath125 1.9 for our fit to the data points .
we note that bj10 obtained nearly the same value by using the div1 model of @xcite ( hereafter gm07 ) after multiplying the density profiles of that model by a factor of 2/3 .
the results indicate that our fit to the data is as good as that obtained by bj10 .
this is not surprising because in both the m1 empirical model and the div1 model the mean temperature is @xmath126 10,000 k and the density profiles are approximately hydrostatic in the relevant altitude range of @xmath7 1.083.0 r@xmath6 .
figure [ fig : light_curve ] shows the m1 model light curve as a function of time from the center of the transit compared with the observed data points from figure 2(a ) of bj07 . in order to calculate this light curve , we used the wavelength limits specified by bj07 and bj08 .
the transit depth at the center of the transit is @xmath96.6 % , which is slightly lower than that given by bj07 but the difference is not statistically significant .
the symmetry of the light curve and a good fit by our model indicate that there is no observational evidence that the extended thermosphere of hd209458b deviates significantly from a spherical shape .
this is in contrast to the findings of vm03 who , based on the asymmetry of their light curve , argued that the planet is followed by a curved cometary tail of escaping hydrogen .
next we explore whether the full width transit depth of 57 % is consistent with the larger depth of 815 % measured over a more limited wavelength range within the line profile ( vm03,bj07,bj08 ) .
vm03 obtained the depth of 15 @xmath10 4 % by comparing the ratio of the integrated flux within the wavelength intervals of [ 1215.15,1215.5 ] and [ 1215.8,1216.1 ] ( the blue and red sides of the in region , respectively ) to the flux within the intervals of [ 1214.4,1215.15 ] and [ 1216.1,1216.8 ] ( the blue and red sides of the out region , respectively ) obtained during the transit . in their analysis , the interval [ 1215.5,1215.8 ] was excluded because contamination by the geocorona and sky background makes the flux measurements in this region unreliable .
bj07 redefined all of these wavelength intervals and obtained the transit depth of only 8.9 @xmath10 2.1 % . in his analysis
the ` contamination window ' was extended to [ 1215.41,1215.84 ] and the blue and red intervals were defined as [ 1214.83,1215.36 ] and [ 1215.89,1216.43 ] , respectively .
another significant difference between the analysis of vm03 and bj07 is the treatment of the time variability of the h ly@xmath0 emissions , either due to stellar variability or noise .
bj07 derived unperturbed h ly@xmath0 profiles by merging the subspectra of a dense time series that was corrected for the transit trend . in the resulting line profiles ( figure 3 in bj07 ) ,
the time variability has been reduced to @xmath91 % level . vm03 , on the other hand , attempted to account for time variability by calculating the transit depth based on the core - to - wing flux ratio instead of evaluating the transit depth based on the ratio of the flux obtained during transit to the flux obtained outside of the transit .
this approach relies on the assumption that the transit depth in the wings of the line is not significant and that the variability in the wings is similar to the variability within the core of the line .
@xcite ( hereafter vm08 ) correctly pointed out that a comparison between the two approaches is not fair because bj07 used different wavelength limits and instead of evaluating the change in the core - to - wing ( in / out ) flux ratio , he calculated the true transit depth i.e. , the flux decrement based on equation ( [ eq : integ_trans ] ) .
however , we believe that the transit depths calculated by vm03 are not necessarily related to the size of the absorbing cloud while those calculated by bj07 and bj08 are . by using the m1 model and the vm03 and bj07 wavelength limits ,
we obtain absorption depths of 2.4 % and 1.3 % , respectively , based on the in / out flux ratio and true transit depths of 7.5 % and 6.6 % , respectively .
our model is a good fit to the transmission curve of bj08 and the behavior of the in / out ratio is roughly in line with the line profiles shown in figure 3(a ) of bj07 and yet the transit depth based on the in / out flux ratio is small .
we note , however , that the m1 model overflows the roche lobe and thus we agree with vm08 that the atmosphere is evaporating .
the large transit depths in the o i and c ii lines reported by vm04 and bj10 can potentially be used to constrain the degree of eddy mixing and the composition of the upper atmosphere .
the o i transit depth , if accurately measured , is a good indicator of the proportion and mixing of heavy elements in the thermosphere .
the interpretation of the c ii transit depth , on the other hand , is much more complicated because of the uncertainties related to the ionization profiles , photochemistry and , if c ii is escaping the thermosphere , the interaction of the atmosphere with the stellar wind or magnetospheric plasmas . statistically , the observed c ii transit depth is also the least significant detection out of the three currently available uv transit depth measurements . for these reasons ,
we limit our discussion and analysis to the interpretation of the o i transit depth .
vm04 reported an integrated transit depth of 12.8 @xmath10 4.5 % in the wavelength range of [ 1300,1310 ] , which covers the o i triplet emission lines listed in table [ table : spectra ] .
bj10 reported a revised and slightly shallower transit depth of 10.5 @xmath10 4.4 % .
these transit depths are only 2.4 and 1.9 @xmath14 , respectively , away from the fuv continuum transit depth of @xmath92 % .
the data were obtained by using the stis g140l grating with a spectral resolution of @xmath92.5 , which does not allow for the individual lines of the triplet to be resolved .
the lines were resolved , however , in an earlier spectrum obtained by vm03 with the e140 m echelle grating . although this spectrum was not used to analyze the transit depths , it is a good reference point for the expected line profiles and can be used to constrain the properties of the ism .
we modeled the emission spectrum of the o i triplet by using the full disk solar line profile parameters derived by @xcite and fitted the line - integrated fluxes to the observations .
as a result we obtained surface fluxes of 3.0 , 3.2 , and 3.4 w m@xmath106 for the o i , o i * , and o i * * lines of hd209458 , respectively . in order to correct the model fluxes for absorption by interstellar oxygen , we assumed that the o i / h i ratio in the lism is @xmath93 @xmath102 10@xmath127 @xcite and that all of the interstellar oxygen is in its ground state .
the latter assumption is consistent with the ratios of the individual emission lines in the high - resolution e140 m spectrum .
bj10 used the e140 m data points to deduce fluxes for the o i , o i * , and o i * * lines that scale as 1:1.5:1.17 .
the solar line - integrated fluxes of these lines scale as 1:1.06:1.16 . after applying interstellar absorption to the solar line profiles
, we obtained model fluxes that scale as 1:1.43:1.55 .
the close agreement between the model and observed o i / o i * flux ratios confirms that most of the interstellar oxygen is in its ground state and that our parameters for the lism are reliable .
the observed o i**/o i flux ratio , on the other hand , is slightly lower than the model flux ratio .
this can not be due to absorption in the ism because a population inversion in the excited levels is unlikely .
thus the disagreement , if real , arises from the intrinsic properties of hd209458 and implies that the stellar o i triplet flux ratios differ slightly from the corresponding solar flux ratios .
we note , however , that the e140 m spectrum is noisy and it is not clear if the difference is real or if it arises from uncertainties in the data . the flux in the core of the o i ground state line is absorbed by the ism , and a significant o i transit depth is possible only if the populations of the excited states of oxygen in the thermosphere of hd209458b are sufficient .
we estimated the relative population of the ground state and the excited states by assuming that the thermosphere is in lte and by using the electronic partition function calculated by @xcite . at a temperature of @xmath1 10,000
k we obtained population ratios of 5.2:3.0:1 for the ground state , lower excited level , and upper excited level , respectively .
these ratios agree closely with the statistical weights of the levels that give rise to the triplet .
the m1 model presented in table [ table : models ] yields an o i transit depth of 4.3 % .
this is similar to the transit depth calculated by bj10 who used the scaled density profiles from the div1 model of gm07 .
again , the agreement between these models is not surprising because the temperature and density profiles are similar and both models assume solar o / h ratios .
we note that the o i transit depth based on the m1 model is within @xmath91.4@xmath14 from the observed depth . assuming gaussian statistics , the probability that the true transit depth is between 1.76.1 % ( 21@xmath14 ) is @xmath914 % . due to the relatively large uncertainty in the data points ,
this probability is significant . in other words ,
the deviation of the model transit depth from the observed value is not statistically significant .
better observations with higher s / n and spectral resolution are required to constrain the transit depth further .
our analysis shows that the stis g140 m grating , rather than the g140l grating , could already be used to obtain better data .
bj10 argued that the o i transit depth of 10.5 % is possible if oxygen is preferentially heated to t@xmath128 100,000 k ( 8.6 ev ) within a layer located between @xmath7 2.25 r@xmath6 and @xmath129 2.9 r@xmath6 .
it is true that hot oxygen populations have been observed in the thermospheres of terrestrial planets ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) but these populations are formed mostly by dissociative recombination of o@xmath130 , which should not be important on hd209458b .
other possible mechanisms for heating oxygen atoms include collisional processes that , as we demonstrate in the next section , would also imply that oxygen is mostly ionized .
thus it is not clear if a neutral cloud at these temperatures is physically possible .
as we already pointed out , the o i transit depth of 4.3 % based on the m1 model is consistent with the observations .
however , the measurements also allow for a larger transit depth and it is thus worth exploring if transit depths close to 10.5 % can be explained by changing the parameters of the model atmosphere , the host star , or the interstellar medium . in the following
we investigate several different possibilities that can give rise to larger o i transit depths without the need to introduce energetic oxygen atoms in the thermosphere .
theoretically , the o i transit depth is sensitive to interstellar absorption . in order to test
if changes in the abundance of interstellar oxygen can increase the calculated transit depth , we assumed that the column density of interstellar oxygen towards hd209458b is negligible . as a result ,
the calculated transit depth increased only by fractions of a percent .
we conclude that , contrary to h i , the full - width o i transit depth of hd209458b is not very sensitive to interstellar absorption . in addition , the e140 m spectra indicate that our ism parameters are accurate and the assumption of a negligible oxygen column would not be reasonable anyway .
consequently , we must explore the properties of the host star and the planet in order to explain the possibility of large o i transit depths .
transit depths in the visible and the ir wavelengths are sensitive to stellar activity and can appear larger if a significant portion of the stellar disk is covered by starspots even if these spots do not actually fall in the path of the transit ( e.g. , * ? ? ?
this is because the effective emitting area of the star appears smaller . in order to crudely estimate the magnitude of this effect , we defined an _ effective _ fractional area , @xmath131 , of the quiet regions as the ratio of the area that would be required to emit zero flux to account for the diminished luminosity during quiet periods to the area of the stellar disk . in terms of @xmath131 , the observed transit depth , @xmath132 ,
can be written as : @xmath133 where @xmath134 is the radius of the transiting planet , and @xmath135 is the radius of the star .
we note that @xmath131 would have to be @xmath960 % for the o i transit depth to increase from 4.3 % to 10.5 % . because the definition of @xmath131 implies that the _ actual _ fractional area of the quiet regions would have to be much larger than 60 % , this is clearly not realistic .
in fact , the o i triplet flux of the sun varies by @xmath930 % during the solar cycle @xcite , and thus we would expect an enhancement at most by 0.7 percentage points ( even this is probably overestimated because the relationship between the flux variability and @xmath131 is complicated ) .
this is not statistically significant and we conclude that the interpretation of the low - resolution transit depths is not affected by the distribution of active or quiet regions .
the transit depth measurements are also affected by stellar limb darkening or limb brightening . according to the solar disk observations of the o i triplet fluxes reported by @xcite , the fluxes are roughly constant between @xmath136 10.2 , with only negligible limb darkening . at @xmath137 0.2 the fluxes are brighter than within the center of the disk by a factor of @xmath91.1 .
if the atmosphere of the transiting planet does not cover the bright ring surrounding the stellar disk , limb brightening actually makes the transit depth appear shallower . we simulated limb brightening by dividing the stellar disk into a central disk region and a bright ring , and neglected limb darkening within the central region . as a result
, we found that limb brightening does not lead to a detectable change in the low - resolution o i transit depth .
several authors have attempted to explain the observed o i transit depth by referring to the velocity dispersion arising from radial outflows .
vm04 invoked this explanation and argued that hydrodynamically escaping oxygen atoms moving at sonic or supersonic velocities could give rise to a transit depth of 13 % .
later , @xcite argued that the velocity dispersion arising from similar supersonic outflows could also explain the large h i transit depth reported by vm03 .
they assumed that the doppler broadening parameter is given by : @xmath138 where @xmath139 is the velocity dispersion . assuming that @xmath140 10 km s@xmath13 , and using the m1 model as a base , we obtain h i and o i transit depths of 6.7 and 7.7 % , respectively . chi - squared for the h i transmission fit in this case is @xmath125 2.1 , which indicates that the h i transit depth is not significantly affected by additional velocity dispersion .
this is because a large portion of the h ly@xmath0 line is already saturated . on surface
, then , velocity dispersion appears to be a promising way to explain the large o i and c ii transit depths .
equation ( [ eqn : broadening ] ) , however , is appropriate for simulating broadening due to _ microturbulence_. it is only accurate if the scale of the turbulence is much smaller than the mean free path .
further , what really matters is not radial velocity but the projected velocity along the los . because most of the absorption during transit is due to the layers near the terminator , radial velocity is not a good measure of the true velocity dispersion . in order to estimate the velocity dispersion
properly , we calculated the appropriate velocities for each layer along each los and used these velocities in integrating equation ( [ eq : tau ] ) . for each layer
@xmath75 along a los , the parallel velocity @xmath141 is given by : @xmath142 where @xmath143 is the radial velocity , and @xmath44 is the constant tangent altitude of the los . assuming a uniform radial velocity of @xmath144 10 km s@xmath13 for the m1 model , the maximum los velocity , obtained at the upper boundary of the model along a sightline through the lower boundary is @xmath1459.2 km s@xmath13 . for a los at @xmath7 2 r@xmath6 ,
the corresponding velocity is only @xmath1466.8 km s@xmath13 .
the m2 model ( see table [ table : models ] ) includes a uniform radial velocity of @xmath73 10 km s@xmath13 .
figure [ fig : profiles ] shows the transmission profile of the o i * line for this model together with a corresponding profile for the m1 model and another profile calculated by using equation ( [ eqn : broadening ] ) .
the absorption profiles differ significantly . by using equation ( [ eqn : broadening ] )
we obtain a much broader absorption profile with a slightly smaller central cross section compared to the m1 model .
if the velocity dispersion is treated properly , as it is in the m2 model , the absorption profile exhibits two symmetric peaks around the central wavelength of the o i * line at @xmath147 1304.86 .
the resulting h i and o i transit depths based on the m2 model are 6.6 and 5.8 % , respectively . a radial velocity of @xmath73 10 km s@xmath13 is not enough to explain an o i transit depth of 10.5 % but it can explain o i transit depths that are comparable to the h i transit depth even when the abundance of oxygen is significantly lower than that of hydrogen .
it is not realistic , however , to assume that @xmath73 10 km s@xmath13 in the lower thermosphere .
hydrodynamic models of the thermosphere indicate that the base of the planetary wind is located near @xmath3 1 nbar , which corresponds to an altitude of @xmath7 1.7 r@xmath6 in the m1 model . assuming that the outflow is effective above this altitude
, we obtain an o i transit depth of only 5.5 % .
consequently , velocity dispersion can only affect the transit depth significantly if the optical depth above the critical level of @xmath148 3 r@xmath6 is significant .
it should be noted that radial velocities higher than @xmath73 10 km s@xmath13 do not necessarily lead to larger transit depth because the absorption is limited by the fact that the central cross section in equation ( [ eq : sigma0 ] ) is proportional to @xmath59 .
in fact , if we introduce a radial velocity of @xmath73 60 km s@xmath13 to the m1 model , we obtain an o i transit depth of only 5.2 % . within the velocity range of @xmath73 1060 km s@xmath13
, we obtain a maximum o i transit depth of 7.2 % by assuming that @xmath73 30 km s@xmath13 .
thus fast hydrodynamic escape of neutral oxygen from the thermosphere alone can not explain an o i transit depth of 10.5 % .
the next logical step is to investigate if the o / h and c / h ratios in the thermosphere of the planet are actually higher than the solar values .
bj10 did not consider this possibility because the density profiles of their best - fitting model are already enhanced with respect to the y04 model .
with respect to the div1 model of gm07 , however , their density profiles are actually depleted and further , the o / h and c / h ratios of their model are not enhanced but solar .
we generated new empirical models m3 , m4 , and m5 ( see table [ table : models ] ) by increasing the o / h and c / h ratios of the m1 model by factors of 5 , 10 , and 40 , respectively .
initially , this caused both the h i and o i transit depths to decrease because the higher mean molecular weight of the thermosphere led to a shorter pressure scale height .
in fact , we found that it is not possible to increase the abundance of the heavier atoms without also increasing the mean temperature of thermosphere to recover the fit to the h i transit depth . in order to do so , we increased the mean temperatures of the m4 and m5 models to @xmath126 12,000 k and 16,000 k , respectively .
the m3 model did not require a significantly increased temperature .
the mean temperature of the thermosphere can be constrained by two additional means .
first , the temperature of the neutral atoms is limited by thermal ionization . in lte conditions
the thermal ionization balance is determined by the saha equation , which implies that atomic hydrogen is fully ionized at temperatures higher than @xmath1 70008000 k. in non - lte conditions of the thermosphere the ion / neutral fractions resulting from thermal ionization are determined by a balance between impact ionization by thermalized electrons and radiative recombination .
the reaction rate coefficients of these processes , respectively , for atomic hydrogen are @xcite : @xmath149 where @xmath150 is the electron temperature and @xmath151 is the ratio of the ionization potential @xmath152 to the energy of the electrons @xmath153 .
the balance of these rate coefficients implies that @xmath154h@xmath155 / @xmath154h@xmath156 100 for @xmath157 25,000 k. based on the impact ionization rate of @xcite and the radiative recombination rate of @xcite , the ionization fraction of atomic oxygen i.e. , @xmath154o@xmath155 / @xmath154o@xmath156 100 in the same temperature range .
second , as bj10 also points out , higher temperatures lead to a worsening fit to the h i transmission data in figure [ fig : transmission ] . as the temperature approaches
@xmath1 15,000 k , @xmath158 2 for the fit .
this is not a very strong constraint at all , of course , but it indicates a general trend that disfavors temperatures much higher than @xmath1 15,000 k. indirectly , the need to retain the fit to the h i transmission data and these temperature limits imply that models with o / h and c / h ratios much higher than 40 times solar ratios are not realistic , and that enhanced oxygen abundance can increase the o i transit depth at most to 7.1 % .
we also experimented with a combination of supersolar oxygen abundance and hydrodynamic escape by introducing a uniform radial velocity of @xmath73 30 km s@xmath13 to the m5 model . as a result
, we obtained an o i transit depth of 10.8 % , which is actually close to 10.5 % . despite being a good fit to the observations ,
there are several good reasons for why this model is unrealistic .
the o / h ratio of the m5 model is 40 times higher than the solar ratio and @xmath73 30 km s@xmath13 below the critical level .
matching the observations is not enough prove a theory the theory also has to make physical sense .
it is possible that , contrary to our assumptions , there are optically thick clouds of neutral hydrogen and oxygen above the upper boundary of the m1 model moving at large los velocities .
these clouds can be formed , for instance , by neutral atoms that escape the atmosphere and react with the stellar wind @xcite .
although the density profile of the empirical model is misleading at high altitudes , we extended the m1 model thermosphere to @xmath159 10 r@xmath6 . as a result
, we obtained h i and o i transit depths of 13.1 and 12.2 % , respectively . in this case , @xmath160 13.4 for the h i transmission fit . in a slightly more realistic case
, we assumed that only 10 % of the atoms are neutral between @xmath87 and @xmath159 10 r@xmath6 .
this led to h i and o i transit depths of 7.8 and 6.3 % , respectively , and @xmath160 2.1 .
we note that external neutral atoms can affect the transit depth but we also agree with bj10 that the h i transmission data points are not consistent with optically thick populations of hydrogen moving at large velocities .
detailed models of the chemical and dynamical interaction of the escaping atmosphere with the stellar wind outside the roche lobe are required to study the distribution of external neutral atoms . with the exception of gm07 ,
most of the modeling efforts so far have concentrated on simulating the distribution of hydrogen ( and sometimes helium ) in the escaping atmosphere of hd209458b .
these models are justified by the assumption that the _ overall _ characteristics of the thermosphere are not affected by small abundances of heavier neutral atoms or ions .
in addition to the empirical model presented in section [ subsc : model_atmosphere ] , we have used two models with different characteristics and limiting assumptions to calculate transit depths .
these models are the one - dimensional outflow model of @xcite ( y04 ) and a thermospheric circulation model developed by @xcite ( k10 ) .
the photochemistry in these models is based on a composition of hydrogen and helium only . in order to roughly estimate the o i transit depth
, we added a solar proportion of oxygen to these models and assumed an oxygen ionization fraction equal to that of hydrogen .
the k10a model is different from the existing outflow models in that it is three - dimensional and can account for horizontal flows that are important in the thermosphere of hd209458b . on the other hand , the model assumes that the thermosphere is in hydrostatic equilibrium and thus it can not properly account for vertical flows that lead to a deviation from hydrostatic conditions . also , the photochemical scheme of the model is not valid in regions where h@xmath15 is the dominant species . to account for these limitations , we placed the upper boundary of the transit depth calculations to @xmath7 2.9 r@xmath6 in line with the arguments we have made for the validity of the empirical model . in order to calculate the optical depth as a function of altitude
, we averaged the density and temperature profiles of the model globally to obtain one - dimensional vertical profiles .
the resulting h i and o i transit depths are 5.2 and 3.7 % , respectively , and for the h i transmission fit we obtain @xmath125 3.1 ( see table [ table : models ] ) .
these values are in rough agreement with our earlier calculations and the observations .
the one - dimensional outflow model developed by y04 solves the vertical component of the navier - stokes equations and includes a self - consistent treatment of much of the important physics in the thermosphere of hd209458b .
the altitude range of the model is limited to @xmath161 3 r@xmath6 because above this altitude stellar gravity , radiation pressure , and interaction of the escaping atmosphere with the stellar wind affect the solution in ways that can not be accounted for by simple one - dimensional , hydrodynamic models . in order to ensure consistency with kinetic theory ,
the velocity at the upper boundary of the model was required to be equal to the weighted mean of the velocity distribution functions of different species .
this is not the same as setting the velocity to be equal to jeans effusion velocity because _ drifting _ maxwellians were used to calculate the boundary condition in an iterative fashion .
the density and temperature profiles of the y04 model yield h i and o i transit depths of 2.9 and 2.7 % , respectively ( see table [ table : models ] ) .
while the full - width h i transit depth is within 1.6@xmath14 of the observed value , @xmath162 for the hi transmission curve is larger than the one obtained for the empirical model or the k10a model .
the reason for the slightly poorer fit lies in the details of the density profiles .
the h@xmath2/h dissociation front of the y04 model is located at the pressure of @xmath3 0.5 nbar at the altitude of @xmath7 1.1 r@xmath6 .
this means that the model is less extended than the empirical model below the 0.5 nbar level because the mean molecular weight of the gas is larger and the thermosphere is cooled effectively by infrared emissions from h@xmath163 ions .
also , the ionization front of the model is located at a relatively low altitude of @xmath110 1.7 r@xmath6 .
the extent and temperature of the thermosphere depend strongly on the location of the h@xmath2/h dissociation front that effectively determines the number density and temperature at the lower boundary of the empirical model . in order to illustrate this point
, we changed the pressure at the lower boundary of the m1 model from @xmath61 0.1 @xmath4bar to @xmath61 1 nbar ( see model m6 in table [ table : models ] ) . as a result we obtained reduced h i and o i transit depths of 2.3 and 2.2 % , respectively .
this explains why @xcite obtained a relatively low h i transit depth of @xmath93 % by using a one - dimensional outflow model with a composition based on h and h@xmath15 only .
the pressure and temperature at the lower boundary of the model are only @xmath72 30 nbar and @xmath164 1000 k , respectively .
thus the model ignores the critical region of the atmosphere that contributes significantly to the optical depth .
if the pressure of their lower boundary was higher , they could have obtained a larger transit depth . in the k10a model the h@xmath2/h dissociation front is located at the pressure of @xmath3 0.1 @xmath4bar due to a combination of thermal dissociation and horizontal mixing . in the div1 model of gm07
it is located at @xmath3 1.5 @xmath4bar because , contrary to the other models , it includes ion and neutral photochemistry with carbon , oxygen , and nitrogen species .
thus h@xmath2 is also dissociated in reactions with atomic oxygen and the oh radical .
we note , however , that @xcite also modeled the neutral photochemistry and found that h@xmath2 was not entirely converted into h at these pressure levels .
motivated by the div1 model , we moved the lower boundary of the m1 model to @xmath61 1 @xmath4bar and obtained h i and o i transit depths of 6.6 and 3.9 % , respectively ( see model m7 in table [ table : models ] ) . in this case , we adjusted the mean temperature of the thermosphere to @xmath1 8250 k in order to retain the fit to the h i transit depth . in terms of optical depth , the m7 model is almost indistinguishable from the m1 model .
our exploration together with the results of bj10 reveal that models of the thermosphere can be used to match the observed h i absorption depths as long as the location of the h@xmath2/h dissociation front , the temperature and the location of the upper boundary agree with the values derived from our empirical model .
these findings provide strong justification for our approach of using a simple empirical model to fit the data .
we would not have been able to identify these important parameters that affect the optical depth profiles in the thermosphere if we had allowed one of the more complicated models to cloud our judgement .
once known , these parameters can help to guide the hydrodynamic and photochemical models towards solutions that agree with the observed transit depths .
in this section we assess the feasibility of our basic assumptions and results . we find that it is easy to generate models that agree with the observed h i transit depths . according to the best - fitting empirical models m1 and m7 ( see table [ table : models ] ) , the h@xmath2/h dissociation front is located between @xmath3 0.11 @xmath4bar ( @xmath5 1.1 r@xmath6 ) and the mean temperature of the thermosphere is between @xmath1 800011,000 k. we note that due to thermal ionization , the temperature of the neutral thermosphere can not be much higher than @xmath126 15,000 k. the m1 model atmosphere is optically thick up to @xmath5 2 r@xmath6 in the wings of the h ly@xmath0 line and at least up to the roche lobe in the core of the line . above our upper boundary at @xmath111 2.9 r@xmath6 , however , the atmosphere is mostly ionized .
our results are sensitive to the pressure level of the h@xmath2/h dissociation front .
additional observations are required to constrain the vertical column density of h i in the thermosphere .
such observations were recently obtained by @xcite who attempted to observe the auroral and dayglow emissions of h@xmath2 from hd209458b .
they did not detect the emissions and argued that this implies that the vertical column density of h i must be at least @xmath165 3 @xmath102 10@xmath121 m@xmath106 to prevent stellar fuv radiation from exciting h@xmath2 in the lower thermosphere .
table [ table : models ] includes substellar column densities of h i for the models discussed in this paper . according to this additional constraint ,
it is possible that the h@xmath2/h dissociation front is deeper than the 0.1 @xmath4bar level . in that case
, m7 is the best - fitting model . the detection or non - detection of infrared emissions from h@xmath163
can also be used to constrain the properties of the thermosphere .
@xcite attempted to observe these emissions from a sample that included the transiting planet gj436b and several non - transiting egps by using the cshell spectrograph at the nasa infrared telescope facility ( nirtf ) .
they reported upper limits for the fluxes that are often several orders of magnitude higher than the predicted fluxes for close - in egps .
more recently , @xcite obtained a near - ir secondary eclipse spectrum of hd189733b by using the spex instrument at nirtf .
the lack of a 3.9 @xmath4 m feature in this spectrum may imply that h@xmath163 emissions are not important on this particular planet .
we note that a lack of emissions implies that there is not enough h@xmath2 to form h@xmath163 or that h@xmath163 is destroyed in reactions with oxygen or carbon species .
modeling the chemistry and dynamics of the lower thermosphere is a complicated problem .
for instance , gm07 has shown that the complex photochemistry that includes oxygen and carbon species affects the location of the h@xmath2/h dissociation front and thus the extent and temperature of the thermosphere .
h@xmath2 can also be dissociated by high - energy electrons that penetrate below the 1 @xmath4bar level and have not been included in the existing photochemical models .
the composition of the atmosphere is also affected by dynamics . because the upper atmosphere is strongly ionized ,
the conductivities are likely to be high . as a result
, the dynamics of the thermosphere is affected by strong electric currents that have not been included in the existing models either .
the results are not as sensitive to the location of the upper boundary as they are to the location of the lower boundary .
we note , however , that the location of the upper boundary at @xmath111 2.72.9 r@xmath6 in both our work and that of bj10 is remarkably consistent .
it is tempting to suggest that this is not a coincidence .
after all , the upper boundary is close to both the boundary of the roche lobe and the ionopause of a weakly magnetized planet .
it is possible that , for some reason , the density of the neutral atoms decreases sharply above this altitude .
we believe that multidimensional plasma models , either kinetic or mhd , are needed to study the density profiles of different species above the upper boundary . based on an earlier study by @xcite , @xcite ( hereafter e10 ) studied the formation of energetic neutral atoms ( enas ) outside the roche lobe of hd209458b .
they pointed out that the enas are created by charge exchange between hydrogen atoms escaping the atmosphere and the protons of the stellar wind .
they argued that absorption by enas can explain the h i transmission curve reported by vm03 . in principle , this is a good idea but some of the details are unclear .
we are not convinced that the boundary conditions of the e10 model are consistent with realistic models of the thermosphere .
also , the e10 model was not used to fit the bj08 transmission data .
the inner boundary of the e10 model is located at @xmath7 2.8 r@xmath6 and the optical depth of the _ atmosphere _ in the h ly@xmath0 line is assumed to be small inside the inner boundary sphere . at the same time the number density at the inner boundary is @xmath166 4 @xmath102 10@xmath167 m@xmath168 because a mass loss rate of @xmath169 1.6 @xmath102 10@xmath170 g s@xmath13 is required to ensure a sufficient production rate of the enas .
we note that the number density of the m1 model at @xmath7 2.9 r@xmath6 is @xmath166 2.6 @xmath102 10@xmath167 m@xmath168 and that the density profile of this model already leads to a full - width h i transit depth of 6.6 % .
thus absorption by enas is not required to fit the data . on the contrary
, too much additional absorption may actually lead to a deteriorating fit to the detailed h i transmission data .
the mass loss rate of @xmath169 1.6 @xmath102 10@xmath170 g s@xmath13 is in line with the gm07 model .
bj10 showed that this model can be used to fully explain the observed absorption .
they also pointed out that an optically thick external population of neutral hydrogen is possible but the los velocities must be slow . in this case
, additional absorption is constrained to the core of the line , which is fully absorbed by the ism and the transmission in the wings is not significantly affected .
our analysis supports this conclusion .
new models that treat the thermosphere and the external region self - consistently are required to study the true distribution of external neutral atoms .
the o i transit depth is less precise than the h i transit depth but we are able to explain the observations with our empirical model and o / h ratios ranging from solar to 40 times solar , although the latter value implies an uncomfortably high temperature .
the large o i transit depth indicates that the thermosphere is uniformly mixed up to the roche lobe although there is a marginal possibility that this is not the case because the fuv continuum transit depth of 2 % is within the 1.96@xmath14 confidence interval of the detected depth .
if diffusive separation takes place , the o i transit depth can not be much higher than the continuum transit depth .
based on the m1 model , we obtain a full - width o i transit depth of 4.3 % , which is statistically consistent with the observed absorption . in this case
, oxygen overflows the roche lobe with hydrogen . assuming radial velocities of @xmath73110 km s@xmath13 at @xmath111 2.9 r@xmath6 , the total mass loss rate from the m1 model is @xmath169 440 @xmath102 10@xmath12 kg s@xmath13 and this leads to a total escape flux of @xmath171 1.818 @xmath102 10@xmath172 s@xmath13 for oxygen .
we have also shown that the velocity dispersion due to radial outflow , supersolar oxygen abundance and external populations of neutral oxygen can explain o i transit depths that are comparable to the full - width h i depth .
we have not , however , been able to identify _ feasible _ mechanisms that give rise to o i transit depths that are significantly higher than the h i depth .
we believe that more precise measurements are required to further constrain the o i transit depth . we have generated synthetic data by simulating the response function of the stis g140 m grating to show that , by assuming the m1 model to be accurate , we would observe a transit depth of 4.3 @xmath10 1 % for hd209458b .
thus for transit depths higher than @xmath95 % , a confident detection in the upper atmosphere can easily be obtained .
in addition , the spectral resolution of the g140 m grating is high enough to resolve the individual emission lines of the o i triplet .
as we have seen , spectrally resolved measurements provide much better constraints on the actual transit depth than unresolved observations .
we believe that it would be useful to repeat the o i transit depth measurements by using this instrument .
transmission spectroscopy in uv wavelength bands is a rich source of information about the upper atmospheres of extrasolar planets . in this paper
we have introduced a generic method that can be used to interpret and analyze uv transit light curves .
this method is based on tracing the emitted flux from the stellar atmosphere through the atmosphere of the planet and the interstellar medium to the observing instrument at earth .
we have demonstrated the method in practice by applying it to the existing uv transit depth measurements of hd209458b in the h ly@xmath0 and o i triplet lines ( vm03,vm04,bj10 ) . in order to interpret the measured transit depths we used a simple empirical model of the thermosphere based on the generic features of more complex models ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) to simulate absorption by the occulting atmosphere .
the h i transit depth is sensitive to interstellar absorption and the full - width transit depth reflects the optical depth of the atmosphere in the wings of the h ly@xmath0 line .
we found that it is easy to generate models that explain the observed absorption without the need to introduce external enas or hydrogen atoms accelerated by radiation pressure outside the roche lobe .
according to the best - fitting models ( see models m1 and m7 in table [ table : models ] ) , the mean temperature of the thermosphere is @xmath1 800011,000 k and the h@xmath2/h dissociation front is located at @xmath61 0.11 @xmath4bar ( @xmath62 1.1 r@xmath6 ) .
the upper boundary of the model is located at @xmath173 2.9 r@xmath6 , which is near the boundary of the roche lobe .
below @xmath87 the density profiles are approximately hydrostatic and above @xmath87 the atmosphere is mostly ionized . by using the m1 model , we obtain a full - width h i transit depth of 6.6 % and the model transmission of the h ly@xmath0 line matches the data points reported by bj08 . the apparent disagreement between bj08 and the earlier analysis by vm03 arises from differences in the treatment of the data and different definitions of the transit depth .
in particular , there is no definite observational evidence for a cometary tail following the planet in the transit light curve and the absorption is not significantly stronger in the blue side of the h ly@xmath0 line .
however , we agree with vm08 that the core of the absorption line is optically thick up to the roche lobe and thus that the atmosphere is evaporating .
we estimate a mass loss rate of @xmath17410@xmath1210@xmath70 kg s@xmath13 based on the density profile of the m1 model and a range of possible radial velocities .
it is possible that enas are present outside the roche lobe as suggested by e10 , but the optical depth of the ena clouds has to be consistent with the underlying models of the thermosphere .
we do not agree with e10 that the optical depth of the thermosphere below @xmath7 2.8 r@xmath6 is small .
we do believe , however , that one - dimensional hydrodynamic models are not adequate in modeling the distribution of ionized gases outside the roche lobe .
multidimensional plasma models are required to study the interaction of the ionosphere with the stellar wind and magnetospheric plasma self - consistently .
the mean o i transit depth of 10.5 % is only 1.93@xmath14 away from the fuv continuum transit depth of @xmath92 % . by using the m1 model
, we obtain an o i transit depth of 4.3 % , which is marginally consistent with the observations .
full - width o i transit depths that are comparable or slightly higher than the h i depth are possible if the abundance of oxygen in the thermosphere is supersolar , the atmosphere is escaping with supersonic velocities or if large external clouds of neutral oxygen are present above the roche lobe .
more precise measurements are required to constrain the o i transit depth further . out of the currently available instruments , the stis g140 m grating
can be used to obtain repeated measurements with higher s / n and spectral resolution .
the authors would like to thank b. e. wood for correspondence about the h ly@xmath0 emission line profile of hd209458b .
t.t.k . would like to thank j. y - k .
cho for helpful discussions regarding the structure and dynamics of the ionosphere .
this research has been supported by the nasa s planetary atmospheres program through grant nnx09ab58 g .
atreya , s. k. 1986 , atmospheres and ionospheres of the outer planets and their satellites ( berlin , germany : springer - verlag ) ben - jaffel , l. 2007 , , 671 , l61 ben - jaffel , l. 2008 , , 688 , 1352 ben - jaffel , l. , & hosseini , s. s. 2010 , , 709 , 1284 charbonneau , d. , brown , t. m. , noyes , r. w. , & gilliland , r. l. 2002 , , 568 , 377 charbonneau , d. , et al .
2009 , , 462 , 891 colonna , g. , % capitelli , m. , spectrochimica acta b , 64 , 863 cotton , d. m. , gladstone , g. r. , & chakrabarti , s. 1993 , , 98 , 21651 czesla , s. , huber , k. f. , wolter , u. , schroter , s. , & schmitt , j. h. m. m. , , 505 , 1277 ekenback , a. , et al .
2010 , , 709 , 670 france , k. et al .
2010 , preprint ( arxiv:1002.3218v1 ) garcia munoz , a. 2007 , plan .
, 55 , 1426 gladstone , g. r. 1992 , , 97 , 19519 gu , p .-
lin , d. n. c. , & bodenheimer , p. h. 2003 , , 588 , 509 griessmeier , j .-
m . et al .
2004 , , 425 , 753 holstrom , m. , ekenback , a , selsis , f. , lammer , h. , & wurz , p. 2008
, , 451 , 970 humlicek , j. 1982 , j. quant .
transfer , 27 , 437 hunten , d. m. , pepin , r. o. , & walker , j. c. g. 1987 , icarus , 69 , 532 knutson , h. a. , charbonneau , d. , allen , l. e. , burrows , a. , & megeath , s. t. 2008 , , 673 , 526 koskinen , t. t. , aylward , a. d. , miller , s. , & smith , c. g. a. 2007 , , 661 , 515 koskinen , t. t. , aylward , a. d. , & miller , s. 2009 , , 693 , 868 koskinen , t. t. , cho , j. y .- k . , achilleos , n. , & aylward , a. d. 2010 , submitted lammer , h. et al .
2003 , , 598 , l121 landsman , w. , & simon , t. 1993 , , 408 , 305 liang , m .- c . ,
parkinson , c. d. , lee , a. y .- t . ,
yung , y. l. , & seager , s. 2003 , , 596 , l247 moos , h. w. et al .
2002 , , 140 , 3 morton , d. c. 1991 , apjs , 77 , 119 morton , d. c. 2003 , apjs , 149 , 205 murray - clay , r. a. , chiang , e. i. , & murray , n. 2009 , , 693 , 23 nagy , a. f. , cravens , t. e. , yee , j .- h . , &
stewart , a. i. f. 1981 , , 8 , 629 palle , e. , osorio , m. r. z. , barrena , r. , montanes - rodriguez , p. , & martin , e. l. 2009 , , 459 , 814 press , w. h. , teukolsky , s. a. , vetterling , w. t. , & flannery , b. p. 1992
, numerical recipes in fortran , 2nd ed .
( cambridge , uk : cambridge university press ) roussel - dupre , d. 1985 , , 153 , 116 schunk , r. w. , nagy , a. f. , ionospheres : physics , plasma physics , and chemistry ( cambridge , uk : cambridge university press ) shematovich , v. i. , bisikalo , d. v. , & gerard , j .- c .
2006 , , 111 , a10301 shkolnik , e. , gaidos , e. , & moskovitz , n. 2006 , , 132 , 1267 smith , g. r. , & hunten , d. m. 1990 , rev .
geophys . , 28 , 117 storey , p. j. , & hummer , d. g. 1995 , , 272 , 41 swain , m. r. , vasisht , g. , & tinetti , g. 2008 , , 452 , 329 swain , m. r. , et al .
2009 , , 704 , 1616 swain , m. r. , et al .
2010 , , 463 , 637 tian , f. , toon , o. b. , pavlov , a. a. , & de sterck , h. 2005 , , 621 , 1049 tinetti , g. , et al .
2007 , , 448 , 169 vidal - madjar , a. , et al .
2003 , , 422 , 143 vidal - madjar , a. , et al .
2004 , , 604 , l69 vidal - madjar , a. , et al .
2008 , , 676 , l57 voronov , g. s. 1997 , atomic data and nuclear data tables , 65 , 1 wood , b. e. , redfield , s. , linsky , j. l. , muller , h .-
, & zank , g. p. 2005
, , 159,118 yelle , r. v. 2004 , icarus , 170 , 167 | transmission spectroscopy at uv wavelengths is a rich and largely unexplored source of information about the upper atmospheres of extrasolar planets .
so far , uv transit observations have led to the detection of atomic hydrogen , oxygen and ionized carbon in the upper atmosphere of hd209458b .
the interpretation of these observations is controversial
it is not clear if the absorption arises from an escaping atmosphere interacting with the stellar radiation and stellar wind , or the thermosphere inside the roche lobe . in this paper
, we introduce an empirical model that can be used to analyze uv transit depths of extrasolar planets .
we use this model to interpret the transits of hd209458b in the h lyman @xmath0 and the 1304 o i triplet emission lines .
the results indicate that the mean temperature of the thermosphere is @xmath1 8,00011,000 k and that the h@xmath2/h dissociation front is located at pressures between @xmath3 0.11 @xmath4bar , which correspond to an altitude of @xmath5 1.1 r@xmath6 .
the upper boundary of the model thermosphere is located at altitudes between @xmath7 2.73 r@xmath6 , above which the atmosphere is mostly ionized .
we find that the h i transit depth reflects the optical depth of the thermosphere in the wings of the h lyman @xmath0 line but that the atmosphere also overflows the roche lobe . by assuming a solar mixing ratio of oxygen
, we obtain an o i transit depth that is statistically consistent with the observations . an o i transit depth comparable to or slightly larger than the h i transit depth is possible if the atmosphere is undergoing fast hydrodynamic escape , the o / h ratio is supersolar , or if a significant quantity of neutral oxygen is found outside the roche lobe .
we find that the observations can be explained solely by absorption in the upper atmosphere and extended clouds of enas or atoms strongly perturbed by radiation pressure are not required . due to the large uncertainty in the data , repeated observations
are necessary to better constrain the o i transit depths and thus the composition of the thermosphere . |
recently , impressive experimental results concerning the quantum degenerate regime of a dilute gas of trapped fermionic atoms have been presented @xcite . by cooling @xmath0 @xmath1k atoms to @xmath2 of the fermi temperature , the onset of fermi degeneracy was observed in the thermodynamic and in the scattering properties of the gas .
also , progress towards achieving fermi degeneracy for trapped @xmath3li @xcite has been presented .
such a weakly - interacting , degenerate fermi - dirac gas provides a new platform for exploring fundamental quantum many - body physics .
several theoretical results dealing with the equilibrium properties of such a gas have been presented @xcite .
for example , it has been shown that a two - component gas of spin - polarized @xmath3li undergoes a bardeen - cooper - schrieffer ( bcs ) transition to superfluidity at experimentally obtainable densities and temperatures @xcite .
one class of experiments that is compatible with trapping protocols , and which has gleaned valuable information on the dynamics of dilute bose - einstein condensates ( becs ) , involves monitoring the response of the gas to a change in trapping potential .
for example , transient modulation of the trapping potential induces free ringing of the gas , which in the case of becs led to a direct determination of low - lying regions of the quasi - particle spectrum @xcite .
complete release of the trapping potential enables one to view the free expansion of the gas ; this established early on the essential validity of the time - dependent gross - pitaevski equation for describing the dynamics of latest generation of becs @xcite , which was later put to stringent quantitative tests @xcite .
a trapped , single - component gas of ultracold fermionic atoms can , for experiments of current interest , be considered to be ideal ( noninteracting ) as a reasonable approximation , since atomic collisions are strongly suppressed @xcite . in this paper , we examine theoretically the dynamics of such an ideal gas in a time - dependent trap . using a scaling formalism similar to that which has been successfully applied to trapped becs @xcite , we derive analytical results for the class of experiments mentioned above .
the essential fermionic character of the systems with which we are concerned is established by the initial distribution of particles in the trap ; the time evolution of this distribution , under changes of the trapping potential that preserve its harmonicity , are rigorously equivalent to that of an ensemble of noninteracting particles , independent of statistics .
we present a simple formula that describes the free expansion of such an ideal gas , and our results suggest a new approach to the problem of quantitative thermometry in the nanokelvin regime .
due to their weak pair interactions , single - component fermi - dirac gases are attractive candidates for nanokelvin thermometry ; with an improved theoretical understanding of finite - temperature properties of becs @xcite , which are much more robust candidates for experiment at present , we can envisage a direct comparison of temperatures of ultracold bose - einstein and fermi - dirac gases .
we also examine in detail the ( non - linear ) response of the gas to a harmonic oscillation of the trapping frequency .
this reveals a domain of driving frequencies and amplitudes which generates a resonant response of the gas .
the non - interacting limit considered here should provide a good approximation for degenerate , single - component cold fermi - dirac gases . since interactions in general play a smaller role for fermi - dirac vs. bose - einstein systems , the present approach may also provide a useful starting point for consideration of the dynamics of multiple - component fermi gases .
we start with a derivation of the equations describing the scaling properties of an ideal gas trapped in a time - dependent harmonic potential . consider a classical particle of mass @xmath4 trapped in a potential @xmath5 with @xmath6 denoting the 3 spatial dimensions .
we set @xmath7 for @xmath8 , i.e. the trap potential is constant prior to @xmath9 .
newton s law is then expressed by @xmath10 and @xmath11 . invoking a scaling transformation @xmath12 and @xmath13
, we obtain @xmath14 , with @xmath15 , if the scaling parameters @xmath16 satisfy the equations @xmath17 in quantum mechanics , the heisenberg equation for a non - interacting gas in a time - dependent harmonic potential takes the form : @xmath18\hat{\psi}({\mathbf{r}},t).\ ] ] where @xmath19 is the field operator for an atom in a single hyperfine state at position @xmath20 , which obeys the usual fermi - dirac anticommutation relations .
the quantum mechanical analogue of the rescaling outlined for the classical case above corresponds to writing @xcite @xmath21}{\sqrt{\gamma_x\gamma_y\gamma_z } } e^{im\sum_jr_j^2\dot{\gamma}_j/2\hbar\gamma_j}\ ] ] with @xmath22 , as defined above . if each @xmath23 satisfies eq.([gammaeq ] ) , by writing @xmath24=\pi_j\hat{\phi}[q_j(t)]$ ] , we obtain @xmath25=[-\frac{\hbar^2}{2m}\partial^2_{q_j } + \frac{1}{2}m\omega_{0j}^2q_j^2]\hat{\phi}[q_j(t)].\ ] ] thus , the time - dependent problem has been reduced to the trivial case of evolution in a time - independent harmonic trap in the rescaled variables @xmath26 .
determination of these variables requires only the solution of the three ordinary differential equations expressed by eq.([gammaeq ] ) ; the three spatial dimensions can be treated independently .
an appealing qualitative description of the solution of these equations , cast in the context of one - dimensional scattering theory , has been given by kagan et al.@xcite . from eq.([transform ] ) , it follows that the density @xmath27 of the gas for a given time @xmath28 is given by @xmath29\ ] ] where @xmath30 is the particle density for @xmath31 .
thus , we can calculate @xmath27 for any time @xmath32 for modulated frequencies @xmath33 , by solving eq.([gammaeq ] ) , subject to the boundary conditions @xmath34 and @xmath35 .
equation ( [ gammaeq ] ) also describes the two - dimensional , non - ideal bec subject to isotropic variations of the trap potential , with eq.([density ] ) also being applicable if @xmath36 is obtained by a solution of the gross - pitaevski equation @xcite .
we now analyze the solution of eq.([gammaeq ] ) for two cases of experimental relevance : free expansion ; and harmonic modulation of the trapping potential .
to model a free expansion experiment , we take @xmath37 for @xmath32 . solving eq.([gammaeq ] ) with the requirement that the gas is in equilibrium for @xmath38
, we obtain @xmath39 for @xmath38 and @xmath40 for @xmath41 .
the time - dependent width of the cloud after the trap has been dropped is given by @xmath42 .
to describe the aspect ratio @xmath43 of a cylindrically - symmetric cloud , we find @xmath44 we now apply these results to an ideal gas in two limiting cases .
we first treat the case of @xmath45 , with the chemical potential @xmath46 for @xmath6 .
this corresponds to the semiclassical limit , appropriate to current experiments with more than a few hundred trapped atoms . in this limit
, the initial density profile is well described by the thomas - fermi ( tf ) approximation @xcite .
this gives the integrated density @xmath47 , @xmath48 ^ 2\ ] ] with @xmath49 and @xmath50 .
we see that the cloud becomes isotropic for @xmath51 .
this is the result one would get for a classical gas at finite temperature , in which the momentum distribution is isotropic , and is consistent with the initial isotropic momentum distribution implicit in the tf approximation ( in becs in anisotropic traps , on the other hand , the long - time expansion is anisotropic due to macroscopic population of an anisotropic initial state , a key result in their first observation @xcite ) .
the initial aspect ratio of @xmath52 and eq.([alpha ] ) evolves to @xmath53 as @xmath54 . in fig.([expandfig ] ) , we plot the integrated density @xmath55 for @xmath45 , @xmath56 , and @xmath31 ( a ) and @xmath57 ( b ) .
we have taken @xmath58 and @xmath59 corresponding to current experiments on trapped @xmath1k atoms @xcite .
the two snapshots of the integrated density clearly show that the gas becomes isotropic for large expansion times .
this is further illustrated in fig.([alphafig ] ) , where we plot @xmath43 as given in eq.([alpha ] ) for the same set of parameters .
note that @xmath52 for any initial density distribution of the form @xmath60 .
such initial distributions will become isotropic in an free expansion experiment .
for instance , for @xmath61 , the density is well described by a classical gaussian profile , i.e. @xmath62 $ ] and fig.([alphafig ] ) thus describes a trapped gas of fermions for any @xmath63 within the tf approximation .
this means that one can not detect the onset of fermi degeneracy by looking at the aspect ratio of the expanding gas . as an example where
@xmath64 , we now consider a case with @xmath65 such that only one level is occupied in the @xmath66-direction and @xmath67 for @xmath68 .
the gas is initially strongly confined in the @xmath66-direction .
the integrated density profile is then @xmath69 for @xmath70 , yielding @xmath71 . in fig.([alphafig ] ) , we plot the aspect ratio for a free expansion for @xmath72 and @xmath73 using eq.([alpha ] ) .
we have @xmath74 and @xmath75 for @xmath54 .
the gas , which initially is strongly confined in the @xmath66-direction will for @xmath51 become most confined in the @xmath76- , and @xmath77-directions .
this is , of course , a direct reflection of the uncertainty relation giving higher average momentum in the @xmath66-direction .
however , observation of such a quantum effect requires a highly anisotropic trap . for the case of @xmath78
trapped atoms we would require @xmath79 . thus , for fermi - dirac particles the anisotropy of the expanded cloud due to the heisenberg uncertainty principle is considerably less than that encountered in the bose - einstein case . from a measurement of the density at any time @xmath28 under free expansion
it is straightforward to determine the initial density distribution @xmath30 using eq.([density])-([gammafree ] ) .
this suggests that a trapped , single - component fermi - dirac gas could serve as a low @xmath63 thermometer . assuming the gas is in thermodynamic equilibrium for @xmath80
, one could infer the temperature of the gas from a knowledge of its density distribution , the calculation of which is a trivial problem of summing over fractionally occupied trap levels .
hence , a determination of @xmath63 is simply a matter of fitting a calculated density to the measured @xmath81 .
density profiles for an ideal fermi - dirac gas for various temperatures have already been given elsewhere @xcite .
however , if only one hyperfine level is present at all times , the atomic collisions required for thermalization are suppressed at low temperatures , and evaporative cooling of the gas will be inhibited .
this can be remedied by initially trapping the gas in two hyperfine states , to allow @xmath82-wave collisions between atoms in different states .
one can then deplete the trap of one hyperfine state by applying a microwave field that depletes one of the states .
if this depletion occurs on a time scale that is long compared to collision times in the sample , the remaining atoms will be in thermodynamic equilibrium , and their expansion can be described by the results in this section .
in fact , a procedure of this type is being used in present experiments on trapped @xmath1k atoms @xcite .
alternatively , one could drop the trap with both hyperfine levels present . the initial density profile and the free expansion of the gas
will then be perturbed away from the ideal gas result by the interactions .
when the effect of interactions is small , the initial equilibrium density and the subsequent expansion can be calculated relatively easily .
a calculation of the effect of the interactions on the equilibrium distribution has been presented elsewhere , @xcite and the effect of the interactions on the free expansion can be calculated within the mean - field approximation , starting from the simple uncoupled solutions given by eq.([gammafree ] ) for each spatial direction @xcite .
as noted above , modulation of the trapping frequency has proven to be a useful technique for probing the low - lying collective modes of trapped becs .
we therefore examine theoretically the non - linear response of a trapped gas to such a modulation .
the non - interacting case treated here is in some sense opposite to the hydrodynamic limit we have treated elsewhere @xcite . to model a typical experiment
, we assume that the trapping frequencies take the form @xmath83 $ ] for @xmath32 with @xmath84 being the driving frequency and @xmath85 the driving amplitude . instead of solving eq.([gammaeq ] ) with this form for @xmath33 ,
it turns out to be easier to go back to the original classical equation of motion @xmath86 by using @xmath87 and @xmath88 $ ] .
thus , by writing @xmath89 $ ] , and @xmath90 , we obtain : @xmath91\xi_j=0\ ] ] with @xmath92 , @xmath93 and @xmath94 .
equation ( [ mathieu ] ) is a variant of mathieu s equation @xcite , whose properties have been extensively studied .
its behavior is easy to explain in one case of experimental interest : that when the trap frequency is modulated during a finite interval @xmath95 , and then returned to its original value
. the subsequent motion of the cloud is described by a solution of the time - independent problem @xmath96 with arbitrary initial conditions : @xmath97 where @xmath98 is a conserved quantity for @xmath99 ; @xmath100 $ ] ; and @xmath101 is the value of @xmath16 immediately after the modulation ceases .
equation ( [ gammagen ] ) has a discrete frequency spectrum ( frequencies of @xmath102 with @xmath103 ) , as expected for a non - interacting gas in a harmonic trap .
the linear response limit is recovered by letting @xmath104 ineq.([gammagen ] ) , to obtain @xmath105 .
in essence , the problem of predicting the response of the gas to a harmonic driving with frequency @xmath84 and amplitude @xmath85 is reduced to an analysis of the well - known solutions to mathieu s equation . in the parameter space @xmath106
, there are regions where the solutions of eq.([mathieu ] ) are unstable , i.e. their amplitude increases exponentially with time . also , there are stable regions where the solutions remain bounded .
the solutions on the boundaries between these regions are the mathieu functions @xcite . using @xmath107 , @xmath92 , and @xmath93 ,
this means that for certain regions in the @xmath108-space , the response of the gas diverges as the driving time @xmath109 increases , i.e. there is a resonant response , whereas in other regions the response of the gas remains finite for any value of @xmath109 .
specifically , for @xmath110 with @xmath103 , the solutions to eq.([mathieu ] ) diverge in time for an arbitrary small @xmath111 @xcite .
thus , for @xmath112 the response of the gas is resonant for an arbitrarely small driving amplitude and the amplitude of its oscillations will diverge with the driving time . in terms of eq.([gammagen ] ) , if the gas remains trapped in the time - independent potential @xmath96 after the driving ( @xmath99 ) , we have @xmath113 and the oscillations of the gas will be large and contain many harmonics .
the resonance for @xmath114 is , of course , the usual excitation frequency for an even - parity perturbation for a non - interacting gas .
however , the resonances for @xmath115 etc . do not correspond to new modes .
they simply reflect the fact that for these driving frequencies , the trapping potential is doing resonant work on the gas .
note that the present scaling formalism does not predict the higher - frequency modes at @xmath116 with @xmath117 , as they do not correspond to simple dilations / contractions of the cloud .
we will now examine the width of the unstable ( resonant ) regions around @xmath118 for @xmath119 .
for the @xmath114 resonance , the unstable region of the mathieu equation is bounded by @xmath120 for @xmath121 @xcite .
this gives , that for driving frequencies and amplitudes such that @xmath122 the response of the gas is divergent .
hence , the resonance region of the gas for finite but small driving amplitude @xmath85 has a reasonable width and should be relatively easy to access experimentally . likewise , for the @xmath123 resonance , the unstable region is bounded by @xmath124 and it should be experimentally accessible . for the lower frequency resonances ( @xmath125 ) , it turns out that the resonance regions are very narrow for @xmath119 as the boundary lines between the stable and unstable solutions of eq.([mathieu ] ) only differ by terms of order @xmath126 or higher @xcite .
thus , these resonances are difficult to find experimentally for @xmath119 .
instead , one could increase the driving amplitude @xmath85 for a given frequency @xmath84 ; for large enough @xmath85 , an unstable region is reached and the response of the gas diverges .
the above results are illustrated in fig.([mathieufig ] ) , where we plot the response of the gas as a function of driving frequency @xmath84 and amplitude @xmath85 .
the lines in the plot separate regions where the response of the gas remains finite from regions where the oscillations of the gas diverge with driving time .
for example , if one modulates the external potential with an off - resonance frequency of @xmath127 , the response of the gas remains finite for small @xmath85 . however , if one gradually increases the driving amplitude @xmath85 for this frequency , then one should observe a divergent response of the gas for @xmath85 larger than @xmath128 . the plot is generated using the well - known properties of the mathieu equation and the mapping @xmath129 .
note the unstable ( resonance ) regions for @xmath119 for @xmath118 .
the regions have zero width for @xmath119 for @xmath130 as predicted above . for increasing driving amplitude @xmath85 ,
the unstable regions grow as expected . for @xmath131 ,
the response of the gas for @xmath132 is finite whereas it diverges for @xmath133 .
physically , this corresponds to the fact that for @xmath134 and @xmath135 , the gas returns adiabatically to its initial state after one period of trap modulation and there is no net work done .
contrary , for @xmath136 the trapping potential becomes inverted for certain times resulting in a divergent response of the gas .
mathematically , the transition region for @xmath137 comes from the fact , that for @xmath138 , the even and odd periodic solutions ( mathieu functions ) of eq.([mathieu ] ) start to differ in `` energy '' ( the mathieu characteristic value @xmath139 @xcite ) , as the tunneling between successive minima of the potential potential @xmath140 becomes significant . again , since a spin - polarized gas of cold fermionic atoms is effectively ideal , the results presented here should apply to the response of such a gas to a harmonic modulation of the trapping potential .
in conclusion , using a scaling formalism we have been able to derive simple analytical results for the dynamics of ideal gases trapped in time - dependent harmonic traps .
we have concentrated on two important classes of experiments : for free expansion , we show how the initial density profile of the gas can easily be determined from a measurement of the density profile at any time @xmath28 after the trap has been dropped .
we proposed a low @xmath63 thermometer based on these results . also , we showed how the problem of determining non - linear response of the gas to a harmonic modulation of the trapping frequency can be mapped to an analysis of the well - known properties of the solutions to mathieu s equation .
we identified regions in @xmath108-space where the response of the gas was divergent with the modulation time .
especially , we were able to predict unstable regions for @xmath119 reflecting the fact , that the trap is doing resonant work on the gas .
since ultracold spin - polarized fermions are non - interacting to a very good approximation , our results should be directly relevant for current experiments in this very active field of research .
we thank y. castin and d. jin for useful discussions .
99 b. demarco and d. s. jin , science * 285 * , 1703 ( 1999 ) m .- o .
mewes , g. ferrari , f. schreck , c. salomon , centennial meeting of the american physical society , march 20 - 26 , 1999 , atlanta , georgia , vol .
44 , no . 1 , fb16 7 d. a. butts and d. s. rokhsar , phys . rev .
a * 55 * , 4346 ( 1997 ) j. schneider and h. wallis , phys .
a * 57 * , 1253 ( 1998 ) g. m. bruun and k. burnett , phys .
a * 58 * , 2427 ( 1998 ) h. t. c. stoof , m. houbiers , c. a. sackett , and r.g .
hulet , phys .
lett . * 76 * , 10 ( 1996 ) g. m. bruun , y. castin , r. dum , and k. burnett , accepted for publication in euro . j. of phys .
d d. s. jin , j. r. ensher , m. r. matthews , c. e. wieman , and e. a. cornell , phys .
lett . * 77 * , 420 ( 1996 ) ; m.edwards , p. a. ruprecht , k. burnett , r. j.dodd , and c. w. clark , phys .
lett . * 77 * , 1671 ( 1996 ) ; m .- o .
mewes , m.r .
andrews , n.j .
van druten , d.m .
kurn , d.s .
durfee , c.g .
townsend , and w. ketterle , phys .
lett . * 77 * , 988 ( 1996 ) ; s. stringari , phys .
lett . * 77 * , 2360 ( 1996 ) m. holland and j. cooper , phys .
a * 53 * , r1954 ( 1996 ) m. j. holland , d. s. jin , m. l. chiofalo , and j. cooper , phys .
lett . * 78 * , 3801 ( 1997 ) b. demarco , j. l. bohn , j. p. burke jr . , m.holland , and d. s. jin , phys .
82 * , 4208 ( 1999 ) b. demarco and d. s. jin , phys .
a * 58 * , r4267 ( 1998 ) y. castin and r. dum , phys . rev . lett .
* 77 * , 5315 ( 1996 ) y. kagan , e. l. surkov , and g. v. shlyapnikov , phys .
a * 54 * , r1753 ( 1996 ) y. kagan , e. l. surkov , and g. v. shlyapnikov , phys .
a * 55 * , r18 ( 1997 ) f. dalfovo , c. minniti , and l. p. pitaevskii , phys.lett . a * 227 * , 259 ( 1997 ) d. s. jin , m. r. matthews , j. r. ensher , c. e. wieman , and e. a. cornell , phys . rev .
* 78 * , 764 ( 1997 ) ; r. j.dodd , m. edwards , c. w. clark , and k. burnett , phys .
a * 57 * , r32 ( 1998 ) ; d. m. stamper - kurn , h .- j .
miesner , s. inouye , m.r .
andrews , and w. ketterlee , phys .
81 * , 500 ( 1998 ) m. h. anderson , j. r.ensher , m. r. matthews , c. e. wieman , and e. a. cornell , science * 269 * , 198 ( 1995 ) g. m. bruun and c. w. clark , to be published .
g. m. bruun and c. w. clark , cond - mat/9905263 m. abramowitz and i. a. stegun ( editors ) , _ handbook of mathematical functions _
( u.s . government printing office , washington dc , 1964 ) | we investigate theoretically the properties of an ideal trapped gas in a time - dependent harmonic potential . using a scaling formalism , we are able to present simple analytical results for two important classes of experiments : free expansion of the gas upon release of the trap ; and the response of the gas to a harmonic modulation of the trapping potential is investigated .
we present specific results relevant to current experiments on trapped fermions . |
on numerous occasions during our research in social media , resource sharing , intention analysis , and dissemination patterns , an interesting question emerged : when did a certain resource first appear on the public web ? upon examining a resource , one could find a publishing timestamp indicating when this resource was created or first made available to the public . for those select few pages ,
the timestamp format varies largely along with the time granularity .
some forum posts could deliver solely the month and the year of publishing , while in other news sites one can extract the timestamp to the second .
time zones could be problematic too : if not clearly stated on the page , the time zone could be that of the webserver , crawler / archive , or gmt .
ideally , each resource should be accompanied by a creation date timestamp but this is not true in most cases .
a second resort would be to ask the hosting web server to return the last modified http response header .
unfortunately , a large number of servers deliberately return more current last modified dates to persuade the search engine crawlers to continuously crawl the hosted pages .
this renders the dates obtained from the resource or its server highly unreliable . in our prior work ,
some of the social media resources we were investigating , ceased to exist .
we needed to investigate the time line of this resource from creation , to sharing , to deletion . depending on the hosting server to provide historic information about
a missing resource is unachievable in most cases .
this places a limitation to services that attempt to parse the resource textual representation or even its uri looking for timestamps
. the following step would be to search the public archives for the first existence of the resource . as we show below
that using this method solely has significant limitations .
thus there is a need for a tool that can estimate the creation date of any resource investigated without relying on the infrastructure of the hosting web server or the state of the resource itself .
some pages are associated with apis or tools to extract its metadata , but unfortunately they are non - unified , extremely specific , and what works on one page would not necessarily work on the other . due to the speed of web content creation and the ease of publishing ,
a certain assumption could be established . in some cases , like in blogs ,
a page could be created and edited before it is published to the public . to facilitate our analysis
, we will assume that the creation and publishing of a resource coincide .
if the creation date of the resource is unattainable , then the timestamp of its publishing or release could suffice as a fairly accurate estimate of the creation date of the resource . as
fire leaves traces of smoke and ashes , web resources leave traces in references , likes , and backlinks . the events associated with creating those shares , links , likes , and interaction with the uri could act as an estimate as well . if we have access to these events , the timestamp of the first event could act as a sufficient estimate of the resource s creation date . in this paper , we investigate using those traces on the web to estimate the creation date of the published resource .
finally , we propose an implementation to this tool based on our analysis to be utilized by researchers .
the problem of estimating the age of web resources has been visited before , but from a different angle .
jatowt et al .
investigated the age of web content posted in dynamic pages @xcite .
they utilized a multiple binary search algorithm to extract the first time the content of a certain dom component of the page started to appear within the page in the archives .
they analyzed multiple versions of the web page provided by the public archives .
after breaking down the page to multiple dom components , the archived versions were explored using binary search for the first existence of each of these components .
the timestamp of this first appearance is recorded indicating an estimate for when the enclosed web content , within each component , was created .
this approach , relies on the archiving coverage of the web provided by the public archives , and the temporal difference between when the content s creation date and the time it was crawled and archived .
this period of time could range from a few hours in heavily archived pages , up to more than a year in other cases . to access and analyze the public archives we utilized the memento framework which facilitated the navigation between the current and the past web @xcite .
we investigated web archival coverage while estimating how much of the web is archived @xcite . in our experiment , we sampled the web forming four different data sets extracted from four sources .
we found that the amount of the web currently archived ( or having at least one accessible past version in the public web archives ) is highly correlated to where the web resource resides .
accordingly , the percentage of coverage ranges from 16% to 79%
. this would be the case for long standing resources that exist on the web at the time of archiving .
our recent study , investigating resources related to multiple historical events since 2009 , showed that the published resources are at continuous risk of disappearance and within the first year of publishing about 11% disappear @xcite .
this is important if the resource whose age we wish to estimate existed on the web only briefly .
this disappearance event might occur prior to the first archival crawl , resulting in complete unattainability of the resource .
an investigation on the web resource itself mining for timestamps in the published content was conducted by inoue and tajima @xcite .
they analyzed web pages for timestamp embedded by content management systems ( cms ) .
this approach supports the most popular date formats but could suffer from ambiguity due to dates the mix in the month versus day order in the uk format versus in the us one .
the authors applied different techniques in attempts to solve this ambiguity .
as accurate the results of this approach could be , it still remains specific to cmss and highly reliant on the content itself , reducing its generality .
we propose analyzing different other sources and services to mine for the first appearance of the resource .
these services vary in reliability and the results they provide which demanded that we conduct an evaluation of each of the services we used and investigate the amount of accuracy lost upon the failure of each service .
it is worth noting that mccown and nelson conducted an experiment to gauge the difference between what some services like google search might provide from both their api versus the web interface @xcite .
they found a significant difference in the results from both sources .
similarly , klein conducted a study analyzing the results from using the delicious.com api vs. screen scraping the web interface @xcite .
he proved that screen scraping provided better results than utilizing the api , which we considered in our analysis .
there are three reasons we can not use just the web archives to estimate the creation date .
first , not all pages are archived .
second , there is often a considerable delay between when the page first appeared and when the page was crawled and archived .
third , web archives often quarantine the release of their holdings until after a certain amount of time has passed ( sometimes 612 months ) .
these three major deficiencies discourage the use of the web archives solely in estimating an accurate creation date timestamp for web resources . in the following sections , we investigate several other sources that explore different areas to uncover the traces of the web resources . utilizing the best of a range of methods
since we can not rely on one method alone , we build a module that gathers this information and provides a collectively estimation of the creation date of the resource .
figure [ fig : timeline ] illustrates the methodology of the age estimation process with respect to the timeline of the resource .
prior to investigating any of the web traces we return back to the basics , to the resource itself .
we send a request for headers to the hosting web server and parse the output .
we search for the existence of last modified date response header and parse the timestamp associated if it exists .
we use the ` curl ` command to request the headers as shown in figure [ fig:0 ] .
we also note the timestamp obtained from the headers can have errors as demonstrated in a study of the quality of etags and last - modified datestamps by clausen @xcite . * ` curl -i http://ws-dl.blogspot.com/2012/02/2012-02-11-losing-my-revolution-year.html ` * -5 mm .... http/1.1 200 ok content - type : text / html ; charset = utf-8 expires : sat , 02 mar 2013 04:04:09 gmt date : sat , 02 mar 2013 04:04:09 gmt cache - control : private , max - age=0 last - modified : we d , 27 feb 2013 17:27:20 gmt etag : " 473ba56b - fd4a-4778-b721 - 3eabdd34154e " x - content - type - options : nosniff x - xss - protection : 1 ; mode = block content - length : 0 server : gse .... typically , we think of backlinks as discoverable through search engines . in the next sections we explore the different forms of backlinks and how we can utilize them in our investigation .
firstly , a backlink refers to the link created on a page _ a _ referring to the intended page _
b_. page _ a _ is considered a backlink of _ b_. if page _ a _ is static and never changed this means that it was created at point in time following the creation of _ b _ , could be by minutes or years . if page _ a _ was change - prone and had several versions , the first appearance of the link to page _ b _ on _ a _ could trigger the same event indicating that that it happened also at a point in time following the creation of _
b_. if we can search the different versions of a throughout time we can estimate this backlink timestamp . to accomplish this , we utilized google api in extracting the backlinks of the uri . note that google api is known to under - report backlinks as shown by mccown and nelson @xcite . to explore the multiple versions of each of the backlinks we utilize the memento framework in accessing the multiple public archives available @xcite
for each backlink we extract its corresponding timemaps .
we use binary search to discover in the time maps the first appearance of the link to the investigated resource in the backlink pages . using binary search
ensures the speedy performance of this section of the age estimating module . with the backlink having the most archived snapshots ( cnn.com > 23,000 mementos ) , the process took less than 15 iterations accessing the web archives .
the minimal of the first appearance timestamps from all the backlinks is selected as the estimated backlink creation date .
similarly , this date can act as a good estimation to the creation date of the resource .
twitter enables users to associate a link with their tweeted text , technically creating a backlink to the shared resource .
when a user creates a web resource and publicizes it on their social network , by tweeting a link to it or posting it on their facebook account , they create backlinks to their resource .
typically , these backlinks are not accessible via a search engine .
the more popular the user and the more the resource gets retweeted or shared , the more backlinks the original resource gains increasing its rank and discoverability in search engines .
to elaborate , we examine the following scenario .
a resource has been created at time @xmath0 , as shown in fig [ fig:1 ] and shortly after a social media post , or a tweet , has been published referring to the resource at time @xmath1 as shown in fig [ fig:2 ] .
this new time @xmath2 , could act as a fairly close estimate to the creation date of the post with a tolerable margin of error of minutes in some cases between the original @xmath0 and @xmath3 .
given this scenario , tweets inherently are published with a creation / posting date which makes it easier to extract . the task remaining is to find the tweets that were published with the targeted resource embedded in the text with incorporating all the shortened versions of the uri as well .
twitter s timeline search facility and its api both provide results of a maximum of 9 days from the current day @xcite .
accordingly , we utilize another service , topsy.com , that enables the user to search for a certain uri and get the latest tweets that incorporated it and the influential users sharing it .
topsy s otter api provides up to 500 of the most recent tweets published embedded a link to the resource and the total number of tweets ever published . except for the highly popular resources ,
the 500 tweets limit is often sufficient for most resources .
the tweets are collected and the corresponding posting timestamps are extracted .
the minimum of these timestamps acts as an indication of the first time the resource was tweeted .
this timestamp in turn signifies the intended @xmath3 mentioned earlier .
another form backlinks could take is uri shortnening .
currently , there are hundreds of services that enables the user to create a short uri that references another longer uri and acts as an alias to it for easier dissemination on the web .
shortened uris could be used for the purposes of customizing the uri or for monitoring the resource by logging the amount of times the short uri have been dereferenced or clicked @xcite . some services , like bitly ,
can provide the users with a lookup capability for long uris .
when a uri is shortened for the first time by a non logged - in user , it creates an aggregate public short uri that is public to everyone .
when other unauthenticated users attempt to shorten the same uri it provides the original first aggregated short uri .
for every logged - in user , the service provides the possibility to create another personal shortened uri . for our purposes we lookup the aggregated short uri indicating the first time the resource s uri have been shortened by this service and from that we query the service once more for the short uri creation timestamp .
bitly has been used as the official automatic shortener for years by twitter before they replaced it with their own shortener . similarly to the previous backlinks method we mine bitly for those creation timestamps and use them as an estimate of the creation date of the resource , assuming the author shortens and shares the resource s uri shortly after publishing it .
the most straightforward approach used in the age estimation module is the web archives analysis .
we utilize the memento framework to obtain the timemap of the resource and from which we obtain the memento datetime for each and then extract the least one indicating the first memento captured .
note that memento datetime is the time of capture at the web archive and is not equivalent to last modified or creation dates @xcite . in some cases
, the original headers in some mementos include the original last modified dates , but all of them have the memento date time fields .
we extract each of those fields , parse the corresponding dates , and pick the lowest of which .
an extra date range filter was added to avoid dates prior to 1995 , before the internet archive began archiving , or more than the current timestamp .
the final approach is to investigate the search engines and extract the last crawled date . except for the highly active and dynamic web pages ,
the resources get crawled once and get marked as such to prevent unnecessary re - crawling .
news sites article pages , blogs , and videos are the most encountered examples of this .
the idea is to use the search engines apis to extract this last crawled date and utilize it as an estimate of the creation date .
this approach is effective due to the relatively short period of time between publishing a resource and its discovery by search engine crawlers .
we use google s search api and modify it to show the results from the last 15 years accompanied by the first crawl date .
unfortunately this approach does not give time granularity ( hh : mm : ss ) , just dates ( yyyy : mm : dd ) .
to validate an implementation of the methods described above , we collect a gold standard dataset from different sources which we can extract the real publishing timestamps .
this could be done by parsing feeds , parsing web templates , and other methods . in the next sections we illustrate each of the sources
utilized and explain the extraction process .
two important factors were crucial in the data collection process : the quality of the timestamps extracted , and the variety of the sources to reduce any bias in the experiment .
thus , we divide data into four categories .
table [ tab:1 ] summarizes the four categories .
each article is associated with a timestamp in a known template that can be parsed and extracted .
the articles are also usually easily accessible through rss and atom feeds or xml sitemaps . for each of the news sites under investigation we extracted as many resources as possible then randomly downsized the sample . to increase the variety of the gold standard dataset we investigate five different social media sources .
these selected sources are highly popular , and it is possible to extract accurate publishing timestamps . as those sources are tightly coupled with the degree of popularity and to avoid the bias resulting from this popularity we randomly extract as many resources as possible from the indexes , feeds , and sitemaps and do not rely solely on the most famous blogs or most shared tumblr posts .
furthermore , we randomly and uniformly sample each collection to reduce its size for our experiment . so as not to limit our gold standard dataset to low level articles , blogs , or posts only , we incorporated top level , long - standing domains . to extract a list of those domains we mined alexa.com for the list of the top 500 sites .
this list of sites was in turn investigated for the dns registry dates using one of the dns lookup tools available online . a final set of 100 was randomly selected from the resolved sites and added to the gold standard dataset .
finally , we randomly select a set of 100 uris that we can visually identify the timestamp somewhere on the page itself .
these uris were selected empirically using random walks on the web .
the 10 uris analyzed in @xcite included within these 100 uris as well .
the corresponding true value of the creation timestamp for each of the 10 uris is the one provided in their analysis .
the collected dataset of 1,200 data points is tested against the developed implementation of the carbon dating methods and the results are recorded .
since the data points are collected from different sources , the granularity varies in some cases , as well as the corresponding time zones . to be consistent , each real creation date timestamp @xmath4 is transformed from the corresponding extracted timestamp to coordinated universal time ( utc ) and the granularity for all the timestamped have been set to be a day .
each data point has a real creation date in the iso 8601 date format without the time portion ( e.g. , yyyy : mm : dd ) .
similarly , the extracted estimations were processed in the same manner and recorded . for each method ,
we record the estimated timestamp @xmath5 and the temporal delta @xmath6 between the estimated timestamp @xmath5 and the actual one @xmath4 as shown in equation [ eq:1 ] .
collectively , we calculate the best estimated timestamp @xmath7 as in equation [ eq:2 ] , the closest delta between all the methods @xmath8 and the real timestamp @xmath4 as shown in equation [ eq:3 ] , and the method that provided this best estimate . @xmath9 @xmath10 @xmath11 table [ tab : auc ] shows the outcomes of the experiment .
the numbers indicate how many times a resource provided the closest timestamp to the real one .
it also shows that for 290 resources , the module failed to provide a single creation date estimate ( 24.90% ) .
as our age estimation module relies on other services to function ( e.g. , bitly , topsy , google , web archives ) ; the next step is to measure the effect of each of the six different age estimation methods and to gauge the consequences resulting in failure to obtain results from each . for each resource
we get the resulting best estimation and calculate the distance between it and the real creation date .
we set the granularity of the delta to be in days to match the real dates in the gold standard dataset .
to elaborate , if the resource was created on a certain date and the estimation module returned a timestamp on the same day we declare a match and in this case @xmath8 = 0 . to measure the accuracy of estimation
, 393 resources out of 1200 ( 32.78% ) returned @xmath8 = 0 indicating a perfect estimation .
for all the resources , we sort the resulting deltas and plot them .
we calculate the area under the curve using the composite trapezoidal rule and the composite simpon s rule with x - axis spacing of 0.0001 units .
we take the average of both approximations to represent the area under the curve ( auc ) .
semantically , this area signifies the error resulting from the estimation process .
ideally , if the module produced a perfect match to the real dates , auc = 0 .
table [ tab : auc ] shows that the auc using the best lowest estimate of all the six methods is 762.64 .
disabling each method one by one and measuring the auc indicates the resultant error corresponding to the absence of the disabled method accordingly .
the table shows that using or disabling the use of backlinks barely affected the results .
disabling the bitly services or the google search index query affected the results slightly ( 0.51% and 2.64% respectively ) . while disabling any of the public archives query , or the social backlinks in topsy and the extraction of the last modified date if exists hugely affects the results increasing the error tremendously .
we utilized polynomial fitting functions to fit the values corresponding to the age estimations corresponding to each uri .
figure [ fig : fitted ] shows the polynomial curve of the second degree used in fitting the real creation times stamps of the gold standard dataset .
figure [ allfitted ] shows the fitted curve resulting from removing each of the methods one by one .
each of the curves signifies an estimate of the best the other methods could provide .
the further the estimated curve is from the real one the less accurate this estimation would be .
+ + * ` curl -i http://cd.cs.odu.edu/cd/http://www.mementoweb.org ` * -5 mm .... http/1.0 200 ok date : fri , 01 mar 2013 04:44:47 gmt server : wsgiserver/0.1 python/2.6.5 content - length : 550 content - type : application / json ; charset = utf-8 { " uri " : " http://www.mementoweb.org " , " estimated creation date " : " 2009 - 09 - 30t11:58:25 " , " last modified " : " 2012 - 04 - 20t21:52:07 " , " bitly " : " 2011 - 03 - 24t10:44:12 " , " topsy.com " : " 2009 - 11 - 09t20:53:20 " , " backlinks " : " 2011 - 01 - 16t21:42:12 " , " google.com " : " 2009 - 11 - 16 " , " archives " : { " earliest " : " 2009 - 09 - 30t11:58:25 " , " by archive " : { " wayback.archive-it.org " : " 2009 - 09 - 30t11:58:25 " , " api.wayback.archive.org " : " 2009 - 09 - 30t11:58:25 " , " webarchive.nationalarchives.gov.uk " : " 2010 - 04 - 02t00:00:00 " } } } ....
after validating the accuracy of the developed module the next step was to openly provide age estimation as a web service .
to fulfill this goal , we created `` * * _ carbon date _ * * '' , a web based age estimation api . to use the api , simply concatenate the uri of the desired resource to the following path : + _ http://cd.cs.odu.edu/cd/ _ and the resulting json object would be similar to the one illustrated in figure [ fig : json ] .
estimating the age of web resources is essential for many areas of research .
previous research investigated the use of public archives as a point of reference to when the content of a certain page appeared . in this study , we investigated several other possibilities in estimating the accurate age of a resource including social backlinks ( social posts and shortened uris ) , search engine backlinks , search engine last crawl date , the resource last modifed date , the first appearance of the link to the resource in its backlinks sites , and the archival first crawl timestamp .
we also incorporated the minimum of the original headers last modified date , and the memento - datetime http response header .
all of these methods combined , where we select the oldest resulting timestamp , proved to provide an accurate estimation to the creation date upon evaluating it against a gold standard dataset of 1200 web pages of known publishing / posting dates .
we succeeded in obtaining an estimated creation date to 910 resources out of the 1200 in the dataset ( 75.90% ) .
40% of the closest estimated dates were obtained from google , topsy came in second with 26% , followed by the public archives , bitly , and last modified header with 17% , 11% , and 6% respectively . using
the backlinks yielded only 3 closest creation dates proving its insignificance .
we also simulate the failure of each of the six services one at a time and calculated the resulting loss in accuracy .
we show that the social media existence ( topsy ) , the archival existence ( archives ) and the last modified date if it exists , are the strongest contributers to the age estimation module respectively .
this work was supported in part by the library of congress and nsf iis-1009392 .
s. g. ainsworth , a. alsum , h. salaheldeen , m. c. weigle , and m. l. nelson .
how much of the web is archived ? in _ proceeding of the 11th annual international acm / ieee joint conference on digital libraries _
, jcdl 11 , 2011 .
d. antoniades , i. polakis , g. kontaxis , e. athanasopoulos , s. ioannidis , e. p. markatos , and t. karagiannis .
we.b : the web of short urls . in _ proceedings of the 20th international conference on world wide web _ , www 11 , pages 715724 , new york , ny , usa , 2011 .
m. inoue and k. tajima .
noise robust detection of the emergence and spread of topics on the web . in _ proceedings of the 2nd temporal web analytics workshop _ , tempweb 12 , pages 916 , new york , ny , usa , 2012 .
a. jatowt , y. kawai , and k. tanaka . detecting age of page content . in _ proceedings of the 9th annual acm international workshop on web information and data management _ , widm 07 , pages 137144 , new york , ny , usa , 2007 .
acm .
h. m. salaheldeen and m. l. nelson .
losing my revolution : how many resources shared on social media have been lost ? in _ proceedings of the second international conference on theory and practice of digital libraries _ , tpdl12 , pages 125137 , berlin , heidelberg , 2012 .
springer - verlag . | in the course of web research it is often necessary to estimate the creation datetime for web resources ( in the general case , this value can only be estimated ) .
while it is feasible to manually establish likely datetime values for small numbers of resources , this becomes infeasible if the collection is large .
we present `` carbon date '' , a simple web application that estimates the creation date for a uri by polling a number of sources of evidence and returning a machine - readable structure with their respective values . to establish a likely datetime , we poll bitly for the first time someone shortened the uri , topsy for the first time someone tweeted the uri , a memento aggregator for the first time it appeared in a public web archive , google s time of last crawl , and the last - modified http response header of the resource itself .
we also examine the backlinks of the uri as reported by google and apply the same techniques for the resources that link to the uri .
we evaluated our tool on a gold standard data set of 1200 uris in which the creation date was manually verified .
we were able to estimate a creation date for 75.90% of the resources , with 32.78% having the correct value . given the different nature of the uris , the union of the various methods produces the best results .
while the google last crawl date and topsy account for nearly 66% of the closest answers , eliminating the web archives or last - modified from the results produces the largest overall negative impact on the results .
the carbon date application is available for download or use via a web api .
= 10000 = 10000 |
observations in the past decade have been interpreted as suggesting that 70% of the matter energy density in the universe at the present epoch is in the form of a smooth vacuum energy , or `` dark energy '' , which does not clump gravitationally .
this is supported by two powerful independent lines of observation .
firstly , type ia supernovae in distant galaxies ( perlmutter 1998 , 1999 ; riess 1998 , 2004 ) are dimmer than would be expected in standard friedmann
walker ( frw ) models , especially when it is noted that many independent dynamical estimates of the present clumped mass fraction , @xmath0 , suggest values of order 2030% .
secondly , observations of the power spectrum of primordial anisotropies in the cosmic microwave background radiation ( cmbr ) , most recently by the wmap satellite ( bennett 2003 ) , indicate that the universe appears to be spatially flat on the largest of scales . a cosmological constant , or alternatively dynamical dark energy , is most commonly invoked to explain the observed cosmological parameters , even though a fundamental origin for such dark energy remains one of the profoundest mysteries of modern physics
. however , even in the presence of dark energy at the present epoch , a number of problems remain .
one of the most significant problems is that the epoch of reionization measured by wmap appears at a redshift of order @xmath3 , indicating that the first stars formed much earlier than conventional models of structure formation would suggest .
the detection of complex galaxies at relatively large redshifts compounds the conundrum ( see , e.g. , cimatti 2004 , glazebrook 2004 ) . in recent work , ( 2005 )
have proposed a profoundly different resolution of the `` cosmological constant problem '' .
they reason that primordial inflation will have produced density perturbations many times larger than the present horizon volume .
we should not view the observable universe as typical of the universe on scales larger than our particle horizon , and the values of cosmological parameters observed should not necessarily be taken as typical of the whole .
furthermore , they suggest that our present observations might be compatible with the observed universe being an underdense bubble in an otherwise spatially flat @xmath4 frw universe with an energy density @xmath5 in ordinary matter . such assumptions are indeed supported by detailed numerical and analytic calculations made in inflationary models a decade ago ( linde , linde and mezhlumian 1996 ) .
the possibility of inhomogeneously defined clock rates is commonly accepted in studies of inhomogeneous cosmologies , such as the lematre tolman
bondi ( ltb ) models .
( for a review see krasiski ( 1997 ) . ) in this _ letter _ , i will argue , however , that primordial inflation can give rise to a particular structure in inhomogeneous cosmologies , which allows for a homogeneous cosmic time on the scales of the first bound systems which form , which differs from the `` comoving '' time parameter of the average late epoch geometry .
this novel feature is what distinguishes the present model from previous studies of inhomogeneous cosmologies , as it leads to a new solution of the _ fitting problem _ ( ellis and stoeger 1987 ) .
it is a consequence of the inflationary paradigm that a spectrum of initially small density perturbations is stretched to all observable scales within our past light cone , and also to scales beyond our particle horizon .
the fact that this is true for the past light cone is well supported by the cmbr . the hypothesis that such perturbations should extend to super
horizon scales is a feature of most inflationary models , independent of their details .
one important realisation is the fact that since primordial inflation ended at a finite very early time , the scale of super horizon sized modes , although huge , must have a cut off at an upper bound . beyond that scale
we assume the universe is described by a spatially flat bulk metric [ frwflat ] ^2 = - ^2 + |a^2()(x^2+y^2+z^2 ) , where @xmath6}$ ] , and we use units with @xmath7 . the precise bulk equation of state @xmath8 proves to be inconsequential if @xmath9 , though in accord with the principles of the model we assume only ordinary matter and radiation .
although the observable universe will undoubtedly be embedded in many regions of under- and over - density , like the smallest figure inside a russian doll , it is nonetheless reasonable to assume that provided the density perturbation immediately containing our observed universe extends sufficiently beyond our horizon then there is a super - horizon sized underdense bubble containing the observable universe , with matter density _ equal to the average matter density we measure _ , which we can model as a super horizon sized underdense region , .
the matter distribution inside is assumed to be _ inhomogeneous and `` fractal '' _ , in accord with calculations from primordial inflation ( linde , linde and mezhlumian 1996 ) and in accord with all the evidence of observations of the actual universe , with its large - scale structure of streaming motions of clusters of galaxies , bubbles and voids .
nonetheless , despite this inhomogeneity we do observe an average isotropic hubble flow .
thus at some level the first step in solving the fitting problem ( ellis and stoeger 1987 ) must involve the approximation of the inhomogeneous geometry of by an average spacetime which depends on a single cosmic scale factor , viz .
, the spatially open frw geometry [ frwopen1 ] s^2 = - ^2 + ^2 ( ) , where parametrically in terms of conformal time , @xmath10 , & = & a_i2(1-)(-1 ) , + h_i&=&2(1-)^3/2(- ) , [ frwopen2]where @xmath11 is an initial density parameter of the bubble , , and @xmath12 a constant to be set .
usually the first step after writing down the metric ( [ frwopen1 ] ) is to identify comoving observers in this geometry with idealised isotropic observers , namely _ observers who measure no dipole anisotropy in the cmbr_. one identifies typical stars in typical galaxies with such observers , apart from the small effects of peculiar velocities .
however , even in an homogeneous isotropic frw model this implicitly assumes a solution of the fitting problem in terms of a complicated matching of asymptotic scales to relate the clocks on geodesics in bound systems , where space is not expanding , to the scale of comoving observers in the hubble flow , where space is expanding .
it is assumed without question that such clocks can be identified . in general relativity with the particular self
similar inhomogeneous geometry that arises from the evolution of the density perturbations of primordial inflation , the standard solution of the fitting problem is not appropriate , as i argue in detail in a subsequent paper ( wiltshire 2005 ) , henceforth ref .
i. in brief , while the underdense bubble , , is on its largest scale a single perturbation away from the average bulk density , _ due to the scale - invariance of the inflationary spectrum , perturbations on the smaller and intermediate scales that give rise to the first structures nonetheless have the same statistical distribution as the bulk universe , with a mean density distributed about that of the bulk_. the particular structure of inhomogeneity we see today , with its large voids , results in part from a _
`` particle horizon volume selection bias '' _ in the initial density perturbations ( see ref .
i ) . while the metric ( [ frwopen1 ] ) is still valid for describing the clocks of an observer at an average _ spatial _ position , in an inhomogeneous underdense bubble such _ average spatial positions are in voids and do not coincide with stars and galaxies_. on the other hand , the first bound systems of stars and star clusters which aggregate to typical galaxies , break away from the hubble flow at early epochs when the average local density of matter in their past light cones is close to the average density of the bulk .
such systems retain fragments of the bulk hypersurface geometry .
a clock in a system which has had an approximate stationary killing vector since the epoch of break away , thus measures the bulk cosmic time parameter , @xmath13 , which is `` frozen in '' from an earlier epoch in the evolution of the universe .
initial perturbations with the selection biased underdensity of the bubble , , evolve to form voids .
given that structure forms in a bottom up manner , a _
`` temporal '' _ copernican principle operates within in the sense that _ as average isotropic observers we assume we are located in a bound system formed from those density perturbations which broke away from the hubble flow when the geometry within their past light cone was indistinguishable from that of the bulk surfaces of homogeneity_. rather than occupying an average position on a spatial hypersurface , we find ourselves in a bound system that formed from matter which broke from an almost homogeneous hubble flow at an average epoch . at the present epoch , these same average galaxies ,
are located in clusters in bubble walls surrounding local voids , in a self similar hierarchical structure , which we loosely call _
fractal_. since recent expansion of space occurs primarily in voids , the question of who does or does not measure an isotropic cmbr depends not only on local peculiar velocities but also on whether an observer s line of sight to the surface of last scattering in any direction on the sky averages over the same number and volume distribution of voids and bubble walls , once such structures form . for observers in average galaxies in bubble walls or observers within small voids one would expect a roughly isotropic distribution of bubbles and voids within the self similar hierarchical matter distribution , and hence an almost isotropic cmbr .
clock rates are determined by local geometry rather than by isotropy , or otherwise , of the cmbr at any location . to define the local cosmic time ,
@xmath13 , in bound systems in average galaxies , as we have done , in reference to bulk hypersurfaces of matter homogeneity which stretch to regions beyond our present particle horizon may seem puzzling at first .
however , prior to inflation , such regions _ were in causal contact with the observed universe_. thus the same basic processes of inflation that lead to isotropy of the cmbr also lead , via structure formation from scale invariant perturbations , to a definition of inertial frames of observers in average galaxies , namely a new variant of mach s principle .
( see ref .
i for further discussion . ) since general relativity is a local theory , the average geometry in our first order solution of the fitting problem is described by , but insofar as measurements , including our own , are referred to the frame of average galaxies we must refer them to a coordinates related to those of ( [ frwopen1 ] ) by a non trivial lapse function @xmath14 , so that ( [ frwopen1 ] ) becomes [ frwopen3 ] s^2 = ^2()^2 , where [ frwopen4 ] ^2= -^2 + ^2(),and @xmath15 . _ since quantities such as the luminosity distance involve null geodesics , for many calculations we can use the conformally related geometry ( [ frwopen4 ] ) .
_ it is not necessary to solve for the geodesics of observers , defined by local geometry . only null geodesics probe the vast cosmological scales over which the average geometry ( [ frwopen1 ] ) is more relevant than other scales in the actual inhomogeneous geometry .
spacelike intervals are never directly measured .
thus by fiat , once we insist on writing the average cosmological geometry in a `` synchronous gauge '' adapted to the proper time , @xmath13 , of the local clocks of observers in average galaxies , then the geometry ( [ frwopen4 ] ) also comes to define our rods when specifying areas , volumes and densities on the largest scales .
the quantity @xmath16 leads to a gravitational time dilation between average galaxies in the gravitational wells of the bubble walls and the ideal comoving clocks in empty voids , as further discussed in ref i. to analyse some quantities will require a specification of the entire inhomogeneous fractal geometry within .
this may be possible in the context of a ltb model .
however , given the hierarchical nature of the inhomogeneity the ltb mass function would be much more complicated than those of the simple single- or few - void ltb models that have typically been studied to date .
initially , we are only interested in properties defined by the average hubble flow , although we might expect some modifications of cosmological parameters due to the different range of scales and expansion rates among local voids ( tomita 2001 ) . while solving for intermediate scales , perhaps in terms of a ltb model , will ultimately be necessary , for the purpose of the first approximation in the fitting problem , it suffices to construct a `` spherical expansion model '' of the underdense region , in parallel to the well known `` spherical collapse model '' ( kolb & turner 1990 ) .
following the standard approach , we assume the initial density parameter is set sufficiently early that it is very close to unity : @xmath17 , @xmath18 . furthermore , since the universe has @xmath19 initially we require the comoving scales to match at that epoch .
this means we set @xmath20 , @xmath21 , @xmath22 and @xmath23 .
there are various possible hubble parameters to take into account:*(i ) * the background hubble parameter @xmath24 $ ] of the metric ( [ frwflat ] ) which can not be directly measured in ; the hubble parameter of comoving observers in ( [ frwopen1 ] ) , measurable only by unbound observers in average voids ( ) ; [ hvoid]*(iii ) * the physical hubble parameter we measure h ( ) ; [ hphys ] an effective ( unmeasured ) parameter ( ) , [ heff]which is useful for the purposes of calculations . by our assumptions , while all these parameters coincide at early times , as the universe expands they may differ . as observers in average galaxies , using null geodesics we measure the cosmological geometry ( [ frwopen4 ] ) referred to global cosmic time , @xmath13 , not to the parameter @xmath25 .
we simply need to determine @xmath26 to determine cosmological quantities . using eqs . , the bulk universe scale factor , @xmath27 , where @xmath28 $ ] and @xmath29 , the density parameter is defined as = ( a_i)^3(|a|a_i)^2/n= 18h_i^2(1-)^3 ^ 2 ^ 2(-1)^3 .
[ bom]comparing this with the standard parametric expression @xmath30 derived from , we obtain a parametric relation for @xmath31 , h_i=(-1)^23(1-)^3/2 . [ age1]the effective parameter ( [ heff ] )
is found to be ( ) = ( 1-)^3/2(+2)(+1)^3/2 ( -1)^3/2(+2 ) [ hubble_eff]where @xmath32 , and a subscript zero refers to the present epoch .
the physically measured hubble parameter ( [ hphys ] ) is h()=(^2 + 2 + 3 ) ( ) ( + 1)(+2 ) .
[ hubble]and the measured hubble constant , @xmath1 , is related to @xmath33 by = ( 2+^22+).observe that at early times , @xmath34 , @xmath35 and @xmath36 , as expected but at late times , @xmath37 .
the lapse function is given by ( ) = = = 3(+1)2(+2 ) .
[ gamma]at the present epoch @xmath38
. we must also be careful to note that the locally measured density parameter , differs from ( [ bom ] ) by a volume factor according to @xmath39 , so that the present epoch measured matter density fraction is = 27(2+)^3 , with inverse @xmath40 - 2 $ ] .
the expansion age ( [ age1 ] ) is usefully rewritten as ( ) = ( 2+^2)(-1)^3/2(1-)^3/2(+2)^2 ( + 1)^1/2 .
[ age2 ] the physically measured deceleration parameter @xmath41 , is given by q()=7 ^ 2 + 10 + 1(^2 + 2 + 3)^2 .
[ decel]it is equal to the bulk deceleration parameter @xmath42 at early times @xmath43 , but at late times as @xmath44 , @xmath45 , so that it is small but positive at the present epoch .
this is just as expected from a model without dark energy , and contradicts the claims of cosmic acceleration by kolb ( 2005 ) .
however , we must recall that the supernovae measurements involve luminosity _ distances _ , and the _ interpretation _ of cosmic acceleration depends on the time parameter assumed in taking derivatives .
the present model may be recognised as mimicking a milne universe at nearby redshifts , and this still provides a reasonable fit to present sneia data ( carter 2005 ) .
to compare with observations eqs .
( [ hubble])([decel ] ) need to be expressed in terms of the observed cosmological redshift , @xmath46 .
great care is needed at this point in identifying physical quantities .
it proves simplest to always use the line element ( [ frwopen3 ] ) , ( [ frwopen4 ] ) remembering that @xmath13 corresponds to clock time for observers in average galaxies .
it follows that 1+z = a0a=0=(1-)(2 + ) ( + 1)(+2)(-1 ) .
[ redshift]the physical solution to eqn .
( [ redshift ] ) is & = & -12+(1-)(2+)2(z+1 ) + & + & 2(z+1 ) .
+ [ solutn ] we are now ready for our first cosmological tests . the luminosity distance is readily computed in the standard fashion and is found to be @xmath47 , and since @xmath48 it follows that = ( 1+z)(2+^2)(2+)\{2-(2- ) } .
[ dl]as shown by carter ( 2005 ) the model gives a good fit to the supernova data , with or without non baryonic cold dark matter in addition to baryonic matter .
the expansion age is given by substituting ( [ solutn ] ) in ( [ age2 ] ) .
it gives a present day age of the universe of = 2(2+^2)(2+)^2.using the best
fit value ( carter 2005 ) for @xmath49 , the age of the universe is @xmath50 gyr for @xmath51 , or alternatively @xmath52 gyr for a universe with only baryonic matter , @xmath53 , using the recalibrated primordial nucleosynthesis bounds of ref .
i. importantly , the expansion age is significantly larger at large @xmath46 : with @xmath54 , for @xmath51 we find @xmath55 gyr at @xmath56 , @xmath57 gyr at @xmath58 , and @xmath59 gyr at @xmath60 . for @xmath53
then @xmath61 gyr at @xmath56 , @xmath62 gyr at @xmath58 , and @xmath63 gyr at @xmath60 .
this would buy precious time for structure formation early on , and help to explain the redshift of the reionization epoch ( bennett 2003 ) , and observations of `` early '' formation of structure ( cimatti 2004 , glazebrook 2004 ) .
as a model cosmology , the present _ fractal bubble universe _ is remarkable in that its gross features depend only on two already observed parameters , and the results obtained thus far are reasonably promising .
if correct , then all observed quantities in cosmology must be systematically reanalysed .
the determination of some quantities will require further development of our understanding of the inhomogeneous matter distribution within , including the largest scale local voids ( tomita 2001 ) , in particular .
however , the present model offers a clear framework for calculations , as is demonstrated in ref .
i , where further steps are taken in the reanalysis of the hot big bang and observational implications for the cmbr
. it would be ironic that einstein s idea concerning the irrelevance of the cosmological constant , and his idea of the importance of mach s principle , may prove to both be right in understanding the universe . * acknowledgement * this work was supported by the mardsen fund of the royal society of new zealand .
i thank jorma louko , jenni adams , roy kerr and other members of the uc gravity group for helpful comments .
cimatti , a. , 2004 , * 430 * , 184 .
ellis , g.f.r . and stoeger , w. 1987 , class .
quantum grav .
* 4 * , 1697 .
glazebrook , k. , 2004 , * 430 * , 181 .
kolb , e.w .
, matarrese , s. , notari , a. and riotto , a. 2005 , arxiv : hep - th/0503117 .
kolb , e.w . and m. s. turner , m.s . 1990 ,
_ `` the early universe '' _ , ( addison wesley , reading , mass . )
linde , a.d . ,
linde , d. , and mezhlumian , a. 1996 , * 54 * , 2504 .
perlmutter , s. , 1997 , * 483 * , 565 ; perlmutter s. , 1999 , * 517 * , 565 .
riess , a.g .
, 1998 , * 116 * , 1009 ; riess , a.g . , 2004 , * 607 * , 665 .
tomita , k. 2001 , * 326 * , 287 .
wiltshire , d.l .
2005 , in preparation . | a new model of the observed universe , using solutions to the full einstein equations , is developed from the hypothesis that our observable universe is an underdense bubble , with an internally inhomogeneous fractal bubble distribution of bound matter systems , in a spatially flat bulk universe .
it is argued on the basis of primordial inflation and resulting structure formation , that the clocks of the isotropic observers in average galaxies coincide with clocks defined by the true surfaces of matter homogeneity of the bulk universe , rather than the comoving clocks at average spatial positions in the underdense bubble geometry , which are in voids .
this understanding requires a systematic reanalysis of all observed quantities in cosmology .
i begin such a reanalysis by giving a model of the average geometry of the universe , which depends on two measured parameters : the present matter density parameter , @xmath0 , and the hubble constant , @xmath1
. the observable universe is not accelerating .
nonetheless , inferred luminosity distances are larger than navely expected , in accord with the evidence of distant type ia supernovae .
the predicted age of the universe is @xmath2 gyr .
the expansion age is larger than in competing models , and may account for observed structure formation at large redshifts .
# 1 # 1#2#3#1#3#1#2 # 1 _ # 1 _ # 1m^#1 = cmr7 # 1 _ # 1 # 1phys .
rev .
lett . *
# 1 * # 1phys .
rev .
* # 1 * # 1astrophys .
j. * # 1 * # 1astron .
j. * # 1 * # 1#2 |
the national basketball association ( nba ) launched a public database in september 2013 containing over 80 new statistics captured by stats llc through their innovative sportvu player tracking camera systems@xcite .
the cameras capture and record the location of players on the court as well as the location of the ball , and the data are used to derive many different interesting and useful stats that expand greatly upon the traditional stats available for analysis of basketball performance .
we can now break down shot attempts and points by shot selection ( e.g. driving shots , catch and shoot shots , pull up shots ) , assess rebounding ability for contested and uncontested boards , and even look at completely new statistics like average speed and distance and opponent field goal percentage at the rim .
the availability of such data enables fans and analysts to dig into the data and uncover insights previously not possible due to the limited nature of the data at hand .
+ for example , techniques to uncover different positions based on grouping statistical profiles have become increasingly popular .
it reflects the mindset of current nba coaches and general managers who are very much aware of the different types of players beyond the five traditional roles , but a recent proposal@xcite has received criticism for its unintuitive groupings and inability to separate out the impact of player talent@xcite .
the nba player tracking data has the ability to differentiate player performance across more dimensions than before ( e.g. shot selection , possession time , physical activity , etc . ) which can provide better ways to evaluate the uniqueness and similarities across nba player abilities and playing styles .
additionally , many research methods for basketball analysis rely on estimation of possessions and other stats to produce offensive and defensive ratings@xcite . with the ability to track players time of possession and proximity to players in possession of the ball through player tracking
, we can develop more accurate representations of possessions and better player offensive and defensive efficiency metrics .
+ however , the high dimensionality of this new data source can be troublesome as it demands more computational resources and reduces the ability to easily analyze and interpret findings .
we must find a way to reduce the dimensionality of the data set while retaining the ability to differentiate and compare player performance .
one method that is particularly well - suited for this application is principal component analysis ( pca ) which identifies the dimensions of the data containing the maximum variance in the data set .
this article applies pca to the nba player tracking data to discover four principal components that account for 68% of the variability in the data for the 2013 - 2014 regular season .
these components are explored in detail by examining the player tracking statistics that influence them the most and where players and teams fall along these new dimensions .
+ in addition to exploring player and team performances through the principal components , a simple measure of similarity in statistical profiles between players and teams based on the principal components is proposed .
the statistical diversity index ( sdi ) can be calculated for any pairwise player combination and provides a fast and intuitive method for finding players with similar statistical performances along any or all of the principal component dimensions .
this approach is also advantageous from the standpoint of scalability .
the possibilities to derive new statistics from the player tracking data are endless , so as new statistics emerge , this approach can again be applied using the new and existing data to reconstruct the principal components and sdi for improved player evaluation and comparisons .
+ numerous applications in personnel management exist for the use of sdi and the principal component scores in evaluating and comparing player and team statistical performances .
two specific case studies are presented to show how these tools can be used to quickly identify players with similar statistical profiles to a certain player of interest for the purpose of identifying less expensive , similarly skilled players or finding suitable replacement options for a key role player within the organization . +
this article is organized as follows .
section [ sec : data ] describes the player tracking data and data processing .
section [ sec : pca ] provides the analysis and interpretations for the four principal components in detail , showing how players and teams can be compared across these new dimensions .
section [ sec : sdi ] introduces the calculation for sdi and two case studies where principal component scores and sdi are used to find players with similar statistical profiles to all - star tony parker and role player anthony morrow for personnel management purposes .
section [ sec : conclusion ] concludes the article with final remarks .
currently , there are over 90 new player tracking statistics , and data for all 482 nba players from the 2013 - 2014 regular season are available .
separate records exist for players who played for different teams throughout the season , including a record for overall performance across all teams .
brief descriptions of the newly available statistics adapted from the nba player tracking statistics website@xcite are provided for reference and will be helpful in better understanding the analysis going forward .
+ * shooting * + traditional shooting statistics are now available for different shot types : * _ pull up shots _ - shots taken 10 feet away from the basket where player takes 1 or more dribbles prior to shooting * _ driving shots _ - shots taken where player starts 20 or more feet away from the basket and dribbles less than 10 feet away from the basket prior to shooting * _ catch and shoot shots _ - shots taken at least 10 feet away from the basket where player possessed the ball for less than 2 seconds and took no dribbles prior to shooting * assists * + new assist categories are available that enhance understanding of offensive contribution : * _ assist opportunities _ - passes by a player to another player who attempts a shot and if made would be an assist * _ secondary assists _ - passes by a player to another player who receives an assist * _ free throw assists _ - passes by a player to another player who was fouled , missed the shot if shooting , and made at least one free throw * _ points created by assists _ - points created by a player through his assists * touches * + location of possessions provides insight into style of play and scoring efficiency : * _ front court touches _ - touches on his team s offensive half of the court * _ close touches _ - touches that originate within 12 feet of the basket excluding drives * _ elbow touches _ - touches that originate within 5 feet of the edge of the lane and the free throw line inside the 3-point line * _ points per touch _ - points scored by a player per touch * rebounding * + new rebounding statistics incorporate location and proximity of opponents : * _ contested rebound _ - rebounds where an opponent is within 3.5 feet of the rebound * _ rebounding opportunity _ - when a player is within 3.5 feet of a rebound * _ rebounding percentage _ - rebounds over rebounding opportunities * rim protection * + when a player is within 5 feet of the basket and within 5 feet of the offensive shooter , opponents shooting statistics are available to measure how well a player can protect the basket . *
speed and distance * + players average speeds and distances traveled per game are also captured and broken out by offensive and defensive plays
. only players who played at least half the 2013 - 2014 regular season , 41 games , are included in this analysis .
this restriction is made to reduce the influence of player statistics derived from only a few games played .
also fields containing season total statistics and per game statistics are dropped from the analysis since they could be influenced by number of games and minutes played throughout the season . instead , per 48 minutes , per touch , and per shot statistics are used .
the final data set contains 360 player records each containing 66 different player tracking statistics .
with numerous player tracking statistics already available and the potential to develop infinitely many more , it is increasingly difficult to extract meaningful and intuitive insights on player comparisons .
now that the data are available for more granular and detailed comparisons , a methodology is needed that can analyze the entirety of the data set to extract a handful of dimensions for comparisons .
these dimensions should be constructed in a way that ensures optimality in differentiating players ( i.e. dimensions should retain the maximum amount of player separability possible from the original data ) and can be understood in terms of the original statistics .
+ principal component analysis ( pca ) , developed by karl pearson in 1901@xcite and later by hotelling in 1933@xcite , is a particularly well - suited statistical tool that can accomplish this task through identifying uncorrelated linear combinations of player tracking statistics that contain maximum variance .
interested readers can find a brief technical introduction to pca in appendix [ app : pcaintro ] .
components of high variance help us to better differentiate player performance in these directions in hopes that the majority of the variance will be contained in a small subset of components .
this simple and intuitive approach to dimension reduction provides a platform for player comparisons across dimensions that best separate players by statistical performance and can be implemented without expensive proprietary solutions , providing more visibility into how the method works at little to no additional cost .
pca is sensitive to different variable scalings in the original data set such that variables with larger variances may dominate the principal components if not adjusted . to set every statistic on equal footing
, all statistics are standardized with mean 0 and variance 1 prior to conducting the analysis .
+ pca is most useful when the majority of the total variance across variables are captured by only a few of the principal components , thus the dimension reduction .
if this is the case for the nba player tracking data set , it means that we are able to retain the ability to differentiate player performance without having to operate in such a high dimensional space .
figure [ fig : scree ] shows the variance captured by each principal component .
note the variance captured in the first principal component is very high and decreases drastically through the first four components .
after that , the change in variance is relatively flat , forming an elbow shape in the plot .
this means that the variances captured by the fifth component onward are very similar and much smaller than the first four components .
moreover , the first four components capture 68% of the variance across the original variables , so we utilize these four components going forward to analyze and compare player performance and playing styles .
each principal component ( pc ) is a linear combination of the original variables in the dataset .
there is a vector for each principal component containing the coefficients associated with each of the variables in the original data set and are called _ loading vectors_. these describe the influence of each variable for each principal component and are used to interpret these new dimensions in terms of the original variables .
figure [ fig : pcload ] plots the categorized loading coefficients for the four principal components and is explored in detail in the following sections .
variables can have a positive or negative contribution to the principal component . while the sign is arbitrary , understanding which variables contribute positively or negatively can help with interpreting the principal components .
the most important statistics for each component are presented in the following sections , but tables containing all loading coefficients for variables contributing significantly to the principal components are also available in appendix [ app : loadings ] for more details .
+ each player can then be given a set of pc scores by multiplying the standardized statistics by their corresponding loading coefficients and then taking a sum ( see appendix [ app : pcaintro ] for details ) .
figure [ fig : pcscore ] contains plots of the pc scores for all players with a select few noted for illustration . using the loading and score plots here
, we can begin to understand and interpret what these new dimensions are capturing and use them for player comparisons .
the first principal component accounts for the most variation , 42% of the total variance .
table [ tab : pc1top ] lists statistics with highly positive and negative loadings for pc 1 , and refer back to figure [ fig : pcload ] for categorized pc 1 loadings for all statistics .
these are used to better understand the meaning of the scores along this dimension .
players with positive scores for pc 1 are able to secure rebounds of all kinds and are responsible for defending the rim and close shots .
notable examples are andre drummond , deandre jordan , and omer asik . while players with negative scores for pc 1 drive the basketball to the hoop and take pull up and catch and shoot shots often , which implies they tend to be outside players .
these players also tend to possess the ball more often and generate additional offense through assists .
examples here are stephen curry , tony parker , and chris paul .
pc 2 accounts for another 12% of the total variance and table [ tab : pc2top ] lists statistics with highly positive and negative loadings . also refer to figure [ fig : pcload ] for categorized pc 2 loading coefficients .
players with positive pc 2 scores generate offense mainly through assists and driving shots .
these players tend to possess the ball often and either kick the ball to teammates for shot attempts or drive the ball to the basket .
many point guards fall into this category with examples like ricky rubio , tony parker , and chris paul .
players with negative pc 2 scores provide offense primarily through catch and shoot shots and are very efficient scorers , especially from behind the 3-point arc .
primary examples are klay thompson , kyle korver , and anthony morrow .
table [ tab : pc3top ] lists statistics with highly positive and negative loadings for pc 3 which explains 9% of the total variance .
also refer to figure [ fig : pcload ] for categorized pc 3 loading coefficients .
players with positive pc 3 scores are extremely quick on both sides of the ball and cover a lot of ground while on the court .
some examples are ish smith , shane larkin , and dennis schroder .
players with negative pc 3 scores are largely responsible for scoring when on the court and provide a significant amount of offensive production per 48 minutes .
scoring and rebounding efficiency characterize many of the superstars in the nba with players like kevin durant , carmelo anthony , and lebron james touting highly negative pc 3 scores .
table [ tab : pc4top ] lists statistics with highly positive and negative loadings for pc 4 which accounts for another 4% of the total variance .
also refer to figure [ fig : pcload ] for categorized loading coefficients for pc 4 .
this component is characterized by players tendencies when they receive possession of the ball .
players with positive pc 4 scores tend to pass or convert catch and shoot shots when the ball goes their way ( e.g. kevin love , spencer hawes , and patty mills ) while players with negative pc 4 scores tend to drive the ball and score efficiently when they get touches ( e.g. tyreke evans , rodney stuckey , and tony wroten ) . not only can we characterize players by principal components , but teams can also be profiled along these new dimensions as well .
there are numerous ways to aggregate the player pc scores to form a team - level score , but here a simple weighted average is used .
the @xmath0th team pc score , @xmath1 , can be found by taking an average of the @xmath0th pc scores across the @xmath2 players weighted by the minutes played throughout the season , @xmath3 .
figure [ fig : pcscoreteam ] shows the distribution of all nba teams across these dimensions as well as their corresponding 2013 - 2014 regular season winning percentage .
this view is useful in seeing the differences and similarities in team playing styles and how they impact success .
+ + for example , the new york knicks had an extremely negative pc 2 score .
further investigation shows it is partially the result of catch and shoot offense from j.r .
smith , andrea bargnani , and tim hardaway jr . who were all in the top 50 in catch and shoot points per 48 minutes
however , another major factor is that 8 of the 12 new york players were below average in passes per 48 minutes ( average was 58 passes per 48 minutes ) which is indicative of poor ball movement .
all - star carmelo anthony is in this group and has long been labeled a `` ball hog''@xcite which is supported by his below average passing and above average number of touches and scoring .
in fact , anthony s top 10 performance in points per 48 minutes , points per touch , and rebounding efficiency helped earn his team the most negative pc 3 team average score .
note that team average pc 3 score is negatively correlated with winning percentage , yet the knicks won only 37 games and failed to make the playoffs . to better understand how these team average pc scores impact winning , table [ tab : reg ] contains the results from a multiple linear regression analysis on winning percentage .
note that negative pc 3 scores are highly correlated with winning while positive pc 2 and pc 4 scores are also highly correlated with winning .
negative pc 3 scores are associated with high average scoring and rebounding efficiency .
referring back to tables [ tab : pc2top ] and [ tab : pc4top ] , passes , touches , and assists contribute positively to pc 2 and pc 4 scores .
regarding new york s lackluster season , it seems that anthony s great offensive production was not enough to offset the negative impact of extremely poor passing and ball movement . | the release of nba player tracking data greatly enhances the granularity and dimensionality of basketball statistics used to evaluate and compare player performance
. however , the high dimensionality of this new data source can be troublesome as it demands more computational resources and reduces the ability to easily analyze and interpret findings .
therefore , we must find a way to reduce the dimensionality of the data set while retaining the ability to differentiate and compare player performance .
in this paper , principal component analysis ( pca ) is used to identify four principal components that account for 68% of the variation in player tracking data from the 2013 - 2014 regular season and intuitive interpretations of these new dimensions are developed by examining the statistics that influence them the most . in this new high variance , low dimensional space , you can easily compare statistical profiles across any or all of the principal component dimensions to evaluate characteristics that make certain players and teams similar or unique .
a simple measure of similarity between two player or team statistical profiles based on the four principal component scores is also constructed .
the statistical diversity index ( sdi ) allows for quick and intuitive comparisons using the entirety of the player tracking data . as new statistics emerge , this framework is scalable as it can incorporate existing and new data sources by reconstructing the principal component dimensions and sdi for improved comparisons . using principal component scores and sdi ,
several use cases are presented for improved personnel management .
team principal component scores are used to quickly profile and evaluate team performances across the nba and specifically to understand how new york s lack of ball movement negatively impacted success despite high average scoring efficiency as a team .
sdi is used to identify players across the nba with the most similar statistical performances to specific players .
all - star tony parker and shooting specialist anthony morrow are used as two examples and presented with in - depth comparisons to similar players using principal component scores and player tracking statistics .
this approach can be used in salary negotiations , free agency acquisitions and trades , role player replacement , and more .
* keywords : * principal component analysis , nba player tracking data , statistical diversity index , dimension reduction , personnel management , national basketball association |
the @xmath1 meson first observed by the cdf collaboration at fnal @xcite is expected to be copiously produced in the future experiments at hadron colliders @xcite with facilities oriented to the study of fine effects in the heavy quark interactions such as the parameters of cp - violation and charged weak current mixing .
so , one could investigate the spectroscopy , production mechanism and decay features of @xmath1 @xcite with the incoming sample of several billion events .
in such circumstances , in addition to the current success in the experimental study of decays with the cp - violation in the gold - plated mode of neutral @xmath3-meson by the babar and belle collaborations @xcite allowing one to extract the ckm - matrix angle @xmath4 in the unitarity triangle , a possible challenge is whether one could get an opportunity to extract some information about the ckm unitarity triangle from the @xmath1 physics in a model independent way or not .
the theoretical principal answer is one can do it . indeed
, there is an intriguing opportunity to extract the angle @xmath0 in the model - independent way using the strategy of reference triangles @xcite in the decays of doubly heavy hadrons .
this ideology for the study of cp - violation in @xmath1 decays was originally offered by m.masetti @xcite , independently investigated by r.fleischer and d.wyler @xcite and extended to the case of doubly heavy baryons in @xcite .
let us point out necessary conditions to extract the cp - violation effects in the model - independent way .
1 . interference .
the measured quantities have to involve the amplitudes including both the cp - odd and cp - even phases
exclusive channels .
the hadronic final state has to be fixed in order to isolate a definite flavor contents and , hence , the definite matrix elements of ckm matrix , which can exclude the interference of two cp - odd phases with indefinite cp - even phases due to strong interactions at both levels of the quark structure and the interactions in the final state .
3 . oscillations .
the definite involvement of the cp - even phase is ensured by the oscillations taking place in the systems of neutral @xmath3 or @xmath5 mesons , wherein the cp - breaking effects can be systematically implemented .
4 . tagging .
once the oscillations are involved , the tagging of both the flavor and cp eigenstates is necessary for the complete procedure .
the gold - plated modes in the decays of neutral @xmath3 mesons involve the oscillations of mesons themselves and , hence , they require the time - dependent measurements .
in contrast , the decays of doubly heavy hadrons such as the @xmath1 meson and @xmath6 baryons with the neutral @xmath7 or @xmath8 meson in the final state do not require the time - dependent measurements .
the triangle ideology is based on the direct determination of absolute values for the set of four decays , at least : the decays of hadron in the tagged @xmath7 meson , the tagged @xmath8 meson , the tagged cp - even state can be used , too .
however , their registration requires the detection of cp - even state of @xmath9 , which can be complicated because of a detector construction , say , by a long base of @xmath9 decay beyond a tracking system .
] of @xmath7 , and the decay of the anti - hadron into the tagged cp - even state of @xmath7 . to illustrate ,
let us consider the decays of @xmath10 the corresponding diagrams with the decay of @xmath11-quark are shown in figs .
[ fig:1 ] and [ fig:1a ] .
we stress that two diagrams of the decay to @xmath7 have the additional negative sign caused by the pauli interference of two charmed quarks , which , however , completely compensated after the fierz transformation for the corresponding dirac matrices .
( 90,25 ) ( -10,0)=55 ( 33,0)=55 ( -2.3,6.5)@xmath12 ( 34.,6.5)@xmath13 ( 34.2,21)@xmath7 ( 47.2,6.55)@xmath12 ( 67.5,6.6)@xmath7 ( 67.5,18.1)@xmath13 ( 5.,11.5)@xmath14 ( 15.,2.)@xmath15 ( 26.,20.8)@xmath16 ( 27,15.)@xmath15 ( 29,11.5)@xmath17 ( 53.5,12)@xmath14 ( 63.5,11.5)@xmath16 ( 57.5,2)@xmath15 ( 63.2,20.2)@xmath17 ( 63,14.2)@xmath15 ( 90,25 ) ( -10,2)=55 ( 42,1)=55 ( -2.3,8.5)@xmath12 ( 34.,8.5)@xmath13 ( 34.2,23)@xmath8 ( 48.2,13.9)@xmath12 ( 87.5,4.2)@xmath13 ( 87.5,23.6)@xmath8 ( 5.,13.5)@xmath14 ( 15.,4.)@xmath15 ( 26.,22.8)@xmath15 ( 27,17.)@xmath16 ( 29,13.5)@xmath17 ( 56.5,16.7)@xmath14 ( 73.5,9.5)@xmath17 ( 56.5,12)@xmath15 ( 73.2,18.7)@xmath16 ( 83,16.8)@xmath15 ( 83,12.)@xmath15 the exclusive modes make the penguin terms to be excluded , since the penguins add an even number of charmed quarks , i.e. two or zero , while the final state contains two charmed quarks including one from the @xmath11 decay and one from the initial state . however , the diagram with the weak annihilation of two constituents , i.e. the charmed quark and beauty anti - quark in the @xmath18 meson , can contribute in the next order in @xmath19 as shown in fig .
[ fig:1a ] for the given final state .
nevertheless , we see that such the diagrams have the same weak - interaction structure as at the tree level .
therefore , they do not break the consideration under interest .
the magnitude of @xmath19-correction to the absolute values of corresponding decay widths is discussed in section [ sec:2 ] .
thus , the cp - odd phases of decays under consideration are determined by the tree - level diagrams shown in figs .
[ fig:1 ] and [ fig:1a ] .
therefore , we can write down the amplitudes in the following form : @xmath20 where @xmath21 denote the cp - even factors depending on the dynamics of strong interactions . using the definition of angle @xmath0 @xmath22,\ ] ] for the cp - conjugated channels , which corresponds to fixing the representation of the ckm matrix , e.g. by the wolfenstein form @xcite . ]
we find @xmath23 we see that the corresponding widths for the decays to the flavor tagged modes coincide with the cp - conjugated ones .
however , the story can be continued by using the definition of cp - eigenstates for the oscillating @xmath2 system mesons are irrelevant here , and we can neglect them in the sound way .
] , @xmath24 so that we straightforwardly get @xmath25 \sqrt{2}{\cal a}(b^-_c\to d_s^- d_1 ) \stackrel{\mbox{\tiny def}}{= } \sqrt{2 } { \cal a}_{d_1}^{\mbox{\sc cp } } & = & e^{-2{\rm i}\gamma } { \cal a}_{d}+{\cal a}_{\bar d}. \label{eq:6}\end{aligned}\ ] ] the complex numbers entering ( [ eq:5 ] ) and ( [ eq:6 ] ) establish two triangles with the definite angle @xmath26 between the vertex positions as shown in fig .
[ fig:3 ] .
( 80,50 ) ( 10,7)=75 ( 32,4)@xmath27 ( 68,22)@xmath28 ( 51.,14)@xmath29 ( 44.5,33)@xmath30 ( 7,23)@xmath31 thus , due to the unitarity , the measurement of four absolute values @xmath32
|{\cal a}_{d_1}| = |{\cal a}(b^+_c\to d_s^+ d_1)| , & \quad & |{\cal a}_{d_1}^{\mbox{\sc cp}}| = |{\cal a}(b^-_c\to d_s^- d_1)|,\end{aligned}\ ] ] can constructively reproduce the angle @xmath0 in the model - independent way .
the above triangle - ideology can be implemented for the analogous decays to the excited states of charmed mesons in the final state .
the residual theoretical challenge is to evaluate the characteristic widths or branching fractions .
we address this problem and analyze the color structure of amplitudes .
so , we find that the matrix elements under interest have the different magnitudes of color suppression , so that at the tree level we get @xmath33 and @xmath34 , while the ratio of relevant ckm - matrix elements , @xmath35 with respect to the small parameter of cabibbo angle , @xmath36 , which one can easily find in the wolfenstein parametrization @xmath37 v_{cd } & v_{cs } & v_{cb } \\[2 mm ] v_{td } & v_{ts } & v_{tb } \end{array}\right ) = \left(\begin{array}{ccc } 1-\frac{1}{2}\lambda^2 & \lambda & a \lambda^3 ( \rho-{\rm i}\eta ) \\[2 mm ] -\lambda & 1-\frac{1}{2}\lambda^2 & a \lambda^2 \\[2 mm ] a \lambda^3 ( 1-\rho-{\rm i}\eta ) & -a \lambda^2 & 1 \end{array}\right).\ ] ] nevertheless , the interference of two diagrams in the decays of @xmath18 to the @xmath7 meson is destructive , and the absolute values of the amplitudes @xmath28 and @xmath27 become close to each other .
thus , we expect that the sides of the reference - triangles are of the same order of magnitude , which makes the method to be an attractive way to extract the angle @xmath0 . in section [ sec:2 ]
we classify the diagrams for the decays of doubly heavy meson @xmath18 by the color and weak - interaction structures .
section [ sec:3 ] is devoted to the numerical estimates in the framework of qcd sum rules .
the results are summarized in conclusion .
in the framework of @xmath38-expansion we have got the following scaling rules of color structures in the processes with the hadrons composed of the quark and anti - quark : 1 .
the meson wavefunction @xmath39 2 . the coupling constant @xmath40 3 .
the casimir operators @xmath41 4 .
the fierz relation for the generators of su(@xmath42 ) group in the fundamental representation @xmath43 next , the non - leptonic weak lagrangian has a form typically given by the following term @xcite : @xmath44 where @xmath45 , and the wilson coefficients @xmath46 in the @xmath38-expansion .
then , we can proceed with the analysis of decays under interest .
two diagrams shown in fig .
[ fig:1 ] scale with the different order in @xmath38 , so that @xmath47 more definitely we get the color factors @xmath48 where @xmath49,\\[1 mm ] a_2 & = & \frac{1}{2n_c}\,\left[c_+(n_c+1)-c_-(n_c-1)\right].\end{aligned}\ ] ] nevertheless , we have to calculate both diagrams in order to take into account their interference which is not suppressed by any kinematical factors except the color factor of @xmath38 .
two diagrams shown in fig .
[ fig:1a ] also scale with the different order in @xmath38 , so that @xmath50 more definitely we get the color factors @xmath51 where we have explicitly included the @xmath19 correction .
however , we can easily find that due to the virtualities of quarks and gluons the kinematical suppression of second diagram in fig .
[ fig:1a ] is given by the factor of @xmath52 where @xmath53 is the characteristic value of wavefunction , the gluon virtuality is determined by the expression @xmath54 with @xmath55 being the masses of the mesons in the initial and final states , respectively , and the virtuality of quark line connected to the virtual gluon is of the order of @xmath56 therefore , @xmath57 therefore , the above @xmath19-corrections can be neglected to the leading order in the @xmath58-expansion .
finally , in this section we have analyzed the color and weak - interaction structures of decay amplitudes and isolate those of the largest magnitude , while the numerical estimates are presented in the next section .
in this section we use the machinery of qcd sum rules @xcite in order to calculate the widths and branching ratios for the gold - plated modes under study .
the calculations of exclusive non - leptonic modes usually involves the approximation of factorization @xcite , which , as expected , can be quite accurate for the @xmath1 , since the quark - gluon sea is suppressed in the heavy quarkonium .
thus , the important parameters are the factors @xmath59 and @xmath60 in the non - leptonic weak lagrangian , which depend on the normalization point suitable for the @xmath1 decays . in this way , we , first , calculate the form factors in the semileptonic transitions @xcite and , second , evaluate the non - leptonic matrix elements in the factorization approach .
we accept the following convention on the normalization of wave functions for the hadron states under study , i.e. for the pseudoscalar ( @xmath61 ) and vector ( @xmath62 ) states : @xmath63 & & \langle 0|{\cal j}_\mu|v\rangle = \epsilon_\mu\ , f_v\,m_v,\end{aligned}\ ] ] where @xmath64 denote the leptonic constants , so that they are positive , @xmath65 @xmath66 is a four - momentum of the hadron , @xmath67 is a polarization vector of @xmath62 , @xmath68 is its mass , and the current is composed of the valence quark fields constituting the hadron @xmath69 in this respect we can easily apply the ordinary feynman rules for the calculations of diagrams , so that the quark - meson vertices in the decay channel are chosen with the following spin structures : @xmath70 then , we get general expressions for the hadronic matrix elements of weak currents in the exclusive decays of @xmath71 and @xmath72 with the definitions of form factors given by the formulae @xmath73 where @xmath74 and @xmath75 . the form factors @xmath76 are dimensionless , while @xmath77 and @xmath78 has a dimension of inverse energy , @xmath79 is of the energy dimension . in the case of nonrelativistic description for both initial and final meson states
we expect that @xmath80 f^a_0 > 0 , & f^a_+ < 0 , & f^a_- > 0 . \end{array}\ ] ] it is important to note that for the pseudoscalar state the hermitian conjugation in ( [ pseudo ] ) does not lead to the change of sign in the right hand side of equation because of the prescription accepted , since the conjugation of imaginary unit takes place with the change of sign for the momentum of meson ( the transition from the out - state to in - one ) .
the same speculations show that the spin structure of matrix element in the quark - loop order does not involve a functional dependence of form factors on the transfer momentum squared except of @xmath79 , so that we expect that the simplest modelling in the form of the pole dependence can be essentially broken for @xmath79 , while the other form factors are fitted by the pole model in a reasonable way .
following the standard procedure for the evaluation of form factors in the framework of qcd sum rules @xcite , in the @xmath1 decays we consider the three - point functions @xmath81 where @xmath82 and @xmath83 denote the interpolating currents for the final states mesons .
the standard procedure for the evaluation of correlators and form factors is described in @xcite .
the most important notes are the following @xcite : * for the heavy quarkonium @xmath84 , where the relative velocity of quark movement is small , an essential role is taken by the coulomb - like @xmath85-corrections . *
we have found that the normalizations of leptonic constants for the heavy quarkonia are fixed by the appropriate choice of effective constant for the coulomb exchange @xmath86 , while the stability is very sensitive to the prescribed value of heavy quark mass .
thus , these parameters of sum rules are extracted from the two - point qcd sum rules with a quite good accuracy .
* in the framework of two - point sum rules for the heavy - light channel the fixed value of threshold energy @xmath87 determines the binding energy of heavy quark in the meson , @xmath88 gev , which yields the same value of mass for the beauty quark , @xmath89 gev , as it was determined from the analysis of two - point sum rules in the @xmath90 channel .
however , taking into account the second order corrections in @xmath91 , we find that the mass of charmed quark is shifted to the value of @xmath92 gev in the heavy - light channel in comparison with the @xmath93 states .
thus , in the transition of @xmath1 to the charmed meson we put the mass of spectator charmed quark equal to @xmath92 gev .
* in the framework of the effective theory of heavy quarks with the expansion in the inverse masses of the heavy quarks @xcite , the spin symmetry relations between the form factors in the soft limit of zero recoil momentum can be derived @xcite .
* we take the following ordinary ratios of leptonic constants for the vector and pseudoscalar states and for the heavy - strange mesons : @xmath94 which agree with both the lattice computations @xcite and the estimates in the framework of potential models taking into account relativistic corrections @xcite . *
the leptonic constant of @xmath1 is taken from a scaling relation for the heavy quarkonia @xcite .
[ cols="^,^,^,^,^,^,^",options="header " , ] in the above estimates we put the following values of parameters : * the leptonic constants : @xmath95 * the ckm elements : @xmath96 so that the numbers can be appropriately scaled at other values of input parameters . in the btev @xcite and lhcb @xcite experiments
one expects the @xmath1 production at the level of several billion events .
therefore , we predict @xmath97 decays of @xmath1 in the gold - plated modes under interest .
the experimental challenge is the efficiency of detection .
one usually get a 10%-efficiency for the observation of distinct secondary vertices outstanding from the primary vertex of beam interaction .
next , we have to take into account the branching ratios of @xmath98 and @xmath7 mesons .
this efficiency crucially depends on whether we can detect the neutral kaons and pions or not .
so , for the @xmath98 meson the corresponding branching ratios grow from 4% ( no neutral @xmath99 and @xmath100 ) to 25% .
the same interval for the neutral @xmath7 is from 11 to 31% .
the detection of neutral kaon is necessary for the measurement of decay modes into the cp - odd state @xmath101 of the neutral @xmath7 meson , however , we can omit this cross - check channel from the analysis dealing with the cp - even state of @xmath102 .
the corresponding intervals of branching ratios reachable by the experiment are from 0.5 to 1.3% for the cp - even state and from 1.5 to 3.8% for the cp - odd state of @xmath7 .
the pessimistic estimate for the product of branching ratios is about @xmath103 , which results in @xmath104 reconstructed events .
thus , an acceptance of experimental facility and an opportunity to detect neutral pions and kaons as well as reliable estimates of total cross section for the @xmath1 production in hadronic collisions are of importance in order to make expectations more accurate .
in this work we have shown how the reference - triangle ideology can be used for the model - independent extraction of ckm - matrix angle @xmath0 from the set of branching ratios of doubly heavy meson @xmath1 exclusively decaying to the neutral @xmath5 mesons .
tagging the flavor and cp - eigenstates of such the @xmath5 mesons allows one to avoid the uncertainties caused by the qcd dynamics of quarks .
we have estimated the characteristic branching ratios in the framework of qcd sum rules , which yields the values of the order of @xmath105 \approx 5\cdot 10^{-6}.\ ] ] accepting the above value , and putting the efficiency of tagging procedure equal to 0.5% for the neutral charmed meson and 4% for the charmed strange meson in the final state as well as the vertex reconstruction efficiency equal to 10% , we can expect the observation of about 10 reconstructed events per year at the lhc collider in such the experiment like lhcb or in btev experiment at fnal .
the author thanks prof .
r.dzhelyadin for the possibility to visit cern in collaboration with the lhcb group , to which members i express a gratitude for a hospitality .
i am grateful to profs .
a.k.likhoded and m.a.ivanov for clarifications and a cooperation , to prof .
v.f.obraztsov , yu.gouz , o.yushchenko and v.romanovsky for useful discussions and valuable remarks . *
* f. abe et al . , cdf collaboration , phys .
lett . * 81 * , 2432 ( 1998 ) , phys . rev . *
d58 * , 112004 ( 1998 ) .
s.s.gershtein et al . , sov . j. nucl .
bf 48 , 327 ( 1988 ) , yad .
fiz . * 48 * , 515 ( 1988 ) ; + s. s. gershtein , v. v. kiselev , a. k. likhoded and a. v. tkabladze , phys .
usp . * 38 * , 1 ( 1995 ) [ usp .
nauk * 165 * , 3 ( 1995 ) ] [ arxiv : hep - ph/9504319 ] ; + e.eichten , c.quigg , phys .
rev . * d49 * , 5845 ( 1994 ) ; + s.s.gershtein et al . ,
phys . rev . * d51 * , 3613 ( 1995 ) ; + a. v. berezhnoi , v. v. kiselev , a. k. likhoded and a. i. onishchenko , phys . atom .
nucl . * 60 * , 1729 ( 1997 ) [ yad .
* 60n10 * , ( 1997 ) ] [ arxiv : hep - ph/9703341 ] .
k. anikeev _ et al .
_ , arxiv : hep - ph/0201071 .
j.d.bjorken , draft report 07/22/86 ( 1986 ) [ unpublished ] ; + m.lusignoli , m.masetti , z. phys .
* c51 * , 549 ( 1991 ) ; + v.v.kiselev , a.k.likhoded , a.v.tkabladze , phys .
nucl . * 56 * , 643 ( 1993 ) , yad .
fiz . * 56 * , 128 ( 1993 ) ; + v.v.kiselev , a.v.tkabladze , yad . fiz . * 48 * , 536 ( 1988 ) .
i.bigi , phys . lett . *
b371 * , 105 ( 1996 ) ; + m.beneke , g.buchalla , phys . rev
. * d53 * , 4991 ( 1996 ) ; + a.i.onishchenko , [ hep - ph/9912424 ] ; + ch .- h.chang , sh .- l.chen , t .- f.feng , x .- q.li , commun .
. phys . * 35 * , 51 ( 2001 ) , phys . rev . *
d64 * , 014003 ( 2001 ) .
p.colangelo , g.nardulli , n.paver , z.phys .
* c57 * , 43 ( 1993 ) ; + e.bagan et al .
, z. phys .
* c64 * , 57 ( 1994 ) . k. abe _ et al . _ [ belle collaboration ] , phys . rev .
d * 66 * , 071102 ( 2002 ) [ arxiv : hep - ex/0208025 ] ; + b. aubert _ et al .
_ [ babar collaboration ] , phys .
lett . * 89 * , 201802 ( 2002 ) [ arxiv : hep - ex/0207042 ] . m. gronau and d. wyler , phys .
b * 265 * , 172 ( 1991 ) .
m. masetti , phys .
b * 286 * , 160 ( 1992 ) .
r. fleischer and d. wyler , phys .
d * 62 * , 057503 ( 2000 ) [ arxiv : hep - ph/0004010 ] . v. v.
kiselev and a. k. likhoded , usp . fiz .
nauk * 172 * , 497 ( 2002 ) [ arxiv : hep - ph/0103169 ] ; + a. v. berezhnoi , v. v. kiselev and a. k. likhoded , phys .
* 59 * , 870 ( 1996 ) [ yad . fiz . *
59 * , 909 ( 1996 ) ] [ arxiv : hep - ph/9507242 ] ; + s. p. baranov , phys .
d * 54 * , 3228 ( 1996 ) ; + m. a. doncheski , j. steegborn and m. l. stong , phys .
d * 53 * , 1247 ( 1996 ) [ arxiv : hep - ph/9507220 ] ; + a. v. berezhnoi , v. v. kiselev , a. k. likhoded and a. i. onishchenko , phys .
d * 57 * , 4385 ( 1998 ) [ arxiv : hep - ph/9710339 ] ; + v. v. kiselev and a. e. kovalsky , phys . atom .
nucl . * 63 * , 1640 ( 2000 ) [ yad . fiz .
* 63 * , 1728 ( 2000 ) ] [ arxiv : hep - ph/9908321 ] ; + v. v. braguta , v. v. kiselev and a. e. chalov , phys .
atom . nucl .
* 65 * , 1537 ( 2002 ) [ yad .
* 65 * , 1575 ( 2002 ) ] .
v. v. kiselev and o. p. yushchenko , arxiv : hep - ph/0211382 . l. wolfenstein , phys .
* 51 * , 1945 ( 1983 ) .
g. buchalla , a. j. buras and m. e. lautenbacher , rev .
* 68 * , 1125 ( 1996 ) [ arxiv : hep - ph/9512380 ] .
m. a. shifman , a. i. vainshtein and v. i. zakharov , nucl .
b * 147 * , 448 ( 1979 ) , nucl .
b * 147 * , 385 ( 1979 ) , nucl .
b * 147 * , 519 ( 1979 ) ; + v. a. novikov , l. b. okun , m. a. shifman , a. i. vainshtein , m. b. voloshin and v. i. zakharov , phys .
rept . * 41 * , 1 ( 1978 ) ; + l. j. reinders , h. rubinstein and s. yazaki , phys .
* 127 * , 1 ( 1985 ) .
p. ball , m. beneke and v. m. braun , nucl .
b * 452 * , 563 ( 1995 ) [ arxiv : hep - ph/9502300 ] , phys .
d * 52 * , 3929 ( 1995 ) [ arxiv : hep - ph/9503492 ] ; + p. ball , v. m. braun and h. g. dosch , phys .
d * 48 * , 2110 ( 1993 ) [ arxiv : hep - ph/9211244 ] ; + a. a. ovchinnikov and v. a. slobodenyuk , sov .
j. nucl .
phys . * 50 * , 891 ( 1989 ) [ yad .
* 50 * , 1433 ( 1989 ) ] , z. phys .
c * 44 * , 433 ( 1989 ) .
m.dugan and b.grinstein , phys . lett . *
b255 * , 583 ( 1991 ) ; + m.a.shifman , nucl . phys . *
b388 * , 346 ( 1992 ) ; + b.blok , m.shifman , nucl .
b389 * , 534 ( 1993 ) .
v.v.kiselev , a.v.tkabladze , phys . rev .
* d48 * , 5208 ( 1993 ) .
v.v.kiselev , a.k.likhoded , a.i.onishchenko , nucl .
b569 * , 473 ( 2000 ) .
v.v.kiselev , a.k.likhoded , a.e.kovalsky , nucl . phys . *
b585 * , 353 ( 2000 ) , hep - ph/0006104 ( 2000 ) ; arxiv : hep - ph/0006104 . v. v. kiselev , arxiv : hep - ph/0211021 . m. neubert , phys .
rept . * 245 * , 259 ( 1994 ) .
g.t.bodwin , e.braaten , g.p.lepage , phys . rev .
* d51 * , 1125 ( 1995 ) [ erratum - ibid . * d55 * , 5853 ( 1995 ) ] ; + t.mannel , g.a.schuler , z. phys . *
c67 * , 159 ( 1995 ) .
a. pineda and j. soto , nucl .
* 64 * , 428 ( 1998 ) [ arxiv : hep - ph/9707481 ] .
m. e. luke , a. v. manohar and i. z. rothstein , phys .
d * 61 * , 074025 ( 2000 ) [ arxiv : hep - ph/9910209 ] ; + a. v. manohar and i. w. stewart , phys .
d * 63 * , 054004 ( 2001 ) [ arxiv : hep - ph/0003107 ] .
a. h. hoang , a. v. manohar and i. w. stewart , phys .
d * 64 * , 014033 ( 2001 ) [ arxiv : hep - ph/0102257 ] . c. bernard
_ et al . _
[ milc collaboration ] , dynamical quarks , arxiv : hep - lat/0206016 .
d. ebert , r. n. faustov and v. o. galkin , mod .
a * 17 * , 803 ( 2002 ) [ arxiv : hep - ph/0204167 ] .
v. v. kiselev , phys . part .
* 31 * , 538 ( 2000 ) [ fiz .
chast . atom .
yadra * 31 * , 1080 ( 2000 ) ] ; + v. v. kiselev , int .
j. mod .
* a11 * , 3689 ( 1996 ) ; + v. v. kiselev , nucl .
* b406 * , 340 ( 1993 ) . v. v. kiselev , mod .
lett . a * 10 * , 1049 ( 1995 ) [ arxiv : hep - ph/9409348 ] ; + v. v. kiselev , int . j. mod .
a*9 * , 4987 ( 1994 ) .
a. abd el - hady , j. h. munoz and j. p. vary , phys .
d * 62 * , 014019 ( 2000 ) [ arxiv : hep - ph/9909406 ] . c. h. chang and y. q. chen , phys .
d * 49 * , 3399 ( 1994 ) .
a.yu.anisimov , i.m.narodetskii , c.semay , b.silvestrebrac , phys .
* b452 * , 129 ( 1999 ) ; + a.yu.anisimov , p.yu.kulikov , i.m.narodetsky , k.a.ter-martirosian , phys .
62 * , 1739 ( 1999 ) , yad . fiz . * 62 * , 1868 ( 1999 ) . m. a. ivanov , j. g. korner and o. n. pakhomova , phys . lett .
b * 555 * , 189 ( 2003 ) [ arxiv : hep - ph/0212291 ] .
p.colangelo , f.de fazio , phys . rev . *
d61 * , 034012 ( 2000 ) .
i. p. gouz , v. v. kiselev , a. k. likhoded , v. i. romanovsky and o. p. yushchenko , arxiv : hep - ph/0211432 . | the model - independent method based on the triangle ideology is implemented to extract the ckm - matrix angle @xmath0 in the decays of doubly heavy long - lived meson @xmath1 .
we analyze a color structure of diagrams and conditions to reconstruct two reference - triangles by tagging the flavor and cp eigenstates of @xmath2 mesons in the fixed exclusive channels .
the characteristic branching ratios are evaluated in the framework of qcd sum rules . |
periodic potentials , usually called optical lattices ( ol ) , which are induced by interference of laser beams illuminating a bose - einstein condensate ( bec ) , are widely used in experimental studies of becs .
not only one - dimensional ( 1d ) ols , but also their multidimensional counterparts are available in the experiment , see refs . @xcite and references therein .
an important type of nonlinear excitations that becs may support are bright solitons ( in this paper , following the trend commonly adopted in the current literature , we apply the term soliton " to robust localized objects , without implying integrability of the underlying mathematical model ) . in an effectively 1d ( cigar - shaped )
condensate , solitons have been observed in the experiment @xcite .
creation of solitons in a bec confined in the ol has not been reported yet , although splitting of wave packets in a @xmath2rb condensate , loaded in a cylindrical trap with an superimposed 1d ol potential , which was reported recently @xcite , might be a manifestation of transient soliton dynamics .
parallel to the experiments , many theoretical predictions for bec solitons in ols have been reported .
in particular , attention was attracted to a possibility to create _ gap solitons _
, that are expected as a result of the interplay of the ol potential and _ repulsive _ interaction between atoms in the bec @xcite . while only 1d solitons have been thus far observed in becs , it was predicted that ols may support multidimensional solitons of the gap type as well @xcite .
a prediction which is especially relevant to feasible experiments is that ols can also stabilize multidimensional solitons in becs with attraction between atoms ( because of the possibility of collapse , such solitons are unstable without a support potential ) @xcite .
moreover , 2d bright solitons carrying intrinsic vorticity can be stabilized too by the ol in the 2d case @xcite .
various other properties of ol - supported solitons were studied too @xcite .
similar localized objects are expected in a different physical setting , in the form of 2d optical spatial solitons in nonlinear photonic crystals ( npcs ) @xcite . in optical media ,
an effective lattice potential is induced by periodic modulation of the local refractive index in the transverse direction(s ) @xcite .
an alternative is to use a _ virtual lattice _
, induced in a photorefractive medium by a transverse array of strong coherent laser beams ; the latter arrangement has made it possible to observe 2d spatial solitons in a real experiment @xcite , as well as their vorticity - carrying counterparts @xcite .
most theoretical studies of solitons in the above - mentioned models did not address their motion .
in a recent work @xcite , we considered in detail the motion and collisions of 1d solitons in the ol - equipped model in both cases of the attractive and repulsive cubic nonlinearity . in that work ,
the ol potential was combined with an external parabolic ( harmonic ) trapping potential , which corresponds to the real situation in any bec experiment .
direct simulations of the corresponding effective 1d gross - pitaevskii ( gp ) equation had demonstrated that , in either case ( self - focusing or self - defocusing nonlinearity ) , the soliton travels freely through the lattice if its norm ( number of atoms ) is below a certain critical value ; above this value , it gets trapped by the ol .
depending on parameters , collisions between moving solitons are either elastic , or lead to their fusion .
the most intriguing result concerns the motion of the soliton in the parabolic trap : in the case of attraction , the soliton behaves , as it may be naturally expected , as a quasi - particle in the harmonic - oscillator potential .
however , in the case of repulsion , the soliton of the gap type is expelled by the parabolic trap , while it is stably confined by an _ anti - trap _ ( repulsive parabolic potential ) .
this result , which calls for direct experimental verification , was supported by analytical considerations which demonstrate that the effective mass of the gap soliton , regarded as a quasi - particle , is _ negative _ ( in the case of the usual soliton , corresponding to the attractive nonlinearity , the effective mass is positive ) .
an issue of straightforward interest is to extend these results to the 2d case , which is quite relevant , as experiments with 2d solitons in becs seem possible right now . besides that , the problem is also of direct relevance for steering spatial optical soliton in npc media . in this work ,
we focus on the 2d bec model with repulsive interactions , as this case is most relevant to the experiment , and produces most interesting physical results .
first , we demonstrate that broad localized states , which we call loosely bound ( lb ) ones , may be represented as a product of a linear bloch function , defined by the ol , which is taken close to the edge to the corresponding brillouin zone ( bz ) , and a slowly varying envelope function , which obeys an asymptotic nonlinear equation that does not contain ol .
such a representation makes it possible to demonstrate analytically that the effective mass of the lb state is negative . this way , we find not only fundamental lb states , but also ones carrying vorticity , whose mass is negative too ; the existence of 2d localized vortices in the repulsive model has not been reported before . in direct simulations ,
the lb vortex features actual rotation around its center .
simulations confirm the stability of the localized lb states , and demonstrate that they can move freely through the ol . in particular , both fundamental and vortex lb states can be readily set in elliptic circular motion in the anti - trapping potential , the frequency of the circulations being very accurately predicted by the asymptotic equation .
direct simulations ( as well as the asymptotic equation ) demonstrate an abrupt transition ( without hysteresis ) from the lb states to much stronger localized tight - bound ( tb ) ones with the increase of the norm ( number of atoms ) .
the tb states are nothing else but the 2d gap solitons , that were first found in ref .
the lb - tb transition also takes place for the vortices , thus revealing the existence of new tb vortex solitons of the gap type .
it is additionally found that the tb vortices undergo an intrinsic transition to a more symmetric shape with subsequent increase of the norm . unlike their lb counterparts
, the tb solitons can not move , being strongly pinned by the ol .
we also construct lb states , starting from the linear bloch function close to the bz center . in that case , the effective mass is positive , no transition to tb solitons occurs ( as they do not exist ) , and the lb states , both fundamental and vortex ones , are mobile , which is accurately described by the corresponding asymptotic equation . the rest of the paper is organized as follows .
the analysis based on the derivation of the asymptotic equations for the lb states , with both negative and positive mass , is presented in section 2 .
numerical results for static solitons ( however , including rotation of the vortex lb states ) , such as the lb - tb transitions , are summarized in section 3 . in section 4
, we present numerical and analytical results concerning motion of the lb states , for the cases of the negative and positive mass .
the paper is concluded by section 5 .
the mean - field description of the bec dynamics is based on the gp equation for the mean - field wave function @xmath3 in three dimensions @xcite , @xmath4 \psi , \label{gp3d}\ ] ] where @xmath5 is the atomic mass , @xmath6 , @xmath7 is the @xmath8 -wave scattering length , @xmath9 is the potential applied to the condensate , and the number of atoms is @xmath10 ) can be reduced to a 2d equation . after appropriate rescalings , such as @xmath11 and @xmath12 , where @xmath13 is the period of the spatially periodic potential , it takes the form @xmath14 where @xmath15 is the renormalized wave function , the sign @xmath16 denoting the rescaled coordinates @xmath17 is omitted , the kinetic - energy operator @xmath18 acts on the coordinates @xmath19 and @xmath20 , the repulsive interaction between atoms is assumed , the ol period is normalized to be @xmath21 in both directions ( we consider the simplest case of the square lattice , although more sophisticated patterns , such as triangular , hexagonal , and quasi - periodic ones , may be interesting too ) , while @xmath22 and @xmath23 are amplitudes of the , respectively , ol and parabolic potential ( in simulations , we will fix @xmath24 and @xmath25 , or @xmath26 and @xmath27 , negative @xmath23 implying the anti - trapping potential )
. we will show the numerical results with the dimensionless equation ( 2 ) .
the unit length corresponds to the spatial period @xmath13 of the optical lattice , and the unit time corresponds to @xmath28s for @xmath29 m and @xmath30 kg . if @xmath31 and the nonlinear term is neglected , eq .
( [ gp ] ) becomes the ordinary 2d linear schrdinger equation with the periodic potential , which has bloch - state solutions .
the bloch function @xmath32 obeys the equation @xmath33 f , \label{f}\ ] ] where @xmath34 is the chemical potential ( or simply energy , in the case of a single atom ) , and the solution is quasi - periodic , i.e. , @xmath35 , where @xmath36 are two wavenumbers , and the function @xmath37 is periodic , such that @xmath38 .
the bloch - wave solution determines the energy - band structure , @xmath39 , the point @xmath40 being the edge of the bz ( brillouin zone ) . following the lines of the analytical approach to the 1d case developed in ref .
@xcite , an approximate solution to the full time - dependent gp equation ( [ gp ] ) for a small - amplitude broad localized state ( soliton " ) may be sought for , in the lowest approximation , as @xmath41 where @xmath32 is the bloch function defined above , and @xmath42 is an envelope function which varies slowly ( as a function of @xmath19 and @xmath20 ) in comparison with @xmath32 .
this ansatz actually assumes that the nonlinearity and parabolic potential in eq .
( [ gp ] ) are small perturbations in comparison with the ol terms .
the most typical gap soliton corresponds to a solution which approaches the linear bloch function with @xmath43 when its norm ( the number of atoms ) , @xmath44 , tends to zero ( cf .
a similar situation in the 1d case @xcite ) . in the general case
, the slowly varying amplitude @xmath45 obeys an asymptotic nls equation , that can be derived by substituting the ansatz ( [ phi ] ) into eq .
( [ gp ] ) and applying a known averaging procedure @xcite : @xmath46 here , the effective mass @xmath47 for linear excitations is determined from the curvature of the energy - band structure of the linear bloch states , @xmath48 , and @xmath49 the energy functional corresponding to the simplified gp equation ( [ asympt ] ) is @xmath50 .
\label{e}\ ] ] stationary solutions to the asymptotic equation ( [ asympt ] ) can be looked for in an obvious form , @xmath51 , \label{phi}\ ] ] where @xmath52 and @xmath53 are polar coordinates in the plane @xmath54 , @xmath55 is a possible vorticity of the solution , and the amplitude @xmath56 obeys an equation @xmath57 f \label{eigen}\ ] ] [ in the general case in particular , in the case of the full gp equation ( [ gp ] ) which is not isotropic the vorticity of a stationary solution may be defined as @xmath58 , where @xmath59 is the total change of the solution s phase along a closed path around the center ] . the linear limit of eq .
( [ eigen ] ) is the linear schrdinger equation for the 2d harmonic oscillator . for fixed integer @xmath5 ,
the ground - state solution to the latter equation is ( in the present notation ) @xmath60 where @xmath61 , with @xmath62 , and @xmath63 .
near the bz edge , i.e. , close to the point @xmath64 , the linear bloch function can be approximated by a combination of four harmonics : @xmath65 @xmath66 the substitution of this into eq .
( [ f ] ) and a usual truncation procedure ( neglecting higher - order spatial harmonics ) lead to an expression @xmath67 /(2m_{\mathrm{eff}}^{\mathrm{(edge)}})$ ] , where @xmath68 @xmath69 and @xmath70 [ recall @xmath71 is defined by eq .
( [ g ] ) ] ; the signs @xmath72 in eq .
( [ e0edge ] ) pertain to two bloch - wave solutions which differ by a phase shift relative to the periodic ol potential in eq .
( [ f ] ) . as is seen , the effective mass in eq .
( [ eff(k = pi ) ] ) is negative for the linear excitations near the bz edge ( unless @xmath73 is very large , in which case the above approximation does not apply ) . equation ( [ asympt ] ) with @xmath74 , @xmath75 , and @xmath76 ( anti - trap ) is tantamount to the usual 2d gp equation with positive mass , negative @xmath71 ( which corresponds to self - attraction ) , and @xmath77 ( usual trap , rather than anti - trap ) . as is well known @xcite , the gp equation in the latter form has localized solutions , unless the norm @xmath78 is too large ( otherwise , collapse will take place ) .
these solutions , and their comparison with numerically found solutions to the full underlying equation ( [ gp ] ) , will be displayed below .
an approximation similar to that based on eq .
( [ bloch ] ) can also be developed close to the bz center . at the bz center
, the linear bloch function may be approximated by @xmath79[1+c_{2}\cos ( 2\pi y ) ] , \label{center}\ ] ] and the eventual expressions for the effective mass and @xmath80 are @xmath81 @xmath82 other coefficients are found to be @xmath83 and @xmath84 . in this case , the effective mass is always positive [ in particular , in the case with @xmath85 and @xmath86 , for which this approximation will be compared to numerical results below , @xmath87 .
obviously , eq . ( [ eigen ] ) with @xmath88 , @xmath75 , and @xmath77 has a localized solution ( actually , of the thomas - fermi type , see below ) for any vorticity @xmath5 and norm @xmath78 .
our first objective is to find lb ( loosely bound ) localized states near the bz edge , which can be constructed using the asymptotic gp equation ( [ asympt ] ) and the relations ( [ phi ] ) and ( [ bloch ] ) , and compare them with the corresponding numerical solutions to the full equation ( [ gp ] ) .
the 2d localized solutions to the asymptotic gp equation ( [ asympt ] ) were found numerically @xcite , and in an approximate form they can be also solved by dint of a variational method @xcite . in particular , the radial equation ( [ eigen ] ) can be solved numerically by the shooting method , which simultaneously makes it possible to find the nonlinear eigenvalue @xmath34 .
figure 1 displays numerical solutions to eq .
( [ eigen ] ) for @xmath89 and @xmath90 , obtained by means of the shooting method .
the norm of the corresponding solutions for the underlying wave function @xmath15 is , according to eq .
( [ phi ] ) , @xmath91 , where @xmath92 is the norm of @xmath45 [ see eq .
( [ phi ] ) ] .
in particular , @xmath93 for @xmath94 , @xmath95 for @xmath96 , and @xmath97 for @xmath98 . ) with @xmath99 and @xmath100 which represent 2d solitons with vorticity @xmath89 and @xmath90.,width=288 ] to characterize families of the solutions , figs .
2(a ) and 2(b ) display , respectively , the absolute value of the energy - per - atom , @xmath101 for the solution [ the energy @xmath102 is given by the functional ( 7 ) ] , and the maximum value of @xmath56 vs. the norm @xmath78 .
in particular , @xmath103 decreases with @xmath78 , because a larger atom density gives rise a larger positive contribution to @xmath102 through the term @xmath104 in the energy density . the maximum value of @xmath56 increases with @xmath78 , since the effective interaction is attractive owing to the negativeness of @xmath47 . ,
( a ) and amplitude ( b ) of the 2d solitons with the vorticity @xmath105 and @xmath90 , as found from eq .
( [ eigen ] ) , vs. the norm.,width=384 ] figures 3(a ) , ( b ) and ( c ) present comparison of the above semi - analytical solutions ( for @xmath89 and @xmath90 ) , based on eqs .
( [ phi ] ) , ( [ bloch ] ) and ( [ asympt ] ) , with direct numerical solutions of the full 2d equation ( [ gp ] ) .
the comparison is performed by displaying the semi - analytical and direct numerical solutions for @xmath106 , taken along the central cross - section , at @xmath107 ( the norms of the numerical and semi - analytical solutions are virtually identical in each case ) .
fairly good agreement is seen for @xmath89 and @xmath90 , although latexmath:[$|\phi for the loosely bound 2d states near the edge of the brillouin zone with ( a ) @xmath94 , ( b ) @xmath96 , and ( c ) @xmath98 , which corresponds to eq . (
[ bloch ] ) .
solid and dashed curves depict , respectively , the solutions obtained from the asymptotic equation ( [ asympt ] ) and full gp equation ( [ gp]).,width=480 ] from the viewpoint of the reduced gp equation ( [ asympt ] ) , which does not explicitly contain the ol potential , the angular velocity @xmath109 , which appears in the expression ( [ phi ] ) , is only a phase velocity .
however , the relation ( [ phi ] ) suggests that the corresponding solution to the full gp equation ( [ gp ] ) may feature _ actual rotation _ of the pattern with the angular velocity @xmath110 .
direct simulations confirm this expectation , as shown in fig .
the figure displays three consecutive snapshots of the region in which the vortex solution with @xmath98 shown in fig .
3(c ) takes sufficiently large values , namely , satisfying conditions @xmath111 \mathrm{re}\phi ( x , y)>0 $ ] and @xmath112 .
the time interval between the snapshots is @xmath21 .
this figure clearly shows that the vortex state indeed rotates with the frequency @xmath113 in the ol potential .
this figure also demonstrates that the vortex state with @xmath98 is _
stable_. the vortex solutions , with @xmath96 and @xmath90 , to the asymptotic equation ( [ asympt ] ) are known to exist @xcite and be dynamically stable @xcite if @xmath78 is smaller than a critical value . we have also checked the stability of the vortex states with @xmath96 and @xmath90 , shown in fig . 3 , by direct numerical simulations of eq .
( 2 ) , starting from initial conditions in the form of perturbed vortex states .
when the norm @xmath78 is increased , more complicated behaviors are expected .
some simulation results such as a transition to a tightly bound state and a symmetry - breaking process are shown in the next section . and
@xmath98.,width=480 ] solutions predicted by the asymptotic equation ( [ asympt ] ) near the bz center , i.e. , corresponding to eqs .
( [ center ] ) and ( [ eff(k=0 ) ] ) , were also found , and compared to direct numerical solutions of the underlying gp equation ( [ gp ] ) .
it should be stressed that , as the effective mass ( [ eff(k=0 ) ] ) is positive in this case , the corresponding localized solutions to the asymptotic equation ( [ asympt ] ) with @xmath75 ( repulsive interaction ) are not soliton - like ones , but are instead close to the thomas - fermi ( tf ) states , that can be obtained neglecting the kinetic - energy term in the gp equation @xcite .
figure [ fig5 ] shows typical examples of thus found solutions for @xmath94 , @xmath96 , and @xmath98 . as is seen , the accuracy provided by eqs .
( [ asympt ] ) and ( [ center ] ) is very good in this case , for all the values of @xmath5 . ) with @xmath27 ( the usual trap , rather than anti - trap ) and positive effective mass , @xmath114 , as found by the shooting method from eq .
these solutions describe loosely bound states close to the center of the brillouin zone , as per eq .
( [ center ] ) .
continuous curves are numerical counterparts of these solutions , obtained from the full gp equation ( [ gp ] ) with @xmath115 .
( a ) @xmath94 , @xmath116 ; ( b ) @xmath96 , @xmath117 ; ( c ) @xmath98 , @xmath118.,width=480 ] the full gp equation ( [ gp ] ) can not be reduced to the asymptotic nls equation ( [ asympt ] ) if the assumption of the slow variation of the envelope function is not valid ; in that case , strongly confined ( tightly bound , tb ) states may be possible , that are not described by eq .
( [ asympt ] ) . to investigate such solutions , and compare them to their loosely bound counterparts , we performed direct numerical integration of eq .
( [ gp ] ) , looking for solutions that could be classified as having the vorticity @xmath94 or @xmath96 [ in terms of eq .
( [ phi ] ) ] .
figures [ fig6](a ) and [ fig6](b ) display the amplitude of the thus found solutions [ the maximum of @xmath119 as a function of the norm , @xmath120 .
the lower branches in these figures precisely correspond to the lb states described above , the dashed curves showing the dependence of the amplitude vs. @xmath78 as found from the asymptotic equation ( [ asympt ] ) .
the solutions which belong to the upper branches in fig .
6 can be identified as tb ( strongly localized ) 2d gap - soliton states in the gp equation with repulsive interaction between atoms , that persist in the limit of @xmath31 and were first found in refs .
there is weak hysteresis for @xmath94 between the lb and tb states for the parameter values of @xmath121 and @xmath25 .
the discontinuous transition between the lb and tb states occurs for sufficiently large @xmath122 .
( the discontinuous transition was not observed for @xmath122=5 and @xmath25 . )
figures 6(a ) and ( b ) also make it obvious that the asymptotic equation ( [ asympt ] ) predicts a jump upward from the lower branch , which corresponds to the collapse in the asymptotic 2d nls equation ( [ asympt ] ) .
however , the collapse does not occur in the full gp equation ( [ gp ] ) with the repulsive nonlinearity ( note that the negative effective mass is irrelevant when the localization length is on the same order of magnitude as the ol period ) .
the critical value of the norm for the collapse in the asymptotic equation ( [ asympt ] ) is somewhat larger than the value at which the lb - tb transition occurs in the numerical solution of eq .
( [ gp ] ) .
it is necessary to estimate the actual number of atoms in the bec at which the lb - tb transition , shown in fig .
6 , is expected . assuming a pizza - shapedcondensate ,
with the diameter @xmath123 ol periods and transverse width ten times smaller , taking the scattering length @xmath124 nm ( which is close to its value for @xmath2 rb ) , and following once again the derivation of the 2d equation ( [ gp ] ) from the full 3d gp equation , we conclude that the real number of atoms is obtained by multiplying the numbers on the horizontal axes in fig . 6 by a factor @xmath125 . ( a ) and @xmath96 ( b ) vs. the norm ( number of atoms ) @xmath78 , in the case of @xmath126 and @xmath25 [ the dashed lines show the same dependencies , as predicted by the asymptotic equation ( [ asympt ] ) for the loosely bound states ] .
the lower and upper branches of the dependence correspond to the loosely and tightly bound states , respectively ( see further details in the text).,width=384 ] to further illustrate the drastic difference between the lb and tb states with zero vorticity ( @xmath94 ) , we display their typical examples , found as direct numerical solutions of eq .
( [ gp ] ) , in figs .
7(a ) and ( b ) .
these figures display regions in which the absolute value of the respective solution exceeds the quarter - amplitude , i.e. , @xmath127 .
additionally , fig .
7(c ) displays the absolute value for both solutions , @xmath128 , as a function of @xmath19 along the diagonal , @xmath129 .
( zero vorticity ) for @xmath126 and @xmath25 .
the norms of the two states are , respectively , @xmath130 and @xmath131 . panel ( c )
displays the diagonal cross sections of the two solutions , @xmath132 ( the solid and dashed lines depict the loosely and tightly bound states , respectively).,width=480 ] for the loosely ( a ) and tightly ( b ) bound vortices with @xmath96 .
the norms of the two states are , respectively , @xmath133 and @xmath134.,width=480 ] , ( b ) @xmath135 , and ( c ) @xmath136 in the simulation of a loosely - bound vortex with @xmath96 , evolving into a tightly - bound state .
the norm is @xmath137.,width=384 ] and [ fig8 ] for the loosely ( a ) and tightly ( b ) bound vortices with @xmath98 .
the norms of the two states are , respectively , @xmath138 and @xmath139 .
( c ) a profile of @xmath140 for the tightly - bound vortex along the line @xmath141.,width=480 ] in a similar manner , the comparison of the lb and tb states with @xmath96 is presented in fig . 8 , where the panels ( a ) and ( b ) again display regions where , respectively , the loose and tight solutions take the values with @xmath142 , and the panel ( c ) shows the diagonal cross sections of both solutions . in particular , the tb vortex in fig .
8(b ) features a four - fold symmetry ( it is invariant with respect to the rotation by @xmath143 ) . in this connection , it is relevant to mention that , while the same symmetry is observed in all the tb vortices for a sufficiently large norm , @xmath144 , the four - fold symmetry is broken for @xmath145 . to illustrate the latter property , fig .
9 displays three consecutive snapshots of regions with @xmath146 , taken from the simulation starting with an lb vortex state that has @xmath96 and @xmath137 . as is seen
, the patterns do not feature the four - fold symmetry , unlike fig .
8(b ) , but a nearly two - fold symmetric tb state is generated .
the transition between the less symmetric and more symmetric tb vortices accounts for the slope breaking in the amplitude - vs .- norm curve for @xmath147 , which is evident in fig .
we have also found the tb vortices with @xmath98 and checked that they are stable as well .
figure 10 displays the tb vortices with @xmath98 for @xmath148 and @xmath149 . figures 10(a ) and ( b ) display the regions where the tb solutions take the values with latexmath:[$|\phi ( x , y)|~>|\phi the line @xmath141 , on which @xmath151 takes the maximum value for @xmath149 . to the best of our knowledge , the existence of tb vortex solitons in the 2d model with self - repulsion has not been reported before , therefore theses result ( displayed here for @xmath96 and @xmath90 ) are novel by themselves too ( similar stable vortices were very recently found by means of a different method @xcite ) .
the localized solutions belonging to the lb and tb branches differ drastically not only in the dependence of @xmath152 vs. @xmath78 ( i.e. , in their static properties ) , but in dynamical properties too .
as it will be shown in the next section , the tb solitons are always pinned by the underlying ol potential , while the lb states may move freely .
in fact , a sharp transition between mobile and immobile 1d solitons with @xmath153 ( i.e. , close to the bz edge ) and @xmath75 ( repulsive interactions ) was reported in ref .
@xcite . as well as in the 2d case
, the immobilization takes place with the increase of the soliton s norm , and no hysteresis between the mobile and immobile solitons is found .
the abrupt transition between the lb and tb localized states , as described above , is a characteristic feature of the localized states with the negative effective mass , found close to the bz edge , where the lb state is well approximated by eqs .
( [ phi ] ) , ( [ phi ] ) , and ( [ bloch ] ) . on the other hand , no similar transition is found for the localized states close to the bz center , where the approximation ( [ bloch ] ) is replaced by ( [ center ] ) .
the absence of the distinction between lb and tb states in the latter case is explained by the fact that , as is commonly known , tb solitons simply do not exist in the case of the self - repulsion ( @xmath75 ) near the bz center .
localized lb states of course exist , as suggested by the tf approximation .
in the 2d model with both repulsive @xcite and attractive @xcite interaction , the tb solitons are strongly pinned by the ol potential , and can not move ( stable tb solitons also exist in the 2d attractive model with a quasi-1d periodic potential , and in the 3d model with a quasi-2d potential , in which case they may freely move in the unconfined direction @xcite ) . on the contrary to that , the lb localized states described above may move freely on top of the underlying 2d periodic potential , keeping their coherent structure .
this feature resembles a similar one that we have recently demonstrated in the 1d version of the present model @xcite ; in particular , the 1d solitons with the negative effective mass are expelled by the ordinary parabolic trap , and are held by the anti - trap . however , the character of the motion in the 2d case may be completely different , see below . a 2d lb state can be set in motion either by an initial shift from the central position [ relative to the ( anti-)trap ] , or by lending it an initial momentum .
we have performed numerical simulations , using both ways to push the localized state .
for instance , fig .
10 displays results obtained with the initial condition @xmath154 , where @xmath155 is the stationary lb state with @xmath156 and @xmath157 , whose peak position is located at the central point , @xmath158 , and @xmath159 . in other words , this initial condition implies that the localized state is pushed by shifting its center in the @xmath19 direction , and giving it an initial momentum along @xmath20 .
we define the instantaneous position of the center of the localized state as @xmath160 .
figure 11(a ) demonstrates that , as a result , the center of the lb state moves along an elliptic trajectory [ in the case displayed in fig .
11(a ) , the simulations were run up to the time @xmath161 , while the numerically measured period of the elliptic motion was @xmath162 , i.e. , the localized state has completed one full circulation ( in the clockwise direction , since the initial velocity @xmath163 is negative ) . to make the difference from the strongly pinned tb solitons clearer , fig .
11(b ) displays a result of a similar attempt to set in motion the localized state of the tb type . as a result , initially its center moves irregularly , and is then trapped by the ol potential .
many other runs of the simulations produced quite similar results ; in particular , the eccentricity of the elliptic orbit depends on the initial push , and effectively 1d periodic motion along straight lines through the center of the domain are possible too . in this connection , it is relevant to mention that , although the underlying ol makes the medium anisotropic , simulations do not reveal any dependence of the frequency on the direction of the 1d oscillations ( the frequency of the elliptic motion does not depend on the orientation of the ellipse either ) . ) and negative effective mass ( @xmath164 ) in the _ anti - trap _ with @xmath25 . in this case ,
@xmath126 , and the norm of the localized state is @xmath157
. for comparison , panel ( b ) displays an attempt to set in motion a tightly - bound soliton with @xmath165.,width=384 ] the periodic motion of the lb state can be easily described within the framework of the asymptotic equation ( [ asympt ] ) . in that case , one can demonstrate that motion for the center @xmath166 of the localized state , which is treated as a rigid quasi - particle , obeys the newton s equation of motion , @xmath167 [ @xmath47 is the same effective mass as in eq .
( [ asympt ] ) , while the right - hand side is the potential force generated by the parabolic ( anti-)trap ] .
equation ( [ newton ] ) indeed describes the observed motion with a good accuracy .
for instance , in the case displayed in fig .
[ fig8 ] ( a ) , eq .
( [ newton ] ) predicts the period of the motion , @xmath168 which should be compared to the above - mentioned numerically found period , @xmath169.@xmath170 as it was explained above , the localized states found close to the bz center are actually always of the lb type , hence one may expect that they may move freely too .
this is indeed observed in simulations .
typical examples are displayed in fig .
12 , for the localized states with @xmath94 and @xmath90 ( quite a similar motion mode was observed for the vortex with @xmath96 ) .
the trajectory corresponds to anti - clockwise elliptic circulation in this case , as the initial velocity @xmath171 is positive .
we stress that the vortex lb states readily exhibit circular motion with the same period as their counterparts with @xmath94 , and the analytical prediction for the circulation period , based on eq .
( [ t ] ) , is very accurate in this case too .
for example , in both cases ( @xmath94 and @xmath98 ) shown in fig . 12
, the predicted period is @xmath172 , while the one extracted from the numerical data is @xmath173 . , and in the ordinary trap with @xmath27 ( and @xmath174 ) .
panel ( b ) shows the corresponding time dependence of the central coordinate @xmath175 of the moving localized state . in panel
( c ) , the same dependence is shown for the circular motion of the vortex with @xmath98.,width=480 ]
in this work , we have developed analysis of dynamics of nonlinear localized excitations in two - dimensional ( 2d ) bose - einstein condensates ( becs ) with repulsive interactions , which are subjected to the action of the optical lattice ( ol ) combined with the external parabolic potential .
first , we have shown , in an analytical form , that broad ( loosely bound , lb ) localized states , based on the 2d bloch function near the edge of the brillouin zone ( bz ) , feature a negative effective mass ( while the mass of the lb localized states is positive near the bz center ) .
the negative - mass pulse can not be confined by the usual trap , but it is held by an inverted parabolic potential ( _ anti - trap _ ) .
direct simulations have demonstrated that the lb states ( including vortices ) are stable and can freely move on top of the ol , and the frequency of their periodic motion along an elliptic orbit in the anti - trapping potential agrees very well with the analytical prediction , which treats the lb solitons as quasi - particles . besides that , the lb vortices feature rotation around the center . with the increase of the number of atoms ,
an abrupt transition occurs , from the negative - mass lb states to their tightly bound ( tb ) counterparts ; unlike the lb states , the tb solitons are rigidly pinned by the ol .
the positive - mass lb states ( including vortices ) , constructed near the bz center , are stable and move freely too .
further dynamical effects may be considered in the 2d situation , such as collisions between moving solitons .
in fact , various types of collisions are possible , such as between zero - vorticity solitons and vortices , as well as between lb and tb states with the negative mass ( the latter is possible in the presence of the anti - trap ) . on the other hand ,
interactions between solitons with positive and negative mass are precluded , as only one type of the solitons may be confined by a given trapping or anti - trapping potential .
the consideration of collisions is , however , beyond the scope of this paper .
we expect that similar lb and tb states may be found in becs subject to the action of the 3d ( three - dimensional ) ol potential .
however , simulations of the corresponding 3d gp equation is a rather difficult problem .
estimates for physical parameters at which the predicted effects may be observed are similar to those in the 1d counterpart of the model @xcite .
in particular , a circular motion of the lb soliton in the anti - trap may be expected at a frequency @xmath176 hz in a domain of the diameter @xmath177 mm ( @xmath178 spatial periods of the ol ) .
the transition from the lb to tb solitons is expected when the number of atoms in the condensate attains the values between @xmath179 for the @xmath94 states and @xmath180 for the vortex with @xmath96 . in the experiment , the lb and tb localized states are expected to form spontaneously from a properly chosen number of atoms loaded in the appropriate potential , as there is virtually no overlap between these two types of the states . as concerns vortex states
, they may be created using the well - known technique of optical stirring .
the effects predicted in this work for the bec can also take place with spatial optical solitons in bulk photonic crystals .
the work of b.a.m . was supported , in a part , by the israel science foundation through a grant no .
99 greiner m , bloch i , mandel o , hansch t w , and esslinger t , appl . phys .
2001 b * 73 * , 769 ; han d j , depue m t , and weiss d s 2001 phys .
a * 63 * , 023405 ; greiner m , mandel o , hansch t w , and bloch i 2002 nature * 419 * , 51 ; mellish a s , duffy g , mckenzie c , geursen r , and wilson a c 2003 phys .
a 68 , 051601 ; bloch i , greiner m , mandel o , and hansch t w 2003 phil . trans .
l. , ser . a * 361 * , 1409 .
khaykovich l , schreck f , ferrari g , bourdel t , cubizolles j , carr l d , castin y and salomon c 2002 science * 296 * , 1290 ; strecker k e , partridge g b , truscott a g and hulet r g 2002 nature * 417 * 150 ; 2003 new j. phys . * 5 * 73
. konotop v v and salerno m 2002 phys .
a * 65 * 021602 ; louis p j y , ostrovskaya e a , savage c m , and kivshar y s 2003 phys .
a * 67 * 013602 ; ostrovskaya e a and kivshar y s 2003 phys .
90 * 160407 .
carusotto i , embriaco d and la rocca g c 2002 phys .
a * 65 * 053611 ; hilligsoe
k m , oberthaler m k and marzlin k p 2002 _ ibid_. a * 66 * 063605 ; kevrekidis p g , frantzeskakis d j , malomed b a , bishop a r and kevrekidis i g 2003 new j. phys .
* 5 * 64 ; scott r g , martin a m , fromhold t m , bujkiewicz s , sheard f w , and leadbeater m 2003 phys .
. lett . * 90 * 110404 ; efremidis n k and christodoulides d n 2003 phys .
a * 67 * 063608 .
jeng c c , shih m f , motzek k , and kivshar y 2004 phys .
lett . * 92 * , 043904 ; fleischer j w , bartal g , cohen o , manela o , segev m , hudock j , and christodoulides d n 2004 phys .
lett . * 92 * , 123904 . | we study the dynamics of nonlinear localized excitations ( solitons " ) in two - dimensional ( 2d ) bose - einstein condensates ( becs ) with repulsive interactions , loaded into an optical lattice ( ol ) , which is combined with an external parabolic potential .
first , we demonstrate analytically that a broad ( loosely bound , lb ) soliton state , based on a 2d bloch function near the edge of the brillouin zone ( bz ) , has a negative effective mass ( while the mass of a localized state is positive near the bz center ) .
the negative - mass soliton can not be held by the usual trap , but it is safely confined by an inverted parabolic potential ( _ anti - trap _ ) .
direct simulations demonstrate that the lb solitons ( including the ones with intrinsic vorticity ) are stable and can freely move on top of the ol .
the frequency of elliptic motion of the lb - soliton s center in the anti - trapping potential is very close to the analytical prediction which treats the solition as a quasi - particle .
in addition , the lb soliton of the vortex type features real rotation around its center .
we also find an abrupt transition , which occurs with the increase of the number of atoms , from the negative - mass lb states to tightly bound ( tb ) solitons .
an estimate demonstrates that , for the zero - vorticity states , the transition occurs when the number of atoms attains a critical number @xmath0 , while for the vortex the transition takes place at @xmath1 atoms .
the positive - mass lb states constructed near the bz center ( including vortices ) can move freely too .
the effects predicted for becs also apply to optical spatial solitons in bulk photonic crystals . |
the stellar mass of galaxies at the present epoch and the build - up of stellar mass over cosmic time has become the focus of intense research in the past few years . in the local universe ,
results on the stellar mass function of galaxies were published using the new generation of wide - angle surveys in the optical ( sloan digital sky survey ; sdss , @xcite ; 2df , e.g. @xcite ) and near - infrared ( two micron all sky survey ; 2mass , @xcite ) . @xcite combined data from 2mass and 2df to derive the local stellar mass function , @xcite used the sdss and 2mass to the same end . at @xmath3 ,
a number of authors studied the stellar mass density as a function of redshift @xcite reaching @xmath4 , while others , using wider field surveys , investigated the evolution of the mass function of galaxies @xcite to @xmath5 .
generally , the high - redshift work relies on fits of multi - color photometry to a grid of composite stellar population ( csp ) models to determine a stellar mass - to - light ratio , since large and complete spectroscopic samples of galaxies are not yet available .
a similar approach was chosen by @xcite and @xcite , too , at @xmath6 . taking advantage of the availability of photometry and spectroscopy for galaxies in the sdss , ( * ? ? ?
* k03 hereafter ) utilized spectroscopic diagnostics ( 4000 break , d@xmath74000 , and the h@xmath8 balmer absorption line index ) to estimate the mean stellar age and the fraction of stars formed in recent bursts in each galaxy . by comparison of the colors predicted by their best - fit model to the object
s broad - band photometry they determine the amount of extinction by dust and hence the stellar mass - to - light ratio .
the purpose of this letter is to compare the stellar masses determined by this spectroscopic technique to masses obtained from multi - passband photometry and to compare both methods to a simple dynamical estimate of mass .
knowing that none of these methods yields a fiducial ( stellar ) mass , a comparison helps to establish the ( relative ) reliability of each method and makes us aware of potential differences between these estimators .
moreover , it can show us whether one can use observationally cheaper estimators as surrogates for more expensive ( or unobtainable ) ones , which is particularly important when dealing with high - redshift datasets .
specifically , we want to know how the two estimators compare to each other , if using k - band @xmath9 yields better masses than using the g - band @xmath9 which is accessible at high @xmath10 , and how these compare to a simple dynamical mass estimator , @xmath11 .
this letter is laid out as follows . in sect .
[ sec : galaxy - sample ] we describe the sample of galaxies we use in this work . in sect .
[ sec : deriving - masses ] we give a brief overview of how we derive stellar masses by fitting csp models to multi - band photometry . in sect .
[ sec : stellmass ] we compare these masses to the values in k03 and in sect . [ sec : dynmass ] to a simple dynamical estimate of mass based on the sdss velocity dispersions .
we also discuss the implications of these comparisons .
we assume @xmath12 , @xmath13 , and @xmath14 throughout this work .
the sample of galaxies we use in this work is selected from the nyu value - added galaxy catalog @xcite .
this is a merged catalog of objects from the sdss data release two ( dr2 ) and 2mass point - source and extended - source catalogs ( and other catalogs which are not relevant here , as well ) .
we select all objects classified as galaxies and having a secure redshift measurement in the sdss and that are detected in the 2mass catalogs . from this set
we randomly sub - select 20% of the objects leaving us with a sample of sample of @xmath15 objects having redshifts and photometry in ugrizjhk .
we cross - correlate this catalog with the data from k03 to obtain their stellar mass estimates .
the galaxies in the sample span the absolute magnitude range @xmath16 , the restframe @xmath17 color range @xmath18 , and the ( stellar ) mass range @xmath19 .
the method we use to infer stellar masses from multi - color photometry is an advancement of the program used in @xcite .
it is based on the comparison of multi - color photometry to a grid of stellar population synthesis models covering a wide range in parameters , especially star formation histories ( sfhs ) .
we base our new model grid on the @xcite stellar population synthesis package .
we parameterize the possible sfhs by a two - component model , consisting of a main component with a smooth analytically described sfh and a burst of star formation .
the main component is parameterized by a star formation rate of the form @xmath20 , with @xmath21 $ ] gyr and a metallicity of @xmath22 } < 0.3 $ ] .
the age , @xmath23 , is allowed to vary between 0.5 gyr and the age of the universe ( at the object s redshift ) .
the smooth component is linearly combined with a burst of star formation , which is modeled as a 100 myr old constant star formation rate episode of solar metallicity .
we restrict the burst fraction , @xmath24 , to the range @xmath25 in mass ( higher values of @xmath24 are degenerate and unnecessary since this case is covered by models with a young main component ) .
we adopt a salpeter initial mass function for both components , with lower and upper mass cutoffs of 0.1 and 100 @xmath26 .
additionally , both the main component and the burst are allowed to exhibit a variable amount of extinction by dust .
this takes into account the fact that young stars are found in dusty environments and that the starlight from the galaxy as a whole may be reddened by a ( geometry dependent ) different amount .
in fact , @xcite find that the extinction derived from the balmer decrement in the sdss sample is independent of inclination , which , on the other hand , is driving global extinction ( ( see , e.g. , * ? ? ? * ) ) .
this is different from the approach taken by k03 , where a single extinction value for the whole galaxy is used .
we compute the full likelihood distribution on a grid in this 6-dimensional parameter space ( @xmath27 } , t , a_v^1 , \beta , a_v^2 $ ] ) , the likelihood of each model being @xmath28 . to compute the likelihood distribution of @xmath9 , we weight the @xmath9 of each model by its likelihood and marginalize over all parameters .
the uncertainty in @xmath9 is obtained from the width of this distribution .
this procedure is illustrated in fig .
[ fig : loglik ] , where we show seds and likelihood functions for two objects , a young object with a high burst fraction , and an older and fairly quiescent object .
we show projections of the likelihood function onto four planes in parameter space , age vs. dust in the main component , burst fraction vs. burst extinction , age of the main component vs. burst fraction , and star formation timescale , @xmath29 , vs. burst fraction .
the figure also shows the resulting likelihood distributions of @xmath9 in the g , i , and k bands .
note that for the quiescent object , the width of the @xmath9 distribution is very similar in the g and k bands , while it is much wider in g than it is in k for the younger star forming object . on average , the width of the likelihood distribution of @xmath9 at 68% confidence level is between @xmath30 and @xmath31 dex ( using @xmath32 ) .
the uncertainty in mass has a weak dependence on mass ( increasing with lower @xmath33 photometry ) and much of the variation is in spectral type : early - type galaxies have more tightly constrained masses than late types ( see also fig .
[ fig : loglik ] ) . using the u band
, the uncertainty in mass grows by @xmath34 dex .
in fig . [ fig : mass - comp ] we compare our photometry - based stellar mass estimates to the stellar masses of k03 .
we show the h@xmath2 equivalent width ( as measured by the sdss ) by color coding .
the overall impression from this figure is that the two different estimators agree remarkably well , within a rms scatter of only @xmath35 dex over almost 4 decades in mass ( and hence they largely agree within their respective uncertainties ) .
this is only a relative statement , though .
it does not imply that the masses are accurate to that level in an absolute sense , although it is very reaffirming .
however , as may be expected , there are systematic differences as a function of star formation activity on the @xmath1 dex level . the d@xmath74000 and h@xmath8 based method of k03 yields masses almost identical to ours for weakly star forming objects ( eqw(h@xmath2 ) @xmath36 ) at masses above @xmath37 @xmath26 .
at lower masses , our estimator tends to give slightly higher masses than k03 s . for more strongly star forming objects , the photometrically determined masses are smaller than the ones of k03 . for objects with eqw(h@xmath2 ) @xmath38 , the discrepancy becomes as large as 0.15 dex , independent of mass .
note that at high redshift , such objects will be more common .
we suspect that there are a multitude of reasons for these differences based on the different sampling of stellar populations by both methods ( if we leave out the jhk bands , our masses become more similar in their trends to k03 s , although with increased scatter ) .
plausibly , though , this is explainable by the fact that the photometrically determined masses sample the light from the whole galaxy , while k03 s sdss - based sampling of d@xmath74000 and h@xmath8 covers only the inner 3 arcseconds . since most galaxies are redder in their centers than in the outer parts , this might lead to higher masses for star forming disk galaxies .
early - type galaxies without blue star forming disks do not suffer from this effect .
this is confirmed by restricting the sample to low redshifts , which maximizes the effect . also , at the lowest masses , galaxies might have more irregular sfhs , and photometric methods might fail in this case @xcite . however , fig.[fig : mass - comp ]
does not show a dependence of the residuals on mass , only on current star formation rate .
[ fig : mass - comp ] also shows that our masses based on the g band are very similar to the ones estimated through the k band . for early type systems and weakly star forming systems , they are statistically indistinguishable . star forming systems , however , tend to have g - band masses lower by @xmath39 dex .
the good agreement is partly due to the fact that the effect of dust extinction and age on the effective @xmath9 are very similar in canonical models , as has been pointed out by @xcite , although they are not completely degenerate .
typically , the spread of the stellar @xmath9 in the galaxy population at any given luminosity is around 0.7 dex in g and 0.35 dex in k. these results is reaffirming since the restframe blue spectral range is accessible to photometry to very high redshift , and thus high-@xmath10 studies mostly rely on @xmath40 .
since we do not have a fiducial mass estimator ( neither for stellar mass nor for total mass , for that matter ) , it is only natural to ask how the stellar mass measurements presented here compare with estimators of mass based on kinematic data .
in fact , it has been suggested that stellar mass ( or , more accurately , baryonic mass ) and total mass are tightly related and that stellar mass can be used as a surrogate for total mass in the context of high-@xmath10 galaxy surveys to probe structure formation ( see , e.g. , * ? ? ? * ) .
we use the measurements of velocity dispersion , @xmath41 , and effective radius in the g band , @xmath42 , provided by the sdss pipeline to plot stellar mass vs. dynamical mass , @xmath43 , in fig .
[ fig : mass - dyn ] .
the left hand panel shows @xmath44 vs. photometric stellar mass and the right hand panel shows @xmath44 vs. the stellar masses of k03 .
color again encodes h@xmath2 equivalent width . above @xmath45 @xmath26 ,
the stellar masses from both methods follow the dynamical masses remarkably well . below @xmath46
@xmath26 , the velocity dispersion measurements of the sdss becomes unreliable as we approach the instrumental resolution of the data ( @xmath47 km s@xmath48 ) . at higher masses , although both estimators generally follow @xmath44 , there are again some differences , and both estimators show similar trends in their residuals although with different amplitudes .
stellar masses agree very well with @xmath44 at the highest masses ( which are mostly populated by old , quiescent objects ) . at lower masses , stellar masses
show a trend to larger values than @xmath44 with decreasing mass and with increasing h@xmath2 emission line equivalent width .
this effect is weak in the photometric estimator , and stronger in k03 s method , which gives stellar masses larger than @xmath44 by 0.1 to 0.4 dex at almost all masses .
this comparison is unchanged by using @xmath49 instead of @xmath32 .
it is important to note that @xmath44 is not a good estimator of total mass , and that this comparison is again only to be taken in relative terms .
in fact @xmath50 can only provide a lower limit to the mass .
however , as long as a bulge is present , the total mass should not be underestimated by more than @xmath51 dex ( see , e.g. fig . 4 in @xcite ; also @xcite , who show that @xmath50 is a reasonable mass estimator , although this paper is concerned with ellipticals only ) .
it is therefore not surprising to find stellar masses in excess of @xmath44 and the difference between the two increasing at lower masses . nevertheless , the point of this work is to assess the general consistency and reliability of stellar mass estimates than to investigate the relationship between stellar and dynamical mass in galaxies .
[ fig : mass - dyn ] shows that the estimators of stellar mass , and especially the photometric estimator which is most easily obtainable for large high redshift samples ( covering a bluer wavelength range , though ) , closely follow @xmath44 as measured by this simple dynamical measure .
we can not see significant systematic deviations which would bias or invalidate this estimator .
this is a very encouraging result , since such an estimator is very likely to be the only one available at @xmath3 for some time to come .
this publication makes use of data products from the two micron all sky survey , a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by nasa and nsf .
funding for the creation and distribution of the sdss archive has been provided by the alfred p. sloan foundation , the participating institutions , nasa , nsf , the u.s .
department of energy , the japanese monbukagakusho , and the max planck society . | the purpose of this letter is to check the quality of different methods for estimating stellar masses of galaxies .
we compare the results of ( a ) fitting stellar population synthesis models to broad band colors from sdss and 2mass , ( b ) the analysis of spectroscopic features of sdss galaxies @xcite , and , ( c ) a simple dynamical mass estimate based on sdss velocity dispersions and effective radii .
knowing that all three methods can have significant biases , a comparison can help to establish their ( relative ) reliability . in this way , one can also probe the quality of the observationally cheap broadband color mass estimators for galaxies at higher redshift .
generally , masses based on broad - band colors and spectroscopic features agree reasonably well , with a rms scatter of only @xmath0 dex over almost 4 decades in mass .
however , as may be expected , systematic differences do exist and have an amplitude of @xmath1 dex , corrleting with h@xmath2 emission strength .
interestingly , masses from broad - band color fitting are in better agreement with dynamical masses than masses based on the analysis of spectroscopic features .
in addition , the differences between the latter and the dynamical masses correlate with h@xmath2 equivalent width , while this much less the case for the broad - band masses .
we conclude that broad band color mass estimators , provided they are based on a large enough wavelength coverage and use an appropriate range of ages , metallicities and dust extinctions , can yield fairly reliable stellar masses for galaxies .
this is a very encouraging result as such mass estimates are very likely the only ones available at significant redshifts for some time to come . |
there are several mechanisms that can lead to the production of kaon antikaon pairs in nucleon
nucleon collisions near threshold . these can be divided mainly into two general classes involving ( a ) the production of a non strange meson that subsequently decays into @xmath5 , and ( b ) the associated production of @xmath6 , where the @xmath7 is formed through the decay of the hyperon @xmath8 .
the most prominent meson that is found almost exclusively in the @xmath9 channels is , of course the @xmath0 , and there have been measurements of the @xmath10 reaction from disto @xcite and cosy
anke @xcite , as well as of the @xmath11 reaction @xcite .
these experiments show strength also away from the region of the @xmath0 peak that could arise from other mesons decaying into @xmath12 .
the lightest of the non strange scalar mesons are the @xmath13 and @xmath14 , which decay mainly into @xmath15 and @xmath16 , respectively .
both have masses around 980mev / c@xmath17 and widths of the order of 50100mev / c@xmath17 @xcite .
since the central mass values fall almost exactly at the @xmath5 threshold , the strong coupling to this channel distorts significantly the upper parts of the mass spectra @xcite .
it has even been suggested that both resonances might be mainly molecular in nature @xcite .
since the @xmath12 system is a mixture of isospin @xmath18 and @xmath19 , one can consider whether the non@xmath0 events observed in might be linked to the production and decay of the @xmath20 and/or the @xmath21 mesons .
the @xmath19 channel is isolated cleanly by looking at the @xmath22 reaction @xcite . on the other hand
, we can also expect kaon pairs to be created in the @xmath23 reaction , where the hyperon decays through @xmath24 .
there are , of course , several excited hyperons that could contribute to such a process . of particular interest for low energy production
are the @xmath25 and the @xmath26 .
though nominally lying below the sum of the @xmath2 and @xmath27 masses , their large widths ( @xmath28mev / c@xmath17 ) ensure that they overlap the threshold @xcite .
results have recently been presented on the production of these states in the @xmath29 and @xmath30 reactions , where the hyperons were detected in the @xmath31 and @xmath32 channels , respectively @xcite .
in addition to being involved in a direct production mechanism , the @xmath26 and the @xmath25 might also lead to a significant final state interaction ( _ fsi _ ) in the reaction .
evidence that the _ fsi _ is indeed important is to be found in the measurement of the cross section at energies below the @xmath0 threshold @xcite .
it was shown that the invariant masses of the final system were on average lower than those of the @xmath1 pairs and it was suggested that this is due to the strong attraction in the system .
it is therefore an interesting but open question whether the data from the non@xmath0 region should be viewed , in first approximation , as scalar resonance production _ via _ @xmath33 or hyperon production through @xmath34 .
it is one aim of the present paper to study this competition by using the data obtained when investigating @xmath0meson production in proton proton collisions at anke @xcite . a program to investigate the production of kaon pairs in nucleon
nucleon collisions has been initiated at the cosy storage ring of the forschungszentrum jlich .
although the methods of the measurement and analysis of the reaction have already been given in ref .
@xcite , for the sake of clarity the main points are repeated in sect [ experiment ] . due to the finite width of the @xmath0 ,
distinguishing between events corresponding to the production of this meson and the rest requires a detailed modelling of the @xmath12 spectrum , as described in sect .
[ kkbar ] .
when more than one pair of particles interact strongly in the final state , there is no reliable prescription to evaluate a corresponding enhancement factor .
our approach to this problem in terms of an effective scattering length @xmath35 is described in sect .
[ fsia ] , with a heuristic justification being given in appendix [ app_a ] .
the relative distributions of the and @xmath1 invariant masses are shown and discussed in sect
. [ ratio ] .
the ratio of the differential cross sections as functions of the and @xmath1 invariant masses indicates that the attraction is very strong .
this confirms the earlier cosy11 findings @xcite but with higher statistics over much wider ranges of energy and mass .
the events that come from the non@xmath0 region in the 2.65gev data allow us to fix the parameter @xmath35 necessary to describe the @xmath36 ratio within our final state interaction model .
the data are mainly sensitive to the magnitude @xmath37 and the value obtained is not dissimilar to that required to describe free scattering .
the approach reproduces also our results at the other two energies as well as the cosy11 data .
if allowance is made for the small possible contribution to this ratio from @xmath0 events , of the order of 20% , the two particle ratio is also well described for all @xmath12 mass intervals .
in addition , we find a completely analogous effect in the three - body @xmath38 system , with the @xmath39 ratio being strongly biased towards the lowest values of the invariant mass .
we show in sect .
[ ratiokpp ] that these distributions are also predicted quantitatively by the same final state interaction model with the same value of the parameter @xmath35 .
the total cross sections for @xmath0 and non@xmath0 production are presented in sect .
[ sigma_total ] .
the energy dependence of the non@xmath0 total cross section is strongly influenced by both the @xmath40 and _ fsi _ but , even after including these effects , the lowest energy data are not well reproduced .
this leads to the suggestion that the effect might be connected with the two or three points in the @xmath12 spectrum at very low invariant masses , which lie consistently above the simulations .
our conclusions and views on the outlook are presented in sect .
[ conclusions ] .
the study of kaon pair production has been performed using data taken with the anke magnetic spectrometer @xcite .
this is placed at an internal target station of cosy , which is the cooler synchrotron of the research center jlich @xcite .
we have already reported an initial analysis of the data , from which values of the @xmath41 cross sections were extracted @xcite .
the measurements were performed at three proton kinetic energies of @xmath42gev , 2.70gev and 2.83gev , corresponding to excess energies of @xmath43mev , 67mev and 108mev with respect to the @xmath44 threshold .
a dense hydrogen cluster jet target @xcite provided areal densities of and the average luminosity during the experiment was typically @xmath45 .
the detection system of the three
dipole magnetic spectrometer anke registers simultaneously slow positively and negatively charged particles , with fast positive particles being measured in the forward system @xcite .
kaons produced in the @xmath46 reaction at 2.65 and 2.70gev have laboratory momenta between 0.2 and 1.0gev / c and this full range was covered by the detection system for both @xmath47 and @xmath2 . for the 2.83gev run , the @xmath47
were detected only up to 0.6gev / c , which reduced the geometrical acceptance .
the momentum range of the protons emerging from the reaction is between 0.8 and 1.6gev / c , for which the forward detector has full acceptance .
the momenta of both the kaons and protons were determined from the track information furnished by sets of multiwire proportional chambers .
the tracking efficiency for kaons fell smoothly from 98% to 93% as the momentum increased , whereas that for protons was about 80% over the whole momentum range .
the efficiency map , including its dependence on the particle momenta and position , was generated for the track reconstruction and used in the analysis .
the data taking efficiency was around 90% and , taken in combination with the tracking efficiency , resulted in a total detector efficiency of about 60% . in order not to reduce the geometric acceptance unnecessarily , only one forward going proton was measured in coincidence with the charged kaon pair .
the @xmath46 reaction was then identified by requiring that the missing mass in the reaction was consistent with that of the non observed proton . as a first step
, positive kaons were selected through a procedure , described in detail in ref .
@xcite , that used the time of flight ( tof ) between start and stop scintillation counters of a dedicated @xmath47 detection system . in the second stage ,
both the @xmath48 and forward going proton were identified from the time of flight differences between the stop counters in the negative and forward detector systems with respect to the stop counter in the positive system that was hit by the @xmath47 . by comparing this with the time difference expected for a kaon pair based upon their measured momenta ,
a clear separation is achieved , with good events lying in an island where the two time differences are consistent @xcite .
these two tof selections , as well as that for the @xmath47 , were carried out within @xmath49 bands . events at @xmath50=2.65 gev .
the hatching indicates the range used for the selection of protons .
the side band events scaled by the solid line are indicated by the dotted shading.,width=257 ] the spectrum of missing masses with respect to the @xmath51 system , shown in fig .
[ immm2.65 ] for the 2.65gev data , is dominated by the proton peak .
the corrected numbers of @xmath52 events obtained in this way are around 3100 ( at @xmath42gev ) , 1300 ( 2.70gev ) , and 650 ( 2.83gev ) .
the estimated background inside the proton cut window is 5% , 12% , and 18% at these three energies , respectively .
this has been subtracted using the side band events scaled by the solid line shown in fig .
[ immm2.65 ] . the uncertainty resulting from this procedure
is estimated to be less than 3% for the total cross sections and negligible compared to the statistical errors for the differential distributions .
after the identification of the events , the data were binned in intervals of the @xmath12 invariant mass , with the 2.65gev results being presented in fig .
[ raw_spectrum_kk ] .
this spectrum shows a very prominent @xmath0 peak positioned close to the nominal mass of the meson .
in addition there is a non@xmath0 contribution that decreases steadily from low to high invariant masses .
this behavior , which is mainly a reflection of the anke acceptance for non
resonant @xmath44 production , has to be simulated in order to derive cross sections .
invariant mass distribution for the reaction at 2.65gev .
the prominent @xmath0 peak sits on a steadily falling background corresponding to the anke acceptance for the four particle final state .
the simulation of these two contributions and their sum are shown , respectively , by the dotted , dashed , and solid histograms .
[ raw_spectrum_kk],width=257 ] the acceptance has been determined through a monte - carlo simulation based upon geant4 @xcite , where the geometrical acceptance , resolution , detector efficiency , and kaon decay probability were taken into account .
in the initial step , the @xmath0production distributions previously published @xcite were taken as the basis of the simulation , whereas four body phase space was used for the non@xmath0 component of fig .
[ raw_spectrum_kk ] .
there are seven degrees of freedom for a four - body final state produced in an unpolarized reaction .
thus for each beam energy , seven independent c.m .
distributions were generated ( five angular distributions , the relative momentum of final protons , and the @xmath12 invariant mass ) .
these were then divided into two groups depending on the value of the @xmath12 invariant mass , _
i.e. _ , @xmath0rich and @xmath0poor regions .
all distributions were jointly fitted to the experimental data and the relative contribution of @xmath0 and non@xmath0 production evaluated for the determination of the acceptances . in the @xmath0poor region , most of the c.m .
distributions are fairly uniform , though the polar angle of the @xmath12 system relative to the beam axis in the overall c.m .
system shows deviations .
fitting this with the function @xmath53 gives anisotropies of @xmath54 and @xmath55 at 2.65gev and the two higher energies , respectively .
the inclusion of this factor increases the total acceptance for non@xmath0 production by about 10% .
in addition , as will be seen in sects . [ ratio ] and [ ratiokpp ] , the @xmath56 and @xmath57 invariant masses deviate strongly from phase space .
these deviations have been taken into account in the simulations , using iteratively the theoretical approach discussed in sect .
[ fsia ] , in order to converge on the acceptance corrected distributions .
the simulations of the @xmath0 and non@xmath0 components of the uncorrected @xmath12 spectrum at 2.65gev are shown in fig .
[ raw_spectrum_kk ] together with their sum .
the estimated total acceptance for kaon - pair production is 1.1% ( 2.65gev ) , 0.7% ( 2.70gev ) and 0.1% ( 2.83gev ) , and the associated systematic uncertainties in the total cross section are 5% , 7% and 12% , respectively .
the shapes of the differential cross sections , especially those for the @xmath58 and @xmath38 invariant masses , are very stable to this iterative procedure . as a consequence ,
the uncertainty that this introduces in the differential cross sections is negligible compared to the statistical errors .
the luminosity required to evaluate absolute cross sections was determined by measuring in parallel proton
proton elastic scattering for laboratory angles between 5.0@xmath59 and 8.5@xmath59 . for this purpose
the momentum of a forward going proton was determined using the anke forward detector . after a missing mass analysis
, the ( small ) background underneath the proton peak was subtracted and the remaining events were corrected for efficiency and acceptance .
figure [ spectrum_kk ] shows the differential cross sections in terms of the @xmath12 invariant masses as obtained at our three energies .
these have been corrected for acceptance , deadtime , _ etc .
_ , and normalized on the basis of the proton proton elastic scattering data , as discussed in sect . [ sigma_total ] .
the resulting non@xmath0 contribution is much flatter than that illustrated in fig .
[ raw_spectrum_kk ] . in order to ensure a clean separation of the @xmath0 from the rest ,
it is necessary to model the shapes as reliably as possible .
invariant mass at ( a ) 2.65gev , ( b ) 2.70gev , and ( c ) 2.83gev .
the dotted histogram and the dashed curve represent the simulations for the @xmath0 and non@xmath0 contributions to the spectrum , with their sum being shown by the solid histogram .
for all three energies the simulations fail to reproduce the lowest mass points.,width=241 ] in our earlier work @xcite , the non@xmath0 contribution was assumed to be described by a four body phase space .
the @xmath0 was taken to have a breit wigner form with a constant width of @xmath60mev / c@xmath17 @xcite , which was then convoluted with an experimental resolution of @xmath61mev / c@xmath17 . only for the @xmath0 component
was the spectrum modified to include effects arising from the final state interaction between the emerging protons .
these three assumptions have now to be reconsidered . since the @xmath0 is a @xmath27wave resonance that decays strongly into @xmath5 , its width depends upon the @xmath12 and @xmath62 relative momenta . due to the narrowness of the state ,
the peak itself is little affected by this refinement but it does suppress slightly the low mass tail of the resonance .
although for low @xmath12 effective masses there is no clear evidence for any @xmath40 _ fsi _ , only minor effects are to be expected there because of the large amount of energy that is then available to excite the @xmath40 system . nevertheless , the _ fsi _ can not be neglected at higher @xmath12 masses , where its inclusion distorts slightly the four body phase space spectrum . finally , as will be seen in sect
. [ ratio ] , it is clear that there is a large final state interaction between the @xmath2 and one or both protons and this also modifies the shape of the non@xmath0 component , enriching the central part of the @xmath12 spectrum . by chance ,
all three effects go in the same direction . in the fits to fig .
[ spectrum_kk ] , the fractions to be associated with @xmath0 production are diminished somewhat compared to those reported in our previous work @xcite .
it must be noted that the lowest mass points lie above the fitted curves for all three data sets of fig .
[ spectrum_kk ] .
these deviations from the simulations are significant because , as seen from the comparison of figs .
[ raw_spectrum_kk ] and [ spectrum_kk]a , the anke acceptance is largest in this region .
although ignored in the fitting process , they were taken into account in the evaluation of the non@xmath0 contribution to the total cross section , contributing 4% ( 2.65gev ) , 6% ( 2.70gev ) , and 10% ( 2.83gev ) .
we wish to describe the data at all energies in terms of final state interactions between the two protons and between the @xmath2 and either one or both of the protons .
there is no reliable way of carrying out such a program without solving multibody equations , which is well outside the scope of the present paper .
as explained in appendix [ app_a ] , we make the _ ad hoc _ assumption that the overall enhancement factor is the product of the enhancements in the @xmath40 and two systems , all evaluated at the appropriate relative momenta @xmath63 : @xmath64 similar approaches are to be found elsewhere in the literature as , for example , in the description of the @xmath65 final state at low energies @xcite . the enhancement factor is taken in the scattering length approximation @xmath66 where some caution should be exercised in the interpretation of the complex parameter @xmath35 as an effective scattering length . in principle
@xmath35 could have an energy dependence , which would correspond to the introduction of an effective range term .
however , it is hard from our data to determine the values of any extra free parameters .
a proton proton enhancement factor of the form @xmath67 is assumed @xcite , where the position of the @xmath68 virtual state is well fixed at @xmath69@xmath70 .
the final state interaction should not influence higher partial waves and , in order to reduce its effects above about 10mev , we take @xmath71@xmath70 . the relative momentum of the two protons in the @xmath40 rest frame , which is very sensitive to the form of the @xmath40 _ fsi _ ,
is then reproduced reasonably , as seen from fig .
[ p_dist ] .
= 6.8 cm it is important to note that the form of neither the @xmath40 nor the enhancement factor is valid for an excitation energy below about 1mev due to the explicit coulomb force .
however , these regions represent only very small fractions of the allowed phase space .
and ( b ) the systems . in panel ( c )
their ratio is shown on a logarithmic scale .
these 2.65gev data have been selected to come from the region @xmath72gev
/ c@xmath17 , where there is only a very small @xmath0 contribution .
the dashed histograms represent the results of four body phase space simulations , where the only distortion is that coming from the @xmath40 _ fsi_. the solid histograms include also the _ fsi_.[265lt1010],width=241 ] in data taken with the cosy11 spectrometer at @xmath73mev , _
i.e. _ at an energy where there is very little @xmath0 production , it was found that the and @xmath1 invariant mass @xmath74 distributions were markedly different @xcite .
the ratio of the acceptance corrected distributions @xmath75 showed a very strong preference for low values of @xmath74 .
figure [ 265lt1010 ] shows the @xmath58 spectra obtained at our lowest energy . in order to consider a situation similar to that of cosy11 ,
the data have been selected such that only events with @xmath12 invariant masses less than 1.01gev / c@xmath17 are retained .
the simulation that takes into account only the @xmath40 _ fsi _ fails to describe the data .
furthermore , the ratio of the to @xmath1 mass distributions is far from constant , as it would be if there were only the @xmath40 _ fsi_. the peaking of @xmath76 to the lowest invariant masses confirms the cosy11 observation @xcite but with higher statistics and over a wider range of @xmath58 masses .
level uncertainty of the fits of the complex parameter @xmath35 of the _ fsi _ _ ansatz _ to the data of fig .
[ 265lt1010 ] .
our best fit is obtained with @xmath77fm ( marked with a cross ) .
also shown are the best fit values ( points ) and corresponding contours ( dashed lines ) for the isoscalar ( @xmath20 ) and isovector ( @xmath78 ) scattering lengths derived from the study of free @xmath2nucleon scattering @xcite .
[ contour],width=289 ] when the final state interaction factors are also introduced into the four body phase space simulation of the non@xmath0 contribution to the reaction , the individual @xmath79 and @xmath80 distributions of fig .
[ 265lt1010 ] , as well as their ratio @xmath76 , are all well described .
the best fit to these data is achieved with an effective scattering length of @xmath81fm and this value of @xmath35 will be retained for all the subsequent simulations . nevertheless , the uncertainties in the real and imaginary parts are quite large and strongly correlated such that @xmath37 is determined much better than its phase , as illustrated by the @xmath82 contour plot of fig .
[ contour ] . the free @xmath18 and @xmath19 values from ref .
@xcite are also shown there along with their one@xmath83 limits . from this
it is seen that there is an overlap with either isospin scattering length .
one has also to bear in mind that our data represent averages over some energy range and that a constant scattering length is necessarily an oversimplification .
ratio for the reaction at 2.65gev extracted for four different selections of the @xmath12 invariant mass , _ viz . _ ( a ) @xmath84gev / c@xmath17 , ( b ) @xmath85gev / c@xmath17 , ( c ) @xmath86gev / c@xmath17 , and ( d ) @xmath87gev / c@xmath88
. the histograms are predictions of the model described in the text , where the @xmath2 mesons from non@xmath0 events are subject to a final state interaction . in the high mass interval ,
the effects of assuming a 20% probability of the @xmath2 from the @xmath0 decay being influenced by a _
fsi _ is considered and this leads to the dashed histograms . [ fourbinsmpk],width=224 ] the situation for higher @xmath12 masses is complicated by the contribution from the @xmath0 meson . at our beam energies , the average separation of the @xmath2 ( produced in the @xmath0 decay ) from one of the final protons is about 7fm .
this large distance reduces considerably the effects of the _
fsi_. we would therefore expect that deviations of @xmath76 from unity should arise primarily from the non-@xmath0 contributions .
the modelling of the data thus depends in the first instance upon the determination of the fraction of @xmath0 production in the overall spectra presented in fig .
[ spectrum_kk ] . however , classical simulations suggest that more than 15% of the @xmath0 decays lead to at least one pair with a separation of less than 1fm , for which the _ fsi _ is not negligible
. ratio for ( a ) the @xmath0poor region and ( b ) the @xmath0rich at 2.70gev with ( c ) and ( d ) showing respectively the analogous results at 2.83gev . the predictions obtained using the standard parameters are shown with no _ fsi _ effect for the @xmath2 from the decay of the @xmath0 ( solid curve ) and 20% ( dashed ) .
[ ratio270283],width=224 ] to investigate this point further , the 2.65gev data have been divided into four regions of @xmath12 invariant mass , _ _
viz.__@xmath89gev / c@xmath17 , @xmath90gev / c@xmath17 , @xmath86gev / c@xmath17 , and @xmath87gev / c@xmath91 . the first two intervals contain almost no @xmath0 events , whereas the last corresponds mainly to @xmath0 production .
the corresponding @xmath76 distributions are shown in fig .
[ fourbinsmpk ] , where they are compared to the same simulation as that shown in fig .
[ 265lt1010 ] .
the non-@xmath0 region is very well described but , for high @xmath12 invariant masses , the asymmetry seems to be diluted too much by kaons from the @xmath0 decay .
the agreement is improved by assuming that 20% of the @xmath0 events are influenced by the _ fsi_.
the same effect could be achieved by artificially reducing the fraction of @xmath0 mesons in the peak of fig .
[ spectrum_kk]a by 10% , but this would be hard to justify . if the strong final state interaction is responsible for the distortion in the @xmath36 ratio seen in fig
. [ fourbinsmpk ] , similar effects should be seen for the other excess energies . for our @xmath92 and 108mev results shown in fig .
[ ratio270283 ] , the statistics are lower and we merely divide each data set into two , comprising @xmath0rich and @xmath0poor regions . without any change in the value of the effective scattering length ,
all the distributions are well described , especially if it is assumed that 20% of the @xmath0 events lead to _ fsi _ effects .
data on the @xmath36 ratio were not published by the disto collaboration @xcite .
however , the same good description of @xmath76 is found for the cosy11 results of fig .
[ cosy11 ] where , in view of the low excess energies , there is no ambiguity regarding the @xmath0 contribution .
given that the systematic uncertainties in this experiment are independent of those at anke , the quantitative agreement shown by the 28mev data is convincing .
unfortunately , in the publication @xcite the values of the ratio at 10mev were scaled upwards to make them lie closer to those taken at 28mev .
since the numbers of and @xmath1 events are the same , the weighted average of the ratio must be unity and this allows one to make an informed guess regarding the value of the scaling that has to be reintroduced .
no account has been taken of the cosy11 invariant mass resolution , which will be of greater significance for the 10mev data .
ratio for the cosy11 data at ( a ) @xmath93mev ( upper panel ) and ( b ) @xmath73mev .
the small contribution from the low mass tail of the @xmath0 has been neglected in the 28mev simulation . since
the 10mev results were arbitrarily scaled in the publication @xcite , they have here been divided by a factor of three before being compared to the predictions of the _ fsi _ model , using the same values of the parameters as for the anke data . ,
we have argued in sect .
[ fsia ] and appendix [ app_a ] that it is natural to take the final state enhancement to be the product of the individual factors corresponding to the @xmath40 and two interactions , which allows the kaon to be attracted simultaneously to both protons .
this in turn suggests that there might also be an enhancement in the three body system at low @xmath94 invariant masses .
this is indeed the case .
the size of the effect at 2.65gev is illustrated in fig .
[ kpp ] for the regions with @xmath95 .
to @xmath4 invariant masses for ( a ) the @xmath0poor region and ( b ) the @xmath0rich at 2.70gev with ( c ) and ( d ) showing respectively the analogous results at 2.83gev .
the predictions obtained using the standard parameters are shown with no _ fsi _ effect for the @xmath2 from the decay of the @xmath0 ( solid curve ) and 20% ( dashed ) .
the simulations that describe the two particle ratios also represent well these three body ratios .
[ kpp2],width=218 ] there is a striking preference for low @xmath38 masses in both @xmath12 intervals and this behavior is reproduced quantitatively by our simulation , which uses the same effective scattering length @xmath35 that was chosen to describe the two
body @xmath76 ratio .
it provides some empirical justification for our choice of the final state interaction enhancement factor as the product of the individual two body enhancements .
this favors configurations where the @xmath2 is attracted simultaneously to both protons .
it must , however , be stressed that the strong mass dependence which is apparent in fig .
[ kpp ] does not necessarily imply that the @xmath2 forms a bound state with the two protons @xcite . though the error bars are considerably larger at the two higher energies , as seen in fig .
[ kpp2 ] , the simulation is consistent with the results . in all these three particle mass distributions
the introduction of a small amount of _ fsi _ for the products of the @xmath0 decay seems to be of less importance than for the @xmath58 spectra .
the total cross section results for both @xmath0 and non@xmath0 production are given in table [ table1 ] for the three energies .
the systematic errors given here arise from the quadratic sum of the uncertainties in the background subtraction , acceptance correction , tracking efficiency correction for kaons ( 3% ) and the analysis of the @xmath40elastic data used for the normalization .
unfortunately , the @xmath40 database is very limited in our region of angle and energy .
this is recognized in the recent sp07 update to the said program where it is stated that _ our solution should be considered at best qualitative between 2.5 and 3gev _
the raw sp07 solution generally overestimates the values of those @xmath40 differential cross sections that are available in this range @xcite .
a normalization option has to be chosen to ensure agreement with the @xmath40 experimental data and this typically leads to reductions of about 15% at our two lower energies but almost 20% at 2.83gev . to eliminate any ambiguities , the values that we have assumed for the @xmath40 elastic cross section , integrated over the solid angle defined by @xmath96 ,
are presented in table [ table1 ] for our three energies .
.[table1]total cross sections for the @xmath97 and @xmath98 reactions at three beam energies @xmath50 and corresponding excess energies @xmath99 with respect to the @xmath44 threshold .
the excess energy with respect to the @xmath0 threshold is given by @xmath100mev .
the @xmath0 cross section has been corrected for the 49.1% branching ratio @xcite . for both channels
the first error is statistical and the second systematic .
the values that we have taken for @xmath101 , the integral of the proton proton elastic cross section over the solid angle range @xmath102 , are also given .
the associated @xmath103 uncertainty has not , however , been compounded with the other errors . [
cols="^,^,^,^,^ " , ] the value that we obtain for the luminosity at 2.65gev on the basis of the @xmath40 cross section given in the table could be checked using a novel technique @xcite .
this depends upon the energy loss of the coasting proton beam caused by its multiple traversals through the very thin target .
the resultant frequency change can be measured with high accuracy using the spectrum of the schottky noise . combined with measurements of the beam current , this yielded luminosities with an expected precision of about @xmath104 .
the two methods give consistent results within this uncertainty and , although the frequency change technique could not be applied as reliably at the higher energies , it is expected that the overall luminosity uncertainty is on about this level .
our total cross section results for non@xmath0 production are plotted in fig .
[ totalxsec ] along with results taken from disto @xcite and cosy11 @xcite .
it is seen that four body phase space can not describe simultaneously the data at high and low excess energies .
the situation is improved only a little if account is taken of the strong attraction between the two protons in the @xmath68 state but a greater improvement is achieved through the introduction of the _ fsi _ in the system .
nevertheless , the cosy11 data at excess energies of 20mev or less still seem to be underestimated .
this might be connected to the incomplete description of the @xmath12 spectra of fig .
[ spectrum_kk ] at very low invariant masses .
however , it must be stressed that in the evaluation of the cosy11 acceptance no account was taken of any _
fsi_. = 1.1
we have investigated the reaction at three energies . in those parts of the @xmath12 mass spectra where @xmath0 production is small
, we find a marked difference between the and @xmath1 invariant mass distributions .
there is a strong peaking in the @xmath36 ratio towards the @xmath58 threshold for both the cosy11 and our data .
this clearly indicates that the reaction can not be dominated by the undistorted production of a _ single _ scalar resonance @xmath20 or @xmath21 . either a mixture of the two is produced or one of the kaons from the @xmath105 decay interacts with a final proton .
as we have shown , these data can be explained quantitatively through the introduction of a simple final state interaction . in kinematic regions where @xmath0 production is significant ,
the @xmath106 asymmetry is reduced because the meson may travel some distance before decaying .
our data in such conditions are well fit by assuming that about 20% of the @xmath2 from the @xmath0 decay interact with one or both final protons .
the uncertainty in this is , however , quite large because of the limited precision with which the @xmath0 contribution to the @xmath12 mass spectrum can be isolated .
there is no universal technique for modelling the _ fsi _ when more than one pair of particles is involved .
the product _ ansatz _ has the benefit that the same simulation with the same parameters reproduces also the strong peaking in the ratio of the @xmath39 mass distributions .
we find an effective scattering length that is of the same order of magnitude as that needed to describe the free scattering data @xcite .
however , the error bars are large and our data are primarily sensitive to @xmath37 rather than its phase .
it is important to stress that we do not know the relative strengths for the production of isospin @xmath18 and @xmath19 pairs in the reaction and hence the relative weights of @xmath20 and @xmath78 .
some information on this might come from looking at the data on the @xmath107 and @xmath108 reactions taken at 2.83gev @xcite .
however , even this would not give access to the phase between the @xmath18 and @xmath19 amplitudes .
it must also be stressed that we have neglected any energy dependence of the parameter @xmath35 .
another important _ caveat _ is that we have not considered any possible interaction of the @xmath47 with the protons .
although there is no evidence for any strong effect of this nature in the @xmath109 reaction @xcite , since the @xmath1 interaction is repulsive , its neglect might be interpreted as extra attraction in the system .
we have clearly shown the importance of the interaction , which is possibly related to the @xmath26 or the @xmath110 , but we must now ask whether there is any evidence for effects that might be connected with the production of the @xmath105 resonances .
the simulation of the energy dependence of the total cross section for the production of the non@xmath0 component in the reaction underestimates the cosy11 data very near threshold .
even more intriguing , and possibly connected , are the lowest mass points in the @xmath12 spectra . at all three energies shown in fig .
[ spectrum_kk ] , these points lie much higher than the simulations and a similar behavior is visible in the disto results @xcite as well as in our data on the @xmath111 reaction @xcite .
the mass scale of the variation is not that of the widths of the scalar resonances , which are quite large .
it is tempting to suggest that this structure might be due to the opening of the @xmath62 channel at a mass of @xmath112mev / c@xmath17 , which induces some cusp structure that changes the energy dependence of the total cross section near threshold @xcite .
this would require a very strong @xmath113 channel coupling , which might be driven by the @xmath105 resonances .
thus , although the reaction may not be ideal for investigating the properties of scalar states , their indirect effects might still be crucial .
given that the @xmath2 is strongly attracted to protons , it is natural to expect that similar _
effects should exist for other reactions .
the @xmath114 ratio in the @xmath115 reaction does indeed show a very strong enhancement near the @xmath116 threshold @xcite .
the interpretation is , however , slightly more complex because spin parity constraints means that there must be at least one @xmath27wave in the final state , even at low energies .
unfortunately , the momo data on @xmath117he @xcite can not be used to isolate any @xmath118he interaction because the charges of the kaons were not identified and only averaged spectra could be studied @xcite .
there is currently an overall uncertainty in our values of the cross sections because of the limitation in the proton proton elastic scattering database in our region of energy and angle , though we hope that this will be alleviated by future measurements @xcite . with our new estimates for the luminosity ,
the @xmath0production cross sections are reduced compared to those of our first analysis @xcite , which relied upon the non
renormalized 2004 said solution @xcite .
furthermore , the better separation of the @xmath0 and non@xmath0 events goes by chance in the same direction .
these two effects lower the ratio of @xmath0 to @xmath119 production in proton proton collisions near threshold so that it is now only about a factor of six above the okubo zweig
iizuka limit @xcite .
we would like to thank the cosy machine crew for their support as well as that of other members of the anke collaboration .
discussions with j. haidenbauer , c. hanhart and a. sibirtsev were very helpful and those with i. strakovsksy provided important clarifications on the uncertainties in the said @xmath40 procedure above 2.5gev .
r. niler and u.g .
meiner kindly provided us with the predictions of ref .
@xcite in numerical form .
correspondence with r. bertini and p. winter is gratefully acknowledged .
this work has been supported in part by the bmbf , dfg , russian academy of sciences , and cosy ffe .
when two particles interact strongly in the final state , the resulting matrix element involves an average of a production operator with the relative wave function @xmath120 of the strongly interacting pair .
if the interaction is of very short range , the wave function may be evaluated at the origin to leave an enhancement factor @xmath121 where @xmath122 is the relative momentum in the pair and @xmath123 is the @xmath124wave jost function @xcite . in the commonly used scattering length approximation ,
one retains only the linear term in @xmath63 , in which case @xmath125 , where @xmath35 is the scattering length of the interacting pair . on the other hand ,
if two or more pairs of particles interact in the final state , there is no reliable prescription to evaluate an analogous enhancement factor since a three body equation then needs to be solved .
if we denote the interacting pairs as 12 and 13 , the corresponding wave function will be @xmath126 .
we now make the _ ad hoc _ assumption that this wave function factorizes in the form @xmath127 in this case , the three body enhancement factor is simply the product of the two body factors , evaluated at the appropriate relative momenta : @xmath128 it must be stressed that eq .
( [ a3 ] ) is merely an _
ansatz _ to try to understand our ensemble of data and , even if it provides a satisfactory description of these , this does not mean that it is applicable more generally .
however , it has also been used to treat the @xmath129 reaction near threshold , where all the final pairs of particles interact strongly @xcite .
the approach does retain the necessary pole structure when @xmath130 and similarly for @xmath131 .
thus particle-1 can interact simultaneously with both 2 and 3 .
furthermore , to lowest order in the momenta , the _ ansatz _ corresponds to the scattering length of particle1 from a composite 2 + 3 with the desired combined scattering length of @xmath132 . although the @xmath124wave @xmath40 interaction is well known , the interaction is far more complex because of the channel couplings to @xmath133 and also because there are two isospins @xmath18 and @xmath19 . in a recent study , within a chiral su(3 )
unitary approach @xcite , data leading to a variety of channels have been fit and values of the ( complex ) scattering lengths deduced for @xmath18 and @xmath19 .
their _ full model _ resulted in @xmath134 these values describe the rapid drop in the elastic amplitudes from threshold . if isospin breaking induced by the significant mass differences is neglected , the scattering length @xmath135 .
however , this is not necessarily the parameter that is relevant for the analysis of the @xmath136 because it is not clear whether the reaction mechanism preferentially excites @xmath18 or @xmath19 @xmath137 states or some combination thereof .
the values in eq .
( [ a4 ] ) should therefore be considered merely as order of magnitude estimates of what might be expected when making fits to our data .
99 f. balestra _ et al .
_ , phys .
c * 63 * , 024004 ( 2001 ) .
m. hartmann _ et al_. , phys .
96 * , 242301 ( 2006 ) .
y. maeda _ et al_. , phys .
97 * , 142301 ( 2006 ) .
yao _ et al .
_ , j. phys .
g * 33 * , 1 ( 2006 ) .
see the short note on scalar mesons by s. spannier and n.a trnquist in s. eidelman _ et al .
_ , phys .
b * 592 * , 506 ( 2004 ) .
j. weinstein and n. isgur , phys .
d * 41 * , 2236 ( 1990 ) .
v. kleber _ et al .
_ , phys .
* 91 * , 172304 ( 2003 ) ; a. dzyuba _ et al .
_ , eur .
j. a * 29 * , 245 ( 2006 ) .
i. zychor _ et al .
_ , arxiv:0705.1039 [ nucl - ex][arxiv:0705.1039 [ nucl - ex ] ] .
p. winter _
et al . _ ,
b * 635 * , 23 ( 2006 ) .
s. barsov _ et al .
_ , nucl .
instrum .
a * 462 * , 364 ( 2001 ) .
r. maier _ et al .
_ , nucl .
instrum .
. phys .
res . , sect .
a * 390 * , 1 ( 1997 ) .
a. khoukaz _ et al .
_ , eur .
j. d * 5 * , 275 ( 1999 ) .
s. dymov _ et al .
_ , part .
* 2 * , 40 ( 2004 ) .
m. hartmann _ et al .
_ , int .
j. mod.phys .
a * 22 * , 317 ( 2007 ) .
m. bscher _ et al .
_ , nucl .
instrum .
meth.phys .
res . , sect .
a * 481 * , 378 ( 2002 ) .
s. agostinelli _ et al .
_ , nucl .
instrum .
a * 506 * , 250 ( 2003 ) ; http://geant4.web.cern.ch/geant4 .
v. bernard , n. kaiser , and u.g .
meiner , eur .
j. a * 4 * , 259 ( 1999 ) . m.l . goldberger and k.m .
watson , _ collision theory _ ,
( wiley , n.y .
, 1964 ) p. 544 .
b. borasoy , u.g .
meiner , and r. niler , phys .
c * 74 * , 055201 ( 2006 ) .
m. agnello _ et al .
_ , phys . rev .
lett.*94 * , 212303 ( 2005 ) .
t. yamazaki and y. akaishi , arxiv:0709.0630 [ nucl - th][arxiv:0709.0630 [ nucl - th ] ] .
shevchenko , a. gal , j. mare , j. rvai , arxiv:0706.4393 [ nucl - th][arxiv:0706.4393 [ nucl - th ] ] .
arndt , i.i .
strakovsky , and r.l .
workman , phys .
c * 62 * , 034005 ( 2000 ) ; r.a .
arndt , w.j .
briscoe , i.i .
strakovsky , and r.l .
workman , phys .
c * 76 * , 025209 ( 2007 ) ; said solution sp07 , available from http://gwdac.phys.gwu.edu .
d. albers _ et al .
_ , eur .
j. a * 2 * , 125 ( 2004 ) ; r.c .
kammerud _ et al .
_ , phys .
d * 4 * , 1309 ( 1971 ) ; s.a .
azimov _ et al .
_ , sov . phys .
jetp * 15 * , 299 ( 1962 ) ; i. ambats _ et al . _ ,
d * 9 * , 1179 ( 1974 ) ; t. fujii _ et al .
_ , phys .
* 128 * , 1836 ( 1962 ) ; w.m .
preston , r. wilson , and j.c .
street , phys .
rev.*118 * , 579 ( 1960 ) .
stein _ et al .
_ , _ in preparation_. m. wolke , phd thesis , university of mnster ( 1997 ) .
c. quentmeier _ et al .
_ , phys .
b * 515 * , 276 ( 2001 ) .
s. abd el - samad _ et al .
_ , phys .
b * 632 * , 27 ( 2006 ) .
w. oelert _ et al .
_ , int .
j. mod .
a * 22 * , 502 ( 2007 ) .
f. bellemann _ et al .
_ , phys .
c * 75 * , 015204 ( 2007 ) .
grishina , m. bscher , and l.a .
kondratyuk , phys .
c * 75 * , 015208 ( 2007 ) .
s.okubo , phys .
lett . * 5 * , 165 ( 1963 ) ; g.zweig , cern report th-*401 * ( 1964 ) ; j.iizuka , prog .
suppl . * 38 * , 21 ( 1966 ) . | the differential and total cross sections for kaon pair production in the reaction have been measured at three beam energies of 2.65 , 2.70 , and 2.83gev using the anke magnetic spectrometer at the cosy
jlich accelerator .
these near threshold data are separated into pairs arising from the decay of the @xmath0meson and the remainder . for the non@xmath0 selection ,
the ratio of the differential cross sections in terms of the and @xmath1 invariant masses is strongly peaked towards low masses .
this effect can be described quantitatively by using a simple _ ansatz _ for the final state interaction , where it is seen that the data are sensitive to the magnitude of an effective scattering length . when allowance is made for a small number of @xmath0 events where the @xmath2 rescatters from the proton , the @xmath0 region is equally well described at all three energies .
a very similar phenomenon is discovered in the ratio of the cross sections as functions of the @xmath3 and @xmath4 invariant masses and the identical final state interaction model is also very successful here .
the world data on the energy dependence of the non@xmath0 total cross section is also reproduced , except possibly for the results closest to threshold . |
the large magellanic cloud ( lmc ) is a low - metalicity irregular dwarf galaxy and a satellite of the milky way ( mw ) .
since the metallicity of the lmc ( which is only @xmath141/4 of that of the mw ; @xcite ) is similar to that of galaxies at red shifts @xmath15 @xcite , it offers opportunities to study the dust properties in distant low - metallicity extragalactic environments by studying the extinction properties of the lmc . the wavelength dependence of interstellar extinction
`` interstellar extinction law ( or curve ) '' is one of the primary sources of information about the interstellar grain population @xcite . in the mw , the interstellar extinction laws in the ultraviolet ( uv ) and visual wavelength ranges vary from sightline to sightline , and can be characterized by the optical total - to - selective extinction ratio @xmath16 @xcite , where @xmath17 is the interstellar reddening , @xmath18 is the extinction at the visual ( @xmath19 ; @xmath20 ) band , and @xmath21 is the extinction at the blue ( @xmath22 ; @xmath23 ) band .
the regional variations of the uv / visual extinction curves ( i.e. , the variations of the @xmath24 values ) reflect the variations in dust size : larger @xmath24 values indicate the predominance of larger grains @xcite .
however , the infrared ( ir ) interstellar extinction laws of the mw , which also vary from one sightline to another , can not be simply represented by the single @xmath24 parameter .
many recent studies show that there does not exist a `` universal '' near - infrared ( nir ) extinction law for the mw @xcite . moreover , the observationally - determined mid - infrared ( mir ) extinction law shows a flat curve while classical dust models for the diffuse interstellar medium ( ism ) predict a much steeper curve , with a pronounced minimum at @xmath147@xmath1 @xcite . and by `` mir '' we mean @xmath25 . ] due to its low metallicity , the dust quantity ( relative to h ) in the lmc is expected to be lower than that of the mw because there is less raw material ( i.e. , heavy elements ) available for making the dust .
the ( relative ) lack of the dust - making raw material could prevent the dust in the lmc from growing and hence the dust in the lmc may be smaller than the mw dust . furthermore , the star - formation activity in the lmc could destroy the dust .
therefore , one would naturally expect the dust size distribution and extinction curve in the lmc to differ from that of the mw .
as illustrated in figure [ fig : mwlmcsmc ] , the galactic interstellar extinction curve rises from the nir to the near - uv with a broad absorption bump at about @xmath26 ( @xmath27 , and continues rising steeply into the far - uv @xcite . in the small magellanic cloud ( smc )
, the extinction curves of most sightlines display a nearly linear steep rise with @xmath28 and lack the 2175@xmath29 hump @xcite .
the lmc extinction curve is intermediate between that of the mw and that of the smc : compared to the galactic extinction curve , the lmc extinction curve is characterized by a _ weaker _ 2175@xmath29 hump and a _ stronger _ far - uv rise @xcite .
strong regional variations in extinction properties have also been found in the lmc @xcite : the sightlines toward the stars inside or near the supergiant shell , lmc2 , which lies on the southeast side of the 30 doradus star - forming region , have a very weak 2175@xmath29 hump @xcite , while the extinction curves for the sightlines toward the stars which are @xmath30500pc away from the 30 doradus region are closer to the galactic extinction curve . @xcite
estimated @xmath31 for the lmc2 supershell and @xmath32 for the lmc as a whole .
the sightlines outside 30 doradus have @xmath33 @xcite .
@xcite had already noticed that the average value of @xmath24 in the lmc ( @xmath34 ) is close ( within 10% ) to that of the mw ( @xmath35 .
we note that the ccm parameterization controlled by the single @xmath24-parameter is not valid for the lmc extinction curve @xcite , i.e. , even the lmc and the mw are close in @xmath24 , their extinction curves differ appreciably .
@xcite derived the nir extinction law of the lmc based on the nir photometry at the @xmath36 , @xmath37 , and @xmath38 bands of early type supergiants .
he obtained @xmath39 , @xmath40 , and @xmath41 , and argued that the nir extinction law of the lmc is very similar to that of the mw .
, @xmath42 , and @xmath43 @xcite , or @xmath44 , @xmath45 , and @xmath46 @xcite . ]
their results correspond to @xmath47 when converted to the 2mass photmetric system @xcite . because of this , the galactic nir extinction law is sometimes adopted for the lmc ( e.g. , @xcite ) . using the near - infrared color - excess ( nice ) method @xcite , @xcite calculated the nir extinction coefficients of the lmc to be @xmath48 and @xmath49 , corresponding to @xmath50 , which is roughly consistent with that of @xcite .
we note that although the lmc nir extinction law is commonly assumed to be universal , @xcite showed that it also differs between the lmc2 supershell and the lmc average ( see their table 4 ) .
so far , little efforts have been put into the mir extinction properties of the lmc due to the paucity of mir data .
nevertheless , the mir extinction is not negligible : the lmc optical extinction @xmath51 @xcite would imply an appreciable amount of mir extinction @xmath6}{\lesssim}0.35{\,{\rm mag}}$ ] if we take the galactic ratio of @xmath6}/{{a_v}}\sim 0.07 $ ] @xcite . with the advent of sensitive ir space facilities ( e.g. , _ iso _ and _ spitzer _ )
, the lmc has been mapped in the mir wavelength bands with high accuracy and this makes the exploration of the lmc mir extinction possible . in this work , based on the _ spitzer_/sage database ,
we probe the mir extinction of the lmc and its regional variations . in
2 we briefly describe the sage data used in this work . 3
presents the method adopted to derive the extinction , including the galactic foreground extinction correction , the lmc sightline selection , and the selection of extinction tracers .
4 reports the derived extinction law and the mean extinction of the lmc .
finally , we summarize our results in 5 .
the data used in this work were obtained through the _ spitzer_/sage legacy program , entitled `` _ _ spitzer _ _ survey of the large magellanic cloud : surveying the agents of a galaxy s evolution '' @xcite .
the sage legacy program mapped the lmc at two different epochs ( epochs 1 and 2 ) separated by three months , using the irac ( [ 3.6 ] , [ 4.5 ] , [ 5.8 ] , and [ 8.0]@xmath52 ) and mips ( [ 24 ] , [ 70 ] , and [ 160]@xmath52 ) instruments on board the _ spitzer _ space telescope .
the sage points source catalog , named sagelmccatalogirac , released in september 2009 , combined both epochs data bandmerged with 2mass and 6x2mass all - sky data @xcite .
the sagelmccatalogirac catalog is the sage catalog with the highest quality , providing @xmath146.4 million point sources at four irac bands and three 2mass or 6x2mass bands .
faint limits are 18.1 , 17.5 , 15.3 , and 14.2@xmath53 for irac [ 3.6 ] , [ 4.5 ] , [ 5.8 ] , and [ 8.0]@xmath1 , respectively .
the sage data also provide the catalogs , such as sagelmccatalogmips24 , sagelmccatalogmips70 and sagelmccatalogmips160 , which include the sources observed by _ spitzer_/mips at [ 24 ] , [ 70 ] , and [ 160]@xmath52 bands .
however , only irac data are used in this work for probing the lmc mir extinction .
we only select the sources with a signal - to - noise ratio of s / n@xmath545 at all three 2mass bands and four irac bands ( see [ tracer ] for details on sample selection ) .
table [ tab : obserr ] shows the mean photometric errors for each band in the sagelmccatalogirac with @xmath55 . because of the high sensitivity of _
spitzer_/irac , there are @xmath56 sources in sagelmccatalogirac with s / n@xmath545 at all seven bands ( see figure [ fig : cmd ] ) .
additionally , @xcite estimated that the galactic foreground stars and background galaxies contribute roughly 18% and 12% of the sage catalog sources .
toward the lmc , many studies have showed that the mean foreground galactic reddening is @xmath57 @xcite .
@xcite estimated the average galactic extinction across the lmc at the visual band to be @xmath58 , implying @xmath57 with the galactic average value of @xmath59 .
may vary from @xmath140.01@xmath53 to @xmath140.14@xmath53 toward different parts of the lmc .
@xcite found the foreground reddening @xmath60 toward 30 doradus , while the inner reddening in this area exceeds 0.14@xmath53 . ]
we correct for the foreground extinction at each 2mass band , taking the foreground visual extinction to be @xmath61 @xcite for all the lmc sources . to see how much the extinction will change using different foreground extinction ,
we re - calculated the extinction toward co-186 ( 30 doradus ) . if we take @xmath62 , the upper limit of the foreground reddening @xcite . with @xmath63 and @xmath59 , the foreground extinction at the ir bands
are @xmath64 , @xmath65 , @xmath66 , @xmath6}\approx0.030 $ ] , @xmath7}\approx0.027 $ ] , @xmath8}\approx0.024 $ ] , and @xmath9}\approx0.026{\,{\rm mag}}$ ] .
taking these foreground extinction quantities to correct the observed magnitude , there is little change to the derived lmc nir extinction ( compared to that with @xmath61 ) : the difference is @xmath672.1% and @xmath672.3% for @xmath68 and @xmath69 , respectively , while the lmc mir extinction @xmath70 at [ 3.6]@xmath1 , [ 5.8]@xmath1 , and [ 8.0]@xmath1 would decrease by @xmath1416% , 24% , and 33% , respectively .
if neglecting the foreground reddening ( i.e. , taking @xmath71 ) , the derived nir extinction also changes little ( @xmath671.3% , @xmath670.7% ) , while the mir extinction at [ 3.6]@xmath1 , [ 5.8]@xmath1 , and [ 8.0]@xmath1 would increase by @xmath140.8% , 6.4% , and 9.2% , respectively .
the extinction at the [ 4.5]@xmath1 band is more complicated * ( see [ ice45])*. ] we take the wavelength - dependence of the foreground extinction to be that of the @xmath59 galactic average extinction law @xcite , i.e. , @xmath72 , @xmath73 , and @xmath74 .
the mean foreground extinction at each 2mass band are thus @xmath75 , @xmath76 , and @xmath77 .
the mean foreground extinction derived from the extinction map of @xcite are @xmath78 , @xmath76 , and @xmath79 @xcite . for the four irac bands ,
the foreground extinction is also corrected by applying the galactic mir extinction of @xmath6}/a_{{{\rm k_{s } } } } \approx0.63 $ ] , @xmath7}/a_{{{\rm k_{s } } } } \approx0.57 $ ] , @xmath8}/a_{{{\rm k_{s } } } } \approx0.49 $ ] , and @xmath9}/a_{{{\rm k_{s } } } } \approx0.55 $ ] , which were obtained for 131 _ spitzer_/glipmse fields along the galactic plane within @xmath80 with red giants and red clump giants ( rcgs ) as tracers @xcite .
therefore , the mean foreground extinction at the four irac bands are @xmath6}\approx0.014 $ ] , @xmath7}\approx0.013 $ ] , @xmath8 } \approx0.011 $ ] , and @xmath9 } \approx0.012$]@xmath53 , respectively .
as mentioned in 1 the lmc uv / visual extinction exhibits substantial regional variation , particularly among sightlines toward stars in and outside the 30 doradus region .
little is known about the mir extinction of the lmc and its variation toward different sightlines , until the _ spitzer_/sage database offers us the unique opportunity . as
the mir extinction is generally small , to probe the mir extinction efficiently we favor the lmc regions with large @xmath18 ( after all , only these highly obscured regions need to be corrected for ir extinction ) .
previous studies have estimated that the mean lmc reddening varies from @xmath81 to @xmath82 ( see table 2 in @xcite ) .
@xcite constructed the visual extinction ( @xmath18 ) map and found the mean value of @xmath18 is @xmath140.38@xmath53 .
@xcite constructed extinction maps for two stellar populations in the central 64 @xmath83 area of the lmc , and derived the average extinction of @xmath84 and @xmath85 for the cold and hot populations , respectively .
@xcite developed a new method by using the color of the _ x _ percentile reddest stars and derived a new @xmath18 map of the lmc , which is similar to the integrated intensity of the co emission as observed by the nanten telescope @xcite .
all these extinction maps showed that the maximum @xmath18 of the lmc is close to @xmath145@xmath53 .
since @xmath86 ( the extinction at the @xmath2 band relative to @xmath18 ) , for @xmath87 @xmath88 is only @xmath140.1@xmath53 . the mean measurement uncertainty of the 2mass survey is @xmath890.109mag for the @xmath2 magnitude ( snr=1 ) .
therefore , it is necessary to only consider the regions with @xmath90 .
in addition , the lmc areas containing special structures ( such as bar or molecular ridge ) and some special interstellar environments also need to be considered in order to probe the variation of mir extinction among different interstellar environments .
@xcite investigated the mid- to far - ir emission of the lmc based on the _ cobe_/dirbe and _ iras _ sky survey atlas data . in their figure 12
, they illustrated the local structures in the molecular ridge ( mr ) and the co arc in the lmc .
these structures are associated with the co molecular clouds , as well as the highly obscured regions in the extinction map derived by @xcite .
combining @xcite s extinction map and @xcite s figure 12 with the nanten @xmath91co map @xcite , we select five regions with large @xmath18 in the lmc molecular clouds and mr and arc structures ( see the red boxes in figure [ fig : fields ] ; in the following we will call these five regions `` the first five regions '' ) . although the average @xmath18 is smaller than 1@xmath53 or even much smaller , the regions in the lmc bar and hi area ( marked as blue boxes in figure [ fig : fields ] ) are also considered for comparison .
all these seven regions are listed in table [ tab : field ] labeled with the structure name and the molecular cloud number .
the determination of dust extinction is most commonly made by comparing the flux densities of obscured and unobscured pair stars of the same spectral type @xcite . in this work
we adopt the `` color - excess '' method to obtain the extinction .
the `` color - excess '' method is widely applied to photometric data and can probe deeper than the spectrum - pair method . for doing this ,
a group of sources which have essentially the same intrinsic color indices ( or with very small scattering in the color indices ) are chosen .
this method calculates the ratio of the two color excesses which can be expressed as following @xmath92 where @xmath93 is the magnitude in the interested band @xmath94 ; @xmath95 is the magnitude in the reference band @xmath96 ( taken to be the @xmath2 band in this work ) ; @xmath97 is magnitude in the comparison band @xmath98 ( taken to be the j band in this work ) .
therefore , the extinction ratio of the @xmath99 band to the @xmath2 reference band is @xmath100 where @xmath101 is used to derive @xmath102 from @xmath103 and can be obtained by fitting the observed color indices @xmath104 and @xmath105 with a linear line @xcite .
the slope of the fitted linear line is @xmath106 .
this is a statistical method as it makes use of a large number of sources and reduces the risk of depending on any individual objects with large uncertainties in the determination of their intrinsic color indices . in figure
[ fig : fit ] , we show the samples of fitting for the 30 doradus region ( co-186 ) .
plots @xmath13 vs. @xmath12 .
the slope of the fitted linear line is @xmath107 . ] from eq.[slope ] and eq.[eq_ext ] , it is seen that the determination of @xmath108 ( i.e. , the ratio of the @xmath99-band extinction to the extinction of the reference band @xmath2 ) requires the knowledge of @xmath109 .
@xcite obtained @xmath110 for the lmc2 supershell near 30 doradus .
we adopt @xmath111 since the regions selected for this study are near 30 doradus ( co-186 ) ( see figure [ fig : fields ] ) . for the sake of comparison
, we also take the galatcic value of @xmath112 @xcite which is often adopted to study the lmc extinction @xcite . in our previous work of probing the variation of the mir extinction law in the galactic plane @xcite , red giant branch stars ( rgbs )
were used as the tracer to study the extinction law at the four irac bands . in the ir ,
red giants are appropriate tracers of interstellar extinction for the following reasons : ( i ) they have a narrow range of effective temperatures so that the scatter of their intrinsic color indices is small ; ( ii ) they are bright in the ir ( @xmath113 ) and remain visible even suffering large extinction and/or observed from a great distance , even in the lmc . in the lmc ,
rgb stars are one of the most prominent and well populated features in the color - magnitude diagram ( cmd ) of stellar populations with ages larger than @xmath141.52.0gyr @xcite .
@xcite performed a morphological analysis on the 2mass cmd of the lmc and distinguished different populations of stars in the lmc .
the populations were identified based on isochrone fitting and matching the theoretical cmd colors of known populations to the observed cmd source density ( see figure 3 and tables 2 , 3 of @xcite ) .
@xcite selected the sources in _ regions _ e , f , g , h , j , l , and part of _ region _ d of @xcite to derive the extinction map of the lmc .
they eliminated _ regions _ a , b , c , and i , which contain much foreground contamination .
although it is free of foreground contamination , _ region _ k is also eliminated because it consists of dusty agb stars whose large @xmath114 colors are due to circumstellar dust .
however , the sources considered by @xcite contain too many different stellar populations .
we note that _ region _ e of @xcite , located within @xmath115 and @xmath116 , covers the upper rgb and includes the tip of the rgb .
therefore , we choose a slightly expanded version of _ region _ e to select the rgb stars : the selected region is constrained to @xmath117 and @xmath118 ( see figures [ fig : cmd],[fig : rgb186 ] ) .
however , it should be noted that evolved red giants may have circumstellar dust shells which would cause circumstellar extinction and produce ir emission , hence affecting our understanding of their intrinsic color indices .
additionally , the selected region ( _ region _ e ) may contain other populations which will affect the extinction determination .
for the mw , astronomers commonly use @xmath119-[4.5]$]@xmath1200.6 and @xmath121-[8.0]$]@xmath1200.2 as the criteria to exclude the sources with ir excess such as pre - main - sequence stars and asymptotic giant branch ( agb ) stars @xcite . for the lmc , @xcite presented initial results on the epoch 1 data of the _ spitzer_/sage program for a region near n79 and n83 , which is near the southwest end of the lmc bar .
they adopt a simplified point - source classification to identify three candidate groups stars without dust , dusty evolved stars , and young stellar objects ( ysos ) on the mir color - color diagrams . in their figure 11 and figure 12 , @xcite showed the stars without dust are almost constrained to @xmath119 - [ 4.5 ] < 0.0 $ ] , @xmath121 - [ 8.0 ] < 0.5 $ ] .
therefore , we will also use @xmath119 - [ 4.5 ] < 0.0 $ ] and @xmath121 - [ 8.0 ] < 0.5 $ ] in order to reduce the contamination caused by ysos and agbs . @xcite .
the typical magnitudes and colors of background galaxies are @[email protected]@xmath53 and @xmath123 @xcite . ] in figures [ fig : fit ] and [ fig : rgb186 ] , green crosses show the sources with @xmath119 - [ 4.5 ] > 0.0 $ ] or @xmath121 - [ 8.0 ] > 0.5 $ ] in the selected _ region _ e. figure [ fig : rgb186 ] shows the nir cmd of the 30 doradus ( co-186 ) region , which is one of the selected parts in the lmc .
the selected rgb sources are denoted by red dots within the red trapezoid .
left panel shows all the sources with s / n@xmath541 at all three 2mass bands in the 30 doradus ( co-186 ) region from the _ spitzer_/sage irac catalog , while the right panel only shows the sources with s / n@xmath545 at all seven bands ( i.e. , three 2mass bands and four irac bands ) in the same region . the red clump giants ( rcgs ) are also often used as tracers to probe the ir extinction in the mw
. however , rcgs in the lmc are often too faint to be observed ( at least in large number ) .
the tip - rgbs in the lmc have @xmath124 @xcite and rgbs are within @xmath125 , while rcg stars have @xmath126 @xcite .
therefore , in this work only rgbs are considered as tracers to probe the mir extinction of the lmc .
@xcite derived the nir extinction coefficient of the lmc using the nice method @xcite and the nicer ( nice revised ) method @xcite .
the nicer technique relies on the mean color of the control group being characteristic of the intrinsic color of the field stars @xcite .
they took the mean color of the control group as the intrinsic color of their interested fields , i.e. , @xmath127 .
however , their control groups include the sources in _ regions _ e , f , g , h , j , l , and part of d of @xcite , so that the dispersion around the mean control color was larger than that of rgb stars .
based on the selected rgb stars , we fit the nir color - color diagrams of @xmath128 versus @xmath129 and @xmath13 versus @xmath129 with the idl robust linear fit procedure .
our results show that the average nir extinction coefficient for the first five lmc regions listed in table [ tab : field ] is consistent with previous studies , i.e. , @xmath3 and @xmath4 ( see table [ tab : ir_ext ] ) . in figure
[ fig : nir186 ] , we show the nir color - color diagrams and the fits for the 30 doradus region ( co-186 ) .
green crosses show the sources with @xmath119 - [ 4.5 ] > 0.0 $ ] or @xmath119 - [ 8.0 ] > 0.5 $ ] in the selected _ region _ e of 30 doradus .
it efficiently reduces the contamination of ysos and agbs , and prevents from distorting the fit by excluding these sources .
as far as individual parts of the lmc are concerned , the nir extinction coefficients vary among different regions of the lmc .
however , the variation is relatively small among the first five regions listed in table [ tab : field ] which have molecular clouds and for which @xmath18 is large . for comparison , table [ tab :
ir_ext ] also shows the nir extinction coefficients for the lmc bar and hi region . because of their small amounts of extinction , we believe that the results ( for the lmc bar and hi region ) are less certain ( compared to those for the first five regions listed in table [ tab : field ] ) and the ir extinction can not be detected efficiently . in table [ tab : ir_ext ] ,
the color ratios of @xmath70 are shown for the four irac bands .
based on these color ratios and taking @xmath130 @xcite , we calculate the relative extinction @xmath131 . in table
[ tab : mir_ext ] we show the extinction at each irac band ( relative to that of the @xmath2 band ) for the selected regions of the lmc .
we note that only the first five regions listed in table 1 are considered in obtaining the average mir extinction of the lmc ( although the mir extinction toward the lmc bar and hi region are also derived and tabulated in table [ tab : mir_ext ] ) . as a whole , with rgbs as the extinction tracers ,
the mean mir extinction ratios of the lmc are @xmath6}/a_{{\rm k_{s}}}\approx0.72\pm0.03 $ ] , @xmath7}/a_{{\rm k_{s}}}\approx0.94\pm0.03 $ ] , @xmath8}/a_{{\rm k_{s}}}\approx0.58\pm0.04 $ ] , and @xmath9}/a_{{\rm k_{s}}}\approx0.62\pm0.05 $ ] . in figure
[ fig : ext_curve ] , we show the mir extinction of the lmc together with the mw model extinction curves of @xmath59 and @xmath132 @xcite .
the red filled squares show the average mir extinction of the lmc at the four irac bands , while the red error bars show the maximum and minimum extinction of the five selected fields , demonstrating the regional variation of mir extinction in the lmc .
if the extinction at [ 4.5]@xmath1 is excluded ( see 4.3 ) , the extinction at [ 3.6 ] , [ 5.8 ] and [ 8.0]@xmath1 of the lmc consists a flat curve , close to that of the mw model extinction curve of @xmath133 , lacking the deep minimum around 7@xmath1 predicted from the @xmath59 model .
the extinction at [ 4.5]@xmath1 is well above the @xmath132 model extinction curve and also well exceeds the extinction at the other three irac bands .
as far as individual sightlines are concerned , the wavelength dependence of the lmc mir extinction @xmath131 also varies from one sightline to another ( see table [ tab : ir_ext ] and [ tab : mir_ext ] ) . in the literature the nir extinction law of the mw
has often been used to represent that of the lmc @xcite . for comparison
, we have also calculated the mir extinction of the lmc taking the galactic value of @xmath112 @xcite .
the results are also tabulated in table [ tab : mir_ext ] : the extinction is larger than that derived with the lmc2 supershell value of @xmath134 @xcite by @xmath148% , 2% , 17% , and 15% at [ 3.6 ] , [ 4.5 ] , [ 5.8 ] , and [ 8.0 ] , respectively .
as clearly seen in figure [ fig : ext_curve ] and table [ tab : mir_ext ] , the mean extinction of the 4.5@xmath1 irac band @xmath7}/a_{{{\rm k_{s}}}}\approx0.94\pm0.03 $ ] is not only much higher than that of the other three irac bands , but also much larger than that of the galactic plane average of @xmath7}/a_{{{\rm k_{s}}}}\approx0.63\pm0.01 $ ] derived from the rgbs tracers @xcite . in galactic and extragalactic dense clouds
various ice features have been observed , including the 3.1@xmath1 feature due to the o
h stretching mode of h@xmath135o ice as well as a number of weaker features at 4.27@xmath1 due to the c o stretching mode of co@xmath135 ice , at 4.67@xmath1 due to the c o stretching mode of co ice , and at 6.02@xmath1 due to the h o h bending mode of h@xmath135o ice . in this work , the selected sightlines are mostly located in molecular clouds , therefore , it is not unreasonable , at a first glance , to attribute the [ 4.5 ] excess extinction to @xmath136 and co ices in the lmc dense clouds .
observational , experimental and theoretical studies have shown that co and co@xmath135 ices could efficiently form both in quiescent clouds and in active star - forming regions ( e.g. , see @xcite ) .
@xcite found a threshold of @[email protected]@xmath1371.0 , [email protected]@xmath53 respectively for the detection of co@xmath135 and co ices in the taurus dark cloud complex ; for h@xmath135o ice , the detection threshold is @[email protected]@xmath1370.1@xmath53 @xcite . with @xmath138 for the selected lmc regions ,
it is unlikely for these regions to have co@xmath135/h@xmath135o or co / h@xmath135o much higher than that of the mw . to examine whether the ice ( particularly co@xmath135 and , to a less degree , co ) absorption bands could account for the [ 4.5 ] excess extinction , we approximate the 3.05 , 4.27 , 4.67 and 6.02@xmath1 bands respectively from h@xmath135o , co@xmath135 , co , and h@xmath135o ices as four drude profiles
. `` xcn '' absorption feature .
this feature is commonly seen toward luminous protostars embedded in dense molecular clouds .
its carrier lacks a specific identification , although it is generally believed that the 4.62@xmath1 feature arises from the c@xmath139n stretch of some sort of cn - bearing organic dust and the carrier may result from energetic processing ( e.g. , uv photolysis or ion bombardment ) of interstellar ice mixtures containing n in the form of nh@xmath140 or n@xmath135 @xcite .
the `` xcn '' abundance is small , with xcn / h@xmath135o@xmath1206% in high - mass young stellar objects which have the highest xcn abundance ( see @xcite ) . ]
the extinction due to these four ice bands is @xmath141 \times e_j \times 2\gamma_j/\pi } { \left(\lambda -\lambda_j^2/\lambda\right)^2+\gamma_j^2}\ ] ] where @xmath142 , @xmath143 , @xmath144 are the peak wavelength , fwhm , and strength of the @xmath145-th ice absorption band , respectively ( see table [ tab : ice ] ) ; and band strength @xmath144 values tabulated in table [ tab : ice ] are in units of @xmath146 and cmmolecule@xmath147 , respectively . when using eq.[eq : ice ] , they are converted so that they are in units of @xmath1 and @xmath148molecule@xmath147 .
] @xmath149 is the h@xmath135o ice column density ; @xmath150 is the abundance of the ice species @xmath151 relative to h@xmath135o ice in typical dense clouds ; @xmath152 is the `` enhancement '' factor for species @xmath151 ( i.e. , @xmath150 is increased by a factor of @xmath152 in order for the ice absorption bands to account for the [ 4.5 ] excess extinction ) . finally , we add @xmath153 to @xmath154 , the wd01 extinction curve ( with @xmath59 or 5.5 ) @xcite and then convolve with the _
spitzer_/irac filter response functions .
band , the addition of the ice extinction does not change the normalization of @xmath88 .
but in any case , we re - normalize the final extinction at @xmath2 . ]
the h@xmath135o ice abundance ( i.e. , @xmath149 ) is constrained not to exceed the irac [ 3.6 ] extinction . with @xmath155 and @xmath156 , typical for quiescent dense clouds @xcite , co and co@xmath135 are not capable of accounting for the [ 4.5 ] excess extinction .
if we are forced to attribute the [ 4.5 ] excess extinction to co and co@xmath135 , we will have too much h@xmath135o ice and the resulting extinction at [ 3.6 ] would be too high .
if we fix the h@xmath135o ice abundance at what is required by the [ 3.6 ] extinction , in order for co and co@xmath135 to explain the [ 4.5 ] excess extinction , we have to enhance their abundances ( relative to their typical abundances in dense clouds ) by a factor of @xmath157 for @xmath59 and @xmath158 for @xmath132 ( see figure [ fig : ice ] ) .
we note that although the extinction at [ 3.6 ] and [ 4.5 ] is well fitted by the @xmath59 model combined with ices , it could not account for the flat extinction at [ 5.8 ] and [ 8.0 ] .
in contrast , the @xmath132 model together with ices could closely explain the mir extinction at all four irac bands provided that the abundances of co and co@xmath135 ices are increased by a factor of @xmath146.5 from their typical abundances in dense clouds .
however , the required co and co@xmath135 abundances are unrealistically too high : @xmath159 and @xmath160 for @xmath132.o and co@xmath161 ices toward massive ysos in the lmc based on the akari lmc spectroscopic survey .
they estimated @xmath162 and @xmath163 ( see their table 4 ) . ] the origin of the [ 4.5 ] excess extinction remains unclear .
some red giants have circumstellar envelopes rich in co gas which absorbs at 4.6@xmath1 @xcite .
with red giants as a tracer of the mid - ir extinction , the 4.6@xmath1 absorption feature of their co gas would result in an overestimation of the irac [ 4.5 ] extinction .
it is worth exploring whether the co gas absorption of red giants could account for the [ 4.5 ] excess extinction . as discussed in
[ sec : cem ] , in deriving the ir extinction law , the @xmath12 vs. @xmath13 color - color diagram is fitted to obtain the slope @xmath106 ( see eq.[slope ] and eq.[eq_ext ] ) . since @xmath12 and @xmath13 both employ the @xmath2 band photometry and therefore may have correlated uncertainties , there could be systematic bias in determining @xmath106 and @xmath164 .
to evaluate the possible systematic bias , we perform simple monte - carlo simulations . using cmd 2.5 @xcite ,
we generate stellar isochrones and take a stellar atmosphere model of @xmath165 , @xmath166 , @xmath167 for the simulated stars .
the stellar absolute magnitudes are approximatly @xmath168 , @xmath169 , @xmath170 , @xmath171 , @xmath172 , @xmath173 , and @xmath174 in @xmath36 , @xmath37 , @xmath2 , [ 3.6 ] , [ 4.5 ] , [ 5.8 ] , and [ 8.0 ] bands , respectively .
we then apply a random amount of extinction to obscure a large number of stars .
the stars are taken to be identical and have the same absolute magnitudes .
let @xmath175 be the amount of extinction to which the stars are subject .
we take the stars to be obscured by @xmath175 , with @xmath175 being a random number between 0 and @xmath176 .
we take the @xmath36 band as the basis wavelength and assume the extinction at other wavelengths to follow either the mw @xmath59 model curve or the mw @xmath132 model curve of @xcite . the thick blue lines in figure [ fig : montecarlo ]
shows the simulated color indices for a large number of stars ( @xmath177 ) in the @xmath13 vs. @xmath12 color - color diagram .
the stars are obscured by the @xmath59-type extinction with @xmath178 .
we then add errors to the obscured stellar magnitudes based on the uncertainties of the sagelmccatalogirac ( see table [ tab : obserr ] ) .
the errors are simulated to follow a poisson distribution , with their mean values being the mean photometric uncertainties of the sources with @xmath179 in all seven bands in sagelmccatalogirac ( see figure [ fig : obserr ] for a comparison of the simulated errors with the mean photometric uncertainties ) . in figure
[ fig : montecarlo ] , the simulated , error - added sources are shown as black dots in the @xmath13 vs. @xmath12 color - color diagram .
the slope for the resulting @xmath13 vs. @xmath12 diagram ( black dots ) is fitted ( see the red lines in figure [ fig : montecarlo ] ) .
figure [ fig : montecarlo ] shows that the extinction laws derived from the obscured , error - added sources are different from the original ones adopted to obscure the stars : the derived slopes of @xmath101 , i.e. @xmath70 , are all smaller than the simulated ones , and therefore the derived @xmath102 ratios are overestimated . we have examined the effects of ( i ) the level of obscuration @xmath180 , ( ii ) the number of stars @xmath181 , ( iii ) the amount of error , and ( iv ) the extinction - type which is employed to obscure the stars . as shown in table [ tab : montecarlo ] , the level of obscuration @xmath180 dominates the systematic bias which decreases with the increasing of @xmath180 . for @xmath182 the systematic bias becomes negligible .
@xcite derived @xmath183 for the lines of sight toward the lmc2 supershell near 30 doradus . with @xmath184 ,
we obtain @xmath185 . therefore , the mir @xmath102 ratios may have been overestimated and the color ratios @xmath70 may have been underestimated . with @xmath177 , @xmath186 , and a @xmath59-type extinction ( see the fifth row in table [ tab : montecarlo ] ) , we perform a monte - carlo simulation to assess the possible systematic bias caused by the correlated uncertainties of @xmath12 and @xmath13 .
we find that the extinction ratios @xmath164 derived in [ sec : mir ] could be overestimated by @xmath146% , 10% , 16% and 9% at [ 3.6 ] , [ 4.5 ] , [ 5.8 ] and [ 8.0 ] , respectively .
the corrected extinction @xmath102 will be @xmath6}/a_{{\rm k_{s}}}\approx0.68\pm0.03 $ ] , @xmath7}/a_{{\rm k_{s}}}\approx0.84\pm0.03 $ ] , @xmath8}/a_{{\rm k_{s}}}\approx0.49\pm0.04 $ ] , and @xmath9}/a_{{\rm k_{s}}}\approx0.57\pm0.05 $ ] .
the _ spitzer_/sage irac catalog provides us a unique opportunity to study the mir extinction of the lmc .
we select five fields with large optical extinction ( @xmath187 ) to explore the mir extinction of the sightlines toward these regions .
taking @xmath111 @xcite and using rgb stars as tracers , we find that : 1 .
the nir color - color diagrams are fitted to derive the nir extinction coefficients : @xmath3 and @xmath4 , which are consistent with earlier studies @xcite .
the corresponding color ratios for the mw are @xmath188 and @xmath189 @xcite .
2 . at the four irac bands ,
the derived mean mir extinction of the lmc are @xmath6}/a_{{\rm k_{s}}}\approx0.72\pm0.03 $ ] , @xmath7}/a_{{\rm k_{s}}}\approx0.94\pm0.03 $ ] , @xmath8}/a_{{\rm k_{s}}}\approx0.58\pm0.04 $ ] , and @xmath9}/a_{{\rm k_{s}}}\approx0.62\pm0.05 $ ] .
the corresponding extinction ratios for the mw are @xmath6}/a_{{\rm k_{s}}}\approx0.63\pm0.01 $ ] , @xmath7}/a_{{\rm k_{s}}}\approx0.57\pm0.03 $ ] , @xmath8}/a_{{\rm k_{s}}}\approx0.49\pm0.03 $ ] , and @xmath9}/a_{{\rm k_{s}}}\approx0.55\pm0.03 $ ] @xcite .
the lmc extinction at [ 3.6 ] , [ 5.8 ] and [ 8.0]@xmath1 is consistent with a flat curve , close to that of the mw model extinction curve predicted by the interstellar grain model of @xmath133 .
similar to that of the mw , the lmc mir extinction law exhibits appreciable regional variations .
the extinction at [ 4.5 ] is much higher than that of the other three irac bands .
it can not be explained in terms of the 4.27@xmath1 absorption band of @xmath136 ice and the 4.67@xmath1 absorption band of co ice .
it may be caused by the 4.6@xmath1 absorption feature of co gas in the circumstellar envelopes of red giants which are used to trace the ir extinction .
the derived mir extinction @xmath164 may be overestimated because of the correlated uncertainties of @xmath12 and @xmath13 which could affect the determination of @xmath106 , the slope of the @xmath12 vs. @xmath13 color - color diagram . with this systematic bias taken into account
, the derived extinction ratios @xmath164 could be overestimated by @xmath146% , 10% , 16% and 9% at [ 3.6 ] , [ 4.5 ] , [ 5.8 ] and [ 8.0 ] , respectively .
cc|cccccc|rrrrrr @xmath181 & @xmath190 & @xmath191 & @xmath192 & @xmath193)}{e({{\rm j}}-{{\rm k_{s}}})}$ ] & @xmath194)}{e({{\rm j}}-{{\rm k_{s}}})}$ ] & @xmath195)}{e({{\rm j}}-{{\rm k_{s}}})}$ ] & @xmath196)}{e({{\rm j}}-{{\rm k_{s}}})}$ ] & @xmath197 & @xmath198 & @xmath199}}{a_{{{\rm k_{s}}}}}$ ] & @xmath200}}{a_{{{\rm k_{s}}}}}$ ] & @xmath201}}{a_{{{\rm k_{s}}}}}$ ] & @xmath202}}{a_{{{\rm k_{s}}}}}$ ] + & 1.693 & 2.693 & 0.404 & 0.498 & 0.556 & 0.539 & 2.494 & 1.554 & 0.396 & 0.256 & 0.169 & 0.195 + 1000 & 0.5 & 1.240 & 1.936 & 0.275 & 0.352 & 0.394 & 0.415 & 2.494 & 1.749 & 0.589 & 0.475 & 0.412 & 0.381 + 1000 & 1.0 & 1.541 & 2.407 & 0.367 & 0.456 & 0.510 & 0.509 & 2.494 & 1.601 & 0.452 & 0.319 & 0.238 & 0.240 + 1000 & 1.5 & 1.620 & 2.549 & 0.387 & 0.478 & 0.534 & 0.526 & 2.494 & 1.568 & 0.422 & 0.285 & 0.201 & 0.214 + 1000 & 5.0 & 1.684 & 2.675 & 0.402 & 0.496 & 0.554 & 0.538 & 2.494 & 1.542 & 0.399 & 0.259 & 0.172 & 0.196 + 1000 & 10.0 & 1.689 & 2.687 & 0.403 & 0.497 & 0.555 & 0.539 & 2.494 & 1.539 & 0.397 & 0.257 & 0.170 & 0.195 + 10000 & 1.5 & 1.612 & 2.545 & 0.385 & 0.479 & 0.533 & 0.520 & 2.494 & 1.569 & 0.425 & 0.284 & 0.204 & 0.222 + 100000 & 1.5 & 1.620 & 2.553 & 0.387 & 0.479 & 0.536 & 0.520 & 2.494 & 1.568 & 0.422 & 0.284 & 0.199 & 0.223 + & 2.222 & 3.222 & 0.276 & 0.352 & 0.414 & 0.421 & 2.449 & 1.450 & 0.600 & 0.490 & 0.400 & 0.389 + 1000 & 0.5 & 1.317 & 2.050 & 0.161 & 0.225 & 0.271 & 0.312 & 2.449 & 1.707 & 0.766 & 0.675 & 0.608 & 0.548 + 1000 & 1.0 & 1.884 & 2.729 & 0.243 & 0.314 & 0.374 & 0.395 & 2.449 & 1.531 & 0.648 & 0.545 & 0.458 & 0.427 + 1000 & 1.5 & 2.058 & 2.969 & 0.260 & 0.334 & 0.395 & 0.411 & 2.449 & 1.488 & 0.623 & 0.516 & 0.427 & 0.405 + 1000 & 5.0 & 2.204 & 3.193 & 0.274 & 0.350 & 0.412 & 0.422 & 2.449 & 1.454 & 0.602 & 0.493 & 0.403 & 0.389 + 1000 & 10.0 & 2.214 & 3.211 & 0.276 & 0.351 & 0.414 & 0.422 & 2.449 & 1.451 & 0.601 & 0.491 & 0.401 & 0.389 + 10000 & 1.5 & 2.051 & 2.965 & 0.259 & 0.335 & 0.393 & 0.405 & 2.449 & 1.489 & 0.625 & 0.515 & 0.430 & 0.413 + 100000 & 1.5 & 2.060 & 2.974 & 0.261 & 0.335 & 0.396 & 0.405 & 2.449 & 1.455 & 0.622 & 0.515 & 0.426 & 0.414 + comparison of the mw ( black solid ) , lmc ( red solid ) , and smc ( green dot - dashed ) extinction curves . also shown is the lmc2 supershell extinction curve ( blue dashed ) .
the mw extinction curve is calculated from the ccm parameterization with @xmath59 .
the lmc and smc extinction curves are taken from @xcite .
, width=480 ] the color - magnitude diagram ( cmd ) of the lmc .
left panel shows all the sources in the catalog of sagelmccatalogirac , while the right panel only shows the sources with s / n@xmath545 at all three 2mass bands and four irac bands .
the red trapezium shows the region of rgbs , which are candidates for extinction tracers .
, title="fig:",width=345 ] the color - magnitude diagram ( cmd ) of the lmc .
left panel shows all the sources in the catalog of sagelmccatalogirac , while the right panel only shows the sources with s / n@xmath545 at all three 2mass bands and four irac bands .
the red trapezium shows the region of rgbs , which are candidates for extinction tracers .
, title="fig:",width=345 ] ( see eq.[eq_ext ] for @xmath101 , which can be used to derive @xmath102 ) .
red and blue dots show all rgb sources selected in the 30 doradus region ( co-186 ) using the method described in
[ tracer ] , while green crosses show the sources in the _ region _ e with @xmath119-[4.5 ] > 0 $ ] or @xmath121-[8.0 ] > 0.5 $ ] .
only red dots are the rgb sources used in the final fitting after applying the @xmath203 principle .
black dots show all the sources with s / n@xmath545 in all seven ( 2mass and irac ) bands in this region .
[ fig : fit],width=691 ] 1 at all three 2mass bands in the 30 doradus ( co-186 ) region from the _ spitzer_/sage irac catalog , while the right panel only shows the sources with s / n@xmath545 at all seven bands ( three 2mass bands and four irac bands ) in the same region .
green crosses show the sources with @xmath119 - [ 4.5 ] > 0.0 $ ] or @xmath119 - [ 8.0 ] > 0.5 $ ] in the selected region .
[ fig : rgb186],title="fig:",width=345 ] 1 at all three 2mass bands in the 30 doradus ( co-186 ) region from the _ spitzer_/sage irac catalog , while the right panel only shows the sources with s / n@xmath545 at all seven bands ( three 2mass bands and four irac bands ) in the same region .
green crosses show the sources with @xmath119 - [ 4.5 ] > 0.0 $ ] or @xmath119 - [ 8.0 ] > 0.5 $ ] in the selected region .
[ fig : rgb186],title="fig:",width=345 ] . for this region ,
the nir color ratios are @xmath3 and @xmath4 , which are shown as black solid lines . in the mw ,
these values are @xmath204 and @xmath189 @xcite , which are shown as black dashed lines .
[ fig : nir186],title="fig:",width=316 ]
. for this region , the nir color ratios are @xmath3 and @xmath4 , which are shown as black solid lines . in the mw , these values are @xmath204 and @xmath189 @xcite , which are shown as black dashed lines . [
fig : nir186],title="fig:",width=316 ] and @xmath132 models ( black dotted lines ) combined with the 3.05 and 6.02@xmath1 absorption bands of h@xmath135o ice , the 4.27@xmath1 band of co@xmath135 ice , and the 4.67@xmath1 band of co ice ( blue solid lines ) .
blue filled triangles are the @xmath59 or @xmath132 model curve plus the ice absorption bands convolved with the _
spitzer_/irac filter functions .
( a , b ) : the abundances of co and co@xmath135 ices are taken to be that of typical dense clouds ( i.e. , @xmath205 , @xmath206 ) .
bottom panels ( c , d ) : the abundances of co and co@xmath135 ices are enhanced relative to their typical values in dense clouds . ,
width=460 ] , @xmath59 , @xmath207 , and the mean photometric uncertainties for the sources with @xmath179 in all three 2mass bands and four irac bands .
the thick blue lines show the simulated colors of the obscured sources ( but no errors are added to these sources ) .
the black dots show the colors of the simulated , obscured , error - added sources .
the red lines fit the @xmath13 vs. @xmath12 color - color diagram for the simulated , obscured , error - added sources .
[ fig : montecarlo],width=480 ] | based on the photometric data from the _ spitzer_/sage survey and with red giants as the extinction tracers , the mid - infrared ( mir ) extinction laws in the large magellanic cloud ( lmc ) are derived for the first time in the form of @xmath0 , the extinction in the four irac bands ( i.e. , [ 3.6 ] , [ 4.5 ] , [ 5.8 ] and [ 8.0]@xmath1 ) relative to the 2mass @xmath2 band at 2.16@xmath1 .
we obtain the near - infrared ( nir ) extinction coefficient to be @xmath3 and @xmath4 .
the wavelength dependence of the mir extinction @xmath5 in the lmc varies from one sightline to another .
the overall mean mir extinction is @xmath6}/a_{{\rm k_{s}}}\approx0.72\pm0.03 $ ] , @xmath7}/a_{{\rm k_{s}}}\approx0.94\pm0.03 $ ] , @xmath8}/a_{{\rm k_{s}}}\approx0.58\pm0.04 $ ] , and @xmath9}/a_{{\rm k_{s}}}\approx0.62\pm0.05 $ ] . except for the extinction in the irac [ 4.5 ] band which may be contaminated by the 4.6@xmath1 co gas absorption of red giants ( which are used to trace the lmc extinction ) ,
the extinction in the other three irac bands show a flat curve , close to the milky way @xmath10 model extinction curve ( where @xmath11 is the optical total - to - selective extinction ratio ) .
the possible systematic bias caused by the correlated uncertainties of @xmath12 and @xmath13 is explored in terms of monte - carlo simulations .
it is found that this could lead to an overestimation of @xmath5 in the mir . |
during the rising phase of the solar cycle , it is becoming more important to understand the physics of the near - earth space . the dynamical phenomena caused by the constant flow of magnetized collisionless plasma from the sun
creates space weather that may have harmful effects on space - borne or ground - based technological systems or on humans in space . while the physics of space weather is being studied with in situ instruments ( e.g. nasa s radiation belt storm probes launched in 2012 - 08 - 30 ) and by means of remote sensing , it is also important to model the near - earth space with numerical simulations .
the simulations can be used both as a context to the one - dimensional data sets from obsevations , as well as a source to discover new physical mechanisms behind observed variations .
present large scale ( global ) simulations are based on computationally light - weight simplified descriptions of plasma , such as magnetohydrodynamics ( mhd , @xcite , @xcite , @xcite and @xcite ) . on the other hand
the complexity and range of spatial scales ( from less than @xmath1 to over @xmath0 km ) in space weather physics signifies the need to incorporate particle kinetic effects in the modeled equation set in order to better model , for example , magnetic reconnection , wave - particle interactions , shock acceleration of particles , ring current , radiation belt dynamics and charge exchange ( see e.g. @xcite for an overview ) .
however , as one goes from mhd towards the full kinetic description of plasma ( from hybrid pic @xcite and vlasov @xcite to full pic @xcite , @xcite ) , the computational demands increase rapidly , indicating that the latest high performance computing techniques need to be incorporated in the design of new simulation architectures .
as the number of cores in the fastest supercomputers increases exponentially the parallel performance of simulations on distributed memory machines is becoming crucial . on the other hand , utilizing a large number of cores efficiently in parallel is challenging especially in simulations using run - time adaptive mesh refinement ( amr ) .
this is largely a data structure and an algorithm problem albeit specific to massively parallel physical simulations running on distributed memory machines . in computer simulations dealing with , for example , continuous matter ( a fluid )
the simulated domain is discretized into a set of points or finite volumes which we will refer to as cells . at any given cell
the numerical solution of a differential equation describing the problem often depends only on data within a ( small ) part of the simulated volume .
this is true for a single time step in a solver for a hyperbolic problem or a single iteration in a solver for an elliptic problem .
this spatial data dependency can be implemented implicitly in the solver function(s ) or explicitly as a separate grid library used by the application . in a simple case
the number of cells in the simulation stays constant and the data dependency of each cell is identical allowing cell data to be stored in an array whose size is determined at grid creation and the spatial neighbors to be represented as indices into this array .
a straightforward amr extension of this concept is to create additional nested grids in specific parts of the simulation domain with higher resolution . by solving each grid separately and interpolating the results from finer grids into coarser grids
one does not have to modify the solver functions .
this technique is used extensively for example by berger ( see @xcite for some of the earliest work ) and by @xcite and @xcite . in the rest of this work
however we will concentrate on amr implementations in which additional overlapping grids are not created but instead cells of the initial grid are refined , i.e. replaced with multiple smaller cells
. a generic unstructured grid ( as provided for example by libmesh @xcite ) does not admit as simple a description as above and is generally described by a directed graph in which vertices represent simulation cells and directed edges represent the data dependencies between cells .
unfortunately the nomenclature of graph theory and geometry overlap to some extent and discussing both topics simulataneously can lead to confusion .
figure [ fig : nomenclature ] shows the nomenclature we use from this point forward , the standard graph theoretical terms are given in parentheses for reference . a cell is a natural unit in simulations using the finite volume method ( fvm ) and hereinafter we will use the term cell instead of vertex when discussing graphs .
also an edge in fvm simulations usually refers to the edges of a cube representing the physical volume of a cell , and hence we will use the term arrow to refer to a directed edge in a graph . furthermore we note that each cell in the grid can also represent , for example , a block of cells similarly to @xcite , but for the purposes of this work the actual data stored in grid cells is largely irrelevant .
since a graph can also be used to represent the cells and arrows of grids simpler than an unstructured mesh , the question arises how does a particular program implement its graph representation of the simulated system , e.g. what simplifying assumptions have been made and how is the graph represented in memory .
a popular representation in ( m)hd amr simulations is to have a fixed number of arrows directed away from each source cell and to store the arrows as native pointers to the destination cells . in case
a cell does not exist all arrows pointing to it are invalidated in neighboring cells .
this technique has been used with different variations by @xcite , @xcite , @xcite and @xcite , for example .
there are several possibilities for representing the cells and arrows of a graph , for example an adjacency list or an adjacency matrix @xcite . in physical simulations
the number of arrows in the graph is usually of the same order as the number of cells in which case a suitable representation is an adjacency list . in an adjacency list
the cells of the graph are separate objects and each cell stores the arrows pointing to and/or from that cell .
the cells of the graph and the arrows of each cell can be stored in different types of data structures .
for example the cells are stored in a contiguous array ( representing a linear octree ) in @xcite , @xcite , @xcite and @xcite , a hash table in @xcite and a ( doubly ) linked list in @xcite . on the other hand the arrows of each cell are stored in a fixed size array of native pointers in @xcite and as single bits in @xcite . in this work
we introduce the distributed cartesian cell - refinable grid ( dccrg ) for rapid development of parallel simulations using , for example , finite volume or finite element methods ( fem ) . in dccrg the graph is represented by an adjacency list in which cells are stored into a hash table , while the arrow lists of cells are stored into contiguous arrays .
we describe the details of the graph representation in section [ sec : hash_table ] . in section
[ sec : implementation ] we describe the c++ implementation of dccrg and present its unique features with respect to other published grid codes : arbitrary data in grid cells , transparent updates of remote neighbor data , user - selectable neighborhood size for cells and ease of use . in section [ sec : tests ] we test the scalability of dccrg using a variety tests in one , two and three dimensions and draw our conclusions in section [ sec : discussion ] .
dccrg represents the grid as an adjacency list in which cells are stored into a hash table .
a hash table has one clear advantage over a tree when used to store the grid : cells can be accessed , inserted and deleted in constant amortized time regardless of the number of cells and their physical size and location .
thus neither the total number of cells nor the number of refinement levels affect the simulating performance of a single core .
each cell is associated with a unique i d which we use as a key into the hash table .
a potential drawback of a hash table is the computational cost of the hash function , but according to our tests the cost is usually not important .
the time to solve one flux between two cells in the mhd tests presented in section [ sec : mhd_tests ] is about four times larger than accessing one random cell in the hash table .
the cell access time can be optimized further , for example , by storing and solving blocks of cells instead of single cells as is done in @xcite and discussed further in section [ sec : arbitrary_data ] .
our cell ids are globally unique integers which offers several advantages : 1 ) the cell i d can be calculated locally , i.e. without communication with other processes , 2 ) the neighbors of a cell can be stored as cell ids instead of pointers that are not consistent across computing nodes , 3 ) the cost of computing the hash function values is minimized , 4 ) the memory required for storing the cell ids is small .
figure [ fig : example_grid ] shows an example of mapping cells to unique ids which is done as follows : cell ids within each refinement level increase monotonically first in x coordinate , then in y and then in z , with cells at refinement level 0 represented by numbers from 1 to @xmath2 , refinement level 1 by numbers from @xmath3 to @xmath4 , etc .
cells at refinement level @xmath5 are half the size of cells at refinement level @xmath6 in each dimension
. cells at equal refinement level are identical in size . hence in three dimensions @xmath7 , where @xmath8 and @xmath9 are the grid size in cells of refinement level 0 in the x , y and z dimensions respectively .
hereinafter cell size refers to the logical size of cells assuming a homogeneous and isotropic cartesian geometry .
when searching for the neighbors of a cell in the hash table ( see section [ sec : neigh_search ] ) it is convenient to use the concept of cell indices : the location of each cell in the grid is represented by one number per dimension in the interval @xmath10 $ ] where l is the maximum refinement level of the grid and @xmath11 is @xmath12 , @xmath13 or @xmath9 respectively .
figure [ fig : example_grid ] shows the possible cell indices for an example grid with @xmath14 and @xmath15 .
the size of a single cell of refinement level @xmath6 is @xmath16 indices in each dimension .
a cell spanning more than one index is considered to be located at indices closest to the origin of the grid , for example cell # 3 in figure [ fig : example_grid ] is located at indices ( 2 , 0 , 0 ) .
similarly to @xcite there is a one - to - one mapping between cell ids and cell indices plus refinement levels , e.g. in addition to its i d a cell can be uniquely identified by its indices and refinement level . in the current implementation of dccrg a cell
is refined by creating all of its children ; in the example grid of figure [ fig : example_grid ] refining cell # 1 would create cells # 2 ... #9 . in principle this is not required and more complex grid structures are possible in which , for example , the grid in figure [ fig : example_grid ] would consist of cells # 1 and # 3 alone .
such an approach has been found useful by @xcite .
complete refinement of cells in our case was a practical decision based on our current simulation needs and it also simplifies the neighbor searching code and enables optimizations described in the next section . in dccrg all cells existing within a certain minimum distance from local cells ( cells owned by the process )
are stored in a hash table with the cell i d as the key and the process owning the cell as the value . since the mapping of cell ids to a location is unique ,
finding the neighbors of a cell in the hash table is straightforward : for all indices neighboring a given cell the hash table is searched for cells of all applicable refinement levels .
figure [ fig : neighbors ] shows an example of neighbor searching in a grid with @xmath17 and a neighborhood size of one .
the siblings of cell # 4 ( # 3 , # 7 , # 8 , # 11 , # 12 , # 15 and # 16 ) are not shown for clarity and some potential neighbors of cell # 4 in the positive x direction have also been omitted . in dccrg
the cells neighborhoods are measured in multiples of their own size , e.g. for cell # 4 all cells that are ( at least partially ) between indices 0 and 11 inclusive in the x direction would be considered as neighbors . in order to find the neighbor(s ) of cell # 4 in positive x direction the hash table is searched for the smallest cell at indices ( 8 , 0 , 0 ) which in this case could correspond to any of the following cells : # 2 , # 5 , # 23 or # 155 . if cell # 23 is the smallest cell found in the hash table the search can stop since it is known that the siblings of cell # 23 also exist ( not shown , cells # 24 , # 31 , # 32 , # 55 , # 56 , # 63 and # 64 ) because all children of a cell are created when a cell is refined .
on the other hand if cell # 155 is the smallest cell found at indices ( 8 , 0 , 0 ) then the search would continue at indices outside of cell # 155 and its siblings , for example at indices ( 10 , 0 , 0 ) or ( 8 , 2 , 0 ) .
the concept of hanging nodes or faces used in unstructured mesh codes is not directly applicable to dccrg because a single cell is the smallest unit that dccrg deals with .
since the user is responsible for defining what is stored in each cell he / she must also define , if required , the data stored at the faces , edges and vertices of cells . hanging nodes , which are the result of cells of different size sharing an edge or a face , must be handled by the solvers used for a particular application .
for example in gumics-4 @xcite where the magnetic field is separated into background and perturbed components the face average background magnetic field is required when solving the flux through a face . with amr a face average value of the background field
is required for every face of every cell because , for example , if only one face average is stored per dimension in cells it would not be possible to solve the flux between cells ( # 4 , # 2 ) and ( # 1 , # 5 ) in figure [ fig : neighbors ] .
this is due to the fact that the face average value of the smaller cell must be used in both cases and it is only available if every face of every cell stores the face average field .
currently dccrg enforces a maximum refinement level difference of 1 between neighboring cells .
hence it is sufficient to search for cells of three refinement levels @xmath18 when finding the neighbors of a cell of refinement level @xmath6 . in principle
the enforcement of maximum refinement level difference is not required .
for example in figure [ fig : neighbors ] cell # 4 ( refinement level 1 ) has cell # 155 ( refinement level 3 ) as a neighbor but in the current version of dccrg such a situation is not permitted and searching the hash table for cells # 2 , # 5 and # 23 is sufficient for finding the neighbors of cell # 4 .
this was a practical decision based on our experience with global mhd modeling of the earth s magnetosphere using gumics-4 . in future
this restriction might be removed .
a similar one is also used in @xcite . even though a maximum refinement level difference of one is enforced between neighboring cells and searching for cells in the hash table
is a quick operation , in practice the ids of neighbors of local cells are also stored explicitly . as mentioned in section [ sec : introduction ] , this neighbor information corresponds to arrows between cells in a graph and hence we will use the term arrow list to refer to the neighbor list of a cell .
dccrg stores both the arrows pointing away from and the arrows pointing to local cells . with amr
it is possible that there exists only one arrow between two cells because cells neighborhoods are measured in units of their own size .
for example in figure [ fig : example_grid ] with a neighborhood size of 1 , cell # 13 would be considered a neighbor of cell # 2 but cell # 2 would not be considered a neighbor of cell # 13 .
explicitly storing the arrows to and from local cells enables fast iteration for example by user code ( solvers , reconstruction functions , etc ) . in dccrg
the arrow lists of local cells are stored as contiguous arrays .
the most important advantage that globally unique cell ids have over a traditional graph implementation using native pointers between cells is that the arrows between cells are not required to be consistent during amr or load balancing .
for example when doing amr , a pointer - based implementation has to be careful not to leave any dangling pointers and to update the pointers of all nearby cells in the correct order so as not to lose access to any cell . on the other hand with unique cell ids the arrow lists of cells
can simply be emptied when needed and new neighbors searched in the hash table as described in section [ sec : neigh_search ] .
in addition to being easy to implement this method also admits simple thread - based parallelization inside dccrg due to different threads modifying only the arrow lists of different cells .
it is also advantageous to use unique cell ids in arrow lists instead of pointers in a parallel program running on distributed memory hardware . in this environment
a pointer to a neighboring cell is only valid on one process and can not be used to refer to the same neighbor on other processes . on the other hand
the same unique cell i d can be used by all processes to refer to the same neighbor regardless of its actual location in memory .
a separate grid library is a natural abstraction probably for any physical simulation but especially for simulations using fvm where the concept of a grid and its cells data dependencies are easy to define and implement . thus following good software development practice dccrg
is implemented independently of any specific physical problem or its solver , while still providing the flexibility required for various types of simulations , for example ( m)hd , advection ( e.g. vlasov ) and kinetic .
dccrg is written in c++ which allows us to easily separate low level functionality of dccrg into subclasses which higher - level classes can use thus also benefiting from a modular internal implementation , a technique also used in @xcite .
for example the physical geometry of the grid is handled by a separate class which is also given to dccrg as a template parameter .
this allows one to easily extend the grid geometries supported by dccrg . in the default
homogeneous and cartesian geometry cells of the same refinement level are identical in size in each dimension . here
we describe the unique user - visible features of dccrg with respect to other grid codes and also present important features of the serial and parallel implementation .
the most important feature distinguishing dccrg from other grid codes is the possibility of trivially storing data of arbitrary type and size in the grid s cells by simply giving the class which is stored in the cells as a template parameter to dccrg when creating an instance of the grid .
this also allows a single simulation to have several independent parallel grids with different geometries and different types of data stored in the grids cells. the amount of data can also vary between different cells of the same grid and in the same cell as a function of time .
this is required for example in kinetic simulations where not only does the total number of particles change but also the number of particles in each grid cell varies . in dccrg
this is handled by each instance of the user s cell data class providing a mpi datatype corresponding to the data to be sent from or received by that particular cell .
an example of this is presented in section [ sec : example_program ] .
completely arbitrary cell data can also be transferred between processes if the cell data class provides a serialize function which the mpi bindings of boost library will use for transferring cell data between processes .
although this method of transferring data between processes the most general it is also the slowest since data is first copied into a contiguous buffer by serialization and subsequently transferred by mpi resulting in at least one additional copy the data being created compared to pure a mpi transfer .
this is also the case in the samrai framework @xcite which supports transferring arbitrary patch data using the same technique .
dccrg can automatically transfer cell data between processes both for remote neighbor data updates and load balancing using simple function calls .
furthermore whenever cell data is sent between processes either one mpi message per cell can be used or , similarly to @xcite , all cells being sent to another process can be transferred using a single mpi message .
updating the remote neighbor data between processes is possible using several methods .
the simplest one is the synchronous update function that updates the remote neighbor data between processes and returns once transfers have completed ( see section [ sec : example_program ] ) .
the most fine - grained communication currently supported can be used by calling a separate function for initiating transfers and functions that wait for the sends and receives to complete respectively .
a typical usage scenario would consist of the following : 1 .
start transferring remote neighbor data 2 .
solve the inner cells of the simulation ( cells without remote neighbors ) 3 .
wait for the data from other processes to arrive 4 .
solve the outer cells of the simulation ( cells with at least one neighbor on another process ) 5 .
wait for the data from this process to be sent the mhd scalability tests we present in section [ sec : mhd_tests ] use this procedure with the exception that step 5 is executed before step 4 due to the technical implementation of the gumics-4 mhd solver . as mentioned in section [ sec : introduction ] the size of cells neighborhood in simulations is highly problem / solver dependent .
specifically the problem / solver used in the simulation dictates the distance from which data is required at a cell in order to advance the simulation for one time step or one iteration in that cell . in many previously published grid codes
the size of cells neighborhood is restricted to 1 either explicitly or implicitly .
for example in @xcite in three dimensions a block ( a single cell from the point of view of the grid ) has 6 neighbors and it is assumed that a block consist of such a number of simulation cells that , for example , a solver needing data from a distance of 3 simulation cells can obtain that data from the neighboring block .
other examples are @xcite , @xcite , @xcite , @xcite , @xcite and @xcite .
dccrg supports an arbitrarily large neighborhood chosen by the user when the grid is initialized .
the size of the neighborhood can be any unsigned integer and all other cells within that distance of a cell ( in units of size of the cell itself ) will be considered as a neighbors of the cell .
this enables the use of high - order solvers with the added possibility of refining each neighboring cell individually .
naturally one can also store a sufficiently large block of simulation cells in one dccrg cell allowing one to use a small 6 cell neighborhood as done in @xcite .
zero neighborhood size is a special case in dccrg signifying that only face neighbors of equal size are considered neighbors ( with amr all of the refined neighbor s children are considered instead ) .
for example in a periodic grid without amr neighborhood sizes of 1 and 2 would result in 26 and 124 neighbors per cell respectively .
naturally the size of the neighborhood affects the amount of data that must be transferred between processes during remote neighbor data updates regardless of the grid implementation thus affecting parallel scalability .
additionally since in dccrg a maximum refinement level difference of one is enforced between neighboring cells the size of the neighborhood does affect for example the amount of induced cell refinement ( see section [ sec : amr ] ) .
even though initially dccrg was developed only for in - house use , it was nevertheless designed to be simple to use for the kinds of simulations it is targeted for . figure [ fig : gol ] shows an example of a complete parallel program playing conway s game of life using dccrg written in less that 60 lines of code ( loc ) including whitespace and comments .
lines 10 ... 14 of the program define the class to be stored in every cell of dccrg along with member functions at and mpi_datatype which dccrg calls when querying the information required for transferring cell data between processes .
the current state of a cell is saved into data[0 ] and the number of its live neighbors is saved into data[1 ] . on line 23 an instance of the grid
is created with the class defined above as cell data . on line 24 the geometry of the grid
is set to 10x10x1 cells at refinement level 0 with minimum coordinate at ( 0 , 0 , 0 ) and cells of size 1 in each dimension . on line 25 the grid is initialized by setting the load balancing function to use in zoltan , the neighborhood size and the maximum refinement level of cells .
lines 26 and 27 balance the load using zoltan and collect the local cells . in this example
the load is balanced only once and the local cell list does not change afterwards . on line 46 the non - existing neighbors of local cells are skipped .
this is because the grid is initialized with non - periodic boundaries and neighbors that would be outside of the grid do not exist .
figure [ fig : particles ] shows relevant excerpts from a simple kinetic simulation showing the use of dccrg in the case of variable amount of cell data , the full program can be viewed in the dccrg git repository .
the remote neighbor update logic in the main simulation loop consists of the following steps : 1 .
the total number of particles in each cell is transferred between processes ( lines 43 and 44 ) 2 .
space for receiving particle data is allocated in local copies of remote cells based on their received total number of particles in step 1 ( lines 47 ... 52 ) 3 .
the particle coordinates are transferred between processes ( lines 55 and 56 ) the cell data class of the example kinetic simulation must provide the correct information to dccrg when updating remote neighbor data : the at and mpi_datatype functions now return a different address and number of bytes respectively depending on whether the total number of particles or the particle coordinates are transferred between processes .
this is decided by the user in the main simulation loop .
additionally the resize function of the cell data class allocates memory for as many particles as is there are in number_of_particles . a similar approach to the one described above
is also used in our vlasov simulation ( further developed from @xcite ) where each real space cell has a separate adaptable velocity grid for ions consisting of a variable number of @xmath19 cell velocity blocks . in the previous example two
communications are required per time step because processes receiving particle data do not know the number of incoming particles in advance . since the mpi standard requires that the maximum amount of data to be received is known before calling the receive function the number of particles has to be communicated separately .
this guarantees that processes receiving particles can specify the size of the data to mpi and allocate the memory required for received particles .
load balancing is also accomplished easily with dccrg .
a user can call the balance_load function to let the zoltan @xcite library create a new partition , and single cells can also be moved manually between processes using the pin and unpin functions .
most of zoltan s load balancing methods can be used , namely : none , random , block , rcb , rib , hsfc , graph , hypergraph and hier . in any case dccrg
will transparently execute the new partition by transferring the necessary cell data between processes using mpi .
the structure of the grid in dccrg includes the owner of a cell in addition to the unique i d of the cell ( i d is the key and owner is the value in a hash table ) .
thus the cell ids themselves are not used for partitioning cells between processes and any cell can be moved to any process ( for example by using the random partitioner of zoltan which we have found to be very useful for testing ) . due to the unique mapping of cells ids and their physical location and size it is straightforward to refine
any given cell in the grid , i.e. to calculate the ids of the children of any cell , and can be done locally ( see section [ sec : hash_table ] ) . in order to enforce a maximum refinement level difference of one between neighboring cells whenever a cell is refined the refinement level of all neighbors
is checked .
if the refinement level of any neighbor is less than that of the cell being refined that neighbor is also refined .
this is continued recursively until no more cells need to be refined .
the size of the cells neighborhood affects induced refinement indirectly by changing the number of neighbors a cell has and hence the potential number of cells whose refinement will be induced . in dccrg
a few simplifications have been made to amr : 1 ) any set of cells can only be refined once before calculating induced refinement ( by stop_refining ) , e.g. induced refinement can only increase the refinement level of cells by one , and 2 ) unrefining a cell does not induce unrefinement , e.g. any cell which has at least one neighbor with refinement level larger than the cell being unrefined ( i.e. it has a smaller neighbor ) can not be unrefined . in a parallel setting the only difference to the above is that whenever a process refines or unrefines a cell that information has to be given to all processes which have cells within a certain distance of the cell that was refined or unrefined . currently this distance is equal to infinity , e.g. all changes to the structure of the grid ( refines and unrefines ) are communicated globally . this has a significant impact on the parallel scalability of amr in dccrg and is discussed in section [ sec : tests ] and a method for making this minimum distance finite is discussed in section [ sec : discussion ] .
the changes in grid structure are exchanged between processes after each recursive step of induced refinement which are continued until no more cells need to be refined .
good scalability in distributed memory machines requires asynchronous point - to - point mpi communication between processes with minimal total amount of communication , and especially minimal amount of global communication .
the number of mpi messages should also be minimized in order not to burden the network with unnecessary traffic .
therefore any process must know which processes require data from local cells and from which remote cells data is required during remote neighbor updates without querying that information from other processes beforehand .
thus in dccrg every process knows the structure of the grid ( e.g. which cells exist and which processes own them ) to at least a certain distance from any of its cells so it can calculate locally which cells data to send and receive from other processes . due to this dccrg
does not require global communication during ordinary time stepping , e.g. remote neighbor data updates , which enables excellent scalability when not doing load balancing or amr as shown in section [ sec : mhd_tests ] .
internally dccrg precalculates these send and receive lists whenever the structure of the grid changes due to amr or cells being moved between processes . even though the replicated mesh metadata of dccrg does not include the data stored in each cell it nevertheless limits the size of dccrg grids in practice to less than 100 m existing cells .
this number does not depend on the refinement level or physical location of the cells and has so far been more than sufficient for our needs . to our knowledge
the only parallel grid library that does not have any persistent global data structures is @xcite . in @xcite and @xcite
only the macrostructure of the grid ( i.e. cells of refinement level 0 ) is known by all processes . according to the authors this limits the size of the grid to the order of @xmath20 @xmath0 cells of refinement level 0 but
does not limit the number of smaller cells .
the time stepping scalability of dccrg ( e.g. without amr or load balancing ) depends mostly on the hardware running the simulation and on three parameters specific to each simulated system : 1 ) the time required to solve the inner cells of a process , 2 ) the amount of data transferred during the remote neighbor data update of a process and 3 ) the time required to solve the outer cells of a process .
we show the dependency of dccrg scalability on these parameters by varying the total number of cells and processes and by using a mhd solver in one , two and three dimensions .
the run - time amr scalability of dccrg is also presented .
the non - amr scalability tests were carried out on three different supercomputers : 1 ) a 2 k core cray xt5 m system with 12 core nodes connected by seastar2 installed at the finnish meteorological institute which we will refer to as meteo , 2 ) a 295 k core ibm blue gene / p system with 128 core nodes ( nodeboards ) installed at the jlich supercomputing centre which we will refer to as jugene , and 3 ) a 12 k core bullx system with 32 core nodes connected by infiniband installed at the trs grand centre de calcul which we will refer to as curie .
first we show the time stepping scalability of dccrg in several mhd problems with a solver developed for the global mhd model gumics-4 ( @xcite ) which solves ideal mhd equations in conservative form .
specifically the solver is a first order roe s approximate riemann solver for a godunov type problem @xcite . in the test results we present here
only the roe solver from gumics-4 is used as we do not experience problems with negative pressures or densities in the presented tests .
the nature of the solver is such that when solving the flux through a face data is only required from cells adjacent to the face , irrespective of the size of cells involved .
thus interpolation of data is not required at any point in the solution and if a cell has more than one face neighbor in any direction the flux through each common face is solved in the usual way . in the tests presented here
we do not include a background magnetic field which is used in gumics-4 to represent the earth s dipole field .
the problems we use are the one - dimensional shock tube presented for example in @xcite , the two - dimensional circularly polarized alfvn wave presented by @xcite and further elaborated on by @xcite , and the three - dimensional blast wave presented by @xcite .
periodic boundary conditions are used in all tests , except for the shock tube test in the direction of the tube where initial conditions are enforced after every time step . in mhd tests
every cell contains the cell - averaged values of the conservative mhd variables ( density , momentum density , total energy density and magnetic field ) giving a total of 128 bytes which must be transferred when updating the data of one cell between two processes . since only the face neighbors of a cell
are required for calculating the next time step we use a neighborhood size of zero in dccrg . in these tests processes
execute one collective mpi communication per time step in order to dynamically calculate the maximum physical length of the time step .
no other global communication is done . since the grid is static in these tests the computational load is balanced only once at the start of the simulation by using a hilbert space - filling curve instead of zoltan .
figure [ fig : mhd - meteo ] shows the results of strong scalability tests using mhd with a static grid in meteo in one , two and three dimensions . the total number of cells ( 10 k , 50 k , 0.1 m
, 0.5 m and 1 m ) is kept constant while the number of mpi processes is increased from 12 to 1536 . in each test case scalability
improves with the total number of cells because processes have more inner cells to solve while remote neighbor data is being transferred .
for example in the shock tube test every process requires the data of two remote neighbors at most while the number of inner cells with 1.5 k processes increases from about 4 ( 10 k total cells ) to 649 ( 1 m total cells ) . with 1 m cells
the one and two dimensional tests scale almost ideally in meteo and the three dimensional test is also quite close to ideal .
the overall decrease in scalability with increasing number of dimensions is due to more data being transferred between processes for the same number of local cells . as suggested by the scalability results above most of the simulation time
is spent solving mhd which is shown in figure [ fig : meteo - profile ] for the three dimensional blast wave test using 1 m cells .
the only global communication executed per time step while simulating is the calculation of the maximum length of the physical time step and is obtained using mpi_allreduce .
this is labeled as allreduce in figure [ fig : meteo - profile ] and basically shows the computational and mpi imbalance between processes due to load balancing .
initialization and file i / o are not included in the profile and other parts of the simulation take an insignificant fraction of the total run time .
the non - amr scalability tests were also carried out in jugene and curie and the results for the three - dimensional blast wave are shown in figures [ fig : blast - wave - jugene ] and [ fig : blast - wave - curie ] respectively .
similarly to meteo the one and two dimensional tests ( not shown ) scale better than the three - dimensional test in both jugene and curie .
the overall results are similar in all tested machines , e.g. scalability improves with increasing number of total cells and decreasing number of dimensions . in jugene very good
scalability up to 32 k processes is obtained for a total number of cells of 1 m and above . the total simulation speed in jugene
is only slightly above that of meteo mostly due to the relatively small single - core performance of jugene .
additionally the average number of cells per process is more than 20 times larger in meteo than in jugene for the maximum number of processes used but this has only a small effect on scalability in jugene . in curie
good scalability up to 8 k processes is obtained only with 64 m total cells but with a maximum solution speed of nearly 400 m solved cells per second which is over twice of that in jugene .
we attribute this to the relatively low node interconnect and high single - core and performance of curie respectively when compared to jugene .
figure [ fig : amr_speed ] shows the speed of pure adaptive mesh refinement in dccrg . initially the grid is @xmath21 or @xmath22 cells and every process refines all local cells until the total size of the grid is @xmath22 or @xmath23 cells .
initially the cells were partitioned using a space - filling curve and this is not included in the timings .
cells also were not transferred between processes during amr .
as can be seen in figure [ fig : amr_speed ] the maximum cell refining speed of dccrg is of the order of 1 m cells per second .
the linear scalability of amr up to some 32 mpi processes is explained by the fact that after changes in the structure of the grid the arrow lists are recalculated only for local cells and mpi communication has not yet become a bottleneck . at 256 processes
the amount of global communication required for updating the structure of the grid between all processes starts to significantly affect the speed of amr .
this is discussed further in section [ sec : amr_blast_wave ] . or @xmath22 cells .
every process refines each local cell until the total size of the grid is @xmath22 or @xmath23 cells . ] here we present the scalability of dccrg with amr in the three - dimensional blast wave test used in section [ sec : mhd_tests ] . in this test a procedure similar to the one in gumics-4 ( eq . 2 in @xcite )
is used to decide whether to refine or unrefine a cell : a refinement index is calculated for each cell based on the relative difference of several variables between a cell and its face neighbors . here
the calculation of refinement index @xmath24 additionally includes velocity shear relative to the maximum wave velocity from the cells interface .
the full equation for the refinement index @xmath24 is : @xmath25 where @xmath26 denotes the difference in a variable between two cells , the hat denotes a minimum of the two values ( as it actually does also in @xcite ) , @xmath27 and @xmath28 is the maximum wave velocity from the cells interface . in this test
a cell is refined if @xmath29 , where @xmath6 is the cell s current refinement level and @xmath30 is the maximum refinement level of the grid .
in other words a cell is refined to the maximum refinement level if its refinement index exceeds 0.02 .
a cell is unrefined if @xmath31 , e.g. a cell is kept at refinement level 0 if @xmath32 and none of the cell s neighbors refinement levels exceed 1 ( due to dccrg enforcing a maximum refinement level difference of one between neighbors ) .
we use a maximum refinement level of 4 in this test with an initial grid of @xmath33 cells which results in an effective resolution of @xmath34 m cells .
the computational load is balanced using zoltan s recursive coordinate bisection ( rcb ) algorithm whenever the fraction of local cells ( @xmath35 , where max and min are the maximum and minimum number of local cells among all processes respectively ) exceeded a specified limit .
animation 1 ( figure [ fig : amr_result ] in the print version ) shows from left to right , top to bottom the grid , pressure , magnetic and kinetic energy density during the simulation ( at the end of the simulation in print version ) in the y = 0 plane when grid is adapted at every time step .
at the end of the simulation the fraction of maximum to minimum values are 15 for density ( not shown ) , 43 for pressure and 2.3 for magnetic energy density . even though the mhd solver we use is simpler than the one in @xcite the results are still close due to the high effective resolution achieved by using run - time amr .
figure [ fig : amr_scalability ] shows the total solution speed during the simulations as a function of the number of mpi processes used . in the reference run a cfl @xcite of 0.4 is used , amr is done at every time step and the load is balanced whenever the local cell fraction @xmath36 .
the @xmath37 runs are otherwise identical to the reference run but cfl is set to @xmath38 and amr is done every nth time step , essentially multiplying the amount of non - amr work in these runs by n. the results between different amr runs are identical by visual inspection except for increased diffusion in runs with low cfl .
the ratio of work required by amr and the rest of the simulation has a significant effect on the total solution speed .
the solution speed is a factor of 5 higher in the @xmath39 run than in the reference run when using about 500 mpi processes . in the reference
amr run the total solution speed is close to @xmath40 of the non - amr version with up to 144 processes and in the @xmath39 the speed is close to @xmath41 with up to 288 processes after which both fractions start to decrease .
we define these as the regions of excellent amr scalability . on the other hand in all of
the amr runs the total solution speed increases up to about 500 to 600 processes after which it starts to decrease .
we define this as the region where amr is scalable . the total number of cells in the amr runs averages to 4.5 m which is about @xmath42 of the non - amr version .
consequently in the region of excellent scalability the time to solution when using amr is about 67 % to 22 % of the non - amr time for the reference and @xmath39 runs respectively .
even with a higher number of mpi processes it can still be advantagous to use amr because the number of simulation cells is over a magnitude lower than without amr . at the end of the amr runs 9.9 m cells
exist in the grid and the total number of cells created and removed is between 40.2 m and 40.7 m , depending mostly on the diffusion , and averages to about 91 k added + removed cells per time step
. exceeded 2 , where @xmath35 and max and min are the maximum and minimum number of local cells among processes respectively .
the @xmath37 runs are otherwise identical to the reference run but cfl is set to @xmath38 and amr is done every nth time step , essentially multiplying the amount of non - amr work in these runs by n. ] figure [ fig : meteo - profile - amr ] shows which parts of the amr blast wave test require the most time . as the number of processes is increased the largest fraction of simulation time is spend in global communication related to amr and load balancing .
the allreduce label again indicates global calculation of the physical time step and the load balancing label indicates the simulation time spent in load balancing related functions . at about 300 mpi processes and above the largest fraction of simulation time
is spent communicating changes in the structure of the grid between all processes .
this includes both refining and unrefining cells as well as load balancing and in each case the mpi_allgatherv function is used for distributing the changes in grid structure between all processes .
only a small fraction of the time spent in load balancing related functions is taken by zoltan .
while the ease of use of a software library is subjective it can be quantified by the number of lines of code required for usage and compared against other libraries when using the same programming language . with dccrg a complete parallel program playing conway s game of life
can be implemented in less than 60 loc including whitespace and comments .
even though the required loc is a crude estimate for a software library s ease of use it is nevertheless telling that such a short parallel program does not seem to be possible with other grid libraries .
the flexibility of dccrg also stands out since as far as we know only @xcite allows one to easily exchange arbitrary cell data between mpi processes .
additionally dccrg supports transferring user - defined mpi datatypes which is critical for performance in some applications .
for example when solving the 6 dimensional vlasov equation in the earth s magnetosphere ( @xcite ) the simulation is heavily memory bound and using mpi datatypes directly for exchanging remote neighbor data is significantly faster than serializing said data into an additional buffer(s ) before transfer .
dccrg also provides automatic neighbor data updates between processes with the ability of easily overlapping computation with communication .
currently the largest drawback of dccrg is the fact that the entire structure of the grid is known by every process , i.e. a part of the mesh metadata is replicated by all processes .
the global data structure prevents grids larger than about 100 m cells but this has not been a problem for us and can be worked around by storing blocks instead of single cells into dccrg ( similarly to @xcite , for example ) .
the global grid data structure of dccrg also reduces the scalability of amr in the worst case to about 200 and overall to about 600 mpi processes .
nevertheless using amr can lead to significant savings in the required memory as the number of cells can be one or even two orders of magnitude lower . depending on the problem the required cpu time can also be significantly reduced when using amr especially when the number of mpi processes used is of the order of 300 or less .
it should also be noted that using threads to parallelize solvers within a shared memory node could effectively multiply the scalability range of simulations by the number of cores within one node , but this was not investigated . removing or significantly reducing the global data structure ( as done in @xcite , @xcite and @xcite ) should improve both the largest attainable grid size and scalability of amr considerably .
intuitively this is straightforward since with the exception of load balancing every process only needs to know the structure of the grid up to some finite distance from local cells . in order to be able to arbitrarily refine and unrefine grid cells without global communication local changes in the structure of the grid
must be communicated between all neighboring processes .
a neighboring process is defined as any process that has one or more of its cells inside the neighborhood of any cell of refinement level 0 that overlaps a local cell . in other words
if only level 0 cells exist in the grid then the owners of all remote neighbors of local cells are considered as neighboring processes ; and this holds no matter how the grid is subsequently refined and unrefined assuming that cells are not transferred between processes ( load balancing ) .
global communication can also be avoided during load balancing if , for example , cells can be transferred only between neighboring processes . even in this case
new neighbor processes have to be recalculated but global communication is not required because cells could only have been transferred to / from a subset of all processes . implementing this completely distributed mesh metadata
is left to a subsequent study .
we presented the distributed cartesian cell - refinable grid ( dccrg ) : an easy to use parallel structured grid library supporting adaptive mesh refinement and arbitrary c++ classe as cell data .
various mhd scalability results were presented and depending on the problem , hardware and whether amr is used excellent to average scalability is achieved .
dccrg is freely available for anyone to use , study and modify under version 3 of the gnu lesser general public license and can be downloaded from https://gitorious.org/dccrg .
this work is a part of the project 200141-quespace , funded by the european research council under the european community s seventh framework programme .
the research leading to these results has also received funding under grant agreement no 260330 of the european community s seventh framework programme .
ih and mp are supported by project 218165 and as is supported by project 251797 of the academy of finland .
results in this paper have in part been achieved using the prace research infrastructure resource curie based in france at tgcc and jugene based in germany at jsc .
ih thanks daldorff , l.k.s . and pomoell , j. for insightful discussions .
p. janhunen , m. palmroth , t. laitinen , i. honkonen , l. juusola , g. facsk , t.i .
pulkkinen , the gumics-4 global mhd magnetosphere - ionosphere coupling simulation , j. atmos .
.- terr . phys .
80 ( 2012 ) 48 - 59 doi:10.1016/j.jastp.2012.03.006 m. palmroth , i. honkonen , a. sandroos , y. kempf , s. von alfthan , d. pokhotelov , preliminary testing of global hybrid - vlasov simulation : magnetosheath and cusps under northward interplanetary magnetic field , j. atmos .
sol .- terr .
phys . accepted ( 2012 )
henshaw , d.w .
schwendeman , parallel computation of three - dimensional flows using overlapping grids with adaptive mesh refinement , j. comput .
( 2008 ) 7469 - 7502 doi:10.1016/j.jcp.2008.04.033 p. macneice , k.m .
olson , c. mobarry , r. de fainchtein , c. packer , paramesh : a parallel adaptive mesh refinement community toolkit , comput .
commun . , 126 ( 2000 ) 330 - 354 doi:10.1016/s0010 - 4655(99)00501 - 9 q.f .
stout , d.l .
de zeeuw , t.i .
gombosi , c.p.t .
groth , h.g .
marshall , k.g .
powell , adaptive blocks : a high performance data structure , proc .
1997 acm / ieee conf .
57 ( 1997 ) doi:10.1109/sc.1997.10027 w. bangerth , c. burstedde , t. heister , m. kronbichler , algorithms and data structures for massively parallel generic finite element codes , acm trans .
38 ( 2011 ) 14/1 - 28 doi:10.1145/2049673.2049678 r. keppens , z. meliani , a.j .
van marle , p. delmont , a. vlasis , b. van der holst , parallel , grid - adaptive approaches for relativistic hydro and magnetohydrodynamics , j. comput .
( 2012 ) 718 - 744 doi:10.1016/j.jcp.2011.01.020 a.m. wissink , r.d .
hornung , s.r .
kohn , s.s .
smith , n. elliott , large scale parallel structured amr calculations using the samrai framework , proc .
2001 acm / ieee conf .
( 2001 ) doi:10.1145/582034.582040 | we present an easy to use and flexible grid library for developing highly scalable parallel simulations .
the distributed cartesian cell - refinable grid ( dccrg ) supports adaptive mesh refinement and allows an arbitrary c++ class to be used as cell data .
the amount of data in grid cells can vary both in space and time allowing dccrg to be used in very different types of simulations , for example in fluid and particle codes .
dccrg transfers the data between neighboring cells on different processes transparently and asynchronously allowing one to overlap computation and communication .
this enables excellent scalability at least up to 32 k cores in magnetohydrodynamic tests depending on the problem and hardware . in the version of dccrg presented here part of the mesh metadata
is replicated between mpi processes reducing the scalability of adaptive mesh refinement ( amr ) to between 200 and 600 processes .
dccrg is free software that anyone can use , study and modify and is available at https://gitorious.org/dccrg .
users are also kindly requested to cite this work when publishing results obtained with dccrg .
parallel grid , adaptive mesh refinement , free open source software * program summary * _ manuscript title : _ parallel grid library for rapid and flexible simulation development + _ authors : _ honkonen , i. , von alfthan , s. , sandroos , a. , janhunen , p. , palmroth , m. + _ program title : _
dccrg + _ journal reference : _
+ _ catalogue identifier : _
+ _ licensing provisions : _ gnu lgpl v3 + _ programming language : _
c++ + _ computer : _ pc , cluster , supercomputer + _ operating system : _ posix + _ ram : _ 10 mb - 10 gb per process + _ number of processors used : _ 1 - 32768 cores + _ supplementary material : _
+ _ keywords : _ dccrg , parallel , grid , amr , mpi , fvm , fem + _ classification : _ 4.12 , 4.14 , 6.5 , 19.3 , 19.10 , 20 + _ external routines / libraries : _ mpi-2 [ 1 ] , boost [ 2 ] , zoltan [ 3 ] , sfc++ [ 4 ] + _ nature of problem : _ + grid library supporting arbitrary data in grid cells , parallel adaptive mesh refinement , transparent remote neighbor data updates and load balancing . + _ solution method : _ + the simulation grid is represented by an adjacency list ( graph ) with vertices stored into a hash table and edges into contiguous arrays .
message passing interface standard is used for parallelization .
cell data is given as a template parameter when instantiating the grid .
+ + + _ restrictions : _
+ logically cartesian grid .
+ _ additional comments : _ + + _ running time : _ + running time depends on the hardware , problem and the solution method .
small problems can be solved in under a minute and very large problems can take weeks . the examples and tests provided with the package
take less than about one minute using default options . in the version of dccrg presented here
the speed of adaptive mesh refinement is at most of the order of @xmath0 total created cells per second .
+ 0 http://www.mpi-forum.org/ http://www.boost.org/ k. devine , e. boman , r. heaphy , b. hendrickson , c. vaughan , zoltan data management services for parallel dynamic applications , comput .
sci .
eng . 4 ( 2002 ) 90 - 97 doi:10.1109/5992.988653 https://gitorious.org / sfc++ |
suppose that @xmath0 is a separable topological vector space and @xmath1 is a continuous linear mapping on @xmath0 . if @xmath2 , then the orbit of @xmath3 under @xmath1 is defined as @xmath4 .
an operator @xmath1 is called hypercyclic if there is a vector @xmath3 such that @xmath5 is dense in @xmath0 and in this case @xmath3 is called a hypercyclic vector for @xmath1 ( see @xcite for an exhaustive survey on hypercyclicity ) .
it is interesting that many continuous linear mappings can actually be hypercyclic .
the first example of hypercyclicity appeared in the space of entire functions , by birkhoff @xcite in 1929 .
he showed the hypercyclicity of the translation operator , while maclane @xcite proved the hypercyclicity of the differentiation operator in 1952 .
hypercyclicity on banach spaces was discussed in 1969 by rolewics @xcite , who showed that @xmath6 is hypercyclic whenever @xmath7 is the unilateral backward shift ( on @xmath8 and @xmath9 ) and @xmath10 .
a nice condition for hypercyclicity is the hypercyclicity criterion ( theorem 1.1 below ) , which was developed by kitai @xcite and independently by gethner and shapiro @xcite .
this criterion has been used to show that certain classes of composition operators @xcite , weighted shifts @xcite , adjoints of multiplication operators @xcite , and adjoints of subnormal and hyponormal operators @xcite , are hypercyclic .
hypercyclicity has also been established in various other settings by means of this criterion @xcite .
salas @xcite showed that every perturbation of the identity by a unilateral weighted backward shift with nonzero bounded weights is hypercyclic , and he also gave a characterization of the hypercyclic weighted shifts in terms of their weights .
but , then montes and leon showed that these hypercyclic operators do satisfy the criterion as well ( 2 of @xcite and proposition 4.3 of @xcite ) .
bes and peris proved that a continuous linear operator @xmath1 on a frechet space satisfies the hypercyclicity criterion if and only if it is hereditarily hypercyclic .
in particular they show that hypercyclic operators with either a dense generalized kernel or a dense set of periodic points must satisfy the criterion .
also , they provide a characterization of those weighted shifts @xmath1 that are hereditarily hypercyclic with respect to a given sequence @xmath11 of positive integers , as well as conditions under which @xmath1 and @xmath12 share the same set of hypercyclic vectors @xcite .
suppose @xmath0 is a separable banach space and @xmath1 is a continuous linear mapping on @xmath0 .
if there exists two dense subsets @xmath13 and @xmath14 in @xmath0 and a sequence @xmath15 such that : 1 .
@xmath16 for every @xmath17 , 2
. there exists functions @xmath18 such that for every @xmath19 , and @xmath20 , then @xmath1 is hypercyclic .
note that the sequence @xmath15 in theorem 1.1 need not be the entire sequence @xmath21 of positive integers .
salas @xcite and herrero @xcite have shown that there are hypercyclic operators on hilbert spaces that do not satisfy the hypercyclicity criterion for the entire sequence @xmath22 , but so far no hypercyclic operator has been found that does not satisfy the hypercyclicity criterion in its general form . in this paper our work was stimulated by the well - known question : does every hypercyclic operator satisfy the hypothesis of the hypercyclicity criterion ?
( see @xcite ) .
we give necessary and sufficient conditions in terms of open subsets for an operator on a separable hilbert space to satisfy the hypercyclicity criterion .
for this , see theorem 2.6 , corollary 2.11 and proposition 2.12 .
also , in the proof of theorem 2.6 , we pay attention to hypercyclicity on the operator algebra @xmath23 and the algebra of hilbert schmidt operators , @xmath24 . recall that if @xmath25 is an orthonormal basis for a separable hilbert space @xmath26 and @xmath27^{1/2},\ ] ] then @xmath28 is independent of the basis chosen and hence is well - defined .
if @xmath29 , then @xmath30 is called a hilbert schmidt operator and by this norm @xmath24 is a hilbert space . indeed , @xmath31 is a special case of the schatten @xmath32-class of @xmath33 when @xmath34 . for more details about these classes of operators ,
see @xcite .
chan @xcite showed that hypercyclicity can occur on the operator algebra @xmath23 with the strong operator topology ( sot - topology ) that is not metrizable .
for example , when @xmath1 satisfies the hypercyclicity criterion , then the left multiplication operator @xmath35 is sot - hypercyclic on @xmath23 , that is , @xmath35 is hypercyclic on @xmath23 with strong operator topology .
from now on we suppose that @xmath33 is a separable infinite - dimensional hilbert space .
@xmath36 let @xmath37 be linear and bounded .
we say that @xmath38 is sot - hypercyclic if there exist some @xmath39 such that the set @xmath40 is dense in @xmath23 in the strong operator topology .
also we say that @xmath41 is @xmath42-hypercyclic if there exists some @xmath43 such that @xmath44 is dense in @xmath24 with @xmath42-topology .
@xmath36 for any operator @xmath39 , define the left multiplication operator @xmath45 by @xmath46 for every @xmath47 .
note that @xmath24 is an ideal of @xmath23 and hence @xmath48 is also well - defined .
we show that @xmath23 and @xmath24 , respectively with the strong operator topology and @xmath42-topology , are separable . for this , see the following lemma 2.3 .
suppose @xmath49 is an orthonormal basis for a separable hilbert space @xmath33 and @xmath50 denotes the set of all finite rank operators @xmath1 such that there exists @xmath51 , satisfying @xmath52 for @xmath53 .
suppose @xmath54 is a basis for a separable hilbert space @xmath33 , then @xmath50 is sot - dense in @xmath23 and also @xmath42-dense in @xmath24 ; moreover , @xmath50 isseparable .
suppose that @xmath55 and @xmath56 .
then there exist @xmath57 such that @xmath58
. now define the finite rank operator @xmath59 by @xmath60 on @xmath61 $ ] and @xmath62 on @xmath63^{\bot}$ ] .
( @xmath64 $ ] means the linear span of @xmath65 ) .
thus @xmath66 and so @xmath50 is @xmath42-dense . also , of @xcite p. 234
implies that every @xmath42-dense subset of @xmath24 is sot - dense in @xmath23 , and so it follows that @xmath50 is sot - dense .
now the proof is complete.@xmath67 the following result is the main tool that we used to show that an operator is hypercyclic .
versions of this result have appeared in the work of godefroy and shapiro ( @xcite , theorem 1.2 ) and kitai ( @xcite , theorem 2.1 ) .
@xmath36 if @xmath1 is a continuous operator on a separable banach space @xmath0 , then @xmath1 is hypercyclic if and only if for any two non - void open sets @xmath68 and @xmath69 in @xmath70 for some positive integer @xmath71 .
godefroy and shapiro ( @xcite , corollary 1.3 ) also gave a sufficient condition for hypercyclicity that is a direct consequence of proposition 2.4 .
@xmath36 an operator @xmath1 on a separable banach space @xmath0 is hypercyclic if for each pair @xmath72 of non - void open subsets of @xmath0 , and each neighborhood @xmath73 of zero in @xmath0 , there are infinitely many positive integers @xmath71 such that both @xmath74 and @xmath75 are non - empty .
@xmath36 1 . in proposition 2.4 ,
the condition @xmath76 is equivalent to the condition @xmath77 .
if an operator @xmath1 is hypercyclic , then it automatically has a dense set of hypercyclic vectors . for
, if a vector @xmath3 is hypercyclic for @xmath1 , then so is @xmath78 for any positive integer @xmath71 .
thus the condition ` @xmath79 for some positive integer @xmath71 ' , in proposition 2.4 , can be replaced by the condition ` @xmath76 for infinitely many positive integers @xmath71 ' .
3 . equivalent to the hypothesis of corollary 2.5 is the apparently weaker requirement that the sets @xmath80 and @xmath81 be non - empty for a single @xmath71 .
the following theorem shows that the converse of the above corollary is equivalent to the hypercyclicity criterion .
remember that for vectors @xmath82 in @xmath33 the operator @xmath83 denotes a rank one operator and is defined by @xmath84 . for any operator @xmath39 ,
the following are equivalent : 1 .
@xmath1 satisfies the hypothesis of the hypercyclicity criterion .
2 . for each pair @xmath72 of non - void open subsets of @xmath33 , and each neighborhood @xmath73 of zero , @xmath85 and @xmath86 for some integer @xmath71 .
it is easy to see that ( i ) implies ( ii ) ( for details see corollary 1.4 in @xcite ) . for the converse , assume that @xmath1 satisfies property ( ii ) .
first we show that for each pair @xmath87 of non - void @xmath42-open subsets of @xmath24 there is an integer @xmath88 such that @xmath89 . for this ,
fix an orthonormal basis @xmath90 for @xmath33 . by using lemma 2.3 there exist finite rank operators @xmath30 and @xmath7 such that @xmath91 and @xmath92 , whence for a certain integer @xmath93 we have @xmath94 for @xmath95 .
but for some @xmath56 we have @xmath96 and @xmath97 now consider the following open sets : @xmath98 @xmath36 for @xmath99 .
note that corollary 2.5 or remark ( iii ) implies that @xmath1 is hypercyclic .
now by using proposition 2.4 repeatedly ( indeed by remark ( ii ) ) , it follows that there exist integers @xmath100 and @xmath101 such that @xmath102 and @xmath103 put @xmath104 where @xmath105 since @xmath1 satisfies the hypothesis ( ii ) of theorem 2.6 , then there exists some @xmath106 and @xmath107 such that @xmath108 and @xmath109 for some integer @xmath71 .
the relations ( 1 ) and ( 2 ) imply that @xmath110 for @xmath99 . now define @xmath111 and @xmath112 .
let @xmath113 .
then @xmath114 is a hilbert schmidt operator , because it has finite rank .
note that by ( 3 ) , @xmath115 .
now by using ( 4 ) we get the following inequalities : @xmath116 & = \left\ { \sum^{n}_{i=1 } \|s_{1 } e_{i } - ae_{i}\|^{2}\right\}^{1/2 } + \left\ { \sum^{n}_{i=1 } \|s_{2 } e_{i}\|^{2}\right\}^{1/2}\\[.4pc ] & = \left\ { \sum^{n}_{i=1 } \|t^{n_{i-1 } } y - ae_{i}\|^{2}\right\}^{1/2 } + \left\ { \sum^{n}_{i=1 } \|t^{m_{i-1 } } x\|^{2}\right\}^{1/2 } < \varepsilon.\end{aligned}\ ] ] hence @xmath117 .
also note that since @xmath118 , by ( 3 ) we get @xmath119 , and thus we have @xmath120 & = \left\ { \sum^{n}_{i=1 } \|t^{n } s_{2 } e_{i } - be_{i } \|^{2}\right\}^{1/2 } + \left\ { \sum^{n}_{i=1 } \|t^{n } s_{1 } e_{i}\|^{2 } \right\}^{1/2}\\[.4pc ] & = \left\ { \sum^{n}_{i=1 } \|t^{n}(t^{m_{i-1 } } x ) - be_{i}\|^{2}\right\}^{1/2}\\[.4pc ] & \quad\ , + \left\ { \sum^{n}_{i=1 } \|t^{n_{i-1 } } ( t^{n } y)\|^{2}\right\}^{1/2 } < \varepsilon.\end{aligned}\ ] ] so @xmath121 .
now it follows that @xmath122 and so by proposition 2.4 , @xmath35 is @xmath123-hypercyclic .
this also implies that @xmath124 is hypercyclic , because the left multiplication operator @xmath125 is unitary equivalent to the operator @xmath124 ( see @xcite , p. 6 ) .
now theorem 2.3 in @xcite implies that @xmath1 satisfies the hypercyclicity criterion , and so the proof is now
complete.@xmath67 @xmath36 if @xmath39 , then the following are equivalent : 1 .
@xmath1 satisfies the hypothesis of the hypercyclicity criterion .
@xmath1 is hypercyclic and for each non - void open subset @xmath68 and each neighborhood @xmath73 of zero , @xmath85 and @xmath126 for some integer @xmath71 . by theorem 2.6
it suffices to show that ( ii ) implies ( i ) .
so let ( ii ) hold . by theorem 2.6
, it suffices to show that ( ii ) in theorem 2.6 holds . since @xmath1 is hypercyclic , by proposition 2.4 , @xmath127 for some positive integer @xmath128 .
let @xmath129 be a neighborhood of zero that is contained in @xmath130 .
by condition ( ii ) , there exists some positive integer @xmath71 such that @xmath131 and @xmath132 . but @xmath133 is a subset of @xmath134 , hence @xmath135 . also @xmath136 is a subset of @xmath137 which implies that @xmath138 .
thus , hypothesis ( ii ) of theorem 2.6 holds and so the proof is complete.@xmath67 we say that the sequence @xmath139 of bounded linear operators on a hilbert space @xmath33 is hypercyclic provided that there exists some @xmath140 such that the collection of images @xmath141 is dense in @xmath33 .
note that theorem 1.1 , proposition 2.4 and corollary 2.5 can be extended to the case where hypercyclicity of @xmath1 is replaced by hypercyclicity for the sequence @xmath142 of bounded linear operators that have dense range .
in particular we say that @xmath142 satisfies the hypothesis of the hypercyclicity criterion if in the hypothesis of theorem 1.1 , we use @xmath143 instead of @xmath144 .
it also implies that if the sequence @xmath142 satisfies the hypothesis of the hypercyclicity criterion , then @xmath142 is hypercyclic ( see theorem 1.2 , corollaries 1.3 and 1.5 in @xcite ) .
it is not difficult to see that theorem 2.6 and proposition 2.7 work for the sequence @xmath142 of bounded linear operators provided that @xmath145 for each pair @xmath146 of positive integers .
hence we can deduce the following corollary .
@xmath36 suppose that @xmath142 is a sequence of bounded linear operators on a hilbert space @xmath33 such that @xmath145 for each pair @xmath146 of positive integers and have dense range
. then the following are equivalent : 1 .
@xmath142 satisfies the hypothesis of the hypercyclicity criterion .
2 . for each pair @xmath72 of non - void open subsets of @xmath33 , and each neighborhood @xmath73 of zero , @xmath147 and @xmath148 for some integer @xmath71 .
@xmath142 is hypercyclic and for each non - void open subset @xmath68 and each neighborhood @xmath73 of zero , @xmath147 and @xmath149 for some integer @xmath71 .
the following definition is introduced in @xcite .
@xmath36 suppose that @xmath39 and @xmath15 is a sequence of positive integers .
we say that @xmath1 is hereditarily hypercyclic with respect to @xmath15 if for any subsequence @xmath150 of @xmath15 , the sequence @xmath151 is hypercyclic .
now we summarize all necessary and sufficient conditions for the hypercyclicity criterion in the following corollary .
@xmath36 for any operator @xmath39 , the following are equivalent : 1 .
@xmath1 satisfies the hypothesis of the hypercyclicity criterion .
@xmath1 is hereditarily hypercyclic with respect to a subsequence @xmath15 of positive integers .
@xmath152 is hypercyclic on @xmath153 .
the left multiplication operator @xmath154 is @xmath42-hypercyclic .
5 . for each pair @xmath72 of non - void open subsets of @xmath0 , and each neighborhood @xmath73 of zero , @xmath85 and @xmath86 for some integer @xmath71 .
the proof is an immediate consequence of theorem 2.6 , proposition 2.7 and theorem 2.3 in @xcite.@xmath67 the following proposition represents some relation between hypercyclicity and the hypercyclicity criterion .
@xmath36 for any operator @xmath39 the following are equivalent : 1 .
@xmath1 satisfies the hypothesis of the hypercyclicity criterion .
2 . there exists a dense subset @xmath13 in @xmath0 and a sequence @xmath15 such that @xmath155 is hypercyclic and @xmath156 for every @xmath17 .
3 . there exists a sequence @xmath15 such that for each pair @xmath72 of non - void open subsets of @xmath33 , there is @xmath93 such that @xmath157 for any @xmath158 .
@xmath36 \(i ) @xmath159 ( ii ) : it follows from condition ( ii ) of corollary 2.11 .
\(ii ) @xmath159 ( i ) : let @xmath160 be any non - void open set and also let @xmath73 be any open neighborhood of zero . then by remark 2.8 , @xmath161 is hypercyclic and so there is some sequence @xmath162 of positive integers such that @xmath163 for every @xmath164 . now
if @xmath165 , then @xmath166 which yields @xmath167 .
it holds condition ( iii ) of corollary 2.11 , hence @xmath168 satisfies the hypothesis of the hypercyclicity criterion and so @xmath155 and consequently @xmath1 satisfy the hypercyclicity criterion .
\(iii ) @xmath159 ( i ) : it suffices to show that condition ( iii ) implies condition ( v ) of corollary 2.11 .
for this let @xmath72 be a pair of non - void open subsets of @xmath33 and @xmath73 be any neighborhood of zero .
then for some integer @xmath169 , we have @xmath170 for any @xmath171 .
thus indeed condition ( v ) of corollary 2.11 is consistent .
\(i ) @xmath159 ( iii ) : note that by condition ( ii ) of corollary 2.11 , @xmath1 is hereditarily hypercyclic with respect to a sequence @xmath15 of positive integers .
now suppose that ( iii ) does not hold .
so there exist some pair @xmath72 of non - void open sets such that @xmath172 for some subsequence @xmath150 of @xmath15 . but @xmath151 is hypercyclic and so it is a contradiction .
hence for every pair @xmath72 of non - void open sets , there is @xmath173 such that @xmath157 for any @xmath158 .
the proof is now complete.@xmath67
the authors thank the referee for many interesting comments and helpful suggestions about the paper . | we give necessary and sufficient conditions for an operator on a separable hilbert space to satisfy the hypercyclicity criterion .
[ theore]*theorem * [ theore]definition [ theore]lemma [ theore]remark [ theore]proposition [ theore]corollary |
given an @xmath0-vertex edge - weighted undirected planar graph @xmath10 , a distance oracle for @xmath10 is a data structure that can efficiently answer distance queries @xmath11 between pairs of vertices @xmath12 in @xmath10 .
one way of achieving this is to simply store an @xmath13-distance matrix where @xmath0 is the number of vertices .
each query can be answered in constant time but the space requirement is large . if one is willing to settle for approximate distances , much more compact oracles exist .
it has been shown that for any @xmath1 , there is a @xmath2-approximate distance oracle for @xmath10 of size @xmath14 which for any query pair @xmath12 outputs in time @xmath15 an estimate @xmath16 such that @xmath17 ( thorup @xcite and klein @xcite ) .
the oracles of thorup and of klein both rely on a recursive decomposition of @xmath10 using shortest path separators from a shortest path tree @xmath18 : first @xmath10 is decomposed into two subgraphs with such a separator and then the two subgraphs are recursively decomposed .
an important observation is that for any vertex @xmath19 and any shortest path separator @xmath20 , there is a size @xmath15 set @xmath21 of so called portals on @xmath20 which are vertices such that for any @xmath22 , there exists a @xmath23 such that @xmath24 .
thus , to get an approximate distance from @xmath19 to any @xmath22 only @xmath15 distances @xmath25 need to be stored in addition to distances in @xmath18 .
the oracle stores distances from @xmath19 to portals on each of the @xmath26 separators above @xmath19 in the recursive decomposition tree , giving a total space of @xmath27 . to answer a @xmath28-query , the oracle identifies the nearest common ancestor separator @xmath29 of @xmath19 and @xmath30 in the recursive decomposition . as @xmath29 separates @xmath19 and @xmath30 , distances from @xmath19 and from @xmath30 to their respective portal sets on @xmath29
can be combined to obtain a @xmath2-approximate distance estimate in @xmath15 time .
additional oracles for planar graphs have since been presented .
kawarabayashi , klein , and sommer @xcite showed how to improve space to @xmath31 at the cost of an increase in query time to @xmath32 and gave generalizations to bounded - genus and minor - free graphs .
kawarabayashi , sommer , and thorup @xcite focused on improving the space - query time tradeoff and gave an oracle with @xmath33 space and @xmath34 query time , where @xmath35 hides @xmath36 and @xmath37 factors , thereby essentially improving the query time - space product from @xmath38 to @xmath14 .
they also showed that if the average edge weight is poly - logarithmic , @xmath39 space and @xmath34 query time can be obtained . except for planar graphs with poly - logarithmic average edge weights ,
every oracle presented so far has a query time - space product of order @xmath40 ( ignoring the dependency on @xmath41 ) .
we finally break this barrier by giving an oracle with @xmath3 space and @xmath4 query time .
the exact bounds are given in the following theorem .
[ thm : main ] let @xmath10 be an @xmath0-vertex undirected edge - weighted planar graph . for any @xmath42 , there is a @xmath2-approximate distance oracle of @xmath10 with query time @xmath43 and space @xmath44 .
our dependency on @xmath41 in the query time - space product is worse than in @xcite but still only a low - degree polynomial in @xmath45 ; it is roughly @xmath46 when @xmath47 and roughly @xmath48 otherwise , compared to @xmath49 in @xcite .
focus in this paper is on improving the dependency on @xmath0 and not @xmath41 which we regard as fixed .
our data structure uses randomization due to hashing and fast integer sorting .
space and query time for hashing can be made worst - case with expected construction time . for sorting ,
we use the algorithm of han and thorup @xcite to get the bound in theorem [ thm : main ] . to make our data structure deterministic
, we can use a standard optimal comparison sort or the slightly faster deterministic integer sorting algorithm of fredman and willard @xcite . with the latter ,
we get a deterministic query time of our data structure of @xmath50 .
a main difference between our oracle and those of thorup and of klein is that we do not store distances from each vertex @xmath19 to portals on all @xmath26 separators above @xmath19 .
instead we save space by introducing a shortcutting system to the recursive decomposition tree so that we can get from @xmath19 to any separator above it using @xmath51 shortcuts .
each shortcut corresponds to a pair of separators on a root - to - leaf path in the tree and we essentially store approximate distances from vertices on one separator to portals on the other separator closer to the root of the recursive decomposition tree .
this complicates the query algorithm and its analysis since approximate distances are found in @xmath51 steps instead of just one .
thorup @xcite also gave an oracle for planar _ digraphs _ which for polynomially bounded edge weights achieves @xmath52 space and close to @xmath15 query time .
exact oracles for planar digraphs with tradeoff between space and query time have been studied but require near - quadratic space for constant or near - constant query time @xcite . for a general undirected @xmath0-vertex graph @xmath10 , thorup and zwick showed that for any parameter @xmath53
, there is a @xmath54-approximate distance oracle for @xmath10 with @xmath55 space and @xmath56 query time which is believed to be essentially optimal due to a girth conjecture of erds @xcite .
variations and slight improvements have since been presented ; see , e.g. , @xcite . in section [ sec :
prelim ] , we give some basic definitions , notation , and a variant of a standard recursive decomposition of planar graphs with shortest path separators .
we then present our oracle in sections [ sec : phasei ] and [ sec : phaseii ] .
section [ sec : phasei ] presents the first phase of the query algorithm .
this phase computes approximate distances from query vertices @xmath19 and @xmath30 to certain portals on the nearest common ancestor separator @xmath29 of @xmath19 and @xmath30 in the recursive decomposition tree but where approximate shortest paths are restricted to the child regions of @xmath29 in the recursive decomposition .
the second phase in section [ sec : phaseii ] uses the output of the first phase to then find approximate distances from @xmath19 and @xmath30 to portals on @xmath29 in the entire graph @xmath10 .
a main challenge is that the vertices of @xmath29 are not represented explicitly on all recursion levels but on various levels on the path from @xmath29 to the root of the recursive decomposition ; the second phase traverses this path to find the desired approximate distances and portals . from the output of phase ii ,
obtaining an approximate @xmath28-distance can then be done efficiently , as we show in section [ sec : approxdist ] .
finally , we make some concluding remarks in section [ sec : conclrem ] .
for a graph @xmath10 , denote by @xmath57 and @xmath58 its vertex set and edge set , respectively . when convenient , an edge @xmath12 with weight @xmath59 is denoted @xmath60 . for a rooted tree @xmath18 and two nodes @xmath61 , denote by @xmath62 the nearest common ancestor of @xmath19 and @xmath30 in @xmath18 . for a path @xmath21 and for two vertices @xmath63 ,
@xmath64 $ ] denotes the subpath of @xmath21 between @xmath19 and @xmath30 . as in previous papers on approximate distance oracles for planar graphs ,
we assume the word - ram model with standard instructions . in the following , @xmath65 denotes an @xmath0-vertex , undirected , edge - weighted planar embedded graph and @xmath18 is a shortest path tree in @xmath10 rooted at a source vertex @xmath66 . by performing vertex - splitting
, we may assume that @xmath10 has degree three .
the oracle of thorup keeps a recursive decomposition of @xmath10 consisting of shortest path separators .
our oracle obtains a similar decomposition but we need it to have some additional properties which we focus on in the following . denote by @xmath67 an arbitrary triangulation of @xmath10 where edges of @xmath68 are called _ pseudo - edges _ and are given infinite weight .
a shortest path separator of @xmath67 w.r.t .
an assignment of weights to triangles of @xmath67 consists of a ( possibly non - simple ) cycle @xmath69 defined by two shortest paths @xmath70 and @xmath71 in @xmath18 and a non - tree edge @xmath12 ; the total weight of triangles on each side of @xmath69 is at most @xmath72 of the total weight of all triangles of @xmath67 .
see @xcite for details .
first , we decompose @xmath67 into two subgraphs enclosed by @xmath69 ; both subgraphs inherit the edges and vertices of @xmath69 , for a suitable weight function on triangles .
degree two vertices @xmath19 are removed from each subgraph by replacing their incident edges @xmath73 and @xmath74 with a single edge @xmath75 whose weight is the sum of weights of @xmath73 and @xmath74 .
then the two subgraphs are recursively decomposed until constant - size subgraphs are obtained . for each subgraph @xmath76 obtained in the above recursive procedure ,
we form a _ region _ @xmath77 as follows .
subgraph @xmath76 contains @xmath26 separators of the form @xmath78 , namely those formed from the root of the recursion down to @xmath76 .
these are separators formed from the root of the recursion down to @xmath76 .
region @xmath77 is obtained from @xmath76 by removing pseudo - edges , except those contained in the separators that formed @xmath76 , and then removing degree two vertices as above .
let @xmath79 be the faces of @xmath76 containing vertices / edges of @xmath10 not belonging to @xmath76 .
each @xmath80 is a separator @xmath78 ( possibly with some degree two vertices removed ) and we call @xmath80 a _ hole _ ( of @xmath77 ) .
ancestor / descendant relations between regions are defined according to their nesting in the recursive decomposition tree which we denote by @xmath81 . in section [ subsec : constructrecdecomp ] ,
we show how to pick shortest path separators such that 1 .
there are @xmath31 regions in total each having @xmath82 holes , 2 . for each child @xmath77 of each region @xmath76 ,
every hole of @xmath76 ( regarded as the closed set of the plane inside the hole ) is fully - contained in a hole of @xmath77 , 3 .
the height of @xmath81 is @xmath26 .
when we refer to a recursive decomposition in the following , we assume it has these properties .
observe that for all regions @xmath76 and @xmath77 where @xmath76 is a descendant of @xmath77 , @xmath83 .
this follows since @xmath76 is obtained from @xmath77 by eliminating subgraphs of @xmath77 and degree two vertices . in the following ,
refer to the triangulated subgraphs obtained when recursively decomposing @xmath67 as _ @xmath84-regions _ ; holes of @xmath84-regions are defined to be the holes of the corresponding regions with pseudo - edges added to form the triangulation .
we now show how to pick the separators so that the following _ region conditions _ are satisfied : 1 .
each @xmath84-region of even resp .
odd depth in @xmath81 has at most three resp .
four holes , 2 . for each @xmath84-region of even depth in @xmath81 , if it contains exactly @xmath85 faces of @xmath86 , each of its grandchildren contain at most @xmath87 faces of @xmath86 , 3 . for each child @xmath77 of each @xmath84-region @xmath76 , every hole of @xmath76 ( regarded as the closed set of the plane inside the hole ) is fully - contained in a hole of @xmath77 .
furthermore , we show that the number of regions is @xmath31 and that the height of @xmath81 is @xmath26 .
this gives the desired properties for a recursive decomposition , as stated above .
all separators are formed from shortest path tree @xmath18 .
the recursion stops once @xmath84-regions with at most two faces of @xmath86 are obtained . for a @xmath84-region @xmath76 of even depth in @xmath81 ,
assume it has at most three holes ( this trivially holds for @xmath67 at the root of @xmath81 ) .
we assign a unit of weight to each face of @xmath76 that is also a face of @xmath67 .
all other faces of @xmath76 are given weight @xmath88 .
using the subtree of @xmath18 in @xmath76 , we find a balanced shortest path separator w.r.t . this weight function .
if @xmath85 is the number of faces of @xmath86 in @xmath76 then each of the subgraphs formed contain at most @xmath87 of these faces .
furthermore , each of these subgraphs have at most four holes since at most one new hole is formed when removing one side of the separator .
now consider a @xmath84-region @xmath76 of odd depth in @xmath81 and assume it has at most four holes .
if @xmath76 has at most two holes , we decompose it as described above for even - depth @xmath84-regions ; the sub-@xmath84-regions formed will have at most three holes .
otherwise , we define a different weight function than that above : for each hole @xmath89 of @xmath76 , exactly one of the triangles of @xmath76 contained in @xmath89 is given unit weight .
all other triangles of @xmath76 are given weight @xmath88 .
the shortest path separator w.r.t .
this weight function ensures that each of the two sub-@xmath84-regions of @xmath76 formed will have at most three holes , namely at most @xmath90 holes inherited from @xmath76 and one hole formed by removing one side of the separator .
it is now clear that the first region condition holds .
the second region condition holds as well from the above and from the fact that each child of an odd - depth @xmath84-region @xmath76 can not contain more faces of @xmath86 than @xmath76 .
assume for the sake of contradiction that the third region condition does not hold for some @xmath84-region @xmath76 and one of its children .
then the separator that was used to decompose @xmath76 must have used one of the pseudo - edges in the triangulated hole @xmath89 of @xmath76 .
since the boundary of @xmath89 consists of two shortest paths from @xmath18 and a single pseudo - edge , one of the children of @xmath76 must be fully contained in @xmath89 , which for both choices of weight function above gives an unbalanced separator , a contradiction .
we conclude that the third region condition holds .
the number of @xmath84-regions is asymptotically bounded by the number of leaves of the recursive decomposition tree .
each leaf @xmath84-region has a parent with at least three faces of @xmath86 .
for any two distinct @xmath84-regions @xmath76 and @xmath77 that are parents of leaf @xmath84-regions , no face of @xmath86 belongs to both @xmath76 and @xmath77 .
it follows that there are only @xmath31 parents of leaf @xmath84-regions .
since the decomposition tree @xmath81 is binary , the total number of leaf @xmath84-regions is @xmath31 .
this implies that the total number of @xmath84-regions , and hence regions , is @xmath31 .
it follows from the second region condition and the termination condition for the recursion that @xmath81 has height @xmath26 .
let @xmath91 be a path in @xmath18 from @xmath66 to some vertex .
for any subpath @xmath92 $ ] of @xmath91 , let @xmath93 be those interior vertices of @xmath91 having an incident edge of @xmath94 emanating to the left of @xmath92 $ ] when looking in the direction from @xmath19 to @xmath30 .
we order the vertices such that @xmath92 = u\leadsto v_1\leadsto\cdots\leadsto v_k\leadsto v$ ] . the _ left side _ of @xmath92 $ ] is the @xmath28-path with edges @xmath95 where each edge has weight equal to the weight of the corresponding subpath of @xmath91 .
we define the _ right side _ of @xmath92 $ ] similarly . for a region @xmath76 , denote by @xmath96 the _ boundary _ of @xmath76 which is the subgraph of @xmath76 contained in the @xmath82 holes of @xmath76 . for each hole @xmath89
, it will be convenient to regard the two shortest paths in @xmath18 bounding @xmath89 as disjoint in @xmath96 by replacing one path with its left side and the other with its right side ; vertices shared by the original two paths are regarded as distinct in the two new paths , see figure [ fig : eulertour ] .
note that @xmath96 represented in this way is now a single face of @xmath76 .
cutting open each hole like this ensures that paths in @xmath76 do not cross its boundary .
a vertex ( edge ) of @xmath76 that is not contained in @xmath96 is called _
interior_. denote by @xmath97 the @xmath82 shortest paths from the shortest path tree @xmath18 that bound @xmath96 .
a vertex which is an endpoint of a path in @xmath97 is called a _ corner _ of @xmath76 ; see figure [ fig : eulertour ] .
we let @xmath98 denote the set of @xmath82 corners of @xmath76 . in some places
, we will instead consider the face @xmath99 obtained from @xmath96 by replacing each edge with its corresponding path in @xmath10 .
the following lemma gives the structure of @xmath96 .
[ lem : deltar ] for any region @xmath76 , @xmath96 consists of @xmath82 subpaths each of which is either a single ( possibly pseudo-)edge between two corners , an edge corresponding to a path in @xmath18 from @xmath66 to some corner of @xmath76 , or the left or right side of a path in @xmath18 .
consider partitioning @xmath96 into subpaths each starting and ending in a corner and with no interior corner vertices .
there are only @xmath82 such subpaths and each of them is either an edge or pseudo - edge ending in corners or the subpath corresponds to a path in @xmath18 incident to one or two holes . if it is incident to one hole , it is the left or right side of a path in @xmath18 and if it is incident to two holes , it is a single edge corresponding to a path ending in @xmath66 . when convenient , we identify regions with their corresponding nodes in @xmath81 . for two regions @xmath76 and @xmath77 , denote by @xmath100 the simple path from @xmath76 to @xmath77 in @xmath81 .
let @xmath65 be an undirected edge - weighted ( not necessarily planar ) graph , let @xmath101 , and let @xmath91 be a shortest path in @xmath10 .
thorup @xcite showed that for any given @xmath1 , @xmath102 contains a set @xmath21 of @xmath15 _ portals _ such that for any @xmath103 , there is a @xmath23 such that @xmath104 .
let @xmath76 be a region and let @xmath105 .
given a value @xmath106 , a vertex @xmath101 , and an undirected ( possibly non - planar ) graph @xmath89 with @xmath107 where every edge @xmath108 in @xmath89 corresponds to a path @xmath109 in @xmath10 of the same weight .
then @xmath21 is a _
@xmath110-portal set of @xmath96 _ if for any @xmath111 , there exists a vertex @xmath23 such that @xmath112 .
we call @xmath113 a _ portal _ ( of @xmath21 ) . applying the portal construction of thorup @xcite
( see also the proof of lemma [ lem : portal ] ) to each path in @xmath97 gives a @xmath110-portal set of @xmath96 of size @xmath114 .
define @xmath110-portal set of @xmath99 similarly ; its size is @xmath115 as well .
we need a slightly more general result regarding portals than that of thorup which we state in the following somewhat technical lemma .
it roughly says that if we have a graph @xmath89 representing some subgraph of @xmath10 such that distances in @xmath10 from a vertex @xmath19 to a shortest path @xmath91 in @xmath10 are approximated well in @xmath89 , then we can pick a small number of portals from @xmath89 along @xmath91 such that these distances are also approximated well with shortest paths in @xmath89 from @xmath19 to @xmath91 through these portals .
[ lem : portal ] let @xmath91 be a shortest path in an edge - weighted undirected graph @xmath65 and let @xmath101 and @xmath116 be given .
let @xmath89 be an undirected graph with @xmath117 such that for any @xmath118 , @xmath119 .
assume that for any @xmath103 , there is a @xmath120 such that @xmath121 .
then there is a subset @xmath122 of @xmath123 of size @xmath124 such that for any @xmath103 there is a @xmath125 such that @xmath126 .
the construction is similar to that of thorup .
the first portal @xmath127 added to @xmath122 is the vertex @xmath128 minimizing @xmath129 .
let @xmath130 be an endpoint of @xmath91 .
we show how to construct @xmath131 $ ] ; the same construction is done for the other subpath of @xmath91 .
let @xmath132 be the latest portal added to @xmath122 and traverse @xmath133 $ ] towards @xmath130 until encountering a vertex @xmath134 such that @xmath135 .
portal @xmath136 is then the next portal added to @xmath122 .
the process stops when reaching the vertex of @xmath123 closest to @xmath130 ; this vertex is added as the final portal @xmath137 to @xmath122 .
let @xmath103 be given . by assumption , there is a @xmath120 such that @xmath121 .
the above construction ensures that there is a @xmath125 such that @xmath138 it remains to prove that @xmath139 . by symmetry , it suffices to show @xmath140| = o(1/\epsilon'')$ ] . for any @xmath141)$ ] , define potential @xmath142 . for @xmath143 , @xmath144
so the potential is reduced by @xmath145 since @xmath146 and @xmath147 ( because @xmath148 ) , we have @xmath149 , implying that @xmath150 .
our data structure answers a query for vertex pair @xmath12 in two phases , _ phase i _ and _ phase ii_. in this section , we describe the preprocessing for phase i and then the query part .
the output and performance of phase i applied to @xmath19 is stated in the following lemma ( @xmath30 is symmetric ) .
phase i starts with @xmath151 which is an arbitrary region @xmath76 ( among at most two choices ) such that @xmath152 and @xmath153 where @xmath77 is the parent of @xmath76 in @xmath81 .
region @xmath154 is defined similarly .
[ lem : phasei ] phase i for vertex @xmath19 can be implemented to run in @xmath155 time using @xmath156 space . for the output @xmath157
, we have that for all @xmath158 , there is a @xmath159 such that @xmath160 where @xmath161 is the child of @xmath162 on the path in @xmath81 from @xmath163 to @xmath164 .
furthermore , @xmath165 .
note that any @xmath28-path of @xmath10 must intersect @xmath166 .
phase i computes approximate distances to this separator but with the restriction that paths must be contained in @xmath161 .
phase ii is considered in section [ sec : phaseii ] and it extends the output of phase i to approximate distances to @xmath167 in the entire graph @xmath10 .
we start by constructing a recursive decomposition of @xmath10 and the associated decomposition tree @xmath81 . in order to traverse leaf - to - root paths of @xmath81 efficiently
, we set up a shortcutting system for @xmath81 .
for any region @xmath168 , let @xmath169 be the largest integer such that the depth of @xmath76 in @xmath81 is divisible by @xmath170 .
for any integer @xmath171 between @xmath88 and @xmath169 , we add a pointer from @xmath76 to the ancestor @xmath77 @xmath172 levels above @xmath76 .
we refer to this pointer as a _ shortcut _ and denote it by @xmath173 .
we can get from any region @xmath174 to any proper ancestor @xmath175 of @xmath174 by traversing only @xmath51 shortcuts : first traverse the shortcut @xmath176 where @xmath77 is the closest region to @xmath175 which is either @xmath175 or one of its descendants .
then recurse on pair @xmath177 until reaching @xmath175 . before describing the preprocessing for phase i ,
we need the following lemma . for any vertex @xmath101 ,
define @xmath178 as the set of regions @xmath76 where @xmath179 .
[ lem : regionpath ] for all @xmath101 , regions of @xmath178 form a subpath of a leaf - to - root path in @xmath81 .
let @xmath76 be a region in @xmath178 .
non - corner vertices of shortest paths in @xmath97 have degree three in @xmath10 ( otherwise , their incident edges would have been merged into one in the construction of @xmath76 ) so @xmath19 must be incident to an interior edge @xmath180 of @xmath76 . this edge can not be a pseudo - edge since @xmath181 but must be an edge of @xmath10 .
the same can not be true for both child regions of @xmath76 in @xmath81 since @xmath10 has degree three so at most one of these regions belongs to @xmath178 .
we sometimes regard @xmath178 as the subpath from the lemma . for any shortcut @xmath182 ,
define @xmath183 as the set of @xmath82 vertices @xmath184 such that @xmath19 is the last vertex from @xmath66 on a path of @xmath185 satisfying @xmath186 ; see figure [ fig : phaseidata ] . for each shortcut @xmath182 and each @xmath187 , we construct and store a size @xmath188 @xmath189-portal set @xmath190 of @xmath191 together with distances
@xmath192 for each @xmath193 ( figure [ fig : phaseidata ] ) ; @xmath194 will be specified precisely in section [ subsec : timestretch ] below .
this completes the description of the preprocessing .
[ lem : typeiportalset ] the total space required for phase i is @xmath195 .
it suffices to give an @xmath195 bound on the total size of portal sets defined above .
let @xmath196 be given . by lemma
[ lem : regionpath ] , there can only be @xmath51 shortcuts @xmath182 where @xmath197 and @xmath198 .
hence there are only @xmath51 shortcuts @xmath182 where @xmath199 . the total number of sets @xmath183 and the total number of corners of @xmath174 over all shortcuts @xmath182 is @xmath200 and @xmath201 . as each set @xmath190
has size @xmath188 , total size of portal sets is @xmath195 . in this subsection ,
we present phase i for query vertices @xmath19 and @xmath30 .
figure [ fig : phaseidata ] is useful to consult in the following .
we assume that starting regions @xmath151 and @xmath154 are not on the same leaf - to - root path in @xmath81 . as we will see later , the other case is easily handled .
pseudocode for vertex @xmath19 can be found in figure [ fig : phasei ] ; the same call is made with @xmath19 replaced by @xmath30 . ' '' ''
+ ddd============**phase i * * for @xmath19 : + + 1 . let @xmath202 be the shortcuts from @xmath203 to @xmath204 + 2 .
let @xmath205 + 3 . for each @xmath206
, let @xmath207 + 4 .
for @xmath208 to @xmath209 + 5 .
let @xmath210 + 6 .
let @xmath211 + 7 .
let @xmath212 be face @xmath213 restricted to vertices that are either in @xmath214 or + are incident to edges in @xmath215 + 8 .
construct the graph @xmath80 consisting of the edges @xmath216 + 9 . for each @xmath217 ,
let @xmath218 + 10.if @xmath219 , construct @xmath220-portal set @xmath221 of @xmath213 of size @xmath188 + 11.construct @xmath222-portal set @xmath223 of @xmath224 of size @xmath225 + 12.output @xmath226 , where @xmath227 + ' '' '' let @xmath161 be defined as in lemma [ lem : phasei ] ( @xmath228 is defined similarly for @xmath30 ) .
let @xmath202 denote the sequence of shortcuts from @xmath229 to @xmath204 . to simplify the code , we assume @xmath230 ; the other case is straightforward .
note that @xmath231 . in lines @xmath232 and @xmath233 , we obtain the precomputed portal set @xmath205 as well as distances @xmath234 for each portal @xmath206 . note that @xmath235 is well - defined by definition of @xmath151 . in the @xmath169th iteration of the for - loop
, we are given @xmath236 constituting portals for @xmath237 and we form a graph @xmath80 containing @xmath19 and a subset of @xmath238 such that all distances from @xmath19 to @xmath213 in @xmath239 can be approximated by going through @xmath80 and then along @xmath213 .
an illustration of @xmath80 can be seen in figure [ fig : hiphaseidata ] .
edges of @xmath240 in line @xmath241 are added to @xmath80 and these represent the approximate distances to @xmath236 in @xmath242 found in the previous iteration . to have @xmath80 approximate distances from @xmath19 in @xmath239
, we add another edge set @xmath243 , defined in line @xmath244 ; for every vertex in @xmath245 , we add to @xmath80 the edge @xmath246 for every @xmath247 , representing a shortest @xmath248-path in @xmath239 .
to allow @xmath80 to traverse @xmath213 , we add @xmath212 in line @xmath249 .
this is a compact representation of face @xmath213 restricted to a subset of its vertices ; each subpath of @xmath213 between two consecutive vertices in this subset is a single edge of the same weight in @xmath212 .
we show below that distances in @xmath80 from @xmath19 to @xmath250 approximate distances from @xmath19 to @xmath213 in @xmath239 . in order to avoid an explosion in the size of future portal sets , we form a @xmath220-portal set @xmath221 of @xmath213 of size
only @xmath188 in line @xmath251 which is then used in the next iteration .
line @xmath252 is identical to line @xmath251 except that we use a value @xmath253 instead of @xmath254 .
we shall pick @xmath255 which gives a much smaller portal set @xmath256 output in line @xmath257 ; this will help speed up phase ii .
we do not use @xmath258 inside the for - loop in line @xmath251 since the approximation error builds up over each iteration so we need the smaller value @xmath254 there .
lemma [ lem : phasei ] follows from the following invariant for the for - loop in lines @xmath259@xmath251 :
invariant : : : at the start of the @xmath169th iteration of the for - loop in lines @xmath259@xmath251 of figure [ fig : phasei ] , for all @xmath260 , there is a @xmath261 such that @xmath262 , and @xmath263 .
note that the invariant holds initially when @xmath208 since @xmath235 is a @xmath264-portal set of @xmath191 and its size is @xmath188 .
[ [ maintenance - of - invariant ] ] maintenance of invariant : + + + + + + + + + + + + + + + + + + + + + + + + + let @xmath265 be given and assume the invariant holds at the beginning of the @xmath169th iteration of the for - loop .
we show that it also holds at the beginning of the @xmath266th iteration .
let @xmath91 be a shortest path in @xmath239 from @xmath19 to a vertex @xmath267 .
we show that there is a @xmath268 such that @xmath269 approximates the weight of @xmath91 up to a factor of @xmath270 .
the second inequality of the invariant will then follow from lemma [ lem : portal ] .
the first inequality follows since @xmath271 is the weight of an actual path in @xmath239 from @xmath19 to @xmath113 for each @xmath272 .
let @xmath273 be the last vertex on @xmath91 such that the subpath of @xmath91 from @xmath19 to @xmath273 is contained in @xmath242 .
in particular , @xmath274 . by the invariant at the beginning of the @xmath169th iteration ,
there is a portal @xmath275 such that @xmath276 .
assume first that @xmath277 .
then @xmath91 is contained in @xmath242 and @xmath80 approximates the distance from @xmath19 to @xmath278 up to a factor of @xmath279 with the path starting with @xmath280 and followed by edges of @xmath212 .
now assume that @xmath281 .
consider first the case where @xmath282 ( figure [ fig : phaseidata ] with @xmath59 playing the role of @xmath283 ) .
we have the precomputed portal set @xmath284 and it contains a portal @xmath285 such that @xmath286 .
hence , @xmath287 now consider the other case where @xmath281 and @xmath288 .
then a shortest path from @xmath283 to @xmath273 in @xmath237 contains a vertex @xmath289 and the subpath from @xmath283 to @xmath290 is contained in @xmath291 .
we have a precomputed portal set @xmath292 containing a portal @xmath285 such that @xmath293 .
this gives @xmath294 as desired .
[ [ termination ] ] termination : + + + + + + + + + + + + the invariant shows that at the beginning of the @xmath209th iteration , for all @xmath295 , there is a @xmath296 such that @xmath297 .
line @xmath252 is identical to line @xmath251 for @xmath298 except that @xmath254 is replaced by @xmath258 so line @xmath252 increases the approximation by a factor of @xmath299 .
below we choose @xmath254 and @xmath258 such that @xmath300 .
this will imply lemma [ lem : phasei ] .
obtaining shortcuts in line @xmath301 can be done in @xmath302 time .
lines @xmath232 and @xmath233 take @xmath303 time as @xmath304 and distances @xmath234 for @xmath206 are precomputed .
we first show how a single iteration @xmath169 of the for - loop in lines @xmath259@xmath251 can be implemented to run in @xmath305 time .
then we show how to improve it to @xmath306 .
finding @xmath240 takes @xmath307 time . as @xmath308 , @xmath243
can be found in time @xmath309 .
face @xmath212 is obtained in @xmath310 time by sorting the vertices according to their cyclic ordering in an euler tour of face @xmath213 .
graph @xmath80 and distances @xmath311 for @xmath217 can then be obtained in @xmath312 time . to obtain @xmath313 ,
apply the portal construction algorithm in the proof of lemma [ lem : portal ] to each shortest path in @xmath314 restricted to @xmath80 .
this takes @xmath315 time .
line @xmath252 takes @xmath225 time .
we improve the time bound to @xmath306 by avoiding the sorting step above . instead , we omit adding edges of @xmath212 to @xmath80 and apply a variant of the portal construction algorithm in the proof of lemma [ lem : portal ] to each shortest path @xmath316 restricted to @xmath80 .
we describe this variant in the following . first observe that @xmath317 can be partitioned into @xmath188 subsets @xmath318 where each subset
@xmath319 is either a singleton set consisting of a vertex in @xmath214 , a singleton set consisting of an endpoint of an edge in @xmath240 , or a set @xmath320 for each @xmath321 . assume that each set @xmath319 is sorted along @xmath322 in non - decreasing distance from the source @xmath66 in shortest path tree @xmath18 ; this sorting can be done in the preprocessing step .
the first portal @xmath127 to be added to @xmath313 is the vertex @xmath323 minimizing @xmath324 .
we can identify @xmath127 in @xmath325 time .
let @xmath130 be the vertex of @xmath322 farthest from @xmath66 .
as in the proof of lemma [ lem : portal ] , we only describe the algorithm for adding portals to @xmath326 $ ] ; adding portals along @xmath327 $ ] is symmetric . for each set @xmath319 , we keep a pointer to the first vertex in its sorted order .
we then make a single pass over the sets @xmath319 and for each of them move its pointer forward to the first vertex in @xmath326 $ ] ( if any ) for which the distance to it can not be approximated by going through the previously added portal @xmath127 , i.e. , the first vertex @xmath328 $ ] such that @xmath329 . among the vertices with pointers to them , the one closest to @xmath127 in @xmath322 is then added to @xmath313 as the next portal , and @xmath127 is updated to this vertex .
additional passes are made until the pointers have moved past all vertices of their respective @xmath319 sets .
correctness follows since the set of portals formed is the same as that obtained by the portal - construction algorithm of thorup .
running time is @xmath306 . to see this ,
note that each pass ( except possibly the last ) adds at least one portal to @xmath313 so the number of passes is @xmath188 .
furthermore , each pass takes @xmath330 time where @xmath331 is the total number of vertices visited in that pass over all sets @xmath319 .
since the total number of vertices visited over all passes is @xmath332 , the time bound follows . we can obtain a stretch of @xmath333 for the approximate distances obtained in the final iteration of phase i as follows . since @xmath334 when @xmath335 , we pick @xmath336 to obtain @xmath337 which is at most @xmath338 when @xmath339 . picking @xmath340 gives @xmath341 for @xmath41 smaller than some positive constant .
this shows lemma [ lem : phasei ] .
phase ii takes as input the sets @xmath256 and @xmath342 with associated approximate distances @xmath343 and @xmath344 that were output by phase i. the output of phase ii has the properties stated in the following lemma .
denote by @xmath29 the shortest path separator in @xmath10 that separates @xmath163 into @xmath161 and @xmath228 . in section [ sec : approxdist ]
, we efficiently obtain from this output an approximate @xmath28-distance .
[ lem : phaseii ] phase ii for @xmath19 can be implemented to run in @xmath345 time using @xmath346 space , given the output from phase i. for the output @xmath347 from phase ii , we have @xmath348 and for any @xmath349 , there is a vertex @xmath350 such that @xmath351 .
the preprocessing for phase ii consists of the following four steps : [ [ step-1 ] ] step @xmath301 : + + + + + + + + + + + + + + + + + + + + + for each @xmath196 and each of the at most two regions @xmath352 with @xmath353 and @xmath354 where @xmath355 is the parent of @xmath352 in @xmath81 , we form a @xmath356-portal set @xmath357 of @xmath358 . from this
we form and store a subset @xmath359 of @xmath360 .
this subset contains @xmath361 .
in addition , for every @xmath362 and every left or right side @xmath363 ( see section [ subsec : recdecomp ] ) such that @xmath91 and @xmath113 are contained in the same path of @xmath18 , @xmath359 contains the successor and predecessor ( if any ) of @xmath113 on @xmath91 ; figure [ fig : pw ] gives an illustration .
note that @xmath364 so @xmath365 . for any @xmath366
there is a @xmath367 with @xmath368 . since @xmath369 the shortest path in @xmath358 from @xmath370 to @xmath30 intersects @xmath360 .
hence there is a @xmath371 which is either @xmath370 , the successor or predecessor of @xmath370 on @xmath91 , or a corner in @xmath372 such that @xmath373 .
we also have @xmath374 .
[ [ step-2 ] ] step @xmath232 : + + + + + + + + + + + + + + + + + + + + + for each shortcut @xmath182 and each @xmath375 , store a @xmath376-portal set @xmath377 of @xmath378 of size @xmath225 ; call this a _ type 1 _ portal set .
[ [ step-3 ] ] step @xmath233 : + + + + + + + + + + + + + + + + + + + + + for any shortcut @xmath182 , _ dual portal set _
@xmath379 is the set of vertices @xmath380 for which a vertex @xmath59 exists with @xmath381 , @xmath382 , such that @xmath383 ; see figure [ fig : phaseiidata ] .
define @xmath384 .
note that @xmath385 .
for each @xmath386 , store a @xmath387-portal set @xmath388 of @xmath378 of size @xmath225 together with distances @xmath389 for each @xmath390 .
refer to it as a _ type 2 _ portal set .
the definition of type 2 portal sets is rather technical so let us give the high - level idea for introducing them ; see figure [ fig : phaseiidata ] in the following . as in phase
i , we jump along shortcuts @xmath391 in phase ii ; a detailed description is given in the next subsection . in phase ii , we need approximate distances in @xmath10 from certain portals @xmath113 in @xmath237 to @xmath392
. however , @xmath113 might also be present in @xmath213 . in this case
, we can not afford to associate portal sets with @xmath113 and shortcut @xmath391 since @xmath113 may occur in several regions of @xmath81 ( see lemma [ lem : regionpath ] ) .
however , vertices @xmath59 for which @xmath352 is sandwiched in between @xmath242 and @xmath239 can pay for dual portal set @xmath393 and the associated type 2 portal sets . as we show below
, we can obtain an approximate distance from @xmath113 to any @xmath394 by first going along @xmath18 from @xmath113 to a nearby @xmath395 , then along a shortest path in @xmath10 from @xmath396 to a portal @xmath397 in the type 2 portal set of @xmath396 , and finally from @xmath370 to @xmath398 along @xmath18 . using hashing
, we can access each type 1 and type 2 portal set in @xmath82 time from the vertex and the shortcut defining it .
[ [ step-4 ] ] step @xmath259 : + + + + + + + + + + + + + + + + + + + + + for each shortcut @xmath182 and for any shortest path @xmath399 , we keep a veb - tree , allowing us to find the successor / predecessor of any vertex of @xmath91 in the subset @xmath400 in @xmath51 time . with hashing ,
space required for the veb - tree is @xmath401 @xcite .
as mentioned in @xcite , both space and query bounds can be made deterministic .
[ lem : typeiiandiiiportalset ] the total space required for phase ii is @xmath402 . the total size of portal sets @xmath403 over all @xmath101 is @xmath404 .
a proof similar to that of lemma [ lem : typeiportalset ] shows that the total size of all type @xmath301 portal sets is @xmath405 . to bound the size of type 2 portal sets , consider two shortcuts @xmath182 and @xmath406 . by lemma [ lem : regionpath ] , @xmath379 and @xmath407
are disjoint if @xmath408 and @xmath409 are not contained in the same leaf - to - root path @xmath21 in @xmath81 .
for any @xmath196 , there are only @xmath51 shortcuts @xmath410 where @xmath411 .
hence @xmath412 summed over all shortcuts @xmath182 is @xmath413 .
hence , the total size of all dual portal sets @xmath379 , and hence also the total size of all veb trees , is @xmath405 . recall that the number of regions is @xmath31 .
each region has @xmath82 corners and each set @xmath183 has constant size so the total size of all sets @xmath414 is @xmath405 .
each element of these sets has a type 2 portal set of size @xmath225 .
let @xmath415 be the path from @xmath161 to the root @xmath10 of @xmath81 . in the following , for any vertex @xmath59 , denote by @xmath352 the region ( if any ) such that @xmath416 , @xmath353 , and @xmath354 where @xmath355 is the parent of @xmath352 in @xmath81 .
observe that @xmath417 .
phase ii for @xmath19 takes the output from phase i and produces output satisfying lemma [ lem : phaseii ] . ' '' '' + ddd============**phase ii * * for @xmath19 : + + 1 .
let @xmath202 be the shortcuts from @xmath418 to @xmath419 + 2 .
let @xmath420 ( portal set output from phase i ) + 3 .
let @xmath89 be the graph initially consisting of edges @xmath421 for all @xmath422 + 4 .
for @xmath423 to @xmath209 + 5 .
add to @xmath89 edges @xmath424 for all @xmath425 and @xmath426 + 6 .
for each @xmath261 and each @xmath427 + 7 .
add to @xmath89 edge @xmath428 for all @xmath429 + 8 .
let @xmath430 + 9 .
+ ' '' '' we give a high - level description of phase ii before going into details .
pseudocode can be seen in figure [ fig : phaseii ] .
the algorithm traverses shortcuts @xmath202 from @xmath418 to the root @xmath419 of @xmath81 and incrementally constructs a graph @xmath89 which at termination will satisfy lemma [ lem : phaseii ] . in line @xmath233 ,
edges of @xmath89 represent approximate paths found in phase i. these paths correspond to subpaths of the final full paths in @xmath10 ( corresponding to the final @xmath89 ) and the subpaths are prefixes of these full paths that are contained in @xmath161 .
consider the @xmath169th iteration of the for - loop . in line @xmath241 , we check if any subpath endpoint @xmath261 disappears as a boundary vertex when jumping from @xmath242 to @xmath239 .
if so , we can extend the subpath to full paths @xmath432 for each @xmath426 .
the other interesting case is when @xmath433 .
then we do not have a type @xmath301 portal set associated with @xmath113 and @xmath391 but it might be that some separator vertices @xmath59 of @xmath29 that are present in @xmath392 are no longer present in @xmath213 and we need to ensure that there is a good path in @xmath434 from @xmath113 to @xmath59 .
this case is handled in lines @xmath244 and @xmath249 where we ensure such a good path from @xmath113 to @xmath59 by using the type @xmath232 portal sets associated with vertices of @xmath435 that are close to @xmath113 . to show correctness ,
i.e. , that the set output in line @xmath436 satisfies lemma [ lem : phaseii ] , let @xmath59 be any vertex on @xmath29 and let @xmath21 be a shortest path in @xmath10 from @xmath19 to @xmath59 .
let @xmath398 be the last vertex on @xmath21 such that @xmath437 $ ] is contained in @xmath161 .
note that @xmath438 . by lemma [ lem : phasei ]
, there is a @xmath439 such that @xmath440 . since @xmath417 , we have @xmath416 .
since @xmath441 , we also have @xmath442 .
let @xmath443 and @xmath444 be the shortcuts such that @xmath445 , @xmath446 , and @xmath447 , @xmath448 .
we consider two cases in the following : @xmath449 and @xmath450 .
[ [ case-1-i_p_1leq - i_w - figurefigcase1 ] ] case @xmath301 , @xmath449 ( figure [ fig : case1 ] ) : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + consider iteration @xmath451 of the for - loop .
since @xmath452 , we have @xmath453 . since @xmath448
, it follows that @xmath454 .
hence , the final @xmath89 contains an edge @xmath455 for each portal @xmath456 ( line @xmath241 ) . since @xmath457 and since @xmath458
, we must have @xmath459 so there is a portal @xmath460 satisfying @xmath461 .
note that @xmath462 .
the path in the final graph @xmath89 consisting of edges @xmath463 and @xmath464 followed by the path in @xmath18 from @xmath465 to @xmath59 has weight at most @xmath466 [ [ case-2-i_p_1-i_w - figurefigcase2 ] ] case @xmath232 , @xmath450 ( figure [ fig : case2 ] ) : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + consider iteration @xmath467 of the for - loop . since @xmath468 , there is a portal @xmath469 such that @xmath470 ( see step @xmath301 of the preprocessing ) . if @xmath471 then @xmath472
otherwise , there is a vertex @xmath473 such that the path in @xmath18 from @xmath474 to @xmath475 contains @xmath370 .
hence , @xmath476 it follows from the above that there is a vertex @xmath477 such that @xmath478 .
thus , for one of the two choices of @xmath396 in line @xmath244 , we have @xmath479 for that choice of @xmath396 , let @xmath480 be the portal in @xmath481 such that @xmath482 .
the path in @xmath89 consisting of edges @xmath463 ( added in line @xmath233 ) and @xmath483 ( added in line @xmath249 ) followed by the path in @xmath18 from @xmath480 to @xmath59 has weight at most @xmath484 to show lemma [ lem : phaseii ] , note that @xmath485 is at most @xmath486 for any @xmath41 less than some positive constant . as for query time , lines @xmath301 and @xmath232 can be executed in @xmath302 time . by lemma [ lem : phasei ] , adding edges to @xmath89 in line @xmath233 can be done in @xmath487 time .
the total time to find successors and predecessors over all iterations of the for - loop is @xmath488 .
the additional time spent in the for - loop is bounded by the number of edges added to @xmath89 .
the total number of edges added in line @xmath249 is @xmath489 .
note that every @xmath113 considered in line @xmath241 is not included in @xmath313 in line @xmath490 .
hence , we add a total of @xmath491 edges in line @xmath241 .
in this section , we show how to obtain an approximate @xmath28-distance within the time and space stated in theorem [ thm : main ] , given the output of phase ii . we execute phase ii for both @xmath19 and @xmath30 , getting outputs @xmath492 and @xmath493 , respectively .
to find an approximate @xmath28-distance , assume first that @xmath494 .
let @xmath91 be one of the two shortest paths from @xmath66 in @xmath18 bounding @xmath29 and sort the vertices of @xmath495 along @xmath91 . to do the sorting efficiently
, we make a dfs traversal of @xmath18 during preprocessing and label each vertex with an integer time stamp denoting when it was first visited in the traversal .
now , sorting the vertices of @xmath495 along @xmath91 corresponds to integer sorting their precomputed labels . with the algorithm of han and thorup @xcite , this takes @xmath496 time .
we then remove all @xmath497 for which there is another @xmath498 such that @xmath499 .
this can be done in @xmath500 time with two linear scans over @xmath495 , one in sorted order and the other in reverse sorted order ; the first resp .
second scan removes @xmath59 if there is a @xmath398 before resp .
after @xmath59 in the order considered such that the inequality holds .
let @xmath501 be the resulting subset and form a similar subset @xmath502 of @xmath503 .
let @xmath504 be the minimum of @xmath505 over all pairs @xmath506 where @xmath507 $ ] has no interior vertices belonging to @xmath508 .
this takes @xmath500 time .
compute a similar value @xmath509 for the other shortest path @xmath510 of @xmath18 bounding @xmath29 .
the approximate distance output is @xmath511 .
we need to show @xmath512 .
pick @xmath349 such that @xmath513 and assume that @xmath514 ; the case where @xmath515 is symmetric . by lemma [ lem : phaseii ]
, there are vertices @xmath516 and @xmath517 such that @xmath518 and @xmath519 .
pick a pair @xmath520 belonging to @xmath507 $ ] such that they occur on this path in the order @xmath521 and such that @xmath522 $ ] has no interior vertices belonging to @xmath508 . then @xmath523 showing the desired . above
, we assumed that @xmath524 .
if this is not the case , we modify @xmath525 as follows .
partition @xmath526 into maximal - size groups where in each group @xmath527 , all vertices @xmath59 have the same nearest neighbor @xmath398 in @xmath18 belonging to @xmath29 .
replace vertices of @xmath527 by @xmath398 in @xmath525 and instead of approximate distances @xmath528 for @xmath529 , use instead @xmath530 for the approximate distance for @xmath398 .
a similar update is done to @xmath503 , ensuring that @xmath524 .
it is easy to see that the above analysis still carries through .
we have shown theorem [ thm : main ] in the case where @xmath164 and @xmath531 are not on the same leaf - to - root path in @xmath81 .
if instead , say , @xmath532 then @xmath533 and phase i and ii for @xmath19 gives a portal set of @xmath29 .
our algorithm above is modified to find the portal @xmath113 nearest to @xmath30 on @xmath29 and outputs @xmath534 , giving the desired stretch .
this shows theorem [ thm : main ] in the remaining case where @xmath164 and @xmath531 are on the same leaf - to - root path .
we gave a @xmath2-approximate distance oracle for undirected @xmath0-vertex planar graphs and fixed @xmath1 with @xmath3 space and @xmath4 query time which improves the previous best query time - space product from @xmath8 to @xmath9 .
we have not focused on preprocessing time . with a simple implementation
, we should get near - quadratic preprocessing time and it is possible that the exact space - efficient oracle in @xcite can speed this up further to @xmath535 as the number of precomputed distances required by our oracle is only @xmath536 . with techniques from , e.g. , @xcite , we can likely get down to @xmath536 .
the dependency on @xmath41 in the query time - space product is slightly worse ; it is roughly @xmath48 ( @xmath46 when @xmath47 ) compared to @xmath49 in @xcite and roughly @xmath45 in @xcite ( where the latter has a slightly worse dependency on @xmath0 than @xcite ) . using mainly monge properties , we believe it should be possible to replace at least one @xmath45 factor by @xmath37 . getting @xmath537 query time and @xmath538 space for some constant @xmath539 seems problematic with our techniques due to the @xmath540 bottleneck from the use of veb trees when answering queries .
99 s. chechik .
approximate distance oracle with constant query time .
stoc@xmath541 , pp .
p. erds .
extremal problems in graph theory . in theory of graphs and its applications ( proc .
smolenice , @xmath544 ) , czechoslovak acad .
prague , @xmath545 , pp .
m. l. fredman and d. e. willard . surpassing the information theoretic bound with fusion trees .
journal of computer and system sciences , @xmath548(@xmath233):@xmath549@xmath550 , @xmath551 .
y. han and m. thorup .
integer sorting in @xmath552 expected time and linear space .
@xmath553rd annual symposium on foundations of computer science ( focs ) , pp .
@xmath554@xmath555 , @xmath7 .
k. kawarabayashi , p. n. klein , and c. sommer .
linear - space approximate distance oracles for planar , bounded - genus , and minor - free graphs . in @xmath556th
international colloquium on automata , languages and programming ( icalp ) , pp .
@xmath554@xmath557 , @xmath558 .
k. kawarabayashi , c. sommer , and m. thorup .
more compact oracles for approximate distances in undirected planar graphs . in @xmath559th acm - siam symposium on discrete algorithms ( soda ) ,
@xmath560@xmath561 , @xmath562 .
p. n. klein .
preprocessing an undirected planar network to enable fast approximate distance queries . in @xmath563th acm - siam symposium on discrete algorithms ( soda ) ,
@xmath564@xmath565 , @xmath7 .
r. j. lipton and r. e. tarjan . a separator theorem for planar graphs
siam journal on applied mathematics , @xmath547(@xmath232):@xmath566@xmath567 , @xmath568 .
m. mendel and a. naor .
ramsey partitions and proximity data structures .
journal of the european mathematical society , @xmath436(@xmath232):@xmath569@xmath570 , @xmath571 . announced at focs@xmath572 .
s. mozes and c. sommer .
exact distance oracles for planar graphs . in @xmath573rd acm - siam symposium on discrete algorithms ( soda ) ,
@xmath574@xmath575 , @xmath576 .
m. ptracu and l. roditty .
distance oracles beyond the thorup - zwick bound . in @xmath577st ieee symposium on foundations of computer science ( focs ) , @xmath578
. m. ptracu and m. thorup . time - space trade - offs for predecessor search .
@xmath556th annual acm symposium on theory of computing ( stoc ) , pp .
@xmath579@xmath580 , @xmath581 .
m. thorup .
compact oracles for reachability and approximate distances in planar digraphs .
journal of the acm , @xmath577(@xmath244):@xmath582@xmath583 , @xmath6 .
announced at focs @xmath5 .
m. thorup and u. zwick .
approximate distance oracles .
journal of the acm , @xmath584(@xmath301):@xmath301@xmath559 , @xmath585 . announced at stoc@xmath586 .
d. e. willard .
log - logarithmic worst - case range queries are possible in space @xmath587 .
information processing letters , @xmath588(@xmath232):@xmath589@xmath590 , @xmath591 . c. wulff - nilsen .
algorithms for planar graphs and graphs in metric spaces .
phd thesis , university of copenhagen , @xmath578 . c. wulff - nilsen .
approximate distance oracles with improved preprocessing time . in @xmath573rd
acm - siam symposium on discrete algorithms ( soda ) , pp .
@xmath592@xmath593 , @xmath576 .
c. wulff - nilsen .
approximate distance oracles with improved query time . in @xmath559th acm - siam symposium on discrete algorithms ( soda ) ,
@xmath594@xmath595 , @xmath562 . | we consider approximate distance oracles for edge - weighted @xmath0-vertex undirected planar graphs . given fixed @xmath1 , we present a @xmath2-approximate distance oracle with @xmath3 space and @xmath4 query time .
this improves the previous best product of query time and space of the oracles of thorup ( focs @xmath5 , j. acm @xmath6 ) and klein ( soda @xmath7 ) from @xmath8 to @xmath9 . |
in papers @xcite and @xcite , we systematically surveyed period variations of superhumps in su uma - type dwarf novae ( for general information of su uma - type dwarf novae and superhumps , see @xcite ) .
these works indicated that evolution of superhump period ( @xmath2 ) is generally composed of three distinct stages : early evolutionary stage with a longer superhump period ( stage a ) , middle stage with systematically varying periods ( stage b ) , final stage with a shorter , stable superhump period ( stage c ) . in short-@xmath2 systems , these stages are often observed as distinct segments ( e.g. see a representative example in figure [ fig : j1025comp ] : @xmath3 in this figure corresponds to stage a , @xmath4 with a parabolic @xmath5 curve corresponds to stage b with an increasing period , and @xmath6 corresponds to stage c ) . although systems with longer @xmath2 tend to show less distinct stages , there is often a systematic `` break '' in the @xmath5 diagram , which we attribute to a stage b - c transition based on the knowledge in the systematic variation of the @xmath5 diagrams against @xmath2 ( cf .
subsection 3.2 and figure 4 in @xcite ) .
it was also shown that the period derivatives ( @xmath7 ) during stage b is correlated with @xmath2 , or binary mass - ratios ( @xmath8 ) .
we used similar , but either phenomenologically or conceptually different , terminologies referring the superhumps observed during the late stages of superoutbursts or post - superoutburst stages . for reader s convenience , we summarized them in table [ tab : latesuperhumps ] .
note that the use of the terminology is not the same between different authors .
although some authors use the term `` late superhumps '' for our stage c superhumps , we restrict the usage of ( traditional ) late superhumps to superhumps with an @xmath90.5 phase shift .
.types of late - stage superhumps . [ cols="^,<",options="header " , ]
we observed two superoutbursts of this object in 2010 and 2011 .
the 2010 superoutburst was well - observed except for the initial part ( table [ tab : foandoc2010 ] ) .
the 2011 superoutburst was only partly observed ( table [ tab : foandoc2011 ] ) .
the 2010 observation confirmed the presence of a stage b - c transition .
the later part of the 2011 observation probably caught a part of stage c , and we listed values based on this interpretation ( figure [ fig : foandcomp ] ) .
fo and has a short supercycle of 100140 d and has similar outburst properties to v503 cyg , v344 lyr @xcite and mn dra @xcite , all of which are known to show distinct negative superhumps in quiescence ( @xcite ; @xcite ; @xcite ) .
the recent discovery of long - persisting negative superhumps in er uma @xcite and its unusual development of superhumps @xcite has led to a question whether these system show unusual behavior similar to er uma . in the case of fo and ,
the @xmath5 behavior and variation of the superhump profile ( figure [ fig : foandprof ] ) of the 2010 superoutburst looked relatively normal for an ordinary su uma - type dwarf nova .
( 88mm,70mm)fig1.eps ( 88mm,110mm)fig2.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55486.9737 & 0.0011 & @xmath120.0041 & 108 + 1 & 55487.0494 & 0.0006 & @xmath120.0026 & 108 + 2 & 55487.1200 & 0.0006 & @xmath120.0064 & 107 + 5 & 55487.3456 & 0.0006 & @xmath120.0036 & 77 + 6 & 55487.4202 & 0.0005 & @xmath120.0033 & 74 + 7 & 55487.4933 & 0.0006 & @xmath120.0045 & 71 + 8 & 55487.5676 & 0.0005 & @xmath120.0044 & 73 + 9 & 55487.6410 & 0.0006 & @xmath120.0053 & 74 + 10 & 55487.7176 & 0.0009 & @xmath120.0030 & 45 + 18 & 55488.3144 & 0.0008 & @xmath120.0005 & 64 + 19 & 55488.3869 & 0.0007 & @xmath120.0022 & 76 + 20 & 55488.4619 & 0.0006 & @xmath120.0015 & 77 + 21 & 55488.5376 & 0.0005 & @xmath120.0001 & 77 + 22 & 55488.6084 & 0.0007 & @xmath120.0036 & 77 + 23 & 55488.6841 & 0.0010 & @xmath120.0022 & 73 + 32 & 55489.3582 & 0.0007 & 0.0034 & 158 + 33 & 55489.4328 & 0.0005 & 0.0037 & 153 + 34 & 55489.5064 & 0.0010 & 0.0030 & 102 + 35 & 55489.5797 & 0.0013 & 0.0021 & 77 + 36 & 55489.6530 & 0.0009 & 0.0010 & 77 + 45 & 55490.3265 & 0.0005 & 0.0060 & 113 + 46 & 55490.4021 & 0.0026 & 0.0073 & 37 + 47 & 55490.4731 & 0.0019 & 0.0040 & 34 + 48 & 55490.5491 & 0.0023 & 0.0057 & 38 + 49 & 55490.6240 & 0.0017 & 0.0064 & 38 + 50 & 55490.6966 & 0.0018 & 0.0047 & 31 + 59 & 55491.3631 & 0.0017 & 0.0027 & 35 + 60 & 55491.4400 & 0.0017 & 0.0053 & 34 + 61 & 55491.5151 & 0.0017 & 0.0061 & 37 + 62 & 55491.5880 & 0.0015 & 0.0047 & 39 + 63 & 55491.6643 & 0.0043 & 0.0067 & 28 + 86 & 55493.3690 & 0.0056 & 0.0029 & 23 + 87 & 55493.4436 & 0.0018 & 0.0032 & 25 + 88 & 55493.5131 & 0.0028 & @xmath120.0016 & 25 + 89 & 55493.5893 & 0.0018 & 0.0004 & 25 + 90 & 55493.6619 & 0.0017 & @xmath120.0013 & 25 + 99 & 55494.3307 & 0.0018 & @xmath120.0010 & 27 + 100 & 55494.3993 & 0.0022 & @xmath120.0067 & 34 + 101 & 55494.4807 & 0.0008 & 0.0003 & 70 + 102 & 55494.5566 & 0.0011 & 0.0020 & 91 + 103 & 55494.6341 & 0.0012 & 0.0052 & 59 + 104 & 55494.7005 & 0.0032 & @xmath120.0027 & 19 + 113 & 55495.3649 & 0.0017 & @xmath120.0068 & 33 + 116 & 55495.5914 & 0.0021 & @xmath120.0032 & 25 + 117 & 55495.6692 & 0.0021 & 0.0004 & 26 + 130 & 55496.6254 & 0.0012 & @xmath120.0092 & 38 + 131 & 55496.7016 & 0.0017 & @xmath120.0072 & 21 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55578.4067 & 0.0009 & @xmath120.0017 & 77 + 39 & 55581.3172 & 0.0015 & 0.0012 & 81 + 40 & 55581.3916 & 0.0008 & 0.0011 & 158 + 41 & 55581.4702 & 0.0015 & 0.0052 & 61 + 51 & 55582.2089 & 0.0006 & @xmath120.0017 & 55 + 52 & 55582.2825 & 0.0004 & @xmath120.0027 & 89 + 53 & 55582.3594 & 0.0006 & @xmath120.0003 & 40 + 54 & 55582.4331 & 0.0011 & @xmath120.0012 & 32 + + + + we observed the 2011 superoutburst ( table [ tab : v402andoc2011 ] ) .
the data demonstrated stage b with a positive @xmath13 of @xmath14 and stage c first time for this object .
a comparison of @xmath5 diagrams for different superoutbursts is shown in figure [ fig : v402andcomp2 ] .
( 88mm,70mm)fig3.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55582.3645 & 0.0004 & 0.0037 & 49 + 14 & 55583.2496 & 0.0009 & 0.0005 & 38 + 15 & 55583.3127 & 0.0003 & 0.0002 & 116 + 16 & 55583.3781 & 0.0006 & 0.0021 & 63 + 35 & 55584.5783 & 0.0004 & @xmath120.0031 & 46 + 36 & 55584.6425 & 0.0007 & @xmath120.0024 & 37 + 46 & 55585.2761 & 0.0005 & @xmath120.0032 & 59 + 47 & 55585.3392 & 0.0012 & @xmath120.0035 & 52 + 78 & 55587.3085 & 0.0006 & @xmath120.0011 & 58 + 94 & 55588.3279 & 0.0008 & 0.0032 & 78 + 125 & 55590.3008 & 0.0027 & 0.0093 & 17 + 141 & 55591.3067 & 0.0009 & 0.0001 & 65 + 142 & 55591.3704 & 0.0008 & 0.0004 & 39 + 156 & 55592.2548 & 0.0033 & @xmath120.0035 & 39 + 157 & 55592.3208 & 0.0012 & @xmath120.0009 & 126 + 158 & 55592.3831 & 0.0032 & @xmath120.0020 & 37 + + + + we observed the 2010 superoutburst of this object ( = pg 0149@xmath0138 , sdss [email protected] ) .
the outburst was well - observed since its earliest stage .
the times of superhump maxima are listed in table [ tab : bgarioc2010 ] .
c stages are clearly present .
there was a marginal hint of positive @xmath13 during stage b , as in the 2009 superoutburst @xcite .
a comparison of @xmath5 diagrams is shown in figure [ fig : bgaricomp ] .
( 88mm,70mm)fig4.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55536.6520 & 0.0014 & @xmath120.0391 & 79 + 1 & 55536.7314 & 0.0014 & @xmath120.0446 & 70 + 2 & 55536.8151 & 0.0018 & @xmath120.0459 & 64 + 19 & 55538.3106 & 0.0005 & 0.0051 & 87 + 20 & 55538.3996 & 0.0007 & 0.0090 & 89 + 21 & 55538.4834 & 0.0005 & 0.0079 & 87 + 22 & 55538.5715 & 0.0004 & 0.0110 & 134 + 23 & 55538.6562 & 0.0004 & 0.0107 & 69 + 32 & 55539.4208 & 0.0003 & 0.0106 & 92 + 33 & 55539.5055 & 0.0003 & 0.0104 & 80 + 35 & 55539.6763 & 0.0003 & 0.0112 & 69 + 47 & 55540.6960 & 0.0006 & 0.0112 & 76 + 48 & 55540.7813 & 0.0007 & 0.0115 & 58 + 55 & 55541.3738 & 0.0002 & 0.0092 & 91 + 56 & 55541.4574 & 0.0002 & 0.0078 & 56 + 66 & 55542.3060 & 0.0004 & 0.0067 & 54 + 67 & 55542.3904 & 0.0005 & 0.0061 & 53 + 68 & 55542.4753 & 0.0004 & 0.0061 & 56 + 78 & 55543.3267 & 0.0006 & 0.0078 & 83 + 91 & 55544.4323 & 0.0004 & 0.0087 & 56 + 102 & 55545.3683 & 0.0027 & 0.0100 & 29 + 148 & 55549.2775 & 0.0007 & 0.0105 & 48 + 175 & 55551.5617 & 0.0013 & 0.0004 & 52 + 176 & 55551.6495 & 0.0025 & 0.0032 & 41 + 187 & 55552.5737 & 0.0017 & @xmath120.0072 & 54 + 211 & 55554.6103 & 0.0014 & @xmath120.0100 & 51 + 221 & 55555.4415 & 0.0017 & @xmath120.0286 & 82 + + + + v496 aur is a dwarf nova discovered by @xcite , reporting the maximum unfiltered ccd magnitude of 15.7 , and the absence of the quiescent counterpart on the palomer observatory sky survey plates .
based on spectroscopy during outburst , they confirmed the dwarf nova - type classification .
@xcite reported another outburst in 1998 .
unpublished photometry by one of the authors ( t. kato ) during this 1998 outburst did not yield significant superhumps .
the 2010 outburst of this object was by chance detected by the catalina real - time transient survey ( crts , @xcite).http://nesssi.cacr.caltech.edu / catalina/@xmath15 .
for the information of the individual catalina cvs , see @xmath16http://nesssi.cacr.caltech.edu / catalina / allcv.html@xmath15 . ]
subsequent observations confirmed the presence of superhumps ( vsnet - alert 12430 , 12439 ; figure [ fig : v496aurshpdm ] ) .
the times of superhump maxima are listed in table [ tab : v496auroc2010 ] .
the relatively small amplitudes might suggest that the object had already entered the late stage ( possibly stage c ) of the superoutburst .
observations taken four days later ( december 5 ) did not detect significant superhump signals , and were excluded from this analysis .
the improved astrometry by the crts is , ( j2000.0 ) .
( 88mm,110mm)fig5.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55530.4937 & 0.0011 & 0.0007 & 60 + 1 & 55530.5521 & 0.0012 & @xmath120.0021 & 67 + 2 & 55530.6167 & 0.0012 & 0.0014 & 52 + 19 & 55531.6556 & 0.0012 & 0.0005 & 48 + 20 & 55531.7158 & 0.0015 & @xmath120.0005 & 61 + + + + we report on two more superoutbursts in 2007 and 2010 ( tables [ tab : ttboooc2007 ] and [ tab : ttboooc2010 ] ) .
the 2007 observations were performed during the middle stage of the superoutburst .
the 2010 observations fairly well covered the superoutburst except the final stage .
the resultant @xmath13 of @xmath17 confirms the rather large positive @xmath13 for this @xmath2 or @xmath18 @xcite .
a comparison of @xmath5 diagrams is presented in figure [ fig : ttboocomp ] .
the figure might suggest that a stage b
c transition already occurred before the 2007 observation .
( 88mm,70mm)fig6.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 54200.1584 & 0.0006 & 0.0013 & 168 + 1 & 54200.2326 & 0.0006 & @xmath120.0024 & 98 + 13 & 54201.1714 & 0.0005 & 0.0019 & 167 + 26 & 54202.1810 & 0.0005 & @xmath120.0009 & 168 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55311.0879 & 0.0001 & 0.0022 & 232 + 14 & 55312.1801 & 0.0003 & 0.0008 & 161 + 64 & 55316.0792 & 0.0008 & @xmath120.0058 & 140 + 65 & 55316.1623 & 0.0007 & @xmath120.0009 & 163 + 78 & 55317.1774 & 0.0009 & @xmath120.0012 & 146 + 79 & 55317.2561 & 0.0006 & @xmath120.0006 & 152 + 90 & 55318.1152 & 0.0006 & @xmath120.0008 & 152 + 91 & 55318.1953 & 0.0009 & 0.0012 & 154 + 103 & 55319.1347 & 0.0007 & 0.0032 & 148 + 104 & 55319.2114 & 0.0006 & 0.0018 & 156 + + + + although gx cas was observed in multiple instances @xcite , no significant variation of period was recorded , mostly due to the short baseline of observations . during the 2010 superoutburst
, we finally obtained a relatively dense set of observations during the later half of the superoutburst .
the times of superhump maxima are listed in table [ tab : gxcasoc2010 ] .
surprisingly , they suggest a positive @xmath13 of @xmath19 for @xmath20 ( see also figure [ fig : gxcasprof ] ; the conclusion is unchanged even using a rather discrepant point at @xmath21 ; @xmath22 corresponds to a point during the rapid decline phase ) .
a combined @xmath5 diagram ( figure [ fig : gxcascomp2 ] ) also seems to support a general trend of a positive @xmath13 .
as discussed in @xcite ( subsection 4.10 ) , most long-@xmath18 objects generally show relatively large negative @xmath13 , while some objects ( mostly rarely outbursting objects ) have nearly zero @xmath13 .
gx cas is clearly an exception .
although this conclusion may have been affected by the lack of observations during the early stage of the superoutburst , the object will be worth studying in more details .
being a bright object , the object would be a good target for a radial - velocity study in calibrating @xmath18@xmath23 relation in long - period systems .
( 88mm,110mm)fig7.eps ( 88mm,70mm)fig8.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55500.5225 & 0.0004 & 0.0065 & 154 + 1 & 55500.6148 & 0.0003 & 0.0058 & 196 + 10 & 55501.4448 & 0.0003 & @xmath120.0008 & 203 + 11 & 55501.5370 & 0.0003 & @xmath120.0015 & 202 + 12 & 55501.6303 & 0.0005 & @xmath120.0013 & 206 + 16 & 55502.0047 & 0.0005 & 0.0014 & 285 + 17 & 55502.0973 & 0.0008 & 0.0010 & 250 + 18 & 55502.1891 & 0.0006 & @xmath120.0002 & 217 + 19 & 55502.2811 & 0.0009 & @xmath120.0011 & 230 + 20 & 55502.3763 & 0.0009 & 0.0011 & 208 + 21 & 55502.4677 & 0.0004 & @xmath120.0004 & 207 + 26 & 55502.9245 & 0.0154 & @xmath120.0085 & 59 + 27 & 55503.0226 & 0.0013 & @xmath120.0033 & 158 + 28 & 55503.1215 & 0.0020 & 0.0026 & 234 + 29 & 55503.2094 & 0.0010 & @xmath120.0025 & 112 + 30 & 55503.2999 & 0.0008 & @xmath120.0049 & 135 + 37 & 55503.9565 & 0.0018 & 0.0011 & 181 + 38 & 55504.0456 & 0.0013 & @xmath120.0028 & 273 + 39 & 55504.1385 & 0.0008 & @xmath120.0029 & 271 + 40 & 55504.2333 & 0.0012 & @xmath120.0011 & 130 + 41 & 55504.3213 & 0.0014 & @xmath120.0060 & 108 + 48 & 55504.9795 & 0.0007 & 0.0014 & 205 + 49 & 55505.0736 & 0.0010 & 0.0026 & 219 + 50 & 55505.1585 & 0.0013 & @xmath120.0055 & 141 + 51 & 55505.2640 & 0.0039 & 0.0071 & 99 + 59 & 55506.0123 & 0.0025 & 0.0118 & 209 + 60 & 55506.1012 & 0.0015 & 0.0076 & 206 + 62 & 55506.2761 & 0.0042 & @xmath120.0033 & 100 + 71 & 55507.1118 & 0.0071 & @xmath120.0043 & 119
+ + + + ht cas is renowned for its deep eclipses and for its historical superoutburst in 1985 .
there were only fragmentary observations of superhumps during the 1985 superoutburst ( @xcite ; reanalyzed by @xcite ) .
although there were known normal outbursts ( 1987 , 1989 , 1995 , 1997 , 1998 , 1999 , 2002 , 2008 ) since then , there have been no confirmed superoutburst .
the 1995 normal outburst was particularly well observed @xcite .
the 2010 superoutburst was detected by t. parsons on november 2.43 ut at a visual magnitude of 12.9 ( cvnet - discussion 1404 ) .
the object further brightened , and growing superhumps were detected @xmath91.5 d later ( vsnet - alert 12353 , 12359 ; figure [ fig : htcasearly ] ) .
the superhumps ( figure [ fig : htcasshpdm ] ) quickly developed to the full amplitude and slowly decayed . due to the overlapping eclipses and very strong effect of beat phenomenon ( figure [ fig : htcasbeat ] ) , we first subtracted the template orbital variation ( cf .
figure [ fig : htcasorbprof ] ) scaled by fitting individual eclipses ( the results were globally smoothed ; cf . figure [ fig : htcasbeat]c ) and removed the part ( within 0.11 @xmath18 of eclipses ) most strongly affected by eclipses , and applied the usual method of fitting maxima .
the ephemeris of eclipse used in this procedure is given in equation [ equ : htcasecl ] , which were determined by one of the authors ( h. maehara ) using published and newly obtained eclipse timings since 2009 january 1 .
@xmath24 although this procedure lacks the detailed treatment of variations in eclipse profiles , it will be sufficient for the present purpose of measuring the global behavior of superhumps .
the times of superhump maxima are listed in table [ tab : htcasoc2010 ] .
three stages of a
c are apparently present ( figure [ fig : htcashumpall ] ) .
stage a and the middle of stage b were very strongly affected by overlapping eclipses . by rejecting the most severely affected maxima ( @xmath25 ) , we obtained an almost zero @xmath26 period derivative for stage b ( @xmath27 ) .
the other values are listed in table [ tab : perlist ] . as judged from the recorded period
, @xcite seems to have observed stage c superhumps .
the resultant period variation was generally normal for an su uma - type dwarf nova with @xmath18 of 0.07365 d , despite the long lack of superoutbursts .
there were no detectable features of early superhumps , which likely excludes the wz sge - like phenomenon suggested for the 1985 superoutburst ( cf .
the very large amplitude of the outburst reported during the 1985 outburst may have resulted from an old aavso scale of comparison stars , combined with strong orbital / superhump modulations . even after the superoutburst , superhumps with slightly shorter periods persisted at least for 16 d ( table [ tab : htcasoc2010post ] ; maxima were measured in the same way as in the superoutburst ) .
there was no phase jump following the stage c superhumps ( figure [ fig : htcashumpall]a , c ) , and these superhumps are not considered to be traditional " late superhumps ( see a discussion in @xcite , subsection 4.2 ) .
before superhumps appeared , there was likely an indication of orbital humps ( figure [ fig : htcasorbprof]a ) .
the orbital humps were not evident during the plateau phase of the superoutburst , and there was a slight bump around orbital phase 0.5 , which might look like a reflection effect on the secondary ( figure [ fig : htcasorbprof]b ) .
the post - superoutburst phase was dominated by double - wave modulations , which probably arise from ellipsoidal variation of the secondary ( figure [ fig : htcasorbprof]c ) .
no evident orbital humps were detected during this phase .
figure [ fig : htcasorbdep ] shows a more detailed dependence of orbital profiles during the plateau phase on the superhump phase ( @xmath28 ) .
the superhump phases were defined so that @xmath29 corresponds to superhump maxima .
although there was a feature resembling orbital humps for @xmath30 , this feature was not persistently present at other @xmath28 .
the phase 0.5 bump was not clearly present independent of @xmath28 , and this bump ( in averaged profile ) appears to have originated from complex features outside the eclipses for @xmath31 . since these features have different profiles and times of maxima for different @xmath28 , we regard it less likely that they arise from the reflection effect on the secondary ( cf .
these features more likely represent the geometrical effect arising from the non - axisymmetrically extended vertical structure resulting from the heating by the superhump source .
figure [ fig : htcasorbdeplate ] illustrate the same dependence after the rapid decline from the superoutburst ( bjd 24555142455522 ) .
although the general tendency of variations outside the eclipses were similar to those during the plateau phase , eclipses became sharper for @xmath32 and a reflection - like effect became prominent .
( 160mm,190mm)fig9.eps ( 88mm,110mm)fig10.eps ( 88mm,110mm)fig11.eps ( 88mm,110mm)fig12.eps ( 88mm,172mm)fig13.eps ( 88mm,110mm)fig14.eps ( 88mm,110mm)fig15.eps cccccc @xmath10 & max & error & @xmath5 & phase & @xmath11 + 0 & 55504.1036 & 0.0004 & @xmath120.0263 & 0.72 & 1374 + 1 & 55504.1706 & 0.0016 & @xmath120.0355 & 0.63 & 1133 + 2 & 55504.2600 & 0.0006 & @xmath120.0223 & 0.85 & 668 + 3 & 55504.3479 & 0.0004 & @xmath120.0106 & 0.04 & 587 + 4 & 55504.4346 & 0.0010 & @xmath120.0000 & 0.22 & 281 + 5 & 55504.5006 & 0.0003 & @xmath120.0101 & 0.11 & 188 + 6 & 55504.5807 & 0.0004 & @xmath120.0062 & 0.20 & 335 + 7 & 55504.6576 & 0.0002 & @xmath120.0055 & 0.24 & 465 + 8 & 55504.7358 & 0.0002 & @xmath120.0035 & 0.31 & 386 + 9 & 55504.8128 & 0.0002 & @xmath120.0026 & 0.35 & 303 + 10 & 55504.8899 & 0.0002 & @xmath120.0017 & 0.40 & 354 + 11 & 55504.9669 & 0.0001 & @xmath120.0008 & 0.44 & 1322 + 12 & 55505.0420 & 0.0001 & @xmath120.0019 & 0.46 & 1490 + 13 & 55505.1173 & 0.0001 & @xmath120.0027 & 0.49 & 1892 + 14 & 55505.1947 & 0.0001 & @xmath120.0015 & 0.54 & 2049 + 15 & 55505.2696 & 0.0001 & @xmath120.0028 & 0.55 & 1326 + 16 & 55505.3468 & 0.0001 & @xmath120.0018 & 0.60 & 1381 + 17 & 55505.4232 & 0.0001 & @xmath120.0015 & 0.64 & 569 + 18 & 55505.5014 & 0.0001 & 0.0006 & 0.70 & 406 + 19 & 55505.5764 & 0.0002 & @xmath120.0006 & 0.72 & 596 + 20 & 55505.6538 & 0.0002 & 0.0006 & 0.77 & 636 + 21 & 55505.7306 & 0.0003 & 0.0012 & 0.81 & 455 + 22 & 55505.8062 & 0.0004 & 0.0007 & 0.84 & 170 + 23 & 55505.8805 & 0.0003 & @xmath120.0011 & 0.85 & 261 + 24 & 55505.9573 & 0.0002 & @xmath120.0006 & 0.89 & 530 + 25 & 55506.0348 & 0.0002 & 0.0009 & 0.94 & 1360 + 26 & 55506.1157 & 0.0006 & 0.0056 & 0.04 & 1522 + 27 & 55506.1954 & 0.0003 & 0.0091 & 0.12 & 2008 + 28 & 55506.2688 & 0.0003 & 0.0063 & 0.12 & 1471 + 29 & 55506.3440 & 0.0004 & 0.0054 & 0.14 & 798 + 30 & 55506.4167 & 0.0003 & 0.0019 & 0.13 & 344 + 31 & 55506.4922 & 0.0003 & 0.0013 & 0.15 & 472 + 32 & 55506.5716 & 0.0002 & 0.0045 & 0.23 & 517 + 33 & 55506.6468 & 0.0003 & 0.0036 & 0.25 & 390 + 34 & 55506.7220 & 0.0003 & 0.0026 & 0.27 & 226 + 35 & 55506.7981 & 0.0004 & 0.0026 & 0.31 & 169 + 36 & 55506.8741 & 0.0003 & 0.0023 & 0.34 & 138 + 37 & 55506.9508 & 0.0002 & 0.0029 & 0.38 & 747 + 38 & 55507.0267 & 0.0001 & 0.0026 & 0.41 & 1334 + 39 & 55507.1023 & 0.0001 & 0.0021 & 0.44 & 1960 + 40 & 55507.1789 & 0.0001 & 0.0026 & 0.48 & 1149 + 41 & 55507.2528 & 0.0002 & 0.0002 & 0.48 & 710 + 42 & 55507.3294 & 0.0001 & 0.0008 & 0.52 & 805 + 43 & 55507.4056 & 0.0002 & 0.0007 & 0.56 & 583 + 44 & 55507.4835 & 0.0002 & 0.0025 & 0.61 & 434 + 45 & 55507.5611 & 0.0002 & 0.0039 & 0.67 & 466 + 46 & 55507.6365 & 0.0002 & 0.0032 & 0.69 & 493 + 47 & 55507.7141 & 0.0003 & 0.0047 & 0.75 & 233 + 48 & 55507.7915 & 0.0003 & 0.0058 & 0.80 & 202 + 49 & 55507.8659 & 0.0008 & 0.0041 & 0.81 & 102 + 50 & 55507.9428 & 0.0003 & 0.0049 & 0.85 & 170 + 51 & 55508.0197 & 0.0003 & 0.0055 & 0.89 & 269 + 52 & 55508.0963 & 0.0008 & 0.0060 & 0.93 & 273 + + + + + cccccc @xmath10 & max & error & @xmath5 & phase & @xmath11 + 53 & 55508.1744 & 0.0004 & 0.0079 & 1.00 & 238 + 54 & 55508.2503 & 0.0006 & 0.0077 & 0.03 & 359 + 55 & 55508.3260 & 0.0003 & 0.0072 & 0.05 & 541 + 56 & 55508.4000 & 0.0002 & 0.0050 & 0.06 & 535 + 57 & 55508.4803 & 0.0004 & 0.0092 & 0.15 & 528 + 58 & 55508.5544 & 0.0003 & 0.0071 & 0.16 & 634 + 59 & 55508.6278 & 0.0003 & 0.0044 & 0.15 & 504 + 60 & 55508.7072 & 0.0009 & 0.0076 & 0.23 & 134 + 61 & 55508.7797 & 0.0007 & 0.0039 & 0.21 & 123 + 62 & 55508.8588 & 0.0005 & 0.0069 & 0.29 & 252 + 63 & 55508.9342 & 0.0004 & 0.0062 & 0.31 & 406 + 64 & 55509.0109 & 0.0003 & 0.0067 & 0.35 & 379 + 65 & 55509.0867 & 0.0002 & 0.0063 & 0.38 & 393 + 66 & 55509.1609 & 0.0003 & 0.0044 & 0.39 & 305 + 67 & 55509.2384 & 0.0002 & 0.0057 & 0.44 & 458 + 68 & 55509.3143 & 0.0002 & 0.0054 & 0.47 & 410 + 69 & 55509.3903 & 0.0002 & 0.0053 & 0.51 & 462 + 70 & 55509.4659 & 0.0002 & 0.0047 & 0.53 & 259 + 72 & 55509.6182 & 0.0005 & 0.0047 & 0.60 & 44 + 73 & 55509.6946 & 0.0003 & 0.0050 & 0.64 & 174 + 74 & 55509.7712 & 0.0003 & 0.0054 & 0.68 & 196 + 75 & 55509.8460 & 0.0006 & 0.0040 & 0.69 & 100 + 76 & 55509.9231 & 0.0002 & 0.0050 & 0.74 & 787 + 77 & 55509.9995 & 0.0002 & 0.0052 & 0.78 & 963 + 78 & 55510.0749 & 0.0001 & 0.0045 & 0.80 & 1396 + 79 & 55510.1502 & 0.0001 & 0.0036 & 0.82 & 1539 + 80 & 55510.2259 & 0.0003 & 0.0031 & 0.85 & 690 + 81 & 55510.2996 & 0.0005 & 0.0006 & 0.85 & 151 + 82 & 55510.3788 & 0.0005 & 0.0037 & 0.93 & 356 + 84 & 55510.5317 & 0.0005 & 0.0043 & 0.00 & 215 + 85 & 55510.6084 & 0.0003 & 0.0048 & 0.05 & 413 + 86 & 55510.6822 & 0.0004 & 0.0024 & 0.05 & 379 + 87 & 55510.7569 & 0.0005 & 0.0010 & 0.06 & 121 + 88 & 55510.8314 & 0.0006 & @xmath120.0006 & 0.07 & 117 + 89 & 55510.9072 & 0.0002 & @xmath120.0010 & 0.10 & 947 + 90 & 55510.9844 & 0.0001 & 0.0000 & 0.15 & 1719 + 91 & 55511.0601 & 0.0001 & @xmath120.0005 & 0.18 & 1678 + 92 & 55511.1369 & 0.0001 & 0.0002 & 0.22 & 2165 + 93 & 55511.2096 & 0.0001 & @xmath120.0032 & 0.21 & 2175 + 94 & 55511.2865 & 0.0002 & @xmath120.0025 & 0.25 & 1316 + 95 & 55511.3644 & 0.0003 & @xmath120.0008 & 0.31 & 413 + 96 & 55511.4395 & 0.0004 & @xmath120.0018 & 0.33 & 265 + 97 & 55511.5132 & 0.0004 & @xmath120.0043 & 0.33 & 221 + 98 & 55511.5889 & 0.0004 & @xmath120.0048 & 0.36 & 209 + 99 & 55511.6654 & 0.0002 & @xmath120.0044 & 0.40 & 379 + 100 & 55511.7413 & 0.0003 & @xmath120.0046 & 0.43 & 286 + 101 & 55511.8178 & 0.0004 & @xmath120.0043 & 0.47 & 153 + 102 & 55511.8963 & 0.0004 & @xmath120.0020 & 0.53 & 1033 + 103 & 55511.9714 & 0.0002 & @xmath120.0030 & 0.55 & 1612 + 104 & 55512.0496 & 0.0002 & @xmath120.0010 & 0.61 & 1866 + 105 & 55512.1239 & 0.0001 & @xmath120.0028 & 0.62 & 2033 + 106 & 55512.1982 & 0.0001 & @xmath120.0047 & 0.63 & 1734 + 107 & 55512.2749 & 0.0002 & @xmath120.0042 & 0.67 & 505 + + + + + cccccc @xmath10 & max & error & @xmath5 & phase & @xmath11 + 108 & 55512.3512 & 0.0003 & @xmath120.0040 & 0.71 & 427 + 109 & 55512.4261 & 0.0002 & @xmath120.0053 & 0.73 & 275 + 113 & 55512.7334 & 0.0009 & @xmath120.0026 & 0.90 & 37 + 114 & 55512.8057 & 0.0008 & @xmath120.0065 & 0.88 & 54 + 115 & 55512.8824 & 0.0003 & @xmath120.0059 & 0.92 & 470 + 116 & 55512.9568 & 0.0005 & @xmath120.0077 & 0.93 & 802 + 117 & 55513.0338 & 0.0012 & @xmath120.0069 & 0.98 & 582 + 118 & 55513.1057 & 0.0013 & @xmath120.0112 & 0.95 & 425 + 124 & 55513.5610 & 0.0022 & @xmath120.0128 & 0.14 & 130 + 125 & 55513.6407 & 0.0007 & @xmath120.0092 & 0.22 & 163 + 126 & 55513.7174 & 0.0007 & @xmath120.0087 & 0.26 & 94 + + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55514.2505 & 0.0009 & @xmath120.0074 & 51 + 1 & 55514.3298 & 0.0012 & @xmath120.0036 & 92 + 2 & 55514.4019 & 0.0011 & @xmath120.0070 & 138 + 5 & 55514.6296 & 0.0016 & @xmath120.0058 & 73 + 10 & 55515.0057 & 0.0008 & @xmath120.0071 & 116 + 11 & 55515.0804 & 0.0011 & @xmath120.0080 & 235 + 12 & 55515.1653 & 0.0006 & 0.0014 & 511 + 13 & 55515.2314 & 0.0037 & @xmath120.0080 & 235 + 14 & 55515.3072 & 0.0011 & @xmath120.0077 & 375 + 15 & 55515.3817 & 0.0011 & @xmath120.0087 & 326 + 16 & 55515.4665 & 0.0015 & 0.0007 & 396 + 17 & 55515.5356 & 0.0007 & @xmath120.0057 & 303 + 18 & 55515.6051 & 0.0017 & @xmath120.0118 & 313 + 19 & 55515.7035 & 0.0011 & 0.0112 & 285 + 22 & 55515.9092 & 0.0013 & @xmath120.0096 & 288 + 23 & 55515.9823 & 0.0010 & @xmath120.0120 & 576 + 24 & 55516.0672 & 0.0012 & @xmath120.0025 & 691 + 25 & 55516.1419 & 0.0022 & @xmath120.0034 & 508 + 26 & 55516.2248 & 0.0008 & 0.0041 & 743 + 28 & 55516.3773 & 0.0014 & 0.0055 & 158 + 30 & 55516.5220 & 0.0027 & @xmath120.0008 & 101 + 38 & 55517.1286 & 0.0003 & 0.0019 & 468 + 39 & 55517.2157 & 0.0014 & 0.0135 & 216 + 50 & 55518.0345 & 0.0008 & 0.0019 & 585 + 56 & 55518.4799 & 0.0015 & @xmath120.0056 & 52 + 61 & 55518.8810 & 0.0011 & 0.0180 & 87 + 62 & 55518.9501 & 0.0005 & 0.0116 & 50 + 64 & 55519.1068 & 0.0004 & 0.0173 & 907 + 65 & 55519.1772 & 0.0003 & 0.0122 & 1225 + 67 & 55519.3260 & 0.0017 & 0.0100 & 122 + 70 & 55519.5551 & 0.0010 & 0.0126 & 30 + 71 & 55519.6294 & 0.0043 & 0.0114 & 33 + 72 & 55519.7060 & 0.0092 & 0.0125 & 19 + 73 & 55519.7864 & 0.0013 & 0.0175 & 31 + 81 & 55520.3829 & 0.0016 & 0.0100 & 60 + 85 & 55520.6823 & 0.0025 & 0.0074 & 69 + 89 & 55520.9690 & 0.0012 & @xmath120.0078 & 226 + 92 & 55521.2005 & 0.0027 & @xmath120.0028 & 406 + 93 & 55521.2736 & 0.0017 & @xmath120.0052 & 309 + 94 & 55521.3474 & 0.0009 & @xmath120.0069 & 102 + 131 & 55524.1552 & 0.0047 & 0.0077 & 117 + 141 & 55524.8969 & 0.0013 & @xmath120.0056 & 193 + 142 & 55524.9816 & 0.0018 & 0.0037 & 322 + 145 & 55525.1885 & 0.0019 & @xmath120.0159 & 473 + 165 & 55526.7035 & 0.0020 & @xmath120.0108 & 26 + 169 & 55527.0147 & 0.0031 & @xmath120.0016 & 212 + 170 & 55527.0888 & 0.0021 & @xmath120.0030 & 206 + 171 & 55527.1650 & 0.0010 & @xmath120.0023 & 507 + 172 & 55527.2378 & 0.0087 & @xmath120.0050 & 290 + 177 & 55527.6099 & 0.0035 & @xmath120.0104 & 84 + 178 & 55527.6894 & 0.0016 &
@xmath120.0063 & 118 + 179 & 55527.7626 & 0.0031 & @xmath120.0087 & 26 + 183 & 55528.0823 & 0.0011 & 0.0091 & 394 + 214 & 55530.4193 & 0.0016 & 0.0058 & 161 + + + + we report additional superoutburst in 2007 ( table [ tab : v1504cygoc2007 ] ) .
although the data were not abundant , stages b and c can be recognized .
the determined periods are in good agreement with the following kepler observations .
the times of maxima derived from kepler observations discussed in subsection [ obj : kepv1504cyg ] are presented in table [ tab : v1504cygoc2009b ] .
ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 54324.4787 & 0.0018 & @xmath120.0061 & 13 + 27 & 54326.4283 & 0.0013 & @xmath120.0019 & 29 + 41 & 54327.4339 & 0.0026 & @xmath120.0049 & 32 + 42 & 54327.5122 & 0.0016 & 0.0013 & 33 + 55 & 54328.4608 & 0.0039 & 0.0133 & 19 + 84 & 54330.5368 & 0.0017 & @xmath120.0002 & 15 + 96 & 54331.4028 & 0.0016 & 0.0013 & 18 + 97 & 54331.4754 & 0.0039 & 0.0018 & 19 + 98 & 54331.5491 & 0.0023 & 0.0034 & 18 + 110 & 54332.4129 & 0.0020 & 0.0026 & 13 + 111 & 54332.4855 & 0.0032 & 0.0032 & 13 + 112 & 54332.5530 & 0.0018 & @xmath120.0014 & 11 + 124 & 54333.4125 & 0.0053 & @xmath120.0064 & 13 + 125 & 54333.4911 & 0.0020 & 0.0000 & 12 + 126 & 54333.5634 & 0.0025 & 0.0004 & 11 + 139 & 54334.4933 & 0.0022 & @xmath120.0064 & 12 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55076.2646 & 0.0009 & @xmath120.0344 & 84 + 1 & 55076.3457 & 0.0014 & @xmath120.0254 & 84 + 2 & 55076.4227 & 0.0008 & @xmath120.0205 & 84 + 3 & 55076.4962 & 0.0004 & @xmath120.0192 & 84 + 4 & 55076.5663 & 0.0006 & @xmath120.0212 & 85 + 5 & 55076.6365 & 0.0003 & @xmath120.0232 & 85 + 6 & 55076.7075 & 0.0004 & @xmath120.0243 & 85 + 7 & 55076.7779 & 0.0006 & @xmath120.0260 & 85 + 8 & 55076.8514 & 0.0004 & @xmath120.0247 & 85 + 9 & 55076.9234 & 0.0005 & @xmath120.0248 & 84 + 10 & 55077.0023 & 0.0003 & @xmath120.0180 & 85 + 11 & 55077.0790 & 0.0004 & @xmath120.0135 & 85 + 12 & 55077.1528 & 0.0004 & @xmath120.0119 & 85 + 13 & 55077.2287 & 0.0003 & @xmath120.0081 & 84 + 14 & 55077.3023 & 0.0003 & @xmath120.0066 & 84 + 15 & 55077.3763 & 0.0002 & @xmath120.0048 & 85 + 16 & 55077.4488 & 0.0002 & @xmath120.0045 & 85 + 17 & 55077.5202 & 0.0002 & @xmath120.0052 & 84 + 18 & 55077.5922 & 0.0001 & @xmath120.0053 & 85 + 19 & 55077.6682 & 0.0002 & @xmath120.0014 & 85 + 20 & 55077.7415 & 0.0001 & @xmath120.0003 & 76 + 21 & 55077.8164 & 0.0002 & 0.0025 & 85 + 22 & 55077.8884 & 0.0001 & 0.0024 & 85 + 23 & 55077.9596 & 0.0001 & 0.0014 & 84 + 24 & 55078.0322 & 0.0001 & 0.0019 & 85 + 25 & 55078.1047 & 0.0002 & 0.0022 & 85 + 26 & 55078.1782 & 0.0002 & 0.0036 & 85 + 27 & 55078.2507 & 0.0002 & 0.0039 & 84 + 28 & 55078.3223 & 0.0002 & 0.0034 & 85 + 29 & 55078.3945 & 0.0003 & 0.0034 & 84 + 30 & 55078.4676 & 0.0002 & 0.0044 & 85 + 31 & 55078.5398 & 0.0002 & 0.0045 & 84 + 32 & 55078.6141 & 0.0002 & 0.0066 & 85 + 33 & 55078.6869 & 0.0003 & 0.0072 & 85 + 34 & 55078.7576 & 0.0002 & 0.0059 & 85 + 35 & 55078.8319 & 0.0002 & 0.0080 & 85 + 36 & 55078.9021 & 0.0003 & 0.0061 & 85 + 37 & 55078.9748 & 0.0003 & 0.0066 & 85 + 38 & 55079.0471 & 0.0003 & 0.0068 & 84 + 39 & 55079.1190 & 0.0003 & 0.0066 & 85 + 41 & 55079.2642 & 0.0003 & 0.0074 & 85 + 42 & 55079.3363 & 0.0002 & 0.0074 & 85 + 43 & 55079.4075 & 0.0004 & 0.0065 & 84 + 44 & 55079.4812 & 0.0003 & 0.0080 & 85 + 45 & 55079.5533 & 0.0003 & 0.0080 & 84 + 46 & 55079.6251 & 0.0002 & 0.0076 & 85 + 47 & 55079.6949 & 0.0003 & 0.0053 & 84 + 48 & 55079.7658 & 0.0003 & 0.0041 & 85 + 49 & 55079.8395 & 0.0004 & 0.0056 & 84 + 50 & 55079.9105 & 0.0004 & 0.0045 & 85 + 51 & 55079.9825 & 0.0003 & 0.0043 & 85 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 52 & 55080.0552 & 0.0004 & 0.0049 & 85 + 53 & 55080.1261 & 0.0004 & 0.0036 & 85 + 54 & 55080.1989 & 0.0003 & 0.0043 & 85 + 55 & 55080.2704 & 0.0005 & 0.0037 & 84 + 56 & 55080.3446 & 0.0005 & 0.0058 & 84 + 57 & 55080.4193 & 0.0005 & 0.0083 & 84 + 58 & 55080.4909 & 0.0003 & 0.0078 & 85 + 59 & 55080.5613 & 0.0004 & 0.0060 & 84 + 60 & 55080.6338 & 0.0004 & 0.0064 & 85 + 61 & 55080.7065 & 0.0003 & 0.0069 & 74 + 62 & 55080.7780 & 0.0003 & 0.0063 & 85 + 63 & 55080.8496 & 0.0003 & 0.0058 & 85 + 64 & 55080.9221 & 0.0004 & 0.0061 & 85 + 65 & 55080.9940 & 0.0003 & 0.0059 & 85 + 66 & 55081.0665 & 0.0003 & 0.0062 & 84 + 67 & 55081.1401 & 0.0003 & 0.0077 & 84 + 68 & 55081.2115 & 0.0003 & 0.0070 & 84 + 69 & 55081.2852 & 0.0004 & 0.0085 & 84 + 70 & 55081.3578 & 0.0005 & 0.0089 & 85 + 71 & 55081.4287 & 0.0003 & 0.0077 & 84 + 72 & 55081.5004 & 0.0003 & 0.0073 & 85 + 73 & 55081.5740 & 0.0003 & 0.0087 & 85 + 74 & 55081.6460 & 0.0002 & 0.0086 & 84 + 75 & 55081.7172 & 0.0003 & 0.0077 & 85 + 76 & 55081.7899 & 0.0003 & 0.0082 & 84 + 77 & 55081.8615 & 0.0002 & 0.0077 & 84 + 78 & 55081.9332 & 0.0003 & 0.0073 & 85 + 79 & 55082.0055 & 0.0003 & 0.0074 & 85 + 80 & 55082.0805 & 0.0003 & 0.0102 & 84 + 81 & 55082.1528 & 0.0004 & 0.0104 & 85 + 82 & 55082.2246 & 0.0002 & 0.0101 & 85 + 83 & 55082.2978 & 0.0004 & 0.0112 & 85 + 84 & 55082.3672 & 0.0003 & 0.0084 & 85 + 85 & 55082.4393 & 0.0003 & 0.0083 & 85 + 86 & 55082.5108 & 0.0003 & 0.0077 & 85 + 87 & 55082.5837 & 0.0003 & 0.0085 & 85 + 88 & 55082.6554 & 0.0003 & 0.0080 & 85 + 89 & 55082.7286 & 0.0004 & 0.0091 & 84 + 90 & 55082.7989 & 0.0003 & 0.0073 & 85 + 91 & 55082.8708 & 0.0003 & 0.0070 & 84 + 92 & 55082.9431 & 0.0003 & 0.0072 & 84 + 93 & 55083.0154 & 0.0005 & 0.0073 & 85 + 94 & 55083.0893 & 0.0004 & 0.0091 & 85 + 95 & 55083.1593 & 0.0004 & 0.0069 & 85 + 96 & 55083.2323 & 0.0004 & 0.0078 & 85 + 97 & 55083.3031 & 0.0003 & 0.0064 & 84 + 98 & 55083.3751 & 0.0004 & 0.0063 & 85 + 99 & 55083.4482 & 0.0003 & 0.0073 & 85 + 100 & 55083.5177 & 0.0003 & 0.0047 & 85 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 101 & 55083.5903 & 0.0003 & 0.0051 & 85 + 102 & 55083.6613 & 0.0003 & 0.0039 & 74 + 103 & 55083.7337 & 0.0005 & 0.0042 & 84 + 104 & 55083.8059 & 0.0004 & 0.0043 & 84 + 105 & 55083.8770 & 0.0003 & 0.0032 & 84 + 106 & 55083.9501 & 0.0005 & 0.0042 & 85 + 107 & 55084.0222 & 0.0004 & 0.0041 & 85 + 108 & 55084.0949 & 0.0004 & 0.0047 & 85 + 109 & 55084.1676 & 0.0004 & 0.0053 & 85 + 110 & 55084.2378 & 0.0003 & 0.0033 & 85 + 111 & 55084.3083 & 0.0003 & 0.0017 & 84 + 112 & 55084.3809 & 0.0004 & 0.0021 & 85 + 113 & 55084.4522 & 0.0003 & 0.0013 & 85 + 114 & 55084.5230 & 0.0003 & 0.0000 & 85 + 115 & 55084.5968 & 0.0003 & 0.0016 & 85 + 116 & 55084.6686 & 0.0004 & 0.0013 & 85 + 117 & 55084.7415 & 0.0004 & 0.0021 & 84 + 118 & 55084.8104 & 0.0003 & @xmath120.0012 & 84 + 119 & 55084.8862 & 0.0004 & 0.0024 & 84 + 120 & 55084.9563 & 0.0004 & 0.0004 & 85 + 121 & 55085.0291 & 0.0003 & 0.0011 & 85 + 122 & 55085.1008 & 0.0004 & 0.0006 & 85 + 123 & 55085.1703 & 0.0004 & @xmath120.0020 & 85 + 124 & 55085.2432 & 0.0004 & @xmath120.0013 & 84 + 125 & 55085.3143 & 0.0004 & @xmath120.0023 & 85 + 126 & 55085.3864 & 0.0003 & @xmath120.0024 & 85 + 127 & 55085.4591 & 0.0004 & @xmath120.0018 & 85 + 128 & 55085.5303 & 0.0004 & @xmath120.0027 & 85 + 129 & 55085.6038 & 0.0003 & @xmath120.0014 & 85 + 130 & 55085.6746 & 0.0004 & @xmath120.0027 & 85 + 131 & 55085.7459 & 0.0004 & @xmath120.0035 & 85 + 132 & 55085.8191 & 0.0004 & @xmath120.0025 & 85 + 133 & 55085.8938 & 0.0003 & 0.0001 & 85 + 134 & 55085.9637 & 0.0004 & @xmath120.0021 & 85 + 135 & 55086.0351 & 0.0004 & @xmath120.0029 & 85 + 136 & 55086.1051 & 0.0004 & @xmath120.0050 & 85 + 137 & 55086.1790 & 0.0004 & @xmath120.0033 & 85 + 138 & 55086.2495 & 0.0003 & @xmath120.0049 & 85 + 139 & 55086.3237 & 0.0003 & @xmath120.0028 & 84 + 140 & 55086.3943 & 0.0003 & @xmath120.0044 & 84 + 141 & 55086.4671 & 0.0003 & @xmath120.0037 & 85 + 142 & 55086.5397 & 0.0003 & @xmath120.0033 & 84 + 143 & 55086.6096 & 0.0004 & @xmath120.0055 & 84 + 144 & 55086.6818 & 0.0004 & @xmath120.0054 & 74 + 145 & 55086.7560 & 0.0003 & @xmath120.0034 & 85 + 146 & 55086.8282 & 0.0005 & @xmath120.0034 & 84 + 147 & 55086.8996 & 0.0004 & @xmath120.0041 & 85 + 148 & 55086.9711 & 0.0004 & @xmath120.0047 & 84 + 149 & 55087.0413 & 0.0005 & @xmath120.0066 & 85 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 150 & 55087.1146 & 0.0003 & @xmath120.0055 & 85 + 151 & 55087.1876 & 0.0004 & @xmath120.0046 & 84 + 152 & 55087.2567 & 0.0003 & @xmath120.0077 & 85 + 153 & 55087.3278 & 0.0004 & @xmath120.0087 & 85 + 154 & 55087.3972 & 0.0005 & @xmath120.0115 & 85 + 155 & 55087.4706 & 0.0006 & @xmath120.0102 & 84 + 156 & 55087.5428 & 0.0008 & @xmath120.0102 & 85 + 157 & 55087.6146 & 0.0004 & @xmath120.0105 & 85 + 158 & 55087.6897 & 0.0005 & @xmath120.0075 & 85 + 159 & 55087.7620 & 0.0010 & @xmath120.0073 & 84 + 160 & 55087.8269 & 0.0005 & @xmath120.0146 & 85 + 161 & 55087.9015 & 0.0005 & @xmath120.0122 & 85 + 162 & 55087.9788 & 0.0009 & @xmath120.0070 & 85 + 163 & 55088.0412 & 0.0012 & @xmath120.0168 & 85 + 164 & 55088.1033 & 0.0012 & @xmath120.0268 & 84 + + + + we observed the middle - to - late stage of the 2011 superoutburst .
the stage b
c transition was well - recorded ( table [ tab : awgemoc2011 ] ) .
a comparison of @xmath5 diagrams between different superoutbursts is shown in figure [ fig : awgemcomp2 ] .
( 88mm,70mm)fig16.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55571.1836 & 0.0002 & @xmath120.0112 & 112 + 1 & 55571.2657 & 0.0002 & @xmath120.0079 & 111 + 2 & 55571.3445 & 0.0005 & @xmath120.0079 & 83 + 12 & 55572.1390 & 0.0003 & @xmath120.0017 & 108 + 13 & 55572.2182 & 0.0003 & @xmath120.0013 & 112 + 25 & 55573.1695 & 0.0003 & 0.0042 & 108 + 26 & 55573.2500 & 0.0003 & 0.0058 & 111 + 27 & 55573.3276 & 0.0005 & 0.0046 & 84 + 38 & 55574.1947 & 0.0003 & 0.0048 & 165 + 39 & 55574.2757 & 0.0003 & 0.0068 & 168 + 40 & 55574.3563 & 0.0006 & 0.0086 & 84 + 50 & 55575.1423 & 0.0003 & 0.0065 & 112 + 51 & 55575.2196 & 0.0003 & 0.0049 & 112 + 52 & 55575.2969 & 0.0004 & 0.0035 & 111 + 76 & 55577.1845 & 0.0006 & @xmath120.0006 & 115 + 77 & 55577.2642 & 0.0006 & 0.0003 & 114 + 78 & 55577.3427 & 0.0013 & @xmath120.0001 & 77 + 88 & 55578.1285 & 0.0005 & @xmath120.0025 & 84 + 89 & 55578.2065 & 0.0007 & @xmath120.0033 & 87 + 90 & 55578.2849 & 0.0006 & @xmath120.0037 & 87 + 101 & 55579.1509 & 0.0013 & @xmath120.0047 & 88 + 102 & 55579.2295 & 0.0019 & @xmath120.0050 & 87 + + + + we reported on the 2010 april may superoutburst of this object in @xcite .
the object underwent another superoutburst in 2010 october
the interval since the april may superoutburst was only 179 d. the current state with frequent superoutbursts started in 2009 , in contrast to earlier state with relatively infrequent superoutbursts ( @xcite ; @xcite ) .
since the present observations were obtained at high airmasses , we corrected observations by using a second - order atmospheric extinction whose coefficients were experimentally determined .
the times of superhump maxima are listed in table [ tab : v844heroc2010b ] . despite rather unfavorable seasonal conditions ,
we have succeeded in obtaining good timing data of superhumps .
the @xmath5 diagram indicates the presence of a stage b
c transition at late epochs .
figure [ fig : v844hercomp3 ] illustrates a comparison of @xmath5 diagrams in different superoutbursts .
it is noteworthy that the stage b c transition in this small - scale ( faint and short ) superoutburst occurred much earlier than in past large - scale superoutbursts .
the period of stage b superhumps appears to be noticeably longer than in other superoutbursts .
although this may have been a result of the lack of the early stage observations , this result suggests that the evolution of small - scale superoutbursts can be much faster than in large - scale ones .
( 88mm,70mm)fig17.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55495.2921 & 0.0005 & @xmath120.0003 & 38 + 18 & 55496.3006 & 0.0009 & @xmath120.0010 & 60 + 34 & 55497.1996 & 0.0008 & 0.0010 & 112 + 35 & 55497.2563 & 0.0015 & 0.0016 & 123 + 36 & 55497.3100 & 0.0009 & @xmath120.0008 & 124 + 52 & 55498.2022 & 0.0017 & @xmath120.0056 & 120 + 53 & 55498.2643 & 0.0009 & 0.0004 & 123 + 54 & 55498.3175 & 0.0034 & @xmath120.0025 & 92 + 70 & 55499.2185 & 0.0012 & 0.0015 & 119 + 71 & 55499.2779 & 0.0012 & 0.0049 & 121 + 72 & 55499.3340 & 0.0033 & 0.0049 & 57 + 106 & 55501.2335 & 0.0007 & @xmath120.0019 & 126 + 107 & 55501.2892 & 0.0008 & @xmath120.0022 & 122 + + + + in @xcite , we reported on rather fragmentary observations of mm hya during the 1998 and 2001 superoutbursts .
since mm hya was initially suggested to be related to wz sge - type dwarf novae @xcite , more detailed observations were needed to examine this possibility .
there has been no report on measurement of @xmath13 for stage b superhumps , which is expected to be useful in characterizing the object ( cf .
@xcite , subsection 5.3 ) .
the 2011 superoutburst was fortunately detected during its rising stage ( r. stubbings , vsnet - alert 13113 ) .
this was the first recorded superoutburst since the last one in 2006 march .
early observations detected long - period superhumps ( vsnet - alert 13130 , 13131 ) , rather than early superhumps in wz sge - type dwarf novae , which further developed into full superhumps within 1 d ( vsnet - alert 12132 ) .
the times of superhump maxima are listed in table [ tab : mmhyaoc2011 ] . due to a short gap in observation between the first and second nights ,
the cycle count is somewhat ambiguous in the earliest phase .
we identified the cycle numbers given in the table since observations on the first night apparently recorded the growing stage ( stage a ) of superhumps and there must have been a large period change before the system started showing fully developed stage b superhumps .
the adopted cycle counts gives a mean period of 0.0629(2 ) d for @xmath33 , which is close to the period [ 0.0617(3 ) d ] obtained from the observations on the first night .
the fractional excess ( 7 % ) of the period of stage a superhumps to that of stage b superhumps is exceptionally large ( cf .
@xcite , subsection 3.7 ) .
if we assume an additional cycle count between the first and second nights , we get a period of 0.0590(4 ) d on these nights . although this value is close to that of stage b superhumps , the @xmath5 variations within the first night become inconsistent with the later evolution .
we therefore favor the former assumption of the cycle counts .
the resultant @xmath13 for stage b superhumps was @xmath34 , which also supports the classification as an usual su uma - type dwarf nova with a short @xmath18 .
ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55660.0258 & 0.0018 & @xmath120.0393 & 113 + 1 & 55660.0952 & 0.0011 & @xmath120.0291 & 113 + 2 & 55660.1546 & 0.0006 & @xmath120.0289 & 112 + 3 & 55660.2149 & 0.0012 & @xmath120.0277 & 59 + 16 & 55661.0355 & 0.0003 & 0.0237 & 76 + 17 & 55661.0944 & 0.0002 & 0.0235 & 113 + 18 & 55661.1528 & 0.0002 & 0.0228 & 106 + 22 & 55661.3880 & 0.0004 & 0.0213 & 58 + 32 & 55661.9756 & 0.0013 & 0.0173 & 47 + 33 & 55662.0327 & 0.0014 & 0.0151 & 64 + 39 & 55662.3856 & 0.0005 & 0.0131 & 60 + 49 & 55662.9702 & 0.0012 & 0.0060 & 47 + 50 & 55663.0342 & 0.0010 & 0.0108 & 118 + 51 & 55663.0898 & 0.0009 & 0.0073 & 125 + 66 & 55663.9713 & 0.0008 & 0.0013 & 159 + 67 & 55664.0301 & 0.0008 & 0.0009 & 181 + 68 & 55664.0915 & 0.0022 & 0.0032 & 67 + 83 & 55664.9723 & 0.0015 & @xmath120.0035 & 101 + 84 & 55665.0337 & 0.0007 & @xmath120.0012 & 108 + 85 & 55665.0925 & 0.0008 & @xmath120.0016 & 113 + 86 & 55665.1519 & 0.0010 & @xmath120.0014 & 110 + 117 & 55666.9788 & 0.0010 & @xmath120.0086 & 113 + 118 & 55667.0397 & 0.0010 & @xmath120.0069 & 106 + 119 & 55667.0972 & 0.0014 & @xmath120.0085 & 113 + 120 & 55667.1553 & 0.0010 & @xmath120.0096 & 87 + + + + v406 hya is an am cvn - type cv ( for a recent review , see @xcite ) which was initially discovered as a possible supernova ( @xcite and references therein ) .
@xcite reported on the unusual behavior of its 2004 superoutburst .
@xcite further reported phase - resolved spectroscopy and obtained an orbital period of 2027.8(5 ) s = 0.023470(6 ) d. a bright ( @xmath915.0 mag ) outburst of this objected in 2010 december was detected by crts .
superhumps with amplitudes of 0.065 mag were detected ( vsnet - alert 12483 ) .
the times of superhump maxima are listed in table [ tab : v406hyaoc2010 ] .
the mean period of superhumps was 0.02365(5 ) d ( pdm analysis ) , which is in agreement with previous measurements [ 0.023628(4 ) d , @xcite ; 0.02357(4 ) d , @xcite ] .
the known outbursts of this object are listed in table [ tab : v406hyaout ] . based on magnitudes and durations ,
all known outbursts appear to be superoutbursts , and the intervals between superoutbursts are apparently longer than the one ( 60 d ) in kl dra @xcite or @xmath9100 d in cp eri based on crts data ( vsnet - alert 11504)http://nesssi.cacr.caltech.edu / catalina/20090925/909250090174119865.html@xmath15 . ] .
ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55539.5314 & 0.0007 & @xmath120.0004 & 34 + 1 & 55539.5564 & 0.0004 & 0.0008 & 49 + 2 & 55539.5800 & 0.0005 & 0.0005 & 47 + 3 & 55539.6026 & 0.0006 & @xmath120.0007 & 47 + 4 & 55539.6267 & 0.0004 & @xmath120.0004 & 48 + 26 & 55540.1552 & 0.0038 & 0.0042 & 62 + 28 & 55540.1955 & 0.0011 & @xmath120.0032 & 43 + 29 & 55540.2201 & 0.0011 & @xmath120.0024 & 72 + 30 & 55540.2405 & 0.0013 & @xmath120.0059 & 87 + 31 & 55540.2776 & 0.0070 & 0.0074 & 50 + + + + cccc month & max & magnitude & source + 2003 2 & 52677 & 17.8 & @xcite + 2004 5 & 53143 & 14.715.2 & @xcite + 2005 4 & 53465 & 15.2 & crts + 2006 11 & 54057 & 17.5 & crts + 2008 4 & 54565 & 15.3 & crts + 2009 12 & 55181 & 15.8 & crts + 2010 12 & 55536 & 15.0 & crts + + the data discussed in subsection [ obj : kepv344lyr ] are presented .
the object showed prominent double peaks during the late course of the superoutbursts .
we therefore used phases @xmath120.2 to 0.2 ( instead of @xmath120.4 to 0.4 , cf .
@xcite ) for fitting the template superhump profile in order to separate individual peaks .
the times of maxima of main peaks are listed in table [ tab : v344lyroc2009 ] .
the times of secondary maxima , which smoothly evolved into persisting superhumps during the next normal outburst and subsequent interval of quiescence , are listed in table [ tab : v344lyroc2009sec ] .
the object underwent another superoutburst in 2009 november december .
the data for this superoutburst are listed in tables [ tab : v344lyroc2009b ] and [ tab : v344lyroc2009bsec ] .
ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55057.6071 & 0.0008 & @xmath120.0327 & 54 + 1 & 55057.7010 & 0.0024 & @xmath120.0303 & 53 + 2 & 55057.7969 & 0.0011 & @xmath120.0258 & 54 + 3 & 55057.8860 & 0.0006 & @xmath120.0283 & 54 + 4 & 55057.9779 & 0.0008 & @xmath120.0279 & 54 + 5 & 55058.0721 & 0.0003 & @xmath120.0253 & 53 + 6 & 55058.1630 & 0.0002 & @xmath120.0259 & 54 + 7 & 55058.2559 & 0.0002 & @xmath120.0245 & 54 + 8 & 55058.3535 & 0.0001 & @xmath120.0184 & 54 + 9 & 55058.4484 & 0.0001 & @xmath120.0150 & 54 + 10 & 55058.5440 & 0.0001 & @xmath120.0109 & 54 + 11 & 55058.6380 & 0.0001 & @xmath120.0084 & 53 + 12 & 55058.7317 & 0.0001 & @xmath120.0063 & 54 + 13 & 55058.8239 & 0.0001 & @xmath120.0056 & 54 + 14 & 55058.9165 & 0.0001 & @xmath120.0046 & 53 + 15 & 55059.0096 & 0.0001 & @xmath120.0029 & 54 + 16 & 55059.1027 & 0.0001 & @xmath120.0013 & 53 + 17 & 55059.1942 & 0.0002 & @xmath120.0014 & 54 + 18 & 55059.2874 & 0.0001 & 0.0003 & 54 + 19 & 55059.3804 & 0.0001 & 0.0017 & 54 + 20 & 55059.4729 & 0.0001 & 0.0028 & 54 + 21 & 55059.5640 & 0.0001 & 0.0023 & 54 + 22 & 55059.6546 & 0.0001 & 0.0014 & 54 + 23 & 55059.7477 & 0.0001 & 0.0030 & 54 + 24 & 55059.8402 & 0.0001 & 0.0040 & 43 + 25 & 55059.9316 & 0.0001 & 0.0038 & 53 + 26 & 55060.0257 & 0.0001 & 0.0064 & 54 + 27 & 55060.1165 & 0.0001 & 0.0057 & 54 + 28 & 55060.2093 & 0.0002 & 0.0070 & 54 + 29 & 55060.3006 & 0.0002 & 0.0068 & 53 + 30 & 55060.3914 & 0.0001 & 0.0060 & 54 + 31 & 55060.4833 & 0.0002 & 0.0064 & 54 + 32 & 55060.5753 & 0.0002 & 0.0069 & 54 + 33 & 55060.6675 & 0.0002 & 0.0076 & 54 + 34 & 55060.7590 & 0.0002 & 0.0076 & 54 + 35 & 55060.8498 & 0.0001 & 0.0069 & 54 + 36 & 55060.9410 & 0.0002 & 0.0065 & 54 + 37 & 55061.0329 & 0.0002 & 0.0069 & 53 + 38 & 55061.1249 & 0.0002 & 0.0074 & 54 + 39 & 55061.2169 & 0.0002 & 0.0079 & 54 + 40 & 55061.3078 & 0.0001 & 0.0072 & 54 + 41 & 55061.3998 & 0.0002 & 0.0077 & 53 + 42 & 55061.4918 & 0.0002 & 0.0082 & 53 + 43 & 55061.5827 & 0.0003 & 0.0075 & 54 + 44 & 55061.6742 & 0.0002 & 0.0075 & 54 + 45 & 55061.7652 & 0.0003 & 0.0070 & 54 + 46 & 55061.8574 & 0.0003 & 0.0076 & 54 + 47 & 55061.9487 & 0.0002 & 0.0075 & 54 + 48 & 55062.0405 & 0.0002 & 0.0077 & 54 + 49 & 55062.1319 & 0.0003 & 0.0076 & 53 + 50 & 55062.2227 & 0.0004 & 0.0069 & 54 + 51 & 55062.3126 & 0.0007 & 0.0053 & 54 + 52 & 55062.4076 & 0.0002 & 0.0087 & 54 + 53 & 55062.4980 & 0.0003 & 0.0076 & 54 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 54 & 55062.5902 & 0.0003 & 0.0083 & 54 + 55 & 55062.6838 & 0.0002 & 0.0104 & 54 + 56 & 55062.7741 & 0.0004 & 0.0092 & 54 + 57 & 55062.8649 & 0.0003 & 0.0084 & 54 + 58 & 55062.9561 & 0.0002 & 0.0081 & 54 + 59 & 55063.0462 & 0.0003 & 0.0067 & 54 + 60 & 55063.1397 & 0.0002 & 0.0087 & 54 + 61 & 55063.2324 & 0.0002 & 0.0099 & 54 + 74 & 55064.4234 & 0.0003 & 0.0111 & 54 + 75 & 55064.5144 & 0.0003 & 0.0105 & 54 + 76 & 55064.6035 & 0.0003 & 0.0082 & 54 + 77 & 55064.6956 & 0.0002 & 0.0087 & 54 + 78 & 55064.7865 & 0.0002 & 0.0081 & 54 + 79 & 55064.8803 & 0.0002 & 0.0104 & 54 + 80 & 55064.9714 & 0.0003 & 0.0099 & 53 + 81 & 55065.0578 & 0.0003 & 0.0048 & 54 + 82 & 55065.1536 & 0.0003 & 0.0091 & 54 + 83 & 55065.2456 & 0.0003 & 0.0096 & 53 + 84 & 55065.3374 & 0.0002 & 0.0098 & 54 + 85 & 55065.4268 & 0.0002 & 0.0077 & 54 + 86 & 55065.5203 & 0.0003 & 0.0097 & 54 + 87 & 55065.6094 & 0.0003 & 0.0073 & 54 + 88 & 55065.6988 & 0.0003 & 0.0052 & 54 + 89 & 55065.7918 & 0.0006 & 0.0067 & 43 + 90 & 55065.8815 & 0.0003 & 0.0048 & 53 + 91 & 55065.9753 & 0.0003 & 0.0071 & 54 + 92 & 55066.0649 & 0.0004 & 0.0051 & 54 + 93 & 55066.1574 & 0.0003 & 0.0062 & 54 + 94 & 55066.2475 & 0.0003 & 0.0047 & 54 + 95 & 55066.3406 & 0.0006 & 0.0064 & 54 + 96 & 55066.4299 & 0.0005 & 0.0041 & 54 + 97 & 55066.5207 & 0.0004 & 0.0033 & 53 + 98 & 55066.6125 & 0.0004 & 0.0037 & 53 + 99 & 55066.7034 & 0.0003 & 0.0030 & 53 + 100 & 55066.7941 & 0.0007 & 0.0022 & 54 + 101 & 55066.8838 & 0.0004 & 0.0004 & 54 + 102 & 55066.9771 & 0.0007 & 0.0022 & 54 + 103 & 55067.0691 & 0.0007 & 0.0026 & 53 + 104 & 55067.1599 & 0.0007 & 0.0019 & 54 + 105 & 55067.2518 & 0.0013 & 0.0023 & 54 + 106 & 55067.3405 & 0.0006 & @xmath120.0005 & 54 + 107 & 55067.4333 & 0.0006 & 0.0008 & 54 + 108 & 55067.5243 & 0.0006 & 0.0003 & 54 + 109 & 55067.6156 & 0.0011 & @xmath120.0000 & 54 + 110 & 55067.7062 & 0.0017 & @xmath120.0009 & 54 + 111 & 55067.7990 & 0.0007 & 0.0004 & 53 + 112 & 55067.8892 & 0.0005 & @xmath120.0009 & 54 + 113 & 55067.9812 & 0.0003 & @xmath120.0004 & 54 + 114 & 55068.0719 & 0.0003 & @xmath120.0012 & 54 + 115 & 55068.1620 & 0.0005 & @xmath120.0028 & 53 + 116 & 55068.2543 & 0.0006 & @xmath120.0019 & 54 + 117 & 55068.3450 & 0.0006 & @xmath120.0027 & 54 + 118 & 55068.4376 & 0.0004 & @xmath120.0017 & 54 + 119 & 55068.5302 & 0.0006 & @xmath120.0006 & 54 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 120 & 55068.6203 & 0.0006 & @xmath120.0020 & 54 + 121 & 55068.7128 & 0.0005 & @xmath120.0011 & 54 + 122 & 55068.8044 & 0.0006 & @xmath120.0009 & 54 + 123 & 55068.8988 & 0.0010 & 0.0019 & 54 + 124 & 55068.9884 & 0.0020 & 0.0000 & 54 + 125 & 55069.0782 & 0.0005 & @xmath120.0017 & 54 + 126 & 55069.1682 & 0.0006 & @xmath120.0033 & 53 + 127 & 55069.2614 & 0.0006 & @xmath120.0015 & 53 + 128 & 55069.3531 & 0.0006 & @xmath120.0014 & 54 + 129 & 55069.4448 & 0.0006 & @xmath120.0012 & 54 + 130 & 55069.5342 & 0.0006 & @xmath120.0033 & 54 + 131 & 55069.6271 & 0.0018 & @xmath120.0019 & 54 + 132 & 55069.7217 & 0.0020 & 0.0012 & 54 + 133 & 55069.8121 & 0.0006 & 0.0000 & 54 + 134 & 55069.9016 & 0.0005 & @xmath120.0021 & 54 + 135 & 55069.9920 & 0.0003 & @xmath120.0031 & 54 + 136 & 55070.0827 & 0.0004 & @xmath120.0039 & 53 + 137 & 55070.1752 & 0.0007 & @xmath120.0030 & 54 + 138 & 55070.2681 & 0.0008 & @xmath120.0016 & 54 + 139 & 55070.3551 & 0.0005 & @xmath120.0061 & 54 + 140 & 55070.4493 & 0.0007 & @xmath120.0035 & 54 + 141 & 55070.5393 & 0.0004 & @xmath120.0050 & 54 + 142 & 55070.6358 & 0.0011 & 0.0000 & 54 + 143 & 55070.7252 & 0.0006 & @xmath120.0021 & 54 + 144 & 55070.8154 & 0.0005 & @xmath120.0034 & 54 + 145 & 55070.9067 & 0.0004 & @xmath120.0036 & 54 + 146 & 55071.0009 & 0.0010 & @xmath120.0010 & 54 + 147 & 55071.0900 & 0.0006 & @xmath120.0034 & 53 + 148 & 55071.1801 & 0.0005 & @xmath120.0048 & 53 + 149 & 55071.2654 & 0.0010 & @xmath120.0110 & 54 + 150 & 55071.3573 & 0.0004 & @xmath120.0107 & 53 + 151 & 55071.4471 & 0.0007 & @xmath120.0124 & 54 + 152 & 55071.5399 & 0.0005 & @xmath120.0111 & 54 + 153 & 55071.6266 & 0.0007 & @xmath120.0159 & 54 + 154 & 55071.7156 & 0.0006 & @xmath120.0184 & 54 + 155 & 55071.8050 & 0.0021 & @xmath120.0206 & 54 + 156 & 55071.9007 & 0.0006 & @xmath120.0164 & 54 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55065.9135 & 0.0009 & 0.0029 & 54 + 1 & 55066.0041 & 0.0021 & 0.0018 & 54 + 3 & 55066.1883 & 0.0012 & 0.0028 & 53 + 4 & 55066.2770 & 0.0009 & @xmath120.0001 & 54 + 5 & 55066.3677 & 0.0011 & @xmath120.0011 & 53 + 6 & 55066.4627 & 0.0008 & 0.0022 & 54 + 7 & 55066.5528 & 0.0008 & 0.0008 & 54 + 8 & 55066.6498 & 0.0006 & 0.0061 & 53 + 9 & 55066.7347 & 0.0009 & @xmath120.0006 & 54 + 10 & 55066.8240 & 0.0004 & @xmath120.0029 & 53 + 11 & 55066.9168 & 0.0007 & @xmath120.0017 & 54 + 12 & 55067.0051 & 0.0012 & @xmath120.0051 & 54 + 14 & 55067.1914 & 0.0013 & @xmath120.0020 & 54 + 15 & 55067.2814 & 0.0012 & @xmath120.0037 & 54 + 16 & 55067.3649 & 0.0012 & @xmath120.0118 & 54 + 18 & 55067.5586 & 0.0010 & @xmath120.0014 & 54 + 19 & 55067.6466 & 0.0007 & @xmath120.0050 & 54 + 20 & 55067.7406 & 0.0007 & @xmath120.0026 & 54 + 21 & 55067.8325 & 0.0017 & @xmath120.0024 & 54 + 22 & 55067.9235 & 0.0007 & @xmath120.0030 & 54 + 24 & 55068.1060 & 0.0009 & @xmath120.0038 & 54 + 25 & 55068.1964 & 0.0007 & @xmath120.0050 & 54 + 26 & 55068.2845 & 0.0013 & @xmath120.0085 & 54 + 27 & 55068.3782 & 0.0009 & @xmath120.0065 & 53 + 28 & 55068.4739 & 0.0036 & @xmath120.0024 & 54 + 29 & 55068.5636 & 0.0014 & @xmath120.0043 & 54 + 30 & 55068.6575 & 0.0005 & @xmath120.0021 & 54 + 31 & 55068.7487 & 0.0008 & @xmath120.0024 & 44 + 32 & 55068.8392 & 0.0011 & @xmath120.0036 & 54 + 34 & 55069.0220 & 0.0006 & @xmath120.0040 & 53 + 35 & 55069.1078 & 0.0011 & @xmath120.0099 & 54 + 36 & 55069.2086 & 0.0005 & @xmath120.0007 & 54 + 37 & 55069.2964 & 0.0009 & @xmath120.0045 & 54 + 38 & 55069.3879 & 0.0010 & @xmath120.0046 & 54 + 39 & 55069.4823 & 0.0003 & @xmath120.0019 & 54 + 40 & 55069.5726 & 0.0008 & @xmath120.0033 & 54 + 41 & 55069.6618 & 0.0007 & @xmath120.0057 & 54 + 42 & 55069.7558 & 0.0013 & @xmath120.0033 & 54 + 43 & 55069.8474 & 0.0013 & @xmath120.0034 & 53 + 44 & 55069.9411 & 0.0008 & @xmath120.0012 & 54 + 45 & 55070.0317 & 0.0004 & @xmath120.0023 & 54 + 46 & 55070.1236 & 0.0006 & @xmath120.0020 & 54 + 47 & 55070.2167 & 0.0009 & @xmath120.0006 & 54 + 48 & 55070.3079 & 0.0006 & @xmath120.0010 & 54 + 49 & 55070.4011 & 0.0004 & 0.0006 & 54 + 50 & 55070.4931 & 0.0006 & 0.0009 & 54 + 51 & 55070.5853 & 0.0004 & 0.0016 & 54 + 52 & 55070.6749 & 0.0004 & @xmath120.0005 & 54 + 53 & 55070.7641 & 0.0006 & @xmath120.0029 & 54 + 54 & 55070.8552 & 0.0009 & @xmath120.0035 & 54 + 55 & 55070.9495 & 0.0003 & @xmath120.0007 & 54 + 56 & 55071.0371 & 0.0006 & @xmath120.0048 & 54 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 57 & 55071.1300 & 0.0006 & @xmath120.0035 & 54 + 58 & 55071.2262 & 0.0010 & 0.0011 & 54 + 59 & 55071.3175 & 0.0011 & 0.0008 & 54 + 60 & 55071.4121 & 0.0005 & 0.0037 & 54 + 61 & 55071.5063 & 0.0010 & 0.0063 & 54 + 62 & 55071.5987 & 0.0007 & 0.0071 & 54 + 63 & 55071.6912 & 0.0006 & 0.0079 & 54 + 64 & 55071.7831 & 0.0007 & 0.0081 & 54 + 65 & 55071.8731 & 0.0006 & 0.0065 & 54 + 66 & 55071.9745 & 0.0016 & 0.0163 & 54 + 67 & 55072.0668 & 0.0019 & 0.0170 & 53 + 68 & 55072.1560 & 0.0014 & 0.0145 & 54 + 69 & 55072.2488 & 0.0023 & 0.0158 & 54 + 70 & 55072.3349 & 0.0018 & 0.0102 & 54 + 71 & 55072.4267 & 0.0009 & 0.0104 & 54 + 72 & 55072.5160 & 0.0009 & 0.0080 & 54 + 73 & 55072.6119 & 0.0010 & 0.0123 & 54 + 74 & 55072.7007 & 0.0010 & 0.0094 & 54 + 75 & 55072.7964 & 0.0015 & 0.0136 & 54 + 76 & 55072.8855 & 0.0017 & 0.0110 & 53 + 77 & 55072.9814 & 0.0007 & 0.0153 & 54 + 79 & 55073.1619 & 0.0014 & 0.0126 & 53 + 80 & 55073.2460 & 0.0010 & 0.0050 & 54 + 81 & 55073.3395 & 0.0007 & 0.0068 & 52 + 82 & 55073.4322 & 0.0013 & 0.0079 & 54 + 83 & 55073.5232 & 0.0006 & 0.0073 & 54 + 84 & 55073.6162 & 0.0005 & 0.0086 & 54 + 85 & 55073.7047 & 0.0009 & 0.0055 & 54 + 86 & 55073.7978 & 0.0006 & 0.0071 & 54 + 87 & 55073.8886 & 0.0006 & 0.0062 & 54 + 88 & 55073.9830 & 0.0005 & 0.0090 & 53 + 89 & 55074.0743 & 0.0013 & 0.0086 & 54 + 90 & 55074.1639 & 0.0005 & 0.0066 & 54 + 91 & 55074.2569 & 0.0006 & 0.0080 & 54 + 92 & 55074.3490 & 0.0014 & 0.0084 & 54 + 93 & 55074.4350 & 0.0010 & 0.0028 & 53 + 94 & 55074.5247 & 0.0007 & 0.0009 & 54 + 95 & 55074.6157 & 0.0008 & 0.0003 & 53 + 96 & 55074.7047 & 0.0005 & @xmath120.0023 & 54 + 97 & 55074.7957 & 0.0006 & @xmath120.0030 & 54 + 98 & 55074.8871 & 0.0006 & @xmath120.0033 & 53 + 99 & 55074.9791 & 0.0005 & @xmath120.0029 & 54 + 100 & 55075.0703 & 0.0006 & @xmath120.0033 & 54 + 101 & 55075.1604 & 0.0007 & @xmath120.0048 & 54 + 102 & 55075.2549 & 0.0008 & @xmath120.0020 & 53 + 103 & 55075.3466 & 0.0005 & @xmath120.0018 & 54 + 104 & 55075.4349 & 0.0008 & @xmath120.0052 & 54 + 105 & 55075.5297 & 0.0015 & @xmath120.0020 & 54 + 106 & 55075.6173 & 0.0010 & @xmath120.0061 & 54 + 107 & 55075.7055 & 0.0009 & @xmath120.0095 & 54 + 108 & 55075.7982 & 0.0012 & @xmath120.0085 & 54 + 109 & 55075.8896 & 0.0020 & @xmath120.0087 & 54 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 112 & 55076.1611 & 0.0008 & @xmath120.0120 & 54 + 113 & 55076.2642 & 0.0010 & @xmath120.0005 & 54 + 117 & 55076.6187 & 0.0019 & @xmath120.0126 & 54 + 118 & 55076.7048 & 0.0015 & @xmath120.0181 & 54 + 119 & 55076.7971 & 0.0007 & @xmath120.0174 & 53 + 120 & 55076.8900 & 0.0007 & @xmath120.0162 & 54 + 121 & 55076.9811 & 0.0008 & @xmath120.0168 & 54 + 122 & 55077.0731 & 0.0008 & @xmath120.0164 & 54 + 123 & 55077.1646 & 0.0015 & @xmath120.0165 & 54 + 124 & 55077.2577 & 0.0013 & @xmath120.0150 & 54 + 125 & 55077.3515 & 0.0010 & @xmath120.0128 & 54 + 126 & 55077.4458 & 0.0010 & @xmath120.0102 & 53 + 127 & 55077.5342 & 0.0011 & @xmath120.0134 & 54 + 128 & 55077.6270 & 0.0007 & @xmath120.0122 & 54 + 129 & 55077.7204 & 0.0011 & @xmath120.0105 & 43 + 130 & 55077.8142 & 0.0021 & @xmath120.0083 & 54 + 131 & 55077.9033 & 0.0007 & @xmath120.0108 & 54 + 132 & 55078.0032 & 0.0020 & @xmath120.0025 & 54 + 133 & 55078.0996 & 0.0027 & 0.0023 & 54 + 134 & 55078.1934 & 0.0012 & 0.0044 & 54 + 135 & 55078.2936 & 0.0030 & 0.0130 & 53 + 136 & 55078.3777 & 0.0011 & 0.0055 & 54 + 137 & 55078.4686 & 0.0015 & 0.0047 & 54 + 138 & 55078.5609 & 0.0010 & 0.0054 & 54 + 139 & 55078.6530 & 0.0016 & 0.0058 & 54 + 140 & 55078.7427 & 0.0013 & 0.0039 & 53 + 141 & 55078.8349 & 0.0008 & 0.0045 & 54 + 142 & 55078.9250 & 0.0010 & 0.0030 & 53 + 143 & 55079.0135 & 0.0010 & @xmath120.0001 & 54 + 144 & 55079.1104 & 0.0012 & 0.0052 & 53 + 146 & 55079.2897 & 0.0005 & 0.0011 & 54 + 146 & 55079.2897 & 0.0005 & 0.0011 & 54 + 147 & 55079.3781 & 0.0010 & @xmath120.0021 & 54 + 148 & 55079.4778 & 0.0008 & 0.0060 & 54 + 149 & 55079.5651 & 0.0007 & 0.0017 & 53 + 150 & 55079.6511 & 0.0029 & @xmath120.0039 & 54 + 151 & 55079.7558 & 0.0016 & 0.0091 & 53 + 152 & 55079.8427 & 0.0008 & 0.0043 & 53 + 154 & 55080.0151 & 0.0050 & @xmath120.0065 & 53 + 155 & 55080.1040 & 0.0007 & @xmath120.0092 & 54 + 156 & 55080.2077 & 0.0030 & 0.0029 & 54 + 157 & 55080.3105 & 0.0023 & 0.0140 & 54 + 158 & 55080.3926 & 0.0010 & 0.0045 & 54 + 160 & 55080.5653 & 0.0040 & @xmath120.0061 & 54 + 161 & 55080.6736 & 0.0021 & 0.0106 & 54 + 162 & 55080.7633 & 0.0015 & 0.0087 & 54 + 163 & 55080.8489 & 0.0009 & 0.0027 & 54 + 164 & 55080.9429 & 0.0034 & 0.0050 & 54 + 165 & 55081.0375 & 0.0025 & 0.0079 & 54 + 166 & 55081.1283 & 0.0021 & 0.0072 & 54 + 167 & 55081.2182 & 0.0023 & 0.0054 & 54 + 168 & 55081.3018 & 0.0020 & @xmath120.0026 & 54 + 169 & 55081.3934 & 0.0014 & @xmath120.0026 & 54 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 170 & 55081.5011 & 0.0059 & 0.0134 & 54 + 171 & 55081.5750 & 0.0017 & @xmath120.0043 & 53 + 172 & 55081.6704 & 0.0008 & @xmath120.0005 & 54 + 173 & 55081.7619 & 0.0013 & @xmath120.0006 & 54 + 174 & 55081.8498 & 0.0011 & @xmath120.0044 & 54 + 175 & 55081.9463 & 0.0022 & 0.0005 & 54 + 176 & 55082.0429 & 0.0020 & 0.0055 & 54 + 177 & 55082.1342 & 0.0054 & 0.0052 & 54 + 178 & 55082.2234 & 0.0091 & 0.0027 & 54 + 179 & 55082.3130 & 0.0008 & 0.0007 & 54 + 180 & 55082.4069 & 0.0015 & 0.0029 & 54 + 181 & 55082.4974 & 0.0015 & 0.0018 & 54 + 182 & 55082.5900 & 0.0007 & 0.0028 & 54 + 183 & 55082.6700 & 0.0013 & @xmath120.0088 & 54 + 184 & 55082.7594 & 0.0012 & @xmath120.0110 & 54 + 185 & 55082.8545 & 0.0009 & @xmath120.0076 & 54 + 186 & 55082.9446 & 0.0018 & @xmath120.0091 & 54 + 187 & 55083.0394 & 0.0018 & @xmath120.0059 & 53 + 188 & 55083.1421 & 0.0022 & 0.0051 & 54 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55161.9728 & 0.0038 & @xmath120.0600 & 54 + 1 & 55162.0796 & 0.0015 & @xmath120.0448 & 54 + 2 & 55162.1711 & 0.0022 & @xmath120.0447 & 54 + 3 & 55162.2755 & 0.0011 & @xmath120.0318 & 53 + 4 & 55162.3654 & 0.0006 & @xmath120.0335 & 54 + 5 & 55162.4583 & 0.0020 & @xmath120.0321 & 54 + 6 & 55162.5509 & 0.0004 & @xmath120.0310 & 54 + 7 & 55162.6437 & 0.0009 & @xmath120.0297 & 54 + 8 & 55162.7356 & 0.0004 & @xmath120.0294 & 54 + 9 & 55162.8323 & 0.0003 & @xmath120.0242 & 54 + 10 & 55162.9261 & 0.0003 & @xmath120.0219 & 54 + 11 & 55163.0186 & 0.0002 & @xmath120.0209 & 54 + 12 & 55163.1135 & 0.0002 & @xmath120.0175 & 54 + 13 & 55163.2077 & 0.0002 & @xmath120.0148 & 54 + 14 & 55163.3022 & 0.0001 & @xmath120.0118 & 53 + 15 & 55163.3965 & 0.0001 & @xmath120.0090 & 54 + 16 & 55163.4910 & 0.0001 & @xmath120.0060 & 54 + 17 & 55163.5833 & 0.0001 & @xmath120.0053 & 54 + 18 & 55163.6774 & 0.0001 & @xmath120.0027 & 54 + 19 & 55163.7713 & 0.0001 & @xmath120.0003 & 53 + 20 & 55163.8634 & 0.0001 & 0.0003 & 53 + 21 & 55163.9565 & 0.0001 & 0.0018 & 54 + 22 & 55164.0483 & 0.0001 & 0.0021 & 53 + 23 & 55164.1404 & 0.0001 & 0.0028 & 54 + 24 & 55164.2318 & 0.0001 & 0.0026 & 53 + 25 & 55164.3240 & 0.0001 & 0.0033 & 54 + 26 & 55164.4170 & 0.0001 & 0.0048 & 54 + 27 & 55164.5094 & 0.0002 & 0.0056 & 54 + 28 & 55164.6011 & 0.0001 & 0.0058 & 54 + 29 & 55164.6930 & 0.0001 & 0.0063 & 54 + 30 & 55164.7856 & 0.0001 & 0.0073 & 54 + 31 & 55164.8771 & 0.0001 & 0.0073 & 54 + 32 & 55164.9702 & 0.0002 & 0.0088 & 54 + 33 & 55165.0609 & 0.0001 & 0.0080 & 53 + 34 & 55165.1531 & 0.0002 & 0.0087 & 54 + 35 & 55165.2456 & 0.0001 & 0.0097 & 54 + 36 & 55165.3359 & 0.0002 & 0.0084 & 54 + 37 & 55165.4286 & 0.0002 & 0.0096 & 54 + 38 & 55165.5203 & 0.0001 & 0.0098 & 54 + 39 & 55165.6117 & 0.0002 & 0.0097 & 54 + 40 & 55165.7044 & 0.0002 & 0.0110 & 54 + 41 & 55165.7946 & 0.0003 & 0.0096 & 54 + 42 & 55165.8860 & 0.0002 & 0.0095 & 53 + 43 & 55165.9772 & 0.0002 & 0.0092 & 54 + 44 & 55166.0699 & 0.0002 & 0.0104 & 54 + 45 & 55166.1612 & 0.0003 & 0.0101 & 54 + 46 & 55166.2528 & 0.0003 & 0.0102 & 54 + 47 & 55166.3447 & 0.0002 & 0.0106 & 54 + 48 & 55166.4352 & 0.0004 & 0.0096 & 53 + 49 & 55166.5268 & 0.0002 & 0.0097 & 53 + 50 & 55166.6205 & 0.0002 & 0.0119 & 54 + 51 & 55166.7106 & 0.0003 & 0.0105 & 54 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 52 & 55166.8022 & 0.0003 & 0.0105 & 54 + 53 & 55166.8921 & 0.0003 & 0.0089 & 54 + 54 & 55166.9838 & 0.0002 & 0.0091 & 54 + 55 & 55167.0757 & 0.0002 & 0.0094 & 53 + 56 & 55167.1675 & 0.0004 & 0.0098 & 53 + 57 & 55167.2611 & 0.0002 & 0.0118 & 54 + 58 & 55167.3504 & 0.0002 & 0.0096 & 54 + 59 & 55167.4437 & 0.0003 & 0.0114 & 53 + 60 & 55167.5347 & 0.0002 & 0.0109 & 54 + 61 & 55167.6283 & 0.0002 & 0.0130 & 54 + 62 & 55167.7192 & 0.0004 & 0.0124 & 54 + 63 & 55167.8107 & 0.0002 & 0.0123 & 54 + 64 & 55167.9021 & 0.0003 & 0.0122 & 53 + 65 & 55167.9947 & 0.0004 & 0.0133 & 54 + 66 & 55168.0846 & 0.0003 & 0.0117 & 54 + 67 & 55168.1762 & 0.0003 & 0.0118 & 53 + 68 & 55168.2682 & 0.0002 & 0.0122 & 54 + 69 & 55168.3602 & 0.0003 & 0.0128 & 53 + 70 & 55168.4492 & 0.0002 & 0.0103 & 53 + 71 & 55168.5438 & 0.0002 & 0.0134 & 54 + 72 & 55168.6331 & 0.0001 & 0.0111 & 53 + 73 & 55168.7239 & 0.0002 & 0.0104 & 54 + 74 & 55168.8165 & 0.0003 & 0.0114 & 54 + 75 & 55168.9084 & 0.0003 & 0.0118 & 54 + 76 & 55168.9978 & 0.0002 & 0.0097 & 54 + 77 & 55169.0909 & 0.0002 & 0.0113 & 54 + 78 & 55169.1822 & 0.0003 & 0.0111 & 54 + 79 & 55169.2711 & 0.0004 & 0.0085 & 54 + 80 & 55169.3650 & 0.0003 & 0.0109 & 43 + 81 & 55169.4541 & 0.0002 & 0.0084 & 54 + 82 & 55169.5478 & 0.0002 & 0.0106 & 53 + 83 & 55169.6376 & 0.0002 & 0.0089 & 54 + 84 & 55169.7288 & 0.0002 & 0.0086 & 53 + 85 & 55169.8210 & 0.0003 & 0.0092 & 54 + 86 & 55169.9117 & 0.0002 & 0.0084 & 54 + 87 & 55170.0059 & 0.0002 & 0.0111 & 54 + 88 & 55170.0972 & 0.0004 & 0.0109 & 54 + 89 & 55170.1867 & 0.0003 & 0.0089 & 54 + 90 & 55170.2782 & 0.0002 & 0.0089 & 54 + 91 & 55170.3698 & 0.0002 & 0.0090 & 54 + 92 & 55170.4611 & 0.0003 & 0.0087 & 53 + 93 & 55170.5505 & 0.0003 & 0.0066 & 54 + 94 & 55170.6441 & 0.0003 & 0.0087 & 54 + 95 & 55170.7346 & 0.0004 & 0.0077 & 54 + 96 & 55170.8248 & 0.0003 & 0.0064 & 54 + 97 & 55170.9145 & 0.0003 & 0.0046 & 54 + 98 & 55171.0087 & 0.0005 & 0.0072 & 54 + 99 & 55171.0973 & 0.0003 & 0.0043 & 54 + 100 & 55171.1899 & 0.0005 & 0.0055 & 54 + 101 & 55171.2798 & 0.0003 & 0.0038 & 54 + 102 & 55171.3725 & 0.0005 & 0.0050 & 54 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 103 & 55171.4609 & 0.0005 & 0.0018 & 54 + 104 & 55171.5544 & 0.0005 & 0.0039 & 54 + 105 & 55171.6454 & 0.0006 & 0.0033 & 54 + 106 & 55171.7352 & 0.0005 & 0.0017 & 54 + 107 & 55171.8287 & 0.0005 & 0.0036 & 54 + 108 & 55171.9198 & 0.0011 & 0.0032 & 54 + 109 & 55172.0102 & 0.0006 & 0.0021 & 54 + 110 & 55172.1012 & 0.0007 & 0.0015 & 54 + 111 & 55172.1939 & 0.0013 & 0.0027 & 54 + 112 & 55172.2829 & 0.0006 & 0.0002 & 54 + 113 & 55172.3741 & 0.0006 & @xmath120.0001 & 54 + 114 & 55172.4662 & 0.0006 & 0.0005 & 54 + 115 & 55172.5578 & 0.0013 & 0.0005 & 54 + 116 & 55172.6468 & 0.0011 & @xmath120.0020 & 53 + 117 & 55172.7399 & 0.0007 & @xmath120.0003 & 54 + 118 & 55172.8318 & 0.0007 & 0.0000 & 54 + 119 & 55172.9256 & 0.0009 & 0.0023 & 54 + 120 & 55173.0126 & 0.0007 & @xmath120.0022 & 54 + 121 & 55173.1047 & 0.0024 & @xmath120.0016 & 54 + 122 & 55173.1982 & 0.0017 & 0.0004 & 53 + 123 & 55173.2902 & 0.0023 & 0.0009 & 54 + 124 & 55173.3778 & 0.0023 & @xmath120.0031 & 54 + 125 & 55173.4699 & 0.0015 & @xmath120.0025 & 54 + 126 & 55173.5618 & 0.0010 & @xmath120.0021 & 54 + 127 & 55173.6550 & 0.0005 & @xmath120.0004 & 54 + 128 & 55173.7456 & 0.0004 & @xmath120.0014 & 53 + 129 & 55173.8366 & 0.0004 & @xmath120.0018 & 54 + 130 & 55173.9276 & 0.0004 & @xmath120.0023 & 53 + 131 & 55174.0180 & 0.0004 & @xmath120.0035 & 54 + 132 & 55174.1120 & 0.0003 & @xmath120.0010 & 54 + 133 & 55174.2027 & 0.0003 & @xmath120.0018 & 54 + 134 & 55174.2930 & 0.0003 & @xmath120.0031 & 54 + 135 & 55174.3858 & 0.0003 & @xmath120.0017 & 53 + 136 & 55174.4775 & 0.0003 & @xmath120.0016 & 54 + 137 & 55174.5685 & 0.0004 & @xmath120.0021 & 54 + 138 & 55174.6600 & 0.0003 & @xmath120.0021 & 54 + 139 & 55174.7515 & 0.0002 & @xmath120.0022 & 54 + 140 & 55174.8438 & 0.0004 & @xmath120.0014 & 53 + 141 & 55174.9362 & 0.0004 & @xmath120.0005 & 54 + 142 & 55175.0274 & 0.0006 & @xmath120.0008 & 54 + 143 & 55175.1174 & 0.0002 & @xmath120.0023 & 54 + 144 & 55175.2096 & 0.0004 & @xmath120.0016 & 54 + 145 & 55175.3010 & 0.0003 & @xmath120.0017 & 54 + 146 & 55175.3922 & 0.0005 & @xmath120.0020 & 54 + 147 & 55175.4828 & 0.0004 & @xmath120.0030 & 54 + 148 & 55175.5755 & 0.0006 & @xmath120.0017 & 54 + 149 & 55175.6669 & 0.0004 & @xmath120.0019 & 54 + 150 & 55175.7580 & 0.0003 & @xmath120.0023 & 54 + 151 & 55175.8477 & 0.0005 & @xmath120.0041 & 54 + 152 & 55175.9415 & 0.0003 & @xmath120.0018 & 54 + 153 & 55176.0329 & 0.0003 & @xmath120.0020 & 54 + 154 & 55176.1198 & 0.0004 & @xmath120.0066 & 54 + 155 & 55176.2125 & 0.0004 & @xmath120.0054 & 54 + 156 & 55176.3000 & 0.0006 & @xmath120.0094 & 54 + 157 & 55176.3901 & 0.0005 & @xmath120.0108 & 54 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 158 & 55176.4829 & 0.0006 & @xmath120.0096 & 54 + 159 & 55176.5708 & 0.0007 & @xmath120.0132 & 53 + 160 & 55176.6610 & 0.0009 & @xmath120.0144 & 54 + 161 & 55176.7532 & 0.0008 & @xmath120.0138 & 54 + 162 & 55176.8418 & 0.0008 & @xmath120.0167 & 54 + 163 & 55176.9339 & 0.0009 & @xmath120.0161 & 53 + 164 & 55177.0264 & 0.0005 & @xmath120.0151 & 54 + 165 & 55177.1167 & 0.0008 & @xmath120.0163 & 54 + 166 & 55177.2072 & 0.0008 & @xmath120.0174 & 54 + 167 & 55177.3015 & 0.0006 & @xmath120.0146 & 54 + 168 & 55177.3911 & 0.0006 & @xmath120.0165 & 54 + 169 & 55177.4827 & 0.0007 & @xmath120.0164 & 54 + 170 & 55177.5723 & 0.0015 & @xmath120.0184 & 54 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55169.2136 & 0.0006 & 0.0047 & 54 + 1 & 55169.2986 & 0.0013 & @xmath120.0018 & 54 + 2 & 55169.4014 & 0.0008 & 0.0094 & 54 + 3 & 55169.4837 & 0.0014 & 0.0001 & 54 + 4 & 55169.5834 & 0.0009 & 0.0083 & 53 + 5 & 55169.6736 & 0.0009 & 0.0070 & 53 + 6 & 55169.7640 & 0.0008 & 0.0058 & 54 + 7 & 55169.8514 & 0.0007 & 0.0017 & 53 + 8 & 55169.9480 & 0.0008 & 0.0067 & 54 + 9 & 55170.0308 & 0.0010 & @xmath120.0020 & 54 + 10 & 55170.1154 & 0.0020 & @xmath120.0090 & 54 + 11 & 55170.2167 & 0.0012 & 0.0008 & 54 + 12 & 55170.3006 & 0.0023 & @xmath120.0070 & 53 + 13 & 55170.3899 & 0.0011 & @xmath120.0092 & 54 + 14 & 55170.4816 & 0.0012 & @xmath120.0091 & 54 + 15 & 55170.5828 & 0.0011 & 0.0005 & 54 + 16 & 55170.6773 & 0.0007 & 0.0036 & 54 + 17 & 55170.7649 & 0.0010 & @xmath120.0004 & 54 + 18 & 55170.8598 & 0.0005 & 0.0029 & 54 + 19 & 55170.9441 & 0.0008 & @xmath120.0043 & 54 + 20 & 55171.0378 & 0.0005 & @xmath120.0022 & 54 + 21 & 55171.1286 & 0.0007 & @xmath120.0029 & 53 + 22 & 55171.2242 & 0.0005 & 0.0011 & 54 + 23 & 55171.3159 & 0.0006 & 0.0012 & 53 + 24 & 55171.4048 & 0.0009 & @xmath120.0014 & 53 + 25 & 55171.4959 & 0.0012 & @xmath120.0019 & 54 + 26 & 55171.5851 & 0.0007 & @xmath120.0043 & 54 + 27 & 55171.6751 & 0.0009 & @xmath120.0058 & 54 + 28 & 55171.7609 & 0.0017 & @xmath120.0115 & 54 + 29 & 55171.8578 & 0.0017 & @xmath120.0063 & 54 + 30 & 55171.9443 & 0.0043 & @xmath120.0113 & 53 + 31 & 55172.0419 & 0.0007 & @xmath120.0052 & 53 + 32 & 55172.1351 & 0.0006 & @xmath120.0036 & 54 + 33 & 55172.2252 & 0.0014 & @xmath120.0051 & 53 + 34 & 55172.3105 & 0.0030 & @xmath120.0114 & 50 + 35 & 55172.4060 & 0.0008 & @xmath120.0074 & 53 + 36 & 55172.4979 & 0.0006 & @xmath120.0071 & 54 + 37 & 55172.5903 & 0.0009 & @xmath120.0062 & 54 + 38 & 55172.6855 & 0.0008 & @xmath120.0026 & 54 + 39 & 55172.7774 & 0.0008 & @xmath120.0023 & 54 + 40 & 55172.8653 & 0.0008 & @xmath120.0059 & 54 + 41 & 55172.9539 & 0.0017 & @xmath120.0088 & 54 + 42 & 55173.0500 & 0.0008 & @xmath120.0043 & 53 + 43 & 55173.1440 & 0.0010 & @xmath120.0019 & 54 + 44 & 55173.2254 & 0.0015 & @xmath120.0120 & 54 + 45 & 55173.3170 & 0.0031 & @xmath120.0119 & 54 + 46 & 55173.4088 & 0.0012 & @xmath120.0117 & 54 + 47 & 55173.5088 & 0.0008 & @xmath120.0033 & 54 + 48 & 55173.6008 & 0.0009 & @xmath120.0028 & 54 + 49 & 55173.6908 & 0.0006 & @xmath120.0044 & 53 + 50 & 55173.7856 & 0.0003 & @xmath120.0012 & 54 + 51 & 55173.8774 & 0.0003 & @xmath120.0009 & 54 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 52 & 55173.9671 & 0.0005 & @xmath120.0027 & 54 + 53 & 55174.0576 & 0.0007 & @xmath120.0038 & 54 + 54 & 55174.1508 & 0.0004 & @xmath120.0022 & 54 + 55 & 55174.2407 & 0.0006 & @xmath120.0038 & 54 + 56 & 55174.3332 & 0.0005 & @xmath120.0029 & 54 + 57 & 55174.4235 & 0.0006 & @xmath120.0042 & 54 + 58 & 55174.5187 & 0.0004 & @xmath120.0005 & 54 + 59 & 55174.6140 & 0.0024 & 0.0032 & 53 + 60 & 55174.6982 & 0.0008 & @xmath120.0041 & 54 + 61 & 55174.7901 & 0.0008 & @xmath120.0038 & 54 + 62 & 55174.8844 & 0.0004 & @xmath120.0010 & 54 + 63 & 55174.9743 & 0.0005 & @xmath120.0027 & 54 + 64 & 55175.0689 & 0.0004 & 0.0003 & 54 + 65 & 55175.1595 & 0.0004 & @xmath120.0006 & 54 + 66 & 55175.2512 & 0.0009 & @xmath120.0005 & 53 + 67 & 55175.3424 & 0.0004 & @xmath120.0008 & 43 + 68 & 55175.4344 & 0.0008 & @xmath120.0004 & 54 + 69 & 55175.5275 & 0.0010 & 0.0012 & 54 + 70 & 55175.6195 & 0.0009 & 0.0016 & 54 + 71 & 55175.7121 & 0.0005 & 0.0026 & 54 + 72 & 55175.8018 & 0.0006 & 0.0008 & 54 + 73 & 55175.8955 & 0.0004 & 0.0029 & 53 + 74 & 55175.9856 & 0.0005 & 0.0014 & 54 + 75 & 55176.0753 & 0.0004 & @xmath120.0004 & 54 + 76 & 55176.1685 & 0.0007 & 0.0012 & 54 + 77 & 55176.2598 & 0.0007 & 0.0010 & 54 + 78 & 55176.3579 & 0.0006 & 0.0075 & 54 + 79 & 55176.4474 & 0.0007 & 0.0055 & 54 + 80 & 55176.5416 & 0.0006 & 0.0081 & 54 + 81 & 55176.6308 & 0.0007 & 0.0057 & 54 + 82 & 55176.7212 & 0.0008 & 0.0046 & 53 + 83 & 55176.8121 & 0.0006 & 0.0039 & 53 + 84 & 55176.9093 & 0.0010 & 0.0096 & 54 + 85 & 55176.9977 & 0.0007 & 0.0064 & 54 + 86 & 55177.1033 & 0.0027 & 0.0205 & 54 + 87 & 55177.1789 & 0.0007 & 0.0045 & 53 + 88 & 55177.2777 & 0.0009 & 0.0118 & 54 + 89 & 55177.3659 & 0.0008 & 0.0084 & 53 + 90 & 55177.4574 & 0.0012 & 0.0083 & 53 + 91 & 55177.5585 & 0.0016 & 0.0179 & 54 + 92 & 55177.6467 & 0.0009 & 0.0145 & 53 + 93 & 55177.7382 & 0.0017 & 0.0145 & 54 + 94 & 55177.8228 & 0.0008 & 0.0075 & 53 + 95 & 55177.9264 & 0.0022 & 0.0196 & 54 + 96 & 55178.0070 & 0.0011 & 0.0086 & 54 + 97 & 55178.1072 & 0.0019 & 0.0172 & 54 + 98 & 55178.2022 & 0.0018 & 0.0207 & 54 + 99 & 55178.2879 & 0.0004 & 0.0148 & 53 + 100 & 55178.3805 & 0.0014 & 0.0159 & 54 + 101 & 55178.4703 & 0.0009 & 0.0141 & 54 + 102 & 55178.5656 & 0.0013 & 0.0179 & 54 + 103 & 55178.6510 & 0.0007 & 0.0117 & 54 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 104 & 55178.7409 & 0.0007 & 0.0100 & 54 + 105 & 55178.8317 & 0.0007 & 0.0092 & 54 + 106 & 55178.9270 & 0.0005 & 0.0130 & 54 + 107 & 55179.0168 & 0.0008 & 0.0113 & 54 + 108 & 55179.1049 & 0.0005 & 0.0078 & 54 + 109 & 55179.1960 & 0.0005 & 0.0074 & 53 + 110 & 55179.2890 & 0.0007 & 0.0088 & 54 + 111 & 55179.3790 & 0.0008 & 0.0073 & 54 + 112 & 55179.4696 & 0.0008 & 0.0063 & 54 + 113 & 55179.5597 & 0.0003 & 0.0048 & 54 + 114 & 55179.6504 & 0.0004 & 0.0040 & 54 + 115 & 55179.7410 & 0.0004 & 0.0030 & 54 + 116 & 55179.8329 & 0.0003 & 0.0033 & 54 + 117 & 55179.9272 & 0.0006 & 0.0061 & 54 + 118 & 55180.0194 & 0.0008 & 0.0067 & 54 + 119 & 55180.1082 & 0.0010 & 0.0040 & 54 + 120 & 55180.1999 & 0.0007 & 0.0041 & 54 + 121 & 55180.2915 & 0.0009 & 0.0041 & 54 + 122 & 55180.3793 & 0.0007 & 0.0004 & 54 + 123 & 55180.4719 & 0.0008 & 0.0014 & 53 + 124 & 55180.5627 & 0.0005 & 0.0007 & 54 + 125 & 55180.6505 & 0.0007 & @xmath120.0030 & 53 + 126 & 55180.7424 & 0.0004 & @xmath120.0028 & 54 + 127 & 55180.8324 & 0.0006 & @xmath120.0043 & 54 + 128 & 55180.9238 & 0.0007 & @xmath120.0044 & 54 + 129 & 55181.0111 & 0.0005 & @xmath120.0087 & 54 + 130 & 55181.1023 & 0.0008 & @xmath120.0091 & 54 + 131 & 55181.1919 & 0.0005 & @xmath120.0110 & 54 + 132 & 55181.2845 & 0.0009 & @xmath120.0100 & 42 + 133 & 55181.3786 & 0.0010 & @xmath120.0075 & 54 + 134 & 55181.4769 & 0.0027 & @xmath120.0008 & 54 + 135 & 55181.5639 & 0.0009 & @xmath120.0053 & 54 + 136 & 55181.6497 & 0.0012 & @xmath120.0110 & 54 + 137 & 55181.7349 & 0.0010 & @xmath120.0174 & 54 + 138 & 55181.8267 & 0.0005 & @xmath120.0172 & 54 + 139 & 55181.9202 & 0.0006 & @xmath120.0152 & 54 + 140 & 55182.0079 & 0.0029 & @xmath120.0191 & 54 + 141 & 55182.0987 & 0.0013 & @xmath120.0199 & 54 + 142 & 55182.1864 & 0.0010 & @xmath120.0237 & 54 + 143 & 55182.2850 & 0.0018 & @xmath120.0166 & 54 + 144 & 55182.3722 & 0.0005 & @xmath120.0210 & 53 + 145 & 55182.4641 & 0.0008 &
@xmath120.0206 & 54 + + + + very little has been known for this dwarf nova .
the object was originally selected as a recurrent nova or a mira - type star @xcite based on two detections at a photographic magnitude of 16.2 on 1956 june 12 and 15.8 on 1959 april 29 .
@xcite favored the latter classification based on its red color .
p. schmeer , however , noted that this object is a dwarf nova and reported an outburst on 1999 june 16 at an unfiltered ccd magnitude of 16.4 ( vsnet - alert 3086 ) .
j. kemp conducted multicolor and time - resolved photometry upon this information and reported the detection of superhumps with an amplitude @xmath90.20 mag and a period of @xmath90.068 d ( vsnet - alert 3087 ) .
there have been three other known outburst detections ( all observations gave unfiltered ccd magnitudes ) : 2004 february 8 ( 16.1 mag , b. monard ) , 2004 september 2 ( 17.3 , b. monard ) and 2005 april 17 ( 16.3 , p. schmeer ) .
the object was again detected in outburst by m. simonsen at a magnitude of @xmath35=17.54 on 2011 april 3.390 and @xmath35=16.00 on april 9.374 .
subsequent observations indeed detected superhumps ( vsnet - alert 13144 , 13151 ) .
the times of superhump maxima are listed in table [ tab : v1195ophoc2011 ] . since the first observations were obtained 9 d after the recorded maximum , the observation likely recorded only stage c superhumps .
ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55664.2522 & 0.0026 & @xmath120.0003 & 132 + 9 & 55664.8595 & 0.0006 & 0.0018 & 70 + 10 & 55664.9236 & 0.0009 & @xmath120.0014 & 69 + 24 & 55665.8666 & 0.0007 & 0.0002 & 69 + 25 & 55665.9334 & 0.0005 & @xmath120.0003 & 70 + + + + v1212 tau was discovered as an eruptive object near m45 @xcite , whose location might suggest one of pleiades flare stars .
although the object remained unnamed for long , members of the variable star observers league in japan ( vsolj ) suspected that the object is more likely a dwarf nova based on published observations , and continued their own monitoring since 1987 .
there was a visual record of a bright outburst in 1987 october , which remained unconfirmed .
the object was detected in outburst in 2007 january by g. gualdoni ( vsnet - alert 9190 ) .
although observations of this outburst was rather fragmentary , j. patterson reported the detection of superhumps with a period of 0.075 d.http://cbastro.org/communications/news/messages/0518.html@xmath15 . ]
the object was again detected in outburst in 2011 january ( baavss - alert 2473 , vsnet - alert 12718 ) .
this outburst was detected sufficiently early , and full evolution of superhumps were recorded ( vsnet - alert 12727 , 12730 , 12736 , 12743 ; figure [ fig : v1212taushpdm ] ) .
the time of superhump maxima are listed in table [ tab : v1212tauoc2011 ] .
stages a c can be well recognized .
the double - wave modulations reported during the early stage were likely stage a superhumps having longer period than stage b superhumps.120308.7 .
the present case clearly excludes this possibility because of its long period ( e.g. vsnet - alert 12727 , 12730 ) .
the result and discussion by @xcite probably need to be re - examined in the light of the knowledge from this object . ]
the evolution of superhumps is quite typical for a system with this @xmath2 .
( 88mm,110mm)fig18.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55589.6637 & 0.0028 & @xmath120.0434 & 99 + 1 & 55589.7459 & 0.0022 & @xmath120.0313 & 98 + 11 & 55590.4712 & 0.0011 & @xmath120.0066 & 55 + 12 & 55590.5479 & 0.0010 & 0.0000 & 68 + 13 & 55590.6165 & 0.0003 & @xmath120.0014 & 97 + 14 & 55590.6863 & 0.0003 & @xmath120.0017 & 99 + 15 & 55590.7588 & 0.0003 & 0.0007 & 99 + 20 & 55591.1074 & 0.0008 & @xmath120.0010 & 131 + 24 & 55591.3915 & 0.0003 & 0.0028 & 100 + 25 & 55591.4617 & 0.0003 & 0.0030 & 126 + 26 & 55591.5353 & 0.0007 & 0.0066 & 70 + 27 & 55591.6027 & 0.0002 & 0.0039 & 94 + 28 & 55591.6722 & 0.0002 & 0.0033 & 99 + 29 & 55591.7429 & 0.0002 & 0.0040 & 97 + 33 & 55592.0232 & 0.0015 & 0.0039 & 134 + 37 & 55592.3031 & 0.0005 & 0.0036 & 50 + 38 & 55592.3727 & 0.0004 & 0.0032 & 80 + 39 & 55592.4430 & 0.0004 & 0.0034 & 135 + 42 & 55592.6531 & 0.0004 & 0.0033 & 74 + 43 & 55592.7228 & 0.0006 & 0.0029 & 49 + 46 & 55592.9302 & 0.0017 & 0.0002 & 104 + 47 & 55593.0036 & 0.0014 & 0.0035 & 128 + 52 & 55593.3513 & 0.0005 & 0.0009 & 101 + 53 & 55593.4217 & 0.0008 & 0.0012 & 106 + 54 & 55593.4926 & 0.0009 & 0.0020 & 69 + 56 & 55593.6330 & 0.0009 & 0.0023 & 32 + 57 & 55593.7030 & 0.0010 & 0.0023 & 36 + 66 & 55594.3340 & 0.0008 & 0.0027 & 100 + 67 & 55594.4041 & 0.0007 & 0.0027 & 103 + 68 & 55594.4734 & 0.0009 & 0.0020 & 82 + 71 & 55594.6872 & 0.0023 & 0.0055 & 37 + 72 & 55594.7543 & 0.0015 & 0.0026 & 36 + 76 & 55595.0391 & 0.0042 & 0.0071 & 144 + 80 & 55595.3159 & 0.0011 & 0.0036 & 90 + 81 & 55595.3886 & 0.0007 & 0.0063 & 177 + 82 & 55595.4588 & 0.0006 & 0.0064 & 165 + 84 & 55595.5992 & 0.0009 & 0.0067 & 24 + 85 & 55595.6695 & 0.0014 & 0.0070 & 35 + 89 & 55595.9553 & 0.0022 & 0.0124 & 135 + 90 & 55596.0145 & 0.0035 & 0.0016 & 147 + 95 & 55596.3732 & 0.0009 & 0.0100 & 67 + 96 & 55596.4406 & 0.0008 & 0.0074 & 68 + 109 & 55597.3511 & 0.0010 & 0.0070 & 107 + 110 & 55597.4186 & 0.0012 & 0.0044 & 107 + 111 & 55597.4879 & 0.0026 & 0.0036 & 67 + 118 & 55597.9668 & 0.0083 & @xmath120.0079 & 119 + 123 & 55598.3245 & 0.0019 & @xmath120.0005 & 66 + 124 & 55598.3951 & 0.0009 & 0.0001 & 104 + 125 & 55598.4645 & 0.0011 & @xmath120.0006 & 96 + 137 & 55599.3017 & 0.0015 & @xmath120.0042 & 40 + 138 & 55599.3693 & 0.0012 & @xmath120.0067 & 104 + 139 & 55599.4410 & 0.0012 & @xmath120.0051 & 97 + 152 & 55600.3453 & 0.0013 & @xmath120.0116 & 28 + 153 & 55600.4174 & 0.0011 & @xmath120.0096 & 36 + 166 & 55601.3246 & 0.0022 & @xmath120.0131 & 33 + 167 & 55601.3966 & 0.0019 & @xmath120.0112 & 27 + + + + we observed the 2010 superoutburst .
this outburst occurred after a relatively long ( @xmath9890 d ) interval following the 2008 superoutburst .
the outburst was relatively well observed despite the rather adverse seasonal conditions .
since some of the present observations were obtained at high airmasses , we used a second - order correction for atmospheric extinction as in v844 her .
the times of superhump maxima are listed in table [ tab : swumaoc2010 ] .
although not all parts were observed , all stages a c are readily identified .
a comparison of @xmath5 diagrams in different superoutbursts is shown in figure [ fig : swumacomp2 ] .
the 2010 superoutburst apparently followed the similar trend to that of the 2006 superoutburst @xcite , another bright superoutburst preceded by relatively long quiescence .
( 88mm,70mm)fig19.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55535.2103 & 0.0004 & @xmath120.0056 & 234 + 1 & 55535.2715 & 0.0005 & @xmath120.0026 & 233 + 2 & 55535.3296 & 0.0006 & @xmath120.0025 & 149 + 17 & 55536.2070 & 0.0002 & 0.0024 & 65 + 18 & 55536.2659 & 0.0009 & 0.0032 & 73 + 52 & 55538.2481 & 0.0003 & 0.0078 & 123 + 53 & 55538.3071 & 0.0002 & 0.0087 & 114 + 54 & 55538.3646 & 0.0002 & 0.0080 & 124 + 67 & 55539.1128 & 0.0003 & 0.0001 & 233 + 68 & 55539.1725 & 0.0003 & 0.0017 & 88 + 84 & 55540.0979 & 0.0002 & @xmath120.0035 & 82 + 85 & 55540.1560 & 0.0002 & @xmath120.0036 & 358 + 86 & 55540.2135 & 0.0002 & @xmath120.0042 & 357 + 87 & 55540.2734 & 0.0001 & @xmath120.0025 & 343 + 88 & 55540.3316 & 0.0001 & @xmath120.0025 & 349 + 102 & 55541.1411 & 0.0006 & @xmath120.0072 & 127 + 103 & 55541.2019 & 0.0002 & @xmath120.0046 & 222 + 109 & 55541.5515 & 0.0002 & @xmath120.0039 & 72 + 110 & 55541.6121 & 0.0006 & @xmath120.0016 & 47 + 111 & 55541.6683 & 0.0003 & @xmath120.0035 & 82 + 112 & 55541.7251 & 0.0003 & @xmath120.0049 & 67 + 119 & 55542.1337 & 0.0009 & @xmath120.0034 & 94 + 120 & 55542.1894 & 0.0003 & @xmath120.0059 & 206 + 121 & 55542.2495 & 0.0002 & @xmath120.0039 & 303 + 122 & 55542.3094 & 0.0002 & @xmath120.0022 & 289 + 123 & 55542.3690 & 0.0004 & @xmath120.0007 & 264 + 171 & 55545.1669 & 0.0006 & 0.0055 & 179 + 172 & 55545.2290 & 0.0004 & 0.0094 & 302 + 173 & 55545.2914 & 0.0005 & 0.0136 & 303 + 174 & 55545.3495 & 0.0007 & 0.0136 & 169 + 205 & 55547.1486 & 0.0002 & 0.0096 & 140 + 226 & 55548.3632 & 0.0008 & 0.0028 & 126 + 240 & 55549.1758 & 0.0004 & 0.0012 & 221 + 241 & 55549.2370 & 0.0004 & 0.0043 & 245 + 242 & 55549.2992 & 0.0005 & 0.0083 & 248 + 275 & 55551.1981 & 0.0017 & @xmath120.0121 & 62 + 276 & 55551.2560 & 0.0049 & @xmath120.0124 & 82 + 277 & 55551.3199 & 0.0020 & @xmath120.0067 & 83 + + + + we observed the 2011 superoutburst of this object ( table [ tab : ciumaoc2011 ] ) .
stage b with positive @xmath13 of @xmath36 and a part of stage c were observed .
the behavior of @xmath5 variation is very similar between different superoutbursts ( figure [ fig : ciumacomp2 ] ) .
( 88mm,70mm)fig20.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55660.4340 & 0.0002 & 0.0005 & 58 + 1 & 55660.4971 & 0.0003 & 0.0011 & 64 + 2 & 55660.5609 & 0.0003 & 0.0022 & 67 + 3 & 55660.6224 & 0.0003 & 0.0011 & 68 + 17 & 55661.4961 & 0.0004 & @xmath120.0023 & 57 + 18 & 55661.5582 & 0.0005 & @xmath120.0028 & 68 + 19 & 55661.6232 & 0.0006 & @xmath120.0005 & 63 + 32 & 55662.4335 & 0.0008 & @xmath120.0047 & 127 + 33 & 55662.4965 & 0.0005 & @xmath120.0043 & 90 + 48 & 55663.4403 & 0.0007 & @xmath120.0003 & 66 + 49 & 55663.5023 & 0.0006 & @xmath120.0009 & 62 + 50 & 55663.5679 & 0.0005 & 0.0021 & 55 + 63 & 55664.3807 & 0.0014 & 0.0005 & 32 + 64 & 55664.4481 & 0.0010 & 0.0052 & 34 + 65 & 55664.5112 & 0.0013 & 0.0057 & 34 + 66 & 55664.5705 & 0.0019 & 0.0023 & 24 + 111 & 55667.3863 & 0.0061 & @xmath120.0010 & 33 + 112 & 55667.4491 & 0.0024 & @xmath120.0008 & 34 + 113 & 55667.5095 & 0.0024 & @xmath120.0031 & 23 + + + + we observed the 2011 superoutburst .
the maxima were determined from observations outside the eclipses ( table [ tab : dvumaoc2011 ] ) .
the 2011 observation only recorded the final portion of stage b and partly stage c ( cf .
figure [ fig : dvumacomp2 ] ) , and the resultant @xmath13 was rather unreliable .
( 88mm,70mm)fig21.eps cccccc @xmath10 & max & error & @xmath5 & phase & @xmath11 + 0 & 55608.4897 & 0.0004 & @xmath120.0020 & 0.40 & 147 + 2 & 55608.6668 & 0.0004 & @xmath120.0023 & 0.46 & 138 + 3 & 55608.7540 & 0.0007 & @xmath120.0038 & 0.48 & 117 + 35 & 55611.5967 & 0.0004 & 0.0007 & 0.59 & 138 + 36 & 55611.6864 & 0.0006 & 0.0016 & 0.63 & 133 + 37 & 55611.7805 & 0.0008 & 0.0070 & 0.73 & 97 + 51 & 55613.0253 & 0.0037 & 0.0102 & 0.23 & 203 + 52 & 55613.1145 & 0.0016 & 0.0106 & 0.27 & 178 + 63 & 55614.0703 & 0.0012 & @xmath120.0092 & 0.40 & 174 + 64 & 55614.1554 & 0.0005 & @xmath120.0128 & 0.39 & 174 + + + + + this object was discovered as an eruptive object by k. itagaki in 2007 november ( vsnet - outburst 8245 ) .
the object was observed to fade by 0.3 mag in seven days .
because of the presence of a rosat x - ray counterpart , we call this object in its rosat name ( hereafter 1rxs j0038 ) .
the object was again reported to be in outburst on 2010 october 21 at an unfiltered ccd magnitude of 14.6 ( e. muyllaert , vsnet - alert 12295 ) .
early observations suggested the presence of superhumps ( vsnet - alert 12313 ) .
the presence of long - period superhumps was finally established ( vsnet - alert 12318 ) . the long outburst in 2007
also must have been a superoutburst .
the times of superhump maxima are listed in table [ tab : j0038oc2010 ] .
the maximum at @xmath37 was recorded after the rapid fading , and is not very reliable .
an analysis of the single - night continuous run yielded a period of 0.0985(6 ) d. the pdm analysis in figure [ fig : j0038shpdm ] used all observations , including those after the rapid fading , and yielded a mean period of 0.09708(8 ) d. we adopted the latter period in table [ tab : perlist ] .
( 88mm,110mm)fig22.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55495.3027 & 0.0008 & @xmath120.0020 & 94 + 1 & 55495.4040 & 0.0009 & 0.0020 & 97 + 42 & 55499.3928 & 0.0032 & @xmath120.0000 & 6
+ + + + this object ( hereafter 2qz j2224 ) was selected as a cv during the 2df survey for quasars @xcite .
the spectrum was typical for a dwarf nova in quiescence with strong balmer and hei emission lines .
there is not a strong indication of an underlying white dwarf as in many wz sge - type dwarf novae ( see also vsnet - alert 12167 ) .
the object was for the first time detected in outburst by the crts siding spring survey ( sss ) ( p. wils , cvnet - outburst 3836 ) .
subsequent observations confirmed the presence of superhumps ( vsnet - alert 12171 , 12180 ) .
the times of superhump maxima are listed in table [ tab : j2224oc2010 ] .
the scatter was relatively large due to the low amplitudes of superhumps .
combined with the apparent lack of a large variation in the period , it is likely we observed the stage c superhumps during their decay phase .
a pdm analysis ( figure [ fig : j2224shpdm ] ) has yielded a mean period of 0.05824(1 ) d , which is adopted in table [ tab : perlist ] .
( 88mm,110mm)fig23.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55452.3543 & 0.0007 & 0.0006 & 127 + 1 & 55452.4112 & 0.0007 & @xmath120.0008 & 130 + 2 & 55452.4665 & 0.0008 & @xmath120.0037 & 129 + 3 & 55452.5278 & 0.0009 & @xmath120.0007 & 129 + 4 & 55452.5829 & 0.0010 & @xmath120.0039 & 115 + 16 & 55453.2838 & 0.0013 & @xmath120.0022 & 129 + 17 & 55453.3415 & 0.0009 & @xmath120.0027 & 129 + 18 & 55453.3999 & 0.0011 & @xmath120.0026 & 129 + 19 & 55453.4587 & 0.0013 & @xmath120.0021 & 129 + 20 & 55453.5170 & 0.0012 & @xmath120.0020 & 129 + 35 & 55454.4010 & 0.0031 & 0.0081 & 120 + 36 & 55454.4573 & 0.0033 & 0.0061 & 129 + 37 & 55454.5074 & 0.0017 & @xmath120.0021 & 114 + 52 & 55455.3948 & 0.0038 & 0.0115 & 131 + 53 & 55455.4424 & 0.0029 & 0.0007 & 132 + 54 & 55455.5176 & 0.0051 & 0.0177 & 131 + 55 & 55455.5533 & 0.0022 & @xmath120.0049 & 132 + 64 & 55456.0832 & 0.0014 & 0.0006 & 120 + 65 & 55456.1285 & 0.0025 & @xmath120.0123 & 121 + 70 & 55456.4307 & 0.0017 & @xmath120.0015 & 131 + 71 & 55456.4994 & 0.0026 & 0.0090 & 116 + 72 & 55456.5493 & 0.0026 & 0.0006 & 124 + 85 & 55457.3044 & 0.0053 & @xmath120.0017 & 77 + 86 & 55457.3627 & 0.0031 & @xmath120.0016 & 96 + 88 & 55457.4791 & 0.0021 & @xmath120.0017 & 56 + 89 & 55457.5396 & 0.0028 & 0.0005 & 89 + 104 & 55458.4094 & 0.0017 & @xmath120.0036 & 129 + 105 & 55458.4649 & 0.0047 & @xmath120.0063 & 129 + 106 & 55458.5300 & 0.0021 & 0.0004 & 129 + + + + we observed the 2010 superoutburst of this object discovered by @xcite . the outburst , second known in its history since , was detected on 2010 december 12 at a visual magnitude of 12.8 ( r. stubbings ) .
the first ccd observation started on december 15 . since some of the observations were recorded at high airmasses ( larger than 3 ) , secondary corrections for atmospheric extinction were applied to these observations .
the times of superhump maxima are listed in table [ tab : asas0918oc2010 ] .
the period was relatively constant except @xmath38 .
considering that the reported magnitude was fainter than that in 2005 and the observation started relatively late , it is likely that we mostly observed stage c superhumps .
the period was indeed similar to that of stage c superhumps recorded in 2005 .
the object underwent a rebrightening on december 30 ( vsnet - alert 12544 ) .
ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55545.9210 & 0.0009 & @xmath120.0044 & 63 + 20 & 55547.1785 & 0.0013 & @xmath120.0001 & 120 + 21 & 55547.2384 & 0.0007 & @xmath120.0028 & 285 + 22 & 55547.3054 & 0.0007 & 0.0015 & 194 + 32 & 55547.9344 & 0.0010 & 0.0040 & 51 + 35 & 55548.1197 & 0.0015 & 0.0013 & 77 + 52 & 55549.1855 & 0.0007 & 0.0019 & 112 + 53 & 55549.2465 & 0.0008 & 0.0003 & 219 + 54 & 55549.3125 & 0.0005 & 0.0036 & 196 + 85 & 55551.2502 & 0.0014 & @xmath120.0009 & 171 + 86 & 55551.3094 & 0.0015 & @xmath120.0044 & 165
+ + + + this object ( hereafter asas j1025 ) is a wz sge - type dwarf nova discovered in 2006 .
the history of observations can be found in @xcite and references therein .
the object was again detected in outburst in 2011 february by the crts ( = css110226:102522@xmath12154222 ) . due to the gap in observation before the crts detection ,
the outburst detection was apparently made in a later stage than in the 2006 superoutburst .
the stage of early superhumps was not recorded .
the times of superhump maxima are listed in table [ tab : asas1025oc2011 ] , which clearly demonstrates stage b superhumps with @xmath39 and short segment of stage c. this value is in good agreement with the 2006 measurement ( see also figure [ fig : j1025comp ] ) . in @xcite
, we suggested that the object would be a `` borderline '' long-@xmath2 wz sge - like dwarf nova based in the large @xmath13 and the large fractional superhump excess .
the short interval of superoutbursts ( 5 yr or shorter ) for a wz sge - type dwarf nova also strengthens this interpretation .
( 88mm,70mm)fig24.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55619.5018 & 0.0003 & 0.0005 & 95 + 3 & 55619.6924 & 0.0006 & 0.0007 & 52 + 4 & 55619.7561 & 0.0005 & 0.0009 & 52 + 5 & 55619.8196 & 0.0006 & 0.0009 & 52 + 6 & 55619.8819 & 0.0010 & @xmath120.0003 & 50 + 18 & 55620.6413 & 0.0010 & @xmath120.0027 & 43 + 19 & 55620.7063 & 0.0008 & @xmath120.0012 & 43 + 20 & 55620.7695 & 0.0005 & @xmath120.0015 & 44 + 34 & 55621.6595 & 0.0008 & @xmath120.0003 & 51 + 35 & 55621.7216 & 0.0007 & @xmath120.0017 & 51 + 36 & 55621.7853 & 0.0013 & @xmath120.0014 & 49 + 56 & 55623.0616 & 0.0010 & 0.0051 & 95 + 57 & 55623.1217 & 0.0008 & 0.0018 & 90 + 72 & 55624.0764 & 0.0005 & 0.0042 & 101 + 87 & 55625.0242 & 0.0004 & @xmath120.0003 & 106 + 88 & 55625.0856 & 0.0010 & @xmath120.0024 & 91 + 103 & 55626.0378 & 0.0054 & @xmath120.0024 & 113
+ + + + misv 1443 was discovered as an eruptive object by y. nakashima at an unfiltered ccd magnitude of 14.4 mag on 2011 january 8.59707 ut . the position of the object is , ( j2000.0 ) ( vsnet - alert 12580 ; @xcite ) .
there was a 20.4-mag quiescent counterpart ( vsnet - alert 12581 ) .
soon after this discovery announcement , time - resolved photometry indicated the presence of double - wave modulations .
although this signal was initially interpreted as early superhumps , this identification requires confirmation because ordinary superhumps quickly developed ( vsnet - alert 12597 ; figure [ fig : misv1443shpdm ] ) .
the dwarf nova nature of the object was spectroscopically confirmed by @xcite and a. arai ( vsnet - alert 12598 ) .
a retrospective analysis of survey materials revealed that the object was already in outburst at an unfiltered ccd magnitude of 12.8 on 2011 january 2.019 ut ( s. korotkiy and k. sokolovsky , vsnet - alert 12592 ) .
the times of superhump maxima are listed in table [ tab : misv1443oc2011 ] . disregarding stage a superhumps ( @xmath40 ) , evidently positive @xmath13 of @xmath41 was recorded .
based on the large outburst amplitude ( @xmath42 mag ) and a long ( @xmath438 d ) delay before the appearance of ordinary superhumps , the object is likely a wz sge - type dwarf nova .
the stage of early superhumps , however , was not sufficiently observed .
the initial detection of double - wave modulations may be interpreted as growing stage of ordinary superhumps rather than genuine early superhumps .
the maximum amplitude of superhumps ( 0.30 mag ) was , however , rather exceptionally large for a wz sge - type dwarf nova ( cf .
subsection [ sec : humpamp ] ) .
misv 1443 may bear intermediate characteristics between wz sge - type dwarf novae and more usual su uma - type dwarf novae .
the final stage of the superoutburst was observed by snapshot observations .
the object was recorded at 15.1 on january 27 and rapidly faded to 17.9 on january 28 .
the entire duration of the outburst was thus longer than 25 d. the object underwent a rebrightening at 15.3 on february 3 ( vsnet - alert 12787 , 12798 ; all observations are unfiltered ccd magnitudes ) .
the object stayed at mag @xmath918 , 2.5 mag above quiescence , at least until march 7 .
( 88mm,110mm)fig25.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55571.9590 & 0.0022 & @xmath120.0046 & 100 + 3 & 55572.1250 & 0.0005 & @xmath120.0090 & 146 + 4 & 55572.1842 & 0.0008 & @xmath120.0065 & 113 + 5 & 55572.2379 & 0.0010 & @xmath120.0096 & 100 + 19 & 55573.0464 & 0.0002 & 0.0038 & 100 + 20 & 55573.1038 & 0.0003 & 0.0045 & 124 + 21 & 55573.1609 & 0.0003 & 0.0048 & 117 + 22 & 55573.2166 & 0.0004 & 0.0037 & 87 + 23 & 55573.2735 & 0.0004 & 0.0038 & 130 + 26 & 55573.4443 & 0.0007 & 0.0043 & 26 + 27 & 55573.5006 & 0.0004 & 0.0038 & 47 + 28 & 55573.5568 & 0.0003 & 0.0032 & 54 + 29 & 55573.6135 & 0.0003 & 0.0031 & 51 + 35 & 55573.9529 & 0.0009 & 0.0017 & 101 + 37 & 55574.0662 & 0.0007 & 0.0015 & 52 + 38 & 55574.1221 & 0.0004 & 0.0006 & 89 + 39 & 55574.1793 & 0.0004 & 0.0010 & 88 + 40 & 55574.2366 & 0.0005 & 0.0015 & 63 + 41 & 55574.2916 & 0.0008 & @xmath120.0003 & 62 + 44 & 55574.4629 & 0.0004 & 0.0007 & 54 + 45 & 55574.5202 & 0.0004 & 0.0012 & 41 + 46 & 55574.5755 & 0.0007 & @xmath120.0003 & 52 + 53 & 55574.9743 & 0.0020 & 0.0011 & 61 + 55 & 55575.0868 & 0.0004 & @xmath120.0001 & 61 + 56 & 55575.1431 & 0.0004 & @xmath120.0006 & 109 + 57 & 55575.1991 & 0.0004 & @xmath120.0013 & 102 + 58 & 55575.2551 & 0.0006 & @xmath120.0021 & 59 + 62 & 55575.4849 & 0.0005 & 0.0005 & 49 + 63 & 55575.5408 & 0.0005 & @xmath120.0004 & 48 + 72 & 55576.0493 & 0.0034 & @xmath120.0030 & 78 + 73 & 55576.1125 & 0.0024 & 0.0035 & 105 + 74 & 55576.1659 & 0.0039 & 0.0001 & 91 + 75 & 55576.2154 & 0.0023 & @xmath120.0072 & 99 + 92 & 55577.1878 & 0.0008 & @xmath120.0002 & 63 + 93 & 55577.2434 & 0.0007 & @xmath120.0013 & 62 + 108 & 55578.0918 & 0.0025 & @xmath120.0048 & 95 + 109 & 55578.1533 & 0.0019 & @xmath120.0000 & 88 + 110 & 55578.2128 & 0.0017 & 0.0026 & 105 + + + + the final stage of the 2009 superoutburst of this object ( hereafter rx j1715 ) was reported in @xcite and @xcite .
i. miller detected a new outburst on 2010 september 15 ( baavss - alert 2386 ) .
subsequent observations confirmed superhumps ( vsnet - alert 12182 , 12183 ) .
the times of superhump maxima are listed in table [ tab : j1715oc2010 ] .
the outburst was apparently observed in its late stage , and these superhumps can be identified as stage c superhumps .
ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55455.4107 & 0.0010 & 0.0005 & 37 + 14 & 55456.3996 & 0.0012 & @xmath120.0011 & 77 + 15 & 55456.4722 & 0.0007 & 0.0008 & 72 + 16 & 55456.5439 & 0.0011 & 0.0018 & 18 + 23 & 55457.0343 & 0.0051 & @xmath120.0031 & 127 + 24 & 55457.1051 & 0.0020 & @xmath120.0030 & 112 + 29 & 55457.4662 & 0.0016 & 0.0042 & 72 + 30 & 55457.5325 & 0.0009 & @xmath120.0001 & 76 + + + + we observed the 2011 superoutburst of this object ( hereafter sdss j0732 ; the object was selected by @xcite , see a comment in @xcite ) .
the times of superhump maxima are listed in table [ tab : j0732oc2011 ] .
the observation most likely recorded stage c superhumps , whose period is in good agreement with the value in @xcite .
ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55582.2543 & 0.0021 & 0.0037 & 79 + 12 & 55583.1975 & 0.0034 & @xmath120.0050 & 159 + 24 & 55584.1582 & 0.0022 & 0.0037 & 132 + 30 & 55584.6289 & 0.0011 & @xmath120.0015 & 153 + 31 & 55584.7073 & 0.0010 & @xmath120.0024 & 153 + 68 & 55587.6474 & 0.0016 & 0.0026 & 152 + 69 & 55587.7276 & 0.0014 & 0.0035 & 151 + 70 & 55587.8042 & 0.0023 & 0.0008 & 152 + 71 & 55587.8815 & 0.0032 & @xmath120.0012 & 150 + 72 & 55587.9580 & 0.0034 & @xmath120.0041 & 122
+ + + + this object ( hereafter sdss j0803 ) was selected as a cv by @xcite .
the 2011 outburst was detected by the crts , which was the second brightest outbursts recorded by the crts .
subsequent observations confirmed the presence of superhumps ( vsnet - alert 13070 , 13073 ; figure [ fig : j0803shpdm ] ) .
the times of superhump maxima are listed in table [ tab : j0803oc2011 ] .
as inferred from the initial large amplitudes of superhumps , the outburst was apparently detected during its early stage ( cf .
vsnet - alert 13075 ) .
we therefore identified these superhumps as stage b superhumps .
the large @xmath13 of @xmath44 is consistent with this interpretation .
( 88mm,110mm)fig26.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55654.2962 & 0.0004 & 0.0017 & 58 + 1 & 55654.3719 & 0.0005 & 0.0023 & 75 + 11 & 55655.1193 & 0.0036 & @xmath120.0013 & 133 + 14 & 55655.3437 & 0.0005 & @xmath120.0023 & 100 + 15 & 55655.4210 & 0.0004 & @xmath120.0000 & 143 + 16 & 55655.4946 & 0.0008 & @xmath120.0015 & 71 + 23 & 55656.0223 & 0.0012 & 0.0005 & 154 + 37 & 55657.0713 & 0.0025 & @xmath120.0019 & 150 + 49 & 55657.9750 & 0.0079 & 0.0005 & 125 + 50 & 55658.0516 & 0.0040 & 0.0021 & 153 + + + + sdss [email protected] ( hereafter sdss j0804 ) is a cv selected during the course of the sloan digital sky survey ( sdss ) @xcite .
the first - ever superoutburst was recorded in 2006 @xcite .
the detection of the 2006 superoutburst was only made during the final stage of the plateau phase and the basic period of superhumps was not very well established ( @xcite ; @xcite ; see @xcite for a discussion of the corrected superhump period ) .
this wz sge - type dwarf nova is particularly notable in its grazing eclipses ( @xcite ; @xcite ) , in the existence of eleven rebrightenings during the 2006 superoutburst ( @xcite ; @xcite ) and in the zz cet - type pulsation of the white dwarf ( @xcite ; @xcite ) .
the object underwent an unexpected outburst in 2010 september , only 4.5 yr after the preceding superoutburst ( h. maehara , vsnet - alert 12184 ) .
this outburst was fortunately detected in its rising stage , and the presence of early superhumps and evolution of ordinary superhumps were well recorded ( vsnet - alert 12186 , 12192 , 12196 ; figure [ fig : j0804eshprof ] ) .
the analysis generally follows the method described in @xcite .
the mean period of early superhumps ( before bjd 24455462 ) was 0.058973(6 ) d , 0.05 % shorter than @xmath18 ( figure [ fig : j0804eshpdm ] ) .
the mean period of ordinary superhumps during the plateau phase ( bjd 245546255472.5 ) was 0.059584(2 ) d , 0.98 % longer than @xmath18 ( figure [ fig : j0804shpdm ] ) .
the times of superhump maxima , determined after excluding the phases of eclipses , during the plateau phase are listed in table [ tab : j0804oc2010 ] .
since the contribution of orbital signature was small , we did not subtract the mean orbital variation .
the stages a
c are clearly demonstrated .
the increase in the period during the stage b was not uniform , and the major increase took place during the final part of stage b. this behavior in the period evolution would explain the relatively long @xmath2 obtained during the 2006 superoutburst @xcite . in table [
tab : perlist ] we adopted @xmath13 = @xmath45 ( @xmath46 ) .
this value is unusually large for a wz sge - type dwarf nova with multiple rebrightenings . if we neglect the final part with the major increase in the period [ cf .
wz sge ( 2001 ) in @xcite ] , we obtain @xmath13 = @xmath47 ( @xmath48 ) .
we will return to this issue in subsection [ sec : wzsgestat ] .
in contrast to the expectation in @xcite , shallow eclipses were present during the entire course of the outburst .
the eclipses around the superoutburst maximum , however , were only observed depending on superhump phases ( figure [ fig : j0804shprof ] ) .
although stage b c was apparently recorded , we did not determine the period of stage c superhumps because the change in the @xmath5 diagram was too large to be interpreted as a typical stage b
c transition .
this may have been a result of contamination of increasing signal of orbital humps or a different component of superhumps as the system faded . during the six rebrightenings ,
persistent superhumps were observed ( figure [ fig : j0804rebpdm ] ) .
the mean period was 0.059656(4 ) d , confirming the slightly longer period observed during rebrightening phase of the 2006 superoutburst @xcite .
the rebrightenings following the current superoutburst was less numerous than in 2006 .
this may have resulted from the short interval ( for a wz sge - type dwarf nova ) since the preceding superoutburst .
the amplitudes of orbital humps and depths of eclipses increased by a factor of @xmath92 as the system faded from the peak of rebrightening ( figure [ fig : j0804rebphprof ] ) .
since the typical amplitudes of rebrightenings were @xmath92 mag ( @xmath96 times in real scale ) , this increase in the amplitudes can not be simply explained by a model assuming the orbital humps with a constant luminosity and varying contribution from the accretion disk ( see also the discussion in @xcite ) , and there was no particular indication of an enhanced hot spot before the start of rebrightenings ( cf .
the times of superhump maxima , determined after subtracting orbital variations and excluding eclipses , during the rebrightening phase are listed in table [ tab : j0804oc2010reb ] . even after rebrightenings , the superhump signal with a mean period of 0.05967(2 ) d persisted ( figure [ fig : j0804postpdm ] ) .
the times of maxima determined after subtracting the orbital signal are listed in table [ tab : j0804ocpostreb ] .
the @xmath5 diagram together with the light curve is shown in figure [ fig : j0804humpall ] .
the global evolution of the @xmath5 diagram is remarkably similar to those of gw lib and v455 and @xcite except the six rebrightenings after the major disturbance in the @xmath5 . the late stage ( during rebrightenings and after rebrightenings )
superhumps can be interpreted as an extension on the stage b superhumps .
the amplitudes of superhumps ( after subtracting for the orbital variation ) during the rebrightening systematically varied depending on the brightness of the system ( figure [ fig : j0804humpamp ] ) .
the amplitudes increased when the system became fainter .
the tendency appears to be well explained by assuming the constant luminosity of the superhump source and assuming variable contribution from the accretion disk .
the general tendency in amplitudes of superhumps is similar to what was observed in persistent negative superhumps in v503 cyg @xcite and mn dra @xcite , and superhumps during the rebrightening phase of eg cnc ( @xcite ; @xcite ) and wz sge ( cf .
@xcite ; reanalysis of the data in @xcite is shown in figure [ fig : wzsgerebhumpamp ] ) .
the tendency showed a deviation from this relation around and after the final rebrightening .
the decreased strength of superhumps , or tidal dissipation , may have prevented further rebrightenings to occur .
the phenomenon ( immediately following the rebrightening phase ) has been also confirmed in the 2001 superoutburst of wz sge ( figure [ fig : wzsgerebhumpamp ] ) .
( 88mm,110mm)fig27.eps ( 88mm,110mm)fig28.eps ( 88mm,110mm)fig29.eps ( 88mm,110mm)fig30.eps ( 88mm,110mm)fig31.eps ( 88mm,110mm)fig32.eps ( 88mm,110mm)fig33.eps ( 160mm,190mm)fig34.eps ( 88mm,110mm)fig35.eps ( 88mm,110mm)fig36.eps cccccc @xmath10 & max & error & @xmath5 & phase & @xmath11 + 0 & 55462.4512 & 0.0007 & @xmath120.0188 & 0.53 & 59 + 1 & 55462.5081 & 0.0006 & @xmath120.0216 & 0.49 & 190 + 2 & 55462.5710 & 0.0006 & @xmath120.0183 & 0.56 & 198 + 3 & 55462.6305 & 0.0006 & @xmath120.0184 & 0.57 & 129 + 15 & 55463.3632 & 0.0008 & @xmath120.0014 & 0.99 & 174 + 16 & 55463.4246 & 0.0001 & 0.0003 & 0.03 & 690 + 17 & 55463.4852 & 0.0002 & 0.0013 & 0.06 & 529 + 18 & 55463.5436 & 0.0002 & 0.0000 & 0.04 & 281 + 19 & 55463.6044 & 0.0003 & 0.0011 & 0.07 & 185 + 24 & 55463.9051 & 0.0007 & 0.0036 & 0.17 & 48 + 25 & 55463.9678 & 0.0008 & 0.0067 & 0.23 & 48 + 30 & 55464.2670 & 0.0003 & 0.0077 & 0.30 & 143 + 31 & 55464.3289 & 0.0002 & 0.0099 & 0.35 & 744 + 32 & 55464.3881 & 0.0001 & 0.0095 & 0.36 & 691 + 33 & 55464.4456 & 0.0004 & 0.0073 & 0.33 & 180 + 34 & 55464.5045 & 0.0002 & 0.0066 & 0.33 & 438 + 35 & 55464.5635 & 0.0001 & 0.0060 & 0.33 & 871 + 41 & 55464.9189 & 0.0009 & 0.0035 & 0.35 & 47 + 42 & 55464.9792 & 0.0008 & 0.0041 & 0.37 & 44 + 45 & 55465.1561 & 0.0008 & 0.0022 & 0.37 & 108 + 46 & 55465.2177 & 0.0002 & 0.0041 & 0.42 & 521 + 47 & 55465.2770 & 0.0002 & 0.0038 & 0.42 & 808 + 48 & 55465.3409 & 0.0004 & 0.0080 & 0.50 & 266 + 50 & 55465.4540 & 0.0007 & 0.0018 & 0.42 & 86 + 51 & 55465.5137 & 0.0005 & 0.0019 & 0.43 & 151 + 52 & 55465.5743 & 0.0003 & 0.0028 & 0.46 & 219 + 53 & 55465.6348 & 0.0004 & 0.0038 & 0.49 & 147 + 54 & 55465.6925 & 0.0004 & 0.0018 & 0.46 & 95 + 58 & 55465.9313 & 0.0006 & 0.0020 & 0.51 & 56 + 59 & 55465.9900 & 0.0008 & 0.0011 & 0.51 & 40 + 65 & 55466.3492 & 0.0005 & 0.0024 & 0.59 & 49 + 67 & 55466.4678 & 0.0005 & 0.0017 & 0.60 & 93 + 68 & 55466.5285 & 0.0004 & 0.0027 & 0.63 & 200 + 69 & 55466.5889 & 0.0004 & 0.0036 & 0.66 & 124 + 70 & 55466.6478 & 0.0003 & 0.0028 & 0.65 & 159 + 71 & 55466.7041 & 0.0004 & @xmath120.0006 & 0.61 & 119 + 75 & 55466.9469 & 0.0007 & 0.0037 & 0.72 & 47 + 81 & 55467.3045 & 0.0007 & 0.0035 & 0.78 & 86 + 82 & 55467.3619 & 0.0003 & 0.0012 & 0.76 & 275 + 83 & 55467.4224 & 0.0002 & 0.0021 & 0.78 & 668 + 84 & 55467.4811 & 0.0003 & 0.0011 & 0.78 & 181 + 85 & 55467.5404 & 0.0003 & 0.0008 & 0.78 & 161 + 86 & 55467.5991 & 0.0005 & @xmath120.0002 & 0.78 & 55 + 87 & 55467.6591 & 0.0003 & 0.0002 & 0.79 & 50 + 91 & 55467.8947 & 0.0005 & @xmath120.0029 & 0.78 & 64 + 92 & 55467.9557 & 0.0003 & @xmath120.0015 & 0.82 & 175 + 96 & 55468.1889 & 0.0054 & @xmath120.0068 & 0.77 & 26 + 97 & 55468.2526 & 0.0005 & @xmath120.0028 & 0.85 & 198 + 98 & 55468.3127 & 0.0004 & @xmath120.0023 & 0.87 & 681 + 99 & 55468.3709 & 0.0008 & @xmath120.0038 & 0.86 & 324 + 100 & 55468.4326 & 0.0003 & @xmath120.0017 & 0.90 & 306 + 101 & 55468.4921 & 0.0002 & @xmath120.0019 & 0.91 & 886 + 102 & 55468.5504 & 0.0003 & @xmath120.0031 & 0.90 & 289 + + + + + cccccc @xmath10 & max & error & @xmath5 & phase & @xmath11 + 103 & 55468.6102 & 0.0003 & @xmath120.0030 & 0.91 & 275 + 104 & 55468.6687 & 0.0004 & @xmath120.0041 & 0.90 & 90 + 108 & 55468.9032 & 0.0006 & @xmath120.0082 & 0.88 & 110 + 109 & 55468.9667 & 0.0005 & @xmath120.0044 & 0.95 & 174 + 120 & 55469.6250 & 0.0004 & @xmath120.0021 & 0.11 & 91 + 121 & 55469.6848 & 0.0004 & @xmath120.0020 & 0.12 & 92 + 124 & 55469.8586 & 0.0014 & @xmath120.0071 & 0.07 & 31 + 125 & 55469.9223 & 0.0005 & @xmath120.0030 & 0.15 & 169 + 126 & 55469.9819 & 0.0005 & @xmath120.0031 & 0.16 & 159 + 130 & 55470.2160 & 0.0021 & @xmath120.0076 & 0.13 & 62 + 134 & 55470.4579 & 0.0012 & @xmath120.0042 & 0.23 & 46 + 135 & 55470.5189 & 0.0013 & @xmath120.0029 & 0.26 & 31 + 137 & 55470.6370 & 0.0004 & @xmath120.0041 & 0.26 & 91 + 138 & 55470.6981 & 0.0003 & @xmath120.0027 & 0.30 & 89 + 141 & 55470.8774 & 0.0004 & @xmath120.0023 & 0.33 & 98 + 142 & 55470.9381 & 0.0002 & @xmath120.0011 & 0.36 & 218 + 143 & 55470.9972 & 0.0003 & @xmath120.0017 & 0.37 & 131 + 146 & 55471.1862 & 0.0013 & 0.0084 & 0.57 & 189 + 147 & 55471.2401 & 0.0009 & 0.0026 & 0.48 & 435 + 148 & 55471.2997 & 0.0004 & 0.0026 & 0.49 & 479 + 149 & 55471.3594 & 0.0012 & 0.0026 & 0.50 & 162 + 151 & 55471.4925 & 0.0028 & 0.0165 & 0.76 & 26 + 152 & 55471.5437 & 0.0017 & 0.0080 & 0.63 & 28 + 153 & 55471.6033 & 0.0012 & 0.0080 & 0.64 & 111 + 154 & 55471.6684 & 0.0013 & 0.0134 & 0.74 & 90 + 155 & 55471.7160 & 0.0009 & 0.0014 & 0.55 & 70 + 158 & 55471.9099 & 0.0006 & 0.0163 & 0.83 & 142 + 159 & 55471.9658 & 0.0009 & 0.0126 & 0.78 & 175 + 175 & 55472.9103 & 0.0041 & 0.0028 & 0.79 & 45 + 176 & 55472.9637 & 0.0042 & @xmath120.0035 & 0.69 & 45 + 186 & 55473.5415 & 0.0040 & @xmath120.0221 & 0.49 & 16 + 187 & 55473.5964 & 0.0024 & @xmath120.0268 & 0.42 & 16 + + + + + cccccc @xmath10 & max & error & @xmath5 & phase & @xmath11 + 0 & 55473.8915 & 0.0019 & @xmath120.0057 & 0.42 & 24 + 1 & 55473.9512 & 0.0015 & @xmath120.0056 & 0.43 & 23 + 17 & 55474.9120 & 0.0022 & 0.0007 & 0.71 & 12 + 18 & 55474.9665 & 0.0018 & @xmath120.0044 & 0.64 & 23 + 22 & 55475.2033 & 0.0013 & @xmath120.0063 & 0.65 & 79 + 23 & 55475.2595 & 0.0010 & @xmath120.0097 & 0.60 & 126 + 24 & 55475.3189 & 0.0011 & @xmath120.0100 & 0.61 & 133 + 25 & 55475.3749 & 0.0024 & @xmath120.0137 & 0.56 & 46 + 32 & 55475.7991 & 0.0023 & @xmath120.0070 & 0.75 & 32 + 34 & 55475.9125 & 0.0024 & @xmath120.0130 & 0.67 & 21 + 39 & 55476.2214 & 0.0009 & @xmath120.0023 & 0.90 & 283 + 40 & 55476.2826 & 0.0013 & @xmath120.0008 & 0.94 & 342 + 45 & 55476.5835 & 0.0010 & 0.0019 & 0.04 & 128 + 46 & 55476.6403 & 0.0012 & @xmath120.0010 & 0.00 & 140 + 50 & 55476.8817 & 0.0018 & 0.0017 & 0.09 & 75 + 51 & 55476.9390 & 0.0019 & @xmath120.0006 & 0.07 & 90 + 56 & 55477.2453 & 0.0007 & 0.0074 & 0.26 & 235 + 57 & 55477.3022 & 0.0008 & 0.0047 & 0.22 & 269 + 61 & 55477.5373 & 0.0013 & 0.0012 & 0.21 & 18 + 62 & 55477.5997 & 0.0014 & 0.0039 & 0.26 & 20 + 63 & 55477.6572 & 0.0012 & 0.0018 & 0.24 & 18 + 68 & 55477.9569 & 0.0005 & 0.0032 & 0.32 & 77 + 76 & 55478.4329 & 0.0010 & 0.0019 & 0.38 & 24 + 77 & 55478.4904 & 0.0011 & @xmath120.0002 & 0.36 & 22 + 78 & 55478.5484 & 0.0007 & @xmath120.0019 & 0.34 & 56 + 79 & 55478.6097 & 0.0014 & @xmath120.0002 & 0.38 & 60 + 80 & 55478.6633 & 0.0010 & @xmath120.0062 & 0.29 & 27 + 84 & 55478.9030 & 0.0008 & @xmath120.0051 & 0.35 & 81 + 85 & 55478.9623 & 0.0005 & @xmath120.0056 & 0.36 & 83 + 93 & 55479.4455 & 0.0022 & 0.0004 & 0.55 & 23 + 94 & 55479.4991 & 0.0032 & @xmath120.0056 & 0.45 & 23 + 95 & 55479.5684 & 0.0012 & 0.0040 & 0.63 & 22 + 96 & 55479.6263 & 0.0017 & 0.0022 & 0.61 & 23 + 97 & 55479.6823 & 0.0053 & @xmath120.0014 & 0.56 & 15 + 107 & 55480.2863 & 0.0008 & 0.0060 & 0.79 & 199 + 109 & 55480.4057 & 0.0014 & 0.0061 & 0.82 & 14 + 110 & 55480.4642 & 0.0005 & 0.0050 & 0.81 & 34 + 111 & 55480.5262 & 0.0005 & 0.0073 & 0.86 & 84 + 112 & 55480.5833 & 0.0005 & 0.0048 & 0.83 & 99 + 113 & 55480.6416 & 0.0005 & 0.0034 & 0.82 & 102 + 114 & 55480.7032 & 0.0006 & 0.0053 & 0.86 & 74 + 117 & 55480.8774 & 0.0007 & 0.0006 & 0.81 & 45 + 118 & 55480.9388 & 0.0007 & 0.0024 & 0.85 & 65 + 119 & 55480.9971 & 0.0010 & 0.0010 & 0.84 & 55 + 127 & 55481.4726 & 0.0009 & @xmath120.0007 & 0.90 & 40 + 128 & 55481.5345 & 0.0005 & 0.0015 & 0.95 & 126 + 129 & 55481.5898 & 0.0007 & @xmath120.0028 & 0.89 & 128 + 130 & 55481.6523 & 0.0008 & 0.0000 & 0.95 & 108 + 131 & 55481.7182 & 0.0013 & 0.0062 & 0.06 & 37 + 134 & 55481.8872 & 0.0015 & @xmath120.0037 & 0.93 & 66 + 135 & 55481.9500 & 0.0015 & @xmath120.0005 & 0.99 & 65 + + + + + cccccc @xmath10 & max & error & @xmath5 & phase & @xmath11 + 140 & 55482.2552 & 0.0032 & 0.0064 & 0.16 & 30 + 143 & 55482.4309 & 0.0047 & 0.0031 & 0.14 & 62 + 144 & 55482.4940 & 0.0019 & 0.0066 & 0.21 & 153 + 145 & 55482.5554 & 0.0017 & 0.0083 & 0.25 & 198 + 146 & 55482.6165 & 0.0020 & 0.0098 & 0.29 & 193 + 147 & 55482.6741 & 0.0018 & 0.0077 & 0.26 & 149 + 156 & 55483.2082 & 0.0039 & 0.0048 & 0.31 & 102 + 157 & 55483.2680 & 0.0005 & 0.0050 & 0.33 & 170 + 158 & 55483.3277 & 0.0006 & 0.0051 & 0.34 & 142 + 160 & 55483.4483 & 0.0006 & 0.0063 & 0.38 & 152 + 161 & 55483.5052 & 0.0007 & 0.0036 & 0.35 & 173 + 162 & 55483.5667 & 0.0006 & 0.0054 & 0.39 & 178 + 163 & 55483.6254 & 0.0007 & 0.0045 & 0.38 & 87 + 167 & 55483.8627 & 0.0016 & 0.0031 & 0.41 & 34 + 169 & 55483.9818 & 0.0031 & 0.0030 & 0.43 & 29 + 176 & 55484.3923 & 0.0046 & @xmath120.0041 & 0.38 & 11 + 177 & 55484.4564 & 0.0013 & 0.0003 & 0.47 & 68 + 178 & 55484.5111 & 0.0015 & @xmath120.0046 & 0.40 & 88 + 179 & 55484.5726 & 0.0015 & @xmath120.0028 & 0.44 & 197 + 180 & 55484.6318 & 0.0010 & @xmath120.0032 & 0.44 & 195 + 181 & 55484.6920 & 0.0015 & @xmath120.0027 & 0.46 & 89 + 184 & 55484.8704 & 0.0028 & @xmath120.0033 & 0.48 & 124 + 185 & 55484.9295 & 0.0012 & @xmath120.0038 & 0.49 & 135 + 186 & 55484.9872 & 0.0011 & @xmath120.0058 & 0.46 & 139 + 200 & 55485.8296 & 0.0118 & 0.0015 & 0.74 & 18 + 201 & 55485.8850 & 0.0022 & @xmath120.0028 & 0.68 & 38 + 207 & 55486.2469 & 0.0021 & 0.0012 & 0.81 & 37 + 210 & 55486.4271 & 0.0079 & 0.0024 & 0.87 & 17 + 211 & 55486.4890 & 0.0010 & 0.0047 & 0.92 & 60 + 212 & 55486.5459 & 0.0028 & 0.0019 & 0.88 & 96 + 219 & 55486.9628 & 0.0015 & 0.0012 & 0.95 & 13 + 229 & 55487.5664 & 0.0048 & 0.0083 & 0.18 & 64 + 230 & 55487.6181 & 0.0011 & 0.0003 & 0.05 & 65 + 231 & 55487.6822 & 0.0031 & 0.0047 & 0.14 & 63 + 246 & 55488.5761 & 0.0005 & 0.0039 & 0.29 & 107 + 247 & 55488.6362 & 0.0003 & 0.0043 & 0.31 & 111 + 248 & 55488.6963 & 0.0005 & 0.0048 & 0.33 & 77 + 263 & 55489.5897 & 0.0008 & 0.0033 & 0.47 & 64 + 264 & 55489.6408 & 0.0039 & @xmath120.0053 & 0.33 & 43 + 265 & 55489.6972 & 0.0023 & @xmath120.0085 & 0.29 & 43 + 268 & 55489.8810 & 0.0011 & @xmath120.0036 & 0.40 & 27 + 269 & 55489.9438 & 0.0011 & @xmath120.0005 & 0.47 & 23 + 270 & 55489.9981 & 0.0034 & @xmath120.0059 & 0.39 & 21 + 286 & 55490.9688 & 0.0018 & 0.0104 & 0.84 & 18 + 297 & 55491.6102 & 0.0010 & @xmath120.0045 & 0.71 & 56 + 311 & 55492.4432 & 0.0025 & @xmath120.0066 & 0.83 & 74 + 312 & 55492.5072 & 0.0014 & @xmath120.0023 & 0.91 & 76 + 313 & 55492.5598 & 0.0015 & @xmath120.0093 & 0.80 & 37 + 328 & 55493.4543 & 0.0012 & @xmath120.0096 & 0.96 & 18 + 330 & 55493.5742 & 0.0017 & @xmath120.0090 & 1.00 & 71 + 331 & 55493.6343 & 0.0013 & @xmath120.0085 & 0.01 & 51 + + + + + cccccc @xmath10 & max & error & @xmath5 & phase & @xmath11 + 0 & 55494.6511 & 0.0022 & @xmath120.0052 & 0.25 & 39 + 1 & 55494.7154 & 0.0015 & @xmath120.0005 & 0.34 & 73 + 21 & 55495.9004 & 0.0051 & @xmath120.0086 & 0.42 & 21 + 27 & 55496.2675 & 0.0023 & 0.0005 & 0.64 & 70 + 28 & 55496.3138 & 0.0055 & @xmath120.0128 & 0.42 & 94 + 47 & 55497.4502 & 0.0040 & @xmath120.0099 & 0.68 & 17 + 48 & 55497.5121 & 0.0020 & @xmath120.0076 & 0.73 & 17 + 49 & 55497.5787 & 0.0106 & @xmath120.0007 & 0.86 & 16 + 50 & 55497.6327 & 0.0027 & @xmath120.0063 & 0.78 & 17 + 53 & 55497.8062 & 0.0043 & @xmath120.0118 & 0.72 & 15 + 54 & 55497.8657 & 0.0031 & @xmath120.0120 & 0.73 & 23 + 55 & 55497.9314 & 0.0010 & @xmath120.0060 & 0.84 & 22 + 56 & 55497.9940 & 0.0017 & @xmath120.0030 & 0.90 & 21 + 63 & 55498.4061 & 0.0035 & @xmath120.0085 & 0.88 & 15 + 64 & 55498.4665 & 0.0020 & @xmath120.0077 & 0.91 & 16 + 100 & 55500.6201 & 0.0009 & @xmath120.0018 & 0.41 & 17 + 118 & 55501.6987 & 0.0022 & 0.0029 & 0.69 & 28 + 126 & 55502.1777 & 0.0039 & 0.0047 & 0.81 & 100 + 127 & 55502.2358 & 0.0014 & 0.0032 & 0.79 & 228 + 128 & 55502.2930 & 0.0014 & 0.0007 & 0.76 & 180 + 129 & 55502.3639 & 0.0061 & 0.0119 & 0.96 & 116 + 135 & 55502.7042 & 0.0028 & @xmath120.0057 & 0.73 & 31 + 145 & 55503.3014 & 0.0014 & @xmath120.0051 & 0.85 & 86 + 161 & 55504.2593 & 0.0023 & @xmath120.0017 & 0.08 & 138 + 162 & 55504.3199 & 0.0024 & @xmath120.0007 & 0.11 & 102 + 178 & 55505.2630 & 0.0020 & @xmath120.0122 & 0.09 & 85 + 179 & 55505.3237 & 0.0010 & @xmath120.0111 & 0.12 & 132 + 195 & 55506.2876 & 0.0054 & @xmath120.0017 & 0.46 & 80 + 196 & 55506.3420 & 0.0013 & @xmath120.0070 & 0.38 & 87 + 262 & 55510.2888 & 0.0015 & 0.0025 & 0.27 & 74 + 347 & 55515.3572 & 0.0014 & 0.0000 & 0.17 & 60 + 363 & 55516.3134 & 0.0022 & 0.0018 & 0.37 & 125 + 379 & 55517.2820 & 0.0018 & 0.0158 & 0.79 & 25 + 380 & 55517.3427 & 0.0015 & 0.0169 & 0.82 & 25 + 381 & 55517.4017 & 0.0033 & 0.0162 & 0.82 & 25 + 382 & 55517.4594 & 0.0013 & 0.0143 & 0.79 & 26 + 384 & 55517.5772 & 0.0016 & 0.0128 & 0.79 & 26 + 385 & 55517.6295 & 0.0036 & 0.0054 & 0.68 & 15 + 396 & 55518.2766 & 0.0032 & @xmath120.0038 & 0.64 & 57 + 398 & 55518.4051 & 0.0213 & 0.0055 & 0.82 & 25 + 399 & 55518.4603 & 0.0024 & 0.0010 & 0.76 & 25 + 400 & 55518.5232 & 0.0052 & 0.0042 & 0.82 & 24 + 401 & 55518.5782 & 0.0043 & @xmath120.0004 & 0.76 & 26 + 414 & 55519.3724 & 0.0015 & 0.0182 & 0.22 & 35 + 415 & 55519.4303 & 0.0015 & 0.0165 & 0.20 & 25 + 416 & 55519.4876 & 0.0008 & 0.0141 & 0.17 & 26 + 417 & 55519.5505 & 0.0007 & 0.0174 & 0.23 & 26 + 418 & 55519.6098 & 0.0008 & 0.0170 & 0.24 & 26 + 433 & 55520.4953 & 0.0008 & 0.0077 & 0.25 & 26 + + + + + cccccc @xmath10 & max & error & @xmath5 & phase & @xmath11 + 434 & 55520.5622 & 0.0017 & 0.0149 & 0.38 & 24 + 668 & 55534.4961 & 0.0006 & @xmath120.0109 & 0.53 & 100 + 670 & 55534.6191 & 0.0004 & @xmath120.0073 & 0.61 & 142 + 680 & 55535.2159 & 0.0029 & @xmath120.0070 & 0.73 & 69 + 702 & 55536.5394 & 0.0010 & 0.0040 & 0.16 & 15 + 703 & 55536.5991 & 0.0008 & 0.0041 & 0.17 & 21 + 752 & 55539.5183 & 0.0012 & 0.0001 & 0.64 & 19 + 753 & 55539.5779 & 0.0013 & 0.0001 & 0.65 & 20 + 754 & 55539.6347 & 0.0006 & @xmath120.0028 & 0.62 & 19 + 1002 & 55554.4324 & 0.0018 & @xmath120.0001 & 0.40 & 15 + 1003 & 55554.4934 & 0.0027 & 0.0013 & 0.44 & 20 + 1004 & 55554.5499 & 0.0020 & @xmath120.0018 & 0.39 & 20 + 1005 & 55554.6148 & 0.0025 & 0.0034 & 0.49 & 21 + 1021 & 55555.5644 & 0.0027 & @xmath120.0016 & 0.59 & 21 + 1022 & 55555.6199 & 0.0023 & @xmath120.0057 & 0.53 & 21 + 1036 & 55556.4460 & 0.0015 & @xmath120.0148 & 0.53 & 20 + 1038 & 55556.5611 & 0.0020 & @xmath120.0189 & 0.48 & 21 + 1039 & 55556.6232 & 0.0010 & @xmath120.0166 & 0.53 & 18 + + + + + this object ( hereafter sdss j0812 ) was selected as a cv by @xcite .
the outburst detection by k. itagaki in 2008 led to a classification as an su uma - type dwarf nova @xcite .
the object was again in outburst in 2011 march ( i. miller , baavss - alert 2561 , vsnet - alert 13019 ) .
subsequent observations detected superhumps ( vsnet - alert 13024 , 13025 ) . although @xcite reported a period of 0.08432(1 ) d , the new observations indicated that this period was a wrong alias ( vsnet - alert 13030 ; figure [ fig : j0812shpdm ] ) .
the times of superhump maxima for 2008 ( updated ) and 2011 superoutbursts are listed in tables [ tab : j0812oc2008 ] and [ tab : j0812oc2011 ] , respectively .
a comparison of @xmath5 diagrams ( figure [ fig : j0812comp ] ) suggests that there is a short ( @xmath9 40 cycles or slightly longer ) stage b with a relatively constant @xmath2 and subsequent stage c common to these two superoutbursts .
since these stages were identifiable in the 2011 superoutburst , we gave values for these stages in table [ tab : perlist ] .
there was an apparent gap in stage b in the 2008 superoutburst and we gave a global value .
the large negative @xmath13 likely referred to a stage b
c transition .
the global behavior of the @xmath5 diagram is not very different from that of ht cas ( subsection [ obj : htcas ] ) with a similar @xmath2 .
( 88mm,110mm)fig37.eps ( 88mm,70mm)fig38.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 54751.2796 & 0.0002 & @xmath120.0150 & 310 + 39 & 54754.3230 & 0.0006 & 0.0035 & 148 + 51 & 54755.2551 & 0.0004 & 0.0049 & 274 + 52 & 54755.3300 & 0.0007 & 0.0023 & 179 + 64 & 54756.2654 & 0.0014 & 0.0070 & 198 + 65 & 54756.3398 & 0.0009 & 0.0038 & 113 + 76 & 54757.2076 & 0.0047 & 0.0184 & 177 + 77 & 54757.2689 & 0.0008 & 0.0022 & 295 + 90 & 54758.2731 & 0.0019 & @xmath120.0019 & 84 + 103 & 54759.2580 & 0.0018 & @xmath120.0253 & 102 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55644.9450 & 0.0010 & @xmath120.0130 & 144 + 1 & 55645.0265 & 0.0011 & @xmath120.0091 & 212 + 2 & 55645.1049 & 0.0013 & @xmath120.0084 & 111 + 3 & 55645.1905 & 0.0022 & @xmath120.0005 & 45 + 6 & 55645.4195 & 0.0003 & @xmath120.0044 & 83 + 7 & 55645.4982 & 0.0003 & @xmath120.0033 & 73 + 13 & 55645.9660 & 0.0003 & @xmath120.0014 & 96 + 14 & 55646.0435 & 0.0003 & @xmath120.0015 & 102 + 18 & 55646.3544 & 0.0010 & @xmath120.0012 & 46 + 19 & 55646.4357 & 0.0005 & 0.0025 & 48 + 26 & 55646.9805 & 0.0002 & 0.0038 & 405 + 27 & 55647.0589 & 0.0003 & 0.0045 & 350 + 28 & 55647.1336 & 0.0008 & 0.0016 & 85 + 39 & 55647.9927 & 0.0004 & 0.0066 & 307 + 40 & 55648.0700 & 0.0006 & 0.0062 & 178 + 41 & 55648.1466 & 0.0013 & 0.0051 & 68 + 52 & 55649.0047 & 0.0004 & 0.0092 & 164 + 53 & 55649.0812 & 0.0005 & 0.0080 & 165 + 77 & 55650.9388 & 0.0037 & 0.0021 & 118 + 78 & 55651.0212 & 0.0006 & 0.0069 & 363 + 79 & 55651.1025 & 0.0014 & 0.0105 & 200 + 90 & 55651.9474 & 0.0014 & 0.0014 & 59 + 91 & 55652.0257 & 0.0017 & 0.0020 & 83 + 103 & 55652.9583 & 0.0008 & 0.0029 & 303 + 104 & 55653.0330 & 0.0015 & @xmath120.0001 & 264 + 129 & 55654.9630 & 0.0016 & @xmath120.0111 & 110 + 130 & 55655.0348 & 0.0025 & @xmath120.0170 & 112 + 131 & 55655.1275 & 0.0047 & @xmath120.0020 & 86
+ + + + this object ( hereafter sdss j0932 ) was selected as a short - period ( @xmath49 hr ) cv by @xcite .
@xcite clarified its eclipsing nature and suspected that it is an intermediate polar based on x - ray observations .
this object was for the first time detected in outburst on 2011 march 18 at an unfiltered ccd magnitude of 14.9 ( j. shears , baavss - alert 2537 ) .
subsequent observations immediately detected superhumps ( vsnet - alert 13006 , 13020 , 13034 , 13042 ; figure [ fig : j0932shpdm ] ) .
the observed eclipse minima during the current outburst confirmed the shorter alias for @xmath18 given by @xcite .
the times of recorded eclipses , determined with the kwee and van woerden ( kw ) method @xcite , after removing linearly approximated trends around eclipses in order to minimize the effect of superhumps , are summarized in table [ tab : j0932ecl ] .
the resultant refined ephemeris is as follows : @xmath50 the times of superhump maxima determined from observations outside the eclipses are listed in table [ tab : j0932oc2011 ] .
there was no significant global trend of period variation .
it is likely that we only observed stage c superhumps .
the values given in table [ tab : perlist ] follows this identification .
the fractional superhump excess was 2.7 % .
we also obtained post - superoutburst observations . due to the faintness of the object and low signal of superhumps
, we only obtained a mean period of 0.06814(4 ) d with the pdm method for the interval of bjd 2455650.32455656.2 .
this period is practically identical with the period in stage c and these superhumps are likely persistent superhumps in post - superoutburst stage of short-@xmath18 systems ( e.g. @xcite ) .
( 88mm,110mm)fig39.eps ccccc @xmath10 & minimum & error & @xmath5 & source + 0 & 53106.68364 & 0.00045 & 0.00023 & 1 + 1 & 53106.74963 & 0.00020 & -0.00008 & 1 + 2 & 53106.81609 & 0.00028 & 0.00008 & 1 + 467 & 53137.64733 & 0.00016 & 0.00017 & 1 + 468 & 53137.71340 & 0.00014 & -0.00007 & 1 + 469 & 53137.77931 & 0.00023 & -0.00046 & 1 + 470 & 53137.84620 & 0.00030 & 0.00013 & 1 + 38202 & 55639.61176 & 0.00005 & 0.00026 & 2 + 38203 & 55639.67801 & 0.00006 & 0.00020 & 2 + 38209 & 55640.07542 & 0.00007 & -0.00021 & 2 + 38210 & 55640.14193 & 0.00006 & 0.00000 & 2 + 38211 & 55640.20853 & 0.00007 & 0.00029 & 2 + 38214 & 55640.40617 & 0.00010 & -0.00097 & 2 + 38215 & 55640.47469 & 0.00011 & 0.00124 & 2 + 38216 & 55640.53892 & 0.00006 & -0.00083 & 2 + 38217 & 55640.60456 & 0.00006 & -0.00149 & 2 + 38218 & 55640.67125 & 0.00005 & -0.00111 & 2 + 38220 & 55640.80509 & 0.00017 & 0.00013 & 2 + 38227 & 55641.26942 & 0.00007 & 0.00033 & 2 + 38228 & 55641.33602 & 0.00010 & 0.00063 & 2 + 38229 & 55641.40250 & 0.00005 & 0.00080 & 2 + 38232 & 55641.60086 & 0.00004 & 0.00025 & 2 + 38244 & 55642.39646 & 0.00007 & 0.00022 & 2 + 38304 & 55646.37486 & 0.00005 & 0.00039 & 2 + 38305 & 55646.44106 & 0.00004 & 0.00030 & 2 + 38314 & 55647.03754 & 0.00009 & 0.00004 & 2 + 38315 & 55647.10351 & 0.00007 & -0.00029 & 2 + 38319 & 55647.36886 & 0.00004 & -0.00015 & 2 + 38320 & 55647.43547 & 0.00005 & 0.00015 & 2 + 38328 & 55647.96488 & 0.00006 & -0.00087 & 2 + 38329 & 55648.03180 & 0.00008 & -0.00025 & 2 + 38330 & 55648.09873 & 0.00007 & 0.00038 & 2 + 38364 & 55650.35295 & 0.00006 & 0.00028 & 2 + 38365 & 55650.41921 & 0.00004 & 0.00023 & 2 + 38366 & 55650.48497 & 0.00006 & -0.00031 & 2 + 38367 & 55650.55199 & 0.00008 & 0.00041 & 2 + 38410 & 55653.40198 & 0.00005 & -0.00066 & 2 + 38440 & 55655.39143 & 0.00010 & -0.00032 & 2 + 38442 & 55655.52489 & 0.00005 & 0.00054 & 2 + + + + cccccc @xmath10 & max & error & @xmath5 & phase & @xmath11 + 0 & 55639.5874 & 0.0013 & @xmath120.0089 & 0.64 & 42 + 1 & 55639.6605 & 0.0007 & @xmath120.0038 & 0.74 & 55 + 7 & 55640.0662 & 0.0016 & @xmath120.0068 & 0.86 & 56 + 8 & 55640.1376 & 0.0016 & @xmath120.0035 & 0.94 & 66 + 9 & 55640.2064 & 0.0008 & @xmath120.0028 & 0.97 & 43 + 11 & 55640.3556 & 0.0017 & 0.0101 & 0.22 & 41 + 12 & 55640.4179 & 0.0014 & 0.0044 & 0.16 & 56 + 13 & 55640.4874 & 0.0019 & 0.0058 & 0.21 & 57 + 14 & 55640.5540 & 0.0012 & 0.0042 & 0.22 & 57 + 16 & 55640.6848 & 0.0034 & @xmath120.0012 & 0.19 & 49 + 15 & 55640.6167 & 0.0018 & @xmath120.0012 & 0.16 & 64 + 17 & 55640.7487 & 0.0027 & @xmath120.0054 & 0.15 & 52 + 18 & 55640.8208 & 0.0044 & @xmath120.0014 & 0.24 & 50 + 19 & 55640.8884 & 0.0036 & @xmath120.0019 & 0.26 & 28 + 20 & 55640.9532 & 0.0068 & @xmath120.0053 & 0.24 & 29 + 25 & 55641.2991 & 0.0007 & 0.0001 & 0.45 & 72 + 26 & 55641.3651 & 0.0008 & @xmath120.0020 & 0.45 & 139 + 30 & 55641.6353 & 0.0015 & @xmath120.0043 & 0.52 & 71 + 41 & 55642.3853 & 0.0019 & @xmath120.0034 & 0.83 & 56 + 56 & 55643.4156 & 0.0043 & 0.0052 & 0.37 & 33 + 70 & 55644.3637 & 0.0007 & @xmath120.0002 & 0.67 & 29 + 71 & 55644.4333 & 0.0008 & 0.0013 & 0.72 & 30 + 72 & 55644.5006 & 0.0020 & 0.0005 & 0.74 & 35 + 81 & 55645.1174 & 0.0032 & 0.0043 & 0.04 & 22 + 85 & 55645.3914 & 0.0019 & 0.0058 & 0.17 & 31 + 86 & 55645.4577 & 0.0016 & 0.0040 & 0.17 & 35 + 87 & 55645.5217 & 0.0024 & @xmath120.0001 & 0.14 & 35 + 88 & 55645.5918 & 0.0017 & 0.0019 & 0.20 & 35 + 99 & 55646.3441 & 0.0012 & 0.0050 & 0.54 & 90 + 100 & 55646.4122 & 0.0008 & 0.0050 & 0.57 & 110 + 101 & 55646.4783 & 0.0006 & 0.0030 & 0.57 & 61 + 102 & 55646.5480 & 0.0011 & 0.0046 & 0.62 & 37 + 103 & 55646.6142 & 0.0008 & 0.0027 & 0.62 & 34 + 109 & 55647.0182 & 0.0022 & @xmath120.0020 & 0.71 & 23 + 110 & 55647.0894 & 0.0027 & 0.0011 & 0.78 & 37 + 111 & 55647.1597 & 0.0027 & 0.0033 & 0.84 & 43 + 114 & 55647.3625 & 0.0009 & 0.0018 & 0.90 & 61 + 115 & 55647.4297 & 0.0009 & 0.0008 & 0.92 & 72 + 123 & 55647.9826 & 0.0018 & 0.0088 & 0.25 & 47 + 124 & 55648.0505 & 0.0019 & 0.0087 & 0.28 & 53 + 158 & 55650.3542 & 0.0024 & @xmath120.0034 & 0.02 & 36 + 159 & 55650.4175 & 0.0016 & @xmath120.0082 & 0.98 & 36 + 160 & 55650.4818 & 0.0015 & @xmath120.0120 & 0.95 & 37 + 161 & 55650.5473 & 0.0022 & @xmath120.0146 & 0.94 & 36 + + + + + this object ( hereafter sdss j1120 ) was initially selected as a dwarf nova by @xcite .
the 2011 outburst was detected on february 14.838 ut at an unfiltered ccd magnitude of 15.7 by j. shears ( baavss - alert 2513 ) .
superhump were detected by s. brady ( baavss - alert 2514 ) .
we analyzed the available data and obtained the times of superhump maxima ( table [ tab : j1120oc2011 ] ) .
it was not possible to identify the stage of superhump evolution .
the result of pdm analysis and mean profile are shown in figure [ fig : j1120shpdm ] .
( 88mm,110mm)fig40.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55607.4060 & 0.0008 & @xmath120.0016 & 95 + 1 & 55607.4797 & 0.0011 & 0.0016 & 94 + 7 & 55607.9019 & 0.0004 & 0.0005 & 73 + 8 & 55607.9716 & 0.0004 & @xmath120.0004 & 73 + 21 & 55608.8893 & 0.0005 & 0.0001 & 72 + 22 & 55608.9594 & 0.0005 & @xmath120.0002 & 73
+ + + + this object ( hereafter sdss j1146 ) is a cv selected during the course of the sdss @xcite .
@xcite reported an orbital period of 1.6 hr based on radial - velocity study .
although the spectrum suggested a dwarf nova in quiescence , no major outburst had been recorded .
a small outburst was recorded on 2009 april 17 at an unfiltered ccd magnitude of 16.9 ( j. shears , cvnet - outburst 3062 ) .
j. shears also detected an outburst on 2011 january 3 during its rising phase ( baavss - alert 2459 ) .
subsequent observations recorded growing superhumps ( vsnet - alert 12567 , 12576 , 12579 ; figure [ fig : j1146shpdm ] ) .
the times of superhump maxima are listed in table [ tab : j1146oc2011 ] .
the values given in table [ tab : perlist ] are somewhat uncertain because the middle of stage b was not very well observed and only the initial part of stage c was observed .
( 88mm,110mm)fig41.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55565.6927 & 0.0043 & @xmath120.0242 & 30 + 1 & 55565.7744 & 0.0016 & @xmath120.0060 & 29 + 2 & 55565.8387 & 0.0015 & @xmath120.0051 & 32 + 3 & 55565.9048 & 0.0016 & @xmath120.0024 & 19 + 9 & 55566.2962 & 0.0018 & 0.0085 & 111 + 10 & 55566.3593 & 0.0015 & 0.0081 & 126 + 14 & 55566.6131 & 0.0004 & 0.0082 & 65 + 15 & 55566.6766 & 0.0005 & 0.0083 & 51 + 16 & 55566.7379 & 0.0004 & 0.0063 & 32 + 42 & 55568.3795 & 0.0006 & @xmath120.0011 & 123 + 43 & 55568.4464 & 0.0005 & 0.0024 & 105 + 72 & 55570.2859 & 0.0010 & 0.0026 & 249 + 73 & 55570.3475 & 0.0009 & 0.0009 & 180 + 75 & 55570.4714 & 0.0012 & @xmath120.0021 & 42 + 76 & 55570.5387 & 0.0010 & 0.0018 & 66 + 77 & 55570.6011 & 0.0007 & 0.0008 & 68 + 78 & 55570.6651 & 0.0010 & 0.0014 & 68 + 79 & 55570.7292 & 0.0008 & 0.0021 & 46 + 80 & 55570.7937 & 0.0016 & 0.0032 & 23 + 81 & 55570.8526 & 0.0008 & @xmath120.0014 & 46 + 82 & 55570.9195 & 0.0007 & 0.0021 & 49 + 83 & 55570.9829 & 0.0013 & 0.0021 & 39 + 88 & 55571.2960 & 0.0018 & @xmath120.0020 & 38 + 89 & 55571.3605 & 0.0007 & @xmath120.0009 & 100 + 90 & 55571.4255 & 0.0009 & 0.0007 & 96 + 91 & 55571.4889 & 0.0008 & 0.0007 & 94 + 92 & 55571.5482 & 0.0019 & @xmath120.0035 & 96 + 93 & 55571.6120 & 0.0009 & @xmath120.0031 & 99 + 94 & 55571.6763 & 0.0012 & @xmath120.0022 & 72 + 105 & 55572.3698 & 0.0009 & @xmath120.0063 & 33 + + + + this object ( hereafter sdss j1227 ) is an eclipsing su uma - type dwarf nova whose superoutburst was first recorded in 2007 .
this 2007 superoutburst was analyzed by @xcite and @xcite .
another superoutburst was observed in 2011 . by using eclipses
observed during the 2011 observations ( table [ tab : j1227ecl ] ) , it has become evident that the alias selection of the orbital ephemeris in @xcite was incorrect .
we have revised the eclipse ephemeris as follows : cccc @xmath10 & minimum & error & @xmath5 + 29202 & 55634.51978 & 0.00002 & 0.00014 + 29203 & 55634.58272 & 0.00002 & 0.00012 + 29215 & 55635.33819 & 0.00004 & 0.00019 + 29216 & 55635.40145 & 0.00014 & 0.00050 + 29217 & 55635.46423 & 0.00002 & 0.00033 + 29218 & 55635.52706 & 0.00003 & 0.00021 + 29230 & 55636.28281 & 0.00009 & 0.00056 + 29231 & 55636.34561 & 0.00007 & 0.00040 + 29232 & 55636.40864 & 0.00005 & 0.00048 + 29246 & 55637.28967 & 0.00005 & 0.00021 + 29247 & 55637.35270 & 0.00003 & 0.00028 + 29248 & 55637.41573 & 0.00007 & 0.00036 + 29284 & 55639.68236 & 0.00008 & 0.00078 + 29300 & 55640.68954 & 0.00003 & 0.00075 + 29391 & 55646.41865 & 0.00008 & 0.00138 + + + @xmath51 we have updated the results of the 2007 superoutburst because the significant addition to the data published in @xcite .
the times of superhump maxima are listed in table [ tab : j1227oc2007 ] .
the data now clearly show all stages a c and a positive @xmath13 for stage b. the updated parameters are listed in table [ tab : perlist ] .
the 2011 outburst was detected by crts (= css110312:122741@xmath0513925 ) .
the outburst was apparently observed during its middle - to - late stage .
the times of superhump maxima outside the eclipses are listed in table [ tab : j1227oc2011 ] .
a stage b
c transition was recorded .
a combined @xmath5 diagram is shown in figure [ fig : j1227comp ] .
( 88mm,70mm)fig42.eps cccccc @xmath10 & max & error & @xmath5 & phase & @xmath11 + 0 & 54256.4379 & 0.0004 & @xmath120.0286 & 0.45 & + 1 & 54256.5069 & 0.0006 & @xmath120.0242 & 0.55 & + 33 & 54258.6069 & 0.0003 & 0.0073 & 0.91 & 122 + 34 & 54258.6736 & 0.0002 & 0.0094 & 0.97 & 120 + 35 & 54258.7360 & 0.0005 & 0.0071 & 0.96 & 118 + 40 & 54259.0627 & 0.0014 & 0.0107 & 0.15 & 107 + 61 & 54260.4136 & 0.0035 & 0.0041 & 0.61 & 106 + 63 & 54260.5428 & 0.0003 & 0.0040 & 0.66 & 192 + 75 & 54261.3100 & 0.0019 & @xmath120.0045 & 0.85 & 9 + 76 & 54261.3722 & 0.0019 & @xmath120.0069 & 0.84 & 18 + 77 & 54261.4411 & 0.0008 & @xmath120.0026 & 0.93 & 206 + 78 & 54261.5108 & 0.0006 & 0.0024 & 0.04 & 196 + 80 & 54261.6413 & 0.0006 & 0.0036 & 0.11 & 31 + 81 & 54261.7049 & 0.0010 & 0.0026 & 0.13 & 35 + 82 & 54261.7706 & 0.0011 & 0.0036 & 0.17 & 35 + 86 & 54262.0321 & 0.0009 & 0.0066 & 0.32 & 49 + 91 & 54262.3468 & 0.0005 & @xmath120.0020 & 0.32 & 15 + 101 & 54262.9967 & 0.0009 & 0.0015 & 0.64 & 82 + 102 & 54263.0615 & 0.0012 & 0.0017 & 0.67 & 114 + 106 & 54263.3175 & 0.0014 & @xmath120.0008 & 0.74 & 15 + 107 & 54263.3773 & 0.0019 & @xmath120.0057 & 0.69 & 28 + 111 & 54263.6418 & 0.0017 & 0.0003 & 0.89 & 35 + 112 & 54263.7040 & 0.0015 & @xmath120.0022 & 0.88 & 35 + 113 & 54263.7683 & 0.0025 & @xmath120.0026 & 0.90 & 35 + 114 & 54263.8312 & 0.0034 & @xmath120.0043 & 0.90 & 35 + 122 & 54264.3573 & 0.0014 & 0.0048 & 0.26 & 28 + 123 & 54264.4203 & 0.0008 & 0.0031 & 0.26 & 157 + 124 & 54264.4892 & 0.0005 & 0.0073 & 0.35 & 140 + 126 & 54264.6180 & 0.0010 & 0.0068 & 0.40 & 41 + 127 & 54264.6821 & 0.0013 & 0.0063 & 0.42 & 39 + 128 & 54264.7453 & 0.0010 & 0.0048 & 0.42 & 40 + 129 & 54264.8122 & 0.0032 & 0.0072 & 0.49 & 39 + 137 & 54265.3267 & 0.0015 & 0.0045 & 0.66 & 28 + 138 & 54265.3906 & 0.0014 & 0.0038 & 0.67 & 29 + 153 & 54266.3706 & 0.0026 & 0.0141 & 0.24 & 29 + 168 & 54267.3243 & 0.0014 & @xmath120.0017 & 0.39 & 24 + 169 & 54267.3903 & 0.0012 & @xmath120.0004 & 0.44 & 23 + 184 & 54268.3527 & 0.0025 & @xmath120.0076 & 0.73 & 29 + 199 & 54269.3141 & 0.0069 & @xmath120.0158 & 0.00 & 18 + 216 & 54270.4171 & 0.0015 & @xmath120.0117 & 0.52 & 17 + 262 & 54273.3961 & 0.0018 & @xmath120.0062 & 0.85 & 25 + + + + + cccccc @xmath10 & max & error & @xmath5 & phase & @xmath11 + 0 & 55634.5319 & 0.0003 & @xmath120.0056 & 0.20 & 88 + 1 & 55634.5973 & 0.0005 & @xmath120.0048 & 0.24 & 63 + 13 & 55635.3749 & 0.0006 & @xmath120.0017 & 0.59 & 56 + 14 & 55635.4394 & 0.0004 & @xmath120.0018 & 0.61 & 190 + 15 & 55635.5029 & 0.0004 & @xmath120.0028 & 0.62 & 121 + 27 & 55636.2833 & 0.0018 & 0.0030 & 0.02 & 41 + 28 & 55636.3485 & 0.0008 & 0.0037 & 0.06 & 50 + 29 & 55636.4141 & 0.0014 & 0.0047 & 0.10 & 30 + 43 & 55637.3181 & 0.0007 & 0.0051 & 0.46 & 63 + 44 & 55637.3826 & 0.0007 & 0.0050 & 0.48 & 65 + 80 & 55639.6999 & 0.0013 & @xmath120.0014 & 0.29 & 53 + 96 & 55640.7334 & 0.0051 & @xmath120.0006 & 0.71 & 38 + 184 & 55646.4115 & 0.0010 & @xmath120.0027 & 0.91 & 59 + + + + + we have observed the 2011 superoutburst of this eclipsing su uma - type dwarf nova ( hereafter sdss j1250 ) . in analyzing the data , we used the refined ephemeris of eclipses ( equation [ equ : j1250ecl ] ) .
@xmath52 the times of superhump maxima are listed in table [ tab : j1250oc2011 ] . although we likely observed stages b and c , we only measured the mean period due to the gap in the observation .
cccccc @xmath10 & max & error & @xmath5 & phase & @xmath11 + 0 & 55665.0937 & 0.0005 & 0.0003 & 0.47 & 124 + 40 & 55667.5096 & 0.0044 & 0.0058 & 0.60 & 62 + 41 & 55667.5603 & 0.0005 & @xmath120.0038 & 0.47 & 133 + 42 & 55667.6206 & 0.0006 & @xmath120.0038 & 0.49 & 134 + 52 & 55668.2272 & 0.0013 & 0.0003 & 0.82 & 106 + 53 & 55668.2868 & 0.0015 & @xmath120.0005 & 0.84 & 104 + 55 & 55668.4099 & 0.0012 & 0.0022 & 0.93 & 48 + 56 & 55668.4674 & 0.0014 & @xmath120.0006 & 0.91 & 55 + 58 & 55668.5902 & 0.0014 & 0.0017 & 0.00 & 112 + 59 & 55668.6474 & 0.0023 & @xmath120.0014 & 0.98 & 71 + + + + + sdss [email protected] ( hereafter sdss j1339 ) is a cv selected during the course of the sdss @xcite .
@xcite reported a possible orbital period of 1.7 hr and the presence of features of a white dwarf in the spectrum , suggesting a wz sge - type dwarf nova .
@xcite obtained an orbital period of 0.05731(2 ) d and showed the presence of zz cet - type pulsation of the white dwarf .
@xcite also found a long - period photometric variation , similar to the ones observed in gw lib @xcite and fs aur @xcite , whose origin is yet unknown .
the 2011 outburst , the first known outburst in its history , was detected by j. shears on february 7.919 at a visual magnitude of 10.4 ( baavss - alert 2501 ) .
p. schmeer also made an independent detection of the outburst ( vsnet - alert 12806 ) .
early observations detected low - amplitude ( @xmath90.0150.025 mag ) variations which can be attributed to early superhumps ( vsnet - alert 12817 , 12819 , 12820 ) . early spectroscopy by a. arai detected h@xmath53 in emission and other balmer lines in absorption .
although heii emission was absent , ciii / niii emission lines were possibly present ( vsnet - alert 12822 ) .
the spectroscopic feature was less striking compared to the 2001 superoutburst of wz sge @xcite .
ordinary superhumps emerged four days after the outburst detection ( vsnet - alert 12835 , 12836 , 12837 ) .
after 13 d , the object started to fade rapidly .
in contrast to many wz sge - type dwarf novae , the object did not show post - superoutburst rebrightenings .
the times of ordinary superhumps are listed in table [ tab : j1339oc2011 ] .
the stages a and b are clearly recognized ( cf .
figure [ fig : j1339humpall ] ) . during the rapid fading phase , the phase of superhumps were apparently reversed and the reversed phase continued during the entire post - superoutburst state . during the post - superoutburst phase , a beat phenomenon between persisting superhumps and the orbital signal was strongly present ( vsnet - alert 12943 ; figure [ fig : j1339latebeat ] ) .
we have obtained the orbital period of 0.057289(1 ) d from these observations ( cf .
figure [ fig : j1339lateshpdm ] ) .
this period is in good agreement with the period determined from radial - velocity study @xcite .
the times of superhumps during the post - superoutburst stage were determined after subtracting the averaged orbital variations ( table [ tab : j1339oc2011late ] ) .
the tables includes the initial eight epochs which were observed as the secondary maxima during the final stage of the plateau ( cf .
subsection [ obj : kepv344lyr ] ) .
these maxima were measured using phases @xmath120.2 to 0.2 of the humps in order to reduce the contamination from the primary maxima .
the mean periods of late - stage superhumps were 0.058175(3 ) d ( @xmath54 ) and 0.058183(5 ) d ( @xmath55 ) .
since the amplitudes of early superhumps were very low , and since some of observations were made under non - ideal conditions ( due to the sparse field , the comparison star was much fainter than the variable in some observations ) , we have analyzed a selected high - quality subset selected from the data before bjd 2455604.2 .
a pdm analysis of the data yielded a signal of 0.05689(2 ) d in addition to the likely signal of emerging ordinary superhumps ( signals around 0.0590.062 d , upper panel of figure [ fig : j1339eshpdm ] ) . since this signal is very weak and the baseline of the observation was not sufficiently long , we used the orbital period as an approximation of the period of early superhumps and extracted the averaged profile .
the resultant profile had double - wave modulations characteristic to early superhumps @xcite , although the full amplitude was only 0.014 mag ( lower panel of figure [ fig : j1339eshpdm ] ) .
the relatively large @xmath13 of @xmath56 during stage b in combination with the lack of post - superoutburst rebrightenings is compatible with the type - d classification of wz sge - type outbursts ( @xcite , subsection 5.3 ) .
the delay times of appearance of ordinary superhumps in these type - d outbursts , however , tend to be around @xmath910 d. this delay , corresponding to the phase with early superhumps , was observed only for 4 d in sdss j1339 .
although this short duration more resembled those of type - b outbursts ( outbursts with multiple rebrightenings ) , it may have been a result of the lack of observations around the true maximum .
indeed , s. ueda reported retrospective positive detections 1.7 d before miller s detection at unfiltered ccd magnitudes of 9.9 ( vsnet - alert 12859 ) .
the last negative observations by s. ueda were reported on january 29 , 9 d before detection of the outburst .
this possible failure to record the earliest stage of the outburst might explain the lack of highly excited emission lines in the spectrum and low amplitudes of early superhumps .
if the true maximum was 6 d earlier ( assuming a typical delay time of 10 d ) , the object should have reached a magnitude of 9.5 or brighter , giving an outburst amplitude of @xmath98.0 mag .
the object is thus one of the brightest members of wz sge - type dwarf novae and deserves further detailed investigation .
( 88mm,110mm)fig43.eps ( 88mm,110mm)fig44.eps ( 88mm,110mm)fig45.eps ( 88mm,110mm)fig46.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55604.4595 & 0.0005 & @xmath120.0113 & 152 + 1 & 55604.5148 & 0.0009 & @xmath120.0141 & 83 + 2 & 55604.5782 & 0.0006 & @xmath120.0088 & 66 + 6 & 55604.8123 & 0.0008 & @xmath120.0071 & 72 + 7 & 55604.8699 & 0.0006 & @xmath120.0077 & 71 + 8 & 55604.9274 & 0.0008 & @xmath120.0083 & 71 + 11 & 55605.1082 & 0.0004 & @xmath120.0018 & 64 + 12 & 55605.1640 & 0.0005 & @xmath120.0042 & 70 + 13 & 55605.2208 & 0.0003 & @xmath120.0054 & 70 + 14 & 55605.2810 & 0.0003 & @xmath120.0033 & 139 + 15 & 55605.3399 & 0.0002 & @xmath120.0026 & 285 + 16 & 55605.3983 & 0.0004 & @xmath120.0023 & 82 + 17 & 55605.4572 & 0.0002 & @xmath120.0015 & 89 + 18 & 55605.5189 & 0.0003 & 0.0021 & 89 + 19 & 55605.5758 & 0.0002 & 0.0009 & 87 + 20 & 55605.6343 & 0.0003 & 0.0013 & 86 + 21 & 55605.6943 & 0.0002 & 0.0032 & 74 + 23 & 55605.8084 & 0.0027 & 0.0011 & 47 + 24 & 55605.8662 & 0.0033 & 0.0007 & 13 + 27 & 55606.0457 & 0.0006 & 0.0059 & 67 + 29 & 55606.1636 & 0.0002 & 0.0075 & 168 + 30 & 55606.2223 & 0.0003 & 0.0081 & 135 + 32 & 55606.3388 & 0.0005 & 0.0085 & 185 + 33 & 55606.3962 & 0.0001 & 0.0077 & 109 + 34 & 55606.4562 & 0.0002 & 0.0096 & 177 + 35 & 55606.5136 & 0.0002 & 0.0089 & 176 + 36 & 55606.5699 & 0.0002 & 0.0071 & 102 + 37 & 55606.6286 & 0.0001 & 0.0076 & 90 + 38 & 55606.6861 & 0.0002 & 0.0071 & 148 + 39 & 55606.7443 & 0.0003 & 0.0071 & 112 + 40 & 55606.8016 & 0.0002 & 0.0064 & 124 + 41 & 55606.8593 & 0.0005 & 0.0059 & 72 + 42 & 55606.9187 & 0.0005 & 0.0072 & 96 + 43 & 55606.9767 & 0.0003 & 0.0071 & 122 + 44 & 55607.0335 & 0.0001 & 0.0058 & 117 + 51 & 55607.4358 & 0.0003 & 0.0013 & 98 + 52 & 55607.4964 & 0.0001 & 0.0038 & 244 + 57 & 55607.7861 & 0.0003 & 0.0029 & 47 + 58 & 55607.8439 & 0.0002 & 0.0026 & 120 + 59 & 55607.9020 & 0.0001 & 0.0026 & 253 + 60 & 55607.9604 & 0.0001 & 0.0029 & 256 + 62 & 55608.0780 & 0.0003 & 0.0042 & 45 + 64 & 55608.1917 & 0.0003 & 0.0017 & 180 + 65 & 55608.2503 & 0.0003 & 0.0022 & 180 + 66 & 55608.3074 & 0.0003 & 0.0012 & 180 + 67 & 55608.3656 & 0.0004 & 0.0013 & 116 + 68 & 55608.4228 & 0.0002 & 0.0004 & 74 + 69 & 55608.4825 & 0.0002 & 0.0020 & 75 + 70 & 55608.5387 & 0.0003 & 0.0001 & 70 + 71 & 55608.5972 & 0.0002 & 0.0004 & 64 + 72 & 55608.6555 & 0.0003 & 0.0006 & 72 + 73 & 55608.7137 & 0.0013 & 0.0007 & 29 + 80 & 55609.1199 & 0.0002 & 0.0001 & 175 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 84 & 55609.3521 & 0.0003 & @xmath120.0001 & 361 + 85 & 55609.4104 & 0.0003 & 0.0001 & 536 + 86 & 55609.4660 & 0.0002 & @xmath120.0024 & 480 + 87 & 55609.5248 & 0.0003 & @xmath120.0017 & 538 + 88 & 55609.5829 & 0.0003 & @xmath120.0017 & 170 + 89 & 55609.6405 & 0.0003 & @xmath120.0022 & 67 + 90 & 55609.6992 & 0.0004 & @xmath120.0017 & 45 + 102 & 55610.3944 & 0.0003 & @xmath120.0038 & 87 + 103 & 55610.4537 & 0.0003 & @xmath120.0026 & 90 + 104 & 55610.5118 & 0.0002 & @xmath120.0026 & 88 + 105 & 55610.5698 & 0.0003 & @xmath120.0027 & 82 + 106 & 55610.6268 & 0.0003 & @xmath120.0038 & 88 + 107 & 55610.6855 & 0.0002 & @xmath120.0032 & 79 + 115 & 55611.1494 & 0.0002 & @xmath120.0042 & 241 + 116 & 55611.2073 & 0.0004 & @xmath120.0045 & 163 + 117 & 55611.2658 & 0.0003 & @xmath120.0041 & 198 + 118 & 55611.3245 & 0.0005 & @xmath120.0035 & 259 + 119 & 55611.3820 & 0.0005 & @xmath120.0041 & 68 + 120 & 55611.4395 & 0.0003 & @xmath120.0048 & 53 + 121 & 55611.4975 & 0.0004 & @xmath120.0048 & 52 + 122 & 55611.5555 & 0.0005 & @xmath120.0049 & 52 + 123 & 55611.6153 & 0.0003 & @xmath120.0033 & 58 + 124 & 55611.6700 & 0.0008 & @xmath120.0067 & 52 + 125 & 55611.7302 & 0.0003 & @xmath120.0046 & 164 + 126 & 55611.7894 & 0.0003 & @xmath120.0035 & 184 + 127 & 55611.8481 & 0.0003 & @xmath120.0029 & 235 + 128 & 55611.9059 & 0.0003 & @xmath120.0032 & 204 + 129 & 55611.9621 & 0.0005 & @xmath120.0051 & 84 + 135 & 55612.3110 & 0.0005 & @xmath120.0049 & 76 + 136 & 55612.3708 & 0.0007 & @xmath120.0032 & 188 + 137 & 55612.4279 & 0.0004 & @xmath120.0043 & 62 + 138 & 55612.4875 & 0.0012 & @xmath120.0027 & 61 + 139 & 55612.5438 & 0.0003 & @xmath120.0046 & 61 + 140 & 55612.6029 & 0.0004 & @xmath120.0036 & 56 + 144 & 55612.8365 & 0.0006 & @xmath120.0024 & 30 + 145 & 55612.8952 & 0.0007 & @xmath120.0018 & 31 + 146 & 55612.9513 & 0.0006 & @xmath120.0038 & 31 + 149 & 55613.1268 & 0.0005 & @xmath120.0026 & 64 + 150 & 55613.1868 & 0.0006 & @xmath120.0008 & 157 + 151 & 55613.2425 & 0.0003 & @xmath120.0032 & 241 + 152 & 55613.3004 & 0.0003 & @xmath120.0033 & 240 + 153 & 55613.3584 & 0.0009 & @xmath120.0035 & 107 + 154 & 55613.4180 & 0.0004 & @xmath120.0020 & 63 + 155 & 55613.4766 & 0.0004 & @xmath120.0015 & 63 + 156 & 55613.5340 & 0.0004 & @xmath120.0023 & 64 + 157 & 55613.5936 & 0.0009 & @xmath120.0007 & 58 + 158 & 55613.6515 & 0.0005 & @xmath120.0009 & 63 + 160 & 55613.7696 & 0.0036 & 0.0009 & 27 + 161 & 55613.8261 & 0.0011 & @xmath120.0007 & 30 + 162 & 55613.8831 & 0.0008 & @xmath120.0018 & 31 + 163 & 55613.9464 & 0.0012 & 0.0034 & 30 + 165 & 55614.0574 & 0.0014 & @xmath120.0019 & 42 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 166 & 55614.1193 & 0.0005 & 0.0020 & 63 + 167 & 55614.1762 & 0.0005 & 0.0007 & 61 + 168 & 55614.2339 & 0.0004 & 0.0003 & 231 + 169 & 55614.2897 & 0.0004 & @xmath120.0020 & 239 + 170 & 55614.3529 & 0.0004 & 0.0031 & 153 + 171 & 55614.4081 & 0.0004 & 0.0002 & 140 + 172 & 55614.4680 & 0.0005 & 0.0020 & 172 + 173 & 55614.5272 & 0.0006 & 0.0031 & 132 + 174 & 55614.5842 & 0.0004 & 0.0019 & 115 + 175 & 55614.6433 & 0.0004 & 0.0029 & 116 + 176 & 55614.6996 & 0.0003 & 0.0011 & 87 + 178 & 55614.8163 & 0.0014 & 0.0016 & 17 + 179 & 55614.8735 & 0.0007 & 0.0007 & 29 + 180 & 55614.9341 & 0.0010 & 0.0032 & 16 + 182 & 55615.0439 & 0.0024 & @xmath120.0032 & 47 + 183 & 55615.1075 & 0.0007 & 0.0023 & 31 + 185 & 55615.2240 & 0.0004 & 0.0025 & 223 + 186 & 55615.2827 & 0.0003 & 0.0031 & 206 + 187 & 55615.3404 & 0.0007 & 0.0027 & 174 + 188 & 55615.3977 & 0.0007 & 0.0019 & 64 + 189 & 55615.4575 & 0.0005 & 0.0036 & 147 + 190 & 55615.5150 & 0.0005 & 0.0030 & 144 + 191 & 55615.5725 & 0.0006 & 0.0024 & 115 + 192 & 55615.6313 & 0.0023 & 0.0030 & 102 + 194 & 55615.7479 & 0.0007 & 0.0034 & 48 + 203 & 55616.2745 & 0.0008 & 0.0070 & 63 + 204 & 55616.3326 & 0.0005 & 0.0070 & 83 + 205 & 55616.3898 & 0.0009 & 0.0061 & 56 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55615.2489 & 0.0014 & @xmath120.0055 & 120 + 2 & 55615.3635 & 0.0016 & @xmath120.0072 & 94 + 3 & 55615.4198 & 0.0005 & @xmath120.0091 & 28 + 4 & 55615.4781 & 0.0005 & @xmath120.0090 & 87 + 5 & 55615.5377 & 0.0016 & @xmath120.0075 & 56 + 18 & 55616.2883 & 0.0005 & @xmath120.0134 & 30 + 19 & 55616.3470 & 0.0006 & @xmath120.0129 & 28 + 20 & 55616.4024 & 0.0007 & @xmath120.0157 & 29 + 37 & 55617.4063 & 0.0046 & @xmath120.0009 & 93 + 38 & 55617.4770 & 0.0008 & 0.0116 & 117 + 39 & 55617.5281 & 0.0006 & 0.0045 & 118 + 40 & 55617.5896 & 0.0010 & 0.0078 & 114 + 41 & 55617.6436 & 0.0009 & 0.0036 & 62 + 48 & 55618.0483 & 0.0020 & 0.0010 & 39 + 49 & 55618.1078 & 0.0011 & 0.0024 & 62 + 50 & 55618.1638 & 0.0010 & 0.0002 & 63 + 51 & 55618.2227 & 0.0005 & 0.0009 & 175 + 52 & 55618.2795 & 0.0009 & @xmath120.0005 & 168 + 53 & 55618.3370 & 0.0009 & @xmath120.0012 & 64 + 56 & 55618.5140 & 0.0004 & 0.0012 & 44 + 57 & 55618.5728 & 0.0005 & 0.0019 & 46 + 58 & 55618.6319 & 0.0006 & 0.0027 & 42 + 59 & 55618.6878 & 0.0007 & 0.0005 & 29 + 71 & 55619.3902 & 0.0005 & 0.0046 & 134 + 72 & 55619.4477 & 0.0003 & 0.0040 & 237 + 73 & 55619.5003 & 0.0005 & @xmath120.0016 & 223 + 74 & 55619.5631 & 0.0002 & 0.0031 & 258 + 75 & 55619.6226 & 0.0003 & 0.0043 & 242 + 77 & 55619.7363 & 0.0004 & 0.0016 & 28 + 87 & 55620.3194 & 0.0007 & 0.0029 & 62 + 88 & 55620.3757 & 0.0009 & 0.0010 & 62 + 89 & 55620.4355 & 0.0012 & 0.0026 & 80 + 90 & 55620.4841 & 0.0016 & @xmath120.0070 & 55 + 91 & 55620.5490 & 0.0008 & @xmath120.0003 & 53 + 92 & 55620.6038 & 0.0021 & @xmath120.0036 & 62 + 93 & 55620.6716 & 0.0027 & 0.0060 & 57 + 116 & 55622.0047 & 0.0011 & 0.0007 & 36 + 121 & 55622.2944 & 0.0007 & @xmath120.0005 & 64 + 122 & 55622.3527 & 0.0007 & @xmath120.0004 & 40 + 124 & 55622.4716 & 0.0009 & 0.0022 & 36 + 126 & 55622.5936 & 0.0012 & 0.0079 & 66 + 127 & 55622.6420 & 0.0007 & @xmath120.0020 & 61 + 137 & 55623.2307 & 0.0010 & 0.0049 & 42 + 138 & 55623.2886 & 0.0004 & 0.0046 & 64 + 139 & 55623.3536 & 0.0004 & 0.0115 & 41 + 140 & 55623.4075 & 0.0005 & 0.0071 & 65 + 141 & 55623.4617 & 0.0005 & 0.0032 & 87 + 142 & 55623.5213 & 0.0005 & 0.0045 & 68 + 152 & 55624.0958 & 0.0032 & @xmath120.0029 & 31 + 153 & 55624.1598 & 0.0007 & 0.0030 & 55 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 156 & 55624.3306 & 0.0029 & @xmath120.0008 & 41 + 157 & 55624.3881 & 0.0005 & @xmath120.0015 & 147 + 158 & 55624.4474 & 0.0005 & @xmath120.0003 & 163 + 159 & 55624.5073 & 0.0003 & 0.0014 & 284 + 160 & 55624.5626 & 0.0003 & @xmath120.0015 & 274 + 161 & 55624.6236 & 0.0003 & 0.0013 & 271 + 162 & 55624.6805 & 0.0008 & 0.0001 & 134 + 170 & 55625.1496 & 0.0007 & 0.0036 & 62 + 171 & 55625.2048 & 0.0012 & 0.0007 & 62 + 172 & 55625.2661 & 0.0008 & 0.0037 & 63 + 173 & 55625.3206 & 0.0016 & 0.0001 & 100 + 174 & 55625.3826 & 0.0012 & 0.0039 & 121 + 175 & 55625.4409 & 0.0011 & 0.0040 & 103 + 176 & 55625.5009 & 0.0010 & 0.0058 & 101 + 177 & 55625.5546 & 0.0011 & 0.0013 & 107 + 178 & 55625.6166 & 0.0009 & 0.0051 & 67 + 185 & 55626.0200 & 0.0016 & 0.0012 & 35 + 188 & 55626.1980 & 0.0006 & 0.0047 & 119 + 189 & 55626.2537 & 0.0009 & 0.0022 & 115 + 192 & 55626.4291 & 0.0011 & 0.0030 & 38 + 193 & 55626.4889 & 0.0011 & 0.0046 & 37 + 207 & 55627.2998 & 0.0010 & 0.0009 & 49 + 208 & 55627.3596 & 0.0005 & 0.0026 & 111 + 209 & 55627.4176 & 0.0004 & 0.0024 & 128 + 210 & 55627.4737 & 0.0004 & 0.0003 & 138 + 211 & 55627.5305 & 0.0005 & @xmath120.0011 & 118 + 212 & 55627.5914 & 0.0005 & 0.0016 & 126 + 213 & 55627.6468 & 0.0011 & @xmath120.0012 & 38 + 220 & 55628.0527 & 0.0017 & @xmath120.0026 & 47 + 221 & 55628.1124 & 0.0010 & @xmath120.0010 & 64 + 222 & 55628.1765 & 0.0012 & 0.0049 & 62 + 223 & 55628.2285 & 0.0008 & @xmath120.0013 & 62 + 224 & 55628.2888 & 0.0015 & 0.0008 & 72 + 225 & 55628.3429 & 0.0011 & @xmath120.0033 & 70 + 226 & 55628.4081 & 0.0005 & 0.0037 & 276 + 227 & 55628.4591 & 0.0005 & @xmath120.0034 & 213 + 228 & 55628.5204 & 0.0004 & @xmath120.0003 & 210 + 229 & 55628.5784 & 0.0005 & @xmath120.0006 & 264 + 230 & 55628.6394 & 0.0005 & 0.0023 & 240 + 231 & 55628.6957 & 0.0009 & 0.0004 & 93 + 240 & 55629.2226 & 0.0010 & 0.0036 & 122 + 241 & 55629.2737 & 0.0007 & @xmath120.0035 & 111 + 244 & 55629.4531 & 0.0027 & 0.0014 & 35 + 245 & 55629.5093 & 0.0010 & @xmath120.0006 & 36 + 246 & 55629.5722 & 0.0020 & 0.0041 & 39 + 247 & 55629.6275 & 0.0021 & 0.0012 & 37 + 255 & 55630.0942 & 0.0010 & 0.0024 & 65 + 256 & 55630.1531 & 0.0011 & 0.0032 & 62 + 257 & 55630.2110 & 0.0012 & 0.0029 & 55 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 258 & 55630.2657 & 0.0009 & @xmath120.0006 & 45 + 259 & 55630.3256 & 0.0008 & 0.0011 & 70 + 271 & 55631.0210 & 0.0014 & @xmath120.0017 & 60 + 272 & 55631.0806 & 0.0007 & @xmath120.0003 & 103 + 273 & 55631.1383 & 0.0005 & @xmath120.0008 & 181 + 274 & 55631.1976 & 0.0007 & 0.0003 & 46 + 278 & 55631.4334 & 0.0006 & 0.0033 & 42 + 279 & 55631.4884 & 0.0005 & 0.0002 & 59 + 280 & 55631.5466 & 0.0005 & 0.0001 & 62 + 284 & 55631.7769 & 0.0007 & @xmath120.0023 & 40 + 285 & 55631.8336 & 0.0008 & @xmath120.0038 & 40 + 286 & 55631.8935 & 0.0010 & @xmath120.0020 & 40 + 287 & 55631.9461 & 0.0027 & @xmath120.0076 & 51 + 288 & 55632.0103 & 0.0014 & @xmath120.0016 & 40 + 289 & 55632.0660 & 0.0013 & @xmath120.0041 & 46 + 290 & 55632.1262 & 0.0007 & @xmath120.0020 & 64 + 291 & 55632.1851 & 0.0022 & @xmath120.0013 & 65 + 292 & 55632.2425 & 0.0010 & @xmath120.0022 & 55 + 293 & 55632.3014 & 0.0008 & @xmath120.0015 & 64 + 296 & 55632.4694 & 0.0034 & @xmath120.0080 & 24 + 300 & 55632.7070 & 0.0023 & @xmath120.0032 & 30 + 301 & 55632.7704 & 0.0007 & 0.0021 & 48 + 302 & 55632.8252 & 0.0010 & @xmath120.0013 & 47 + 303 & 55632.8841 & 0.0011 & @xmath120.0006 & 48 + 304 & 55632.9441 & 0.0060 & 0.0012 & 59 + 305 & 55632.9949 & 0.0009 & @xmath120.0062 & 48 + 307 & 55633.1131 & 0.0013 & @xmath120.0043 & 122 + 308 & 55633.1743 & 0.0008 & @xmath120.0014 & 122 + 309 & 55633.2292 & 0.0011 & @xmath120.0046 & 109 + 312 & 55633.4077 & 0.0014 & @xmath120.0007 & 42 + 318 & 55633.7545 & 0.0007 & @xmath120.0030 & 59 + 319 & 55633.8168 & 0.0007 & 0.0011 & 25 + 321 & 55633.9358 & 0.0029 & 0.0037 & 30 + 322 & 55633.9867 & 0.0020 & @xmath120.0035 & 48 + 324 & 55634.1086 & 0.0009 & 0.0020 & 49 + 325 & 55634.1612 & 0.0009 & @xmath120.0036 & 64 + 326 & 55634.2204 & 0.0006 & @xmath120.0026 & 64 + 327 & 55634.2813 & 0.0009 & 0.0001 & 51 + 328 & 55634.3389 & 0.0006 & @xmath120.0005 & 79 + 329 & 55634.4005 & 0.0007 & 0.0030 & 71 + 332 & 55634.5727 & 0.0007 & 0.0006 & 62 + 344 & 55635.2716 & 0.0008 & 0.0013 & 49 + 345 & 55635.3271 & 0.0009 & @xmath120.0014 & 67 + 352 & 55635.7326 & 0.0011 & @xmath120.0032 & 48 + 354 & 55635.8471 & 0.0012 & @xmath120.0051 & 24 + 356 & 55635.9645 & 0.0006 & @xmath120.0040 & 45 + 361 & 55636.2628 & 0.0011 & 0.0033 & 98 + 362 & 55636.3178 & 0.0004 & 0.0001 & 155 + 363 & 55636.3762 & 0.0004 & 0.0003 & 99 + 364 & 55636.4325 & 0.0025 & @xmath120.0016 & 28 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 365 & 55636.4895 & 0.0012 & @xmath120.0028 & 40 + 366 & 55636.5468 & 0.0009 & @xmath120.0036 & 38 + 367 & 55636.6053 & 0.0020 & @xmath120.0033 & 38 + 382 & 55637.4759 & 0.0014 & @xmath120.0055 & 38 + 383 & 55637.5382 & 0.0011 & @xmath120.0014 & 38 + 384 & 55637.5931 & 0.0011 & @xmath120.0047 & 40 + 393 & 55638.1195 & 0.0007 & @xmath120.0019 & 64 + 394 & 55638.1789 & 0.0013 & @xmath120.0008 & 61 + 395 & 55638.2401 & 0.0013 & 0.0023 & 48 + 416 & 55639.4520 & 0.0021 & @xmath120.0078 & 55 + 417 & 55639.5127 & 0.0014 & @xmath120.0052 & 56 + 418 & 55639.5744 & 0.0012 & @xmath120.0017 & 45 + 424 & 55639.9204 & 0.0020 & @xmath120.0048 & 29 + 428 & 55640.1475 & 0.0051 & @xmath120.0104 & 65 + 495 & 55644.0591 & 0.0018 & 0.0027 & 64 + 496 & 55644.1106 & 0.0025 & @xmath120.0040 & 174 + 497 & 55644.1796 & 0.0009 & 0.0068 & 168 + 498 & 55644.2314 & 0.0014 & 0.0004 & 208 + 499 & 55644.2889 & 0.0058 & @xmath120.0003 & 87 + 502 & 55644.4616 & 0.0019 & @xmath120.0021 & 61 + 503 & 55644.5245 & 0.0018 & 0.0026 & 64 + 504 & 55644.5751 & 0.0010 & @xmath120.0050 & 63 + 505 & 55644.6358 & 0.0015 & @xmath120.0025 & 59 + 513 & 55645.1073 & 0.0014 & 0.0035 & 65 + 514 & 55645.1713 & 0.0023 & 0.0094 & 62 + 515 & 55645.2223 & 0.0013 & 0.0021 & 173 + 516 & 55645.2734 & 0.0013 & @xmath120.0049 & 179 + 517 & 55645.3237 & 0.0015 & @xmath120.0129 & 189 + 534 & 55646.3257 & 0.0035 & 0.0000 & 40 + 540 & 55646.6777 & 0.0024 & 0.0029 & 23 + 546 & 55647.0160 & 0.0023 & @xmath120.0079 & 64 + 547 & 55647.0867 & 0.0033 & 0.0045 & 104 + 548 & 55647.1377 & 0.0022 & @xmath120.0026 & 183 + 549 & 55647.2072 & 0.0017 & 0.0088 & 230 + 550 & 55647.2567 & 0.0020 & 0.0001 & 181 + 551 & 55647.3236 & 0.0042 & 0.0087 & 179 + 567 & 55648.2442 & 0.0009 & @xmath120.0016 & 120 + 568 & 55648.2992 & 0.0008 & @xmath120.0049 & 116 + 584 & 55649.2344 & 0.0009 & @xmath120.0006 & 120 + 585 & 55649.2944 & 0.0025 & 0.0013 & 114 + 587 & 55649.4091 & 0.0005 & @xmath120.0005 & 101 + 588 & 55649.4695 & 0.0006 & 0.0017 & 131 + 599 & 55650.1147 & 0.0035 & 0.0070 & 111 + 600 & 55650.1555 & 0.0035 & @xmath120.0104 & 118 + 601 & 55650.2258 & 0.0025 & 0.0016 & 159 + 602 & 55650.2799 & 0.0034 & @xmath120.0025 & 115 + 606 & 55650.5100 & 0.0010 & @xmath120.0051 & 63 + 607 & 55650.5644 & 0.0011 & @xmath120.0089 & 66 + 619 & 55651.2770 & 0.0024 & 0.0055 & 113 + 620 & 55651.3286 & 0.0020 & @xmath120.0011 & 96 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 622 & 55651.4525 & 0.0015 & 0.0065 & 61 + 623 & 55651.5095 & 0.0007 & 0.0053 & 65 + 624 & 55651.5646 & 0.0009 & 0.0022 & 62 + 625 & 55651.6254 & 0.0010 & 0.0048 & 40 + 634 & 55652.1414 & 0.0011 & @xmath120.0029 & 120 + 636 & 55652.2611 & 0.0012 & 0.0004 & 122 + 637 & 55652.3183 & 0.0025 & @xmath120.0006 & 109 + 640 & 55652.4943 & 0.0010 & 0.0009 & 31 + 641 & 55652.5509 & 0.0008 & @xmath120.0006 & 33 + 650 & 55653.0841 & 0.0037 & 0.0089 & 79 + 653 & 55653.2451 & 0.0076 & @xmath120.0047 & 115 + 658 & 55653.5394 & 0.0008 & @xmath120.0013 & 28 + 659 & 55653.5966 & 0.0025 & @xmath120.0023 & 28 + 675 & 55654.5327 & 0.0012 & 0.0027 & 87 + 676 & 55654.5980 & 0.0012 & 0.0099 & 52 + 677 & 55654.6536 & 0.0012 & 0.0073 & 29 + 686 & 55655.1733 & 0.0009 & 0.0033 & 53 + 687 & 55655.2290 & 0.0009 & 0.0008 & 241 + 688 & 55655.2891 & 0.0011 & 0.0028 & 300 + 702 & 55656.1012 & 0.0019 & 0.0002 & 123 + 703 & 55656.1623 & 0.0008 & 0.0032 & 186 + 704 & 55656.2207 & 0.0022 & 0.0034 & 142 + 705 & 55656.2695 & 0.0010 & @xmath120.0060 & 182 + 720 & 55657.1572 & 0.0034 & 0.0089 & 120 + 721 & 55657.2173 & 0.0085 & 0.0109 & 131 + 722 & 55657.2533 & 0.0039 & @xmath120.0114 & 120 + 726 & 55657.4933 & 0.0019 & @xmath120.0041 & 28 + 727 & 55657.5696 & 0.0034 & 0.0140 & 20 + 776 & 55660.4034 & 0.0030 & @xmath120.0033 & 55 + 777 & 55660.4613 & 0.0009 & @xmath120.0036 & 64 + 778 & 55660.5321 & 0.0025 & 0.0090 & 64 + 779 & 55660.5802 & 0.0018 & @xmath120.0011 & 64 + 780 & 55660.6391 & 0.0020 & @xmath120.0003 & 65 + 810 & 55662.3797 & 0.0017 & @xmath120.0053 & 32 + 811 & 55662.4383 & 0.0008 & @xmath120.0049 & 19 + 812 & 55662.5043 & 0.0024 & 0.0029 & 29 + 813 & 55662.5637 & 0.0009 & 0.0041 & 33 + 825 & 55663.2569 & 0.0029 & @xmath120.0009 & 115 + 826 & 55663.3159 & 0.0085 & @xmath120.0001 & 88 + 860 & 55665.3007 & 0.0024 & 0.0064 & 94 + 863 & 55665.4785 & 0.0023 & 0.0096 & 33 + 911 & 55668.2702 & 0.0041 & 0.0083 & 98 + 934 & 55669.5992 & 0.0019 & @xmath120.0009 & 27 + 934 & 55669.5983 & 0.0017 & @xmath120.0018 & 25 + 963 & 55671.2864 & 0.0009 & @xmath120.0011 & 22 + 965 & 55671.4028 & 0.0016 & @xmath120.0011 & 30 + 981 & 55672.3217 & 0.0007 & @xmath120.0132 & 31 + 982 & 55672.3719 & 0.0007 & @xmath120.0212 & 33 + 983 & 55672.4392 & 0.0016 & @xmath120.0121 & 31 + 985 & 55672.5833 & 0.0051 & 0.0157 & 23 + + + + this transient was detected by k. itagaki on 2010 november 21.36782 ut at an unfiltered ccd magnitude of 12.0 ( vsnet - alert 12414 ) .
a. drake informed that this object was spectroscopically recorded as an evident cv in sdss and there was a possible faint outburst in 2005 ( vsnet - alert 12413 ) .
based on this information , we refer to this object by the sdss designation ( hereafter sdss j1605 ) .
although the outburst of the amplitude was sufficiently suggestive of an outburst of a wz sge - type dwarf nova , the short visibility in the evening and morning sky made observations very difficult .
the rapid fading from the main superoutburst took place on december 24 ( vsnet - alert 12525 ) , 33 d after the initial detection .
the object still remained bright on 2011 january 21 ( 61 d after the initial detection ) . since the later stage of the outburst was poorly observed and the state of the object ( either continuation of the plateau phase or some kind of rebrightening phenomenon ) was unknown , we restrict our analysis to the earlier part of the observation . since some of the observations were recorded at high airmasses ( larger than 3 ) , secondary corrections for atmospheric extinction were applied to these observations .
the early data showed some hint of double - wave modulations , which are possibly early superhumps ( figure [ fig : j1605eshpdm ] ) . due to the short visibility ,
it was impossible to uniquely determine the period , we selected the most likely period based on our knowledge in periods of known wz sge - type dwarf novae with long plateau phases ( @xcite , subsection 5.3 ) .
this period should be tested by future observations . during the later stage of the outburst
, there was a signal around 0.05718 d ( 0.9 % longer than the period of possible early superhumps ) attributable to ordinary superhumps , but we did not adopt this value due to very poor statistics .
( 88mm,110mm)fig47.eps we reported on the 2007 superoutburst of this object ( hereafter sdss j2100 ) in @xcite .
we further obtained observations on three consecutive nights during the 2010 superoutburst ( table [ tab : j2100oc2010 ] ) .
the @xmath5 values suggest that this object is either a long-@xmath2 system with a large negative @xmath13 or we incidentally spotted an epoch of stage transition .
the global period with the pdm method was 0.08743(2 ) d , close to the value of 0.08746(8 ) d in @xcite .
ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55530.2355 & 0.0021 & @xmath120.0030 & 41 + 1 & 55530.3224 & 0.0003 & @xmath120.0034 & 82 + 35 & 55533.3065 & 0.0009 & 0.0127 & 81 + 69 & 55536.2554 & 0.0012 & @xmath120.0064 & 77 + + + + this object (= css101007:000938@xmath12121017 , hereafter ot j0009 ) was discovered by the crts on 2010 october 10 .
the object has a blue counterpart on the digitized sky survey ( dss ) . following the initial detection , an inspection of the backlog by the crts team clarified another outburst in 2005 ( cf .
vsnet - alert 12235 ) .
subsequent observations confirmed superhumps ( vsnet - alert 12234 , figure [ fig : j0009shpdm ] ) , identifying the object a dwarf nova near the period gap ( vsnet - alert 12238 , 12241 , 12244 ) .
there was a distinct shortening of the superhump period after @xmath57 , which we attribute to a stage b - c transition ( table [ tab : j0009oc2010 ] ) .
although we list our tentative identifications of periods in table [ tab : perlist ] , this variation in the period may be interpreted as a continuous change with a global @xmath13 of @xmath58 ( see also figure [ fig : j0009prof ] ) .
although the object resembles ef peg in its long @xmath2 , large outburst amplitude and rare outbursts , the large variation in the superhump period resembles those of more frequently outbursting su uma - type dwarf novae ( cf .
@xcite , subsection 4.10 ) .
as proposed for v342 cam ( = 1rxs j042332.8 + 745300 ) @xcite , the appearance of orbital signature in some stage of the outburst may have contributed this change of the period .
further detailed observations during future superoutbursts and in quiescence are desired .
( 88mm,110mm)fig48.eps ( 88mm,110mm)fig49.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55477.0085 & 0.0006 & @xmath120.0048 & 65 + 1 & 55477.0983 & 0.0003 & @xmath120.0041 & 227 + 2 & 55477.1877 & 0.0004 & @xmath120.0039 & 331 + 3 & 55477.2770 & 0.0004 & @xmath120.0037 & 94 + 12 & 55478.0893 & 0.0004 & 0.0065 & 93 + 13 & 55478.1778 & 0.0004 & 0.0059 & 94 + 14 & 55478.2674 & 0.0004 & 0.0065 & 94 + 34 & 55480.0493 & 0.0008 & 0.0060 & 111 + 35 & 55480.1349 & 0.0006 & 0.0025 & 222 + 45 & 55481.0229 & 0.0024 & @xmath120.0007 & 50 + 46 & 55481.1100 & 0.0006 & @xmath120.0028 & 94 + 47 & 55481.1989 & 0.0008 & @xmath120.0030 & 94 + 48 & 55481.2864 & 0.0009 & @xmath120.0045 & 94
+ + + + this object ( hereafter ot j0120 ) was discovered as a large - amplitude transient by k. itagaki on 2010 november 30.50663 at an unfiltered ccd magnitude of 12.3 ( vsnet - alert 12431 ) .
immediate follow - up observations clearly indicated the presence of early superhumps ( vsnet - alert 12436 ; figure [ fig : j0120eshpdm ] ) .
the object started to show ordinary superhumps 11 d after the discovery ( vsnet - alert 12491 , 12496 ; figure [ fig : j0120shpdm ] ) . since some of the observations were recorded at high airmasses ( larger than 3 ) , secondary corrections for atmospheric extinction were applied to these observations .
the times of superhump maxima are listed in table [ tab : j0120oc2010 ] . due to the gap in the observations , stage
a superhumps were not recorded .
following the stage b with a significant increase in the period [ @xmath59 , the period quickly shortened .
the mean period for the interval bjd 2455554.42455556.3 was 0.05672(8 ) d ( pdm method ) , which is 0.8 % shorter than the period of early superhumps ( close to @xmath18 ) .
this rapid excursion to a shorter period is similar to the case in sdss j0804 ( subsection [ obj : j0804 ] ) , and it might be a common property in wz sge - type dwarf novae . whether this period is related to negative superhumps or not requires further detailed examination .
the object underwent multiple rebrightenings ( vsnet - alert 12538 , 12562 , 12577 , 12596 , 12602 , 12610 , 12674 ) . although the exact number of rebrightenings was unclear , the shortest intervals of rebrightenings were @xmath92 d , close to those of wz sge ( 2001 ; @xcite ; @xcite ; @xcite ) and al com
( 2007 ; @xcite ) , a group of rebrightening classified as type - a wz sge - type superoutbursts in @xcite and @xcite .
a comparison with wz sge ( 2001 ) ( figure [ fig : j0120humpall ] ) suggests at least nine rebrightenings during the observable period .
an analysis of selected high signal - to - noise observations during the rebrightening phase supports the presence of persistent superhumps with a period of 0.05787(3 ) d , which is close to @xmath2 during the main superoutburst [ fig : j0120rebpdm ] .
the observed amplitude of these modulations were small because the observations were relatively restricted to maxima of rebrightening ( see a discussion in sdss j0804 , subsection [ obj : j0804 ] ) .
these modulations apparently persisted , even with larger amplitudes , 18 d after the final observation of the rebrightening phase ( vsnet - alert 12762 ) .
( 88mm,110mm)fig50.eps ( 88mm,110mm)fig51.eps ( 88mm,110mm)fig52.eps ( 88mm,110mm)fig53.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55544.4060 & 0.0003 & 0.0031 & 189 + 1 & 55544.4631 & 0.0005 & 0.0024 & 100 + 2 & 55544.5221 & 0.0003 & 0.0036 & 55 + 4 & 55544.6375 & 0.0004 & 0.0034 & 53 + 5 & 55544.6949 & 0.0006 & 0.0030 & 47 + 6 & 55544.7524 & 0.0004 & 0.0026 & 50 + 10 & 55544.9825 & 0.0006 & 0.0015 & 77 + 11 & 55545.0401 & 0.0004 & 0.0013 & 95 + 12 & 55545.0997 & 0.0005 & 0.0031 & 94 + 13 & 55545.1567 & 0.0006 & 0.0022 & 95 + 16 & 55545.3300 & 0.0004 & 0.0022 & 56 + 17 & 55545.3863 & 0.0003 & 0.0007 & 77 + 18 & 55545.4446 & 0.0004 & 0.0011 & 79 + 19 & 55545.5015 & 0.0005 & 0.0002 & 51 + 21 & 55545.6170 & 0.0006 & 0.0001 & 35 + 22 & 55545.6743 & 0.0008 & @xmath120.0003 & 48 + 34 & 55546.3688 & 0.0005 & 0.0004 & 86 + 35 & 55546.4252 & 0.0002 & @xmath120.0010 & 109 + 36 & 55546.4831 & 0.0004 & @xmath120.0009 & 73 + 44 & 55546.9463 & 0.0010 & @xmath120.0002 & 127 + 45 & 55547.0044 & 0.0010 & 0.0001 & 83 + 47 & 55547.1127 & 0.0010 & @xmath120.0072 & 216 + 48 & 55547.1743 & 0.0016 & @xmath120.0034 & 42 + 49 & 55547.2366 & 0.0007 & 0.0011 & 29 + 61 & 55547.9294 & 0.0010 & 0.0001 & 104 + 62 & 55547.9860 & 0.0007 & @xmath120.0010 & 134 + 72 & 55548.5637 & 0.0008 & @xmath120.0015 & 45 + 73 & 55548.6206 & 0.0008 & @xmath120.0024 & 47 + 74 & 55548.6773 & 0.0009 & @xmath120.0034 & 47 + 75 & 55548.7353 & 0.0007 & @xmath120.0032 & 52 + 78 & 55548.9143 & 0.0019 & 0.0023 & 105 + 79 & 55548.9589 & 0.0022 & @xmath120.0109 & 109 + 80 & 55549.0229 & 0.0020 & @xmath120.0047 & 108 + 81 & 55549.0815 & 0.0021 & @xmath120.0039 & 115 + 82 & 55549.1393 & 0.0025 & @xmath120.0039 & 93 + 88 & 55549.4903 & 0.0018 & 0.0003 & 30 + 89 & 55549.5510 & 0.0020 & 0.0031 & 32 + 101 & 55550.2428 & 0.0015 & 0.0012 & 13 + 102 & 55550.2924 & 0.0018 & @xmath120.0070 & 17 + 106 & 55550.5285 & 0.0008 & @xmath120.0022 & 46 + 107 & 55550.5858 & 0.0008 & @xmath120.0027 & 39 + 108 & 55550.6420 & 0.0010 & @xmath120.0043 & 53 + 109 & 55550.6995 & 0.0012 & @xmath120.0046 & 48 + 110 & 55550.7599 & 0.0060 & @xmath120.0020 & 28 + 113 & 55550.9334 & 0.0020 & @xmath120.0019 & 169 + 114 & 55550.9896 & 0.0016 & @xmath120.0035 & 158 + 115 & 55551.0373 & 0.0028 & @xmath120.0136 & 169 + 116 & 55551.1146 & 0.0028 & 0.0059 & 165 + 118 & 55551.2208 & 0.0008 & @xmath120.0035 & 31 + 120 & 55551.3335 & 0.0034 & @xmath120.0064 & 10 + 134 & 55552.1490 & 0.0003 & @xmath120.0002 & 239 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 135 & 55552.2052 & 0.0003 & @xmath120.0019 & 180 + 136 & 55552.2619 & 0.0003 & @xmath120.0030 & 253 + 137 & 55552.3190 & 0.0005 & @xmath120.0037 & 204 + 152 & 55553.1998 & 0.0011 & 0.0100 & 255 + 153 & 55553.2536 & 0.0003 & 0.0059 & 118 + 154 & 55553.3109 & 0.0007 & 0.0055 & 253 + 155 & 55553.3626 & 0.0005 & @xmath120.0007 & 188 + 159 & 55553.6016 & 0.0015 & 0.0071 & 47 + 160 & 55553.6578 & 0.0016 & 0.0055 & 40 + 161 & 55553.7143 & 0.0016 & 0.0042 & 44 + 164 & 55553.8902 & 0.0021 & 0.0066 & 79 + 165 & 55553.9474 & 0.0020 & 0.0061 & 104 + 166 & 55554.0102 & 0.0024 & 0.0110 & 105 + 167 & 55554.0671 & 0.0035 & 0.0102 & 219 + 168 & 55554.1198 & 0.0016 & 0.0051 & 216 + 169 & 55554.1802 & 0.0009 & 0.0076 & 226 + 170 & 55554.2370 & 0.0003 & 0.0066 & 174 + 171 & 55554.2950 & 0.0004 & 0.0068 & 204 + 172 & 55554.3519 & 0.0003 & 0.0059 & 254 + 173 & 55554.4119 & 0.0006 & 0.0081 & 99 + 174 & 55554.4671 & 0.0008 & 0.0055 & 62 + 184 & 55555.0344 & 0.0021 & @xmath120.0053 & 168 + 185 & 55555.1030 & 0.0022 & 0.0055 & 213 + 186 & 55555.1580 & 0.0017 & 0.0026 & 285 + 187 & 55555.2166 & 0.0008 & 0.0035 & 111 + 188 & 55555.2705 & 0.0009 & @xmath120.0004 & 186 + 190 & 55555.3914 & 0.0012 & 0.0048 & 192 + 191 & 55555.4326 & 0.0024 & @xmath120.0118 & 136 + 195 & 55555.6597 & 0.0103 & @xmath120.0159 & 62 + 196 & 55555.7167 & 0.0028 & @xmath120.0167 & 61 + 197 & 55555.7706 & 0.0023 & @xmath120.0206 & 61 + + + + this object (= css101127:014150@xmath0090822 ) , hereafter ot j0141 ) was discovered by the crts on 2010 november 27 .
no previous outburst was known .
soon after the discovery , superhumps were detected ( vsnet - alert 12422 ; figure [ fig : j0141shpdm ] ) .
the times of superhump maxima are listed in table [ tab : j0141oc2010 ] .
stage b with a @xmath13 of @xmath60 and the subsequent stage c were apparently observed , and we listed parameters in table [ tab : perlist ] based on this identification .
( 88mm,110mm)fig54.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55527.9421 & 0.0010 & 0.0025 & 126 + 1 & 55528.0072 & 0.0015 & 0.0051 & 125 + 2 & 55528.0677 & 0.0008 & 0.0032 & 128 + 19 & 55529.1230 & 0.0007 & @xmath120.0037 & 128 + 34 & 55530.0618 & 0.0022 & @xmath120.0021 & 96 + 35 & 55530.1254 & 0.0010 & @xmath120.0010 & 131 + 67 & 55532.1189 & 0.0027 & @xmath120.0067 & 79 + 68 & 55532.1811 & 0.0033 & @xmath120.0070 & 114 + 114 & 55535.0687 & 0.0033 & 0.0065 & 128 + 115 & 55535.1298 & 0.0041 & 0.0051 & 93 + 130 & 55536.0598 & 0.0026 & @xmath120.0020 & 128 + 131 & 55536.1245 & 0.0020 & 0.0001 & 115
+ + + + this object (= css110125:041350@xmath0094515 ) , hereafter ot j0413 ) was discovered by the crts on 2011 january 25 .
short - period superhumps were immediately detected ( vsnet - alert 12713 , 12719 ; figure [ fig : j0413shpdm ] ) .
the times of superhump maxima are listed in table [ tab : j0413oc2011 ] . a relatively small @xmath61 was detected .
the superhump period resembles those of wz sge - type dwarf novae , such as gw lib and uw tri .
although the early superhumps were not observed , this object may have been detected well after the true maximum ( the last preceding crts negative observation was on 2011 january 9 ) .
although the lack of past outbursts might also support the interpretation as a potential wz sge - type dwarf nova , the object might belong to group `` x '' in @xcite , a small group of short-@xmath18 su uma - type dwarf novae without noticeable wz sge - type characteristics .
observations of future outbursts are desired .
( 88mm,110mm)fig55.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55587.6103 & 0.0005 & 0.0007 & 54 + 1 & 55587.6645 & 0.0006 & 0.0001 & 54 + 2 & 55587.7205 & 0.0004 & 0.0013 & 54 + 3 & 55587.7736 & 0.0006 & @xmath120.0004 & 54 + 18 & 55588.5962 & 0.0005 & @xmath120.0001 & 54 + 19 & 55588.6502 & 0.0006 & @xmath120.0008 & 54 + 20 & 55588.7055 & 0.0007 & @xmath120.0004 & 54 + 21 & 55588.7597 & 0.0006 & @xmath120.0010 & 53 + 36 & 55589.5831 & 0.0007 & 0.0002 & 54 + 37 & 55589.6360 & 0.0006 & @xmath120.0018 & 54 + 38 & 55589.6918 & 0.0007 & @xmath120.0008 & 53 + 39 & 55589.7477 & 0.0007 & 0.0003 & 53 + 54 & 55590.5699 & 0.0012 & 0.0002 & 27 + 55 & 55590.6239 & 0.0008 & @xmath120.0005 & 38 + 56 & 55590.6796 & 0.0007 & 0.0003 & 38 + 57 & 55590.7340 & 0.0008 & @xmath120.0001 & 37 + 58 & 55590.7890 & 0.0012 & 0.0001 & 22 + 73 & 55591.6125 & 0.0009 & 0.0014 & 36 + 74 & 55591.6676 & 0.0010 & 0.0017 & 38 + 75 & 55591.7226 & 0.0009 & 0.0018 & 37 + 76 & 55591.7766 & 0.0012 & 0.0010 & 29 + 100 & 55593.0881 & 0.0024 & @xmath120.0031 & 87
+ + + + this object (= css110113:043112@xmath12031452 ) , hereafter ot j0431 ) was discovered by the crts on 2011 january 13 .
subsequent observations detected both superhumps and eclipses ( vsnet - alert 12612 , 12613 , 12615 ; figure [ fig : j0431shpdm ] ) .
the times of recorded eclipses , determined with the kw method , after removing linearly approximated trends around eclipses in order to minimize the effect of superhumps , are summarized in table [ tab : j0431ecl ] .
we obtained an ephemeris of @xmath62 the times of superhump maxima outside the eclipses are listed in table [ tab : j0431oc2011 ] .
well - refined stage b and c superhumps were recorded .
the @xmath13 and fractional excess of stage b superhumps are @xmath63 and 2.3 % , respectively , and are usual values for this @xmath18 .
it is noteworthy that this object is the first eclipsing su uma - type dwarf nova with a very distinct positive @xmath13 and with distinct stages b and c [ the best observed such an object dates back to xz eri @xcite , with poorer statistics ] . as in ht cas ( subsection [ obj : htcas ] ) ,
a strong beat phenomenon was observed ( figure [ fig : j0431shprof ] ) .
( 88mm,110mm)fig56.eps ( 88mm,110mm)fig57.eps cccc @xmath10 & minimum & error & @xmath5 + 0 & 55575.27191 & 0.00012 & -0.00020 + 1 & 55575.33844 & 0.00018 & 0.00028 + 2 & 55575.40389 & 0.00007 & -0.00032 + 15 & 55576.26287 & 0.00007 & 0.00002 + 16 & 55576.32920 & 0.00030 & 0.00030 + 23 & 55576.79243 & 0.00005 & 0.00118 + 37 & 55577.71577 & 0.00006 & -0.00017 + 38 & 55577.78178 & 0.00004 & -0.00021 + 39 & 55577.84786 & 0.00006 & -0.00018 + 52 & 55578.70661 & 0.00005 & -0.00007 + 53 & 55578.77274 & 0.00005 & 0.00001 + 58 & 55579.10194 & 0.00010 & -0.00103 + 67 & 55579.69737 & 0.00005 & -0.00005 + 68 & 55579.76367 & 0.00009 & 0.00020 + 69 & 55579.82847 & 0.00006 & -0.00106 + 82 & 55580.68795 & 0.00005 & -0.00022 + 83 & 55580.75397 & 0.00005 & -0.00025 + 84 & 55580.81997 & 0.00006 & -0.00029 + 113 & 55582.73495 & 0.00009 & -0.00075 + 114 & 55582.80158 & 0.00006 & -0.00017 + 128 & 55583.72630 & 0.00005 & -0.00014 + 129 & 55583.79240 & 0.00006 & -0.00009 + 143 & 55584.71737 & 0.00004 & 0.00019 + 144 & 55584.78362 & 0.00006 & 0.00038 + 158 & 55585.70787 & 0.00006 & -0.00005 + 159 & 55585.77401 & 0.00008 & 0.00004 + 173 & 55586.69850 & 0.00003 & -0.00017 + 174 & 55586.76447 & 0.00003 & -0.00024 + 187 & 55587.62349 & 0.00004 & 0.00013 + 188 & 55587.68957 & 0.00006 & 0.00016 + 189 & 55587.75559 & 0.00006 & 0.00013 + 203 & 55588.68015 & 0.00004 & -0.00000 + 204 & 55588.74617 & 0.00002 & -0.00003 + 205 & 55588.81211 & 0.00003 & -0.00014 + + + cccccc @xmath10 & max & error & @xmath5 & phase & @xmath11 + 0 & 55575.2417 & 0.0006 & 0.0021 & 0.54 & 51 + 1 & 55575.3078 & 0.0004 & 0.0007 & 0.54 & 66 + 2 & 55575.3750 & 0.0004 & 0.0004 & 0.56 & 59 + 3 & 55575.4422 & 0.0005 & @xmath120.0001 & 0.57 & 62 + 15 & 55576.2565 & 0.0011 & 0.0039 & 0.90 & 30 + 16 & 55576.3209 & 0.0009 & 0.0007 & 0.88 & 54 + 23 & 55576.7924 & 0.0011 & @xmath120.0005 & 0.02 & 50 + 37 & 55577.7350 & 0.0007 & @xmath120.0034 & 0.29 & 56 + 38 & 55577.8011 & 0.0006 & @xmath120.0048 & 0.29 & 56 + 40 & 55577.9387 & 0.0012 & @xmath120.0023 & 0.37 & 102 + 45 & 55578.2738 & 0.0006 & @xmath120.0048 & 0.45 & 45 + 46 & 55578.3437 & 0.0006 & @xmath120.0025 & 0.50 & 42 + 47 & 55578.4088 & 0.0006 & @xmath120.0050 & 0.49 & 48 + 52 & 55578.7490 & 0.0008 & @xmath120.0024 & 0.64 & 59 + 56 & 55579.0109 & 0.0028 & @xmath120.0107 & 0.61 & 26 + 57 & 55579.0860 & 0.0029 & @xmath120.0031 & 0.74 & 77 + 66 & 55579.6945 & 0.0020 & @xmath120.0023 & 0.96 & 32 + 67 & 55579.7617 & 0.0019 & @xmath120.0027 & 0.97 & 55 + 81 & 55580.7120 & 0.0006 & 0.0021 & 0.36 & 59 + 82 & 55580.7798 & 0.0006 & 0.0024 & 0.39 & 62 + 83 & 55580.8459 & 0.0011 & 0.0010 & 0.39 & 56 + 110 & 55582.6780 & 0.0024 & 0.0097 & 0.13 & 41 + 111 & 55582.7444 & 0.0014 & 0.0085 & 0.13 & 55 + 112 & 55582.8100 & 0.0012 & 0.0066 & 0.13 & 55 + 125 & 55583.6898 & 0.0011 & 0.0085 & 0.45 & 49 + 126 & 55583.7572 & 0.0005 & 0.0083 & 0.47 & 68 + 127 & 55583.8255 & 0.0009 & 0.0091 & 0.50 & 66 + 140 & 55584.7010 & 0.0005 & 0.0066 & 0.76 & 55 + 141 & 55584.7681 & 0.0007 & 0.0062 & 0.77 & 56 + 155 & 55585.7060 & 0.0008 & @xmath120.0014 & 0.97 & 57 + 156 & 55585.7705 & 0.0010 & @xmath120.0045 & 0.95 & 56 + 170 & 55586.7186 & 0.0009 & @xmath120.0018 & 0.30 & 56 + 171 & 55586.7854 & 0.0009 & @xmath120.0025 & 0.31 & 55 + 183 & 55587.6027 & 0.0026 & 0.0044 & 0.69 & 64 + 184 & 55587.6666 & 0.0012 & 0.0007 & 0.65 & 66 + 185 & 55587.7316 & 0.0010 & @xmath120.0018 & 0.64 & 70 + 199 & 55588.6665 & 0.0028 & @xmath120.0124 & 0.79 & 55 + 200 & 55588.7333 & 0.0044 & @xmath120.0131 & 0.80 & 56
+ + + + + this object (= css071115:044216@xmath12002334 ) , hereafter ot j0442 ) was discovered by the crts in 2007 january .
three outburst have been known , and the 2011 january one was the brightest .
superhumps were subsequently detected ( vsnet - alert 12784 , 12785 , 12786 ; figure [ fig : j0442shpdm ] ) . since the observations were obtained at high airmasses , we corrected observations by using a second - order atmospheric extinction .
the times of superhump maxima are listed in table [ tab : j0442oc2011 ] .
( 88mm,110mm)fig58.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55596.3030 & 0.0030 & @xmath120.0009 & 23 + 1 & 55596.3821 & 0.0011 & 0.0015 & 41 + 13 & 55597.3014 & 0.0023 & @xmath120.0003 & 24 + 14 & 55597.3777 & 0.0013 & @xmath120.0008 & 40 + 26 & 55598.2999 & 0.0017 & 0.0002 & 33 + 27 & 55598.3767 & 0.0019 & 0.0003 & 34 + + + + this object (= css091026:064805@xmath0414702 , hereafter ot j0648 ) was discovered in outburst by the crts in 2009 october . during its 2011 january outburst ,
superhumps were detected ( vsnet - alert 12735 , 12737 ; figure [ fig : j0648shpdm ] ) .
the times of superhump maxima are listed in table [ tab : j0648oc2011 ] .
the outburst was apparently already in stage b , and a transition to stage c superhumps was recorded five days later .
( 88mm,110mm)fig59.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55590.5903 & 0.0006 & 0.0011 & 67 + 1 & 55590.6583 & 0.0002 & 0.0027 & 126 + 2 & 55590.7236 & 0.0003 & 0.0017 & 126 + 3 & 55590.7902 & 0.0003 & 0.0020 & 126 + 4 & 55590.8552 & 0.0004 & 0.0007 & 126 + 5 & 55590.9218 & 0.0004 & 0.0011 & 125 + 10 & 55591.2531 & 0.0011 & 0.0008 & 133 + 12 & 55591.3849 & 0.0008 & 0.0001 & 76 + 13 & 55591.4514 & 0.0007 & 0.0003 & 88 + 14 & 55591.5168 & 0.0009 & @xmath120.0006 & 68 + 23 & 55592.1114 & 0.0013 & @xmath120.0026 & 105 + 24 & 55592.1804 & 0.0013 & 0.0001 & 129 + 25 & 55592.2461 & 0.0018 & @xmath120.0005 & 119 + 26 & 55592.3094 & 0.0021 & @xmath120.0035 & 82 + 27 & 55592.3749 & 0.0013 & @xmath120.0043 & 60 + 28 & 55592.4402 & 0.0023 & @xmath120.0053 & 66 + 29 & 55592.5112 & 0.0013 & @xmath120.0006 & 58 + 42 & 55593.3715 & 0.0022 & @xmath120.0021 & 63 + 43 & 55593.4412 & 0.0017 & 0.0013 & 23 + 44 & 55593.5072 & 0.0023 & 0.0010 & 68 + 69 & 55595.1644 & 0.0056 & 0.0008 & 134 + 70 & 55595.2399 & 0.0102 & 0.0100 & 81 + 83 & 55596.0875 & 0.0028 & @xmath120.0042 & 122 + + + + this object (= css110414:075414@xmath0313216 , hereafter ot j0754 ) was discovered by the crts on 2011 april 14 . two earlier outbursts ( 2007 february and 2008 november )
were also detected by the crts , and the 2008 one reached a maximum of 14.2 mag .
subsequent observations detected superhumps ( vsnet - alert 13166 , 13186 , 13202 ) . due to the limited nightly baselines , the possibility of one - day alias is not completely excluded ( figure [ fig : j0754shpdm ] ) .
the times of superhump maxima are listed in table [ tab : j0754oc2011 ] .
although both the overall light curve and , @xmath5 variation and amplitudes of superhumps ( cf . subsection
[ sec : humpamp ] ) showed a typical course of a stage b
c transition ( vsnet - alert 13219 ) , the baseline was not sufficient to determine @xmath13 for stage b. ( 88mm,110mm)fig60.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55666.3975 & 0.0012 & @xmath120.0050 & 65 + 1 & 55666.4626 & 0.0015 & @xmath120.0027 & 55 + 16 & 55667.4058 & 0.0009 & @xmath120.0019 & 64 + 17 & 55667.4713 & 0.0016 & 0.0008 & 43 + 31 & 55668.3503 & 0.0011 & 0.0002 & 47 + 32 & 55668.4154 & 0.0009 & 0.0025 & 65 + 33 & 55668.4844 & 0.0023 & 0.0086 & 38 + 63 & 55670.3587 & 0.0015 & @xmath120.0018 & 33 + 64 & 55670.4252 & 0.0018 & 0.0019 & 34 + 65 & 55670.4858 & 0.0022 & @xmath120.0004 & 29 + 79 & 55671.3661 & 0.0015 & 0.0004 & 31 + 80 & 55671.4291 & 0.0026 & 0.0005 & 34 + 81 & 55671.4993 & 0.0038 & 0.0079 & 18 + 95 & 55672.3662 & 0.0050 & @xmath120.0048 & 34 + 96 & 55672.4276 & 0.0045 & @xmath120.0061 & 34 + + + + this object (= css101130:102616@xmath0192045 , hereafter ot j102616 ) was discovered by the crts on 2010 november 30 .
two earlier outbursts ( 2006 april and 2007 april ) were also detected by the crts .
subsequent observations immediately clarified the presence of superhumps ( vsnet - alert 12447 , 12452 , 12459 ; figure [ fig : j102616shpdm ] ) .
the large amplitudes ( @xmath90.3 mag ) of superhumps at the start of the observation suggests that the observation succeeded in recording the relatively early stage of the superoutburst .
the times of superhump maxima are listed in table [ tab : j102616oc2010 ] .
there were a large systematic decrease in the superhump period , amounting to a global @xmath64 and a mean global @xmath65 d from timing analysis .
there appeared to have been a discontinuous decrease of the period at @xmath66 and early epochs ( @xmath67 ) may have resulted from stage a superhumps .
although we list values in table [ tab : perlist ] based on this interpretation , there remains a possibility that the entire observation should be interpreted as stage b , as discussed in @xcite ( subsection 4.10 ) .
it is noteworthy that a relatively large negative @xmath13 was recorded in a system with a long-@xmath18 and with relatively infrequent outbursts .
this object should require further attention .
( 88mm,110mm)fig61.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55533.4881 & 0.0006 & @xmath120.0094 & 131 + 1 & 55533.5744 & 0.0001 & @xmath120.0058 & 245 + 13 & 55534.5738 & 0.0004 & 0.0016 & 74 + 22 & 55535.3196 & 0.0003 & 0.0034 & 321 + 25 & 55535.5677 & 0.0002 & 0.0035 & 291 + 36 & 55536.4789 & 0.0003 & 0.0053 & 127 + 47 & 55537.3859 & 0.0005 & 0.0029 & 92 + 48 & 55537.4685 & 0.0003 & 0.0029 & 131 + 61 & 55538.5434 & 0.0004 & 0.0030 & 109 + 62 & 55538.6257 & 0.0006 & 0.0026 & 73 + 70 & 55539.2847 & 0.0008 & 0.0003 & 105 + 82 & 55540.2738 & 0.0006 & @xmath120.0026 & 100 + 95 & 55541.3495 & 0.0023 & @xmath120.0017 & 100 + 107 & 55542.3373 & 0.0012 & @xmath120.0059 & 185 + + + + this object (= sss110314:102706@xmath12434341 , hereafter ot j1027 ) was discovered by the crts sss on 2011 march 14 .
there were four past outbursts in crts sss data , and has an x - ray counterpart 1rxp [email protected] .
subsequent observations detected superhumps ( vsnet - alert 12997 , 13009 ; figure [ fig : j1027shpdm ] ) .
the times of superhump maxima are listed in table [ tab : j1027oc2011 ] . the period given in table [ tab : perlist ]
was determined by the pdm method .
( 88mm,110mm)fig62.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55637.0723 & 0.0007 & @xmath120.0001 & 109 + 1 & 55637.1532 & 0.0006 & @xmath120.0001 & 109 + 2 & 55637.2351 & 0.0006 & 0.0009 & 109 + 3 & 55637.3142 & 0.0007 & @xmath120.0009 & 110 + 12 & 55638.0444 & 0.0008 & 0.0013 & 79 + 13 & 55638.1242 & 0.0009 & 0.0003 & 84 + 14 & 55638.2024 & 0.0010 & @xmath120.0024 & 84 + 15 & 55638.2867 & 0.0011 & 0.0010 & 85 + + + + this object (= css110205:120053@xmath12152620 , hereafter ot j1200 ) was discovered by the crts on 2011 february 5 . both crts and asas-3 @xcite
recorded numerous past outbursts , and the brightest outburst reaching @xmath68 on 2001 february 14 recorded by asas-3 was indicative of a superoutburst ( vsnet - alert 12792 ) . upon this information
, follow - up observations indeed clarified the presence of superhumps ( vsnet - alert 12803 , 12815 ; figure [ fig : j1200shpdm ] ) .
the times of superhump maxima are listed in table [ tab : j1200oc2011 ] . due to the short baseline of observations
, we adopted the most likely one - day alias .
other aliases still remain viable .
we adopted the period determined by the pdm method in table [ tab : perlist ] .
( 88mm,110mm)fig63.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55599.6371 & 0.0005 & @xmath120.0004 & 87 + 9 & 55600.5242 & 0.0011 & 0.0041 & 62 + 10 & 55600.6145 & 0.0004 & @xmath120.0037 & 97
+ + + + this object (= sss110403:132901@xmath12365859 ) , hereafter ot j1329 ) was discovered by the crts sss on 2011 april 3 .
subsequent observations detected apparent superhumps ( vsnet - alert 13105 ; figure [ fig : j1329shpdm ] ) .
the times of superhump maxima are listed in table [ tab : j1329oc2011 ] .
there appears to have been a stage transition between @xmath69 and @xmath70 , and we listed a period from @xmath33 for the period of stage b superhumps in table [ tab : perlist ] . since there was no earlier record of outbursts in crts sss , the object seems to have a relatively low outburst frequency .
( 88mm,110mm)fig64.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55656.4074 & 0.0011 & @xmath120.0050 & 118 + 1 & 55656.4796 & 0.0014 & @xmath120.0033 & 162 + 13 & 55657.3296 & 0.0011 & 0.0007 & 162 + 14 & 55657.4017 & 0.0010 & 0.0023 & 161 + 15 & 55657.4743 & 0.0009 & 0.0045 & 162 + 16 & 55657.5439 & 0.0020 & 0.0035 & 155 + 59 & 55660.5723 & 0.0020 & 0.0007 & 162 + 60 & 55660.6386 & 0.0020 & @xmath120.0034 & 159 + + + + this object (= css110428:154545 + 442830 ) , hereafter ot j1545 ) was discovered by the crts on 2011 april 28 .
subsequent observations successfully detected superhumps ( vsnet - alert 13275 ; figure [ fig : j1545shpdm ] ) .
the times of superhump maxima are listed in table [ tab : j1545oc2011 ] .
the object started rapid fading only three days after the initial observation , and the outburst was observed only during its final phase .
we identified the superhumps as being stage c superhumps .
( 88mm,110mm)fig65.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55681.1443 & 0.0050 & @xmath120.0045 & 83 + 1 & 55681.2289 & 0.0007 & 0.0030 & 157 + 17 & 55682.4600 & 0.0037 & 0.0020 & 18 + 18 & 55682.5354 & 0.0021 & 0.0004 & 25 + 29 & 55683.3812 & 0.0020 & @xmath120.0008 & 22 + 30 & 55683.4587 & 0.0019 & @xmath120.0003 & 21 + 31 & 55683.5344 & 0.0017 & @xmath120.0017 & 22 + 32 & 55683.6176 & 0.0021 & 0.0045 & 14 + 42 & 55684.3821 & 0.0011 & @xmath120.0011 & 25 + 43 & 55684.4567 & 0.0018 & @xmath120.0035 & 23 + 44 & 55684.5342 & 0.0018 & @xmath120.0029 & 25 + 45 & 55684.6191 & 0.0040 & 0.0049 & 15
+ + + + this object (= css090910:223418@xmath12035530 ) , hereafter ot j2234 ) was discovered by the crts on 2009 september 10 .
we observed this object during two superoutbursts in 2009 and 2010 .
although the existence of superhumps were already apparent in 2009 ( cf .
vsnet - alert 11480 ) , the short observations were unable to select a unique period . using the 2010 observations ,
we have selected a unique period , and the @xmath5 s referring to this period are listed in tables [ tab : j2234oc2009 ] and [ tab : j2234oc2010 ] .
since the amplitudes of superhumps were low ( @xmath9 0.03 mag ) and they grew to 0.18 mag six days later , the initial observations were apparently performed during the growing stage of superhumps . in table
[ tab : perlist ] , we listed averaged periods for the entire runs since observations were too insufficient for determining stages and period changes .
the superhumps recorded during the 2010 superoutburst is shown in figure [ fig : j2234shpdm ] .
( 88mm,110mm)fig66.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55084.9749 & 0.0184 & 0.0014 & 134 + 1 & 55085.0638 & 0.0052 & @xmath120.0018 & 194 + 66 & 55091.0591 & 0.0012 & 0.0010 & 194 + 109 & 55095.0217 & 0.0026 & @xmath120.0006 & 186 + + + + ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55482.0178 & 0.0008 & 0.0001 & 93 + 1 & 55482.1094 & 0.0007 & @xmath120.0002 & 95 + 2 & 55482.2017 & 0.0011 & 0.0001 & 78 + + + + this transient was originally reported as a possible nova @xcite . soon after the announcement , the color of the quiescent counterpart inferred a dwarf - nova outburst ( vsnet - alert 12548 ) .
initial time - resolved photometry recorded relatively small variations ( a. arai , vsnet - alert 12558 ) .
subsequent observations indicated the presence of superhumps with amplitude of 0.06 mag ( a. arai , vsnet - alert 12563 ) .
this finding was confirmed by subsequent observations ( vsnet - alert 12564 , 12578 , 12633 ; figure [ fig : j2304shpdm ] ) . since the observations were obtained at high airmasses , we corrected observations by using a second - order atmospheric extinction .
the resultant times of superhump maxima are listed in table [ tab : j2304oc2011 ] .
since the observing windows were limited to the low evening sky , there remained an ambiguity in selecting the alias ( cf .
vsnet - alert 12578 ) .
the superhumps queerly grew since the detection until @xmath71 , when the period and amplitudes suddenly decreased ( figure [ fig : j2304prof ] ) .
since the fractional decrease ( 1.4 % ) is exceptionally large for a stage b - c transition @xcite , we leave the interpretation of the later stage open though we tentatively assign the values in table [ tab : perlist ] .
although the results may have been affected by adverse observing conditions , the unusual development of superhumps in this system should warrant further study during future outbursts .
( 88mm,110mm)fig67.eps ( 88mm,110mm)fig68.eps ccccc @xmath10 & max & error & @xmath5 & @xmath11 + 0 & 55563.9791 & 0.0028 & @xmath120.0354 & 108 + 29 & 55565.9214 & 0.0011 & @xmath120.0273 & 274 + 30 & 55565.9956 & 0.0023 & @xmath120.0198 & 64 + 59 & 55567.9479 & 0.0017 & @xmath120.0019 & 104 + 60 & 55568.0077 & 0.0015 & @xmath120.0087 & 70 + 74 & 55568.9571 & 0.0016 & 0.0069 & 215 + 88 & 55569.8972 & 0.0014 & 0.0132 & 347 + 89 & 55569.9608 & 0.0018 & 0.0101 & 427 + 103 & 55570.9046 & 0.0013 & 0.0201 & 116 + 104 & 55570.9694 & 0.0011 & 0.0182 & 107 + 118 & 55571.9022 & 0.0009 & 0.0172 & 217 + 119 & 55571.9743 & 0.0014 & 0.0226 & 211 + 122 & 55572.1788 & 0.0022 & 0.0271 & 15 + 123 & 55572.2380 & 0.0014 & 0.0195 & 12 + 137 & 55573.1684 & 0.0071 & 0.0162 & 16 + 149 & 55573.9517 & 0.0014 & @xmath120.0009 & 148 + 164 & 55574.9499 & 0.0013 & @xmath120.0032 & 107 + 179 & 55575.9383 & 0.0022 & @xmath120.0153 & 150 + 224 & 55578.9187 & 0.0043 & @xmath120.0364 & 237 + 254 & 55580.9340 & 0.0014 & @xmath120.0221 & 192 + + + +
the empirical relation by @xcite between @xmath2 and @xmath18 have been widely used in estimating @xmath18 from @xmath2 ( e.g. @xcite ) .
since we now have a plenty of objects and know the difference between stage b and c superhumps , we provide an updated calibration .
since it was already evident that @xmath18@xmath23 relation deviates from a linear one ( figure 15 in @xcite ) , we have introduced a @xmath72-type dependence instead of the linear one in @xcite . using all samples with known @xmath18 and well - defined stage b superhumps ( following the criteria used in @xcite )
, we have derived relations for objects below the period gap : @xmath73 for the mean period of stage b superhumps , and @xmath74 for the mean period of stage c superhumps .
figure [ fig : p1porbres ] represents the residuals for estimated @xmath18 from mean @xmath2 of stage b superhumps .
the relation by @xcite systematically ( 0.0003 d in average ) gives longer @xmath18 .
the 1-@xmath75 error for the estimated @xmath18 from the updated relation is 0.0003 d. figure [ fig : p2porbres ] represents the residuals for estimated @xmath18 from mean @xmath2 of stage c superhumps .
although the relation by @xcite generally well reproduces the real @xmath18 , there remains a small systematic trend ( shorter values for short-@xmath18 and long-@xmath18 systems and longer values for intermediate-@xmath18 systems .
we conclude that the relation by @xcite appears to be generally useful for stage c superhumps , and suggest to use our updated relation for stage b superhumps especially when only early - stage observations and mean superhump periods are available .
the 1-@xmath75 error for the estimated @xmath18 from the updated relation is 0.0003 d. ( 88mm,77mm)fig69.eps ( 88mm,77mm)fig70.eps in the following subsections , orbital periods were estimated using these relations for objects without known orbital periods .
figure [ fig : pdotporb3 ] represents the relation between @xmath18 and @xmath13 during stage b ( note that we use @xmath18 instead of @xmath2 used in earlier papers ) . the enlarged figure ( corresponding to figure 10 in @xcite )
is only shown here .
the new data generally confirmed the predominance of positive period derivatives during stage b in systems with orbital periods shorter than 0.07 d , in agreement with the tendency reported in @xcite and @xcite . although longer period systems ( @xmath76 d ) tend to show zero or slightly negative @xmath13 , the new study suggest the presence of positive-@xmath18 systems in this period regime .
the most striking example is gx cas ( 2010 , subsection [ obj : gxcas ] ) .
( 88mm,77mm)fig71.eps since this table was missing in @xcite , we list values for the second year and this ( third ) year ( table [ tab : pera ] ) following the manner in @xcite . the combined figure is presented in figure [ fig : ppre2 ] .
the fractional period excess ( against periods of stage b superhumps ) has a tendency to increase as @xmath18 increases .
the result for high - accuracy kepler data ( subsection [ sec : kepler ] ) agrees with other ground - based observations .
as shown in subsection [ obj : mmhya ] , mm hya likely had an unusually large fractional period excess for stage a superhumps .
this needs to be confirmed by future observations .
we discussed the origin of stage a superhumps in @xcite and presented two possibilities : larger precession rate for larger disks or longer periods during the growing stage suggested by smoothed particle hydrodynamics ( sph ) simulations
. there have been growing evidence favoring the latter with sph simulations ( @xcite ; @xcite ) and most recently in @xcite , which first successfully produced the growing eccentric mode with a grid - based simulation .
( 88mm,77mm)fig72.eps cccc object & year & period ( d ) & err + nn cam & 2009 & 0.07755 & 0.00025 + v592 her & 2010 & 0.05728 & 0.00029 + ct hya & 2010 & 0.06718 & 0.00015 + ef peg & 2009 & 0.09077 & 0.00063 + ek tra & 2009 & 0.06562 & 0.00054 + iy uma & 2009 & 0.07717 & 0.00014 + 1rxs j0423 & 2010 & 0.07930 & 0.00010 + sdss j1610 & 2009 & 0.05881 & 0.00009 + sdss j1625 & 2010 & 0.09849 & 0.00025 + ot j0506 & 2009 & 0.06964 & 0.00005 + ot j1044 & 2010 & 0.06084 & 0.00003 + ot j1440 & 2009 & 0.06503 & 0.00051 + bg ari & 2010 & 0.08754 & 0.00015 + ht cas & 2010 & 0.07868 & 0.00069 + v1504 cyg & 2009b & 0.07395 & 0.00021 + mm hya & 2011 & 0.06295 & 0.00023 + v344 lyr & 2009 & 0.09314 & 0.00009 + v344 lyr & 2009b & 0.09351 & 0.00011 + v1212 tau & 2011 & 0.07286 & 0.00027 + sw uma & 2010 & 0.05850 & 0.00002 + sdss j0804 & 2010 & 0.06077 & 0.00007 + sdss j1146 & 2011 & 0.06623 & 0.00103 + sdss j1227 & 2007 & 0.06568 & 0.00009 + sdss j1339 & 2011 & 0.05874 & 0.00003 + + + + + + + + table [ tab : wztab ] summarizes the properties of new outbursts of wz sge - type and related objects reported in @xcite and this paper ( the selection criteria are the same as in @xcite ) .
although many of recently discovered wz sge - type objects were not sufficiently observed during the earliest stages of their outbursts , the fractional superhump excesses were relatively well determined .
figure [ fig : wzpdoteps3 ] shows the relation between @xmath13 versus @xmath23 for wz sge - type dwarf novae studied in these series of papers . in @xcite and this paper , higher values @xmath13 were recorded for short-@xmath18 systems , and the relation described in @xcite became less clear .
these outbursts are v592 her ( 2010 ) , ot j2138 ( 2010 ) and sdss j0804 ( 2010 ) . as already described in subsection [ obj : j0804 ]
, the large @xmath13 in sdss j0804 ( 2010 ) is an outlier among objects with multiple rebrightenings .
the large @xmath13 in this system may have been affected by a large `` disturbance '' in the @xmath5 diagram near the end of the superoutburst plateau , and the @xmath13 during the earlier part of the stage b superhumps was much smaller [ @xmath13 = @xmath47 ] . the same phenomenon is sometimes obvious in other wz sge - type dwarf novae [ v455 and ( 2007 ) , figure 44 and the secondary component in wz sge ( 2001 ) , figure 127 in @xcite ] and all of them are eclipsing objects
. the high inclination may be somehow responsible for this phenomenon ( see also a discussion in subsection [ sec : humpamp ] ) , and we might better exclude the `` disturbance '' part of the @xmath5 diagram in determining the representative @xmath13 in such systems .
the interpretation of this phenomenon needs to be investigated further .
wz sge - type dwarf novae show various activities during the post - superoutburst stage . in @xcite , we introduced a classification : type - a ( long - lasting post - outburst rebrightening ) , type - b ( multiple discrete rebrightenings ) , type - c ( single rebrightening ) and type - d ( no rebrightening ) .
it is known that there is a dependence of the types of outbursts on @xmath18 and @xmath23 ( figure 37 in @xcite ; note @xmath2 was instead used ) .
the updated statistics ( figure [ fig : wzpdottype3 ] ) basically strengthens the earlier finding : type - b objects mostly occupy low-@xmath23 and middle-@xmath18 region , type - a and type - d objects occupy middle-@xmath23 and short-@xmath18 region , and type - c objects are more scattered .
the type - d outlier is sdss j0804 as described above .
cccccccccccc object & year & @xmath2 & @xmath18 & @xmath13 & err & @xmath23 & type & @xmath78 & delay & max & min + vx for & 2009 & 0.061327 & & 1.0 & 1.4 & & b & 5 & & ] 12.6 & 20.6 + v592 her & 2010 & 0.056607 & 0.0561 & 7.4 & 0.6 & 0.009 & d & 0 & @xmath794 & ] 14.2 & 21.3 + bc uma & 2009 & 0.064553 & 0.06261 & 9.5 & 2.7 & 0.031 & & & & ] 11.6 & 18.6 + sdss j1610 & 2009 & 0.057820 & 0.05695 & 6.8 & 1.4 & 0.015 & d ? & 0 ? & @xmath792 & ] 13.9 & 20.1 + ot j1044 & 2010 & 0.060236 & 0.05909 & & & 0.019 & c & 1 & @xmath795 & ] 12.6 & 19.3 + ot j2138 & 2010 & 0.055019 & 0.05450 & 7.9 & 0.7 & 0.010 & d & 0 & 6 & 8.7 & [ 15.4
+ ot j2230 & 2009 & & 0.05841 & & & & & & @xmath794 & ] 14.4 & 21.0 + asas j1025 & 2011 & 0.063538 & 0.06136 & 9.9 & 2.4 & 0.035 & & & & ] 12.9 & 19.3 + misv 1443 & 2011 & 0.056725 & & 5.5 & 1.5 & & c ?
& & ] 12.8 & 19.7 + sdss j0804 & 2010 & 0.059630 & 0.059005 & 9.6 & 1.1 & 0.011 & b & 6 & 4 & 12.0 & 17.8 + sdss j1339 & 2011 & 0.058094 & 0.057289 & 5.4 & 0.2 & 0.014 & d & 0 & @xmath796 & ] 9.9 & 17.7 + sdss j1605 & 2010 & & 0.05666 & & & & a ? & ? & @xmath793 & ] 12.0 & 19.9 + ot j0120 & 2010 & 0.057833 & 0.057157 & 4.3 & 0.5 & 0.012 & a & @xmath799 & @xmath7911 & ] 12.3 & 20.0 + + + + + ( 88mm,70mm)fig73.eps ( 88mm,70mm)fig74.eps the data for two su uma - type dwarf novae are available in kepler @xcite public release data .
one of them ( v344 lyr ) has already been published @xcite .
@xcite compared the kepler data with model simulations based on thermal - tidal disk instability .
we examined kepler observations from the viewpoint of the current study .
the magnitudes were calculated from the corrected aperture - integrated flux obtained from the kepler pipeline , with an arbitrary photometric zero - point .
kepler recorded the 2009 september superoutburst of this object ( see subsection [ obj : v1504cyg ] for detailed data ) .
this outburst was composed of a precursor outburst lasting for @xmath92 d , and superhumps started to develop around the maximum of this precursor .
as the superhumps developed , the object stopped fading and entered the plateau phase of the superoutburst ( figure [ fig : v1504kep ] ) .
the plateau phase lasted for @xmath99 d , followed by a steeper decline . during the entire course of the plateau phase and decline phase ,
ordinary superhumps were present .
the amplitudes of superhumps were large around the maximum , initially decayed relatively rapidly for 3.2 d. after reaching a minimum , the amplitudes slightly increased , but again decayed in @xmath9 2 d. the @xmath5 diagram indicates the presence of fine structures rather than our discrete three stages .
the transitions between stages were smoother than in many objects in our present and past studies . in table
[ tab : perlist ] , we assigned stage a before the epoch of maximum amplitude , and stages b and c were divided by a kink in the @xmath5 diagram .
the relatively smooth variation in the @xmath5 diagram and the lack of segment with a positive @xmath13 might be attributable to the notably high frequency of outbursts in this system ( @xcite ; @xcite ) , suggesting a high mass - transfer rate .
although the tendency that objects with high outburst frequencies tend to show smoother @xmath5 variation was originally proposed for long-@xmath18 systems ( @xcite , subsection 4.10 ) , it might also be extended to shorter-@xmath18 systems as in v1504 cyg . yet another interesting feature is in the fine structure during stage a. during the first 8 superhump cycles , the period of superhumps once shortened while the amplitudes stayed low . following this stage of initial development , the amplitudes quickly grew and the period stayed at a relatively constant value .
this later stage has likely been recognized as stage a superhumps in ground - based observations . during the growing stage , small - amplitude ( @xmath90.01 mag )
quasi - periodic oscillations ( qpos ) with time - scales of @xmath910 min were superimposed on superhumps while they disappeared when superhumps quickly grew .
these modulations resemble so - called `` super - qpos '' in some su uma - type dwarf novae during the evolutionary stage of superhumps ( cf .
@xcite ; @xcite ) .
these qpos must be somehow related to the evolutionary process of superhumps .
although @xcite reported humps with a period close to the orbital period on the first day of the superoutburst , there was no special indication of such humps in kepler data . using quiescent segments of kepler observations ,
we have obtained a refined orbital period of 0.069549(2 ) d with a mean amplitude of orbital modulation of 0.016 mag ( figure [ fig : v1504cygquipdm ] ) .
( 170mm,90mm)fig75.eps ( 88mm,110mm)fig76.eps during the normal outburst starting on 2009 august 10 ( two outburst cycles before this superoutburst ) , a prominent `` precursor '' ( or a shoulder ) preceding this outburst was recorded just as in the superoutburst . during the fading stage of its second maxima , negative superhumps with a period of 0.0677(2 )
d [ 2.6(3 ) % shorter than @xmath18 ] appeared ( figure [ fig : v1504prec1 ] ) .
this outburst may have been a failed superoutburst in which tidal instability was excited but it was not strong enough to trigger a full superoutburst .
it is very suggestive that negative superhumps , instead of ordinary ( positive ) superhumps , were excited .
this suggests that ordinary superhumps and negative superhumps can be excited under similar conditions ( or have the common underlying mechanism ) within a relatively short time . in er uma , @xcite reported the presence of negative superhumps during its 2011 january superoutburst and the state with strong negative superhumps appeared to prevent a long - lasting superoutburst .
@xcite also discussed the prevention of disk - instability in the state with strong negative superhumps in v344 lyr . during this normal outburst of v1504 cyg
, excitation of negative superhumps by chance may have prevented an evolution of a full superoutburst . during the next normal outburst
starting on august 23 ( outburst just prior to the superoutburst ) , transient low - amplitude superhumps with a period of 0.07402(7 ) d appeared during its fading branch ( figure [ fig : v1504prec2 ] ) .
this period is very close to the period [ 0.0739(2 ) d ] of stage a superhumps , and the excitation of the superhumps apparently occurred around the maximum of this outburst .
this outburst appears to be phenomenologically similar to the precursor part of the main superoutburst , but the strength of the tidal instability may have not been sufficient to trigger a full superoutburst .
@xcite also reported the presence of short outbursts with durations of less than 1 d and maximum brightness not exceeding 15.2 mag .
@xcite referred to these outburst as `` third type '' of outbursts .
these outbursts are known to occur during the early phase of the supercycle .
they also noted the light variations close to superhump period we detected during one normal outburst ( immediately following a superoutburst ) .
these short outbursts may be somehow related to `` failed '' outbursts observed by kepler , and it is expected that further kepler observations can detect these outbursts .
( 88mm,110mm)fig77.eps ( 88mm,110mm)fig78.eps the kepler observations of v344 lyr were analyzed by @xcite , who reported the presence of ( most likely ) negative superhumps in quiescence and during normal outbursts . during superoutbursts ,
postive superhumps were the predominant signal .
we analyzed kepler observations between 2009 june 20 and december 17 . during this period ,
two superoutbursts were recorded ( 2009 august and 2009 november december ) .
both superoutbursts showed prominent precursors as in v1504 cyg .
@xcite compared calculations based on the thermal - tidal instability model and the enhanced mass - transfer model , and indicated that the thermal - tidal instability model generally well reproduces the precursor although some differences remained between observations and model calculations .
we applied our methods to these kepler observations .
in contrast to v1504 cyg , the superhumps had secondary maxima ( a bump around superhump phase of @xmath9 0.4 ) which became stronger in later stages .
the @xmath5 diagrams for these two superoutbursts are similar ( figures [ fig : v344lyrhumpall ] and [ fig : v344lyrhumpall2 ] ) .
the @xmath5 diagrams of primary maxima showed stages a and b , although the transitions between them were not sharp , as in v1504 cyg .
these relatively smooth transitions may have been caused by a high mass - transfer rate , as in v1504 cyg , which enables short recurrence times of normal outbursts and superoutbursts ( @xcite ; @xcite ; the interval of superoutbursts was 104 d in kepler observations ) .
stages a were less structured compared to v1504 cyg , and there were smooth variations in the periods .
stages b with relatively constant superhump periods apparently lasted for @xmath950 cycles , then the periods switched to slightly shorter ones . as in other su uma - type dwarf novae , this change in period can be attributed to a stage b c transition . after this transition ,
the period gradually increased for approximately @xmath81 , then suddenly decreased .
since the periods after @xmath10 = 150 were similar to the periods seen in the early stage c , we interpreted that there were temporary ( @xmath81 ) excursions to more positive @xmath5 in addition to the basic period of stage c. we therefore adopted mean periods of entire stage c in table [ tab : perlist ] .
after this stage b
c transition , the secondary maxima of superhumps became stronger , and this appearance of secondary maxima may be somehow related to the origin of stage b
c transition .
the periods derived from times of secondary maxima were initially close to superhump periods of stage b , despite that the periods of primary superhumps had already decreased after stage b c transitions . due to the difference in periods ,
the primary and secondary maxima of superhumps approached and became indistinguishable approximately after @xmath10 = 160170 .
the period then suddenly shortened .
the interference between these two components of superhump maxima appeared to cause temporary excursions of primary maxima toward larger @xmath5 .
the merged two components of maxima then formed single - humped superhumps which lasted for two cycles of normal outbursts in 2009 august , and at least for one cycle in 2009 november december .
although individual times of maxima became inapparent , the signal can be traced in phase - averaged light curves for one more cycle of normal outbursts .
the origin of the secondary maxima is not evident .
since the periods of the secondary component of maxima were longer than the periods of the primary component , the precession rate should be larger .
considering that the periods of the secondary component are close to those of stage b superhumps ( primary maxima ) , it would be possible to assume that the secondary component was formed in the outermost region of the accretion disk , which had already cooled , and the hot region of the accretion disk had shrunken to a smaller radius , resulting shorter superhump periods ( primary maxima ) .
this picture would naturally explain why the secondary component was not apparent before the period of superhumps ( primary maxima ) shortened , i.e. stage b
c transition .
the mass - transfer from the secondary might be responsible for producing a light source in the outermost region of the disk ( cf .
@xcite ) , although it will be premature to directly associate the source with an ordinary stream - impact hot spot .
the behavior of amplitudes ( middle panels in figures [ fig : v344lyrhumpall ] and [ fig : v344lyrhumpall2 ] ) was similar to those of persisting superhumps in sdss j0804 and wz sge ( subsection [ obj : j0804 ] ) , and the luminosity from this source can be estimated to be approximately constant .
it would be noteworthy that secondary maxima were temporarily present in v1504 cyg , another high mass - transfer ( @xmath82 ) system , but did not evolve further as in v344 lyr .
the behavior in v1504 cyg seems to be difficult to explain by a simple model based on stream - impact hot spot . as a possibility alternative to the emergence of stream - impact hot spot , we propose the ( 1,3)-mode wave ( the @xmath83 mode designation follows @xcite ) , which is a one - armed wave traveling with an angular velocity third times larger than that of the secondary .
this wave meets the tidal stress of the secondary twice in a single binary rotation , and might produce observable double - wave modulations .
@xcite has shown that this ( 1,3)-mode wave tends to grow and persist at later epochs .
it is known that short-@xmath18 systems with distinct stage b - c transitions tend to show low - amplitude double - wave modulations prior to the stage b - c transition ( e.g. @xcite ; @xcite ; @xcite ) .
such a phenomenon would be understood if the ( 1,0)-mode wave , the main superhump component , weakens and the other low - amplitudes modes become observable .
these objects are likely low-@xmath82 systems and it would be more difficult to attribute the secondary maxima to stream - impact hot spot than in v344 lyr . the interpretation based on
the ( 1,3)-mode wave might have a potentiality in explaining the double - wave superhump modulations in wide range of objects , and should be further investigated .
( 88mm,110mm)fig79.eps ( 88mm,110mm)fig80.eps after selecting quiescent segments without strong negative superhumps , we searched a signal of the orbital period with a help of estimated orbital period ( 0.08770.0880 d ) from superhump periods ( using the relation of @xcite ) .
we have found a candidate period 0.087903(2 ) d with a mean amplitude of 0.026 mag ( figure [ fig : v344lyrquipdm ] ) . assuming that this is the orbital period , the fractional excesses for positive ( stage b ) and negative superhumps are 4.2 % and @xmath122.1 % , respectively .
( 88mm,110mm)fig81.eps kepler obtained highest precision , uninterrupted observations of two su uma - type dwarf novae .
some of the features , including substructures in stage a in v1504 cyg and complex behavior in late stages in v344 lyr , were new and were outside descriptions derived from our large set of ground - based samples .
apart from them , the @xmath5 behavior of kepler observations generally fits the descriptions obtained from ground - based samples .
the two systems observed by kepler were either long-@xmath18 or high-@xmath82 , and neither of them were short-@xmath18 and low-@xmath82 systems which show distinct stage a c and long - persisting superhumps .
since there were not sufficient ground - based observations for long-@xmath18 or high-@xmath82 systems particularly after the termination of superoutbursts , it is not clear whether persisting superhumps in v1504 cyg and v344 lyr have the origin common to persistent stage c superhumps in short-@xmath18 and low-@xmath82 systems .
future key kepler observations should include these short-@xmath18 and low-@xmath82 systems . in
well - observed wz sge - type dwarf novae , late - stage superhumps having periods longer than those observed during the main superoutburst have been reported .
although there was already a suggestion in eg cnc @xcite , @xcite established the prevalence of such superhumps in gw lib , v455 and and wz sge . @xcite further reported further detections in fl psc ( = asas [email protected] ) , asas [email protected] , sdss j0804 and ot j074727.6@xmath0065050 . in ot j213806.6@xmath0261957 ( hereafter ot j2138 ) , however , a slightly shorter period was detected @xcite . in this paper , we further detected long - period late - stage superhumps in a new superoutburst of sdss j0804 and in sdss j1339 .
since the variation in sdss j1339 looks somewhat unique , we first examine the phenomenon in this object , and present a survey of the existing data of wz sge - type dwarf novae and similar systems for detecting weaker periodicities other than the reported late - stage superhumps .
the late - stage superhumps in sdss j1339 were unique in that they apparently showed a phase reversal , as in traditional " late superhumps ( @xcite ; @xcite ) , despite the fact that most of late - stage superhumps ( stage c superhumps ) in ordinary su uma - type dwarf novae do not show a phase reversal @xcite .
the overall tendency of the @xmath5 diagram , however , resembles those of gw lib @xcite and sdss j0804 ( subsection [ obj : j0804 ] ) , and the @xmath5 variation could be understood without a phase jump if the phase of rapid period decrease immediately following the end of stage b was not observed due to a gap in observation . in interpreting the structure of these @xmath5 diagrams in @xcite
, we did not assume a phase reversal because the phases of superhumps were observationally continuous .
since this continuity may have been superficial and there could have been multiple components or periods as in v344 lyr diagram in er uma was presented in @xcite . ] , we made an alternative analysis following the method used in v344 lyr .
since there were sometimes small secondary superhump maxima during the plateau phase , we also measured the times of these secondary maxima ( the first eight epochs in table [ tab : j1339oc2011late ] ) .
as seen in figure [ fig : j1339humpall ] , these maxima do not seem to be on a smooth extension of late - stage superhumps unless we assume a discontinuous period change .
the situation appears to be different from the case of v344 lyr .
the long period of persisting superhumps , however , is analogous to that of persistent superhumps in v344 lyr , and these superhumps are expected to arise from the outermost region of the accretion disk . as in the case of kepler observations of v344 lyr , it would be controversial whether these late - stage superhumps are traditional " late - superhumps arising from the stream impact point or superhumps arising from varying tidal dissipation .
being a wz sge - type dwarf nova , the system parameters , including the mass - transfer rate , are expected to be different from v344 lyr , and the light from the stream impact point would be expected to be much less in v344 lyr . although one might assume an irradiation - induced enhanced mass - transfer , the lack of post - outburst rebrightening suggests that the mass - transfer rate was not sufficient to produce a normal outburst in contrast to v344 lyr
. the situation may be closer to low-@xmath82 systems described in subsection [ obj : kepv344lyr ] .
we might speculate that the ( 1,0)-mode wave temporarily vanished as the cooling front passed the disk , and the other weaker modes remained and coupled with others to excite the ( 1,0)-mode , which was by chance formed in the opposite direction to the original ( 1,0)-mode .
the presence of the beat phenomenon might also provide a clue to the origin of late - stage superhumps . in sdss j1339 ,
the singly peaked orbital signal was clearly present during the post - superoutburst stage .
the actually observed light curve was not explained by a simple addition of the orbital modulation to the superhumps with constant amplitudes , and the amplitudes ( subtracted for orbital modulations ) of the superhumps cyclically varied with a period of the beat period . as a working hypothesis
, we propose that these cyclic variations were produced by the cyclically variable eccentricity of the accretion disk precessing with the beat period ( cf .
@xcite ) : when the disk becomes more circular , the amplitudes become smaller , and when the disk becomes more eccentric , the amplitudes becomes larger .
this explanation would apply either to traditional " late - superhumps - type explanation , assuming the varying release of the gravitational energy , and or to the cyclic variation of the superhump light source itself from varying tidal dissipation .
we have analyzed post - superoutburst observations of gw lib @xcite .
although it was less striking than in sdss j1339 , gw lib also showed the orbital signal with an amplitude of 0.045 mag and a period of 0.053310(2 ) d ( figure [ fig : gwlibpdmlate ] ) , which is in agreement with the radial - velocity study [ 0.05332(2 ) d , @xcite ] . a beat phenomenon in the superhump amplitudes
was also recorded with a smaller degree than in sdss j1339 .
the detection of orbital signal is surprising since the inclination of the system is suggested to be very low ( 11@xmath84 , @xcite ) , and none of quiescent observations detected orbital modulations ( @xcite ; @xcite ; @xcite ) .
it appears that gw lib and sdss j1339 are very similar in their outburst behavior , evolution of superhumps and the post - superoutburst superhumps and beat phenomenon .
( 88mm,110mm)fig82.eps an analysis of post - superoutburst observations of fl psc @xcite yielded a strong periodicity at 0.054210(8 ) d with a mean amplitude of 0.08 mag ( figure [ fig : flpscpdmlate ] ) .
the period is too short for an orbital period ( 5 % shorter than the superhump period during the main outburst ) , and is most likely identified as negative superhumps .
this is the first indication of negative superhumps during the post - superoutburst stage of wz sge - type dwarf novae .
as discussed in subsection [ obj : kepv1504cygtrans ] , the state with negative superhumps is considered to have a tendency to suppress the disk instability to occur .
the appearance of negative superhumps might explain the reason why multiple rebrightenings did not occur .
there has been a suggestion that there is a minimum mass - transfer rate to generate a tilt , which has been proposed as the mechanism for negative superhumps @xcite .
although the quiescent mass - transfer rates in wz sge - type dwarf novae generally are lower than this limit , the mass - transfer rate may have exceeded this limit in outburst and generated a tilt and negative superhumps .
the generation of tilts in wz sge - type dwarf novae may affect the dynamics of superoutbursts than had been supposed and requires further investigation .
further systematic , sufficiently long , observations during the post - superoutburst states of wz sge - type or su uma - type dwarf novae are needed .
we also propose a candidate orbital period of 0.056096(5 ) d , with a mean amplitude of 0.06 mag .
this candidate period is well in agreement with the predicted orbital period of 0.05609 d from stage b superhumps ( subsection [ sec : pshporb ] ) . assuming this orbital period , fractional superhump excesses for positive ( stage b ) and negative superhumps are 1.8 % and @xmath123.4 % , respectively .
it is known that fractional excesses ( in absolute values ) for positive superhumps are approximately twice larger in long-@xmath18 systems ( cf .
@xcite ; table 2 of @xcite ) , and the larger fractional excess for negative superhumps in fl psc is unusual .
it has been established , however , the fractional excesses ( in absolute values ) for positive and negative superhumps in er uma are almost equal @xcite , and there was a suggestion of a large negative fractional excess for the supposed negative superhumps in v1159 ori @xcite .
the negative superhumps in very short-@xmath18 systems may have different properties from those in traditional negative superhumps in long-@xmath18 systems .
( 88mm,110mm)fig83.eps an analysis of the post - superoutburst observations of ot j2138 ( 2010 , bjd after 2455348 ) yielded two periodicities ( figure [ fig : j2138pdmlate ] ) : persistent superhumps with a period of 0.054853(6 ) d and an amplitude of 0.11 mag , the orbital period of 0.054523(4 ) d and an amplitude of 0.05 mag .
the other signals were one - day or two - day aliases of the superhumps .
the orbital period determined from late - stage observations is consistent with the period of early superhumps .
the fractional excess of stage b superhumps has been refined to be 0.91 % .
( 88mm,110mm)fig84.eps the origin of orbital signals in these systems during the post - superoutburst stage is unclear , although they were detected in several wz sge - type dwarf novae .
although it might have originated from the irradiated secondary by the heated white dwarf , the low inclination in gw lib would make this interpretation difficult .
the assymmetric profile in ot j2138 resembles an combination of a shallow eclipse and a orbital hump . although the amplitudes were too small to be regarded as true
eclipses , geometrical effect can be responsible for systems with intermediate inclinations .
@xcite recently raised a question against the widely believed descriptions about amplitudes of superhumps .
@xcite summarized as follows : 1 .
superhumps first appear around superoutburst maximum and reach their highest amplitude either at maximum or 1 - 2 days later .
2 . superhump amplitudes decrease during superoutburst ; this effect , however , has been well documented only for relatively few cases .
3 . at their full development
the superhumps have a range of 0.3 - 0.4 mag , and are equally prominent in all su uma stars independent of inclination ( @xcite , p.194 ) .
4 . there is no strong modulation of the superhump profile at the beat period ( @xcite , p.372 ) .
the claim by @xcite was that the points 3 and 4 are wrong .
these observational findings were , however , derived from observations in the 1970 s and 1980 s , when most of the knowledges in superhumps came from regularly outbursting bright southern su uma - type dwarf novae , i.e. relatively long-@xmath18 objects with higher mass - transfer rates . since the modern accumulated data cover much wider range of objects , we first present statistical descriptions of superhump amplitudes , and examine the items individually .
note that we already presented a typical variation of superhump amplitudes in figure 3 of @xcite . in this paper
, we present a statistical analysis and treat only a pilot study of representative objects .
we are going to devote ourselves in a separate work to more comprehensive survey of the amplitudes of superhumps as well as detailed data on individual objects .
the samples are observations in @xcite , @xcite and this paper .
we restricted analysis to superoutbursts in which the starts of stage b were determined .
additional criterion for choosing samples was that more than 20 epochs of superhumps maxima were measured .
the resultant samples are listed in table [ tab : ampoutlist ] .
the amplitudes of superhumps were measured using the template fitting method described in @xcite , and the derived amplitudes are not literally full minimum - to - maximum amplitudes in usual sense but more reflect a kind of first - order moment .
it has become evident that some ( less than 1 % ) of recorded maxima tabulated in @xcite were inadequate for measuring amplitudes because they were only observed around superhump maxima .
we rejected these maxima in the analysis .
lcclcc object & year & @xmath2 & object & year & @xmath2 + v455 and & 2007 & 0.0571 & wz sge & 2001 & 0.0572 + v466 and & 2008 & 0.0572 & v551 sgr & 2003 & 0.0676 + dh aql & 2002 & 0.0800 & v1212 tau & 2011 & 0.0701 + vy aqr & 2008 & 0.0647 & fl tra & 2005 & 0.0599 + eg aqr & 2006 & 0.0790 & su uma & 1999 & 0.0791 + bg ari & 2010 & 0.0849 & sw uma & 2000 & 0.0583 + tt boo & 2004 & 0.0781 & sw uma & 2006 & 0.0582 + nn cam & 2009 & 0.0743 & sw uma & 2010 & 0.0582 + ht cas & 2010 & 0.0763 & bc uma & 2003 & 0.0646 + v1040 cen & 2002 & 0.0622 & bz uma & 2007 & 0.0702 + wx cet & 1998 & 0.0595 & dv uma & 1997 & 0.0888 + go com & 2003 & 0.0631 & dv uma & 2007 & 0.0885 + v632 cyg & 2008 & 0.0658 & iy uma & 2000 & 0.0758 + v1028 cyg & 1995 & 0.0617 & iy uma & 2006 & 0.0761 + v1504 cyg & 2009b & 0.0722 & iy uma & 2009 & 0.0762 + ho del & 2008 & 0.0644 & ks uma & 2003 & 0.0702 + kv dra & 2009 & 0.0601 & hv vir & 2002 & 0.0583 + xz eri & 2007 & 0.0628 & hv vir & 2008 & 0.0583 + xz eri & 2008 & 0.0628 & 1rxs j0423 & 2008 & 0.0784 + uv gem & 2003 & 0.0935 & 1rxs j0423 & 2010 & 0.0785 + v592 her & 2010 & 0.0566 & asas j1025 & 2006 & 0.0634 + v1108 her & 2004 & 0.0575 & asas j1536 & 2004 & 0.0646 + ru hor & 2008 & 0.0710 & asas j1600 & 2005 & 0.0650 + mm hya & 2011 & 0.0589 & mis v1443 & 2011 & 0.0567 + rz leo & 2000 & 0.0787 & sdss j0804 & 2010 &
0.0596 + gw lib & 2007 & 0.0541 & sdss j0812 & 2011 & 0.0779 + v344 lyr & 2009 & 0.0916 & sdss j1146 & 2011 & 0.0633 + v344 lyr & 2009b & 0.0916 & sdss j1227 & 2007 & 0.0646 + v585 lyr & 2003 & 0.0604 & sdss j1339 & 2011 & 0.0581 + dt oct & 2003 & 0.0748 & sdss j1627 & 2008 & 0.1097 + v2527 oph & 2004 & 0.0721 & ot j0238 & 2008 & 0.0537 + uv per & 2000 & 0.0666 & ot j1443 & 2009 & 0.0722 + uv per & 2003 & 0.0667 & ot j1610 & 2009 & 0.0578 + uv per & 2007 & 0.0663 & ot j1625 & 2010 & 0.0961 + qy per & 1999 & 0.0786 & ot j2138 & 2010 & 0.0550 + + + + + + + the variation of superhump amplitudes in non - eclipsing systems are shown in figures [ fig : humpamp1][fig : humpamp4 ] .
the superhump amplitudes are plotted against cycle numbers after the start of stage b. since the starts of stage b were measured from @xmath5 variations ( shown in tables in @xcite , @xcite and in this paper ) , they can be different from the maxima of superhump amplitudes .
there is also a tendency of 1-d periodicities in the density of data , resulting from uneven distribution of observations against longitudes .
there should have been typically this degree ( 1015 cycles ) of uncertainty in determining the starts of stage b. since the quality of the data were sometimes low , there are sometimes a number of scattered points in large - amplitude regions .
they should be better considered as a result of a selection bias when the errors in the original data were large .
although there must have equally been scattered points in low - amplitude regions , they were not detected either due to poor convergence in numerical fitting or because the resultant amplitudes were negative .
nevertheless , the majority of the measured data are in relatively narrow strips in each figure , and we can regard these strips as typical behavior of amplitude variation .
the large increase in the amplitude at late stages ( e.g. @xmath85 in figure [ fig : humpamp1 ] ) corresponds to the final fading of superoutbursts .
several features can be seen in these figures : 1 .
the amplitudes of superhumps usually reach a maximum around the stage a b transition .
this means that the time before the system reaches the peak of superhump amplitude is variable between objects or between superoutbursts .
although the amplitudes of superhumps globally decrease with time , secondary peaks of amplitudes are usually seen @xmath950 cycles after the main peak in long-@xmath18 systems ( figure [ fig : humpamp1 ] ) and @xmath97080 cycles after the main peak in intermediate- or short-@xmath18 systems ( figures [ fig : humpamp2 ] , [ fig : humpamp3 ] ) . in short-@xmath18 systems , the secondary peaks become less evident ( figure [ fig : humpamp4 ] ) .
the epochs of the secondary peaks are close to the epochs of stage b c transitions . in some cases , the secondary peaks appear later than stage b c transitions . 3 .
the amplitude of superhumps are well - correlated with @xmath18 : longer-@xmath18 systems show larger amplitudes .
the relation for maximum amplitudes of superhumps is shown in figure [ fig : humpampporb ] .
figures for other stages of superoutburst basically show the same trend .
warner s point 3 ( 0.30.4 mag for maximum amplitudes ) apparently came from the long-@xmath18 samples in the older times .
the maximum amplitudes seem to decrease for longer-@xmath18 ( @xmath86 d ) objects .
two systems ( rz leo and qy per ) show systematically larger amplitudes of superhumps .
these systems have long recurrence times similar to wz sge - type dwarf novae , but with longer @xmath18 . 5 . the amplitudes of superhumps in extreme wz sge - type ( we refer to systems with recurrence times longer than @xmath910 yr or systems with multiple rebrightenings ) are generally smaller than in ordinary su uma - type dwarf novae .
( 88mm,70mm)fig85.eps ( 88mm,70mm)fig86.eps ( 88mm,70mm)fig87.eps ( 88mm,70mm)fig88.eps ( 88mm,70mm)fig89.eps we further studied the dependence of superhump amplitudes on orbital inclination ( @xmath87 ) . as in subsection [ sec : shampgeneral ] , we only used epochs @xmath88 to illustrate the maximum superhump amplitudes . due to the strong dependence of superhump amplitudes on @xmath18 , we first removed this effect by using a relation derived in subsection [ sec : shampgeneral ] ( curve in figure [ fig : humpampporb ] ) .
the normalization was taken so that the average of amplitudes agrees with the pre - normalized value of the sample ( 0.25 mag ) .
we used @xmath87 values in the on - line version 7.15 of @xcite .
as seen in figure [ fig : humpampinc ] , the normalized maximum amplitudes do not strongly depend on @xmath87 for systems with @xmath89 , including an eclipsing system wz sge ( @xmath90 ) .
the right panel of the figure shows the amplitudes for non - eclipsing su uma - type dwarf novae without known inclinations .
the 90 % of the normalized amplitudes without known inclinations fall in the range of 0.140.35 mag , well in agreement with @xmath89 samples .
this result seems to confirm that warner s point 3 is basically valid for non - eclipsing systems .
it is worth noting that there is no indication of smaller amplitudes in systems with very low inclinations .
( 88mm,70mm)fig90.eps as we have already shown , the beat phenomenon in amplitudes of superhumps are clearly present in eclipsing objects ( e.g. ht cas , subsection [ obj : htcas ] and sdss j0804 subsection [ obj : j0804 ] ) . while it is not clear what the term `` profile '' exactly mentioned in warner s point 4 , the present observations seem to support the presence of a beat phenomenon ( at least in amplitudes ) in eclipsing systems . in figure
[ fig : iyuma2009amp ] , we show variations of superhump amplitudes in iy uma ( 2009 ) , which was reported in @xcite . although the beat phenomenon was strongly seen during stage b , it became weaker with time and almost disappeared during stage c. in ht cas ( 2010 , figure [ fig : htcasbeat ] ) , however , the beat phenomenon persisted during its entire plateau phase
. it would be worth noting that double - wave modulation in amplitude mentioned in @xcite ( @xmath91 in his designation ) is not apparent even in the best observed case of ht cas .
( 88mm,70mm)fig91.eps the beat phenomenon is much weaker in sdss j0804 ( 2010 , figure [ fig : j0804beat ] ) than in iy uma ( 2009 ) , and became inapparent during the late plateau phase .
( 88mm,70mm)fig92.eps among non - eclipsing systems , rz leo , likely a high - inclination system ( @xcite ; @xcite ) showed the presence of the beat phenomenon [ modulations with a 30.6(1)-cycle period is already apparent in figure [ fig : humpamp2 ] ] .
the mean amplitude of modulations ( @xmath92 in @xcite ) was 0.07(2 ) mag .
qy per , another outlier in figure [ fig : humpamp2 ] , has a candidate weak beat phenomenon with a period of 28.5(2 ) cycles and @xmath92 of 0.02(1 ) mag . since the signal was weak and the true orbital period is unknown , this needs to be confirmed by further studies .
we could not confirm the presence of the beat phenomenon in tt boo ( 2004 ) . in ks uma ( 2003 )
, we found a weak beat phenomenon with @xmath92 of 0.02(1 ) mag whose beat period is consistent with the known @xmath18 .
the degree of the beat phenomenon during the superoutburst plateau appears to be related to the maximum amplitudes of superhumps , and appears to be smaller in wz sge - type dwarf novae .
the beat phenomenon in non - eclipsing systems , especially in short-@xmath18 systems , was undetected or close to the detection limit .
this finding seems to be consistent with numerical modeling of inclination effect on superhumps @xcite , in which only systems with @xmath87 larger than 4560@xmath1 show significant orbital modulation of superhumps .
although @xcite discussed a new interpretation on the origin of superhumps in combination with his earlier findings ( e.g. @xcite ; @xcite ) , we restrict our discussion within the traditional framework ( cf .
@xcite ; @xcite ; @xcite ) since we do not have sufficient materials to compare with the analysis by earlier works by smak . )
to smak s idea . since the difference in the presence of a hot spot during superoutburst or the irradiated secondary might be dependent on object [ for example , in ht cas , the frequency of outbursts is unusually low , and orbital humps in quiescence is often inapparent ( @xcite ; @xcite ) .
ht cas therefore might not be an ideal object representing the entire su uma - type population ] , and we should leave the problem as a future work .
detailed modeling of the eclipse and superhump light curve assuming a vertically extended disk is indispensable for further discussion , which is clearly outside the scope of this paper . ] in non - eclipsing systems , there is evidence that our normalized amplitudes of superhumps were fairly constant over a wide range of orbital inclinations ( @xmath93).-dependence while @xcite introduced a concept of normalization to the average luminosity of the disk .
this concept by @xcite implicitly assumes that the observed light from the disk suffers from the @xmath87-dependent projection and obscuration effects , while the emission from the superhump light source is isotropic , or observable independent of @xmath87 .
] this can be understood if the superhump light source has almost the same @xmath87-dependence as the background disk light . for geometrically thin , and optically thick accretion disks , the projection effect ( @xmath94 )
plays the strongest role in low - to - intermediate @xmath87 .
this would imply that the @xmath87-dependence of the superhump light source can be explained as well , such as in a form of the ( superhump phase - dependent ) variations of the surface brightness ( or the temperature ) on a geometrically thin disk .
the apparent lack of deviations from this relation up to @xmath95 suggest that the vertical structure in the superhump light source is smaller than @xmath96 in slope against the remaining disk .
this value is not much greater than the flaring angle of 36@xmath1 inferred for a normal outburst of z cha @xcite , and may not require a special mechanism other than additional tidal heating around the superhump light source . in deeply eclipsing systems ,
the amplitudes are systematically higher than the above simple assumption of the projection effect .
although this phenomenon can likely be understood as a consequence of the vertical expansion around the superhump light source , even a simple treatment should require numerical estimation of different contribution from a vertically extended and structured disk , self - obscuration and eclipse by the secondary , and will be an elaborate future work .
the dependence of amplitudes on @xmath18 may be a combined effect of the strength of the tidal torque ( larger in higher-@xmath97 systems ) and the strength of the 3:1 resonance ( which will be expected to be smaller in higher-@xmath97 systems around the border of the instability condition for tidal instabilities ) .
this would explain a maximum of amplitudes around @xmath98 d. although not included in our samples , the low ( @xmath90.100.15 mag ) maximum amplitudes in rz lmi ( @xcite ; @xcite ; @xcite ) would be an outlier for a @xmath2 = 0.0594 d object may be similarly explained by a supposed lower tidal torque than in other su uma - type dwarf novae as proposed in @xcite and @xcite .
there is an indication that amplitudes of superhumps are higher in systems with longer intervals of outbursts ( rz leo and qy per ) among long-@xmath18 systems .
these systems may have stored larger disk mass , either as a result of a lower quiescent viscosity or by other mechanisms , and the superhump light source or the vertical extent of the disk may be more pronounced at the time of the outburst .
the higher vertical extent might also explain the very strong beat phenomenon in ht cas . in extreme wz sge - type dwarf novae ,
the superhump amplitudes tend to be lower and the degree of the beat phenomenon is weaker than in other su uma - type dwarf novae .
this may be explained if the most of the disk mass outside the 3:1 resonance is efficiently accreted during the stage of early superhumps [ the development of the 2:1 resonance is expected to suppress the 3:1 resonance ( @xcite ; @xcite ) ] and there is relatively little amount of mass at the onset of development of ordinary superhumps . following this interpretation
, the maximum amplitudes of superhumps in wz sge - type dwarf novae would be a discriminating feature in estimating the strength of the 2:1 resonance prior to the development of ordinary superhumps .
the relatively weak beat phenomenon in stage c superhumps ( subsection [ sec : shampbeat ] ) would also deserve attention .
the weaker signal of the beat phenomenon would suggest a lower vertical extent for stage c superhumps .
an examination whether the lower temperature of the light source of stage c superhumps or its location , can explain the phenomenon could shed light on clarifying the nature of still poorly understood stage c superhumps .
we studied the characteristics of superhumps for 51 su uma - type dwarf novae whose superoutbursts were mainly observed during the 20102011 season . in addition to the purpose of our previous surveys focusing on systematic variation of superhump periods , we extended our analysis to post - superoutburst variations and a survey of variations of superhump amplitudes for selected objects .
we also analyzed public kepler data for v1504 cyg and v344 lyr .
most of the new data for systems with short superhump periods basically confirmed the earlier findings .
we recorded early superhumps for four wz sge - type dwarf novae ( sdss j0804 , sdss j1339 , ot j0120 , and likely sdss j1605 ) .
we also refined the ephemerides of eclipsing systems ht cas , sdss j0932 , sdss j1227 and ot j0431 , and provided firm estimates of fractional superhump excesses for these objects .
we have also updated statistics in relations @xmath2@xmath18 for stage b and c superhumps , @xmath18@xmath13 and @xmath18@xmath23 for stage b superhumps .
the traditional @xcite relation gives systematically longer @xmath18 estimates for stage b superhumps .
in addition to them , we found : * the spread of period derivative is broader in long-@xmath18 systems .
in particular , gx cas unexpectedly showed a large positive @xmath13 .
* we recorded the long - waited superoutburst of the eclipsing su uma - type dwarf nova ht cas , and described the full evolution of superhumps during its earliest stage to post - superoutburst stage .
a very strong beat phenomenon was recorded particularly when superhumps reached the full amplitudes .
* there was no particularly indication of the reflection effect and conspicuous appearance of the hot spot in ht cas as judged from the light curves outside the eclipses .
these results do not favor an interpretation assuming an irradiation - induced enhanced mass - transfer . * in sdss j0804 , we for the first time recorded full evolution of early superhumps and ordinary superhumps , as well as persisting superhumps during six rebrightenings .
* shallow eclipses were detected during the entire course of outbursts in sdss j0804 .
the beat phenomenon was also recorded with a smaller degree than in ht cas .
* persisting superhumps during the rebrightening phase of sdss j0804 and wz sge ( 2001 ) had amplitudes well - correlated with the brightness of the system . the relation can be understood assuming the almost constant luminosity of the superhump light source .
the luminosity of the light source showed a decrease just before or around the final rebrightening . * the @xmath5 diagrams for three wz sge - type dwarf novae sdss j0804 , sdss j1339 and ot j0120 showed a similar pattern to that of gw lib during the final fading phase .
in sdss j0804 , a rapid excursion to a shorter period was recorded .
this phase was probably missed in sdss j1339 , and the resultant @xmath5 diagrams even looked like a phase reversal as would be expected for `` traditional '' late superhumps . * in sdss j1339 , there was a strong beat phenomenon in the amplitudes of persistent superhumps even during the post - superoutburst phase .
we interpret this feature as a manifestation of cyclically variable shape of the eccentric disk with the beat period . *
the overall behavior in ot j0120 was very similar to that of wz sge ( 2001 ) , consisting of a post - outburst `` dip '' and multiple rebrightenings with short intervals .
* we studied a new superoutburst of the helium dwarf nova v406 hya and examined its past outburst activities .
* kepler data for v1504 cyg and v344 lyr confirmed the presence of stage a and b superhumps , and the details of stage a evolution , as well as relation to the precursor phenomenon , were described . * in v344 lyr , secondary maxima of superhumps appeared during the late stage of superoutbursts , and this signal gradually merged with the primary maxima to form singly - humped superhumps .
the initial periods of secondary maxima were longer than those of primary maxima , suggesting that the secondary maxima were formed in the outer part of the disk .
these superhumps persisted for at least two cycles of subsequent normal outbursts .
we discussed the role of ( 1,3)-mode wave in manifestation of doubly - humped superhumps . * in v1504 cyg ,
the behavior was different fro v344 lyr in that secondary maxima only transiently appeared . * in v1504 cyg , we detected a `` failed superoutburst '' in which a precursor outburst and a subsequent further rise was recorded . during this outburst ,
negative superhumps instead of ordinary ( positive ) superhumps were excited .
the appearance of negative superhumps may be related to premature quenching of the superoutburst .
* sdss j0804 ( 2010 ) showed a large @xmath13 for a wz sge - type dwarf nova with multiple rebrightenings .
this may have been resulted from a rapid increase of the superhump period at the end of the plateau phase .
this phenomenon seems to be commonly seen in eclipsing wz sge - type dwarf novae .
* we have detected likely negative superhumps during the post - superoutburst stage of fl psc in 2004 , first times recorded among wz sge - type dwarf novae .
we suggested a possibility that the state with negative superhumps ( possibly a state with a tilted disk ) can affect the development of the superoutburst itself , particularly the presence of rebrightenings .
* we have detected orbital modulations in a low - inclination wz sge - type dwarf nova gw lib during its post - superoutburst in 2007 .
we have also identified the orbital period from post - superoutburst observations of ot j2138 in 2010 .
* the amplitudes of superhumps usually show complex variations .
the amplitudes quickly grow during stage a and reach a maximum around the stage a b transition .
although the amplitude decay approximately linearly during stage b , many objects show significant regrowth of amplitudes around or after the transition to stage c. * the amplitudes of superhumps are strongly related to the orbital period .
long - period systems have larger amplitudes . * after correcting for the dependence on the orbital period , there is little indication of dependence on the system inclination for @xmath89 .
* in eclipsing systems , the amplitudes of superhumps are larger , and strongly modulated with the beat period . * the beat phenomenon of superhumps in eclipsing systems decrease with the progress of the superoutburst .
the beat phenomenon is significantly reduced in stage c superhumps . * in wz sge - type eclipsing systems , the beat phenomenon is not as striking as ordinary su uma - type dwarf novae .
* we interpret that the superhumps in non - eclipsing systems can be approximated by varying surface brightness of the superhump light source on a relatively geometrically thin disk .
we interpret the behaviors in eclipsing systems as being a result of vertical extent of the superhump light source .
the dependence on the system inclinations suggests that the slope of the vertical extension does not exceed 10@xmath1 .
this work was supported by the grant - in - aid for the global coe program
the next generation of physics , spun from universality and emergence " from the ministry of education , culture , sports , science and technology ( mext ) of japan .
the authors are grateful to observers of vsnet collaboration and vsolj observers who supplied vital data .
we acknowledge with thanks the variable star observations from the aavso international database contributed by observers worldwide and used in this research . we have been benefited by discussions with makoto uemura , akira imada and daisaku nogami .
this work is deeply indebted to outburst detections and announcement by a number of variable star observers worldwide , including participants of cvnet , baa vss alert and avson networks .
we are grateful to the catalina real - time transient survey team for making their real - time detection of transient objects available to the public .
we thank the kepler mission team and the data calibration engineers for making kepler data available to the public .
the following objects have been named in @xcite during the proofreading period : asas [email protected] = dt pyx , asas [email protected] = qz lib , sdss [email protected] = ez lyn , sdss [email protected] = v355 uma , ot j074727.6@xmath0065050 = dy cmi . | continuing the project described by @xcite , we collected times of superhump maxima for 51 su uma - type dwarf novae mainly observed during the 20102011 season .
although most of the new data for systems with short superhump periods basically confirmed the findings by @xcite and @xcite , the long - period system gx cas showed an exceptionally large positive period derivative .
an analysis of public kepler data of v344 lyr and v1504 cyg yielded less striking stage transitions . in v344 lyr , there was prominent secondary component growing during the late stage of superoutbursts , and the component persisted at least for two more cycles of successive normal outbursts .
we also investigated the superoutbursts of two conspicuous eclipsing objects : ht cas and the wz sge - type object sdss [email protected] .
strong beat phenomena were detected in both objects , and late - stage superhumps in the latter object had an almost constant luminosity during the repeated rebrightenings .
the wz sge - type object sdss [email protected] showed a phase reversal around the rapid fading from the superoutburst .
the object showed a prominent beat phenomenon even after the end of the superoutburst .
a pilot study of superhump amplitudes indicated that the amplitudes of superhumps are strongly correlated with orbital periods , and the dependence on the inclination is weak in systems with inclinations smaller than 80@xmath1 . |
recently , we have considered the problem of the low energy electronic states in spheroidal fullerenes @xcite .
the main findings were a discovery of fine structure with a specific shift of the electronic levels upwards due to spheroidal deformation .
in addition , three twofold degenerate modes near the fermi level with one of them being the true zero mode were found .
an interesting question is the modification of this structure under the influence of a uniform magnetic field .
the problem of zeeman splitting and landau quantization of electrons on a sphere was studied in ref .
@xcite . to this end ,
the schrdinger equation for a free electron on the surface of a sphere in a uniform magnetic field was formulated and solved . in this paper , we explore the field - theory model suggested in ref .
@xcite , which describes the electronic states near the fermi energy and takes into account the specific structure of carbon lattice , geometry , and the topological defects ( pentagons ) .
the euler s theorem for graphene requires the presence of twelve pentagons to get the closed molecule . in the framework of continuum description
we extend the dirac operator by introducing the dirac monopole field inside the spheroid to simulate the elastic vortices due to twelve pentagonal defects . the k spin flux which describes
the exchange of two different dirac spinors in the presence of a conical singularity is included in a form of thooft - polyakov monopole .
our studies cover slightly elliptically deformed molecules in the weak uniform external magnetic field .
let us start with writing down the dirac operator for free massless fermions on the riemannian spheroid @xmath0 .
the dirac equation on a surface @xmath1 in the presence of the abelian magnetic monopole field @xmath2 and the external magnetic field @xmath3 is written as @xcite @xmath4\psi = e\psi , \label{eq:1}\ ] ] where @xmath5 is the zweibein , @xmath6 is the metric , the orthonormal frame indices @xmath7 , the coordinate indices @xmath8 , and @xmath9 with @xmath10 \label{eq:2},\ ] ] being the spin connection term in the spinor representation ( see @xcite for details ) . the energy in ( [ eq:1 ] ) is measured from the fermi level .
let us start from the analysis of an electron state at the fermi level ( so - called zero - energy mode ) .
we will use the projection coordinates in the form @xmath11 where @xmath12 and @xmath13 are the spheroidal axles .
the riemannian connection with respect to the orthonormal frame is written as @xcite @xmath14 the only nonzero component of the gauge field @xmath2 in region @xmath15 for spheroidal fullerenes reads ( see ref .
@xcite ) @xmath16 where @xmath17 notice that the monopole field @xmath2 in eq .
( [ eq:5 ] ) consists of two parts .
the first one comes from the k - spin connection term and implies the charge @xmath18 while the second one is due to elastic flow through a surface .
this contribution is topological in its origin and characterized by charge @xmath19 . for total elastic flux from twelve pentagonal defects one has @xmath20 .
the parameter @xmath21 is introduced in eq.([eq:5 ] ) .
the uniform external magnetic field @xmath22 is chosen to be pointed in the @xmath23 direction so that @xmath24 . in projection
coordinates the only nonzero component of @xmath25 is written as @xmath26 let us study eq.([eq:1 ] ) for the electronic states at the fermi energy ( @xmath27 ) .
the dirac matrices can be chosen to be the pauli matrices , @xmath28 . by using the substitution @xmath29
we obtain @xmath30 @xmath31 where @xmath32 .
we assume that the eccentricity of the spheroid is small enough . in this case , one can write down @xmath33 and @xmath34 ( @xmath35 ) is a small dimensionless parameter characterizing the spheroidal deformation . therefore , one can follow the perturbation scheme using @xmath34 as the perturbation parameter . to the leading in @xmath34 approximation eqs .
( [ eq:11 ] ) and ( [ eq:12 ] ) are written as @xmath36}[j - m(1+\delta)\frac{r^2-r^2}{r^2+r^2}+ m\delta\frac{(r^2-r^2)^{3}}{(r^2+r^2)^{3}}]\ ] ] @xmath37}+\frac{r^2-r^2}{2r(r^2+r^2)})v_{j}(r ) , \label{eq:11}\ ] ] @xmath38}[j - m(1+\delta)\frac{r^2-r^2}{r^2+r^2 } + m\delta\frac{(r^2-r^2)^{3}}{(r^2+r^2)^{3}}]\ ] ] @xmath39}+\frac{r^2-r^2}{2r(r^2+r^2)})u_{j}(r ) .
\label{eq:12}\ ] ] the exact solution to eqs .
( [ eq:11 ] ) and ( [ eq:12 ] ) is found to be @xmath40 @xmath41 where @xmath42 , @xmath43 .
the function @xmath44 can be normalized if the condition @xmath45 is fulfilled . in turn
, the normalization condition for @xmath46 has the form @xmath47 . finally , there are five normalized solutions for @xmath44 with @xmath48 when @xmath49 , and one solution for @xmath46 with @xmath50 when @xmath51 . in the limit @xmath52
we arrive at the solution for spherical fullerenes @xmath53 since normalization conditions do not depend on the parameter @xmath34 they are the same as for spherical case .
thus the perturbation does not change the number of zero modes .
notice that eq .
( [ eq:15 ] ) can be obtained from eqs .
( [ eq:13 ] ) and ( [ eq:14 ] ) by using @xmath54 ^{a}=e^{x } , a\rightarrow\infty$ ] .
let us study now the structure of electron levels near the fermi energy .
it is convenient to consider this problem by using of the cartesian coordinates @xmath55 , @xmath56 , @xmath23 in the form @xmath57 the riemannian connection reads ( cf .
( [ eq:4 ] ) ) @xmath58 within the framework of the perturbation scheme the spin connection coefficients are written as @xmath59 where the terms to first order in @xmath34 are taken into account . in spheroidal
coordinates , the only nonzero component of @xmath2 in region @xmath15 is found to be ( see ref .
@xcite ) @xmath60 and the external magnetic field reads @xmath61 by using the substitution ( [ eq:8 ] ) we obtain the dirac equation for functions @xmath62 and @xmath63 in the form @xmath64 where @xmath65 a convenient way to study the eigenvalue problem is to square eq . ( [ eq:21 ] ) . for this purpose ,
let us write the dirac operator in eq .
( [ eq:21 ] ) as @xmath66 .
one can easily obtain that ( similar way as in ref .
@xcite ) @xmath67 notice that to first order in @xmath34 the square of the dirac operator is written as @xmath68 where @xmath69 in an explicit form @xmath70 @xmath71 where the appropriate substitution @xmath72 is used .
the equation @xmath73 takes the form @xmath74\left(% \begin{array}{c } u_j(x ) \\
\end{array}% \right)\nonumber\ ] ] @xmath75 where @xmath76 .
since we consider the case of a weak magnetic field the terms with @xmath77 and @xmath78 in eq .
( [ eq:25 ] ) can be neglected . as a result
, the energy spectrum for spheroidal fullerenes is found to be @xmath79 where @xmath80 the function of spheroidal deformations @xmath81 is specified in ref .
@xcite . finally , in the linear in @xmath34 approximation , the low energy electronic spectrum of spheroidal fullerenes takes the form @xmath82 with @xmath83 where @xmath84 and @xmath85 for @xmath86 , respectively .
here we came back to the energy variable @xmath87 ( in units of @xmath88 where @xmath89 is the fermi velocity ) . as is seen from eq .
( [ eq:29 ] ) , for @xmath90 the spheroidal deformation gives rise to an appearance of a additional structure .
as an example , table 1 shows all five contributions ( in compliance with eq .
( [ eq:29 ] ) ) to the first and second energy levels for yo - c@xmath91 ( yo means a structure given in @xcite ) . as is seen ,
the energy levels become shifted due to spheroidal deformation .
the uniform magnetic field provides the well - known zeeman splitting . the difference between topological and zeeman splitting
is clearly seen . in the second case , the splitted levels
are shifted in opposite directions while for topological splitting the shift is always positive . as an illustration , fig .
1 schematically shows the structure of the second level . in this case , the initial ( for @xmath92 ) degeneracy of @xmath93 is equal to six .
the spheroidal deformation provokes an appearance of three shifted double degenerate levels ( fine structure ) .
the magnetic field is responsible to zeeman splitting .
we admit that the obtained values of the splitting energies can deviate from the estimations within some more precise microscopic lattice models and density - functional methods ( see , e.g.@xcite ) . in our paper
, we focused mostly on the very existence of physically interesting effects .
for this reason , the values in both table and the schematic picture are presented with accuracy at about one percent of @xmath94 .
in conclusion , we have considered the structure of low energy electronic states of spheroidal fullerenes in the weak uniform magnetic field provided the spheroidal deformation from the sphere is small enough . for the states at the fermi level
, we found an exact solution for the wave functions .
it is shown that the external magnetic field modifies the density of electronic states and does not change the number of zero modes . for non - zero energy modes , electronic states near the fermi energy of spheroidal fullerene
are found to be splitted in the presence of a weak uniform magnetic field .
it should be mentioned that the zero - energy states in our model corresponds to the homo ( highest occupied molecular orbital ) states in the calculations based on the local - density approximation in the density - functional theory ( see , e.g. @xcite ) . in particular , the homo - lumo energy gap is found to be about @xmath95 ev for yo - c@xmath91 fullerene within our model ( here lumo means the lowest unoccupied molecular orbital ) . 0.2 cm 0.2 cm the work was supported in part by vega grant 2/7056/27 . of the slovak academy of sciences , by the science and technology assistance agency under contract
apvt-51 - 027904 and by the russian foundation for basic research under grant no .
05 - 02 - 17721 .
m. pudlak , r. pincak and v.a .
osipov , phys .
b * 74 * ( 2006 ) 235435 . h. aoki and h. suezawa , phys .
a * 46 * ( 1992 ) r1163 .
n.d . birrell and p.c.w .
davies , _ quantum fields in curved space _ , ( cambridge 1982 ) .
m. nakahara , _ geometry , topology and physics _ ,
( institute of physics publishing bristol 1998 ) .
m.gckeler and t.schcker , _ differential geometry , gauge theories , and gravity _
( cambridge university press 1989 ) .
d.v . kolesnikov and v.a .
osipov , eur.phys.journ .
b * 49 * ( 2006 ) 465 . m. yoshida and e. osawa , fullerene sci . tech . * 1 * ( 1993 ) 55 .
j. p. lu and w. yang , phys .
b * 49 * ( 1994 ) 11421 . r.a .
broglia , g. col , g. onida , h.e .
roman , _ solid state physics of finite systems _ , ( springer 2004 ) .
s. saito and a. oshiyama , phys .
* 66 * ( 1991 ) 2637 .
.the structure of the first and higher energy levels for yo - c@xmath91 fullerene in uniform magnetic field .
the hopping integral and other parameters are taken to be @xmath96 and @xmath97 , @xmath98 , @xmath99 , @xmath100 , @xmath101 .
@xmath102 is average bond length , @xmath103 is average radius , @xmath104 is standard deviation from a perfect sphere ( see refs .
[ 7,8 ] ) , so that @xmath105 . [ cols="<,^,^,^,^ " , ] | the effect of a weak uniform magnetic field on the electronic structure of slightly deformed fullerene molecules is studied within the continuum field - theory model .
it is shown how the existing due to spheroidal deformation fine structure of the electronic energy spectrum modifies in the presence of the magnetic field .
exact analytical solutions for zero - energy modes are found . 0.1 cm 0.1 cm |
the concept of _ webs _ was invented by w. blaschke and connected with many parts of geometry @xcite .
let us recall , that a _ trivial @xmath0-web _ in a planar domain @xmath1 are three families of smooth curves such that there is a diffeomorphism @xmath2 taking the families to sets of lines parallel to the sides of a fixed triangle .
see ( * ? ? ?
* lecture 18 ) as an introduction to the topic .
there are several non - trivial @xmath0-webs formed by line- and circle - families , see @xcite for history of the problem and new examples of such webs .
in this paper we consider webs in the positive quadrant @xmath3 formed by confocal conics and pencils of _ apollonian circles _
, which are defined in the following way : let @xmath4 and @xmath5 be two points in the plane called _
foci_. the family of circles passing through @xmath4 and @xmath5 is called an _ elliptic apollonian pencil_. the pencil of circles orthogonal to all the circles of the first family is called a _ hyperbolic apollonian pencil_. each circle from the latter family is a locus of points @xmath6 such that @xmath7 .
we refer to @xcite for this and other classical results related with circles and conics .
we construct four webs defined in the positive quadrant @xmath3 .
these webs consist of the following families of curves : + _ 1 ) both families of apollonian circles and confocal hyperbolas with foci @xmath4 and @xmath5 ; + 2 ) both families of apollonian circles and confocal ellipses with foci @xmath4 and @xmath5 ; + 3 ) families of confocal ellipses , confocal hyperbolas and the hyperbolic appolonian pencil with foci @xmath4 and @xmath5 ; + 4 ) families of confocal ellipses , confocal hyperbolas and the elliptic appolonian pencil with foci @xmath4 and @xmath5 .
_ we prove the existence of these webs by showing a diffeomorphism from a domain @xmath8 to the positive quadrant @xmath3 which maps horizontal , vertical and `` diagonal '' ( @xmath9 or @xmath10 is a constant ) lines to considered curves .
for webs @xmath0 and @xmath11 the images of both diagonal direction are remarkable curve : apollonian circles and lines ( vertical or horizontal ) . before going to the proof ,
let us say how these pencils relate with each other in an algebraic sense .
all circles in the plane can be considered as conics passing through two fixed points of the complex infinite line .
these points are called _ circular _ and have homogeneous coordinates @xmath12 and @xmath13 .
foci @xmath4 and @xmath5 of any conic @xmath14 can be described as the intersection of tangent lines to @xmath14 from the circular points .
the other pair of points of the intersection , denote it by @xmath15 and @xmath16 , can also be considered as foci . on figure
[ fig : projective description ] the projective picture is drawn . through each point @xmath17 in the plane pass the four conics : two conics touching sides of the quadrilateral @xmath18 and two passing through its opposite vertices and @xmath19 and @xmath20 .
the lines appearing in webs @xmath0 and @xmath11 can be described as lines passing through intersection of @xmath21 and @xmath22 ( horizontal lines ) and through the intersection of @xmath21 and @xmath23 ( vertical lines ) .
some triples of these conics have another joint intersections which correspond to points symmetric to @xmath17 in the axes ( figure [ fig : four curves ] ) .
so , in general we describe webs formed by the family of conics tangent to four fixed lines and passing though points of their intersection .
w. bhm in @xcite constructed a net consisting of lines touching a conic .
quadrilaterals formed by lines of this net can be circumscribed around circles and points of intersection of these lines can be split into families lying on confocal conics .
this construction was rediscovered and generalized by the author and a. bobenko in @xcite , where also it was noticed that bhm s net is a special case of the poncelet grid introduced and investigated by r. schwarz @xcite , see also @xcite for an additional discussion .
the current constructions were inspired by the work of h. edelsbrunner @xcite , where he invented a new approach for designing smooth surfaces from a set of spheres .
his method of connecting spheres by circumscribed hyperboloids is based on the following observation .
let @xmath24 , @xmath25 satisfy @xmath26 . if we scale each elliptic apollonian circle of points @xmath4 and @xmath5 @xmath24 times and each hyperbolic apollonian circle @xmath25 with respect to its center , the obtained circles touch a fixed hyperbola with foci @xmath4 and @xmath5 .
figure [ fig : edelsbrunner construction ] shows the case @xmath27 , when the hyperbola is equilateral .
note that if @xmath28 , then scaled elliptic apollonian circles touch an ellipse with foci @xmath4 and @xmath5 ( figure [ fig : edelsbrunner construction ellipse ] ) .
( the touching points may have complex coordinates . )
let us fix the notation we will use .
we suppose that foci of pencils have coordinates @xmath29 and @xmath30 .
denote by @xmath31 and @xmath32 the distances from point @xmath17 to foci @xmath4 and @xmath5 .
define the following functions : * @xmath33 .
the locus of points @xmath34 , is a hyperbolic apollonian circle corresponding to the points @xmath4 and @xmath5 . for points
@xmath35 we have @xmath36 .
* @xmath37 .
the locus of points @xmath38 , @xmath35 , is the upper arc of a circle passing through @xmath4 and @xmath5 ( and the arc symmetric to it in @xmath39-axis ) .
indeed , for any point @xmath17 on this arc we have @xmath40 . *
the locus of points @xmath42 is a branch of a hyperbola with foci @xmath4 and @xmath5 .
note that for @xmath43 we have @xmath44 .
* @xmath45 .
the locus of points @xmath46 is an ellipse with foci @xmath4 and @xmath5 .
note that @xmath35 uniquely determined by values of any two function from this list .
for simplicity in the following sections we denote the values of considered functions at point @xmath17 by @xmath47 , @xmath48 , @xmath49 , @xmath50 , @xmath51 , and @xmath52 .
in this section we prove that two pencils of apollonian circles with foci at @xmath4 and @xmath5 , and the family of hyperbolas with foci at @xmath4 and @xmath5 form a web in the positive quadrant @xmath53 . applying the map to vertices of a shifted lattice @xmath54
we obtain a configuration shown on figure [ fig : cir - cir - hyp ] .
[ thm : cir - cir - hyp ] suppose @xmath1 be a set of points @xmath55 with positive @xmath39-coordinate , and the map @xmath56 is defined by the following way : @xmath57 then @xmath58 is surjective and maps lines @xmath59 to arcs of circles of the elliptic apollonian pencil , rays @xmath60 to arcs of circles of the hyperbolic apollonian pencil and rays @xmath61 to arcs of hyperbolas with foci @xmath4 and @xmath5 . for the proof we need to show that the value of @xmath49 depends only on @xmath62 , @xmath50 only on @xmath39 , and @xmath51 only on @xmath10 .
we start the calculation from finding values of @xmath47 and @xmath48 .
suppose the point @xmath17 has coordinates @xmath63 and it is an image of the point @xmath64 under the map @xmath58
. then @xmath65 and @xmath66 .
therefore @xmath67 analogously @xmath68 now let us compute @xmath49 , @xmath50 , and @xmath51 .
@xmath69 @xmath70 @xmath71 the surjectivity of the map easily follows from the fact that varying @xmath39 and @xmath62 we can obtain all circles from apollonian pencils .
we prove that two pencils of apollonian circles with foci at @xmath4 and @xmath5 , and the family of ellipses with foci at @xmath4 and @xmath5 form a web in the positive quadrant @xmath53 ( figure [ fig : cir - cir - el ] ) .
[ thm : cir - cir - el ] suppose @xmath1 is the forth quadrant , the set of points in the plane with positive @xmath39-coordinate and negative @xmath62-coordinate . the map @xmath56 is defined by the following way : @xmath72 then @xmath58 is surjective and maps rays @xmath59 to arcs of circles of the elliptic apollonian pencil , rays @xmath60 to arcs of circles of the hyperbolic apollonian pencil and rays @xmath61 to arcs of ellipses with foci @xmath4 and @xmath5 . for the proof we need to show that the value of @xmath49 depends only on @xmath62 , @xmath50 only on @xmath39 , and @xmath52 only on @xmath10 . as in the proof of theorem [ thm : cir - cir - hyp ] , we start from the calculation of @xmath47 and @xmath48 .
@xmath73 analogously @xmath74 now let us compute @xmath49 , @xmath50 and @xmath52 .
@xmath75 @xmath76 @xmath77 again , the surjectivity of the map easily follows from the fact that varying @xmath39 and @xmath62 we can obtain all circles from apollonian pencils .
in this section we prove that there is a web formed by confocal ellipses and hyperbolas , and the hyperbolic apollonian pencil .
the family of vertical lines can be adjoint to this web in a very natural way ( figure [ fig : hip - el - ap ] ) .
[ thm : hip - el - ap ] suppose @xmath1 is the forth quadrant and the map @xmath56 is defined by the following way : @xmath78 then @xmath58 is surjective and maps rays @xmath59 and @xmath60 to arcs of ellipses and hyperbolas with foci @xmath4 and @xmath5 , rays @xmath61 to arcs of circles of the hyperbolic apollonian pencil with foci @xmath4 and @xmath5 , and intervals @xmath79 to vertical lines . in this case @xmath47 and @xmath48 have a very nice formulas : @xmath80 @xmath81 we see that the values @xmath82 and @xmath83 defined only by @xmath39 and @xmath62 , respectively , it is left to check that the ratio @xmath84 depends only on @xmath10 : @xmath85 the fact that image of points with @xmath79 is a vertical lines follows directly form the definition of @xmath58 . since varying @xmath39 and @xmath62 we can obtain all ellipses and hyperbolas from the confocal family the image of @xmath58 is the whole quadrangle @xmath53
the last example of a web is formed by confocal ellipses and hyperbolas , and and elliptic apollonian pencil , and family of horisontal lines , which can be adjoint to the web as well ( figure [ fig : hip - ell - pas ] ) .
[ thm : hip - ell - pas ] suppose @xmath1 is the forth quadrant and the map @xmath56 is defined by the following way : @xmath86 then @xmath58 is surjective and maps rays @xmath59 and @xmath60 to arcs of ellipses and hyperbolas with foci @xmath4 and @xmath5 , rays @xmath61 to arcs of circles of the elliptic apollonian pencil with foci @xmath4 and @xmath5 , and intervals @xmath79 to horizontal lines .
we have @xmath89 and @xmath90 .
let us calculate the value of @xmath50 .
@xmath91 we see , that it depends only on @xmath10 .
it is left to notice that from the definition of @xmath58 , follows that images of intervals @xmath92 are horizontal lines .
the map is surjective by the same arguments as in the previous theorem . | we consider families of confocal conics and two pencils of apollonian circles having the same foci . we will show that these families of curves generate _ @xmath0-webs _ and find the exact formulas describing them . |
several experimental groups have produced vortices in bose einstein condensates ( bec ) @xcite .
one type of experiments consists in imposing a laser beam on the magnetic trap holding the atoms to create a harmonic anisotropic rotating potential . for a prolate trap ,
it has been observed @xcite that when a single vortex exists , the vortex line is not straight along the axis of rotation , but bending .
theoretical works @xcite establish a simpler expression of the gross pitaevskii energy that only depends on the vortex lines . in @xcite , it is proved that bending occurs for prolate condensates , but not for oblate ones .
minimization algorithms @xcite have been used to compute local minima of the gross pitaevskii energy and provide an evidence of the bending in the same setting as the experiment . bending ( or @xmath0 )
vortices are described in detail , and multiple vortex configurations are addressed in these studies . recently , the ens group @xcite has further studied configurations with a single vortex line .
they have observed planar bent vortices ( @xmath0 ) but also different configurations ( @xmath1 ) .
they study the length of the line , its deviation from the center and its angular momentum . in this paper , motivated by the recent experiments at the ens @xcite
, we numerically look for local minimizers of the gross pitaevskii energy and we want to understand the various vortex configurations observed in the experimental setting : @xmath0 vortices but also @xmath1 vortices .
we compute solutions with up to 4 vortices and describe their three - dimensional structure .
different solution branches are followed and the evolution of the corresponding energy and angular momentum are shown . the framework of this study is the case of a prolate condensate where bending is an important phenomenon .
we consider a pure bec of @xmath5 atoms confined in a harmonic trapping potential rotating along the @xmath6 axis at angular velocity @xmath4 .
the equilibrium of the system corresponds to local minima of the gross - pitaevskii energy in the rotating frame @xmath7 where @xmath8 and the wave function @xmath9 is normalized to unity @xmath10 . here , for any complex quantities @xmath11 and @xmath12 and their complex conjugates
@xmath13 and @xmath14 , @xmath15 . for numerical applications , it is more convenient to rescale the variables as follows : @xmath16 , @xmath17 , where @xmath18 and @xmath19 in this scaling the thomas - fermi limit of @xmath11 is [ rtf ] ( * r*)=0 -(x^2+^2y^2+^2z^2).then , we use the dimensionless energy introduced in @xcite [ be3d ] e(u ) = h(u ) - l_z(u),with [ h ] h(u)=_12 |u|^2 -12 ^ 2|u|^2 + 14 ^ 2|u|^4,[lz ] l_z(u)=i _ |u ( y - x ) , defined in the domain @xmath20 in the present study we compute critical points of @xmath21 by solving the norm - preserving imaginary time propagation of the corresponding equation : [ eqstrong ] -12u + i().u=1u(-|u|^2)+_u , with @xmath22 on @xmath23 and @xmath24 the lagrange multiplier for the norm constraint @xmath25 .
a hybrid 3 steps runge - kutta - crank - nicolson scheme @xcite is used to advance the equation in time : @xmath26 where @xmath27 contains the remaining non - linear terms .
the corresponding constants for every step ( @xmath28 ) are : @xmath29 the resulting semi - implicit scheme is second order time accurate and allows reasonably large time steps , making it appropriate for long time integration .
the large sparse matrix linear systems resulting from the implicit terms are solved by an alternating direction implicit ( adi ) factorization technique . for the spatial discretization
we use finite differences on a cartesian uniform mesh with periodic boundary conditions in all directions . to accurately resolve sharp gradients of the variable in presence of vortices , low numerical dissipation and very accurate schemes
are required for the spatial derivatives .
a sixth - order compact finite difference scheme @xcite with spectral - like resolution was chosen to this end .
the values of constants in ( [ eqstrong ] ) are set to @xmath30 corresponding to the experiments of the ens group @xcite ( @xmath31 , @xmath32 , @xmath33 and @xmath34 ) .
the angular frequency @xmath4 will be varied from 0 to the maximum value of @xmath35 , for which no deformation of the condensate has to be taken into account .
equation ( [ eqstrong ] ) is propagated in imaginary time until the evolution of the energy ( [ be3d ] ) has a gradient in time smaller than @xmath36 . for the considered range of @xmath37 , the numerical domain
is fixed to an elongated box @xmath38\times[-0.6,0.6]\times[-8.5 , 8.5]$ ] .
a refined grid using @xmath39 nodes is employed , which is sufficient to achieve grid - independence for all considered numerical experiments .
different initial conditions are used in to trigger single or multiple vortex configurations and follow the corresponding branches as @xmath4 is varied .
the simplest initial condition assumes a steady - state solution @xmath40 and is useful to study vortex - free configurations and their degeneracy into multiple vortex configurations when increasing the value of @xmath37 .
initial conditions with vortices are obtained by superimposing to the steady - state a simple ansatz for the vortex .
for example , an initial condition with a centered straight vortex of radius @xmath41 is obtained by imposing @xmath42\right\ } } \cdot exp(i \varphi),\end{aligned}\ ] ] where @xmath43 are the polar coordinates in the @xmath44 plane .
the 3d shape of the vortex can be easily modified by shifting the center @xmath45 of the vortex in successive @xmath44 planes ; for instance , to obtain a planar s shape vortex , the following function can be used : @xmath46\big / tanh(\alpha_v ) , & z < 0 \\ 1 + tanh\left[\alpha_v \left(-1+{\displaystyle z\over \beta_v}\right)\right]\big / tanh(\alpha_v ) , & z \geq 0 \end{array } \right.\ ] ] the constants @xmath47 control , respectively , the curvature and the height of the vortex .
we first focus on single vortex configurations and describe later multi vortex configurations .
we have observed three different types of single vortex configurations as shown in figure [ fig-1v - all ] : planar @xmath0 vortices , planar @xmath1 vortices and non - planar @xmath1 vortices .
the @xmath0 vortices are the bent vortices computed in @xcite and theoretically studied in @xcite .
they are global minimizers of the energy .
the @xmath1 configurations were observed experimentally very recently @xcite and are only local minimizers of the energy .
vortex configurations for @xmath48 ( a ) , @xmath49 ( b ) , @xmath50 ( c ) . ]
the @xmath0 vortex is a planar vortex formed of 2 parts : the central part is a line which stays on the @xmath6 axis and the outer part reaches the boundary of the condensate perpendicularly .
when @xmath4 increases , the central straight part gets longer ( figure [ fig - u1-omega ] ) and the angular momentum ( @xmath51 ) increases to 1 ( figure [ fig - e1v ] ) .
the @xmath0 vortex is obtained by starting the simulation with an initial condition containing a straight vortex away from the @xmath6 axis .
in fact , the @xmath0 vortex lies either in the @xmath2 or @xmath3 plane . starting with an initial condition which is not in one of these plane yields a final state in the @xmath3 plane , which is the plane closest to the @xmath6 axis .
the shape of the the @xmath0 vortex and its preferred location in the @xmath3 plane can be analyzed using the approximate energy derived in @xcite : setting the vortex free solution to 0 energy , then the energy of a vortex line @xmath52 can be approximated by [ env ] e_= _ dl - c _ ^2 dz , where @xmath53 is a constant which depends on the experimental parameters and @xmath54 is given by ( [ rtf ] ) .
if @xmath52 is not in the @xmath2 or @xmath3 plane , then one can construct small perturbations of @xmath52 that preserve @xmath54 and lower the energy .
this implies that @xmath52 can not be a critical point of the energy because the gradient is not zero .
of course , if the ellipticity of the cross section is small , the gradient is small , which may allow to observe these configurations . in order to understand the existence of the straight central part of the @xmath0 vortex
, one can also refer to the analysis of @xcite : from equation ( [ env ] ) we can infer that a vortex line with a lower energy than the vortex free solution is obtained when the quantity @xmath55 is negative , _
i.e. _ @xmath56 .
let @xmath57 be such that @xmath58 ; it corresponds to the 2d critical velocity for the existence of a vortex in the plane @xmath59 . for @xmath4 close to @xmath60 , the inner region where @xmath61 is concentrated near the center of the condensate . in this region ,
the vortex line has to be straight ( see @xcite ) .
this straight part is getting longer as @xmath4 increases since the region where @xmath62 is getting bigger .
this region corresponds to @xmath63 , where @xmath64 is the critical velocity for the existence of a vortex in the 2 dimensional section where @xmath6 is constant . in the outer region
, the vortex reaches the boundary using the shortest path .
figure [ fig - e1v ] shows the energy and angular momentum variation with @xmath37 for the single vortex configurations .
the @xmath0 vortices exist only for @xmath4 bigger than a critical value @xmath65 .
it is interesting to note that at @xmath66 , the energy of the @xmath0 vortex is bigger than the energy of the vortex free solution ( we have set to zero the energy of the vortex free solution ) .
a zoom in this region shows that @xmath66 is very close to the angular velocity @xmath67 for which the energy of the vortex free solution is equal to the energy of the @xmath0 vortex . ) and angular momentum per particle ( in units of @xmath68 ) for the single vortex configurations . ]
figure [ fig - e1v ] also shows that the angular momentum @xmath51 of the @xmath0 vortex for @xmath69 does not go to 0 .
this suggests that in fact there could be another @xmath0 solution for @xmath70 . using an ansatz ,
another type of @xmath0 solution is obtained in @xcite which is a saddle point of the energy : it is away from the axis and has lower angular momentum . in @xcite , it is proved rigorously that for @xmath4 small , there is no @xmath0 as a critical point of the energy .
for an initial condition with a straight vortex centered on the @xmath6 axis , if @xmath71 , the straight vortex is unstable and the final configuration is a @xmath0 , but if @xmath72 , the straight vortex is stable .
this is in agreement with the result of @xcite where the local stability of the straight vortex for @xmath4 larger is proved . for small @xmath4
, the @xmath0 vortex disappears and a vortex - free configuration is obtained , while for @xmath4 larger the @xmath0 vortex degenerates into a three - vortex configuration ( described later ) . motivated by the experiments of @xcite , we compute new critical points of the energy , which are @xmath1 configurations ( see figure [ fig-1v - all ] ) .
several numerical experiments were performed , starting from different initial conditions containing an ansatz for the @xmath1 vortex ( see section [ sec - param ] ) .
the planar @xmath1 can be regarded as a @xmath0 , with the half - part in the plane @xmath73 rotated with respect to the @xmath6 axis by 180 degrees ( see figure [ fig - u1s1 ] ) .
the non planar @xmath1 are such that the projections of the branches on the @xmath74 plane are orthogonal , _ i.e. _ the rotation of the branches is of 90 degrees .
we could check that non planar @xmath1 configurations with an angle between the branches different from 90 degrees do not exist . .
superposition of the @xmath0 and @xmath1 vortex ( a ) and the planar and non - planar @xmath1 vortex ( b ) .
] as already mentioned for the @xmath0 vortex , stable planar @xmath1 configurations lie either in the @xmath2 or @xmath3 plane . as for the @xmath0 ,
this can be explained using the limiting energy obtained in @xcite and considering separately the upper or lower part of the @xmath1 .
as soon as the cross section is not a disc , if the upper or lower branch of the @xmath1 configuration does not lie in the @xmath2 or @xmath3 plane , then the gradient of vortex line energy ( [ env ] ) can never be zero when @xmath52 is varied .
vortex configuration for @xmath75 ( a ) , @xmath76 ( b ) , @xmath77 ( c ) . ]
the @xmath1 vortices exist for all values of @xmath4 while the @xmath0 only exist for @xmath70 .
when @xmath4 decreases , the extension of the @xmath1 along the @xmath6 axis goes downwards as shown in figure [ fig - s1-omega ] , the angular momentum decreases to 0 ( figure [ fig - e1v ] ) and the vortex tends to the horizontal axis .
note that a vortex along the horizontal axis has @xmath78 , but a positive energy . on the other side ,
when @xmath4 increases , the @xmath1 gets straighter and it tends to the vertical axis .
the global minimum of the energy is never an @xmath1 .
but the difference in energy ( and angular momentum ) between @xmath0 and @xmath1 vortices is very small , as illustrated in figure [ fig - e1v ] because an @xmath1 vortex is almost like a @xmath0 with a half - part rotated by 180 degrees .
as pointed out in @xcite , the minimization problem which is related to the experiments , is rather to minimize @xmath80 ( see ( [ h ] ) while fixing @xmath51 , rather than minimizing @xmath81 .
this has been studied in the 2 dimensionnal setting in@xcite .
one can notice that if a given configuration with @xmath82 and @xmath83 minimizes @xmath81 for some @xmath4 , then @xmath84 minimizes @xmath80 under the constraint that @xmath83 : indeed if @xmath85 with @xmath86 , then @xmath87 , since @xmath88 minimizes @xmath89 , and this implies that @xmath90 .
moreover @xmath4 is the slope to the curve @xmath91 at the point @xmath88 and the property of minimizing @xmath89 that is for all @xmath92 , @xmath93 , @xmath94 implies that the curve @xmath91 lies above its tangent at this point .
we have plotted @xmath80 as a function of @xmath51 .
we can check that the curve is convex , and above its tangent , which is consistent with the fact that we have computed minimizers of the energy .
we know that the @xmath0 solution exists for @xmath95 and has @xmath96 . for @xmath97 , we expect that the process of minimizing @xmath80 with fixed @xmath51 would produce @xmath0 vortices and the curve @xmath91 should be concave in this region . in @xcite , we have proved that for @xmath51 close to 0 , @xmath98 , which is a first indication to the concavity of the curve .
multiple vortex configurations are obtained based upon different numerical strategies .
the first one is to start the computation from a vortex - free steady state and to abruptly increase @xmath37 to a very hight value ; multiple vortices are thus obtained . the second strategy is to generate an initial condition with vortices as described in section [ sec - param ] ( the advantage being the control of the shape and initial arrangement of the vortices ) .
both techniques are used to follow solution branches with two , three or four vortices in the condensate .
figures [ fig - eall ] and [ fig - lzall ] display energy and angular momentum vs @xmath37 for all studied configurations . ) for all studied configurations . ]
( in units of @xmath68 ) for all studied configurations . ]
when @xmath4 is increased , the single @xmath0 vortex solution switches to a 3 vortex configuration ( @xmath99 ) .
as shown in figure [ fig - u3-omega]a , the configuration is invariant under rotation in a central plane near @xmath59 but not near the edges . for large @xmath37 ,
three - dimensional views show ( figure [ fig - u3-omega ] a , b ) that there are 2 vortices of similar size and a longer one which is bending near the boundary . for @xmath100 , all vortices display contorted shapes ( figure [ fig - u3-omega]c ) , very similar to those reported in @xcite .
let us point out that the angular momentum of all these 3 vortex configurations is lower than 3 ( see figure [ fig - lzall ] ) .
( a ) , @xmath101 ( b ) , @xmath102 ( c ) .
lower pictures show iso - contours of @xmath103 in the central @xmath59 cut plane . ] when we put as initial condition a configuration with 3 identical @xmath0 vortices at 120@xmath104 , in the final state , one of them gets a little longer ( figure [ fig - u3-ligne]a ) and the symmetry is lost .
this configuration has almost the same energy and angular momentum as the configuration displayed in figure [ fig - u3-omega]b . in exchange
, the initial condition with three straight vortices on the @xmath105-axis has its symmetry preserved ( figure [ fig - u3-ligne ] b ) , but with a higher energy that the previous one .
, from different initial conditions : 3 identical @xmath0 vortices at 120@xmath104 ( a ) and 3 straight vortices in a row on the @xmath105-axis ( b ) .
lower pictures show iso - contours of @xmath103 in the central @xmath59 cut plane . ] when further decreasing @xmath4 , the 3 vortex branch switches to a 2-vortex displaying irregular shapes ( figure [ fig - u2-fromu3 ] ) .
is decreased to @xmath106 ( a ) and @xmath107 ( b ) . ]
the two - vortex branch presented in this section was obtained by starting from a vortex - free solution and suddenly increasing @xmath37 to a value of @xmath108 .
the configuration is planar and symmetric , like twice a single @xmath0 vortex , but away from the axis ( there is a repulsion between the lines ) .
when @xmath4 increases , the lines are almost straight and get closer to each other .
this is in agreement with the fact that when @xmath4 gets large , the straight vortex is a local minimizer of the energy .
hence the bending is no longer the important phenomenon .
( a ) , @xmath107 , ( b ) @xmath109 ( c ) . ]
we recall that decreasing @xmath4 from a configuration with 3 vortices , we obtained 2 vortices which are not symmetric , one being longer than the other ( figure [ fig - u2-fromu3 ] ) .
this configuration has slightly bigger energy than the 2 symmetric vortices .
starting from an initial condition without vortices and increasing @xmath37 to @xmath110 , we have obtained stable configurations with 4 curved vortices ( figure [ fig - u4 ] a ) .
when decreasing @xmath37 , this configuration rapidly degenerates into a three - vortex state . for lower @xmath37 we could obtain stable configurations with four symmetric vortices ( figure [ fig - u4 ] b ) , but with bigger energy .
the location of the vortices in the plane @xmath59 is the same .
- obtained from an initial condition without vortices and ( b ) @xmath111 - obtained from an initial condition with four symmetrical vortices . ]
we have to point out that with the initial condition of 4 identical vortices , the symmetry is preserved as displayed in figure [ fig - u4]b , which was not the case for 3 vortices .
we have studied different vortex configurations in a prolate bose enstein condensate by solving the gross pitaevskii equation .
we have computed @xmath0 and @xmath1 vortices , motivated by the recent experiments of @xcite .
our computations involve a parameter @xmath41 , which is small when the number of atoms @xmath5 is large .
decreasing @xmath41 , that is increasing the number of atoms forces the vortex lines to be almost straight in their central part , while for @xmath41 larger , the central straight part is not so obvious as in some figures of @xcite .
we have found that the @xmath1 vortices are only local minimizers of the energy and exist for all values of the angular velocity @xmath4 , while @xmath0 vortices are global minimizers existing for @xmath112 .
a planar @xmath1 vortex can be regarded as a @xmath0 vortex with a half - part rotated by 180@xmath104 .
moreover , @xmath0 or planar @xmath1 vortices lie only in the @xmath2 or @xmath3 plane while non planar @xmath1 vortices exists only for an angle of 90@xmath104 between the two branches . | we consider a rotating bose - einstein condensate in a harmonic trap and investigate numerically the behavior of the wave function which solves the gross pitaevskii equation . following recent experiments
@xcite , we study in detail the line of a single quantized vortex , which has a @xmath0 or @xmath1 shape . we find that a single vortex can lie only in the @xmath2 or @xmath3 plane .
@xmath1 type vortices exist for all values of the angular velocity @xmath4 while @xmath0 vortices exist for @xmath4 sufficiently large .
we compute the energy of the various configurations with several vortices and study the three - dimensional structure of vortices . |
electrical detection of the spin accumulation is one of the important technologies to achieve semiconductor - based spintronic applications.@xcite though excellent studies of these technologies were reported for gaas - based devices , almost all the spin - related phenomena were observed below room temperature.@xcite in general , it has been understood that the decay of the spin signals originates from the spin - flip scattering induced by the spin - orbit interaction , hyperfine interaction between electrons and nuclei , and so on.@xcite to minimize such intrinsic factors , silicon - based spintronic technologies have been proposed and developed,@xcite and room - temperature detections of the spin - dependent signals have been reported recently.@xcite the temperature - dependent spin - related phenomena in si - based devices were explored experimentally.@xcite for nondoped si channels,@xcite the relaxation of the injected spins can be explained by yafet s @xmath0 power law , indicating that the decay of the spin signals is attributed to the spin - flip scattering in the channel . for heavily doped si ( degenerated si ) channels , on the other hand , the decrease in the spin polarization of the injected spins is one of the main factors for the decay of spin signals.@xcite however , there is no study of the temperature - dependent spin signals for nondegenerated si channels . in this article
, we report on temperature evolution of spin signals by measuring the three - terminal hanle effect in the lateral device with a nondegenerated silicon channel . in our device with a cofe / si schottky - tunnel - barrier contact , a clear difference in the temperature - dependent spin signals between spin - extraction and spin - injection conditions is observed . under a certain condition
, the sensitivity of the spin detection can contribute dominantly to the magnitude of spin signals detected .
the cofe epitaxial layer was grown on si(111 ) by low - temperature molecular beam epitaxy ( lt - mbe ) at 60 @xmath1c and the cofe / si interface was atomically flat.@xcite a three - terminal lateral device ( channel thickness @xmath2 100 nm , carrier density @xmath3 6.0 @xmath4 10@xmath5 @xmath6 ) with one single - crystalline co@xmath7fe@xmath8 contact and two ausb ohmic ones with lateral dimensions of 10 @xmath4 200 @xmath9m@xmath10 and 100 @xmath4 200 @xmath9m@xmath10 , respectively , was fabricated by using conventional processes with photolithography , ar@xmath11 ion milling , and reactive ion etching.@xcite the schematic diagram of the device structure is shown in the inset of fig .
the distance between cofe and ausb is about @xmath2 50 @xmath9 m . to achieve tunneling conduction through the high - quality cofe / si interface
, we inserted sb @xmath12-doped @xmath13-si layer ( sb @xmath2 5@xmath410@xmath14 @xmath6 ) with a thickness of 5 nm between the epitaxial cofe layer and @xmath15-si channel.@xcite as a result , we obtained tunneling conduction having nonlinear @xmath16 characteristics through the interface and the rectification is quite small , as shown in fig .
therefore , we can regard the fabricated cofe/@xmath13-si/@xmath15-si contact for spin injection and extraction as a schottky - tunnel - barrier contact . hanle - effect measurements with a three - terminal geometry were performed by a dc method at 40 @xmath2 300 k. in the measurements , a small magnetic field perpendicular to the plane , @xmath17 , was applied after the magnetic moment of the cofe contact aligned parallel to the plane along the long axis of the contact .
spin accumulation in semiconductor channels can be detected electrically by measuring three - terminal voltage changes ( @xmath18@xmath19 ) via hanle - type spin precessions.@xcite figures 1(c ) and 1(d ) show representative @xmath18@xmath19-@xmath17 curves for @xmath20 -1.0 and 1.0 @xmath9a at 40 k. the electrons are injected into and extracted from , respectively , the si conduction band for reverse ( @xmath21 0 ) and forward ( @xmath22 0 ) biases , where quadratic background voltages depending on @xmath17 are subtracted from the raw data .
we note that clear hanle - effect signals were observed for both bias - current conditions but the sign of the voltage change is opposite .
such sign reversal was seen only when the polarity of @xmath23 was switched .
these curves were fitted with a simple lorentzian function,@xcite @xmath18@xmath19(@xmath17 ) @xmath24 @xmath18@xmath25/[1+(@xmath26@xmath27)@xmath10 ] , where @xmath29 @xmath30@xmath17/@xmath31 is the larmor frequency , @xmath32 is the electron @xmath32-factor ( @xmath33 2 ) , @xmath34 is the bohr magneton , @xmath27 is the lower limit of spin relaxation time . the fitting results are denoted by the red solid curves .
the @xmath27 values for @xmath20 -1.0 and 1.0 @xmath9a are estimated to be @xmath2 4.37 and @xmath2 2.05 nsec , respectively .
the precise origin of the difference in @xmath27 between @xmath20 -1.0 and 1.0 @xmath9a is unclear yet , but we can consider a possible difference in the relative position of the spin accumulation in @xmath15-si to the @xmath13-si layer , as illustrated in fig .
there may be some differences in the influence of the @xmath13-si layer on spin relaxation between spin injection ( @xmath21 0 ) and extraction ( @xmath22 0 ) conditions .
characteristics for various temperatures .
( b ) schematic illustration of spin accumulation in spin - injection and spin - extraction conditions . @xmath18@xmath19-@xmath17 curves at ( c ) @xmath23@xmath35 -1.0 and ( d ) 1.0 @xmath9a at 40 k , and fitting results with lorentzian function ( solid curves).@xcite ( e ) the magnitude of the observed spin accumulation signals , @xmath36@xmath18@xmath37@xmath36 , as a function of bias current @xmath23 for various temperatures . , width=283 ] we hereafter focus on the magnitude of @xmath18@xmath19 , @xmath38@xmath39 , for various temperatures . figure 1(e ) displays @xmath38@xmath39 versus bias current @xmath23 ( -2.0 @xmath9a @xmath40 2.0 @xmath9a ) at 40 , 50 , 75 , and 100 k. we can see clear asymmetric variation in @xmath38@xmath39 with respect to the polarity of @xmath23.@xcite for @xmath22 0 , @xmath38@xmath39 decreases with increasing temperature in all @xmath23 region while , for @xmath21 0 , complicated variations in @xmath38@xmath39 are seen , particularly , in -1.0 @xmath9a @xmath41 @xmath42 0 @xmath9a .
we can see that @xmath38@xmath39 at 100 k is larger than that at 50 k. concentrating on this interesting phenomenon , we further explored temperature - dependent @xmath38@xmath39 for various @xmath23 in detail . figures 2(a ) and 2(b ) show @xmath38@xmath39 as a function of temperature for spin - extraction ( @xmath22 0 ) and spin - injection ( @xmath21 0 ) conditions , respectively .
the features observed are summarized as follows : ( i ) temperature evolution of @xmath38@xmath39 shows a clear difference between spin - extraction ( @xmath22 0 ) and spin - injection ( @xmath21 0 ) conditions .
( ii ) @xmath38@xmath39 can be detected even at room - temperature with a large injection current of - 20 @xmath9a in @xmath21 0 [ see the inset of fig .
2(b ) ] , whereas @xmath38@xmath39 is markedly reduced with increasing temperature and disappears at 250 k irrespective of @xmath23 in @xmath22 0 .
@xmath27 for @xmath43 -20 @xmath9a at room temperature can be estimated to be 1.36 nsec . the @xmath27 value decreases with increasing temperature from 3.02 to 1.36 nsec , consistent with the feature observed in the heavily doped si.@xcite ( iii ) in @xmath43 -0.1 and -1.0 @xmath9a we can see the partial increases in @xmath38@xmath39 despite the rise of temperature [ see arrows in fig .
2(b ) ] . @xmath18@xmath37@xmath36 for ( a ) spin - extraction ( @xmath22 0 ) and ( b ) spin - injection ( @xmath21 0 ) conditions .
the inset of ( a ) shows carrier density and resistivity as a function of temperature for the si channel used in this study .
the inset of ( b ) displays room - temperature hanle - effect signal at @xmath43 -20 @xmath9a and the fitting curve with lorentzian function.,width=321 ] to examine these phenomena , first of all , temperature dependences of resistivity ( @xmath44 ) and carrier density ( @xmath15 ) were shown in the inset of fig .
2(a ) , where @xmath44 and @xmath15 were measured directly by four - terminal transport and hall effect measurements . both @xmath44 and @xmath15 have strong temperature dependence in @xmath45 100 k , nearly consistent with the feature of a general nondegenerated si.@xcite in this context , we should consider the component of @xmath38@xmath39 associated with changing @xmath44 on the basis of the spin diffusion model.@xcite since the spin - related voltage changes in si are proportional to @xmath44@xmath4@xmath46,@xcite the variation in @xmath44 with changing temperature can affect @xmath38@xmath39 in our measurements .
the detail is discussed later . next , in fig .
3 we examined @xmath38@xmath39 versus @xmath47 , i.e. , the bias voltage at the cofe / si interface for various temperatures .
the @xmath38@xmath39 in @xmath48 0 has an evident maximum value at a certain @xmath47 , where @xmath48 0 indicates spin - extraction conditions ( @xmath22 0 ) .
this means that there is the most sensitive @xmath47 value for detecting spin - accumulation signals in @xmath22 0 .
we would like to define the detectability for the spin accumulation signals as the sensitivity of the spin detection at the biased contact .
in general , the sensitivity of the spin detection at the schottky - tunnel - barrier contacts has already been discussed for ferromagnet / semiconductor lateral devices.@xcite the observed feature also implies that we should consider the sensitivity of the spin detection for our cofe / si schottky - tunnel - barrier contacts .
however , when we focus on the temperature evolution of @xmath38@xmath39 at a certain @xmath47 , the feature of the reduction in @xmath38@xmath39 with increasing temperature is relatively simple in @xmath48 0 .
namely , we do not need to consider the change in the spin - detection sensitivity with temperature evolution in @xmath22 0 .
on the other hand , there is almost no clear correlation between @xmath38@xmath39 and @xmath47 in @xmath49 0 , where @xmath49 0 indicates spin - injection conditions ( @xmath21 0 ) .
we find that in -0.1 v @xmath41 @xmath47 @xmath41 0 v @xmath38@xmath39 at 75 and 100 k is higher than that at 40 and 50 k. namely , for @xmath21 0 , we should also consider the sensitivity of the spin detection with temperature evolution .
@xmath18@xmath37@xmath36 as a function of @xmath47 for various temperatures .
the insets show temperature - dependent @xmath50 for various @xmath23 conditions.,width=283 ] here we also concentrate on the variation in the interface resistance ( @xmath50 ) .
the inset of fig .
3 shows @xmath50 as a function of temperature for various @xmath23 conditions .
since the @xmath16 characteristics of our device are nonlinear features , the @xmath50 values can be varied with changing @xmath23 .
it should be noted that @xmath50 slightly decreases with increasing temperature for all @xmath23 conditions .
hence @xmath47 can also slightly be changed by the variation in temperature even if we use the same @xmath23 condition for the spin injection or extraction .
considering these features described so far , we discuss the temperature evolution of @xmath38@xmath39 . under spin - extraction conditions
( @xmath22 0 ) [ see fig .
4(a ) ] , we do not have to consider the temperature - dependent sensitivity of the spin detection at a certain @xmath47 , as described in previous paragraph .
we explain the detailed picture as follows . when the electrons tunnel from the si conduction band into the spin - polarized empty states in cofe at low temperature ( e.g. 40 k ) under @xmath47 ( @xmath51 ) , the spin accumulation ( @xmath18@xmath9 ) can be formed in the si conduction band , as shown in the left figure of fig .
in this situation , the spin - dependent tunneling of electrons is dominant at the fermi level ( see - dashed pink line ) , leading to the detection of @xmath18@xmath9 .
as the temperature rises [ e.g. 100 k , see right figure of fig .
4(a ) ] , @xmath47 is reduced , resulting in a small decrease in @xmath51 .
simultaneously , @xmath15 is steeply enhanced from @xmath5210@xmath53 to @xmath210@xmath5 @xmath6 , causing the rapid decrease in @xmath44 . as a result , though the @xmath18@xmath9 value decreases markedly based on the spin diffusion model,@xcite there is almost no difference in the basic situation for the detection of @xmath18@xmath9 after the rise of temperature . in other words , we can detect @xmath18@xmath9 induced by the spin extraction irrespective of temperature because the spin - dependent tunneling of electrons is maintained .
therefore , there is the sensitivity of the spin detection . in this case , since @xmath18@xmath9 in the si conduction band is related directly to the observed @xmath38@xmath39 values , the change in @xmath44 can dominantly affect the @xmath38@xmath39 values on the basis of the spin diffusion model . in fig .
2(a ) we can find the steep and gradual decreases in @xmath38@xmath39 in @xmath45 100 k and @xmath54 100 k , respectively .
since @xmath44 in @xmath54 100 k is almost constant , the contribution of @xmath44 to the decrease in @xmath38@xmath39 can be ignored in @xmath54 100 k. we infer that , for spin - extraction conditions , the features in @xmath45 100 k and @xmath54 100 k are dominated by the reduction in @xmath44 and @xmath55 ( increase in the spin - flip scattering in si ) , respectively .
it seems that the change in @xmath38@xmath39 influenced by the spin - flip scattering is relatively small in the device with a nondegenerated si channel . under spin - injection conditions ( @xmath21 0 ) , by contrast
, we have to consider two different cases whether the injection current is relatively large or not , as shown in fig .
when the large injection current ( e.g. @xmath43 -20 @xmath9a ) is used , the relatively large @xmath18@xmath9 can be demonstrated.@xcite thus @xmath18@xmath9 induced in the si conduction band can reach the fermi level of the spin - polarized tunneling electrons at any temperatures .
when the temperature rises , @xmath18@xmath9 is merely reduced ( middle figures ) by the same mechanism of fig .
namely , we can still obtain the sensitivity of the spin detection . on the other hand ,
if the injection current is relatively low ( e.g. @xmath43 -0.1 @xmath9a ) at 40 k , @xmath18@xmath9 induced in the si conduction band is quite small .
since the relatively large @xmath51 is applied in the low bias current regime at 40 k , we can understand that the position of the quasi fermi level of the spin - polarized electrons in the si conduction band is quite lower than the fermi level ( dashed pink line ) of the spin - polarized tunneling electrons from cofe .
in this situation , the spin - polarized electrons can tunnel into the unpolarized states ( unoccupied states ) in the si conduction band .
thus we could not obtain @xmath18@xmath9 as a consequence of the @xmath38@xmath39 measurement at @xmath43 -0.1 @xmath9a and 40 k. this means that there is no sensitivity of the spin detection .
the similar situations have already been reported for fe / gaas devices previously,@xcite and we have also reported the above features in detail for the cofe / si devices.@xcite why we can see that @xmath38@xmath39 appears at 100 k for @xmath43 -0.1 @xmath9a despite the rise of temperature ? shown in the right figure of the bottom in fig .
4(b ) is a possible schematic illustration of the interpretation of the spin detection .
as previously described , when the temperature rises up to 100 k , @xmath15 is enhanced from @xmath52 10@xmath53 to 10@xmath5 @xmath6 , causing the rapid decrease in @xmath44 . as a result
, @xmath18@xmath9 is also reduced on the basis of the spin diffusion model.@xcite simultaneously , since @xmath47 also decreases with decreasing @xmath50 , the reduction in @xmath51 also occurs .
accordingly , the fermi level of the spin - polarized tunneling electrons is located on @xmath18@xmath9 induced in the si conduction band , leading to the appearance of the sensitivity of the spin detection .
this can be considered to be the same situation for @xmath43 -20 @xmath9a at 40 k shown in the left figure of fig .
this is a possible mechanism of the appearance of the spin accumulation signals ( @xmath38@xmath39 ) with increasing temperature , observed in fig .
such qualitative consideration has already been explained in terms of the change in the spin - detection sensitivity at the tunnel contact.@xcite we emphasize that the tuning of @xmath47 ( @xmath51 ) is a key to detect @xmath18@xmath9 induced by the spin injection .
we finally comment on the difference in the actual temperature evolution of the decrease in @xmath38@xmath39 between spin - extraction ( @xmath22 0 ) and spin - injection ( @xmath21 0 ) conditions for the large currents of @xmath43 @xmath5620 @xmath9a [ see figs .
2(a ) and ( b ) ] .
as mentioned above , the temperature dependent @xmath38@xmath39 for @xmath22 0 can be explained mainly by the change in @xmath44 .
on the other hand , for @xmath21 0 , we should consider not only the contribution of @xmath44 and @xmath55 to @xmath38@xmath39 but also that of the spin - detection sensitivity to @xmath38@xmath39 .
though @xmath38@xmath39 is reduced with decrease in @xmath44 and @xmath55 , the sensitivity of the spin detection is enhanced in a low current bias regime at the same time .
this opposite feature causes the relatively gradual decrease in @xmath38@xmath39 at @xmath43 -20 @xmath9a compared with that at @xmath43 + 20 @xmath9a , as shown in fig .
2 . in our opinion
, it is very important for room - temperature device operation@xcite to understand the temperature - dependent sensitivity of the spin detection at the schottky - tunnel - barrier contact in devices with a nondegenerated si channel .
we have studied temperature evolution of spin accumulation signals obtained by the three - terminal hanle effect measurements for the device with a nondegenerated si channel . the clear difference in the temperature - dependent spin signals between spin - extraction and
spin - injection conditions was seen .
we found that it is important for the consideration of the temperature evolution to understand not only the mechanism based on the spin diffusion model but also the sensitivity of the spin detection at the schottky - tunnel - barrier contact .
these results are also important to enhance the spin signals in the device applications at room temperature . 11 s. a. wolf , d. d. awschalom , r. a. buhrman , j. m. daughton , s. v. molnr , m. l. roukes , a. y. chtchelkanova , and d. m. treger , science * 294 * , 1488 ( 2001 ) ; h. dery , p. dalal , .
cywiski , and l. j. sham , nature ( london ) * 447 * , 573 ( 2007 ) .
x. lou , c. adelmann , m. furis , s. a. crooker , c. j. palmstrm , and p. a. crowell , phys .
lett . * 96 * , 176603 ( 2006 ) ; x. lou , c. adelmann , s. a. crooker , e. s. garlid , j. zhang , s. m. reddy , s. d. flexner , c. j. palmstrm , and p. a. crowell , nat . phys . * 3 * , 197 ( 2007 ) .
m. ciorga , a. einwanger , u. wurstbauer , d. schuh , w. wegscheider , and d. weiss , phys .
b * 79 * , 165321 ( 2009 ) . g. salis , a. fuhrer , and s. f. alvarado , phys . rev .
b * 80 * , 115332 ( 2009 ) ; g. salis , s. f. alvarado , and a. fuhrer , phys . rev .
b * 84 * , 041307(r ) ( 2011 ) .
very recently , room - temperature detection of spin transport was demonstrated even for gaas - based devices ; t. uemura , t. akiho , m. harada , k. matsuda , and m. yamamoto , appl .
lett . * 99 * , 082108 ( 2011 ) .
j. m. kikkawa and d. d. awschalom , phys .
80 * , 4313 ( 1998 ) .
b. c. min , k. motohashi , c. lodder , and r. jansen , nat .
mater . * 5 * , 817 ( 2006 ) .
i. utic , j. fabian , s. c. erwin , phys .
lett . * 97 * , 026602 ( 2006 ) .
i. appelbaum , b. huang , and d. j. monsma , nature * 447 * , 295 ( 2007 ) .
o. m. j. vant erve , a. t. hanbicki , m. holub , c. h. li , c. awo - affouda , p. e. thompson , and b. t. jonker , appl .
phys . lett . *
91 * , 212109 ( 2007 ) .
y. ando , k. hamaya , k. kasahara , y. kishi , k. ueda , k. sawano , t. sadoh , and m. miyao , appl .
. lett . * 94 * , 182105 ( 2009 ) .
t. sasaki , t. oikawa , t. suzuki , m. shiraishi , y. suzuki , and k. noguchi , ieee trans . mag . * 46 * , 1436 ( 2010 ) .
s. p. dash , s. sharma , r. s. patel1 , m. p. jong , and r. jansen , nature ( london ) * 462 * , 491 ( 2009 ) . t. suzuki ,
t. sasaki , t. oikawa , m. shiraishi , y. suzuki , and k. noguchi , appl .
express * 4 * , 023003 ( 2011 ) .
y. ando , y. maeda , k. kasahara , s. yamada , k. masaki , y. hoshi , k. sawano , k. izunome , a. sakai , m. miyao , and k. hamaya , appl . phys .
lett . * 99 * , 132511 ( 2011 ) .
b. huang , d. j. monsma , and i. appelbaum , phys .
lett . * 99 * , 177209 ( 2007 ) .
t. sasaki , t. oikawa , t. suzuki , m. shiraishi , y. suzuki , and k. noguchi , appl .
phys . lett . *
96 * , 122101 ( 2010 ) .
y. maeda , k. hamaya , s. yamada , y. ando , k. yamane , and m. miyao , appl .
* 97 * , 192501 ( 2010 ) .
y. ando , k. kasahara , k. yamane , y. baba , y. maeda , y. hoshi , k. sawano , m. miyao , and k. hamaya , appl .
lett . * 99 * , 012113 ( 2011 ) .
the influence of the electric fields on the spin injection efficiency , discussed in phys .
b * 66 * , 235302 ( 2002 ) , can be ignored because our device has the relatively large interface resistance and there is almost no influence of the spin absorption into the cofe contact .
a. fert and h. jaffrs , phys .
b * 64 * , 184420 ( 2001 ) ; a. fert , j .-
george , h. jaffrs , and r. mattana , ieee trans .
electron devices * 54 * , 921 ( 2007 ) .
g. l. pearson and j. bardeen , phys . rev . * 75 * , 865 ( 1949 ) ; f. j. morin and j. p. maita , phys .
rev . * 96 * , 28 ( 1954 ) .
r. jansen and b. c. min , phys .
rev . lett . * 99 * , 246604 ( 2007 ) .
a. n. chantis and d. l. smith , phys .
b * 78 * , 235317 ( 2008 )
. s. a. crooker , e. s. garlid , a. n. chantis , d. l. smith , k. s. m. reddy , q. o. hu , t. kondo , c. j. palmstrm , and p. a. crowell , phys .
b * 80 * , 041305(r ) ( 2009 ) . | we study temperature evolution of spin accumulation signals obtained by the three - terminal hanle effect measurements in a nondegenerated silicon channel with a schottky - tunnel - barrier contact .
we find the clear difference in the temperature - dependent spin signals between spin - extraction and spin - injection conditions . in a spin - injection condition with a low bias current
, the magnitude of spin signals can be enhanced despite the rise of temperature . for the interpretation of the temperature - dependent spin signals ,
it is important to consider the sensitivity of the spin detection at the schottky - tunnel - barrier contact in addition to the spin diffusion in si . |
the first topic is the electromagnetic field inside a hollow cylinder cavity with conducting walls and the spectrum of resonance frequencies associated to the various modes of the field .
we will assume that the field has a monochromatic time dependence @xmath0 .
the symmetry axis will be taken as the z - axis of the coordinate system . with this choice
, the field will split in two orthogonal components as @xmath1 the walls of the cavity are situated at @xmath2 , @xmath3 and @xmath4 .
the tangential component of the field should vanish on the conducting walls , that is @xmath5 on the cylinder surface @xmath2 and @xmath6 at @xmath7 ( fig.1 ) .
inside the hollow cavity the field will obey the wave equation @xmath8 or @xmath9 for the case of harmonic field of frequency @xmath10 .
the axial component of the space part of the electric field obeys the same equation : @xmath11 and so does the transverse part @xmath12 .
the equation can be written in cylindrical coordinates as @xmath13 and assumes a separate solution of the type @xmath14 .
thus , the helmholtz pde above becomes @xmath15 or @xmath16 the reflections at the top @xmath17and bottom @xmath3 surfaces impose a z - dependence appropriate for standing waves .
thus , in the eq . ( 1.6 ) the constant @xmath18 and the solution @xmath19 will be a combination of @xmath20 which will obey the boundary conditions .
further , in order to have a uniform @xmath21 dependence , @xmath22 should be of the form @xmath23 with integer @xmath24 thus , the radial equation becomes @xmath25 where @xmath26 .
the equation ( 1.7 ) is the bessel equation of order @xmath27 and the solution is a combination of bessel and neumann functions of order m , @xmath28 . if we take into account that the solution should be finite for any @xmath29 , including @xmath30 and remember the behavior of bessel and neumann functions in the vicinity of the origin @xmath31 respectively @xmath32 for @xmath33 and @xmath34\ ] ] than only @xmath35 is acceptable , because the second solution is singular at the origin . in the case of transverse magnetic ( tm )
modes @xmath36 everywhere in the cavity , while the transverse components of the electric and magnetic field are @xmath37 and , respectively @xmath38 where @xmath39 is the hamilton s operator in polar coordinates . than @xmath40 must assume the form @xmath41 in order that @xmath42 to satisfy the boundary conditions at @xmath43 .
also , the boundary condition @xmath44 for @xmath2 requires that the argument @xmath45 should be one of the zeros of the bessel function @xmath46 .
thus , @xmath47 will assume only a set of discrete values @xmath48 where @xmath49 is the n - th zero of @xmath50 .
the general solution of the field equations ( 1.10 ) inside the cavity that obeys the appropriate boundary conditions will be a superposition of the normal modes : @xmath51 the resonance frequencies of the cavity will depend on three parameters and will be given in the case of tm modes by @xmath52 with m=0,1,2 , ... n=1,2 , ... p=0,1,2 , ...
the second topic is the electromagnetic field inside a hollow cavity of the form of a cylindric ring with conducting walls , with inner radius @xmath53 , outer radius @xmath54and height @xmath55 .
the pde and the boundary conditions that the fields obey will be the same as in the case of the cylinder cavity , except that a supplementary boundary at @xmath56 appears .
since the origin @xmath30 no longer belongs to the domain of values that @xmath57 assumes , than the particular solution of the radial equation will be now a linear combination of both bessel and neumann functions : @xmath58 in order that the field to satisfy homogenous boundary conditions on both @xmath59and @xmath2 boundaries , the radial funtion above should assume the form @xmath60\ ] ] where @xmath61 are the solutions of the equation @xmath62 they will be determined by appropriate numerical methods . in the case of tm modes , the general solution for the axial component
@xmath40 is the linear combination : @xmath63e^ { \pm im\phi } \cos \left ( { p\pi \frac{z}{l } } \right ) } } } \ ] ] while the resonant frequencies are @xmath64 with m=0,1,2 , ... n=1,2 , ... p=0,1,2 , ...
the presented formulae allow an exact calculus of the electric field inside axial cavities with conducting walls , which may be used for a rigorous evaluation of the radiation thermal flux in such a system . for cryogenic systems ,
the numerical calculations need such accurate formulae , taking into account that a large number of coaxial insulators will be used and any errors will propagate and will be recursively amplified . the only way to minimize the global errors is to keep them as low as possible at any stage , thus involving from the beginning these exact analytical solutions .
that is why the numerical calculus of the special functions involved must be carefully conducted , implying an increased computing effort in order to precisely describe the total radiation heat transfer rate in cryogenic installations .
the bessel functions @xmath65are solutions of bessel equation @xmath66r(x ) = 0\ ] ] and assume the following series expansions @xmath67 @xmath68 they satisfy the following orthogonality relation on the interval @xmath69 $ ] @xmath70 ^ 2 \delta _ { nk}\ ] ] the above solutions are independent for all @xmath71 except the case of integer @xmath72 when @xmath73 in this case a second independent solution of bessel equation is the neumann function @xmath74 both bessel and neumann functions as well as their combinations @xmath75 and @xmath76 i.e the hankel functions , satisfy the recursion formulae @xmath77 and @xmath78 where x(x ) denotes a bessel , neumann or hankel function .
this work was partially supported by the romanian national research authority ( ancs ) under grant 22 - 139/2008 .
\1 ) j. d. jackson , classical electrodynamics , 3-rd ed .
john wiley and sons , 1999 chapters 3 and 8 . | * abstract * in the present article we intend to accomplish a rigorous analysis which allows to elaborate useful symbolic and numerical codes leading to an accurate evaluation of the thermal radiation .
we will consider the case of the electromagnetic field in axially symmetric systems , a case of real interest in cryogenics . |
classification of a star as a be star has long been recognised as consignment to a loosely defined phenomenological group , rather than as a definition of the evolutionary status of the star .
presently the be stars may be divided into three main groups : ( i ) the classical be stars , generally thought of as the most rapidly rotating near main sequence b stars ( see e.g. slettebak , 1988 ) ; ( ii ) the herbig be stars , first identified by herbig ( 1960 ) as stars in the b spectral type range whose association with star - forming regions and emission line character might indicate that they are very young ; ( iii ) the b[e ] stars ( allen & swings , 1976 ; zickgraf et al . 1985 ) , noted for the presence of forbidden line as well as hi emission in their spectra and strong ir continuum excesses ( also seen in herbig be stars ) .
while it was initially thought that b[e ] stars are preferentially supergiants , recent work ( gummersbach et al .
1995 ) has demonstrated from deep lmc observations that b[e ] characteristics may also be seen at significantly lower luminosities .
it is now widely accepted that classical be stars are encircled at their equators by ionized , low opening angle , almost keplerian , disks , that are optically - thick in h@xmath3 . by contrast , the circumstellar geometry of herbig be and b[e ] stars is rather more of an open question . in this work , we will open up another avenue for exploring this issue .
we present medium resolution spectropolarimetry across the h@xmath3 line , a technique that , in the case of the detection of polarization changes across the line can provide an answer to the most basic question `` is the ionized material around these stars spherically symmetric or not ? '' . by comparing h@xmath3 polarization with that of the continuum one can exploit the fact that line and continuum respectively form within a larger and smaller volume and
subsequently ` see ' different scattering geometries .
essentially , h@xmath3 is not significantly scattered by the ionized envelope in which it forms , whereas the continuum arising primarily from the central star embedded in the envelope undergoes electron scattering . in the case
that the ionized envelope s projection on to the plane of the sky is non - circular , a net linear polarization is imprinted on the continuum light , but not on h@xmath3 producing a drop in the polarization percentage across the line ( ` line - effect ' ) .
the addition of further continuum polarization by either a dusty envelope or the ism modifies this change and may even produce a net percentage rise across the line
but , significantly , it can not nullify the change .
for example , schulte - ladbeck et al . (
1994 ) showed that the h@xmath3 emission line of ag car displayed enhanced polarization at one epoch while on other occasions a de - polarization across the line was observed .
after the correction for isp however , h@xmath3 was de - polarized with respect to the continuum on all occasions .
the advantage of spectropolarimetry over broadband polarimetry is that a result can be obtained even where it is not possible to distinguish the various contributions to the total continuum polarization .
furthermore , by spectrally - resolving the h@xmath3 line profile one can hope to pick out more subtle effects arising in cases where the assumption that the h@xmath3 emission is unscattered , and hence unpolarized , breaks down .
qualitatively these were demonstrated in model calculations by wood , brown & fox ( 1993 ) .
when there is significant scattering of h@xmath3 the line profile in linear polarized light becomes a probe of the velocity field in the electron - scattering medium . using this tool we have already shown in the case of the b[e ] star , hd 87643 , that there is direct evidence of a rotating and expanding outflow ( oudmaijer et al 1998 ) .
the pioneering work in this area was made in the seventies , when clarke & mclean ( 1974 ) , poeckert ( 1975 ) and poeckert & marlborough ( 1976 , hereafter pm ) conducted narrow - band polarimetric studies of be stars that compared the linear polarization on and off h@xmath3 .
many instances of line de - polarization were found , showing that the envelopes of be stars do not project as circles onto the sky . after this time
, polarimetric studies were made of several classes of object , but due to observational difficulties it remained a specialist activity .
however in the past few years there has been rising interest in the technique .
spectropolarimetry has been performed on several strong h@xmath3 emitting evolved stars , such as ag car and hr car ( schulte - ladbeck et al .
1994 , clampin et al .
1995 ) , where the position angle of the spatially - unresolved flattened electron scattering region has been shown to agree with the observed extension of the optically visible nebulae surrounding these objects .
both the b[e ] and herbig be stars are ideal objects to subject to this style of observation , since they are strong h@xmath3 emitters and often optically bright enough to render studies at medium resolution with high photon counts feasible with 4-meter class telescopes .
furthermore there is a clear need for this type of observation since a change in the linear polarization across h@xmath3 can be the only direct evidence of electron scattering operating on the scale of a few stellar radii as opposed to polarization by a dusty envelope . in the first instance , the observations presented here were motivated by the aim of examining herbig be stars for the presence of ionized circumstellar disks .
these reputedly intermediate mass objects present a phenomenology that suggests they are approaching or have recently achieved a main sequence location on the hr diagram
they are the higher mass counterparts of the t tauri stars .
the paradigm for star formation invokes a collapsing cloud and conservation of angular momentum that results in the formation of a flattened circumstellar ( accretion ) disk , that eventually accretes or is blown away by an outflow .
however there is not yet a consensus that accretion disks are commonly associated with the known herbig be ( and ae ) stars .
there is a certain irony that t tauri stars , their lower mass counterparts , are generally accepted to have disk - like envelopes ( e.g. hh30 in burrows et al .
1996 ) , while evidence is accumulating that their higher mass counterparts , the optically obscured massive young stellar objects ( ysos ) are also surrounded by disk - like structures ( e.g. hoare & garrington 1995 , and references therein ) .
merlin radio data on mwc 297 , a nearby radio - bright early herbig be star , also reveals an elongated ( but ionized ) structure on a spatial scale of @xmath6100 au ( drew et al .
more direct high resolution imaging is clearly worthwhile and of course spectropolarimetry can help identify interesting targets .
nevertheless , at the present , there persists a debate that the observed spectral energy distributions require , on the one hand , dusty disks ( e.g. malfait , bogaert & waelkens , 1998 ) or , on the other , that they can be fit satisfactorily by spherically symmetric dusty envelopes ( miroshnichenko , ivezi & elitzur , 1997 ; see also the overview of this issue in pezzuto , strafella & lorenzetti 1997 ) . the recent direct detection of a rotating disk around a herbig ae star by mannings , koerner & sargent ( 1997 ) indicates that at least some of these objects have disk - like geometries .
broad - band polarimetry of a number of herbig stars has revealed variability of the polarization of the objects which could imply deviations from spherical symmetry of the dusty envelopes ( e.g. grinin et al .
1994 , who studied ux ori ; jain & bhatt , 1995 ) .
by contrast , the h@xmath3spectropolarimetry traces scales even closer to the star , the ionized material . with regard to the b[e ] stars , first picked out by allen & swings ( 1976 ) , the argument for embedding them in disk - like equatorial structures has largely been won in that there is widespread acceptance of zickgraf s phenomenological model ( zickgraf et al 1985 , 1986 ) .
this is because there is compelling spectral evidence of a fast , presumably polar , wind at uv wavelengths , that combines with a high emission measure , much more slowly expanding , presumably equatorial , flow traced by strong optical emission lines .
broad - band polarimetry by zickgraf & schulte - ladbeck ( 1989 ) and magalh@xmath7es ( 1992 ) indicate that for a sub - sample of b[e ] objects , the circumstellar dust , located at larger distances from the star , is distributed in a geometry deviating from spherically symmetric .
the unresolved issue is how these axially - symmetric structures arise and indeed what the stellar evolutionary status of this object class really is .
the fact that b[e ] stars are far from being exclusively supergiants deepens the mystery . in this context , herbig s ( 1994 )
concern about the difficulty of distinguishing herbig be from b[e ] stars becomes all the more intriguing .
to progress in understanding how b[e ] disks arise , a more complete description of the disk density and velocity field is highly desirable .
it is in this respect that h@xmath3 spectropolarimetry has the potential to provide unique insights . because of the problems of distinguishing between the b[e ] and herbig be categories
, there is always a significant probability that a herbig be sample contains some b[e ] stars .
indeed , for galactic b[e ] stars it is often difficult to determine whether an object is a luminous evolved object or a less luminous pre - main sequence object ( see e.g. the discussions on hd 87643 ; oudmaijer et al.1998 , mwc 137 ; esteban & fernndez 1998 , and hd 45677 ; de winter & van den ancker 1997 ) .
here we exploit this in that our programme of h@xmath3spectropolarimetry programme includes as targets relatively clear - cut examples of post main sequence b[e ] stars alongside undisputed herbig be stars and objects that might be either . in this paper
we give an overview of our observing campaign to date . in section 2 , the way in which targets were selected and the observations are discussed .
the results and their interpretation are presented on a case - by - case basis in sec .
sec . 4 contains a discussion on the power of spectropolarimetry and what we have learned from this program .
we conclude in sec . 5 .
[ cols="<,<,<,<,<,<,<,<",options="header " , ] , + continuum defined from 6400 - 6500 and 6600 - 6700 @xmath8 .
the errors on the equivalent widths of the h@xmath3 lines are expected to be of order 5% , the errors cited for the polarization data are based on photon - statistics only , and rounded upwards .
systematic ( i.e. external ) errors in the polarization are estimated to be of order 0.05% 0.1% .
the optical linear spectropolarimetric data were obtained using the rgo spectrograph with the 25 cm camera on the 3.9-metre anglo - australian telescope during three observing runs in january 1995 , december 1995 and december 1996 respectively . during the first two runs ,
the weather provided some spectacular views of lightning from the telescope , but only limited data . during clear time
, we aimed at observing the brightest objects in order to make the best of lower - than - desired count rates .
nevertheless , the resulting polarization measurements proved to be very stable .
the last run was mostly clear , opening the way for time to be spent on some of our fainter targets .
the instrumental set - up was similar during all observing runs and consisted of a rotating half - wave plate and a calcite block to separate the light into perpendicularly polarized light waves .
two holes of size 2.7 arcsec and separated by 22.8 arcsec in the dekker allow simultaneous observations of the object and the sky .
four spectra are recorded , the o and e rays of the target object and the sky respectively .
one complete polarization observation consists of a series of consecutive exposures at four rotator positions . per object ,
several cycles of observation at the four rotator positions were obtained in order to check on the repeatability of the results .
indeed , we find that multiple observations of the same star result in essentially the same polarization spectrum .
to prevent the ccd from saturating on the peak of h@xmath3 , shorter integration times were adopted for those objects with particularly strong h@xmath3emission .
spectropolarimetric and zero - polarization standards were observed every night .
a 1024 @xmath9 1024 pixel tek - ccd detector was used which , combined with the 1200v grating , yielded a spectral range of 400 @xmath8 , centered on h@xmath3 .
wavelength calibration was performed by observing a copper - argon lamp before or after each object was observed . in all observations reported here a slit width of 1.5@xmath10
a log of the observations is provided in table 1 .
bias - subtraction , flatfielding , extraction of the spectra and wavelength calibration was performed in iraf ( tody 1993 ) . the resulting spectral resolution as measured from arc lines
is 60 km s@xmath2 .
the e and o ray data were then extracted and imported into the time series / polarimetry package ( tsp ) incorporated in the figaro software package maintained by starlink .
the stokes parameters were determined and subsequently extracted . a slight drift of a few degrees in position angle ( pa )
was calibrated by fitting its wavelength dependence in nightly 100% polarized observations of bright unpolarized stars ( obtained by inserting an hn-22 filter in the light - path ) and removed from the polarization spectra .
the instrumental polarization deduced from observations of unpolarized standards proved to be smaller than 0.1% in all cases . in july 1998 , mwc 297 was observed in service time with the isis spectrograph and polarimetric optics on the 4.2 m william herschel telescope , la palma .
the instrumental set - up included the 1200r grating and a 1124@xmath91124 tek2 detector , providing a wavelength coverage of 400 @xmath8 around h@xmath3 and a spectral resolution of 40 km s@xmath2 .
the data reduction was the same as for the aat data .
polarization accuracy is in principle only limited by photon - statistics .
one roughly needs to detect 1 million photons per resolution element to achieve an accuracy of 0.1% in polarization ( the fractional error goes at @xmath6 1/@xmath11 ) .
however , although it is probably fair to say that the internal consistency of a polarization spectrum follows photon - statistics , the external consistency ( i.e. the absolute value for the polarization , checking for variability ) is limited due to systematic errors . for example , when calculating the polarization of a given spectral interval one can reach polarization percentages with a statistical error of several thousandth of a percent .
however , instrumental polarization ( less than 0.1% ) , scattered light and low - level intrinsic variability of the polarization standards may influence the zero - points .
the quality and amount of data taken of spectropolarimetric standard stars is at present not yet sufficient to reach absolute accuracies below the 0.1% mark ( see manual by tinbergen & rutten , 1997 ) .
a feeling for the possible accuracies in our data can be obtained by studying some of the objects that have been observed on different occasions .
seven , respectively 6 independent observations of the be star hd 76534 and the polarization standard hd 80558 yield a mean polarization and rotation of ( 0.49 % with an r.m.s .
scatter of 0.03 % , 124@xmath12with a scatter of 3@xmath12 ) and ( 3.19 @xmath13 0.11% , 162 @xmath13 1.6@xmath12 ) respectively .
it is encouraging to note that our independent continuum measurements stretching over more than a year are mostly within 0.1% , and often within 0.05% , in polarization .
some h@xmath3 parameters and continuum polarizations are presented in table 2 . in the following ,
the results across the full range of targets observed are summarized .
these are grouped such that we begin with those objects showing no discernable polarization changes across h@xmath3 ( 3.1 ) , and then move on to objects which do show percentage changes and/or rotations ( 3.2 ) .
unless specifically stated , we have made no attempt below to correct for the interstellar polarization ( isp ) .
this decision is based on the following : the main goal of this study is the detection of polarimetric changes across the h@xmath3 line . since the wavelength dependence of the interstellar polarization only becomes apparent on wavelength ranges larger than our spectra provide , the isp will only contribute a constant polarization vector in ( q , u ) space to the observed spectra .
a further reason to refrain from isp corrections here , is that the methods commonly used for this ( field - star method and continuum variability , see e.g. mclean & clarke 1979 ) do not always return unambiguous values .
however , in the absence of isp correction , it is useful to remember the point raised in the introduction that the isp can change what might otherwise be a reduction of the linear polarization percentage across the h@xmath3 line into an increase in polarization , or an apparently constant polarization , but accompanied by a significant rotation in the position angle .
the same effect can occur in the event of additional polarization due to circumstellar dust . nevertheless , regardless of the influence of the isp and polarization due to circumstellar dust , it is possible to derive the _
intrinsic _ angle of the electron - scattering material ( e.g. schulte - ladbeck et al.1994 ) . assuming the line is depolarized , the vector connecting the line- and continuum polarization in the qu plane will have a slope that is equivalent to the intrinsic angle of the scattering material responsible for the continuum polarization .
since the wavelength dependence of both circumstellar dust polarization and isp is small , they add only a constant qu vector to all points in both line and continuum , and thus will not affect the difference in line - to - continuum polarization .
this slope is measured as @xmath14 = 0.5@xmath9atan(@xmath15u/@xmath15q ) . in this subsection
we discuss the objects that do not show a line - effect . in principle
such an observation implies that the projection of the ionized region on the plane of the sky is ( mostly ) circular .
we will find that this does not necessarily have to be the case .
the objects falling into this group are lk h@xmath3 218 , hen 3 - 230 , as 116 , hd 52721 , v380 ori , hd 76534 , @xmath4 ori and mwc 297 .
their polarization spectra are shown in figure 1 .
[ [ hen3 - 230-as116-lkhalpha218 ] ] hen 3 - 230 , as 116 , lk h@xmath3 218 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + these objects are relatively faint targets for which the signal - to - noise ratios in our data are not so high . hence , the absence of change across h@xmath3 for the time being should be viewed as an absence of any marked contrast .
for example in the case of lk h@xmath3 218 , there seems enhanced polarization at the position of the stronger redshifted emission component in the line .
however this is not strictly even a 2@xmath16 detection .
coarser binning can yield a 3@xmath16 enhanced polarization in the line ( at smaller resolution this is only present in one pixel however ) but in truth it would appear that 100 minutes exposure is not enough for this object .
the null results for hen 3 - 230 and as 116 are more sure .
both targets have extremely bright line emission . as previous observations of them
are extremely sparse , their evolutionary status remains undetermined .
based on its low excitation spectrum stenholm & acker ( 1987 ) argue that hen 3 - 230 is not a planetary nebula , despite having figured in many previous papers to be one . as 116
was appeared in a catalogue of emission line stars of miller & merrill ( 1951 ) , since then not much work has been published .
the iras flux peaks at 25@xmath17 m , which could point at a detached dust shell , but it was not detected in the oh maser by blommaert , van der veen & habing ( 1993 ) . [ [ hd52721-v380ori ] ] hd 52721 , v380 ori + + + + + + + + + + + + + + + + + + these are two quite convincing examples of no line effect .
both , nevertheless , present significant continuum polarizations . since hd 52721 presents little of an infared continuum excess ( hillenbrand et al .
1992 ) , it might seem plausible that this star is a low - inclination classical be star behind a significant interstellar column . indeed the single - peaked h@xmath3 emission is consistent with this , but in contrast to the @xmath18 measurement of 400@xmath1340 km s@xmath2 reported by finkenzeller ( 1985 ) suggesting high inclination .
v380 ori exhibiting a strong infrared continuum excess , would appear to be optically - veiled and hence sits more convincingly in the haebe object class .
the absence of a line - effect in v380 ori may simply imply lower inclination to the line of sight .
[ [ hd-76534 ] ] hd 76534 + + + + + + + + the initial 1995 data on this source have already been presented by oudmaijer & drew ( 1997 ) .
there it was shown that these data did not indicate any changes across h@xmath3 .
our new data confirm this and show no hints of pronounced polarization variability in its modest @xmath60.5% level .
this is despite the source s propensity for spectral variability clearly illustrated by the 11 january 1995 transformation of h@xmath3 absorption into well - developed double - peaked emission within hours ( oudmaijer & drew 1997 ) .
table 2 lists the h@xmath3 equivalent widths at the various occasions that the object was observed .
the ew changes strongly , but situations similar to the january 1995 data were not observed .
[ [ omegaori ] ] @xmath4 ori + + + + + + + + + + + + + + + + + + + + + + + + it is not clear whether @xmath4 ori should be considered a herbig be star , or simply a classical be star ( sonneborn et al .
the absence of a detectable change across h@xmath3 ( fig .
[ nondet ] ) stands in contrast to reports in the literature that the hydrogen recombination lines show de - polarization
pm find that h@xmath3 shows a polarization dip while clarke & brooks ( 1984 ) find the same in h@xmath19 .
this difference is presumably connected with the stronger h@xmath3 emission reported by pm ( line to continuum ratio of 1.8 versus our figure of 1.4 - which was not binned to the same narrow band , and is thus a strong upper limit ) , and higher linear polarization ( 0.38% versus our 0.30% ) . since classical
be stars and , indeed , herbig be stars are known to be emission line variables , this change is probably due to a lowering of the ionized emission measure of the equatorial disk around this star . given the great disparity between the thomson scattering and h@xmath3 absorption cross - sections , a relatively modest drop in the h@xmath3 equivalent width could well be accompanied by a collapse in the percentage of linearly - polarized scattered starlight .
hence it would seem that isp contributes around 0.3% linear polarization in @xmath4 ori , a figure not out of line with pm s estimate of 0.24% .
[ [ mwc297 ] ] mwc 297 + + + + + + + the weak - to - non - existent effect across h@xmath3 is startling in view of the evidence gathered by drew et al .
( 1997 ) that this early herbig be star is viewed at relatively high inclination .
furthermore , the 5 ghz radio image ( see drew et al .
1997 ) , which provides an extinction - free view of the ionized circumstellar medium around mwc 297 , indicates an elongated geometry that would suggest a line effect ought to be apparent in such a bright emission line source .
although at first sight very surprising , we consider two different hypotheses that may explain this apparent paradox .
firstly , we can conclude we are seeing the h@xmath3 line directly , and that the line - forming region is indeed round .
since the h@xmath3 line is formed in a potentially much smaller volume than the continuum 5 ghz radiation , the rounder appearance of the h@xmath3 line - forming region indicates that the geometry changes between the near - stellar scale and the larger scale sampled at radio wavelengths .
spatial evolution of this type has been predicted for lower mass stars ( see frank & mellema 1996 ) .
secondly , it may be that h@xmath3 is formed in an edge - on disk like structure , but that the optical light does not reach us directly , and is completely obscured in the line - of - sight .
the light that we see could then be ` mirrored ' by scattering dust clouds located above and/or beneath the obscuring material .
if the scattering dust - clouds ` see ' a nearly circularly symmetric h@xmath3 emitting region , it will not see any de - polarization across the line either .
consequently , the light reaching us will not show any polarization changes across h@xmath3 .
that dust - scattering plays a role in this object is already suggested by the spectral energy distribution , which shows a notable excess in the @xmath20-band ( bergner et al . 1988 ; hillenbrand et al 1992 ) , possibly due to the ` blueing ' effect .
we may therefore have a similar situation to that in the red rectangle ( see e.g. osterbart , langer & weigelt 1997 , waelkens et al 1996 ) , where it was only recently realized that the central star is actually not the star itself , but its reflection against dusty knots located above and below a very optically thick dust lane .
this finding explained the long - standing problem of the energy balance ; the apparently absorbed light from a star with such a modest reddening ( @xmath21 of order 1 ) is orders of magnitude less than that being re - radiated in the infrared .
if the circumstances are similar in mwc 297 , the reddening ( @xmath22@xmath1 8 - see discussion of drew et al , 1997 ) commonly assigned to this source on the basis of conventional extinction measurements is a severe underestimate .
this would not be completely unexpected , as mwc 297 is in certain respects an intermediate object between the optically visible herbig be stars , and their more massive counterparts , the becklin - neugebauer ( bn ) type objects , which suffer from large optical extinctions ( @xmath22 often in excess of 20 ) .
while in the optical , mwc 297 has much in common with the herbig be stars , at infrared and radio wavelengths it shows evidence of substantial mass loss associated with bn - type objects ( drew et al .
1997 ) . here
we present the objects for which the line - effect is observed .
first , the herbig be stars in this sample are discussed , hd 259431 , hd 53367 and hd 37806 , then mwc 137 , a herbig be star that has been recently proposed to be a massive evolved b[e ] star instead is shown , and we end with the well - known b[e ] objects hd 50138 , hd 87643 and hd 45677 .
[ [ hd-259431 ] ] hd 259431 + + + + + + + + + we start with the least certain detection , hd 259431 .
this object ( fig .
[ 259533 ] ) shows a hint of de - polarization across h@xmath3 , from ( 1.1% , 102@xmath12 ) to 0.8% in the line center .
the intrinsic polarization angle in qu space measured from the change from the continuum to line polarization , @xmath14 = 0.5@xmath9atan(@xmath15u/@xmath15q ) , gives 17@xmath12 , but with a large uncertainty .
the length of this vector is small , of order 0.3% . our two observations , taken one year apart
, do not show any variability within the small error - bars .
the compilation by jain & bhatt ( 1996 ) which contains broad - band polarimetric observations only hints at a slight variability .
the high resolution data , only when binned to errors of 0.1% or less , show the line - effect . [
[ hd-53367 ] ] hd 53367 + + + + + + + + the polarization spectrum of hd 53367 is shown in fig .
[ 259533 ] .
since there was no difference within the errorbars of the data taken on 2 consecutive nights , these were added to increase the signal - to - noise ratio .
the h@xmath3 line is clearly de - polarized with respect to the continuum , while no rotation across the line is present .
the intrinsic polarization angle in qu space measured from the slope is 47@xmath12 .
if the line center is assumed to be completely de - polarized , one can use this information to correct the observed polarization for the isp ( the ` emission line method ' ) .
reading off the polarization in the line - center ( q = 0.0% , u = 0.3% ) , and subtracting this value from the spectrum then gives an intrinsic polarization of 0.2 @xmath13 0.01 % and a position angle of 44.5 @xmath13 0.5@xmath12 ( measured in the bins 6400 6500 @xmath8 and 6700 6800 @xmath8 .
note that the error - bar reflects the internal consistency and not the external consistency ) , consistent with the slope in the qu plane .
this low value of intrinsic continuum polarization is what one would expect from modest electron - scattering ( see e.g. pm ) .
let us now comment on the significance of the measured polarization angle . on the sky ,
hd 53367 is located on the periphery of the canis majoris ` hole ' noted for its low reddening sightlines ( welsh 1991 ) .
herbst , racine & warner ( 1978 ) designated this star a member of the cma r1 cluster , which they placed at a distance of 1150 pc , the same distance as to the cma ob1 association ( claria 1974 ) .
this distance is suspect as it makes hd 53367 more luminous than a supergiant at the same b0 spectral type .
a more reasonable distance estimate would be around 550 pc ( adopting a dereddened @xmath23 magnitude of @xmath24 ; herbst et al .
1982 , and @xmath25 ; for a b0iv star , schmidt - kaler 1982 ) .
it is then less surprising that the hipparcos catalogue ( esa , 1997 ) contains a finite , although very uncertain , parallax measurement for this star ( see also van den ancker , de winter & tjin a djie 1998 ) . in any event there is strong evidence that the cumulative interstellar extinction towards cma r1 is not more than @xmath26 , implying that the remaining observed extinction is local ( herbst et al .
1982 ; vrba , baierlein & herbst 1987 ) . more important , vrba et al .
( 1987 ) demonstrate quite convincingly that within cma r1 the polarization angle tends to follow the sweep of the southern dust arc up through sh2 292 , the hii region ionized by hd 53367 , and that the polarization is most likely due to grain alignment . at hd 53367
this angle is about 40@xmath12 and entirely comparable to that of the non - emission line b3v star bd -10@xmath271839 just @xmath620 arcmin away , and also at a photometric distance of about 600 pc .
the interesting feature of hd 53367 is that the separable intrinsic and foreground polarization angles are the same as indicated by the lack of rotation across h@xmath3 in the observed spectrum .
this suggests an orderly star formation process in which the rotation axis of hd 53367 ` remembers ' the larger scale circumstellar field direction , in preference to a more dynamical mechanism such as the ` accretion induced collision ' merger model of bonnell et al ( 1998 ) , which would result in randomly oriented polarization angles .
[ [ hd-37806 ] ] hd 37806 + + + + + + + + the double peaked h@xmath3 profile of this object is remarkably different during our two observing epochs ( fig .
[ h378 ] ) . in december 1996 , both peaks were equally bright , while in january 1995 the blue peak is much weaker .
although the signal - to - noise of the january 1995 data is not high , it appears that the object does not show significant changes in polarization . but
a rotation is present , which is especially visible in the december 1996 data when the source was observed for longer .
the rotation almost exactly occurs in the central dip in h@xmath3 , rather than across the entire adequately - resolved line profile .
the fact that we observe only rotation is interesting in its own right , as the principle of line de - polarization without superimposed foreground polarization would imply a constant angle across h@xmath3 .
clearly there is isp and perhaps circumstellar polarization present .
if we assume that the underlying , rotated part , of the line - profile is unpolarized , we may attempt a correction for the intervening interstellar- and circumstellar dust polarization , to retrieve the intrinsic spectrum of this object .
the ( q , u ) vector measured in the central dip of the rotation , corresponds to ( -0.06%,0.35% ) or a polarization of 0.36% with pa 50@xmath12 . subtracting these values from the observed spectra results in fig.[h378int ] .
this figure also shows the ` polarized flux ' ( polarization @xmath9 intensity ) .
the polarized flux indicates that the double - peaked h@xmath3 line has the same polarization as the continuum , but by virtue of the manner in which the isp was corrected for the central dip between the peaks is de - polarized .
the 1995 spectrum shows the same behaviour . despite its much lower signal - to - noise
, the large red / blue ratio has virtually disappeared in the polarized flux spectrum , suggesting a large part of the red peak is not associated with the line - forming region responsible for the double peaks .
a consistency check can be made as to whether the choice of isp for the correction is reasonable .
we have searched the matthewson et al.(1978 ) catalogue for field stars nearby the object and found 49 objects within a radius of 180 arcmin ( one object with p = 12% was excluded for the analysis ) .
the catalogue also provides photometric estimates of the total extinction @xmath22 to these objects . a relatively tight relation exists between the observed polarization and @xmath22 .
a least - squares fit to the data gives the relation p ( % ) = 1.5 @xmath9 @xmath22(mag . )
the pa shows mostly a scatter diagram , and gives a mean of 76@xmath12@xmath13 39@xmath12 for the total sample .
the @xmath22 towards hd 37806 is ambiguous , but likely to be low : van den ancker et al .
( 1998 ) reclassify hd 37806 as an a2vpe star and give its extinction as @xmath28 .
in contrast , malfait , bogaert and waelkens ( 1998 ) find an _
v ) _ of 0.14 ( @xmath22= 0.43 if the ratio of total to selective reddening , @xmath29 , is 3.1 ) for a b9 spectral type . based on the extinction and the field stars ,
the isp towards hd 37806 should be between 0.1% and 0.7% .
the ` emission line ' method gives 0.36% , a value that is at least consistent with the value returned from the field stars .
taking the derived intrinsic polarization spectrum at face value , it appears that the h@xmath3 line profile is a composite of two unrelated components : a double - peaked polarized component and a single , unpolarized component .
since both the photospheric continuum radiation and the double - peaked component are equally polarized , it would appear that they both appear point - like to the scattering material , while the single component is formed further away from the star , betraying the asymmetry of the scattering region .
perhaps due to the signal - to - noise in the data , no de - polarization is visible in the blue peak .
this could suggest that the de - polarization in the red peak is not necessarily due to electron scattering , since one would expect the electron scatterers to be located close to the star .
instead , the data do not exclude the possibility that the red peak is located in an extended nebula ( which is not resolved in our data however ) , while the underlying broader emission and the photosphere are polarized by circumstellar dust , which , by implication , is not distributed spherically symmetric around the star . intriguingly , the line - to - continuum ratios of the red peak are constant ( see table 2 ) , while broad - band photometry of this object also appears constant ( van den ancker et al .
1998 ) so the blue peak has increased in strength .
this fact , combined with the relatively equal red / blue ratio of the double - peaked line in polarized flux , suggests that the polarized red part of the line has also become stronger .
this leads to the enigmatic situation that the polarized part of the red peak increased in strength while the unpolarized part of the red peak decreased in strength in such a way that their total has remained constant in time .
clearly , this object needs further study , both spectropolarimetric , and from a modelling perspective to gain more understanding as to the origin of the observed polarization .
[ [ mwc-137 ] ] mwc 137 + + + + + + + although th et al . ( 1994 ) labelled this source as a probable herbig be star , in a recent study esteban & fernndez ( 1998 ) argued that is much more likely to be an evolved b[e ] supergiant .
their arguments , based on a kinematical association with molecular clouds at more than 5 kpc , are reasonably convincing , but we note that the nature of this source has long been controversial ( see references in esteban & fernndez ) . the polarization spectrum and qu diagram of mwc 137 are shown in fig .
the continuum polarization of the object is very large ( 6% ) , and slight polarization changes across h@xmath3 are visible .
much clearer is the broad , observed rotation of h@xmath3 , centered on the line peak .
the rotation is at most only 3@xmath12 , but it is real as is evident from the significant loop apparent between line and continuum in qu space ( fig .
[ m137 ] ) ; the shift is along the qu vector ( + 0.3% , + 0.5% ) , corresponding to an intrinsic angle in the continuum of 30@xmath12 and a depolarization of @xmath1 0.6% ( measured from the length of the polarization vector , @xmath30 ) .
the same situation as for hd 37806 described above occurs in this case : the polarization from intervening material has transformed a de - polarized line into a rotated line .
slight changes in the observed polarization are still present in the wings of the emission , suggesting a different polarization of the line wings than both the continuum or the centre of the emission line .
the interstellar reddening towards the object is very large .
it was redetermined by esteban & fernndez ( 1998 ) to be @xmath22 = 3.77 .
this is very high , but according to the authors consistent with a very large distance to the object ( the authors mention 6 kpc ) .
if true , then the large continuum polarization can be explained mostly by the interstellar reddening .
the polarization of the field stars within a radius of 300@xmath31 ( taken from matthewson et al .
1978 ) increases linearly with @xmath22 up to 4% at @xmath22 @xmath6 2.2 , the largest @xmath22among the field stars in the sample .
unfortunately no information for stars more distant or more reddened is available , but it is clear that a large isp can be expected for the object .
the rotation across the h@xmath3 line is best interpreted as a depolarization across the line , but modified by the intervening isp and circumstellar dust polarization . using the ` emission line ' method
, we find a dust polarization of p=6.2% , @xmath14=159@xmath12 , in agreement with the large expected isp .
the intrinsic pa of the electron scattering medium of 30@xmath12 appears to be parallel with the bright north - western component of the ring nebula around the object ( see again esteban & fernndez 1998 ) .
this may suggest that asymmetries at very small scales , traced by the electron scattering , and at large scales ( from the image ) are still aligned .
[ [ hd-50138 ] ] hd 50138 + + + + + + + + in their search for herbig ae / be stars , th et al . ( 1994 ) found a subsample of objects with many herbig characteristics , that nevertheless did not fulfil all their criteria .
based on the strong emission lines , they called this group ` extreme emission line objects ' .
most of these stars are also classified as b[e ] stars , because of the presence of forbidden lines in the spectrum .
hd 50138 is one of these .
our two observations of hd 50138 ( fig . [ h501pf ] ) , taken two years apart in january 1995 and january 1997 , show essentially the same behaviour .
the emission line is double peaked , with a large red to blue ratio of the peaks .
the velocity separation of the two peaks is 160 km s@xmath2 and the central minimum is at 18 km s@xmath2 ( helio - centric ) , which is 15 km s@xmath2 blueshifted from the forbidden [ o i ] line at 6363 @xmath8 , for which we measure a central velocity of 33 @xmath13 5 km s@xmath2(helio - centric ) , in agreement with the radial velocity determination by pogodin ( 1997 ) .
the h@xmath3 line shows strong de - polarization across the red peak , while the blue peak shows at most a slight de - polarization .
the ` intrinsic ' polarization angle , as deduced from the shift between line and continuum in qu space is about 155@xmath12 , close to the measured one , implying that the isp , if any , has had no great effect on the observed polarization characteristics .
this is more or less in keeping with the moderate reddening towards this source ( van den ancker et al .
1998 assign @xmath32 , possibly an upper - limiting value considering that @xmath33 for this mid - b star is close to zero ) and the low interstellar polarization in the line of sight , as many objects around hd 50138 have very low polarizations for typical extinction values as 0.5 ( matthewson et al . 1978 ) .
the polarized flux spectrum ( polarization @xmath9 flux , fig . [ h501pf ] ) reveals equally strong blue and red peaks at both epochs .
this may indicate that part of the red emission is formed in the same region as the blue peak , such as a rotating disk , but that the excess emission compared to the blue line is formed in a larger volume , resulting in the observed de - polarization . the most straightforward explanation to account for the ` excess ' flux in the red peak , as in the case of hd 37806 , is that the intensity spectrum is a composite of a rotating disk type of geometry close to the star , and an additional , extended single peaked component .
a clue to the line formation may be provided by the spectrum taken around the lower opacity h@xmath19 line by jaschek & andrillat ( 1998 ) .
their published h@xmath19 profile shows the blue and red peaks roughly equal , with some underlying photospheric absorption still visible .
the much larger red - to - blue ratio in h@xmath3 could imply that the excess red emission in h@xmath3 is optically thin - since for thin hi emitting gas the balmer decrement may be substantially steeper than for optically - thick gas .
the double - peaked part of the line could then be an optically - thick line formed very close to the star , while the unpolarized line is optically - thin .
it is clear that any picture of this object must be simplified since the spectrum of hd 50138 shows many more peculiarities .
grady et al . ( 1996 ) include this star in their sample of objects exhibiting @xmath19 pic type infall phenomena .
pogodin ( 1997 ) drew attention to the variable hei @xmath345876 line profile , which sometimes shows an inverse p cygni behaviour .
he also attributed this to infall of material .
[ [ hd-87643 ] ] hd 87643 + + + + + + + + + hd 87643 is a b[e ] star , for which evidence suggests that it is located at several kpc , indicating a massive and evolved nature of the object .
this , and its spectroscopic and spectropolarimetric data have been analysed in more detail in oudmaijer et al .
( 1998 ) . for completeness
we show the data and the spectrum corrected for isp and circumstellar polarization in fig .
the polarization of this object shows some striking features .
after correction for the intervening polarization , it turns out that most of this structure appears an artefact due to the polarization vector additions ; the spectrum corrected for isp and circumstellar dust has a much smoother behaviour . a comparison with the schematic model calculations by wood
, brown & fox ( 1993 ) indicates that the polarization profile can be best reproduced with a circumstellar disk that is both rotating and expanding .
[ [ hd-45677 ] ] hd 45677 + + + + + + + + + from their uv - optical low resolution spectropolarimetry schulte - ladbeck et al . ( 1992 ) infer that hd 45677 is surrounded by a bipolar nebula . after correcting for isp ,
the polarization angle of the blue / uv spectrum they present is rotated by about 90@xmath12 with respect to the red part of the spectrum .
the explanation advanced for this behaviour is that the red emission is scattered through a dust torus , while the blue emission , to which the dust is optically thick , comes from a scattering bipolar flow , perpendicularly oriented with respect to the torus .
the spectra taken in january 1995 and december 1996 are shown in fig .
[ hispec ] . on both occasions
the polarization across h@xmath3 is enhanced with respect to the continuum .
the variability of the continuum polarization is very strong , with the polarization changing from 0.4% to 0.2% , while the position angle changed from 11@xmath12to 140@xmath12 .
the peak of h@xmath3 shows in both cases roughly the same polarization and position angle ( @xmath60.8% , 75@xmath12 ) .
although h@xmath3 is enhanced in the observed polarization spectrum , this does not necessarily imply that the intrinsic spectrum exhibits the same effect . in the following
we discuss the different elements , isp , circumstellar dust scattering and electron - scattering that shape the observed polarization spectra .
as explained below , we assume that the polarization changes across h@xmath3 are due to electron - scattering .
the spectra are plotted in qu space in fig .
both circumstellar dust polarization and isp add only a constant qu vector to all points . the continuum points cluster at different positions on the dates of observation , while the qu vectors across h@xmath3 are almost parallel .
a least squares fit through the qu - points between 6550 and 6570 @xmath8 , returns intrinsic polarization angles of 168 @xmath13 3@xmath12 and 163 @xmath13 3@xmath12 for the 1995 and 1996 spectra respectively . depending on the quadrant where the intrinsic qu vectors are located , these values could be rotated by 90@xmath12 and define a projected angle on the sky of @xmath1 75@xmath12 .
the length of the vector on both occasions corresponds to p @xmath1 0.95% , assuming the line to be unpolarized .
the electron - scattering region is thus aspherical , has a pa of @xmath6 75@xmath12 on the sky at a relatively constant amplitude of about 0.95% .
how does this relate to the view taken by schulte - ladbeck et al .
( 1992 ) of their similar detection of enhanced h@xmath3 linear polarization ? in their isp - corrected spectrum , h@xmath3 is still enhanced with respect to its adjacent continuum .
they explained this in terms of an ionized region much closer to the circumstellar dust than the stellar point source : the h@xmath3 line then sees a larger solid angle of scattering material and is thus more polarized than the continuum .
this seems to us an unlikely alternative to the conventional view that the ionized region , with a temperature of @xmath610000 k , is located within a very much smaller volume around the star than the dust , which should have an equilibrium temperature below @xmath61500 k , the dust condensation temperature ( see e.g. the spectral energy distribution modelling of hd 45677 by sorrell 1989 ) .
furthermore , the isp correction adopted by schulte - ladbeck et al.(1992 ) was the _ ad hoc _ value proposed by coyne & vrba ( 1976 ) on the basis that it should be comparable with the relatively steady observed polarization in the blue ( the red is much more variable ) .
whatever this correction actually does correspond to , there is no reason to suppose that it accounts for both the isp and circumstellar dust polarization .
if both sources of foreground polarization can be removed with confidence , only then can it be discerned whether there is an intrinsic linear polarization enhancement across h@xmath3 .
hence , for the timebeing , we retain the more conventional view that the observed polarization change across h@xmath3 is due to electron scattering in the ionized region , which may project as an oblate ` disk ' or as a prolate ` bipolar flow ' .
now we turn to consider the change in polarization of the continuum points .
as it is likely that the isp is constant in time , the additional , potentially variable , mechanisms involved are electron - scattering and circumstellar dust scattering . in the same manner as the polarization changes across
h@xmath3 move along a constant angle , temporal changes , due to a polarization mechanism that becomes stronger or weaker , may only affect the magnitude of polarization and not the pa .
if we adopt this principle here , the intrinsic pa of the polarizing material responsible for the change in continuum polarization is then @xmath6 20@xmath12 .
since the intrinsic pa of the electron scattering region is @xmath6 75@xmath12 , and its effect on the h@xmath3 line appears not to have changed significantly , the variable component does not seem to be electron scattering .
this leaves circumstellar dust scattering as the likely variable .
although the polarization variability of hd 45677 is well known , dating back to the paper in 1976 by coyne & vrba , it is only the high spectral resolution of the current data that allows us to discriminate between electron and dust scattering .
the major question remaining is why the apparent rotation of about 55@xmath12 between the major axis of the ( variable ) dust polarization and the h@xmath3 line forming region exists . in principle , this would suggest that multiple geometries , such as a combination of an equatorial disk and a bipolar flow , would show a rotation of 90@xmath12 , regardless of the respective opening angles .
the cancellation of perpendicularly - oriented polarization vectors tends to increase for larger opening angles , decreasing the observed polarization percentage , but the 90@xmath12 rotation will remain intact . an answer may lie in the clumpiness of the dusty material around the object .
hd 45677 is photometrically variable , a property attributed to the presence of various dust - clouds surrounding and even orbiting the object ( de winter & van den ancker 1997 ) .
clumpy material is also revealed by the spectroscopic variability .
grady et al . ( 1993 ) show the presence of variable redshifted absorption lines which are attributed to infalling and evaporating cometary bodies .
these indicate patchiness of the circumstellar material as well .
the result of such clumpy material on the polarization angle is relatively easy to understand .
if ( one of ) the scattering regions is clumpy , a rotation of 90@xmath12 is only retrieved if the clumps are symmetrically distributed around the central star .
if not , not all perpendicularly oriented polarization vectors will cancel out , and the net effect is that the observed position angle does not represent the time - averaged mean orientation of the scattering material . a similar argument has been brought forward by trammell et al .
( 1994 ) to explain the rotation of 70@xmath12 ( instead of 90@xmath12 ) in the spectrum of iras 08005 - 2356 .
it seems thus that the rotation between the dust ring and the h@xmath3 line forming region , which is less than 90@xmath12 , could be the result of scattering with incomplete cancellation in an inhomogeneous region .
a word of caution should be given here with regard to the ` true ' orientation of the dusty material .
the 20@xmath12 that was measured between the continuum points in fig .
[ qu ] , and its comparison with the intrinsic angle of the electron scattering region points to the clumpiness .
however , this direction along which the variation occurred should not now be associated with the orientation of the scattering region , since we only measure incomplete cancellation of dust clumps at changing positions
. only sequences of observations of this type may give a clue as to whether the variable dust component arises from a region perpendicular to or colinear with the electron scattering region .
the main result of the new observations of hd 45677 is that it is possible to discriminate between the _ electron _ scattering and the _ dust _ scattering regions .
the former can be probed by the change across h@xmath3 while the latter is probed by the variability of the polarization of the continuum .
this paper concerned medium resolution spectropolarimetry of a relatively large sample of b[e ] and herbig be stars , objects which so far have never been observed in this way .
we have described the results of these exploratory observations in a qualitative way and we summarize the highlights of both non - detections and detections below , and then briefly discuss the implications of the results .
the main goal of the observations was to answer the basic question `` does the ionized material around these stars project to circular symmetry on the sky , or not ? '' by investigating whether or not a ` line - effect ' is seen across h@xmath3 . in principle , a non - detection should imply a circular projection .
this encompasses three - dimensional geometries that are spherically - symmetric , or disk - like seen close to face - on .
two of the non - detections failed in quite different ways to fit into this simple picture : first , in the case of mwc 297 a young b1.5 star in the aquila rift ( drew et al .
1997 ) observation over two hours or so produced evidence of , at most , a subtle line effect .
this was despite the fact that the radio image of this object shows a clearly elongated ionized gas distribution .
an intriguing interpretation , testable by high resolution imaging , is that we view mwc 297 only indirectly at optical wavelengths .
if the direct sightline to the disk should reveal an edge - on structure , but if the disk is obscured from view due to the large extinction , scattering dust clouds may see a more circularly symmetric structure , such as a face - on disk , and reflect a polarization spectrum without a line effect to the observer .
second , we have presented data on @xmath4 ori , a star that has been reported twice before in the literature as showing an h@xmath3 line effect ( in both instances the observations were narrow - band rather than spectropolarimetric ) .
the absence of any such effect in our data suggests that the ionized envelope is smaller than at previous occasions because the , optically thin , electron - scattering is more sensitive to changes in ionization than the optically thick h@xmath3 emission .
the non - detection in @xmath4 ori implies that single epoch measurements are not always sufficient to provide a definitive answer on the circumstellar geometry of these objects . in both cases ,
it is clear that h@xmath3 spectropolarimetry is best judged in the context of other observational constraints on the target .
indeed they warn not to assume _ too _ quickly that the ionized regions in stars without a line effect are face - on disks or spherically - symmetric .
in contrast , the detection of a line - effect immediately tells us that the scattering region does not project to circular symmetry on the sky .
the data presented in this paper provide a new , richer variety of line effects than has hitherto been seen in the literature .
this is in part due to their relatively high spectral resolution .
the curious cases of the somewhat similar situations of hd 37806 and hd 50138 offer great opportunities to understand the conditions close to the star .
both objects show double peaked h@xmath3 line profiles , that have different @xmath35 ratios in normal intensity spectra , but which turn out to be of equal strength in the polarized flux data .
both stars also exhibit a superposed single component of h@xmath3 emission that is also picked out by a change in linear polarization at much the same wavelength .
this suggests that the line profile as a whole may be the result of two kinematically - distinct phenomena .
these may be a rotating disk ( or self - absorbed compact nebula ) , and a spatially more extended region of less - rapidly expanding hii whose emission is only polarized by the ism . a particularly striking result to emerge from our data concerns the probable herbig
be star , hd 53367 .
we have observed depolarization across h@xmath3 without angle rotation .
the co - alignment of the local interstellar magnetic field and stellar rotation axis together with the findings of vrba et al .
( 1987 ) , favour formation of this relatively massive star ( @xmath36 m@xmath37 ) by collapse rather than by the merger of less massive stars .
the power of repeated observations is shown by the case of hd 45677 . due to the continuum variability and the constant polarization arising from electron scattering , it is possible to distinguish between the electron and circumstellar dust scattering mechanisms .
this is the first time that it has been possible to do this . since the measured intrinsic angles of the dusty and the ionized region differ by 55@xmath12 rather than 90@xmath12 ( for perpendicular geometries ) or 0@xmath12 ( for parallel geometries ) we can conclude that the the dust component is clumpy . among the probable herbig
be stars we have observed , around half of them have shown no detectable line effect ( lkh@xmath3 218 , hd 52721 , v380 ori and mwc 297 ) .
this of course means that almost half have ( namely hd 259431 , hd 37806 and hd 53367 ) and should encourage further campaigns of this nature . using narrow bands , poeckert & marlborough ( 1976 ) surveyed 48
be stars and found that 21 showed the line effect at a 3@xmath16 level , while a further 8 show the line effect at a 2 - 3@xmath16 level .
they investigated the relation between the intrinsic polarization of these be stars ( measured from the h@xmath3polarization change ) with @xmath38 ( as measure of inclination ) and found that their observations could well be explained by inclination effects .
although based on very small number statistics , the comparable incidence of line effects in the herbig be stars observed does hint that flattened ionized circumstellar structures are quite common for this object class as well , and that the non - detections in our sample could be due to random sampling of the full range of inclinations .
one of our non - detections is of course mwc 297 , an object revealed by radio imaging to be non - circular ( albeit on a scale of tens of au ) .
it is important to appreciate that the h@xmath3 polarization effect is sensitive to much smaller structures than are presently probed directly by imaging .
analytical calculations such as those by by cassinelli , nordsieck & murison ( 1987 , their fig .
7 ) demonstrate that the bulk of electron scattering occurs on scales as small as two to three stellar radii .
if the herbig be stars we have observed are indeed young or even pre - main sequence objects , then the deviation from spherical symmetry of the ionized region is presumably a consequence of the way in which they have formed . viewed in these terms , the structures that we detect now via these polarization measurements could well be accretion disks that reach to within a stellar radius or so of the stellar surface .
this conclusion is at variance with the magnetospheric accretion model widely regarded as applicable to t tau stars , wherein magnetic channeling inhibits the formation of the inner disk ( see shu et al .
1994 ) . for herbig
be stars it remains a more open question as to how far magnetic fields determine the accretion geometry .
since the main sequence destiny of these more massive stars is to possess radiative envelopes , it leaves more room to doubt that magnetic fields must play a big role at the stage in which we are able to observe them
. it will be interesting to see if further h@xmath3 spectropolarimetry continues to uncover plausible disk accretors .
a conceptual model of how these objects may look , still embedded in accretion disks reaching into the stellar surface , has recently been devised by drew , proga & stone ( 1998 , building on the work of proga , stone & drew 1998 ) .
this work shows how observationally - significant disk winds , driven by radiation pressure , would be created .
a further piece in the puzzle of herbig be and bn objects that these predicted flows can help explain is the high contrast , quite narrow hi line emission often observed at earlier b spectral types ( drew , 1998 ) .
a further very strong outcome of this study is that all objects that can be classified as ( evolved ) b[e ] stars , presented significant polarization changes across h@xmath3 .
a factor that clearly helps increase the likelihood of detecting a spectropolarimetric line effect in b[e ] stars is that their h@xmath3 profiles are typically extremely high contrast and often somewhat broader than in herbig be stars .
pre - eminent among our b[e ] group is hd 87643 which has already been discussed in a separate paper ( oudmaijer et al .
1998 ) . here
we have presented mwc 137 ( probably an evolved b[e ] star ) , hd 45677 and hd 50138 .
the fact that _ all _ observed galactic b[e ] stars in our sample show the line - effect in one incarnation or another lends strong support to the zickgraf et al .
( 1985 , 1986 ) model .
the variety of line - effects observed in our data illustrate that the structures around these stars have their deviation from spherical symmetry in common , but that the details in each case are different .
so far , the discussion of these data has been largely qualitative . as models simulating
these phenomena begin to be calculated there will no doubt be a considerable sharpening of insight .
apart from providing some striking insights into a number of the targets observed , the programme of observations we have described here has offered some lessons in how best to obtain single - line spectropolarimetric data .
it is clear that the spectral resolution available to us ( @xmath39 ) has in most instances been just enough . as
numerical modelling becomes more commonplace , the case for increased spectral resolution will become stronger .
the main issue , nevertheless , is the achievement of high enough data quality .
our 8@xmath40 magnitude and brighter objects have come out well in under an hour s telescope time , while 10@xmath40 to 12@xmath40 magnitude objects require several hours observation with a 4-metre class telescope in at least middling weather conditions .
ultimately , these fainter sources will be best served by 8-metre class facilities where the shorter total integration times will be less subject to weather influence
presently they can be an uncertain struggle .
the overall conclusion of this study is that this relatively unexplored mode of observing does yield valuable new insights . in some instances
we have encountered deepening mysteries that suggest conclusions drawn from other data have missed something .
mwc 297 and hd 45677 are both good examples of this . at the same time
, h@xmath3spectropolarimetry readily throws up examples that demand sophisticated numerical modelling of a type that is just beginning to become available ( hillier , 1996 ; harries , 1996 ) .
we thank the staff at the anglo - australian telescope for their expert advice and support .
conor nixon and graeme busfield are thanked for their help during some of the observing runs .
the allocation of time on the anglo - australian telescope was awarded by patt , the united kingdom allocation panel .
rdo is funded by the particle physics and astronomy research council of the united kingdom .
the data analysis facilities are provided by the starlink project , which is run by cclrc on behalf of pparc .
part of the observations are based on data obtained from the william herschel telescope , tenerife , spain , in the isaac newton group service scheme .
this research has made use of the simbad database , operated at cds , strasbourg , france .
allen d.a . , swings j.p .
1976 , a&a 47 , 293 bergner yu.k .
et al 1988 , afz 28 , 529 blommaert j.a.d.l . , van der veen w.e.c.j . , habing h.j . 1993 ,
a&a 267 , 39 bonnell i.a . ,
bate m.r . , zinnecker h. 1998 , mnras 298 , 93 burrows c.j .
1996 , apj 473 , 437 cassinelli j.p .
, nordsieck k.h .
, murison m.a .
1987 , apj 317 , 290 clampin m. , schulte - ladbeck r.e .
, nota a. , robberto m. , paresce , clayton g.c .
1995 , aj 110 , 251 claria j.j .
1974 , a&a 37 , 229 clarke d. , mclean i.s .
1974 , mnras 167 , 27 clarke d. , brooks a. 1984 , mnras 211 , 737 corcoran m. , ray t.p .
1997 a&a 321 , 189 coyne g.v . &
1976 , apj 207 , 790 de winter d. , van den ancker m.e .
1997 , a&as 121 , 275 drew j.e .
1998 , in asp conf .
series 131 _ boulder - munich ii : properties of hot , luminous stars _ , ed .
howarth , astron .
, san francisco , p. 14
, busfield g. , hoare m.g .
, murdoch k.a . ,
nixon c.a . ,
oudmaijer r.d .
1997 , mnras 286 , 538 drew j.e .
, proga d. , stone j.m .
1998 , mnras 296 , l6 esa , `` the hipparcos and tycho catalogues '' , 1997 , esa sp-1200 esteban c. , fernndez m. 1998 , mnras 298 , 185 finkenzeller u. 1985 , a&a 151 , 34 frank a. and mellema g. 1996 , apj 472 , 684 gnedin iu . n. , kiselev n.n .
, pogodin m.a . ,
rozenbush a.e . ,
rozenbush v.k .
1992 , azhl 18 , 454 grady c.a .
, prez m.r .
, talavera a. et al .
1996 , a&as 120 , 157 grady c.a . , bjorkman k.s . , shepherd d. , schulte - ladbeck r.e . , perez m.r
. 1993 , apj 415 , l39 grinin v.p .
1994 , a&a 292 , 165 gummersbach c.a .
, zickgraf f .- j . , wolf b. 1995 , a&a 302 , 409 harries t.j .
1996 , a&a 315 , 499 herbig g.h .
1960 , apjs 4 , 337 herbig g.h . , 1994 , in asp conf.sers .
62 _ the nature and evolutionary status of herbig ae / be stars _ , eds .
th , m.r .
prez , e.p.j .
van den heuvel , astron .
, san francisco p. 3
herbst w. , racine r. , warner j.w .
1978 , apj 223 , 471 herbst w. , warner j.w . ,
miller d.p .
, herzog a. 1982 , aj 87 , 98 hillenbrand l.a . ,
strom s.e . ,
, keene j. 1992 , apj 397 , 613 hillier d.j .
1996 , a&a 308 , 521 hoare m.g .
, garrington s.t .
1995 , apj 449 , 874 jain s.k .
bhatt h.c .
1995 , a&as 111 , 399 jaschek c. , andrillat y. 1998 , a&as 128 , 475 magalh@xmath7es a.a .
1992 , apj 398 , 286 malfait k. , bogaert e. , waelkens c. 1998 , a&a 331 , 211 mannings v. , koerner d.w . ,
sargent a.i .
1997 , nat .
388 , 555 matthewson d.s .
, ford v.i . , klare g. , neckel t. , krautter j. 1978 , _ bull .
, 14 , 115 mclean i.s . , clarke d. 1979 , mnras 186 , 245 miller w.c . , merrill p.w .
1951 , apj 113 , 624 miroshnichenko a.s . ,
ivezi , elitzur m. 1997 , apj 475 , l41 osterbart r. , langer n. , weigelt g. 1997 , a&a 325 , 609 oudmaijer r.d .
, drew j.e .
1997 , a&a 318 , 198 oudmaijer r.d .
, proga d. , drew j.e . , de winter d. 1998 , mnras 300 , 170 pezzuto s. , strafella f. , lorenzetti d. 1997 , apj 485 , 290 poeckert r. 1975 , apj 152 , 181 poeckert r. , marlborough j.m .
1976 , apj 206 , 182 ( pm ) pogodin m.a .
1997 , a&a 317 , 185 proga d. , stone j.m . , drew j.e .
1998 , mnras 295 , 595 schmidt - kaler th .
1982 , in landolt - brnstein , numerical data and functional relationships in science and technology , new series group vi , vol . 2b , springer - verlag schulte - ladbeck r.e . , shepherd d.s .
, nordsieck k.h .
1992 , apj 401 , l105 schulte - ladbeck r.e . ,
clayton g.c . ,
hillier d.j . , harries t.j . ,
howarth i.a .
1994 , apj 429 , 846 shu f. , najita j. , ostriker e. , wilkin f. , ruden s. , lizano s. 1994 , apj 429 , 781 slettebak a. 1988 , pasp 100 , 770 sonneborn g. et al .
1988 , apj 325 , 784 sorrell w.h .
1989 , mnras 241 , 89 stenholm b. acker a. 1987 , a&as 68 , 51 th p.s .
, de winter d. , perez m.r .
1994 , a&as 104 , 315 tinbergen j. , rutten r. 1997 , ` measuring polarization with isis ' tody d. 1993 , in asp conf .
series 52 _ astronomical data analysis software and systems ii _ , eds .
hanisch , r.j.v .
brissenden , j. barnes , astron . soc .
, san francisco , p. 173 .
trammell s.r .
, dinerstein h.l . ,
goodrich r.w .
1994 , aj 108 , 984 van den ancker m.e .
, de winter d. , tjin a djie h.r.e .
1998 , a&a 330 , 145 vrba f.j . , baierlein r. , herbst w. 1987 , apj 312 , 207 waelkens c. , van winckel h. , waters l.b.f.m . , bakker e.j .
1996 , a&a 314 , l17 welsh b.y .
1991 , apj 373 , 556 wood k. , brown j.c . , fox g.k .
1993 , a&a 271 , 492 zickgraf f .- j .
, schulte - ladbeck r.e .
1989 , a&a 214 , 274 zickgraf f .- j .
, wolf b. , stahl o. , leitherer c. , klare g. 1985 , a&a 143 , 421 zickgraf f .- j . , wolf b. , stahl o. , leitherer c. , appenzeller i. 1986 , a&a 163 , 119 | we present the results of medium resolution ( @xmath0 @xmath1 60 km s@xmath2 ) spectropolarimetric observations across h@xmath3 of a sample of b[e ] and herbig be objects .
a change in linear polarization across h@xmath3 is detected in a large fraction of the objects , with characteristics ranging from simple depolarization in a couple of herbig be stars , to more complex behaviour in the probable post main sequence b[e ] stars .
h@xmath3 in the spectra of hd 37806 and hd 50138 each consist of a double - peaked polarized line and a superposed unpolarized single emission peak , suggesting two distinct line - forming regions . multiple observations of hd 45677 allow for the separation of electron and dust scattering effects for the first time : the difference between derived intrinsic polarization angles of the two components indicate that the dust - scattering region is clumpy .
two unexpected results are the non - detections of h@xmath3 polarization changes in @xmath4 ori , where depolarization has previously been detected , and in mwc 297 , which exhibits source elongation at radio wavelengths . in @xmath4 ori time variability is probably responsible such that this star s electron - scattering disk was much weakened at the time of observation .
two hypotheses are advanced that might explain the mwc 297 result .
the general findings are that roughly half of the observed herbig be stars show polarization changes across h@xmath3 , implying immediately that their ionized envelopes are not spherically symmetric . this pattern , if confirmed by observations of a larger sample , could indicate that the non - detection rate is simply a consequence of sampling randomly - oriented circumstellar disks able to scatter starlight within a few stellar radii .
the stars classified as b[e ] stars all show startling polarization changes across h@xmath3 .
the details in each case are different , but the widely accepted concept of dense h@xmath3-emitting equatorial disks around these objects is supported . epsf # 1#2 # 1#2#1@xmath5roman#2 = = = 1=1=0pt = 2=2=0pt biblabel[1 ] stars : circumstellar matter stars : pre - main sequence stars : emission line , be stars : mass loss polarization |
cataclysmic variables ( cvs ) are close binary star systems in which a white dwarf primary star accretes matter from the roche lobe overflow of a less - compact secondary star .
@xcite gives an overview of the field .
the orbital period , @xmath0 , of a cv is its most easily measurable fundamental property .
mass transfer in cvs is thought to be driven by the slow loss of orbital angular momentum . during most of a cv
s life , this leads to gradual shrinkage of the orbit and consequent tightening of secondary s roche lobe .
the orbital period therefore gives a rough idea of a system s evolutionary state .
once an orbital period is known , the roche - filling criterion sets strong constraints on the secondary s radius , which in favorable circumstances can be exploited to yield a distance estimate .
cvs with high orbital inclinations @xmath2 will eclipse , which makes orbital period determination straightforward and also makes it possible to estimate component masses and the like .
non - eclipsing systems sometimes show periodic photometric modulations , but these do not always appear at @xmath0 ; other phenomena such as superhumps @xcite can mimic orbital modulations .
radial velocity spectroscopy is therefore the most reliable way of obtaining @xmath0 for a non - eclipsing cv . ;
@xcite discuss a particularly interesting example .
] for this reason we have therefore been collecting radial - velocity time series for a large number of cvs , with the aim of determining @xmath0 . in this paper
we collect together observations and analyses of twenty cvs , all of which proved to have periods well longward of the so - called period gap , a range of @xmath0 from roughly 2 to 3 hours in which relatively few cvs are found @xcite .
table [ tab : star_info ] gives basic information about the stars studied here .
the periods of most of these stars are published here for the first time ; a few others are included because we have improved the accuracy of the period .
we did not use any particular astrophysical criterion to select the present sample , but rather observed these stars because period determinations appeared feasible .
section 2 describes the observations and analysis , and 3 gives results for the individual objects .
our spectroscopic observations are from the 2.4 m hiltner or 1.3 m mcgraw - hill telescopes at mdm observatory in kitt peak , arizona . at the hiltner
, we used the ` modular ' spectrograph , a 600 line mm@xmath3 grating , a 1-arcsec slit , and a site 2048@xmath4 ccd detector .
most of the spectra cover from 4210 to 7560 ( vignetting severely toward the ends ) , and have typical resolution of 3.5 fwhm . on the 1.3 m
, we used the mark iii spectrograph , a site 1024@xmath4 ccd detector , a 2-arcsecond slit , and achieved typical resolution of 4 - 5 fwhm from 4645 to 6960 . on earlier observing runs , we took spectra of comparison lamps whenever the telescope was moved to achieve accurate wavelength calibration , but on later runs we used the night - sky spectrum to shift the zero point of a master wavelength calibration obtained during twilight . when the sky was clear , we observed bright early - type stars to derive corrections for the telluric absorption features , and flux standards to calibrate the instrument response .
our spectrophotometry suffers from uncalibratable errors due to losses at the spectrograph slit , but experience suggests that these typically amount to @xmath5 mag .
* hereafter tft04 ) discuss the observational protocols in more detail .
we measured centroids for the emission lines using convolution techniques explained by @xcite and @xcite , in which the line profile is convolved with an antisymmetric function and the zero of the convolution is taken as the line center .
for the most part we used a convolution function consisting of a positive gaussian and a negative gaussian separated by an adjustable amount .
we usually set the separation of the gaussians to maximize sensitivity to the steep sides of the line profile .
the roche geometry dictates that longer - period cvs will have larger and generally more massive secondary stars than shorter - period cvs , so their secondary stars often contribute significantly to the total light . when possible , we measured absorption line velocities using the cross - correlation radial velocity package _ xcsao _ @xcite . in most cases
the template spectrum was a composite of 76 observations of late - type iau velocity standard stars , which had been shifted individually to zero velocity before averaging ; for later - type secondaries we used a similar template constructed from m dwarfs .
the spectral region used in the cross - correlation was either 6000 to 6500 , or a window from 5100 to 6500 , excluding the complicated region near hei @xmath6 and nad . uncertainties , based on the @xmath7-statistic of @xcite , were typically on the order of 20 km s@xmath3 .
we omitted unphysically large velocities ( which result when the ` wrong ' cross - correlation peak is selected ) and velocities with large estimated uncertainties .
we analyzed the velocity time series with a ` residualgram ' period - search algorithm @xcite which fits general least - squares sinusoids of the form @xmath8 to the time series over a densely - spaced grid of trial frequencies @xmath9 .
the results are presented as plots of 1/@xmath10 , where @xmath11 ^ 2 $ ] ; here @xmath12 is a measured velocity , @xmath13 is the best - fitting sinusoid evaluated at the corresponding time @xmath14 , and @xmath15 is the formal uncertainty of the velocity , obtained by propagating the estimated counting - statistics uncertainties of the individual wavelength channels .
ideally , a graph of @xmath16 as a function of @xmath9 shows a strong spike at the orbital frequency , but the imperfect sampling of the time series inevitably results in aliasing .
we used a monte carlo algorithm developed by @xcite to assess the confidence with which the highest peak in the periodogram can be identified with the true orbital frequency .
once we were satisfied that the period was correct , we fit the time - series with sinusoids of the form @xmath17\ ] ] using a hybrid linear least - square algorithm and @xmath18 , respectively ] .
this procedure is described in detail in tft04 . for those cvs with absorption - line radial velocity solutions , we shifted the individual spectra into the absorption - line rest frame before averaging them . as noted earlier , the secondary - star spectrum is often detectable in longer - period cvs .
the absorption - line radial velocities should track the orbital motion of the secondary , though the radial velocities may be affected by irradiation and other effects @xcite .
however , a late - type contribution need not arise from the secondary ; if the late - type spectrum is nearly fixed in velocity , it may come from an unrelated field star , or the outer member of a hierarchical triple ; this appears be the case for v542 cyg , discussed further below .
@xcite and tft04 discuss our spectral decomposition technique and the inferences based on the secondary star , which we summarize briefly here . to estimate the secondary star s spectral type and its contribution to the light , we scale spectra of cool spectral - type standard stars , subtract them from the target spectrum , and look for the best possible cancellation of the late - type features .
this yields estimates of the spectral type of the secondary and its contribution to the total light .
the spectral type constrains the secondary s surface brightness , and the orbital period ( together with the condition that the secondary fills its roche critical lobe ) tightly constrains the secondary s radius @xmath19 .
the precise value of @xmath19 does depend weakly on the secondary s mass , @xmath20 ; to compute a realistic range for @xmath19 , we use the evolutionary calculations of @xcite to guide the selection of a range for @xmath20 .
the estimates of the radius and surface brightness together yield the secondary s absolute magnitude , which in turn gives the distance , without any assumption that the secondary is a typical main - sequence star .
where applicable , we estimate the visual extinction using the @xcite maps ; these give the extinction to the edge of the galaxy , so we use their tabulated @xmath21 as an upper limit . for most of the stars studied here ,
we have collected direct images using the hiltner 2.4 m telescope and a site 2048@xmath4 ccd detector .
most of the celestial positions in table [ tab : star_info ] were measured from these images .
in a few cases we give magnitudes based on @xmath22 or @xmath23 image sets , which were reduced using observations of standard stars from @xcite .
our results for the individual cvs are summarized in tables and figures .
figs.[fig : spectra1 ] and [ fig : spectra2 ] show the mean spectra of most of the stars ( excluding a few spectra of dwarf novae in outburst ) .
a small sample of radial velocity measurements included in the analysis are listed in table [ tab : velocities ] . figs .
[ fig : pgrams1 ] and [ fig : pgrams2 ] give periodograms of the velocity time series , computed as described above .
table [ tab : parameters ] lists parameters of the best - fit sinusoids to the velocity time series .
for all parameters , the number in parentheses represents the uncertainty in the last digit .
finally , figs .
[ fig : folpl1 ] , [ fig : folpl2 ] , and [ fig : folpl3 ] show folded radial velocity curves ; when both absorption- and emission - line velocities are available , the radial velocity curves are folded about the weighted mean period .
v709 cas is a dq her star , or intermediate polar ( ip ) , with a pulsation period near 313 s. @xcite found @xmath24 d from a radial - velocity study .
they also detected absorption wings around the higher balmer lines , which they interpreted as arising from the white dwarf photosphere .
our h@xmath25 emission - line radial velocities were taken in the interval from 1996 to 2005 , and define a unique cycle count over the entire time span .
the period we find , 0.2222033(7 ) d , or 5.33288 hr , agrees within the uncertainties with the @xcite period , but is more precise because of the longer time base .
[ fig : v709casspec ] shows the average spectrum from 1999 october ; the flux level in this spectrum corresponds to @xmath26 .
while the spectrum does not cover the higher balmer lines well , we do not see absorption wings around h@xmath27 comparable to those found by @xcite .
we examined the summed spectra from other observing runs , and none of these showed balmer absorption either .
we therefore can not confirm the white dwarf detection claimed by @xcite .
after some consideration , we are skeptical that the balmer absorption wings detected by @xcite arise from the white dwarf photosphere .
they get a good fit to the absorption using a @xmath28 k white dwarf model atmosphere . at @xmath29 ( in cgs units ) ,
a white dwarf at this @xmath30 should have @xmath31 @xcite . if the system s luminosity is low enough for such a faint white dwarf to be seen , then where is the secondary ?
@xcite compiled a semi - empirical ` donor sequence ' for cvs , and at the 5.3 hr orbital period of v709 cas , this predicts the spectral class of the secondary to be near m2 , and the absolute magnitude @xmath32 to be near 9 . for their interpretation to be correct , the flux contributed by the white dwarf puts the system at @xmath33 pc , or @xmath34 ; the @xmath35 secondary should therefore have @xmath36 , and contribute very substantially to the spectrum , but we do not detect the secondary s contribution in our long cumulative exposures .
we do see nad features , with each of the d lines having an equivalent width of @xmath37 , but these are sharp and motionless , indicating an interstellar origin .
we feel that the balance of the evidence suggests that the line wings are an intermittent feature of the optically thick disk , and that the system is more luminous and distant than in the @xcite interpretation .
@xcite reported spectroscopy of af cam .
they detected the k - type secondary star , and radial velocity variability in both emission and absorption lines at a weighted mean period of @xmath38 d. we have since observed the star occasionally with the aim of refining @xmath0 and deriving an ephemeris .
the period and epoch given in table [ tab : parameters ] is based on an unambiguous cycle count over the interval from 2000 to 2007 .
@xcite show a spectrum and analyze the secondary star s contribution .
@xcite discovered this cv as part of the isaac newton telescope photometric h@xmath25 survey of the northern galactic plane , the strong k - type contribution evident in their spectrum , and the relatively bright magnitude ( @xmath39 ) made this a tempting target for an absorption - line velocity study , which we undertook starting in 2008 january .
the expected modulation did prove to be present in both the h@xmath25 emission velocities and the absorption velocities , with @xmath0 near 7.5 hr .
both time series favor a period near 0.31390 d. an alias period near 0.31609 d , reflecting an alternate choice of cycle count between observing runs , is much less likely .
the secondary spectrum is best matched by a k5.5 @xmath40 1.5 star , but the continuum that remains after the secondary star is removed is unusually red , suggesting that the reddening is substantial .
the galactic latitude is low , @xmath41 , and the @xcite maps give the reddening @xmath42 mag at this location .
we experimented with de - reddening by various amounts . with reddenings @xmath43 ,
the balmer decrement became implausibly steep ( h@xmath27 stronger than h@xmath25 ) , while at small reddenings the continuum remained too red .
the best values appeared to be in the range @xmath44 .
the procedure outlined earlier then led to the inferences in table [ tab : inferences ] .
the distance estimate is very rough , mostly because of the large uncertainty in the reddening .
sinusoidal fits to the emission and absorption velocities show them to be @xmath45 cycles out of phase . in double - lined spectroscopic binaries ,
the ratio @xmath46 gives an estimate of the mass ratio , but in order for this to be accurate , the velocities must trace the center - of - mass motion of the two stars . if this is the case , the phase lag between emission and absorption must be 0.5 cycles this is a necessary , but not sufficient , condition for the estimate to be trustworthy . in the present case ,
this condition is marginally fulfilled the discrepancy is formally three standard deviations .
if we assume the @xmath47-velocities do reflect the center - of - mass motion , we find @xmath48 .
combining this with a broadly typical white dwarf mass of 0.7 m@xmath49 yields a secondary mass near 0.55 m@xmath49 , similar to what the @xcite models would suggest at this period and spectral type . while these mass calculations are somewhat fanciful , the modest value of @xmath18 does firmly constrain the orbital inclination @xmath2 to be fairly low .
even we take @xmath50 to be a very low @xmath51 m@xmath52 and @xmath53 to be 0.6 pushing both quantities to values that increase @xmath2 the inclination is only @xmath54 25 degrees .
the apparent lack of eclipses is therefore not unexpected .
the @xcite evolutionary models that best match this system are those in which mass transfer begins after hydrogen has been substantially depleted in the core .
the spectral type of the secondary is similar to that in other cvs with similar orbital periods @xcite .
no dwarf nova outbursts have been reported from this system . in many respects
it resembles ly uma (= cw 1045 + 525 ) , a 0.271-d emission - line binary with a strong late - k contribution and no reported outbursts @xcite .
v1062 tau is the optical counterpart of the x - ray source 1h0459 + 246 , which was discovered in the heao-2 large area sky survey @xcite .
@xcite found an x - ray modulation near 62 minutes , and classified the object as an intermediate polar , or dq her star .
the 62-minute x - ray modulation was confirmed and refined by @xcite .
@xcite also published an optical spectrum , and noted the features of a cv and a k star .
they obtained @xmath55 and @xmath56 time series photometry that showed ellipsoidal variation at a period of @xmath57 h. @xcite monitored v1062 tau photometrically , and refined the orbital period to @xmath58 h. they also detected the 62-minute period , and the beat between this and the orbit .
their data showed two short , low - amplitude outbursts , with @xmath59 mag .
for the most part , our spectra of v1062 tau appeared similar to the average shown in fig .
[ fig : spectra1 ] , which in turn is generally similar to that shown by @xcite ( their figure 3 ) .
the excitation is high ; heii @xmath60 emission is similar in strength to h@xmath27 , and our long cumulative exposure also shows emission at heii @xmath61 . the @xcite map indicates a reddening @xmath62 at this position ( @xmath63 ) , and the continuum does suggest substantial reddening .
the presence of diffuse interstellar bands near @xmath64 and @xmath65 @xcite gives further evidence for reddening , and the rather strong na d absorption lines near @xmath66 are almost motionless in the spectrum , indicating that they are mostly interstellar . in computing the secondary - based distance , we assume that v1062 tau is beyond most of the reddening in the @xcite map , which yields a distance of @xmath67 pc , which would put it over 200 pc from the midplane of the galaxy , where it would indeed be expected to be outside most of the dust layer .
our spectrum has a synthetic @xmath55 magnitude near 16.7 .
however , we also obtained two spectra 2002 feb 20.53 ut that show the star much brighter , with synthetic magnitudes near @xmath68 ; another pair of spectra taken at feb 20.78 ut , a few hours later , have @xmath69 and 15.3 . by a remarkable coincidence
, @xcite observed the same outburst earlier in the same night ( from a different longitude ) ; our first pair of spectra was taken some 4.5 hours after their last point . if the decline between our spectra was the decline from outburst , the flare s duration was considerably less than one day .
only a few magnetic cvs have been observed to undergo such short - lived flares @xcite . in our spectra
the velocity of the late - type feature is modulated at @xmath70 d , or 9.9812(3 ) h , consistent with the @xcite period .
the emission - line velocities include some points from earlier observing seasons ; the signal - to - noise of these spectra was not adequate for measuring absorption - line velocities , but the increased time base does result in a more precise period .
the emission - line velocities do not determine the period on their own
the single best fit to the emission is near 0.4187 d ( 10.05 h ) .
however , when the absorption line velocities are fitted at the 10.05 h period , the fit is very poor , while the fit to the emission - line velocities at 9.9812 h ( the best absorption - line period ) is reasonably good . because the absorption lines arise in a relatively uncomplicated stellar photosphere
, their velocities should be much less vulnerable to state changes ( and the like ) than those of the emission lines ; because of this , and the good agreement with the independent period determination by @xcite , we are confident that the absorption - line period is essentially correct . in table [ tab : parameters ] and fig . [
fig : folpl1 ] we adopt the weighted average of the emission- and absorption - line periods .
the two periods do disagree by 3.6 standard deviations , so some caution is in order extending the ephemeris into the future , but the actual difference of the periods is only 4.3 s. the spectral decomposition shows the secondary to be of type k5 @xmath71 subclasses . at this orbital period and spectral type , modeling suggests that cv secondaries depleted much of their core hydrogen before the beginning of mass transfer @xcite .
the emission - line velocity modulation is far from antiphased with the absorption ( see fig . [
fig : folpl1 ] ) , so it is evidently not a faithful tracer of the white dwarf motion . in contrast , the absorption lines should follow the secondary star . without a reliable @xmath72 we can not make firm mass estimates .
however , the absorption velocity amplitude @xmath18 , @xmath73 km s@xmath3 , is fairly large for this @xmath0 , and suggests that the white dwarf is relatively massive and that the inclination is not too far from edge - on . as an illustration , a system @xmath50 = 1 m@xmath52 and @xmath20 = 0.5 m@xmath52 at @xmath74 degree would produce the observed @xmath18 .
note that @xcite do not detect eclipses despite intensive coverage .
our spectrum appears similar to that published by @xcite .
we clearly detect an m - dwarf secondary . the gross period ( table [ tab : parameters ] )
is based on emission - line velocities .
it is unambiguous , but the fine period is aliased on a scale of one cycle per 47 d due to uncertain cycle count between observing runs ; the allowed precise periods satisfy @xmath75 where the denominator is an integer .
the cross - correlation technique was sensitive enough to measure m - dwarf velocities for some of our spectra , but these were not extensive or accurate enough to shed light on the period ; however , they did confirm that the m - dwarf light is from the binary companion and not from an interloper .
combining our spectra into a single - trailed greyscale representation confirmed the presence of secondary features ; in particular , the cai @xmath76 6122 absorption line appeared faintly , with a velocity semiamplitude of @xmath77 km s@xmath3 .
v344 ori is at @xmath78 , for which @xcite give @xmath79 to the edge of the galaxy .
however , its spectrum shows neither diffuse interstellar bands nor interstellar nad absorption , which would be expected if the reddening were this large .
standardized @xmath22 observations taken 2008 jan .
16.0 ut give @xmath80 , @xmath81 , @xmath82 , and @xmath83 . at minimum light ,
dwarf novae typically have @xmath84 and @xmath85 ( see , e.g. , @xcite ) ; the very blue @xmath86 suggests near - zero reddening , while the redder @xmath87 suggests @xmath88 .
the poorly - constrained reddening contributes substantially to the distance uncertainty ( table [ tab : inferences ] ) .
@xcite obtained a spectrum of this rosat all - sky survey source that showed the broad emission lines of a cv . @xcite
obtained more extensive spectrophotometry from which they deduced a spectral type of m2 for the secondary .
they also found a radial - velocity periodicity that corroborated an early version of the period reported here .
nearly all our observations are from 2001 march , but we also have two velocities from 2000 april . the period measured in 2001 march
is not precise enough to extrapolate the cycle count to the previous year , but the extra data do constrain the precise period to a group of aliases which can be expressed as @xmath89 where the denominator is an integer .
the m - dwarf contribution to the spectrum was strong enough that we could measure absorption velocities for the majority of our spectra by cross - correlating the m - dwarf radial velocity template over the 6000 to 6500 region .
the absorption velocities were not precise enough to yield an accurate @xmath18 , but they did move approximately in antiphase to the emission , confirming that the m dwarf contribution arises from the binary companion .
we also estimated the secondary star s spectral type , and found the best match around m1 @xmath40 1 subtype , slightly earlier than the type found by @xcite .
this is just a little warmer than is typical at this @xmath0 ( 3.57 hr ) ; the @xcite models that predict this early spectral type begin with relatively massive secondaries and have depleted nearly all the core hydrogen at the start of mass transfer .
@xcite identified this object as a cv , most likely a dwarf nova of the ugss subclass , after patrick wils first called attention to the nature of its light curve from the northern sky variability survey .
it appears in the sloan digital sky survey , but was not selected for a spectrum , probably because its colors ( @xmath90 , @xmath91 , @xmath92 , @xmath93 , and @xmath94 ) place it near the ` stellar locus ' used to exclude spectroscopic targets @xcite .
this illustrates that while the sdss has been extraordinarily effective at finding new cvs and illuminating their properties @xcite , it does not find all the cvs within its footprint ( and indeed , @xcite warn that sdss does not claim completeness for cvs ) .
our spectra show modulation of both the emission and absorption at essentially the same period , with a secure cycle count on all timescales over the 1456 d span of our data set .
the weighted average @xmath0 is 0.376308(1 ) d ( 9.03 hr ) .
if we define orbital phase @xmath95 at the blue - to - red crossing of the absorption velocities through their mean velocity @xmath96 , then the emission velocities cross blue - to - red at @xmath97 , which is close to the antiphased behavior expected if the emission lines trace the white - dwarf velocity .
if the emission lines do in fact trace the white - dwarf velocity ( which is far from proven ) , @xmath98 , yielding @xmath99 . from @xcite , we find @xmath100 for a k4 or k5 secondary star with a @xmath0 of roughly 9 h ; this gives @xmath101 from @xmath53 .
these values of @xmath0 , @xmath50 and @xmath53 would imply an inclination near 45 degrees , so eclipses are not expected . at @xmath102 , @xmath103 , and @xmath104
pc ( table [ tab : inferences ] ) , this object is at least several hundred pc away from the galactic plane , so it may belong to the thick disk . the secondary s spectrum does not appear unusual , so it is probably not a low - metallicity halo object .
it does not have a significant proper motion listed in the usnob1.0 @xcite or ucac3 @xcite catalogs .
the general catalog of variable stars @xcite lists this star as having variability type is , that is , showing rapid irregular variability . @xcite
assign a superhump period of 0.12 d ( 2.88 hr ) , which would imply that @xmath0 is slightly shorter than that , which would put v478 her in the ( roughly ) 2- to 3-hour ` period gap ' .
the source for the 2.88-hr period is a photometric time series , 3.7 hours in duration , taken by t. vanmunster on 2001 june 7 ut , during an outburst .
this observation showed a variation that vanmunster interpreted as a probable superhump ; on this basis , the star is listed in the final edition of the @xcite cv catalog as an su - uma type dwarf nova .
we were therefore puzzled when our initial observations showed no sign of a short - period radial velocity modulation , despite sufficient signal - to - noise in the h@xmath25 emission lines to make such a modulation apparent if it were present .
more extensive observations extending over several seasons indicate that @xmath0 is much longer than had been thought , over 15 h. the secondary spectrum is just detectable , and the radial velocities of both the the h@xmath25 emission line and the absorption spectrum show the long period .
the short period appears to be mistaken , and the star has therefore been misclassified . because of its short duration , the light curve on vanmunster s site shows only a single minimum , and we note that he characterized his detection of superhumps as `` likely '' , rather than definitive .
we do not have time series photometry , and so can not comment further on the cause of the variation he detected .
the emission - line velocities we measure do not fit a sine curve accurately , and are not in antiphase to the absorption velocities .
the slightly double - peaked structure visible in fig .
[ fig : folpl1 ] may be real , or might result from limited sampling .
the secondary - star radial velocity amplitude , @xmath105 km s@xmath3 , does not suggest that the masses are out of the ordinary ; for example , a 0.8 m@xmath52 white dwarf and a 0.7 m@xmath52 secondary would produce the observed @xmath18 at @xmath106 .
the orbital period of v478 her is unusually long for a cv it lies well outside the range tabulated by @xcite , so the secondary - star mass range given in table [ tab : inferences ] is even more of a guess than usual .
even so , the long orbital period ( hence large roche lobe ) , warm secondary , and faint apparent magnitude indicate that v478 her very distant for a cv ; at @xmath107 and @xmath108 , it is also at least 1 kpc from the galactic plane .
this is an old nova , having outburst in 1898 .
@xcite describe the spectrum as having a noisy flat continuum , with balmer and heii emission .
the continuum in our spectrum is indeed fairly flat , but also shows weak broad undulations that superficially suggest a companion s contribution .
however , we could not generate a convincing spectral decomposition with our library spectra , so these might not indicate a companion . in part because of the unfavorable declination , we had to observe this target on several observing runs before we were certain about the gross period .
this led to an uncertain cycle count between runs , which in turn created a complex structure of fine - scale aliases .
the 6.87 h period given in table [ tab : parameters ] is based on the velocities from our most extensive single observing run .
@xcite found that the v1082 sgr shows high and low states ; when it is in a high photometric state ( @xmath109 14.0 - 14.5 ) , it shows emission lines , with heii @xmath60 comparable in strength to h@xmath27 , while in its low state ( @xmath109 15.0 - 15.5 ) , the emission lines disappear .
a k - type absorption spectrum is present as well , so they suggested that v1082 sgr may be a symbiotic star .
the orbital period we find , near 0.868 d , would be very short for a symbiotic , and is consistent with a long - period novalike variable .
the swift satellite detected v1082 sgr in hard x - rays ( 14 195 kev ; @xcite ) ; it is one of 29 cvs found in the swift bat 22-month survey catalog , most of which have proven to be magnetic cvs of the dq her ( or intermediate polar ) subtype @xcite .
our spectra confirm the state behavior noted by @xcite , but we never did see the emission lines disappear entirely .
strong he ii @xmath60 is often seen in magnetic cvs ; this , combined with the hard x - ray detection , suggests that this is indeed a magnetic system .
time - series measurements in the optical and x - ray might reveal a white - dwarf spin period .
it proved difficult to get a good match to the secondary spectrum , so the spectral type in table [ tab : inferences ] is fairly uncertain .
the unusually long period puts it outside the realm of the @xcite models , so a wide range of @xmath20 is assumed in the distance calculation ; fortunately , this has little effect on @xmath19 .
the @xcite map gives @xmath110 at this location ( @xmath111 ) ; this is small enough so that it does not contribute significantly to the distance uncertainty , which is dominated by the crude spectral decomposition .
this dwarf nova is among the few studied by @xcite that does not have a known orbital period .
the velocities from 2007 june and july define a period of 0.21777(4 ) d , or 5.23 h. we also have a few velocities in 2005 and 2006 . the cycle count from year to year is ambiguous , but the two most likely choices give periods of 0.217735(3 ) d and 0.217856(4 ) d , which differ in frequency by one cycle per 393 d. some of our spectra were taken above minimum light . while an early m - type secondary star spectrum is evident , it did not yield usable radial velocities .
the spectral type and contribution were fairly uncertain , so the inferences in table [ tab : inferences ] are fairly crude .
the interstellar extinction at this position ( @xmath112 @xmath113 ) is fairly sizeable
@xcite estimate @xmath114 to the edge of the dust layer and the summed spectrum does show weak diffuse interstellar bands . as a result of all these factors ,
the final distance estimate is uncertain by a factor of around 2 .
v587 lyr , also designated tk 5 , was discovered by @xcite , who classified it as a dwarf nova based on the observation of two outbursts on astrograph plates .
the 6.6-hour orbital period found here is within the normal range for dwarf novae .
the spectrum shows a clear contribution from a k - type secondary , and broad emission lines that appear rather weaker than those in otherwise similar stars .
all of our usable spectra are from a single observing run in 2006 june .
because the absorption velocities quickly yielded @xmath0 , we took a comparatively small number of spectra .
we also obtained standardized @xmath22 photometry on 2003 june 19.34 and 2006 august 29.20 ; the 2003 june observation gave @xmath115 , @xmath116 , @xmath117 , and @xmath118 , and the 2006 august observation was very similar .
as is typical , the uncertainty in the distance estimate ( table [ tab : inferences ] ) is dominated by the uncertainty in the secondary s spectral type .
v792 cyg was discovered by @xcite , who found several typical u gem - type outbursts on photographic plates . @xcite
published a spectrum taken with the source well above minimum light , and noted a possible secondary - star contribution .
we have observations from runs in 2007 july and 2007 august , which were separated by about 35 days .
the cycle count between those two runs is ambiguous .
both the absorption and emission velocities slightly favor a period of 0.2972(1 ) d , but periods of 0.2996(1 ) and 0.2948(1 ) d are also possible .
these alternate periods differ by about 3 minutes from the most likely choice , so the rough period is well - determined .
the period quoted in table [ tab : parameters ] is computed assuming the most likely choice of cycle count is correct , but the monte carlo simulation gives a discriminatory power of only @xmath119 for the alias choice .
the @xmath54 k5 spectral type of the secondary is about as expected given the 7.1 h period @xcite .
the mass ratio indicated by the ratio of the absorption and emission velocity amplitudes is @xmath120 , but the emission line radial velocity modulation differs from antiphase with the absorption by @xmath121 cycles , so this calculation is somewhat suspect . for component masses broadly typical of cvs ( @xmath122 m@xmath52 , @xmath123 m@xmath52 ) , the @xmath47-velocities indicate an orbital inclination of @xmath124 .
the system is therefore unlikely to eclipse .
this star was mis - identified in earlier editions of the downes catalog , but we were able to contribute a correction based on our imaging , and the correct object is indicated in @xcite .
the identification images show the star at @xmath125 , @xmath126 , @xmath127 , and @xmath128 .
the radial velocities are from two observing runs only @xmath129 d apart , followed by another several months later .
the choice of cycle count between these runs is fairly secure , so the 4.35 h period ( table [ tab : parameters ] ) is nearly unambiguous , but it is marginally possible that the true orbital frequency is higher by @xmath130 cycle d@xmath131 ( corresponding to a period @xmath132 min shorter ) .
the @xcite maps give a total @xmath133 in this direction ( @xmath134 , @xmath135 ) .
we found that the secondary spectrum fit was easiest when we de - reddened the spectrum by this amount .
this is somewhat puzzling , since our measured colors ( especially @xmath86 ) are typical for an unreddened dwarf nova . in view of this
we assume a generous uncertainty in the reddening when computing the secondary - based distance ( table [ tab : inferences ] ) .
our averaged spectrum resembles that obtained by @xcite , who described it as a text - book example of a cv spectrum " .
the 3.76-hour period is short compared to most of the objects studied here , and we do not detect the secondary star s contribution .
most of our data are from observing runs in 2007 june and 2007 july , only @xmath136 d apart , and the cycle count between these runs is secure ; we also have a few observations from 2005 .
the cycle count between 2005 and 2007 is unknown , but the precise periods allowed by the data are well - fitted by @xmath137 where the denominator is an integer .
@xcite present a spectrum of this star , which they describe as ` extremely faint ' at the time of their observations ; they did not convincingly detect emission lines , and pronounced the dwarf - nova classification as ` doubtful ' on the basis of the spectrum .
our spectrum appears much more robust , with strong , broad balmer lines and the 4.36-hour orbital period is typical of dwarf novae , so we confirm that v542 cyg is indeed a cv .
we have one set of direct images taken 2005 sep . 15.20 ut with the hiltner 2.4 m telescope , and calibrated with observations of @xcite standard stars ; these gave @xmath138 , @xmath139 @xmath140 , and @xmath141 .
for comparison , the flux - calibrated spectrum published by @xcite implies @xmath142 , so it seems possible that they did not have the star in the slit ( they found some of their fainter targets by blind offset ) .
there is an obvious late - type contribution in our spectrum .
the late - type contribution appears to explain the coexistence of the blue @xmath86 color with the relatively red @xmath87 and @xmath143 colors .
however , the late - type component does not show a significant radial - velocity modulation at the period derived from the emission lines ( see fig .
[ fig : folpl2 ] ) , so the late - type spectrum clearly is not from the secondary
. the late - type spectrum could be from an interloper elsewhere along the line of sight , but in our best direct image ( in the @xmath56-band ) , the point - spread function has a fwhm of 1.0 arcsec ( at 0.275 arcsec pixel@xmath3 ) , yet v542 cyg shows no sign of incipient resolution . in addition , the position of the blended image does not shift between the images in different filters , with an upper limit of @xmath144 arcsec .
the @xmath145-filter image should be dominated by the cv , while the @xmath56-filter light is dominated by the late - type contribution , so the lack of a shift means that the two objects are co - located in the sky to good accuracy .
this suggests that the late - type contributor is physically associated with the cv , even if it is not the roche - filling companion .
v542 cyg would seem to be a triple system . v. goranskij clarified the identification of this star ; he noted its blue color and that it appears to be in outburst on the first palomar observatory sky survey , and suggested that it may be a cataclysmic binary .
our spectra show the typical dwarf nova emission lines , confirming his suggestion .
nearly all of our spectra were taken in quiesence ; they are from two closely - spaced observing runs runs in 2007 june and july .
the h@xmath25 emission - line radial velocities give a best period of 0.24675 d ; a much less likely choice of cycle count between the observing runs yields 0.2436 d. in some of the spectra we were able to measure the velocity of the secondary star .
the absorption velocities were less accurate than the emission velocities , but they independently gave the same orbital period .
the secondary s spectral type , m1.5 @xmath40 1.0 , is approximately as expected for the 5.9-h orbital period @xcite .
@xcite reviewed the discovery history of this star and published a spectrum that shows it somewhat brighter and bluer than our mean spectrum .
they described their spectrum as resembling that of a dwarf nova on the decline from outburst , while ours appears to represent a more nearly quiescent state .
the continuum shows weak , broad features that resemble the absorption bands of a secondary star , but we were unable to match these convincingly to our spectral - type standards , so they may not arise from a secondary .
our data come from two short runs separated by two years ; the 4.1 h period is present in both data sets , and is definitive in the combined data , but the cycle count between runs is not determined .
our spectrum resembles the one obtained by @xcite , except that the flux level in our spectrum is somewhat lower and the emission equivalent width appears to be greater . like them
, we detect an m - type companion ; they estimated the spectral type to be in the range m0 to m5 , and from that estimated the distance to be @xmath146 pc , for their latest possible spectral type .
they did not have a @xmath0 with which to constrain the size of the roche lobe .
our 3.85 h period is based on h@xmath25 radial velocities from a single observing run in 2005 september .
we obtained photometry in 2005 september that showed the star near @xmath147 on two separate nights ; the synthesized @xmath55 magnitude from our mean spectrum is also near 18.0 . using the secondary star spectrum ( table [ tab : inferences ] ) we find vz aqr to be considerably greater than the minimum distance found by @xcite .
@xcite found an empirical relationship between the absolute visual magnitudes of dwarf novae at maximum light and their orbital periods .
this predicts @xmath148 at vz aqr s period .
vz aqr reaches @xmath149 , which would imply a distance near 320 pc .
given this , our estimate of @xmath150 pc is probably somewhat greater than the true distance .
as noted above , the normalization of the spectrum used for the distance calculation is corroborated by the photometry derived from direct imaging .
the largest uncertainty in the distance calculation is the spectral classification of the secondary star .
a closer distance would suggest a spectral type somewhat cooler than our estimate .
the x - ray source rxj2133.7 + 5107 was discovered by the @xmath151 galactic plane survey as an x - ray source , and subsequently identified as a dq her - type cv by @xcite .
@xcite presented photometry which revealed a 570 s modulation , related to the white - dwarf spin period , and spectroscopy showing a radial - velocity period of 7.193(16 ) h , or 0.2998(7 ) d , which is evidently @xmath0 .
our h@xmath25 radial velocities show modulation at 0.297432(3 ) d ; while this close to @xcite period , it differs by 3.4 standard deviations .
the interval between our first and last observations is 624 days , and there is no ambiguity in the cycle count over that time . @xcite observed the star on two brief runs some six years apart , and were naturally unable to determine the cycle count between their two epochs , so our period should supersede theirs .
our average spectrum ( not shown ) is very similar in appearance to that that that published by @xcite .
the only new information we present for this star is the improved period .
our main goal in observing these stars was to determine their orbital periods , or , if the periods were already known , to refine them .
the objects discussed here were selected to be those with longer orbital periods . at longer periods
the secondary star tends to be more conspicuous than at shorter periods ; in accordance with expectation , we detect the secondary star s spectrum in most of these objects .
the secondaries spectral types are approximately as expected @xcite for the orbital periods we find .
most of the present sample appear to be typical long - period dwarf novae , without obvious abnormalities .
other types in the cv zoo are also represented .
some of our more noteworthy ( or at least unanticipated ) findings are as follows . 1 .
we do not confirm the presence of a white - dwarf contribution in v709 cas @xcite and argue that it is probably not detectable .
2 . iphas 0345 resembles a long - period dwarf nova , strongly dominated by its secondary - star contribution , but it has never been observed to outburst . in this respect
it resembles ly uma .
we find that v478 her is a long - period novalike variable , rather than a short - period su uma - type dwarf nova .
we report the first secondary - star radial velocity curve for the dq her star v1062 tau , and confirm reports of flaring activity in the literature .
5 . in v542 cyg
, we find a contribution from a late - type star that is evidently not the secondary star , and argue that it is likely to be a triple system .
we gratefully acknowledge support from nsf grants ast-9987334 and ast-0307413 .
we thank the mdm staff for observing assistance , and dartmouth undergraduate jimmy chiang for taking some of the data on v1082 sgr .
john southworth provided a timely and helpful referee report .
as always , we thank the tohono oodham nation for letting us use their mountain for a while .
lrrrrrrl v709 cas & 0 28 48.81 & 59 17 22.1 & 2007.7 & dq & 14.75 b & 15.35 b & 1,2 + af cam & 3 32 15.51 & 58 47 22.4 & 2002.8 & ug & 13.4 p & 17.6 p & 1,2 + iphas j0345 & 3 45 11.59 & 53 35 14.5 & & cv & & 16.0 r@xmath152 & 3 + v1062 tau & 5 02 27.48 & 24 45 23.2 & 2007.7 & dq & 15.9 v & 16.2 v & 1,6 + v344 ori & 6 15 18.95 & 15 30 59.3 & 2008.0 & ugz & 14.2 p & 18.8 v & 1,2,9 + vz sex & 9 44 31.72 & 3 58 06.1 & 2000.3 & ug : & 12.8 v & 16.8 v & 1,2,9 + nsvs 1057 + 09 & 10 57 56.30 & 9
23 15.0 & & ug & 13.0 & 15.5 & 4,8 + v478 her & 17 21 05.62 & 23 39 36.8 & 2006.2 & & 15.5 p & 17.1 p & 1,2 + v1059 sgr & 19 01 50.56 & @xmath15313 09 41.9 & 2003.5 & na & 4.5 & 17.7 v & 1,2,9 + v1082 sgr & 19 07 21.87 & @xmath15320 46 50.5 & 2007.7 & & 14.2 p & 15.9 p & 1,2 + fo aql & 19 16 38.17 & 0 07 37.0 & 2006.4 & ugss & 13.6 p & 17.5 p & 1,2 + v587 lyr & 19 17 26.47 & 37 10 40.8 & 2006.7 & ug & 14.9 & 17.1 & 1,2,9 + v792 cyg & 19 31 01.03 & 33 47 04.0 & 1992.7 & ugss & 14.1 & 17.0 & 5,2 + v795 cyg & 19 34 34.09 & 31 32 11.9 & 2003.5 & ugss & 13.4 & 18.4 v & 1,2,9 + v811 cyg & 19 48 23.31 & 36 26 23.3 & 2004.5 & ugss & 12.7 & @xmath154 & 1,2 + v542 cyg & 19 49 10.51 & 58 31 59.6 & 2005.7 & ugss & 13.1 & 18.3 & 1,2 + pq aql & 19 53 06.67 & 12 59 00.8 & 2002.8 & ug : & 13.0 & 15.0 & 1,2 + v516 cyg & 20 47 09.81 & 41 55 26.3 & 1992.7 & ugss & 13.8 p & 16.8p & 5,2 + vz aqr & 21 30 24.62 & @xmath1532 59 17.5 & 2005.7 & ugss & 11.3 & 18.0 v & 1,2,9 + rx j2133 + 51 & 21 33 43.63 & 51 07 25.0 & 2002.8 & dq & & 16 b & 1,7 + [ tab : star_info ] 51549.8366 & @xmath155 & @xmath156 & @xmath157 & @xmath158 + 51549.8407 & @xmath159 & @xmath160 & @xmath157 & @xmath161 + 51549.8447 & @xmath162 & @xmath163 & @xmath164 & @xmath165 + 51550.8136 & @xmath166 & @xmath167 & @xmath168 & @xmath169 + 51550.8197 & @xmath170 & @xmath171 & @xmath172 & @xmath173 + [ tab : velocities ] v709 cas & 50801.6342(16 ) & 0.2222041(3 ) & 77(4 ) & @xmath174 & 166 & 19 + af cam ( abs ) & 53073.570(4 ) & 0.3240785(14 ) & 105(7 ) & @xmath175 & 36 & 22 + af cam ( emn ) & 52233.700(4 ) & 0.3240780(12 ) & 63(3 ) & @xmath176 & 45 & 12 + af cam ( avg ) & & 0.3240782(9 ) & + iphas j0345 ( abs ) & 54528.802(3 ) & 0.313902(8 ) & 77(5 ) & @xmath177 & 41 & 15 + iphas j0345 ( emn ) & 54528.663(4 ) & 0.313901(11 ) & 99(7 ) & @xmath178 & 38 & 24 + iphas j0345 & & 0.313902(7 ) & & & & + v1062 tau ( abs ) & 53617.988(6 ) & 0.415883(13 ) & 202(13 ) & @xmath179 & 21 & 30 + v1062 tau ( emn ) & 53450.716(9 ) & 0.415933(5 ) & 143(19 ) & @xmath180 & 27 & 46 + v1062 tau & & 0.415926(4 ) & & & & + v344 ori & 54525.633(2 ) & 0.2338(17 ) & 86(4 ) & @xmath181 & 41 & 15 + vz sex ( emn ) & 51994.6631(12 ) & 0.1487(3 ) & 85(5 ) & @xmath182 & 42 & 16 + vz sex ( abs ) & 51994.727(5 ) & 0.1489(15 ) & 42(10 ) & @xmath183 & 33 & 22 + nsvs 1057 ( abs ) & 54277.7603(16 ) & 0.3763076(11 ) & 127(4 ) & @xmath184 & 46 & 12 + nsvs 1057 ( emn ) & 54297.509(3 ) & 0.376312(4 ) & 107(4 ) & @xmath185 & 54 & 15 + nsvs 1057 & & 0.3763079(11 ) & & & & + v478 her ( abs ) & 53182.904(7 ) & 0.629049(5 ) & 138(10 ) & @xmath186 & 51 & 41 + v478 her ( emn ) & 53182.688(12 ) & 0.629048(14 ) & 146(16 ) & @xmath187 & 135 & 70 + v1059 sgr ( emn ) & 54648.932(3 ) & 0.2861(7 ) & 138(11 ) & @xmath188 & 53 & 36 + v1082 sgr ( abs ) & 53562.065(12 ) & 0.867547(19 ) & 55(5 ) & @xmath189 & 101 & 17 + fo aql & 54277.9139(16 ) & 0.217735(3 ) & 96(6 ) & @xmath190 & 31 & 12 + v587 lyr ( abs ) & 53907.941(4 ) & 0.2750(10 ) & 81(7 ) & @xmath191 & 15 & 16 + v587 lyr ( emn ) & 53907.813(5 ) & 0.2731(10 ) & 144(15 ) & @xmath192 & 15 & 32 + v587 lyr & & 0.2740(7 ) + v792 cyg ( emn ) & 54302.618(4 ) & 0.29715(12 ) & 118(9 ) & @xmath193 & 28 & 24 + v792 cyg ( abs ) & 54300.083(2 ) & 0.29722(19 ) & 132(8 ) & @xmath194 & 19 & 18 + v792 cyg & & 0.29717(10 ) + v795 cyg ( emn ) & 54296.736(2 ) & 0.181299(17 ) & 90(6 ) & @xmath195 & 71 & 21 + v811 cyg ( emn ) & 54282.715(2 ) & 0.15656(4 ) & 115(13 ) & @xmath196 & 45 & 31 + v542 cyg ( emn ) & 53907.738(3 ) & 0.1815(5 ) & 88(8 ) & @xmath197 & 20 & 17 + pq aql ( emn ) & 54282.002(3 ) & 0.24675(4 ) & 147(9 ) & @xmath198 & 50 & 35 + pq aql ( abs ) & 54281.885(7 ) & 0.2470(3 ) & 155(23 ) & @xmath199 & 25 & 62 + pq aql & & 0.24675(4 ) + v516 cyg ( emn ) & 54337.0034(13 ) & 0.1712(4 ) & 88(5 ) & @xmath200 & 51 & 15 + vz aqr ( emn ) & 53625.867(2 ) & 0.1606(3 ) & 44(5 ) & @xmath201 & 63 & 13 + rx 2133 ( emn ) & 53074.033(3 ) & 0.297431(5 ) & 69(5 ) & @xmath202 & 49 & 10 + [ tab : parameters ] iphas j0345 & k5.5 @xmath40 1.5 & @xmath203 & @xmath204 & @xmath205 & @xmath206 & @xmath207 & @xmath208 + v1062 tau & k5 @xmath40 2 & @xmath209 & @xmath210 & @xmath211 & @xmath212 & @xmath213 & @xmath214 + v344 ori & m3 @xmath40 0.75 & @xmath215 & @xmath216 & @xmath217 & @xmath218 & @xmath219 & @xmath220 + vz sex & m1 @xmath40 0.75 & @xmath221 & @xmath222 & @xmath223 & @xmath224 & @xmath225 & @xmath226 + nsvs 1057 + 09 & k4.5 @xmath40 2 & @xmath227 & @xmath228 & @xmath229 & @xmath230 & @xmath231 & @xmath232
+ v478 her & k5 @xmath40 2 & @xmath233 & @xmath234 & @xmath235 & @xmath236 & @xmath237 & @xmath238 + v1082 sgr & k4
@xmath40 0.2 & @xmath239 & @xmath240 & @xmath241 & @xmath242 & @xmath243 & @xmath244 + fo aql & m2 @xmath40 1.5 & @xmath245 & @xmath216 & @xmath246 & @xmath247 & @xmath240 & @xmath248 + v587 lyr & k7 @xmath40 1.5 & @xmath249 & @xmath204 & @xmath250 & @xmath251 & @xmath252 & @xmath253 + v792 cyg & k5 @xmath40 2 & @xmath254 & @xmath255 & @xmath256 & @xmath257 & @xmath216 & @xmath258 + v795 cyg & m3.5 @xmath40 1.3 & @xmath259 & @xmath216 & @xmath260 & @xmath261 & @xmath262 & @xmath263 + pq aql & m1.5 @xmath40 1.0 & @xmath264 & @xmath265 & @xmath266 & @xmath267 & @xmath268 & @xmath269 + vz aqr & m2.0 @xmath40 1.5 & @xmath270 & @xmath271 & @xmath272 & @xmath273 & @xmath225 & @xmath150 + [ tab : inferences ] | we obtained time - series radial velocity spectroscopy of twenty cataclysmic variable stars , with the aim of determining orbital periods @xmath0 .
all of the stars reported here prove to have @xmath1 h. for sixteen of the stars , these are the first available period determinations , and for the remaining four ( v709 cas , af cam , v1062 tau , and rx j2133 + 51 ) we use new observations to improve the accuracy of previously - published periods .
most of the targets are dwarf novae , without notable idiosyncracies . of the remainder ,
three ( v709 cas , v1062 tau , and rx j2133 + 51 ) are intermediate polars ( dq her stars ) ; one ( iphas 0345 ) is a secondary - dominated system without known outbursts , similar to ly uma ; one ( v1059 sgr ) is an old nova ; and two others ( v478 her and v1082 sgr ) are long - period novalike variables .
the stars with new periods are iphas 0345 ( 0.314 d ) ; v344 ori ( 0.234 d ) ; vz sex ( 0.149 d ) ; nsvs 1057 + 09 ( 0.376 d ) ; v478 her ( 0.629 d ) ; v1059 sgr ( 0.286 d ) ; v1082 sgr ( 0.868 d ) ; fo aql ( 0.217 d ) ; v587 lyr ( 0.275 d ) ; v792 cyg ( 0.297 d ) ; v795 cyg ( 0.181 d ) ; v811 cyg ( 0.157 d ) ; v542 cyg ( 0.182 d ) ; pq aql ( 0.247 d ) ; v516 cyg ( 0.171 d ) ; and vz aqr(0.161 d ) .
noteworthy results on individual stars are as follows .
we see no indication of the underlying white dwarf star in v709 cas , as has been previously claimed ; based on the non - detection of the secondary star , we argue that the system is farther away that had been thought and the white dwarf contribution is probably negligible . v478
her had been classified as an su uma - type dwarf nova , but this is incompatible with the long orbital period we find .
we report the first secondary - star velocity curve for v1062 tau . in v542 cyg , we find a late - type contribution that remains stationary in radial velocity , yet the system is unresolved in a direct image , suggesting that it is a hierarchical triple system . |
correlated percolation@xcite is a useful theoretic model in statistical physics .
it provides us with fundamental understanding of spread processes of message , disease and matter in nature and society . linking probability between any two nodes in it takes the form of @xmath4 , where @xmath5 is @xmath6-dimensional distance between the nodes , and
@xmath0 is a positive real number , namely , distance - decay exponent of links .
weinrib and halperin@xcite analytically studied whether the correlations change the percolation behavior or not .
weinrib@xcite pointed out , for @xmath7 , the correlations are relevant if @xmath8 , where @xmath9 is the percolation - length exponent for uncorrelated percolation ; while for @xmath10 the correlations are relevant if @xmath11 .
it is a generalization of the harris criterion@xcite appears earlier .
recently , network models referring to correlated percolation have gradually appeared@xcite .
achlioptas process(ap)@xcite for link - adding networks , which is an attractive topic at present@xcite , could be viewed as a new kind of correlated percolation if we put all nodes uniformly on a two - dimensional(2d ) plane .
starting from a set of isolated nodes , two candidate links are put to nodes randomly at every time step , but only the link with smaller product @xmath12 is retained , where @xmath13(or @xmath14 ) is the mass ( the number of nodes ) of the cluster that node i(or j ) belongs to , which is called product rule(pr ) . a link chosen with pr from both inter - cluster candidates is called a type - i link .
when two candidate - links are of different types , i.e. , one is an inter - cluster link , the other is a intra - cluster one , always the later is retained , and it is called a type - ii link . while for both intra - cluster ones , the retained link is arbitrarily chosen no matter they are in the same or different clusters , and it is called a type - iii link . generally speaking , in the way of ap , network percolation
is inhibited , which postpones the appearance of the threshold @xmath15 at which a giant component g starts to grow , and results in a sharp growth of g called an explosive percolation . in our point of view , if we put ap on a 2d plane , it gives rise to a new mechanism of long - range correlation for the nodes based on co - evolutionary growing masses of components they connected .
the selective rule for topological links relies on mass - product instead of 2d geometric length of them , which prevents the property exhibited in the previous correlated percolation . and
correlation feature in ap - type of percolation has not been revealed up till now . a recent model of ours@xcite based on the observation of phenomena in different real systems@xcite describes another kind of correlated percolation in growing networks , which could be viewed as a overlapping of traditional correlated percolation with ap in a 2d space .
the link - occupation function in the model takes the form @xmath16 which looks like newton s gravity rule .
it resumes the classical erdos - renyi(er ) random graph model when exponent @xmath17 , and it gives another extreme of ap when @xmath18 .
different properties of such kind of new correlated percolation are expected , since er random graph grows without any bias , ap takes a strong bias to inhibit network percolation independent of geometric distance , while the new model with gravity - like rule have some distance - related relax on such bias , which produces a new type of correlated percolation . in this paper , we report the simulation results on objective competition between type - i , ii and iii links in both gravity model@xcite and ap model , and we point out a new mechanism to support critical points , which bears the scaling relations revealed in our recent work .
different saturation effect is manifested , which distinguishes it from the traditional correlated percolations .
suppose @xmath19 isolated nodes are uniformly scattered on a 2d plane . for convenience of calculating distance ,
the plane is discretized with a triangular lattice , each minimal edge with the length of two units for the convenience of algorithm .
each vertex of the triangles is occupied by a node so that we exclude all possible biases except link - adding rules . for any two pairs of nodes @xmath20 and @xmath21 possibly with the same product @xmath12 , a type - i link connects the the pair with longer distance if both the links ends hit the nodes belonging to different clusters ; while a type - ii link connects the nodes inside the same cluster if the other one is an inter - cluster link ; a type - iii one connects arbitrarily chosen pair of nodes if both candidate links are intra - cluster ones .
parallel to pr , we pick randomly two pairs [ @xmath22 and @xmath23 of nodes in the plane at every time step . for the pair @xmath22 ( and for @xmath24 likewise ) , we calculate the generalized gravity defined by @xmath25 , where @xmath26 is the geometric distance between @xmath20 and @xmath21 , and @xmath0 is an adjustable decay exponent .
once we have @xmath27 and @xmath28 , we have two choices in selecting which pair to connect . for the case of the maximum gravity strategy ( we call it @xmath29 ) we connect the pair with the larger value of the gravity , e.g. , the link @xmath22 is made if @xmath30 and the link @xmath24 otherwise .
we also use the minimum gravity strategy ( @xmath31 ) in which we favor the smaller gravity pair to make connection .
the two strategies , @xmath32 and @xmath31 , lead the link - adding networks to evolve along the opposite paths of percolation processes . generally speaking , @xmath29 facilitates the percolation process , whereas @xmath31 inhibits it .
all such generalized gravity values are calculated inside the circular transmission range with the radius @xmath1 centered at one of nodes @xmath20 and @xmath21 as the speaking node @xcite in a mobile ad hoc network@xcite . for the different limits of parameters @xmath1 and @xmath6
, we have three cases in the model .
case i : with the transmission range @xmath33 , we have a generalized gravitation rule which is an extension @xcite of widely used gravitation model @xcite @xmath34 with the tunable decaying exponent @xmath0 .
case ii : with the exponent @xmath35 , we assume that node pairs can be linked with pr topologically inside the transmission range with a limited radius @xmath1 .
case iii : with both limited values of radius @xmath1 and exponent @xmath0 , we have the gravity rule inside the transmission range .
it can describe the communication or traffics with constrained power or resources . for case
i and case iii in the model , three scaling relations have been found with large scale simulations .
when strategy @xmath36 is adopted in 2d free space(case i ) , we have @xmath37 where @xmath38 is dimensionless time - step with @xmath39 , @xmath0 is the decay exponent of connection probability .
@xmath40 , @xmath41 , and @xmath42 is a universal function . when strategy @xmath43 is adopted inside the transmission range with radius @xmath1(case iii )
, we have @xmath44\ ] ] for certain parameter ranges of @xmath0 and @xmath1 , where @xmath38 , @xmath45 , @xmath46 , @xmath47 , @xmath48 , and @xmath49 is a universal function .
in addition , when strategy @xmath36 for case iii is adopted inside transmission range defined by @xmath1 , we have another scaling relation @xmath50 for @xmath51 , where @xmath52 , @xmath53 , @xmath54 , @xmath55 , @xmath56 and @xmath57 is a universal function . to understand three scaling relations above , we should look into the mechanism of the evolution processes underlying @xmath58 . to see what happens in such critical points @xmath59 , and what are particular of them in certain link - adding processes , we count the temporal link fractions @xmath60 , and calculate the average lengths @xmath3 of links which is defined as the summation of all lengths of links for a certain type over its number in a window @xmath61 time - steps . by observation of the time - dependent behaviors of fractions of type - i , ii and iii links ,
new properties were found out for our gravity - like model together with ap producing explosive percolations .
all simulations are carried out on the triangular lattice of the size @xmath62 with @xmath63 and @xmath64 , respectively .
we simulate either of strategy @xmath29 or @xmath31 for either case i or iii .
the total number of links equating to that of time - steps is divided by @xmath19 , which is defined as @xmath65 .
the mass of the largest component divided by @xmath19 makes up the observable @xmath66 , the node fraction of the largest component .
all results presented in this work are obtained from 5000 different realizations of network configurations with @xmath67 if not specially indicated . inspired by cho and kahng s @xcite work and a referee of ref.@xcite
, we have gone further by calculating fractions of three types of links and arithmetic average lengths of their links .
our attention was pointed at ap first . in fig.1
we illustrated the evolution of fractions of 3 types of links .
just at the threshold the fraction of type - i@xmath68 links has a sharp drop - down , meanwhile that of type - ii @xmath69 shoots up , crossing @xmath70 at @xmath71 . a little after it , @xmath70 crosses with growing fraction of type - iii links ( @xmath72 ) at the level @xmath73 , while @xmath74 gets its summit(@xmath75 ) at the same point , which has not been concerned by previous works . however , it is this property that pervades all cases in the present correlated percolation . in fig.2 , the average lengths of type - ii(@xmath76 )
merges that of type - iii(@xmath77 ) at @xmath15 after an abrupt growth , @xmath76 starts to grow earlier than @xmath77 .
the level of @xmath3 for both of them keep invariant for @xmath78 , while @xmath79 starts to decrease from @xmath15 .
we see from both the figures that in explosive percolation the system undergoes a sharp transition from a type - i link dominant phase into a type - ii and iii dominant phase at @xmath15@xcite .
besides , average lengths undergo a parallel transition at the same point .
actually , 3 levels of @xmath3 go to infinity in dynamic limit from finite size scaling transformation(not shown ) . and probability decay exponent @xmath80 in case i ( @xmath33).,width=336 ] of type - i , type - ii and type - iii links with @xmath36 and probability decay exponent @xmath81 in case i.,width=336 ] of the largest component with strategy @xmath29 and the same parameters in fig.3.,width=336 ]
now we turn to the fraction of 3 types of links in case i of the present gravity model@xcite . with strategy @xmath36 in free 2d space , we have scaling relation ( 1 ) for distance - decay exponent @xmath82 $ ] .
fig.3 shows the evolution of fractions corresponding to it .
curves for @xmath70 cross those of @xmath74 with @xmath83 and @xmath84 around @xmath85 quite clearly , even those with @xmath86 barely cross near it . but
fractions for @xmath87 and @xmath88 shift the cross point rightward obviously . on the other hand ,
the curves for @xmath89 cross @xmath72 at @xmath90 for almost all values @xmath0 except @xmath86 . to determine which one would be the candidate of another critical point
, we cast ourselves on the assistance from the observation of average lengths of links . in fig.4 @xmath91 merges @xmath91 for @xmath87 and @xmath88 at @xmath90 , separating themselves from others .
however , almost all other ones collect at @xmath85 , which gives hint to us for @xmath92.(here a better resolution is needed in further calculation ) .
correspondingly , in fig.5 for scaling relation ( 1 ) with exponents @xmath93 , @xmath40 , and @xmath94 , @xmath58 for @xmath95 $ ] collapse into the universal function very well , with that for @xmath86 barely collapsing onto it .
but those @xmath58 for @xmath87 and @xmath88 do not behave well in collapse .
the separation from others at the turning middle part indicates the deviation of their @xmath92 from @xmath96 with which others share . in the description of the average lengths for case i with @xmath36 , simulated results @xmath97 in fig.4 with all values
@xmath0 demonstrate the same steady level ( @xmath98 for @xmath67 ) .
variation of parameter @xmath0 only shifts starting points of up - growing @xmath74 and @xmath72 as @xmath0 increases .
the saturation effect of large decay exponents @xmath99 appears clearly and is shown by dash lines , which demonstrates the inheritance from traditional correlated percolation@xcite . in this case with @xmath36 , special level of fractions at cross point of @xmath100 and @xmath101 keeps 0.25 just as in ap without any distance - decay included , so does @xmath74 at hiking its summit . of the system with @xmath36 in case i. @xmath63 and @xmath64.(b)percolation thresholds @xmath15 for @xmath102 and @xmath88 with @xmath36 in case i.,width=336 ] of type - i , type - ii and type - iii links with strategy @xmath29 in case i. parameters are the same as in fig.6,width=336 ] as in a usual way , we determine the critical point @xmath15 of percolation by observation of tips of susceptibility @xmath103 ( fig.6 ) . comparing @xmath15 in fig.6 with @xmath97 in fig.4
, we find that these @xmath15 approximately hit the horizontal coordinates of middle point of growing fraction of type - ii links , which means that @xmath15 is the transition point from the inter - cluster - link dominant phase to the intra - cluster - link dominant phase .
besides @xmath15 , we have another ( sub)-critical point @xmath92 which is in certain range independent of decay exponent @xmath0 in gravity model , and @xmath92 indicates the balance between the fractions of type - i and type - ii links , yielding a new scaling behavior of @xmath104 in formula ( 1 ) not revealed by previous works .
moreover , the steady level @xmath105 is always @xmath106independent , type - independent which takes the inherited value of that in ap . actually , 131.5 is the value for l=128 only .
we have size effect since a free boundary condition instead of a periodic one is adopted .
the finite size effect is shown in fig.7 which gives that @xmath107 , i.e. , @xmath108 where @xmath19 is the number of nodes on the 2d plane .
hopefully it goes towards infinity in thermodynamic limit .
however , the finite size effect of @xmath109 ( fig.8 for an example ) is not strong enough for us to identify the scaling exponents @xmath110 as usual .
therefore , we can check the validation of scaling laws presented by weinrib @xcite for correlated percolation in the present model only by rescaling susceptibility @xmath111 . fig . 9
illustrates the results of it for examples @xmath112 and @xmath84 , respectively .
we have scaling relation of the largest component with strategy @xmath29 in case i. ( a)@xmath112 ; ( b)@xmath113 . @xmath114 and @xmath64.,width=336 ] in case i for ( a ) @xmath112 ; ( b ) @xmath113 . @xmath114 and @xmath64.,width=336 ] @xmath115 where @xmath116 , @xmath117 , and @xmath118 is a universal function . with these values of scaling exponents , the scaling law @xmath119 is not applicable to the present model , where @xmath120 is correlation - length exponent in the long - range case .
with power - law form for the correlation function g(r ) , weinrib had derived the extended harris criterion : the long - range nature of the correlations is relevant if @xmath121 , which means the correlations change the percolation critical behavior .
it has been violated since now they all behave differently from traditional short range percolation in a 2d triangular lattice ( @xmath122 ) and the correlations are relevant no matter @xmath123 is less(fig.9(a ) ) or larger(fig.9b ) than zero .
this is because in strategy @xmath36 we have overlapped the power - law correlation function g(r ) with ap which is another kind of autocorrelation process with positive feed back effect of mass - growing .
and probability decay exponent @xmath124 and @xmath88 in case i. @xmath125.,width=336 ] and probability decay exponent @xmath124 and @xmath88 in case i. @xmath125.,width=336 ] for the strategy @xmath43 which prefers smaller gravity , it tends to retain a longer link under the comparison of the same product of masses @xmath126 .
that is to say , long range links have predominance .
evolution of @xmath100 and @xmath127 links do not cross at any common point . in fig.10 ,
the cross points for @xmath89 and @xmath74 shift leftward from @xmath15 of ap as decay exponent @xmath0 increases , which means that @xmath43 as a correlated percolation mechanism weaken the explosive effect caused by ap .
but the starting position of fraction - ii still provides hints of thresholds @xmath15 .
however , it is hard to locate a common cross point for @xmath89 and @xmath72 in a range of @xmath0 .
therefore , we have no scaling relation for them .
the steady average lengths of links ( fig.11 ) have the same level of @xmath36 and ap cases , and size - effect ( not shown ) also tells the divergence of @xmath3 , but all of them do not tell any possible hint for critical points
. generally speaking , strategy @xmath43 facilitates longer links for certain geometric distribution of clusters or nodes . in the evolution ,
strategy @xmath43 emphasis the assignments for different types of links , encourage longer and intra - cluster links .
humps above the steady @xmath74 and @xmath72 implies out - of - pace growing of link lengths of @xmath89 .
that is , @xmath89 surpasses the growing speed of the giant component . here
, it is the geometric distance - dependent strategy that makes @xmath43 alleviate effect of ap . with smaller @xmath0 ( e.g. , a=0.2 in fig.11)the strategy has the opportunity to exhaust long links before @xmath15 ; while with middle values of @xmath0 ( e.g. , @xmath128 and @xmath84 ) it may take longer time to exhaust them . however , with too large @xmath0 ( e.g. , @xmath129 )
, @xmath43 fall off quicker than the natural dimension , we can only see the ap - type short - range effect of saturation . in this limit ,
i.e. , @xmath130 , percolations are no longer relevant , which causes saturation of curves @xmath58 in fig.1b @xcite .
however , we should not expect the short - range percolation exponent @xmath131 of correlation length here for 2d triangular lattice , since ap has been included in the strategy @xmath43 . and probability decay exponent @xmath132 and @xmath133 in case iii .
@xmath134.,width=336 ] and probability decay exponent @xmath132 and @xmath133 in case iii .
@xmath134.,width=336 ] with strategy @xmath43 , @xmath135 and @xmath136 in case iii .
@xmath134.,width=336 ] with strategy @xmath43 , @xmath137 and @xmath133 in case iii .
@xmath134.,width=336 ] and probability decay exponent @xmath138 and @xmath88 in case iii .
@xmath139.,width=336 ] the distinct feature in case iii for both @xmath43 and @xmath36 is that the candidate links are selected not only by comparing gravities , but also constrained inside a transmission range @xmath1 ( @xmath140 in geometric distance ) , which ruins the effect comes from the divergence of average link lengths .
it is well known that all possible singularities at critical points come from the singularity of correlation length .
however , here no length could goes to infinity in any way , which ruins possible common cross point relies on the balance between @xmath89 and @xmath74 as in the 2d free space ( case i of the present model)and induces the possibility to yield novel scaling relation other than any previous ones . for possible critical point , we seek help from the evolution of link fractions of 3 types .
fig.12 shows the behaviors of @xmath89 , @xmath74 and @xmath72 with @xmath43 for all simulated distance - decay exponents @xmath0 and @xmath134 .
correspondingly , fig.13 shows the behaviors of @xmath79 , @xmath76 and @xmath77 the the same set of parameters .
critical point @xmath45 ( @xmath141 to be precise ) distinguishes itself from others by intuitive observation . the cross point for @xmath129 and @xmath142 go rightward from which others share , which means they could not share the same @xmath59 hence the same scaling relation with other decay exponents . besides , in rescaling process for @xmath104 , curves for @xmath86 and @xmath112 failed in collapse , because smaller transmission range @xmath1 inhibits the effect of slower(long - range ) decay for connection probability .
the rescaled function @xmath143 for @xmath134 is shown in fig.14 .
it seems that we could go further with @xmath144 to include more exponents @xmath0 in the scaling as shown in fig.15 .
however , it is meaningless in physics due to above mentioned reasons . actually , scaling behaviors are r - dependent , but exponents @xmath145 and @xmath146 need not to vary .
the variation of @xmath1 only shifts @xmath59 as illustrated in fig.16 for @xmath139 . therefore , we keep @xmath147 , @xmath46 for all values of @xmath1 with @xmath43 , but take @xmath141 for @xmath134 , @xmath148 for @xmath139 , and so on .
the humps above the steady level of @xmath76 and @xmath77(all independent of exponents @xmath0 ) come from similar mechanism as in free 2d space but now at much lower level constrained by transmission radius @xmath1 , and they are independent of size @xmath149 of the system . and probability decay exponent @xmath113 for @xmath150 and @xmath151 in case iii . @xmath125 .
500 realizations of network configurations.,width=336 ] and probability decay exponent @xmath129 for @xmath152 and @xmath151 in case iii.,width=336 ] and probability decay exponent @xmath113 for @xmath150 and @xmath151 in case iii .
500 realizations of network configurations.,width=336 ] and probability decay exponent @xmath129 for @xmath152 and @xmath151 in case iii.,width=336 ] with strategy @xmath36 , @xmath113 for @xmath153 and @xmath151 in case iii.@xmath125 .
500 realizations of network configurations.,width=336 ] with strategy @xmath36 , @xmath129 for @xmath152 and @xmath151 in case iii.,width=336 ] scaling relation ( 3 ) for case iii with strategy @xmath36 inside transmission range with radius @xmath1 is checked for various decay exponents @xmath0 and for different sizes ( l=32 , 64 , 128 and 256 ) .
its validation is independent of size @xmath149 simulated . in fig.17 and fig.18 ,
the evolution of @xmath89 , @xmath74 and @xmath72 for both @xmath113 and @xmath129 behave much similarly .
@xmath89 and @xmath74 cross at the level a little bit lower than @xmath154 , while @xmath89 crosses @xmath72 at the level @xmath155 , which keeps the same as in all previous cases .
the changes of exponent @xmath0 and @xmath1 only shift fractions along horizontal direction of figures , i.e. , to change starting points and growing / dropping speed instead of levels of them .
however , @xmath56 keeps as their common fixed point for @xmath89 and @xmath72 to cross . in fig.19 and fig.20 , @xmath89 and @xmath72 for both @xmath113 and @xmath129
arrive at the same level hitting @xmath45 , which distinguishes this point from totally 3 cross points , and makes up a candidate of critical point @xmath92 for scaling relations .
the scaling exponents @xmath54 and @xmath55 have been checked for lower values of parameter @xmath0(@xmath156 , fig.21 )
. however , for @xmath129 , we have to choose a new set of exponents : @xmath157 and @xmath158(fig.22 ) .
it is not strange that steady levels of @xmath3 keep unchanged for certain @xmath1 , independent of @xmath0 or @xmath149 , just as that with @xmath43 in case iii .
@xmath3 inside a circle defined by @xmath1 can not go to infinity under any circumstance , but still support a critical point , which distinguishes correlated percolation in case iii from case i and traditional models .
it deserves further investigation .
in this paper , we have proposed a new network model of correlated percolation in which geometric distance - dependent power - law decay connection probability overlaps achlioptas process to form a gravity model .
it can be tuned to facilitate or inhibit percolation with strategy @xmath36 or @xmath43 , cover a wide range of thresholds @xmath15 , yield a set of new scaling relations . and it provides a scheme for better description of practical processes in complex systems . we have developed a new approach to find out candidate critical points with physical meanings other than that of traditional ones .
there are objective competition and balance between type - i and type - ii , type - i and type - iii links , meanwhile , competition of average lengths between type - ii and type - iii links . along this line threshold @xmath15
is found to overlap the balance point between factions @xmath70 and @xmath74 in the explosive percolation of achlioptas process , and the steady average lengths of three types of links are all divergent to infinity in thermodynamic limit .
the percolation is indeed a transition from type - i link dominant phase to type - ii and type - iii dominant phase . by observing evolutions of fractions of type - i , type - ii and type - iii links ,
a candidate critical point can be chosen combined with the message on evolutions of average lengths of them . with strategy @xmath36 in 2d triangular lattice ,
fraction @xmath89 get balance with @xmath74 , makes up a critical point @xmath92 which supports scaling relation ( 1 ) in case i of the model . with strategy @xmath43 inside certain transmission range with radius @xmath1
, a duet balance exists for @xmath89 and @xmath72 meanwhile @xmath76 and @xmath77 , makes up another critical point @xmath92 which supports scaling relation ( 2 ) in case iii . with strategy @xmath43 and certain range @xmath159 of decay exponent @xmath0 ,
again a duet balance exists for @xmath89 and @xmath72 meanwhile @xmath76 and @xmath77 , makes up another critical point @xmath92 which supports scaling relation ( 3 ) for a mini - scale of @xmath1 in case iii .
this approach serves an assistant tool in seeking critical points of order parameter @xmath58 which is usually not easy to determine in an intuitive way . in numerical calculations , besides percolation threshold @xmath15 , two fixed points , @xmath39 and @xmath56 emerge as distinct points not only for special temporal crux but also for unchanged levels of @xmath70 , @xmath74 and @xmath72 inherited from ap , which is expected to be further proved in analytical ways .
however , they have different physical meanings .
the former corresponds to a divergent average length of links , while the later corresponds to confined average lengths by transmission range @xmath1 , which distinguishes itself from traditional critical points in percolations .
correlated percolations are relevant since long - range correlation drastically changes the critical properties .
the validation ranges of decay exponents @xmath0 with various strategies in different cases define the relevance of correlation .
they have demonstrated novel scaling relations different from traditional 2d short - range percolation in triangular lattice .
the intervention of distance - dependent power - law decay ingredients alleviates the explosive effect of percolation transition by horizontal adjustment of evolutions along temporal axis , separates @xmath92 from @xmath15 , while the overlapped ap included in the present gravity model always conquers the vertical levels of three fractions and average lengths , which are found neither in traditional correlated percolations of continuities in 2d space nor in complex networks .
moreover , the node fraction @xmath58 of the largest component , fractions @xmath60 , and average lengths @xmath97 of three types of links all show saturation phenomena as pointed out by weinrib but with different values of exponent of @xmath0 since now ap overlaps in the present gravity model . and scaling law of weinrib
is no longer obeyed according to the evidence of numerical results of average - length exponents .
we are indebt to anonymous referees for stimulating comments .
zhu thanks h. park , p. holm , x .- s .
chen and z .- m .
gu for useful discussion .
we acknowledge financial support from national natural science foundation of china ( nnsfc ) under grants no .
11175086 , 10775071 and 10635040 .
kim was supported by the national research foundation of korea ( nrf ) funded by the korea government ( mest ) under grant no .
2011 - 0015731 . c. e. perkins , _ ad hoc networking_. ( addison - wesley , new york , 2000 ) . l. wang _
et al_. , phys .
e * 78 * , 066107 ( 2008 ) .
j. tinbergen , twentieth century fund , new york ( 1962 ) ; p. poyhonen , weltwirtschaftliches archiv , 93 ( 1963 ) ; j. e. anderson , the american economic review , 69 , 106(1979 ) .
j. h. bergstrand , the review of economics and statistics.1985 .
a. v. deardorff , the regionalization of the world economy , chicago univ . press , chicage , usa ( 1998 ) .
e. helpman , j. eco . lit .
, 44 , 589(2006 ) . | motivated by the importance of geometric information in real systems , a new model for long - range correlated percolation in link - adding networks is proposed with the connecting probability decaying with a power - law of the distance on the two - dimensional(2d ) plane . by overlapping it with achlioptas process
, it serves as a gravity model which can be tuned to facilitate or inhibit the network percolation in a generic view , cover a broad range of thresholds . moreover , it yields a set of new scaling relations . in the present work ,
we develop an approach to determine critical points for them by simulating the temporal evolutions of type - i , type - ii and type - iii links(chosen from both inter - cluster links , an intra - cluster link compared with an inter - cluster one , and both intra - cluster ones , respectively ) and corresponding average lengths .
numerical results have revealed objective competition between fractions , average lengths of three types of links , verified the balance happened at critical points .
the variation of decay exponents @xmath0 or transmission radius @xmath1 always shifts the temporal pace of the evolution , while the steady average lengths and the fractions of links always keep unchanged just as the values in achlioptas process .
strategy with maximum gravity can keep steady average length , while that with minimum one can surpass it . without the confinement of transmission range , @xmath2 in thermodynamic limit , while @xmath3 does not when with it .
however , both mechanisms support critical points . in two - dimensional free space ,
the relevance of correlated percolation in link - adding process is verified by validation of new scaling relations with various exponent @xmath0 , which violates the scaling law of weinrib s . |
chandra and xmm - newton observations have shown that there is a strong feedback mechanism between the central agn and the energetics of the hot gas in groups and clusters of galaxies ( e.g. , peterson & fabian 2006 and mcnamara & nulsen 2007 ) .
agn outbursts in the central dominant galaxy in groups and clusters , which are themselves fueled by the accretion of cooling gas , can produce shocks , buoyant cavities and sound waves , all of which lead to re - heating of the cooling gas .
this agn - cooling flow feedback mechanism is also an important process in galaxy formation and can explain the observed correlation between bulge mass and central black hole mass ( e.g. , gebhardt et al . 2000 ) and the cut - off in the number of massive galaxies ( croton et al .
2006 ) .
agn outbursts can have a much more significant impact on the energetics of the hot gas in groups compared to that in richer clusters , due to their shallower potential wells . in david et al .
( 2009 ) , hereafter paper i , we presented the results of an 80 ksec chandra observation of the x - ray bright group , ngc 5044 , and showed that the central region of this group has been perturbed by several recent agn outbursts and also by motion of the central galaxy relative to the surrounding intragroup gas .
gmrt observations at 235 mhz and 610 mhz show direct evidence for uplifting of low entropy gas from the center of ngc 5044 behind buoyant , agn - inflated x - ray cavities .
most of the x - ray cavities in ngc 5044 , however , remain radio quiet in our gmrt observations .
the roughly isotropic distribution of the small radio quiet cavities suggests that the group weather has a significant impact on the dynamics of these bubbles as they buoyantly rise .
we also showed in paper i that the total mechanical power of the small radio quiet cavities is sufficient to suppress about one - half of the total radiative cooling within the central 10 kpc . in this paper , we discuss how agn driven turbulence has produced a multiphase medium in the center of ngc 5044 and how and this impacts estimates of the abundance and distribution of heavy elements .
this paper is organized in the following manner .
details of the acis data analysis are described in @xmath3 .
section 3 presents temperature and abundance maps for the center of ngc 5044 and @xmath4 contains a discussion about systematic effects in the derivation of elemental abundances when fitting emission from multiphase gas with non - solar abundance ratios . in @xmath5 we show how abundance profiles for ngc 5044 depend on the assumed spectral model . section 6 presents a comparison between xmm - newton and chandra derived abundances .
correlations between the x - ray and h @xmath6 emission are discussed in @xmath7 .
the implications of our results are discussed in @xmath8 and @xmath9 contains a brief summary of our main results .
ngc 5044 was observed by _ chandra _ on march 7 , 2008 ( obsid 9399 ) for a total of 82710 sec .
all data analysis was performed with ciao 4.2 and caldb 4.2.1 .
since the x - ray emission from ngc 5044 completely fills the s3 chip , the back - side illuminated chip s1 was used to screen for background flares .
light curves were made in the 2.5 - 7.0 kev and 9.0 - 12.0 kev energy bands for the diffuse emission on s1 .
no background flares exceeding a count rate threshold of 20% above the quiescent background count rate were detected during the observation .
background images were generated from the standard set of blank sky images in the chandra caldb and the exposure times in the background images were adjusted to yield the same 9.0 - 12.0 kev count rates as in the ngc 5044 data ( see paper i for more details about the data analysis ) .
the gmrt data for ngc 5044 was obtained as part of a low frequency radio survey of elliptical - dominated galaxy groups ( see giacintucci et al .
2010 ; gitti el at .
2010 ) . throughout this paper
we use a luminosity distance of @xmath10 mpc for ngc 5044 , which corresponds to @xmath11 pc .
a temperature map covering the acis - s3 chip ( approximately 90 kpc on a side ) is shown in figure 1 .
the temperature map was derived using the same method described in osullivan et al .
all spectra were fit to an absorbed _ apec _ model with the absorption fixed to the galactic value ( @xmath12 @xmath13 ) and the temperature , abundance and normalization treated as free parameters .
regions were chosen to include at least 1600 net counts .
the middle panel of figure 1 shows the best - fit temperature and the left and right hand panels show the 90% lower and upper limits on the temperature .
the temperature map shown in figure 1 is very similar to the temperature map presented in paper i , which was derived from the energy centroid of the blended fe - l lines .
figure 1 shows that there is a great deal of structure in the temperature map and that the coolest gas extends toward the se .
as discussed in paper i , all of the bright x - ray filaments contain cooler gas compared to the surrounding regions .
a more in depth discussion of the features in the ngc 5044 temperature map is given in paper i. the corresponding abundance map for ngc 5044 is shown in figure 2 .
all 3 panels in figure 2 show the presence of cloud - like and filament - like regions of high abundance gas .
there is also a large extended region of very low abundance gas in the se corner of the s3 chip .
most of the high abundance clouds are located at larger radii than the central x - ray cavities ( see figure 3 ) , but interior to the se cold front noted in paper i ( see figure 4 ) .
the high abundance filaments are primarily located at larger radii than the se cold front . except for the spatial coincidence between the high abundance cloud toward the ne and an x - ray bright filament ,
there are no significant correlations between the high abundance regions , features in the acis image , the unsharp masked image , or the temperature map .
as discussed in paper i , the 235 mhz radio emission passes through the southern x - ray cavity , bends to the west just behind the se cold front and then undergoes another sharp bend toward the south , at the location of the se cold front .
figure 4 shows that the 235mhz radio emission passes through regions with the lowest abundances and threads the region between the high abundance clouds .
the lowest abundance gas is located in the sw corner of s3 , at the same position as the detached 235 mhz radio lobe . to determine the accuracy of the method used for deriving the abundances shown in figure 2 , we extracted spectra from 7 regions with super solar abundances and 5 regions with sub solar abundances ( see figure 5 ) .
these spectra were then fit to a single _ apec _ model in the 0.5 - 3.0 kev energy band with the absorption fixed at the galactic value ( see the results in table 1 ) .
these results are in good agreement with the trends seen in the abundance map and show that the abundances differ by a factor of approximately 3 between the high and low abundance regions .
care must always be taken when estimating abundances for gas which is likely multiphase ( e.g. , buote et al .
1999 ; rasia et al . 2008 ) . in general ,
abundance estimates derived from fitting ccd resolution spectra of multiphase gas are highly model - dependent .
abundances are usually underestimated when fitting the emission from multi - phase gas with temperatures of @xmath141 kev with single temperature models . the abundance map shown in figure 2
is based on fitting the projected spectra to an absorbed _ apec _ model with the absorption fixed at the galactic value .
in addition to the presence of multiphase gas , the derived abundances can also be affected by unresolved lmxbs , non - solar ratios of heavy elements , excess absorption and projection effects . in paper
i , we found that the power - law emission from unresolved lmxbs can affect the derived spectral properties in the central region of ngc 5044 .
we also derived the ratio between the x - ray luminosity of the lmxbs and the k - band luminosity of ngc 5044 in paper i. using this ratio , and the 2mass image of ngc 5044 , we re - fit the spectra extracted from the high and low abundance regions shown in figure 5 with an absorbed _ apec _ plus power - law model with the normalization of the power - law component frozen at the value derived from the k - band luminosity within each region .
we found that the improvement in the quality of the fits is statistically insignificant ( based on a f - test ) and that there is no significant effect on the derived abundances .
this is mainly due to the use of a soft energy band ( 0.5 - 3.0 kev ) in our spectral analysis .
paper i showed that unresolved emission from lmxbs primarily affects the spectroscopic results when higher energy photons are included in the analysis .
we next fit the set of spectra to an absorbed _ apec _ model with free absorption , an absorbed _ vapec _ model with free o , mg , ne , si , s and fe abundances ( freeing up one element at a time ) and an absorbed two temperature _ vapec _ model . the single greatest improvement in the quality of the spectral fits ( based on a f - test ) was obtained by fitting the spectra with an absorbed _ vapec _ model with the absorption fixed at the galactic value and o and fe allowed to vary independently .
the results of this spectral analysis are shown in table 2 .
this table shows that for 6 of the 7 high abundance regions , there is a statistically significant improvement in the quality of the spectral fits . for the low abundance regions ,
the quality of the fits is only improved in 1 out of 5 regions .
while the derived fe abundances in the high abundance regions are still significantly greater than the fe abundances in the low abundance regions ( compare tables 1 and 2 ) , the ratio of the average fe abundance between the high and low abundance regions is significantly reduced by allowing o and fe to vary independently .
figure 6 shows that the high and low abundance regions also have different o / fe ratios .
the high abundance regions tend to have low o / fe ratios , while the low abundance regions have o / fe ratios mostly consistent with the solar ratio ( which is why the low abundance regions are well fit with an _ apec _ model ) .
to demonstrate the dependence of the abundance , z , derived from fitting an _ apec _ model to gas with a non - solar o / fe ratio , we simulated a set of 1 kev spectra using a _ vapec _ model with a solar abundance of fe and a range of o / fe values .
these spectra were then fit with an _ apec _ model .
figure 7 shows that the best - fit abundance derived from an _ apec _ model can be significantly overestimated when fitting the emission from gas with an intrinsically low o / fe ratio .
by excising the o emission in the spectral analysis , and only fitting the data in the 0.8 - 3.0 kev energy band , we find that the abundance derived from an _ apec _ model is consistent with the fe abundance derived from a _ vapec _ model ( see table 3 ) . freeing up additional heavy elements in the _ vapec _ model ( i.e. , mg , ne , si and s ) did not produce a statistically significant improvement in the spectral analysis or alter the best - fit o and fe abundances given in table 2 .
it is possible that the low o / fe ratios derived in the high abundance regions are due to excess absorption which would preferentially absorb the o emission - lines .
we therefore generated a set of simulated _ vapec _ spectra with solar o / fe ratios and a range of excess absorption .
these spectra were then fit to a _ vapec
_ model with the o and fe abundances treated as free parameters and the absorption fixed at the galactic value .
the sensitivity of the derived o / fe ratio as a function of excess absorption is shown in figure 8 .
this figure shows that at least @xmath15 @xmath13 of excess absorption is required to produce the low o / fe ratios observed in the high abundance regions .
table 4 shows the results of fitting the sample spectra to a _
vapec _ model with @xmath16 and the o and fe abundances treated as free parameters .
based on a f - test , there is very little statistical improvement in the quality of the fits compared to that obtained with @xmath16 fixed at the galactic value .
in addition , only 2 of the 12 regions have a best - fit excess @xmath16 of more than @xmath17 @xmath13 at the 90% confidence limit .
there is also no correlation between the o / fe ratio and the best - fit excess @xmath16 among these regions .
the low energy qe of the acis detectors has continued to decrease during the course of the _ chandra _ mission due to out gassed material condensing on the cold acis filters .
our analysis uses the latest update to the acis contamination model which was released in caldb 4.2 on december 15 , 2009 , however , systematic uncertainties in the acis contamination model could affect our derived o abundances .
in addition to the energy dependence of the absorbing material , the acis - s contamination model also includes a time - dependence and a dependence on chipy ( parallel to the read - out direction ) . to estimate the systematic uncertainty in the caldb 4.2 version of the acis contamination model
, we extracted a spectrum from the 2009 observation of the coma cluster ( which has a very low galactic column density of @xmath18 ) spanning the same range in chipy as that covered by the high abundance clouds in the ngc 5044 observation .
the spectrum was then fit to an absorbed _ apec _ model with the absorption treated as a free parameter .
we obtained a best - fit @xmath16 of @xmath19 with a 90% upper limit of @xmath20 .
thus , excess absorption due to systematic uncertainties in the acis contamination model should be less than @xmath21 around the time of the ngc 5044 observation .
such a small systematic uncertainty can not account for the low observed values of the o / fe ratio ( see figure 8) .
table 5 shows the results of fitting the spectra to a two - temperature _ vapec _ model with the o and fe abundances linked between the two temperature components and the absorption fixed at the galactic value . for regions h1 , h7 and l5 , the two temperature fits were essentially degenerate with respect to the single temperature fits ( i.e. , the values for the two temperatures overlap within their uncertainties ) . for these regions ,
we replicate the results from the single temperature analysis in table 5 .
table 5 shows that the improvement in the quality of the spectral fits for 7 of the 9 remaining spectra is highly significant with the addition of a second temperature component .
one obvious difference between the low and high abundance regions is the fraction of the total flux from the cooler component . in the high abundance regions ,
the cooler component accounts for approximately 80% of the total flux , while in the low abundance regions , the flux from the cool and hot components is more evenly distributed . due to the complexity in the abundance map
, we can not perform a simple deprojection analysis .
since we are fitting projected spectra , some of the improvement obtained by fitting a two - temperature model to the spectra probably arises from the range in gas temperatures along the line - of - sight .
using the deprojected temperature and density profiles for ngc 5044 , we calculated the flux arising from gas at different temperatures
. figure 9 shows the cumulative fraction of the 0.5 - 3.0 kev flux , relative to the total flux , arising from gas cooler than a given temperature along three different projected distances ( b=10 , 20 and 30 kpc ) from the center of ngc 5044 .
most of the spectral extraction regions shown in figure 5 are located approximately 20 kpc from the center of ngc 5044 .
figure 9 shows that for a projected distance of b=20 kpc , approximately 20% of the emission should arise from gas hotter than 1 kev due to projection effects .
this is in good agreement with the @xmath22 ratios for the high abundance regions ( see table 5 ) , indicating that the emission from within the high abundance regions themselves is nearly single - phased .
the greater fraction of emission from hotter gas in the low abundance regions indicates that the gas within these regions has a greater predominance of multiphase gas .
the resulting o / fe ratios and fe abundances from a two - temperature spectral analysis are shown in fig . 10 .
fitting the spectra to two - temperature models lessens the fe abundance contrast between the regions initially identified as having high and low abundances based on the spectral analysis with an _ apec _ model . from table 1 ,
the ratio of the average abundance in the high abundance regions to the average abundance in the low abundance regions is 2.5 .
fitting the spectra with a _
vapec _ model with fe and o treated as free parameters gives an average fe abundance ratio between the high and low abundance regions of 1.6 ( see table 2 ) .
the average fe abundance contrast between the high and low abundance regions is further reduced to 1.3 , when the spectra are analyzed with a two - temperature model ( see table 5 ) .
tables 2 and 5 show that the primary reason for the reduction in the fe abundance contrast is a 52% increase in the fe abundance in the low abundance regions compared to only a 26% increase in the fe abundance in the high abundance regions , when a second temperature component is added .
even though the fe abundance contrast between the low and high abundance regions is significantly reduced by fitting the spectra with a two - temperature model , there is still a clear separation between the high and low abundance regions in the o / fe and fe plane ( see figure 10 ) . on average ,
the low abundance regions have higher o / fe ratios than those derived for the high abundance regions . to investigate the dependence of the o / fe ratio on the presence of multiphase gas , we simulated a set of two - temperature spectra and fit the spectra to a single temperature model .
all of the simulated two - temperature models have a solar o / fe ratio , a lower temperature of 0.7 kev , an upper temperature of 1.3 kev and a range in flux ratios between the two temperature components .
the spectra were then fit to a single _ vapec _ model with o and fe allowed to vary independently .
the results are shown in fig
this figure shows that the o / fe ratio can be overestimated by 30 - 50% , due to the presence of multi - phase gas which can explain the segregation seen in the o / fe and fe plane shown in figure 10 , if the gas in the low abundance region has a greater predominance of multiphase gas and the actual o / fe ratios are similar between the different regions .
these results imply that the apparently low abundance regions in the abundance map have a more inhomogeneous gas temperature distribution compared to the apparently high abundance regions , but that the actual fe abundances and o / fe ratios are fairly similar .
to further support the claim that the low abundance regions reside in highly multiphase gas , we fit the spectra to a three temperature model with the temperatures fixed at 0.7 , 1.0 and 1.3 kev and the o and fe abundances linked between the different temperature components .
figure 12 shows the relative distribution of the emission measures for the three temperature components .
this figure clearly shows that the low abundance regions have a broader temperature distribution of emission measures compared to that among the high abundance regions . in summary , there are four reasons to suspect that the _ apec _ derived abundance map shown in figure 4 is more a map of temperature inhomogeneities rather than a true abundance map .
these are : 1 ) the higher values of @xmath22 in the high abundance regions compared to the low abundance regions , 2 ) the narrower temperature distribution in the emission measures within the high abundance regions compared to that in the low abundance regions , 3 ) the greater increase in the derived fe abundance in the low abundance regions compared to the high abundance regions when a second temperature component is added and 4 ) the smaller values of the o / fe ratio in the high abundance regions compared to that in the low abundance regions derived from a single temperature analysis .
the model - dependent nature of derived abundances for emission from multiphase gas with non - solar abundance ratios makes it difficult to determine azimuthally averaged abundance profiles .
such information is essential for determining the masses and mass ratios of different elements and the relative enrichment from type ia and type ii supernovae . to study the sensitivity of derived abundance profiles on the assumed spectral model
, we produced 3 sets of spectra ( with 10000 , 20000 and 40000 net counts in each spectrum ) extracted from concentric annuli centered on ngc 5044 .
figure 13 shows the projected fe abundance profile derived by fitting an absorbed _ vapec _ model with all the elemental abundances linked to fe ( this is equivalent to fitting the spectra to an _ apec _ model ) .
also shown in figure 13 is the projected fe abundance profile if the o and fe abundances are allowed to vary independently .
the presence of non - solar abundance ratios produces a sharp peak in the fe abundance if all of the elements are linked together . allowing the o abundance to vary produces a fairly flat fe abundance profile within the central 100 kpc . to investigate the distribution of other elements
, we also fit the sets of spectra with 20000 and 40000 net counts to an absorbed _ vapec _ model with o , mg , ne , si , s and fe treated as free parameters .
figure 14 shows the projected fe , si and o abundance profiles and figure 15 shows the projected mg and s abundance profiles .
we do not show the ne abundances due to their large uncertainties .
figures 14 and 15 show , that except for o , the other elements have essentially solar ratios .
this explains why the spectral analysis of the low and high abundance regions analyzed in @xmath23 did not improve significantly when elements other than o were allowed to vary independently of fe .
the deprojected fe , o and si abundance profiles derived from fitting single and two - temperature models are shown in figures 16 and 17 .
the spectra were deprojected using the xspec task _ projct_. these figures show that there is little difference between the projected and deprojected fe and si abundance profiles , since they are both essentially flat within the central 100 kpc .
the o / fe ratio is also significantly less than the solar ratio for both the single and two - temperature deprojections .
buote et al . ( 2003 ) derived azimuthally averaged abundance profiles for ngc 5044 from a joint analysis of an earlier 20 ksec _ chandra _ observation and an _ xmm - newton _ observation .
they found significant improvements in the spectral analysis when the data were fit with two - temperature or differential emission measure models compared to single temperature models and discussed the sensitivity of the derived abundances on the presence of multiphase gas .
most of their abundance profiles have a peak abundance between 10 - 30 kpc with lower values both interior and exterior to this region .
the peak abundances in their profiles correspond to the ring of high abundance clouds seen in the abundance map ( see figure 2 ) .
the peak fe abundances in buote et al .
are approximately 0.8 and 1.2 solar in the projected and deprojected analysis , which are in good agreement with the results presented here .
tamura et al . ( 2003 ) presented an analysis of the _ xmm - newton _ reflection grating spectrometer ( rgs ) observation of ngc 5044 and found that the emission within the central @xmath24 ( 22 kpc ) was well fit by a two - temperature model with temperatures of 0.7 and 1.1 kev ( consistent with the range in temperatures along the line - of - sight ) , an fe abundance of 0.55 and an oxygen abundance of 0.25 ( relative to the abundance table in anders & grevesse 1989 ) .
converting these values using the abundance table in grevesse & sauval ( 1998 ) gives fe=0.81 and o / fe=0.39 .
these values for the fe abundance and o / fe ratio fall between the values for the high and low abundance regions in figure 10 .
a narrow - band h@xmath0 image of ngc 5044 was obtained on april 10 , 2010 with the 4.1 m southern observatory for astrophysical research ( soar ) telescope using the soar optical imager ( soi ) .
two ctio narrow - band filters were used : 6649/76 for the h@xmath0+[nii ] lines and 6520/76 for the continuum .
each image was reduced using the standard procedures in the iraf mscred package and the spectroscopic standard is ltt 3864 .
more details on the soi data reduction can be found in sun et al .
( 2007 ) . figure 18 shows the h@xmath0 emission along with gas temperature contours ( red ) and the locations of the southern , northwestern and southwestern caviities ( blue ) .
the total h@xmath0 flux and luminosity is @xmath25 erg @xmath13 s@xmath26 and @xmath27 erg s@xmath26 .
the h@xmath0 emission is primarily extended in a north - south direction , as previously reported by macchetto et al .
figure 18 shows that there are many emission - line filaments extending at least 10 kpc from the nucleus .
most of the filaments extend toward the nw cavity and along the sw filament , which is located between the sw and s cavities .
there is a clump of h@xmath0 emission within the nw cavity and some h@xmath0 emission in the sw cavity , but very little emission within the southern cavity .
both the nw and sw cavities are radio quiet based on our recent gmrt observations
. the brightest and longest of the southern h@xmath0 filaments borders the western edge of the southern cavity .
the temperature contours in figure 18 show that there is a strong correlation between the location of the coolest x - ray emitting gas and the h@xmath0 emission .
temi et al .
( 2007 ) reported evidence for distributed pah emission with the same morphology as the southern extension of the h@xmath28 emission based on a _
observation of ngc 5044 .
as discussed in paper i , the kinematics of the h@xmath28 emitting gas are inconsistent with an external origin for the gas ( caon et al .
2000 ) and are similar to the kinematics of the h@xmath28 filaments in ngc 1275 ( hatch et al .
we also showed in paper i that the most likely origin for the h@xmath28 emitting gas was due cooling of the hot x - ray emitting gas since the stellar mass loss rate in ngc 5044 is more than an order magnitude less than the mass cooling rate .
all of these results are consistent with a scenario in which the gas responsible for the h@xmath28 originates from cooling of the hot x - ray emitting gas within the central few kpcs of ngc 5044 , followed by dredge up behind agn - inflated , buoyantly rising bubbles .
the sw h@xmath0 filament is apparently being dredged up along the outer edge of the 235 mhz radio emitting plasma which threads the southern cavity ( see paper i ) .
the multiphase gas in the center of ngc 5044 is likely produced by agn induced stirring or turbulence .
if all agn activity were to cease , then the subsequent cascading of the turbulence to smaller scales and the eventual dissipation of the turbulent kinetic energy into heat , along with thermal conduction , will eventually produce a more homogeneous distribution of gas temperature .
turbulent kinetic energy is dissipated on approximately the eddy turnover time of the largest eddy which is given by @xmath29 , where @xmath30 is the length scale of the largest eddy and @xmath31 is corresponding turbulent velocity .
the dissipation rate of turbulent kinetic energy per unit volume is given by ( e.g. , david & nulsen 2008 ) : @xmath32 where @xmath33 is a constant of order unity and @xmath34 is the gas density . we can estimate @xmath35 by assuming that the dissipation rate of turbulent kinetic energy locally balances radiative losses , i.e. , @xmath36 , where @xmath37 is the bolometric luminosity of the gas per unit volume . assuming the turbulence can be characterized by a local mixing length prescription , then the size of the largest eddies can be written as @xmath38 , where @xmath0 is a parameter between 0 and 1 and @xmath39 is the radial distance from the cluster center . substituting these expressions into eq .
( 1 ) and solving for @xmath35 gives : @xmath40 adopting @xmath41 ( based on the discussion in dennis & chandran 2005 ) , we computed @xmath35 as a function of radius for ngc 5044 for @xmath42 and @xmath43 ( see figure 19 ) .
this figure shows that without further agn outbursts , the kinetic energy associated with the present level of turbulence will be thermalized within approximately @xmath1 yr within the central 10 kpc .
the required turbulent velocity to locally balance radiative losses within the central 10 kpc is @xmath44 km s@xmath26 for @xmath45 , which gives a turbulent gas pressure less than 1% of the thermal gas pressure within this region .
a cool cloud embedded in hotter gas will be heated by thermal conduction at a rate ( see the discussion in voit et al .
2008 ) : @xmath46 where @xmath47 is the radius of the cool cloud , @xmath48 is the temperature difference between the cool cloud and the ambient medium , @xmath49 is a reduction factor relative to spitzer conduction and @xmath50 erg s@xmath26 k@xmath26 cm@xmath26 .
the time scale for conduction to heat the cool cloud to the ambient gas temperature at constant pressure is : @xmath52 where @xmath53 is the mass of the cool cloud .
this equation can be written as : @xmath54 where @xmath55 is the gas density within the cool cloud .
the conduction time scale as a function of radius is shown in fig .
19 assuming a 20% difference between the gas temperature in the cool cloud and ambient medium , or equivalently , a 20% contrast in density , assuming the clouds are in pressure equilibrium , @xmath56 kpc and @xmath57 or @xmath58 .
this figure shows that @xmath59 is comparable to @xmath35 within the central 10 kpc .
the conduction time scale declines beyond the central 10 kpc due to the increasing gas temperature ( see the temperature profile presented in paper i ) and the @xmath60 factor in the heat conduction rate .
figure 19 shows that both turbulence and heat conduction will produce a more homogeneous distribution of gas temperature within the central 10 kpc in ngc 5044 within approximately @xmath1 yr in the absence of continued agn driven turbulence .
the fact that there is multiphase gas within this region implies that the delay between recent agn outbursts has been less than @xmath1 yr .
while conduction may be efficient at re - heating small cool clouds beyond the central 10 kpc in ngc 5044 ( see figure 19 ) , conduction has a negligible affect on preventing the bulk of the central gas in ngc 5044 from cooling .
figure 20 shows the value of @xmath49 required for heat conduction to locally balance radiative losses . within the central 20 kpc , where the gas is fairly isothermal , heat conduction rates in excess of the spitzer value
are required to suppress cooling . beyond 50 kpc , near the peak in the temperature profile , heat conduction actually helps to cool the gas ( negative values of @xmath49 ) , since the divergence of the heat flux is negative in this region .
thus , there is only a small range in radii , between 25 - 50 kpc , where heat conduction could be important in preventing the gas from cooling .
in paper i , we showed that the central region of ngc 5044 has been strongly perturbed by at least three recent agn outbursts . by combining gmrt and chandra data , we found direct evidence for the uplifting of low entropy gas from the center of ngc 5044 by agn - inflated , radio emitting x - ray cavities .
the nearly continuous production of agn - inflated bubbles over the past few @xmath1 years has probably produced some gas turbulence within the central regions of ngc 5044 .
this agn driven turbulence can have a significant impact on the distribution of heavy elements and the production and longevity of multiphase gas in the center of ngc 5044 . in this paper
, we have shown that fitting a single temperature thermal plasma model with the ratios of the elemental abundances fixed at the solar values ( e.g. , _ apec _ ) can produce significant artifacts in abundance maps . to investigate the origins of the artifacts in the _ apec _ derived abundance map for ngc 5044
, we extracted spectra from several regions with apparently high and low abundances and fit these spectra to a variety of spectral models with different combinations of free parameters .
the single greatest improvement in the spectral analysis was obtained by fitting the spectra with a _
vapec _ model with the o and fe abundances allowed to vary independently .
for the apparently high abundance regions , the best - fit fe abundances derived with the _ vapec _ model are significantly less than the abundances obtained with the _ apec _ model .
these high abundance regions also have best - fit values of the o / fe ratio significantly less that the solar value .
for the apparently low abundance regions , the best - fit fe abundances derived with the _ vapec _ model are essentially consistent with the abundances derived with the _ apec _ model .
the low abundance regions have near solar o / fe ratios . freeing up the abundance of other heavy elements does not further improve the spectral fits in any of the extracted spectra , indicating that the abundance ratios of the other elements are essentially consistent with the solar ratios .
this analysis showed that a substantial portion of the artifacts in the _ apec _ derived abundance map are due to the enforcement of a solar o / fe ratio in the spectral analysis .
this result was confirmed by generating a set of simulated _ vapec _ spectra with a range of o / fe values and fitting the data to an _ apec _ model .
we found that the best - fit abundance obtained with an _ apec _ model increases with a decreasing o / fe ratio .
after allowing for non - solar o / fe values in the spectral analysis , the next greatest single improvement in the quality of the spectral fits was obtained by including a second temperature component .
the addition of a second temperature component produces a slight increase in the fe abundance in the apparently high abundance regions and a much greater increase in the fe abundance in the apparently low abundance regions .
the o / fe ratios are consistent between the single and two temperature spectral analysis , given the uncertainties .
the flux from the second temperature component in the apparently high abundance regions is consistent with that expected from projection affects given the deprojected temperature profile of ngc 5044 . for the spectra extracted from the regions with apparently low abundances ,
the emission from the two different temperature components is more evenly distributed .
in addition to fitting the spectra to a two temperature model , we also fit the spectra to a three temperature model with the temperatures fixed at 0.7 , 1.0 and 1.3 kev .
this analysis showed that the emission from the apparently low abundance regions is more evenly distributed among the different temperature components compared to the emission from the apparently high abundance regions , confirming that inhomogeneities in the temperature distribution are partly responsible for producing the artifacts in the _ apec _ derived abundance map .
we also found that excess absorption and emission from unresolved lmxbs can not produce the artifacts observed in the _ apec _ derived abundance map .
we have shown that deprojected abundance profiles are very sensitive to the assumed spectral model . a single temperature _
apec _ model produces a deprojected abundance profile for ngc 5044 that increases by a factor of approximately 2 from 100 to 10 kpc , and then decreases inward . by fitting the same set of spectra with a _
vapec _ model with the o and fe abundances allowed to vary independently , we obtain deprojected o and fe abundance profiles that are fairly uniform within the central 100 kpc .
central dips in abundance profiles are a fairly common occurrence in cooling flow clusters ( e.g. , sanders et al . 2004 ,
churazov et al . 2004 ) .
our results show that such dips in the central abundance profile could be due to non - solar ratios of the heavy elements or the presence of multiphase gas . excluding oxygen , all of the other elements that produce strong emission lines in 1 kev gas ( i.e. , mg , si and s ) have essentially flat abundance profiles with near solar ratios within the central 100 kpc .
the o / fe ratio is a useful diagnostic for determining the relative enrichment from type ia and type ii supernovae .
however , estimates of the o / fe ratio are very sensitive to the presence of multiphase gas . by generating a set of simulated spectra
, we have shown that the o / fe ratio can be overestimated by factors of 2 to 3 when the emission from multiphase gas is fit with a single temperature model .
while the uncertainties on the o / fe ratios obtained from our spectral analysis are fairly large when more than one temperature component is included in the spectral analysis , on average , we find significantly sub - solar values of o / fe within the central 100 kpc of ngc 5044 .
this implies a greater relative enrichment from type ia enrichment compared to solar abundance gas which is consistent with previous studies of the central regions in groups and clusters ( e.g. , finoguenov et al . 2000;2001 ) . the multiphase gas in ngc 5044 is probably produced by agn induced turbulence .
we have shown that without further agn outbursts , the gas in the center of ngc 5044 should acquire a more homogeneous distribution of temperature within @xmath1 yr due to the dissipation of turbulent kinetic energy and the effects of thermal conduction .
the fact that there is a significant component of multiphase gas in the center of ngc 5044 shows that the delay between successive agn outbursts must be less than @xmath1 yr .
we have recently obtained deeper gmrt data for ngc 5044 to better probe the past agn history in this system and we will present the results in a subsequent paper . we would like to acknowledge seth bruch for obtaining the narrow band h@xmath0+[nii ] images at the soar .
this work was supported by grants go8 - 9122x and go0 - 11003x .
eos acknowledges the support of the european community under the marie curie research traning network .
buote , d. 1999 , mnras , 309 , 685 .
buote , d. , lewis , a. , brighenti , f. & mathews , w. 2003 , apj , 595 , 151 .
caon , n. , macchetto , d. & pastoriza , m. 2000 , apjsuppl , 127 , 39 .
chandran , b. 2005 , apj , 632 , 809 .
churazov , e. , forman , w. , jones , c. , sunyaev , r. & bhringer , h. 2004 , mnras , 347 , 29 .
croton , d. , springel , v. , white , s. , de lucia , g. , frenk , c. , gao , l. , jenkins , a. , kauffmann , g. , navarro , j. , yoshida , n. 2006 , mnras , 365 , 11 .
david , l. & nulsen , p.e.j .
2008 , apj , 689 , 837 .
david , l. , jones , c. , forman , w. , nulsen , p. , vrtilek , j. , osullivan , e. , giacintucci , s. & raychaudhury , s. 2009 , apj , 705 , 624 .
finoguenov , a. , david , l. & ponman , t. 2000 , apj , 188 , 203 .
finoguenov , a. , reiprich , t. & bohringer , h. 2001 , a&a , 368 , 749 .
gebhardt , k. et al .
2000 , apj , 539 , l13 .
giacintucci , s. , osullivan , e. , vrtilek , j. , raychaudhury , s. , david , l. , venturi , t. , athreya , r. , murgia , m. & ishwara - chandra , c. 2010 , apj , ( submitted ) .
gitti , m. , osullivan , e. , giacintucci , s. , david , l. , vrtilek , j. , raychaudhury , s. & nulsen , p. 2010
, apj , ( in press ) .
grevesse , n. & sauval , a. 1998 , sp .
rev.,.85 , 161 . hatch ,
n. , crawford , c. , johnstone , r. , fabian , a. 2006 , mnras , 367 , 433 .
macchetto , f. , pastoriza , m. , caon , n. , sparks , w. , giavalisco , m. , bender , r. , capaccioli , m. 1996 , a&a suppl .
, 120 , 463 .
mcnamara , b. & nulsen , p. 2007
, ara&a , 45 , 117 .
osullivan , e. , vrtilek , j. , kempner , j. , david , l. & houck , j. 2005 , mnras , 357 , 1134 .
peterson , j. & fabian , a. 2006 , phys rep .
, 427 , 1 .
rasia , e. , mazzotta , p. , bourdin , h. , borgani , s. , tornatore , l. , ettori , s. , dolag , k. & moscardini , l. 2008 , apj , 674 , 728 .
sanders , j. , fabian , a. , allen , s. & schmidt , r. 2004 , mnras , 349 , 952 .
sun , m. , donahue , m. & voit , g. m. 2007 , apj , 671 , 190 .
tamura , t. , kaastra , j. , makishima , k. & takahashi , i. 2003 , a&a , 399 , 497 .
temi , p. , brighenti , f. & mathews , w. 2007 , apj , 666 , 222 .
voit , g.m .
, cavagnolo , k. , donahue , m. , rafferty , d. , mcnamara , b. & nulsen , p. 2008 , apj , 681 , l5 .
notes : all spectra were fit in the 0.5 - 3.0 kev energy band to a _
phabs*(vapec+vapec ) _ xspec spectral model with the absorption frozen at the galactic value . the lower best - fit temperature is given by @xmath61 and the upper best - fit temperature is given by @xmath62 .
the ratio between the fluxes in the cooler spectral component to the total flux in the 0.5 - 3.0 kev energy band is given by @xmath63 .
see the notes to table 1 for further details . | a deep _ chandra _ observation of the x - ray bright group , ngc 5044 , shows that the central region of this group has been strongly perturbed by repeated agn outbursts .
these recent agn outbursts have produced many small x - ray cavities , cool filaments and cold fronts .
we find a correlation between the coolest x - ray emitting gas and the morphology of the h@xmath0 filaments .
the h@xmath0 filaments are oriented in the direction of the x - ray cavities , suggesting that the warm gas responsible for the h@xmath0 emission originated near the center of ngc 5044 and was dredged up behind the buoyant , agn - inflated x - ray cavities .
a detailed spectroscopic analysis shows that the central region of ngc 5044 contains spatially varying amounts of multiphase gas .
the regions with the most inhomogeneous gas temperature distribution tend to correlate with the extended 235 mhz and 610 mhz radio emission detected by the gmrt .
this may result from gas entrainment within the radio emitting plasma or mixing of different temperature gas in the regions surrounding the radio emitting plasma by agn induced turbulence .
accounting for the effects of multiphase gas , we find that the abundance of heavy elements is fairly uniform within the central 100 kpc , with abundances of 60 - 80% solar for all elements except oxygen , which has a significantly sub - solar abundance . in the absence of continued agn outbursts ,
the gas in the center of ngc 5044 should attain a more homogeneous distribution of gas temperature through the dissipation of turbulent kinetic energy and heat conduction in approximately @xmath1 yr .
the presence of multiphase gas in ngc 5044 indicates that the time between recent agn outbursts has been less than @xmath2 yr . |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.