article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
a promising approach to study backscattering in the helical edge states on a local scale is to generate small , local potential fluctuations and test whether they perturb transport in the edge channels . to achieve this goal
, one must induce a local potential perturbation with precise control over its position and strength .
this allows either modulation of the strength of a pre - existing scattering site or the generation of an artificial scattering site . in scanning gate microscopy ( sgm )
@xcite , a charged tip is scanned above the device of interest and the effect of the induced potential fluctuation on the conductance in the device is monitored . over the past few years , scanning gate microscopy has been used to investigate transport phenomena in a variety of low - dimensional systems like quantum dots @xcite , quantum point contacts @xcite and two - dimensional electron gases @xcite .
of particular relevance to qsh edge state transport , sgm has been applied to one - dimensional systems like carbon nanotubes @xcite and nanowires @xcite where it was used to identify and manipulate localized states controlling the transport through the device .
m ) allow for precise alignment of the tip to the device .
the mesa extends upward from the dotted yellow line .
the grainy areas at the left and right edge , respectively , are the ohmic contacts to the quantum well .
the orange markers demarcate the scan window used for the conductance maps in figs . 2 and 3 . *
( b ) * a back - gate electrode is used to tune the device into the qsh regime with @xmath1 . * ( c ) * illustration of the experimental configuration for sgm studies of the qsh state . ] our devices were fabricated from undoped hgte / hg@xmath2cd@xmath3te quantum well structures with a nominal hgte layer thickness @xmath4 nm . the devices were patterned using optical lithography , ion - milling to define mesas , and evaporation of in / au ohmic contacts to the quantum well .
1a shows a section of a typical device in a topographic image taken by atomic force microscopy ( afm ) .
our home - built sgm probes feature self - sensing piezoresistive deflection readout @xcite which allows for precise in - situ alignment of the tip to the device .
the separation between the contacts in the transport direction is @xmath5 m , and the lateral mesa width is @xmath6 m . as
this width is several orders of magnitude greater than the predicted extension of the qsh edge states into the bulk @xcite , any effect of inter - edge tunneling @xcite can be ruled out .
more relevant to our sgm experiments , the large width of the device ensures that the tip - induced potential perturbation only affects the transport along the edge located within the scan window , while the far edge remains unaffected and thus provides a constant contribution to the device conductance .
we perform our experiments in a two - terminal configuration using standard lock - in technique ( @xmath7 hz , @xmath8v ) at @xmath9 k. despite the absence of intentional doping in the heterostructure , the studied quantum well structures consistently show a finite density of @xmath10-type bulk carriers , typically on the order of a few times @xmath11 . for the device discussed extensively in this paper , we found a density @xmath12 , prior to gating @xcite .
1b shows the device conductance as a function of the voltage applied to the back gate electrode which is used to tune the fermi level in the device .
the observed saturation of the conductance slightly below @xmath13 for @xmath14 v suggests that the sample is in the qsh state and transport happens only along the edges .
the basic concept of sgm measurements on a sample in the qsh state is illustrated in fig .
1c : a dc voltage @xmath15 is applied to the tip located slightly above the sample surface ( in our experiments , height @xmath16 nm ) .
the electric field from the tip produces a local modulation of the potential in the sample , below the tip . in our sgm experiments
, we induce such a potential perturbation near the device edge with the goal of introducing and controlling backscattering of the helical edge channels .
we studied several devices with similar properties , formed from different heterostructures , and all devices showed comparable results except as noted . in this paper , we focus our discussion on one particular device , which shows the same qualitative features as other devices as well as striking and enlightening behavior seen only on this device . unfortunately , a tip crash damaged the device before we could collect a full data set for the sample in the qsh regime .
however , earlier measurements on the same device in the presence of bulk carriers complement the data and allow for a detailed analysis . while the limited amount of data does not allow for a fully conclusive study of scattering mechanisms affcting the qsh edge channels , our results nonetheless provide a substantial step forward over the available transport experiments .
with @xmath17 v for @xmath18 .
the dotted lines indicate the edge of the device .
the labeled crosses mark the positions of the scattering sites . *
( e ) - ( h ) , * gradient @xmath19 of panels * ( a ) - ( d)*. ] in the qsh regime at @xmath20 v , we observe a strong modulation of the conductance in the sgm maps on this device ( fig .
2a - d ) : as a function of the tip position , the conductance varies approximately from @xmath21 ( similar to the conductance value when the unbiased tip is far away ) down to @xmath22 .
the rather smooth variation of @xmath23 over a length scale of several microns in tip position perpendicular to the edge of the device , most clearly visible in fig .
2a , is mostly caused not by a local gating by the tip but rather by long - range gating by the large conductive area of the cantilever biased to the same voltage as the tip .
in contrast , the superimposed modulations on length scales below 1 micron , highlighted by taking the gradient of conductance with tip position in fig .
2e - h , are associated with local gating by the sgm tip , showing the effect of a well - localized potential perturbation on the edge state conductance .
the suppression of conductance by @xmath24 by local gating suggests that transport at one of the two edges is fully suppressed by the tip potential .
this result supports our assumption that the sample is in the qsh state , with transport occurring primarily along the edges of the device . for a strongly negative tip voltage @xmath25 v ( fig .
2a ) , regions with @xmath26 can be observed in the conductance map .
this can not be explained solely by transport in the qsh edge states , and we attribute the excess conductance to the tip - induced emergence of bulk conductance ( see supplemental material @xcite ) . as noted above , some prior transport measurements @xcite may be understood by invoking a single fully edge - equilibrating metallic region along a single edge .
this would decrease the conductance of that edge by a factor of 2 , from @xmath22 to @xmath27 .
based on a simple cartoon picture for our sgm experiments , we expected the tip potential to induce such a local metallic region below the tip , so we were surprised to see a stronger suppression than @xmath27 associated with a single site .
several theoretical models have been proposed that could in principle lead to full suppression of edge conduction @xcite . however , as neither of those theoretical predictions seems sufficient to fully explain the observed behavior @xcite and we found only a single occurrence of full suppression of edge state transport in all of the measured devices , we focus now on the more commonly seen features in the sgm scans .
besides the pronounced conductance suppression attributed to a single site , the conductance maps show multiple sets of concentric rings representing a reduced conductance . around site 1 ,
the rings are superimposed on the strong conductance modulation ( the full suppression ) .
two more sets of concentric rings are centered around sites 2 and 3 , respectively , in fig . 2 .
while these features are partly masked in the conductance maps by the dominant conductance modulation originating from site 1 , they are clearly visible when the gradient of the conductance is plotted ( fig .
for all three scattering sites , the conductance modulation associated with the sharp rings of conductance suppression is on order @xmath28 .
circular patterns with a similar conductance modulation were found in all devices we studied . in general ,
circular patterns in sgm maps are a signature that transport is a function of the potential at one sensitive site , the common center of the circles .
this can be understood intuitively : when the tip moves along one of the circles , the tip - induced potential at the circle s center does not change and
the conductance remains unaffected .
however , when the tip moves toward or away from the sensitive site , the potential at that site and consequently the conductance change .
the canonical example of this behavior is a quantum dot , where the local potential at the dot determines the occupancy , and the conductance peaks in coulomb - blockaded transport show up as rings in sgm experiments @xcite .
the observation of ring - like patterns in our data suggests that the suppression of conductance in the qsh edge channel is linked to individual sites located at or near the physical edge of the device , though in place of coulomb blockade we below suggest another mechanism dependent on the local potential . in our experiments ,
the rings of conductance suppression appear elongated due to spatial variation in the dielectric environment : when the tip is located above the mesa , the tip potential experienced at the scattering site is screened by the hg@xmath2cd@xmath3te cap layer in the heterostructure with @xmath29 whereas for a tip position not above the mesa the relevant dielectric constant should be between @xmath30 and @xmath31 .
the magnitude and other experimental signatures of conductance modulation resulting in multiple rings about the three indicated scattering sites may be understood if these sites are associated with small bulk conducting regions adjoining the edge and serving as sources of decoherence @xcite . for simplicity
, we will discuss transport only through the helical edge state we scan over and omit the contribution of @xmath32 by the far edge . a macroscopic metallic contact
well - coupled to the edge is expected to fully equilibrate the counterpropagating edge states as noted above . as there is no relationship between the spin orientation of an electron impinging on this contact and one emerging back into the edge
, the electron should emerge with equal probability into the forward - moving ( say , spin up ) and backward - moving ( say , spin down ) channel , respectively .
consequently , conductance along the edge should be reduced from @xmath33 to @xmath34 .
a metallic region of intermediate size could also reduce the conductance if the edge states can lose their spin information during interaction with the charge carriers in this metallic puddle . in ref .
, electron - electron interaction or electron - phonon interaction were given as possible sources for this decoherence " of the helical edge state , though the role of these particular types of interaction has not been discussed quantitatively in the model .
the amount of conductance decrease depends on the exact size of the dephasing region and the strength of the dephasing process . for a finite - sized metallic region , resonances in dephasing should enhance the suppression of the edge state conductance for particular sizes of the dephasing region . the conductance suppression caused by this resonant process
is predicted to be around @xmath35 @xcite , comparable to the size of the modulation observed in our sgm maps . in the experimental data accompanying the model put forward in ref .
, however , similar resonances were not observed , pointing towards the presence of a metallic puddle resulting in strong decoherence of the qsh edge state , e.g. , due to a very large size .
our experimental results indicate that backscattering of the qsh edge state is caused by well - localized scattering sites along the edge of the device .
the observed pattern in the conductance modulation is consistent with the theoretical model described above based on partial dephasing of the edge state in a small metallic region at each of these sites .
subsequently , we will use this picture to interpret the backscattering observed in our measurements .
when either the tip position or the tip voltage is changed , the potential at the puddle will vary , modulating the puddle size and possibly the strength of the dephasing mechanism .
this should tune the dephasing process in the metallic region in and out of resonance , resulting in an oscillatory conductance modulation .
electrostatic simulations show that the size of a metallic region - either induced by the tip potential itself or already existing in the sample - with a charge density exceeding a certain threshold value can be tuned by a few hundred nm within the tip voltage range used in our experiments @xcite . at the same time , the maximum density in the puddle , which should affect the strength of the dephasing mechanism , changes with the tip voltage as well .
this variation in the size and strength of the dephasing region is expected to result in multiple resonances @xcite as seen in our experimental data .
our explanation for the observed backscattering is corroborated by additional features which appear at strongly negative tip voltages @xmath36 v. the maps of @xmath37 show multiple lines running approximately parallel to the mesa edge along the entire length of the device ( indicated by arrows in fig .
this pattern of the gradient corresponds to an oscillatory conductance modulation as a function of the tip position perpendicular to the mesa edge .
we can rule out simple instrumental artifacts as follows : the sgm data are recorded during line scans in the direction of the observed conductance modulation , so a slow drift in the sample conductance can be ruled out as an explanation .
instrumental oscillations on a timescale comparable to the measurement time between individual lines can not account for the observed pattern either as the lines only appear near the edge , evolve consistently with the applied tip voltage @xcite , and only occur for particularly large tip voltages .
though the amplitude of this conductance modulation is less than @xmath38 - weaker than the effect caused by the localized scattering sites - it can be explained by the same mechanism . for a sufficiently strong potential perturbation induced by the sgm tip , a metallic puddle ( here @xmath10-type because of the negative tip voltages )
is induced at an otherwise unperturbed section of the edge . for a dephasing region of appropriate size
, resonant suppression of conductance should be seen .
these lines are only visible for much stronger tip voltages @xmath36 v compared to the ring features described above .
this difference in tip voltage required to cause backscattering can provide a measure for the strength of the intrinsic potential fluctuations and the robustness of the otherwise unperturbed edge states , as discussed in more detail below .
while the sgm maps for the device in the qsh regime provide clear evidence for the presence of scattering sites , experimental data for this transport regime are only available for a limited tip voltage range , so we move next to conditions of mixed bulk and edge conduction .
at @xmath39 , the sample displays a conductance @xmath40 .
the conductance value larger than @xmath41 indicates that the device must have some bulk conductance . however , we expect qsh edge states to coexist with bulk carriers and contribute considerably to the total conductance . in this regime ,
the strongest local gating effect close to the edge reduces the conductance by @xmath42 ( fig .
as this conductance modulation can be induced within a few hundred nm of the edge , it is likely associated with an edge state and constitutes a lower limit for the conductance of a @xmath5 m stretch of a single edge state under these slightly @xmath10-type conditions .
this would be consistent with a reduced edge conductance in comparison to the qsh regime , possibly because coexisting bulk carriers allow backscattering of the edge states in the regions between the local scattering sites identified earlier .
the reduced edge state conductance compared to the qsh regime at @xmath17 v further confirms that the conductance along the edge is indeed due to qsh edge states and not caused by trivial edge currents as they may occur due to an inhomogeneous potential landscape . at @xmath43
v , the dominant conductance modulation @xmath44 occurs when the tip is located directly above the scattering site - within 125 nm ( half width at half maximum ) , giving an estimate for the size of the dephasing region ( fig .
the identification of this feature as suppression of edge state conductance is supported by the observation of an associated ring pattern centered at the edge of the mesa ( fig .
3a - c ) , similar to those observed in the qsh regime .
we can see two more sets of rings , and all three sets are centered around the same three locations as in the qsh regime and presumably to the same scattering sites .
thus , we can use these data for a more complete analysis of the properties of the dephasing regions .
while the suppression caused by site 1 is unusually strong and the underlying mechanism is not understood yet , the superimposed resonances are qualitatively comparable to the features centered at scattering sites 2 and 3 . as the conductance modulation associated with site 1
is most clearly visible in the sgm maps , we use scattering site 1 for further analysis .
v ( black upper trace ) and @xmath43 v ( red lower trace ) show strongest conductance modulation @xmath45 and most localized gating effect @xmath46 m ( hwhm ) , respectively . *
( b ) * evolution of line scans as a function of @xmath15 . ] fig .
4b shows the conductance measured along the line marked in fig .
3a as a function of @xmath15 .
we observe resonant suppression of conductance corresponding to rings in the sgm maps mainly for negative tip voltages .
as the tip voltage becomes more negative , the spatial separation between the conductance dips increases .
this shift can be understood based on the electrostatics of the sgm configuration . when the tip voltage is changed , the potential at a particular sensitive site within the sample can be kept constant if the distance between tip and site is adjusted accordingly , as traced out by the bright and dark curves in fig .
more precisely , the tip voltage needed to trigger dephasing is more negative the farther away from the site the tip is , showing that the potential at the site needs to be pushed up for dephasing . while the strong local gating effect with @xmath47 can only be observed for @xmath48 v , weaker resonances with @xmath49
can still be observed for smaller negative tip voltages . in the slightly @xmath10-conducting regime at @xmath39 , backscattering at the three identified scattering sites occurs for tip voltages as small as @xmath50 v ( sites 1 and 2 ) or @xmath51 v ( site 3 ) , whereas detectable backscattering can not be induced in the adjacent regions for tip voltages as strong as -10 v. this shows that the potential at the sensitive sites need to be pushed up less to trigger dephasing than would be needed in surrounding regions . while there might be quantitative differences between backscattering in the qsh regime ( @xmath17 v ) and in the presence of weak bulk transport ( @xmath39 ) , respectively , we will use @xmath52 v , which was necessary in the qsh regime to introduce backscattering in otherwise unperturbed regions of the edge , as an estimate for the strength of a potential perturbation required to cause backscattering .
this corresponds to a locally induced density @xmath53 @xcite .
let us assume that densities associated with backscattering at the identified sites are also around @xmath54 .
but the low tip voltages necessary to turn on backscattering at those sites correspond to tip - induced density changes only on the order of several @xmath11 .
this implies that the metallic puddles at the scattering sites are already present in the absence of the tip for @xmath39 .
in addition , tuning the sample into the qsh regime by applying @xmath17 v changes the bulk density by @xmath55 , i.e. , significantly less than the inferred carrier density in the dephasing regions , indicating that the resonant backscattering observed in the qsh regime at @xmath17 v can be attributed to @xmath10-type puddles interrupting the edge state .
all observed scattering sites can be identified as @xmath10-type whereas experimental signatures of pre - existing @xmath56-type puddles are absent .
this may be related to the band structure of the hgte structures : hole densities of a few @xmath57 can already occur in the quantum well if the fermi level is just 1 mev below the valence band maximum , much less than the shift into the conduction band required to induce comparable @xmath56-type densities @xcite .
this suggests that potential fluctuations of a few mev might be responsible for the observed metallic scattering sites . as our measurements imply that densities on the order of @xmath57 are required to cause substantial backscattering , a conductance @xmath58 carried by a helical edge state coexisting with a bulk carrier density @xmath59 at @xmath39 , as determined earlier , appears reasonable . for positive tip voltages ,
the pattern of conductance modulation in the sgm maps gets significantly more complex ( fig .
3d - f ) so that attribution of individual features to particular scattering sites is rarely possible .
the clearest features at the mesa edge are centered around site 3 , where circular patterns are visible for both positive and negative tip voltages , suggesting that the site is not simply the location of a pre - existing potential fluctuation .
a variation of the well width by just a single monolayer can change the bulk 2d energy gap by several mev , significant compared to the 13 mev gap predicted for our 8 nm quantum well @xcite .
this could naturally lead to a small region where local gating could enable both @xmath10- and @xmath56-type puddles at the same location . at the same time , a variation of the bulk gap could also result in a relative shift of the bands , locally shifting the fermi level from the gap into one of the bulk bands .
when we position the tip over the slightly @xmath10-conducting bulk at @xmath39 , we see an overall trend from high conductance @xmath23 for negative tip voltages to low conductance for positive tip voltages ( fig .
3**a - f * * ) .
the increase in @xmath23 for negative tip voltages ( fig .
3**a - c * * ) can be understood as due to a tip - induced accumulation of @xmath10-type bulk carriers . for positive tip voltages ( fig .
3**d - f * * ) , the conductance is suppressed by approximately @xmath42 compared to the conductance in the absence of a tip effect . though the sign of this effect makes sense , its magnitude is several times stronger than one would expect for full suppression of bulk transport through only a small fraction of the device - probably around a few percent of the device width - in the case of a homogeneously conducting bulk with @xmath60 . but this assumption of homogeneous conduction is not justified .
if we assume a conductance @xmath61 for each edge state in the slightly @xmath10-conducting device at @xmath39 as determined above , we obtain a bulk resistivity exceeding a few hundred k@xmath62 , a range characteristic of percolative or hopping bulk transport . in this transport regime ,
signatures of localized states are visible in the sgm maps not just along the edge of the device , but also in the bulk several microns away from the edge ( fig .
3 g - l ) .
in contrast to the superficially - similar sets of rings centered at the edge , however , the localized states in the bulk manifest themselves in a variety of conductance patterns . for negative tip voltages ( fig . 3 g - i ) , for example , a set of concentric sharp rings with _ enhanced _
conductance is visible in the upper right corner of the scan window , marking resonantly enhanced transport , as is typical for a quantum dot @xcite . in the scans with @xmath63
v , a single elongated region of enhanced conductance emerges at the upper edge of the scan window and grows with increasing tip voltage ( fig .
3j - l ) .
such a response to a locally induced potential is reminiscent of the behavior seen in sgm studies of quantum point contacts @xcite .
the enhancement of conductance with increasingly positive tip voltages implies that the tip induces @xmath56-type carriers .
this is very plausible , as @xmath10-type densities extracted above for negative tip voltages are significantly larger than the @xmath10-type background density in the absence of the tip , and the small band gap of hgte quantum wells allows for a tip - induced shift of the fermi level from the valence band into the conduction band . for large positive tip voltages , a set of sharp arcs of suppressed conductance ( fig .
3l ) could point to the presence of another localized state strongly affecting transport in our device .
the center of these arcs appears to be in the bulk outside the scan window , but probably still inside the channel in which we measure conductance .
all these observations are consistent with a two - dimensional system with strong potential disorder , in which quantum dots and quantum point contacts can form accidentally as the fermi level fluctuates between bands and gap .
the key results of our experiments are evidence of edge state transport in the qsh regime , identification of pre - existing scattering sites and demonstration that backscattering of the qsh states caused by these sites can be enhanced resonantly if the local perturbation is tuned appropriately .
the separation between these well - localized scattering sites is typically between 1.5 and 2 microns , and the edge states appear to propagate unperturbed between them .
this spacing qualitatively explains the size limit of 1 to a few microns for ballistic transport in qsh devices determined in earlier transport experiments @xcite , while also suggesting that if scattering sites could be individually tuned off - resonance the ballistic length could be extended substantially .
the detection of localized states in the bulk supports our interpretation that the scattering sites along the edge can be attributed to local disorder which is inherent to the quantum well structure and not a result of damage imposed on the edge of the device during the fabrication process .
in addition , we have demonstrated that backscattering can be introduced in otherwise unperturbed qsh states by a sufficiently strong tip - induced perturbation .
this gives insight into the robustness ( or vulnerability ) of the qsh states against potential fluctuations .
however , as the conductance modulation solely induced by the tip potential is weaker than the effect of the pre - existing scattering sites identified in our current experiments , it would be worthwhile to study these effects in more detail in subsequent experiments .
similar sgm experiments in an external magnetic field could elucidate the role of time - reversal symmetry for the predicted suppression of backscattering .
+ we thank m.r .
calvo , j. maciejko , k. moler , k. nowack , and t. schmidt for valuable discussions .
this work was funded by the department of energy , office of basic energy sciences , division of materials sciences and engineering , under contract de - ac02 - 76sf00515 ( sgm and transport investigation of the qsh state in hgte devices ) , by the center for probing the nanoscale , an nsf nsec , supported under grant no .
phy-0830228 ( development of the sgm technique ) , and by the darpa meso project under grant no .
n66001 - 11 - 1 - 4105 ( mbe growth of the hgte heterostructures ) .
m.k . acknowledges financial support from the daad postdoctoral fellowship ( deutscher akademischer austausch - dienst , german academic exchange service ) .
the two - terminal configuration used in the sgm experiments is not well - suited for determining the sample properties like carrier density and mobility . for that purpose , we fabricated a hall bar structure with @xmath64 from the same material .
like in the sgm device , we used a metallic layer at the bottom of the cdte substrates ( thickness @xmath65 m ) in our samples as a back gate electrode . when a voltage is applied to this back gate electrode to tune the fermi level in the device
, the hall bar enters the qsh regime around @xmath66 and @xmath56-type bulk conduction occurs for @xmath67 v ( fig .
the conductance value of @xmath68 for the qsh state is comparable to values reported for other devices of similar size @xcite . in the vicinity of the conductance minimum , several sharp peaks can be observed .
these might be caused by resonances in the scattering mechanism as they have been observed in our sgm data , now with the back gate electrode tuning the properties of multiple dephasing regions simultaneously .
the voltage range for which the hall bar is tuned into the qsh regime is shifted by @xmath69 v in comparison to the sgm device .
based on a parallel plate capacitor model , this corresponds to a difference in density of approximately @xmath70 toward lower p - type densities in the hall bar .
such shifts in density have been observed in several devices . they can be explained by fluctuations of the carrier density on length scales much larger than the typical device size and , to a lesser extent , by small variations in the sample properties between different cooldowns .
v , the device is clearly p - type bulk conducting and carrier density and mobility can be determined from magnetotransport measurements . ]
magnetotransport measurements were performed to determine the density and mobility in the quantum well ( fig .
s1b ) . in close proximity to the qsh state ,
the edge states contribute significantly to the transport , making extraction of bulk properties from magnetotransport data difficult . for @xmath71
v , the sample has a @xmath10-type carrier density and mobility of @xmath72 and @xmath73 , respectively .
the clear reduction of carrier mobility in comparison to n - type quantum wells with comparable thickness @xcite can be explained by the large effective mass in the valence band which is about one order of magnitude higher than the effective electron mass in the conduction band .
based on these results , we estimate the carrier density in the sgm device to be @xmath74 at @xmath39 . this estimate does not take into account uncertainties based on the large disorder in the sample , or on contributions of the edge state to the hall resistance even in the presence of substantial bulk conduction at @xmath71 v. the device geometry itself
, in particular the short length of the device between the metallic contacts to the device , may also result in a gate response different than expected from the assumed parallel plate model .
however , as the discussion in the main paper does not rely on the exact carrier density in the slightly @xmath10-type device , the given estimate is sufficient for our purposes .
the typical analysis of subband energies in hgte structures as a function of the quantum well thickness @xcite considers the energies of the subbands at @xmath75 . for narrow hgte quantum well layers , the band structure shows a direct gap at @xmath75 so that the given energies for the respective subbands represent the bulk gap correctly . however , already for @xmath76 nm the band structure shows an indirect gap with a valence band maximum at @xmath77 @xcite .
thus , the separation of valence and conduction band at @xmath75 overestimates the bulk gap for wider quantum wells .
we performed band structure calculations for hgte quantum wells within a @xmath78 model @xcite to determine the exact gap size in hgte qw structures with a layer sequence identical to our samples ( fig .
s2a ) . for qw layers with a thickness up to @xmath79 nm
, the band structure has a direct gap at @xmath75 . for wider qw layers
, the inverted gap becomes indirect and reaches its maximum size of @xmath80 mev at @xmath76 nm ( fig .
s2b ) . for even wider quantum wells ,
the gap size decreases continuously until a transition to a semimetallic band structure occurs around @xmath81 nm .
the semimetallic behavior was recently observed experimentally in 20 nm wide quantum wells @xcite
. band structure calculations . *
( b ) * dispersion for a hgte structure with @xmath4 nm as it was used in our sgm experiments .
the dashed horizontal lines indicate the fermi energy for @xmath82 ( green ) and @xmath54 ( blue ) , respectively .
inset : valence band maximum and fermi level for @xmath54 . ] the flat dispersion of the valence band near its maximum at finite @xmath83 allows for large carrier densities already for a fermi level only a few mev below the valence band maximum whereas much higher fermi energies , now with respect to the minimum of the conduction band , are required to obtain comparable @xmath56-type densities ( fig .
in the main text , we showed that the conductance of the edge state can be fully suppressed by local gating ( see fig . 2 ) .
however , as we have seen such a strong effect of the tip potential on the conductance only in one case , our data is not sufficient to extract details regarding the underlying mechanism . in the following
, we will discuss several proposed models that in principle can result in a full suppression of transport in a qsh edge state and assess their applicability in our samples .
the first possible scenario relies on coupling between the qsh edge and a nearby magnetic impurity @xcite .
while conventional magnetic impurities are not expected to be present in our samples , an accidentally - formed quantum dot could play the same role .
if the dot is occupied by an odd number of electrons , one electron spin remains unpaired and serves as a magnetic impurity . in ordinary 1d luttinger liquids " , coupling to a local site should fully suppress conductance as @xmath84 , even for weak electron - electron interaction .
in contrast , in a qsh holographic liquid " at the 1d boundary of a 2d system , for weak electron - electron interaction a magnetic impurity should be screened by the formation of a kondo singlet , restoring conductance of the qsh edge state at low temperatures . for strong electron - electron interactions ,
two - particle backscattering is predicted to result in the formation of a
luttinger liquid insulator " characterized by full suppression of edge state transport for @xmath84 .
the strength of electron - electron interaction can be defined by the luttinger parameter @xmath85 for the helical liquid @xcite with @xmath86 where lower @xmath85 connotes stronger interaction .
while it is difficult to determine @xmath85 for a given system exactly , it can be estimated when the relevant interactions are taken into account @xcite . in our specific case
, the strength of coulomb interaction in the hgte devices depends on the device geometry : @xmath87 for the devices with a top - gate electrode as used in earlier transport experiments @xcite , whereas the absence of such an electrode in our sgm devices leads to a reduced screening of electron - electron interaction and the luttinger parameter decreases to @xmath88 .
this value is close to the regime of @xmath89 required for the luttinger liquid insulator " and a small error - either inherent to the approximation or due to a slight deviation of the sample parameters used in calculating @xmath85 - might result in an overestimation of @xmath85 in our devices .
thus , interactions in our devices might be stronger than anticipated . however , it is not clear whether our experimental temperature should be low enough to see nearly full suppression under this scenario .
rashba spin - orbit interaction has also been proposed to lead to localization of qsh edge states if it is spatially nonuniform along the edge @xcite .
hgte quantum well structures show very strong spin - orbit interaction , and a rashba splitting in the conduction band of up to 30 mev @xcite .
in addition , the rashba interaction can be tuned externally by an electric field , making the proposed model a seemingly good candidate for an explanation of the observed behavior . however , the proposed rashba - induced localization length shows a strong dependence on the electron - electron interaction strength @xcite . for @xmath88 , the localization length would exceed the size of our device , so the full suppression we observe can not be explained by spatially inhomogeneous rashba interaction in this case . to induce localization on a scale of order 100 nm , as we infer it to happen in our sample ,
@xmath90 is required which is significantly smaller than the estimated value @xmath88 based on our device properties .
although we apply no magnetic field in our measurements and use no magnetic materials in the fabrication of the devices , for completeness we note that backscattering becomes possible when time - reversal symmetry is broken .
several theoretical proposals have considered the effect of a magnetic field , applied either locally or globally , with a resulting gap in the linear dirac spectrum of the qsh edge states .
two similar models @xcite considered magnetic field applied locally along the edge to form a quantum dot in the qsh edge channel , suppressing conductance when the dot is tuned into coulomb blockade .
more recently , a local potential perturbation was predicted to suppress transport in the qsh state in the presence of an external magnetic field @xcite .
however , with no magnetic field in our experiments , these mechanisms can not explain the observed full suppression either .
[ hbt ] while existing theoretical models do not appear to account for the full suppression of edge state transport , the observed behavior might be related to the presence of strong defects in our samples . as can be seen in fig .
s3 , multiple defects of a characteristic shape can be found within an area of tens of @xmath91 on the mesa surface .
these defects typically have a circular shape with a diameter of approximately 1 micron and a height of a few nanometers .
several of these defects feature a central ridge , occasionally elongated along one of the @xmath92 directions , with an additional height of approximately 10 nm .
one of these particularly tall defects can be found at the center of the region of strong suppression shown in fig . 2 in the main paper .
we believe that these defects originate at the surface of the substrate and subsequently propagate through the mbe - grown layers of the heterostructure
. the resulting lattice distortion can affect the layer structure in multiple ways .
if the quantum well thickness locally is reduced below the critical value @xmath93 nm , the affected region would constitute a topologically - trivial insulator .
a trivially insulating region could also occur if the quantum well is completely interrupted by the defect ( which is taller than @xmath4 nm ) or if the well and barrier material , respectively , get intermixed @xcite due to imperfect growth conditions in the vicinity of the defect . in either case , the qsh edge state should propagate along the boundary between topologically trivial and non - trivial insulator .
the creation of trivially - insulating regions may also cause a strong deformation of the edge states which could form loops in the edge state @xcite or antidots of helical edge states coupled to the qsh state at the edge of the mesa .
such perturbations of the helical edge state likely will cause some degree of backscattering , but it is expected to be a weak effect , not full suppression @xcite . on the other hand , a strong local increase in the qw thickness could result in a semimetallic band structure ( see above and ref . ) , resulting in a bulk - conducting region coupling to the qsh state .
finally , the defect could also cause a local shift of the chemical potential relative to the band edge , possibly into the bulk bands . in either case , the resulting metallic puddle could couple to the edge state and cause a suppression by up to @xmath27 , but again no full suppression @xcite . in summary ,
it is not clear that the models already discussed theoretically suffice to explain our observed full suppression ( and some can even be ruled out completely as explanations in our case ) . nor do the obvious consequences of a strong defect appear to give rise to a full suppression .
thus , it will require further effort to understand the observed full suppression caused by local gating near the edge .
we used comsol@xcite to calculate the electrostatic potential profile of the tip and the resulting induced modulation of the carrier density .
we simulated the effect of a conical tip over a quantum well structure with a lateral extension exceeding the expected long - range effect of the tip .
an axially symmetric layout was chosen to reduce the computational complexity
. this simplified configuration will not provide quantitatively exact results as it does not take into account details of our device geometry such as the edge of the heterostructure , giving rise to a spatially varying dielectric environment , and the nearby ohmic contacts which might shield the tip potential .
nonetheless , the simulations can give a rather accurate representation of the tip - induced density profile .
s4a shows the density induced in the quantum well layer for @xmath94 v. at the center , i.e. , directly beneath the tip , a hole density of @xmath95 is induced .
the density profile decays rapidly with a full width at half maximum of 520 nm .
v. * ( b ) * size of induced metallic regions with densities exceeding certain threshold densities ( labels given in @xmath57 ) . ]
if we assume that the carrier density in a finite - sized region translates directly into the strength of the dephasing mechanism , we can estimate how the size of a region with a given dephasing strength evolves as a function of the tip voltage . in fig .
s4b , we plot the evolution of quasi - two - dimensional regions above a given threshold density as a function of the applied tip voltage . as discussed in the main paper , a density on the order of @xmath57
is required to induce a measurable amount of backscattering . for the corresponding tip voltage of @xmath96 v ,
the diameter of the induced dephasing region is a few 100 nm which is comparable to our size estimate for the pre - existing scattering sites and the size of the metallic region studied theoretically in ref . .
thus , the simulations of the electrostatic potential profile further substantiate our interpretation of the sgm results . for simplicity
, we neglect the likely effect that higher density at the center of the puddle will result in an increase of the dephasing strength .
we already briefly discussed in the main paper and showed in fig .
2a that conductance values exceeding @xmath13 - the expected upper limit for the conductance in the qsh regime - can be observed when large negative tip voltages @xmath36 v are used to manipulate the edge states .
we attribute this excess conductance to an emergence of bulk conductance caused by a long - range effect of the tip potential .
simulations show that densities on the order of several @xmath97 can be induced even microns away from the actual tip position for such large tip voltages ( fig .
our sgm devices are @xmath5 m long so that they are affected along their entire length by the tip potential .
thus , the bulk region close to the tip can get tuned into the valence band , resulting in p - type bulk transport .
s5b and s5c show sgm maps taken with @xmath98 v and -13.5 v , respectively .
it can be seen that a more negative tip voltage results in a larger bulk contribution to the total conductance as it is expected for tip - induced bulk conductance . in the maps plotting the gradient of the conductance ( figs .
s5d and e ) , parallel lines similar to the ones shown in fig
. 2e are visible .
these additional maps clearly demonstrate the evolution of the lines as a function of the tip voltage : for more negative tip voltages , the lines are more widely spaced .
this confirms that the solely tip - induced conductance modulation in the otherwise unperturbed sections of the mesa edge is a function of the potential at the position of the edge states which is consistent with the picture of dephasing occuring in finite - sized metallic regions .
future sgm experiments with coaxially shielded tips @xcite which produce a much more localized potential perturbation could help with many of these complications .
in particular , we expect the long tails in the induced potential to be significantly suppressed relative to the potential directly below the tip .
this improvement will consequently eliminate the long - range gating effect responsible for the emergence of bulk conductance and thus should allow for a more quantitative analysis of the observed conductance modulation . | the discovery of the quantum spin hall state , and topological insulators in general , has sparked strong experimental efforts .
transport studies of the quantum spin hall state confirmed the presence of edge states , showed ballistic edge transport in micron - sized samples and demonstrated the spin polarization of the helical edge states . while these experiments have confirmed the broad theoretical model , the properties of the qsh edge states have not yet been investigated on a local scale . using scanning gate microscopy
to perturb the qsh edge states on a sub - micron scale , we identify well - localized scattering sites which likely limit the expected non - dissipative transport in the helical edge channels . in the micron - sized regions between the scattering sites , the edge states appear to propagate unperturbed as expected for an ideal qsh system and
are found to be robust against weak induced potential fluctuations .
ever since the prediction and realization of topological insulators , this new class of material has not only been studied for fundamental scientific reasons , but also garnered significant interest due to potential applications . in particular , the edge channels of the quantum spin hall ( qsh ) state are considered a promising candidate for low - power information processing because of the expected dissipationless transport in the one - dimensional helical edge channels . in the quantum spin hall state
@xcite , backscattering between the counterpropagating channels in one helical edge state is expected to be suppressed as long as the fermi level is located in the bulk gap , time - reversal symmetry is preserved , and interactions are weak @xcite . in hgte
quantum well structures , the qsh state was predicted to exist if the thickness of the quantum well layer exceeds a critical value of @xmath0 nm @xcite .
following the initial observation of the qsh state in hgte @xcite , the edge - state nature of transport @xcite and the spin - polarization of the edge states @xcite were experimentally demonstrated in the same material .
thus it can serve as a model system for the investigation of the qsh state even though its properties make room - temperature applications appear unlikely .
the good agreement between the experimental results and theoretical predictions based on the landauer - bttiker formalism confirmed that transport in the helical edge channels is indeed ballistic over short distances .
however , conductance values indicative of ballistic transport were only observed on edges shorter than a few microns @xcite . in ref .
, a measured nonlocal conductance value could be explained within the landauer - bttiker formalism , assuming a single scattering site that fully equilibrates the counterpropagating channels along one particular segment of the device edge .
the equilibration was attributed to decoherence of the helical states in a metallic region coupled to the edge states .
such a metallic puddle could form in an inhomogeneous potential landscape introduced by sample growth or processing . in multi - terminal devices ,
the metallic contacts to the qsh edge states can be treated as macroscopic regions causing equilibration , explaining the experimentally observed quantized resistance of the qsh state @xcite .
besides dephasing in metallic regions , various other ways to induce backscattering in a helical edge state have been studied theoretically , including mechanisms based on magnetic @xcite or non - magnetic @xcite impurities , spin - orbit interaction @xcite , or phonons @xcite . against this background of theoretical effort to investigate scattering mechanisms in the qsh state and the obvious presence of backscattering in transport measurements , more detailed experimental investigations of scattering
are urgently needed . |
modern nucleon - nucleon ( @xmath14 ) potentials @xcite give successful description of the @xmath14 scattering data at low energies and have been used for precision calculations in light nuclei @xcite .
similarly , hyperon - nucleon ( @xmath15 ) and hyperon - hyperon ( @xmath16 ) potentials in the coordinate space are known to be quite useful and important for studying the properties of hypernuclei @xcite as well as the equation of state for hyperonic matter in neutron stars @xcite . however , there are still large experimental uncertainties in @xmath15 and @xmath16 interactions at present , because the short life - time of the hyperons makes the scattering experiments difficult . accordingly ,
phenomenological @xmath15 and @xmath16 potentials are not well constrained from data even under some theoretical guides @xcite .
in such a situation , it would be desirable to analyse hyperon interactions on the basis of the first principle lattice qcd simulations .
studies along this line for nucleon interactions was initiated in ref.@xcite , where the @xmath14 scattering lengths were extracted from the quenched simulations by using the lscher s finite volume method @xcite . the same method was applied later to the ( 2 + 1)-flavor simulations with the mixed action @xcite .
the scattering length of @xmath17 system was first examined in @xcite in quenched simulations on a small lattice box .
subsequently , @xmath17 and @xmath18 scattering lengths were studied in ( 2 + 1)-flavor simulations with the mixed action @xcite .
such low - energy scattering parameters calculated in lattice qcd would be valuable inputs to construct phenomenological @xmath15 and @xmath16 potentials to be used for studying hypernuclei and hyperonic matter .
recently , an alternative but closely related approach to @xcite has been proposed to define the @xmath14 potential from lattice qcd @xcite .
since the potential is not a direct physical observable , one can not make quantitative comparison of this lattice @xmath14 potential with the phenomenological @xmath14 potentials .
however , they are both designed to reproduce the correct scattering phase shifts , and their spatial structures are found to have common features @xcite ; the attraction at long and intermediate distances and the strong repulsion at short distance in the @xmath19-wave channel @xcite . the purpose of the present paper is to report our first attempt to apply the above approach to the hyperon - nucleon systems . for the @xmath14 potential , only two channels ( isovector and isoscalar )
exist in the flavor su(2 ) space , i.e. , @xmath20 . including the strange quark extends this to @xmath21 in the flavor su(3 ) space .
the isovector ( isoscalar ) channel in the @xmath14 sector is assigned to be a subset in the @xmath22-plet ( @xmath23-plet ) representation .
the potentials in newly arising channels are hardly determined from experiments so far . in this paper , we focus on the @xmath24 potential in the isovector ( @xmath25 ) channel as a first step .
( a preliminary account of this system has been reported in @xcite . )
there are two main reasons for picking up this channel : ( i ) theoretically , it is the simplest generalization of the @xmath14 system ; @xmath8 is obtained from @xmath26 by replacing the @xmath27-quarks in the neutron by the @xmath6-quarks .
also it is a channel which do not have strong decay into other @xmath15 systems . in the isoscalar
( @xmath28 ) channel is above the @xmath29 threshold . ]
( ii ) experimentally , not much information has been available on the @xmath24 interaction except for a few studies ; a recent report gives the upper limit of elastic and inelastic cross sections @xcite , and earlier publications suggest weakly attractive @xmath30-nucleus potential @xcite .
the @xmath30-nucleus interaction will be soon studied as one of the day - one experiments at j - parc @xcite via @xmath31 reaction with nuclear target .
this paper is organized as follows . in section [ sec : formulation ] , we describe the basic formulation to derive the @xmath8 potential through the bethe - salpeter amplitude measured in the lattice qcd simulations .
our lattice setup is explained in section [ sec : setup ] . in section [ sec : results ] , we show numerical results of the potentials in the spin - singlet and spin - triplet channels and their quark mass dependence .
we also show the estimate of the scattering lengths in these channels .
section [ sec : summary ] is devoted to summary and concluding remarks .
our methodology to obtain the baryon potentials is along the lines of @xcite .
( see also @xcite for the seminal attempts to introduce similar notion of the potential . )
we consider the low - energy @xmath32-@xmath30 scattering and start with an effective schrdinger equation for the equal - time bethe - salpeter ( bs ) wave function @xmath33 obtained from the lippmann - schwinger equation @xcite ; @xmath34 here @xmath35 and @xmath36 are the reduced mass of the @xmath24 system and the non - relativistic energy in the center - of - mass frame , respectively .
( [ effschreq ] ) is derived from qcd by adopting local three - quark operator as the nucleon interpolating operator .
it can be shown that the non - local potential @xmath37 is energy independent .
also , it is unique given the interpolating operator as long as the bs - amplitude for all energies is known @xcite . if we take other nucleon interpolating operator , both the bs wave function and the non - local potential change their spatial structure without affecting the observables such as the binding energy and the phase shift . at low energies ,
the nonlocality of the potential can be expanded as @xcite @xmath38 where the dimensionless expansion parameter is @xmath39 with @xmath40 being the typical scale of the strong interaction such as the excitation energy of a single baryon .
the general expression of the potential @xmath41 is known to be @xcite @xmath42 here @xmath43 is the tensor operator with @xmath44 , @xmath45 are symmetric ( @xmath46 ) and antisymmetric ( @xmath47 ) spin operators , @xmath48 is the orbital angular momentum operator , and @xmath49 ( @xmath50 ) is isospin operator for @xmath51 ( @xmath52 ) .
we note that the antisymmetric spin - orbit forces ( @xmath53 and @xmath54 ) do not arise in the @xmath14 case because of the identical nature of the nucleon within the isospin symmetry .
according to the above expansion , the wave function should be classified by the total isospin @xmath55 , the total angular momentum @xmath56 and the parity .
a particular spin ( isospin ) projection is made in terms of @xmath57 ( @xmath58 ) , e.g. , for the isospin projection we have @xmath59 and @xmath60 .
the equal - time bs wave function for @xmath61 and @xmath62 ( @xmath19-wave ) on the lattice is obtained by @xmath63+\vec{x } ) \xi^0_\beta(\vec{x } ) \right| p \xi^0 ; k \right\rangle , \\
p_\alpha(x ) & = & \varepsilon_{abc } \left ( u_a(x ) c \gamma_5 d_b(x ) \right ) u_{c\alpha}(x ) , \label{proton } \\
\xi^0_\beta(y ) & = & \varepsilon_{abc } \left ( u_a(y ) c \gamma_5
s_b(y ) \right ) s_{c\beta}(y ) , \label{xi}\end{aligned}\ ] ] where @xmath64 and @xmath65 denote the dirac indices , @xmath66 , @xmath67 and @xmath68 the color indices , and @xmath69 the charge conjugation matrix .
the summation over @xmath70 is taken for cubic transformation group to project out the @xmath19-wave state .
, which however are expected to be small in the ground state . ]
the summation over @xmath71 is to select the state with zero total momentum .
here we take local field operator @xmath72 and @xmath73 for the proton and @xmath0 .
the wave function and the potential ( or equivalently the off - shell behavior of the scattering amplitude ) depend on the choice of interpolating operators .
this is the situation common to any field theories . in this paper
, we focus exclusively on the local operators as introduced above and leave further discussions on the operator dependence to @xcite .
we take the upper components of the dirac indices to construct the spin singlet ( triplet ) channel by @xmath74 ( @xmath75 ) .
the bs wave function @xmath33 is understood as a probability amplitude to find `` nucleon - like '' three - quarks located at point @xmath76 and `` @xmath30-like '' three - quarks located at point @xmath71 .
our bs wave function has information not only of the elastic amplitude @xmath77 but also of the inelastic amplitudes : the inelastic effects are localized in coordinate space at low energies and do not affect the asymptotic behavior of @xmath33 and hence the phase shift . in the actual simulations , the bs wave function is obtained from the four - point correlator , @xmath78 here @xmath79 is a wall source located at @xmath80 , which is defined by @xmath81 with @xmath82 and @xmath83 .
the eigen - energy and the eigen - state of the six quark system are denoted by @xmath84 and @xmath85 , respectively , with the matrix element @xmath86 . for @xmath87 , the @xmath88 and hence the wave function @xmath89
are dominated by the lowest energy state .
the lowest energy state @xmath90 created by the wall source @xmath91 contains not only the @xmath19-wave @xmath24 component but also the components which can mix by qcd dynamics , such as the @xmath92-wave due to the tensor force and @xmath93 state due to the rearrangement of the quarks inside baryons . in principle
, these components can be disentangled by preparing appropriate operator sets for the sink .
study along this line to extract the mixing between the @xmath19-wave and the @xmath92-wave in low energy @xmath14 interaction was put forward recently in @xcite . in the present paper , instead of making such decomposition
, we define an effective central potential @xmath94 for the @xmath19-wave component according to our previous works @xcite : @xmath95 such a potential in the @xmath25 sector has only the spin dependence as @xmath96 where effects of the tensor force and coupled - channel effects are implicitly taken into account in @xmath97 .
it is in order here to make some remarks on the potential we have defined . 1 .
the description by the energy - independent non - local potential @xmath37 is equivalent to that by the momentum - dependent local potential , @xmath98 .
the potential @xmath94 in eq.([effcpot ] ) is the leading order term of the derivative expansion of @xmath98 and can be determined by the bs wave function at @xmath99 .
higher order terms can be extracted successively by the bs wave functions with different @xmath100 .
it was recently reported that momentum - dependence of @xmath98 is small at least up to the center of mass momentum @xmath101 mev for the @xmath14 system @xcite .
the non - local potential @xmath37 has one - to - one correspondence to the baryon interpolating operator adopted in defining the bethe - salpeter amplitude .
different choices of the interpolating operator would give different bs wave function and the baryon - baryon potential , although the phase shifts and binding energies are unchanged .
an advantage of working directly in lattice qcd is that we can unambiguously trace the relation between the baryon interpolating operator and the bs wave function or equivalently the associated potential .
we use the plaquette gauge action and the wilson fermion action with the gauge coupling @xmath102 on the @xmath103 lattice . the heatbath algorithm combined with the overrelaxation
is used to generate quenched gauge configurations . after skipping 3000 sweeps for thermalization
, measurement is made for every 200 sweep .
the dirichlet ( periodic ) boundary condition is imposed for quarks in the temporal ( spatial ) direction . the wall source is placed at @xmath104 with the coulomb gauge fixing .
these setup are same as our previous calculation for the @xmath14 potential @xcite . masses of pseudo - scalar and vector mesons obtained at hopping parameter @xmath105 are fitted as @xmath106 for @xmath107 ( @xmath108 ) .
the fit gives @xmath109 for the critical hopping parameter . from @xmath110 mev and @xmath111 mev
, we determine the hopping parameter for physical @xmath4 quarks and the lattice spacing as @xmath112 and @xmath113 fm ( @xmath114 gev ) , respectively , while the hopping parameter for the strange quark mass is given by @xmath115 from @xmath116 mev .
a corresponding lattice volume is @xmath117 , which is large enough to accommodate two baryons . in order to check the thresholds of two baryon systems with strangeness @xmath118 ( @xmath119,@xmath120,@xmath121 and @xmath122 )
, we calculate the two - point correlator , @xmath123 , for the octet baryons ( @xmath124 ) , and @xmath125 is the wall - source for @xmath126 .
the interpolating fields for @xmath127 and @xmath128 employed in this work are @xmath129 @xmath130 table [ masses ] lists the meson and baryon masses measured at two values of @xmath131 , 0.1665 and 0.1678 , with fixed @xmath132 . at @xmath133
corresponding to @xmath134 mev , 17 exceptional configurations are excluded from totally 1300 gauge configurations for the average , while no such exceptional configuration appears at @xmath135 ( @xmath136 mev ) .
as seen in table [ masses ] , the present results for the baryon masses are consistent with the experimentally observed ordering of the two - baryon thresholds in the strangeness @xmath1 sector : @xmath137 .
this warrants that @xmath24 in the @xmath25 channel treated in this paper is indeed the lowest energy scattering state .
.hadron masses in the unit of mev calculated for the hopping parameters @xmath131 and @xmath138 , with the number of gauge configurations @xmath139 . [ cols="^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ]
, in @xmath9 ( circle ) and @xmath10 ( triangle ) channels , obtained at @xmath140 . the green s functions @xmath141 with @xmath142 in @xmath9 ( solid line ) and @xmath10 ( dotted line ) are also shown .
the inset shows its enlargement .
( right ) the effective central potential for @xmath8 , in the @xmath9 ( circle ) and @xmath10 ( triangle ) , obtained from the wave function at time slice @xmath140 .
the inset shows its enlargement .
, title="fig : " ] , in @xmath9 ( circle ) and @xmath10 ( triangle ) channels , obtained at @xmath140 . the green s functions @xmath141 with @xmath142 in @xmath9 ( solid line ) and @xmath10 ( dotted line ) are also shown .
the inset shows its enlargement .
( right ) the effective central potential for @xmath8 , in the @xmath9 ( circle ) and @xmath10 ( triangle ) , obtained from the wave function at time slice @xmath140 .
the inset shows its enlargement .
, title="fig : " ] the left panel of fig .
[ wavepot ] shows the wave functions obtained at time slice @xmath140 with @xmath143 .
the open circle ( triangle ) corresponds to the case of the @xmath9 ( @xmath10 ) channel .
they are normalized to be unity at the spatial boundary @xmath144 . for @xmath145 fm
, the wave function is calculated for all possible values of @xmath146 , while , for the outer region , it is calculated only on the @xmath147 axis and their nearest neighbors to reduce the numerical cost . as seen in the figure ,
the wave functions are suppressed at short distance and are slightly enhanced at medium distance in both @xmath9 and @xmath10 channels .
there is a sizable difference of the suppression pattern at short distance between @xmath9 and @xmath10 ; the repulsion is stronger in the @xmath9 channel . to derive the @xmath24 potential through eq .
( [ effcpot ] ) , we need to know the non - relativistic energy @xmath148 .
it has been shown in ref .
@xcite that @xmath149 and thus @xmath100 can be accurately determined by fitting the wave function in the asymptotic region in terms of the lattice green s function @xmath150 which is the solution of @xmath151 with @xmath152 being the periodic delta function @xcite .
results of the fit in the range @xmath153 are shown in the left panel of fig .
[ wavepot ] .
the fitting range is determined so that it is outside the range of the interaction ( see the right panel of fig .
[ wavepot ] ) .
the non - relativistic energies thus obtained in the @xmath9 ( @xmath10 ) channel becomes @xmath154 mev ( @xmath155 mev ) .
we have checked that the results of the fit in different ranges @xmath156 and @xmath157 introduce systematic errors only less than half of the statistical errors for @xmath100 .
note that @xmath100 can be negative for the scattering state in a finite box if the interaction is attractive .
dependence of the potential in the @xmath9 channel ( left panel ) and the @xmath10 channel ( right panel ) for several values of @xmath158 .
the horizontal lines denote the central values at @xmath140 which are adopted for the potential in fig.1 ( right ) .
the results indicated by @xmath159 $ ] are obtained by a constant fit to the data for @xmath160 with a single - elimination of the jackknife method . ]
dependence of the potential in the @xmath9 channel ( left panel ) and the @xmath10 channel ( right panel ) for several values of @xmath158 .
the horizontal lines denote the central values at @xmath140 which are adopted for the potential in fig.1 ( right ) .
the results indicated by @xmath159 $ ] are obtained by a constant fit to the data for @xmath160 with a single - elimination of the jackknife method . ] the effective central potentials for @xmath8 system obtained from eq .
( [ effcpot ] ) with the wave function @xmath89 and the energy @xmath100 at @xmath140 are shown in the right panel of fig .
[ wavepot ] for the @xmath9 and @xmath10 channels at @xmath161 mev . in order to check
whether the ground state saturation of the @xmath8 system is achieved in the present results , we plot the @xmath162-dependence of the potential at distances @xmath163 , @xmath164 , @xmath165 , @xmath166 and @xmath167 fm in fig . [ pots ] .
the fact that the potential is stable against @xmath162 for @xmath168 within errors indicates that the ground state saturation is indeed achieved at @xmath140 .
this is the reason why we adopted the values at @xmath140 in the right panel of fig.[wavepot ] .
note that a constant fit to data for @xmath169 with a single elimination of the jackknife method does not introduce appreciable change of the final values and the errors of the potential as shown in fig.[pots ] . in the right panel of fig .
[ wavepot ] , the potential in the @xmath170 system shows a repulsive core at @xmath171 fm surrounded by an attractive well , similar to the @xmath14 system @xcite .
in contrast to the @xmath14 case , however , one finds that the repulsive core of the @xmath8 potential in the @xmath9 channel is substantially stronger than that in the @xmath10 channel .
such a large spin dependence is also suggested by the quark cluster model @xcite . the relatively weak attraction in the medium to long distance region that ( @xmath12 fm @xmath13 fm ) is similar in both @xmath9 and @xmath10 channels .
as the energies determined above and scattering lengths given later indicate , the present potentials are weakly attractive on the whole in both spin channels in spite of the repulsive core at short distance .
also , the attraction @xmath10 channel seems a little stronger than that in the @xmath9 channel . in the @xmath9 channel at @xmath172 mev ( circle ) and @xmath173 mev ( box ) . the central part of the opep ( @xmath174 in eq .
( [ eq : opep ] ) is also given by solid line .
( right ) same as the left figure , but in the @xmath10 channel at @xmath172 mev ( triangle ) and @xmath173 mev ( diamond ) .
, title="fig : " ] in the @xmath9 channel at @xmath172 mev ( circle ) and @xmath173 mev ( box ) . the central part of the opep ( @xmath174 in eq .
( [ eq : opep ] ) is also given by solid line .
( right ) same as the left figure , but in the @xmath10 channel at @xmath172 mev ( triangle ) and @xmath173 mev ( diamond ) .
, title="fig : " ] figure [ fig : potqmass_w.opep ] compares the @xmath8 potential at @xmath172 mev with that at @xmath173 mev in the @xmath9 channel ( left ) and in the @xmath10 channel ( right ) . at @xmath173 mev ,
the potentials are evaluated at @xmath175 .
the non - relativistic energy in this case is @xmath176 mev ( @xmath177 mev ) in the @xmath9 ( @xmath10 ) channel , where the region that @xmath178 is adopted to fit the wave function by the green s function .
we have checked that the results of the fit in different ranges @xmath179 and @xmath156 introduce systematic errors only less than half of the statistical errors for @xmath100 . the height of the repulsive core increases as the @xmath4 quark mass decreases , while the significant difference is not seen in the medium to long distances within the error bars .
the exchange of a single @xmath180 would conduct the interaction at long distance in the @xmath8 system : @xmath181 the pseudo - vector @xmath182 coupling @xmath183 and the @xmath184 coupling @xmath185 are related as @xmath186 with the parameter @xmath187 @xcite .
also we define @xmath188 .
the solid lines in fig.[fig : potqmass_w.opep ] is the one pion exchange potential ( opep ) obtained from eq.([eq : opep ] ) with @xmath172 mev , @xmath189 mev ( corresponding to @xmath133 ) and the empirical values , @xmath190 @xcite and @xmath191 @xcite .
unlike the @xmath14 potential in the @xmath19-wave , the opep in the present case has opposite sign between the spin - singlet channel and the spin - triplet channel .
also , the absolute magnitude of opep is weak due to the factor @xmath192 . as is seen from fig.[fig : potqmass_w.opep ] , we do not find clear signature of opep at long distance ( @xmath193 fm ) in our potential within statistical errors .
on the other hand , there is a clear departure from opep at medium distance ( @xmath194 ) in both @xmath9 and @xmath10 channels .
these observations may indicate a mechanism of state - independent attraction such as the correlated two pion exchange .
it should be mentioned here that there is in principle a quenched artifact to the baryon - baryon potentials from the flavor singlet hairpin diagram ( the ghost exchange ) @xcite .
its contribution to the central potential has a spin - dependent exponential tail @xmath195 , which dominates over the yukawa potential at large distances .
its significance can be estimated by comparing the sign and the magnitude of @xmath196 in the spin - singlet and spin - triplet channels . our present data at lightest quark mass @xmath197 mev shows no evidence of the ghost contribution at large distances within errors .
this may indicate the weak coupling of the @xmath198 to @xmath32 and @xmath30 . in the @xmath9 ( circle ) and @xmath10 ( triangle ) , obtained by the lscher s formula from the asymptotic behavior of the wave function .
the vertical dashed line is the physical point at @xmath199 gev . ] using the energy @xmath148 obtained from the asymptotic behavior of the wave function in sec.[sec : central potential ] , the @xmath24 scattering lengths can be deduced from the lscher s formula @xcite , @xmath200 where @xmath201 with @xmath202 is obtained by the analytic continuation of the generalized zeta - function @xmath203 defined for @xmath204 .
the sign of the @xmath19-wave scattering length @xmath205 is defined to be positive for weak attraction .
the results are plotted in fig.[sctlng ] for two different values of the pion masses , @xmath161 mev and @xmath206mev .
both @xmath207 ( the scattering length in the @xmath10 channel ) and @xmath208 ( the scattering length in the @xmath9 channel ) slightly increase as the quark mass decreases : @xmath209 0.05(1 ) @xmath210 0.10(6 ) fm and @xmath211 0.09(1 ) @xmath210 0.21(6 ) fm .
the positive sign indicates that the @xmath8 interaction is attractive on the whole in both @xmath9 and @xmath10 channels .
due to the possible non - linear dependence of the scattering length as a function of the quark mass as is well - known in the case of the @xmath14 system @xcite ( see also @xcite and references therein ) , we are not able to obtain a reliable estimate of the scattering length of the @xmath8 system at the physical @xmath4 quark mass . in addition , a possible contamination from the ghost exchange may arise at small quark masses in the quenched qcd simulations @xcite .
therefore , it is necessary in the future to carry out the full qcd calculation toward lighter quark masses to predict the precise value of @xmath205
. it would be interesting here to summarize diverse results on the @xmath24 scattering length in other approaches .
the @xmath8 interaction in chiral effective field theory @xcite predicts weak repulsive scattering lengths of @xmath212 fm and @xmath213 fm .
the phenomenological boson exchange model ( e.g. , sc97f ) @xcite gives @xmath214 fm and @xmath215 fm .
the quark cluster model ( fss2 ) @xcite gives @xmath216 fm and @xmath217 fm , while qcd sum rules @xcite gives @xmath218 fm and @xmath219 fm .
we study the @xmath8 interaction in the @xmath9 and @xmath10 channels through the equal - time bethe - salpeter amplitude measured by the quenched lattice qcd simulations on a ( @xmath220 fm)@xmath2 lattice with the quark masses corresponding to @xmath221 and @xmath222 .
we adopt specific choice of the baryon interpolating operators as given in eqs .
( [ proton ] ) and ( [ xi ] ) and extract the potential associated with these operators .
the effective central potential deduced from the bethe - salpeter wave function has repulsive core at short distance surrounded by attractive well at the medium and long distances .
the scattering lengths determined by fitting the asymptotic wave function with the solution of the helmholtz equation indicates that the @xmath8 interaction is attractive on the whole in both @xmath9 and @xmath10 channels .
there is a slight tendency that the @xmath10 interaction is more attractive than that in the @xmath9 channel . to reduce the uncertainties due to the lattice discretization at short distances , we need to carry out the simulations with smaller lattice spacing and/or with the improved lattice action .
systematic studies in various other channels such as @xmath17 , @xmath18 and @xmath29 are not only interesting by themselves but also important for studying the structure of hypernuclei and the interior of neutron stars . also , ( 2 + 1)-flavor qcd simulations with realistic quark masses will be the ultimate goal to make qualitative predictions of the @xmath14 and @xmath15 interactions : studies along this line with the pacs - cs gauge configurations@xcite are now under way .
this work is supported by the large scale simulation program no.07 - 07 ( fy2007 ) of high energy accelerator research organization ( kek ) .
we are grateful for authors and maintainers of cps++@xcite , of which a modified version is used for measurement done in this work .
h. n. is supported by the special postdoctoral researchers program at riken .
this research was partly supported by grants - in - aid for young scientists ( b ) ( no .
17740174 ) from the japan society for promotion of science ( jsps ) , and by the ministry of education , science , sports and culture , grant - in - aid ( nos .
13135204 , 15540254 , 18540253 , 19540261 , 20340047 ) .
t. a. rijken , v. g. j. stoks and y. yamamoto , phys . rev .
c * 59 * , 21 ( 1999 ) [ arxiv : nucl - th/9807082 ] .
+ t. a. rijken and y. yamamoto , phys .
c * 73 * , 044008 ( 2006 ) [ arxiv : nucl - th/0603042 ] .
+ t. a. rijken and y. yamamoto , arxiv : nucl - th/0608074 .
j. haidenbauer , u. g. meissner , a. nogga and h. polinder , lect .
notes phys .
* 724 * , 113 ( 2007 ) [ arxiv : nucl - th/0702015 ] .
+ h. polinder , j. haidenbauer and u. g. meissner , phys .
b * 653 * , 29 ( 2007 ) [ arxiv:0705.3753 [ nucl - th ] ] .
+ h. polinder , arxiv:0708.0773 [ nucl - th ] .
r. jastrow , _ phys .
rev . _ * 81 * , 165 ( 1951 ) .
+ y. nambu , _ phys .
_ * 106 * , 1366 ( 1957 ) .
+ m. taketani _
* 39 * ( 1967 ) .
+ n. hoshizaki _
et al . _ , _ prog .
* 42 * ( 1968 ) .
+ g. e. brown and a. d. jackson , _ nucleon - nucleon interaction _ , ( north - holland , amsterdam , 1976 ) .
+ r. machleidt , _ adv .
_ * 19 * , 189 ( 1989 ) . | we calculate potentials between a proton and a @xmath0 ( hyperon with strangeness @xmath1 ) through the equal - time bethe - salpeter wave function , employing quenched lattice qcd simulations with the plaquette gauge action and the wilson quark action on ( 4.5 fm)@xmath2 lattice at the lattice spacing @xmath3 fm .
the @xmath4 quark mass in our study corresponds to @xmath5 and 0.51 gev , while the @xmath6 quark mass corresponds to the physical value of @xmath7 .
the central @xmath8 potential has a strong ( weak ) repulsive core in the @xmath9 ( @xmath10 ) channel for @xmath11 fm , while the potential has attractive well at medium and long distances ( @xmath12 fm @xmath13 fm ) in both channels .
the sign of the @xmath8 scattering length and its quark mass dependence indicate a net attraction in both channels at low energies . , , , and lattice qcd , hyperon - nucleon interaction , nuclear forces , hypernuclei 12.38.gc , 13.75.ev , 21.30.-x , 21.80.+a |
in the main manuscript , we state that the essential features in the electronic structure of graphene superlattices ( gss ) revealed by considering the kronig - penney type of potential remain valid for gss made with different types of periodic potentials . in this section ,
we support this by showing the results for gss with sinusoidal types of external periodic potentials [ fig .
[ sfig1](a ) ] .
the function @xmath31 defined by eq .
( 6 ) in ref . for a sinusoidal type of external periodic potential @xmath32 is @xmath33 .
therefore , as shown in eqs .
( 9 ) and ( 15 ) of ref .
, the group velocity at the original dirac point perpendicular to the periodic direction is given by @xmath34 , where @xmath35 is the group velocity in pristine graphene and @xmath36 .
here , @xmath37 is the zeroth order bessel function of the first kind .
our calculations show that a new pair of massless dirac points are generated whenever @xmath38 , i.e. , @xmath39 is equal to @xmath40 where @xmath41 is the _ n_-th root of @xmath37 ( e.g. , @xmath42 , @xmath43 , etc . ) [ fig .
[ sfig1](b ) ] .
figure [ sfig1](c ) shows the energy bandstructure of a sinusoidal type of gs with @xmath44 . because this value of @xmath39 is between @xmath45 and @xmath46 , a pair of new zero - energy massless dirac cones
are generated , and they clearly affect the landau level degeneracy [ fig .
[ sfig1](d ) ] in the same way as discussed in the main manuscript for a kronig - penney type of gs . [
[ effects - of - symmetry - breaking - on - the - newly - generated - massless - fermions ] ] 2
. effects of symmetry breaking on the newly generated massless fermions ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ in this section , we discuss the effect of symmetry breaking of the external periodic potential on the newly generated massless fermions .
( the case of random perturbation is discussed in ref .
42 of the main manuscript . in this section ,
we focus on a periodic perturbing potential that breaks the even or odd symmetry . )
figure [ sfig2](a)-(c ) repeats the results shown in the main manuscript for a kronig - penney type of periodic potential having both even and odd symmetries . if the odd symmetry is broken by adding an appropriate perturbation [ fig .
[ sfig2](d ) ] , new branches of massless fermions are still generated [ fig .
[ sfig2](e ) ] . the energy at these new massless dirac points however is different from that of the original dirac point [ fig .
[ sfig2](e ) ] .
even though @xmath25 landau level does not have the degeneracy coming from multiple dirac points , some lower - index landau levels still show this kind of degeneracy [ fig .
[ sfig2](f ) ] . if the even symmetry is broken through a perturbing potential [ fig .
[ sfig2](g ) ] , new dirac points are not generated [ fig .
[ sfig2](h ) ] .
however , the signature of newly generated states may still be probed with photoemission experiments or transport measurements @xcite . and @xmath47 , @xmath48 , in a gs depicted in fig .
[ sfig1](a ) with @xmath49 versus @xmath50 and @xmath51 which are the angles between the @xmath52 axis and wavevectors @xmath53 and @xmath54 , measured from the newly - generated massless dirac point ( appearing when @xmath55 and moving along the @xmath15 direction as @xmath11 is increased further ) , respectively . the overlap is shown in a gray scale ( 0 in black and 1 in white ) .
the results show negligible dependence on @xmath56 and @xmath57 when they are smaller than @xmath58 ( in the figures , @xmath59 ) .
the two states are in the same band ( @xmath60 ) in ( a ) and are in different bands ( @xmath61 ) in ( b ) .
( c ) and ( d ) , and , ( e ) and ( f ) : same quantities as in ( a ) and ( b ) for gss with @xmath62 and @xmath63 , respectively . ]
the pseudospin character of the newly - generated massless states are different from that of the original dirac fermions . in order to illustrate the pseudospin character of these states ,
we numerically calculate the overlap @xmath64 of two quasiparticle states @xmath65 and @xmath66 in a gs having wavevectors @xmath53 and @xmath67 measured from the newly - generated dirac point ( appearing when @xmath55 and moving along the @xmath15 direction as @xmath11 is increased further ) . the behavior shown in fig .
[ sfig3 ] , which corresponds to the overlap of the pseudospin part of the wavefunctions , is robust if the magnitudes of @xmath53 and @xmath67 are smaller than @xmath58 .
as mentioned in the main manuscript , the pseudospin character of these additional massless fermions [ e.g. , the left and the right dirac cones ( not the center one ) in fig . [ fig1](c ) ] are different from that of the original massless dirac fermions .
backscattering amplitude due to a slowly varying potential within one of the new cones does not vanish ( fig .
[ sfig3 ] ) .
the zigzag form of the vector potential we used results in the perpendicular magnetic field of a uniform strength pointing along the @xmath68 direction for half the artificial periodicity and along the @xmath69 direction for the other half .
the group velocity along the @xmath15 direction is given by @xmath70 where @xmath31 for a kronig - penney type of superlattice is given by @xmath71 ( see ref . ) .
when eq .
( [ eq : u0 ] ) is satisfied , @xmath72 .
the effect of up to @xmath310 % disorder in the on - site potential of graphene on the ll broadening is small ( ref . ) . since the low - energy electronic band structure of a gs
can effectively be decomposed into a finite number of dirac cones , we expect that the lls in gss with moderate disorder can be observed . for a kronig - penney superlattice with @xmath27 nm and @xmath73 ( corresponding to @xmath74 t )
, the magnetic energy @xmath75 is 17 mev . in this case
, temperature has to be lowered to observe the ll quantization . | we show that , when graphene is subjected to an appropriate one - dimensional external periodic potential , additional branches of massless fermions are generated with nearly the same electron - hole crossing energy as that at the original dirac point of graphene .
because of these new zero - energy branches , the landau levels at charge neutral filling becomes @xmath0-fold degenerate ( with _ n_=0,1,2 , ... , tunable by the potential strength and periodicity ) with the corresponding hall conductivity @xmath1 showing a step of size @xmath2 .
these theoretical findings are robust against variations in the details of the external potential and provide measurable signatures of the unusual electronic structure of graphene superlattices .
the physical properties of graphene @xcite are currently among the most actively investigated topics in condensed matter physics .
graphene has the unique feature that the low - energy charge carriers are well described by the two - dimensional ( 2d ) massless dirac equation , used for massless neutrinos , rather than by the schrdinger equation @xcite .
moreover , graphene is considered to be a promising candidate for electronics and spintronics applications @xcite .
it has been shown that , because of their gapless energy spectrum and chiral nature , the charge carriers in graphene are not hindered by a slowly varying electrostatic potential barrier at normal incidence @xcite , analogous to the klein tunneling effect predicted in high - energy physics .
direct evidences of klein tunneling through a single barrier in graphene @xcite have been observed in recent experiments @xcite .
application of multiple barriers or periodic potentials , either electrostatic @xcite or magnetic @xcite , to graphene has been shown to modulate its electronic structure in unique ways and lead to fascinating new phenomena and possible applications .
periodic arrays of corrugations @xcite have also been proposed as graphene superlattices ( gss ) .
experimentally , different classes of gss have been fabricated recently .
patterns with periodicity as small as 5 nm have been imprinted on graphene through electron - beam induced deposition of adsorbates @xcite .
epitaxially grown graphene on the ( 0001 ) surface of ruthenium @xcite and that on the ( 111 ) surface of iridium @xcite also show superlattice patterns with @xmath33 nm lattice period .
the amplitude of the periodic potential applied to graphene in these surface systems has been estimated to be in the range of a few tenths of an electron volt @xcite .
fabrication of periodically patterned gate electrodes is another possible way of making gss with periodicity close to or larger than @xmath320 nm .
the quantum hall plateaus in graphene take on the unusual values of @xmath4 where @xmath5 is a non - negative integer @xcite .
the factor 4 comes from the spin and valley degeneracies . in bilayer graphene ,
the quantum hall plateaus are at @xmath6 with @xmath5 a positive integer @xcite .
these unconventional quantum hall effects have been experimentally verified @xcite , providing evidences for 2d massless particles in graphene and massive particles in bilayer graphene . in this paper , we investigate the lls and the quantum hall effect in gss formed by the application of a one - dimensional ( 1d ) electrostatic periodic potential and show that they exhibit additional unusual properties .
we find that , for a range of potential shapes and parameters , new branches of massless fermions are generated with electron - hole crossing energy the same as that at the original dirac point of pristine graphene .
these additional massless fermions affect the lls qualitatively .
in particular , the lls with energy corresponding to the fermi energy at charge neutrality ( i.e. , zero carrier density ) become @xmath0-fold degenerate ( @xmath7 ) , depending on the strength and the spatial period of the potential ( pristine graphene corresponds to @xmath8 ) . accordingly , when sweeping the carrier density from electron - like to hole - like , the quantum hall conductivity in such a gs is predicted to show an unconventional step size of @xmath2 that may be tuned by adjusting the external periodic potential . in our study ,
the electronic structure of the gss is evaluated using the methods developed in ref .
; we evaluate the bandstructure of the gs numerically by solving the 2d massless dirac equation with the external periodic potential included using a planewave basis . similarly , to obtain the lls , the eigenstates of the gss under an external perpendicular magnetic field
are expanded with planewaves .
we work in a landau gauge with the vector potential depending on the position coordinate along the direction of the periodicity of the gs , and a zigzag form for the vector potential with a very large artificial periodicity ( large compared to the gs periodicity ) is employed to mimic a constant magnetic field near the origin in position space @xcite .
we have checked that the lls are converged in energy to within less than 1 % with respect to the size of the supercell for the vector potential and the kinetic energy cutoff for the planewaves .
the size of the largest supercell and that of the smallest sampling distance in real space used are 400 and 0.05 in units of a single unit cell , respectively .
figure [ fig1](a ) shows a gs formed by a kronig - penney type of electrostatic potential periodic along the _ x _ direction , with lattice parameter @xmath9 and barrier width @xmath10 .
remarkably , unlike that in graphene [ fig .
[ fig1](b ) ] , the bandstructure in a gs [ fig .
[ fig1](c ) ] can have , depending on the potential barrier height @xmath11 , more than one dirac point with @xmath12 having exactly the same electron - hole crossing energy @xcite . as fig . [
fig1](c ) shows , the number of dirac points for this type of gss increases by two ( without considering the spin and valley degrees of freedom ) whenever the potential amplitude exceeds a value of @xmath13 with @xmath14 a positive integer . the value of the potential barrier given in eq .
( [ eq : u0 ] ) corresponds to special gss in which the group velocity along the @xmath15 direction vanishes for charge carriers whose wavevector is near the original dirac cone [ e.g. , the dirac cone at the center in fig .
[ fig1](c ) ] @xcite .
all the findings in this study apply in general to gss made from a periodic potential which has both even and odd symmetries , like a sinusoidal type of potential .
the results for gss whose odd or even symmetry is broken are discussed in ref . .
) versus @xmath15 with @xmath12 in gss shown in fig . [ fig1 ] for several different values of barrier height @xmath11 ( specified in each panel in units of @xmath16 ) .
] figure [ fig2 ] shows the evolution of the energy of the electronic states with @xmath12 for a gs depicted in fig .
[ fig1 ] for several different values of @xmath11 . as stated above , the group velocity along the @xmath15 direction becomes zero near @xmath17 when the barrier height is given by eq .
( [ eq : u0 ] ) [ figs . [ fig2](c ) and [ fig2](e ) ] .
when @xmath11 has a value between those specific values , the position of the additional new dirac points move away from the @xmath17 point along the @xmath15 direction with increasing @xmath11 .
the complex behavior of the zero - energy dirac cones revealed by our numerical calculations can not be derived using perturbation theory @xcite because @xmath15 is not small compared to the superlattice reciprocal lattice spacing @xmath18 . moreover , the pseudospin character of these additional massless fermions [ e.g. , the left and the right dirac cones ( not the center one ) in fig .
[ fig1](c ) ] are different from that of the original massless dirac fermions . for example , backscattering amplitude due to a slowly varying potential within one of the new cones does not vanish @xcite .
( in units of @xmath19 with @xmath20 ) versus the landau level index @xmath21 ( @xmath22 ) in gss formed with a 1d kronig - penney potential for several different values of barrier height @xmath11 , with lattice period @xmath23 .
the lls now have a finite width @xmath24 ( shown not to scale and exaggerated in the figure ) arising from the @xmath15 dependence of the energy of the electronic states in a perpendicular magnetic field @xcite .
note the 3-fold and the 5-fold degeneracies around @xmath25 in ( d ) and ( f ) , respectively .
( if the spin and valley degeneracies are considered , those become 12-fold and 20-fold , respectively . ) ] a natural question arising from this peculiar behavior in the electronic structure of a gs , which is topologically different from that of pristine graphene , is how the lls are distributed .
figure [ fig3 ] shows the calculated lls of the 1d kronig - penney gss depicted in fig .
[ fig1 ] for various values of @xmath11 @xcite . when the superlattice potential modulation is moderate [ fig .
[ fig3](b ) ] , the spacings between neighboring lls become smaller than those in pristine graphene [ fig .
[ fig3](a ) ] , owing to a reduction in the band velocity .
once @xmath11 becomes larger than @xmath26 (= 0.4 ev for @xmath27 nm ) , the zero - energy lls ( corresponding to zero carrier density ) become three - fold degenerate [ fig . [ fig3](d ) ] .
an important point to note is that this degeneracy is insensitive to @xmath11 over a range of @xmath11 near @xmath28 because the topology of the electron bands does not change with this variation @xcite .
moreover , even though the massless particles of the different dirac cones may have different band velocities , the degeneracy of the zero - energy lls is not affected . versus carrier density ( with an artificial broadening for illustration ) for a 1d kronig - penney gs with @xmath11 near @xmath29 ( solid red line )
is compared to that of pristine graphene ( dashed blue line ) . ]
the dependence of the hall conductivity @xmath1 on the charge carrier density @xmath30 most directly reflects the degeneracy of the lls .
figure [ fig4 ] schematically shows that , depending on the superlattice potential parameters , @xmath1 of the gss considered has a @xmath2 step as the density is scanned from hole - like to electron - like carriers .
( we have put in the additional factor 4 coming from the spin and valley degeneracies in this discussion and in fig .
[ fig4 ] . ) because the degeneracy of the lls in the 1d gss is insensitive to a variation in @xmath11 , this qualitative difference in @xmath1 of the 1d gss from that of pristine graphene ( fig .
[ fig4 ] ) is expected to be robust , and will provide a measurable signature of the unique electronic structure of the 1d gss . in conclusion , we have shown that the electronic structure of 1d graphene superlattices can have additional dirac cones at the same energy as the original cones at the k and k points of pristine graphene .
these new massless particles contribute to a @xmath0-fold degeneracy in the zero - energy landau levels , whose signature is reflected in a @xmath2 hall conductivity step where @xmath7 depending on the superlattice potential parameters .
this feature of the electronic structure of the 1d graphene superlattices gives rise to new properties for the quantum hall effect .
equally importantly , these new phenomena may provide a direct way to characterize the peculiar electronic structure of these systems experimentally .
c .- h.p .
thanks dmitry novikov and jay deep sau for fruitful discussions .
this work was supported by nsf grant no.dmr07-05941 and by the director , office of science , office of basic energy sciences , division of materials sciences and engineering division , u.s .
department of energy under contract no.de- ac02 - 05ch11231 . y .-
w.s . was supported by the krf ( krf 2008 - 314-c00111 ) and quantum metamaterials research center ( no . r11 - 2008 - 053 - 01002 - 0 ) through the kosef funded by the mest .
computational resources have been provided by nersc and teragrid .
_ note added._ after submission , we became aware of a recent theoretical work @xcite confirming the newly generated massless fermions reported in this manuscript , with applications to transport properties . |
the corot satellite detected two neptune - size planets with transit radii of @xmath83.7 and @xmath84.9 with orbital periods of @xmath85.11 and @xmath811.76days around an old main - sequence k1v star @xcite .
radial velocity measurements led only to an upper limit of 5.7 for the hotter planet , while for the cooler , more massive planet a mass of 28@xmath911 was obtained . from the deduced low average densities of @xmath10g@xmath11 and @xmath12g@xmath11 of corot-24b and corot-24c
respectively , one can expect that the cores of both planets are surrounded by h@xmath13-dominated envelopes .
this assumption agrees well with previous studies , concluding that protoplanetary cores with masses @xmath141.5 and radii @xmath151.6 would hardly lose their h@xmath13-envelopes under the action of the stellar high - energy flux @xcite .
we aim at understanding whether the low density and large radius of the smaller planet are physically possible . in sect .
[ sec : modelling ] and [ sec : results ] we describe the applied hydrodynamic model , and present the obtained upper atmosphere structure and loss rates . in sect .
[ sec : discussion ] we compare the results and discuss the implications of our findings .
we conclude in sect .
[ sec : conclusions ] .
to study the euv - heated upper atmosphere structure and hydrogen thermal escape rates , we apply an energy absorption and 1-d hydrodynamic upper atmosphere model described in detail in @xcite and @xcite .
the model solves the system of hydrodynamic equations for mass , @xmath16 momentum , @xmath17 and energy conservation , @xmath18}{\partial t}+ \frac{\partial r^2 v\left[\frac{\rho v^2}{2 } + e_{\rm t } + p \right]}{\partial r}= \nonumber\\ -\rho v \frac{\partial{\phi}}{\partial r}r^2+q_{\rm euv}r^2 + \frac{\partial}{\partial r } \left(\kappa r^2\frac{\partial t}{\partial r}\right).\label{eq : energy_conservation}\end{aligned}\ ] ] because the influence of the roche lobe can not be neglected , we consider also the energy @xmath19 per unit mass of a test particle in the ecliptic plane @xmath20 where @xmath21 is newton s gravitational constant , is the planetary mass , @xmath22 is the distance from a given point to the center of mass , @xmath23 is the stellar mass and , @xmath24 is the distance between the star and the planet . in the equations above , @xmath25 is the planetocentric distance , @xmath26 is the time , @xmath27 the atmospheric temperature , @xmath28 is the euv volume heating rate , @xmath29 the atmospheric mass density , @xmath30 the gas flow velocity , @xmath31 the atmospheric pressure , and @xmath32 erg@xmath33s@xmath34k@xmath34 is the thermal conductivity @xcite .
the thermal energy @xmath35 of the atmospheric particles per unit volume can then be written as , @xmath36 k t , \label{eq : thermalenergy}\ ] ] with @xmath37 being the boltzmann constant and @xmath38 , @xmath39 neutral particle and @xmath40 , @xmath41 ion densities .
the continuity equations for these number densities are solved as described in @xcite . to calculate q@xmath42
, we adopt the average euv photoabsorption cross sections @xmath43 of @xmath44@xmath45 and @xmath46@xmath45 for h atoms and h@xmath13 molecules , respectively @xcite , in agreement with what is provided by experimental and theoretical data @xcite .
to estimate the euv ( 10 - 92 nm ) luminosity of corot-24 , we consider a stellar rotation period of 27.3days @xcite .
using the scaling laws of @xcite and a @xmath47 ( the power - law index correlating the rossby number at saturation and the x - ray flux ; see eq.1 of * ? ? ?
* ) of @xmath482.18 , we obtain @xmath49ergs@xmath34 , hence @xmath50ergs@xmath34 @xcite .
therefore , we adopted euv fluxes of 4330 and 1410 for corot-24b and corot-24c , respectively . for comparison , the x - ray luminosity of hd189733 , which has similar parameters to corot-24 is @xmath51ergs@xmath34 @xcite . by using instead the other value of @xmath47 of @xmath482.7 given by @xcite , we obtain @xmath52ergs@xmath34 .
the 1-d simulations also include the effects of ionization , dissociation and recombination .
the photoionization cross sections , the dissociation and recombination coefficients , the continuity equations for the number densities of the atomic hydrogen atoms and ions , and those for the neutral and ionized molecules were taken from @xcite and @xcite and scaled proportionally to the euv fluxes derived for each planet .
the electron density is determined from the quasi - neutrality condition .
the upper boundary of the simulation domain is set at the roche lobe of @xmath824 and @xmath871.7 for corot-24b and 24c , respectively .
the hydrodynamic model can be applied when enough collisions occur , which is the case if knudsen number @xmath53 @xcite , where @xmath54 is the mean free path and @xmath55 is the local scale height .
this knudsen criterion is fulfilled within the calculation domain .
for all calculations we considered a heating efficiency of 15% @xcite .
calculations have been done for two planetary radii and a planetary mass of 5.7 .
the points show the integral of the convolution of the two model curves with the transmission function of the corot filter at the center of the wavelength span covered by the filter , which is indicated by the horizontal bars.,width=302 ] we set the lower boundary of the simulations at @xmath56 , where we assume a temperature equal to the planetary equilibrium temperature . here
we also assume an h@xmath13 number density corresponding to an atmospheric pressure of 100 mbar , which is near the level where the tangential optical depth @xmath57=1 for most transparent continuum bands @xcite . to prove the validity of this assumption we use a radiative - transfer transmission - spectrum model for a solar composition atmosphere and the @xmath27-@xmath31 profiles obtained from the hydrodynamic code extended with isothermal hydrostatic profiles at high pressures , assuming two different planetary radii ( see sect .
[ sec : results ] ) .
the main contribution to the absorption comes from h@xmath13 rayleigh scattering ( @xmath5600 nm ) , alkali lines ( na and k ; 500800 nm ) , h@xmath13-h@xmath13 collision - induced absorption ( @xmath14600 nm ) , and h@xmath13o ( @xmath8900 nm ) .
corot s observing band covers the range 3501000 nm , with the effective wavelength at @xmath8690 nm .
figure [ fig : tau ] shows that @xmath57@xmath81 lies at a pressure of @xmath86070 mbar , but this is for a slant geometry ( transmission ) and therefore in a radial geometry the pressure at is slightly higher than the average pressure probed by transmission spectroscopy
. differences of a factor of 10 in metallicity do not affect this result .
we run a consistency check of the adopted hydrodynamic code by comparing the calculated h loss rates with those of the well studied hot - jupiter hd209458b , for which upper atmosphere hydrodynamic modelling has been performed by several groups .
if we assume similar planetary and atmospheric parameters at the lower boundary @xcite we obtain a mass - loss rate of @xmath87@xmath5810@xmath59gs@xmath34 , which agrees well with what previously derived by several other groups ( 47@xmath5810@xmath59gs@xmath34 ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
for the first set of calculations , we assumed that = = 3.7 for corot-24b and 4.9 for corot-24c .
the equilibrium temperatures of corot-24b and 24c are 1070@xmath9140k and 850@xmath980k , respectively @xcite . because these temperatures are cooler than the h@xmath13 thermal dissociation temperature ( @xmath82000k ) , h@xmath13 is the dominant species near the 100 mbar level . above the altitude level where dissociation takes place ,
h atoms become the main species and can escape from the planet s gravitational well .
the solutions of our model runs are quasi - stationary with slow time variations of the atmospheric parameters . for corot-24b and corot-24c
, the hydrodynamic model gives h escape rates of @xmath82.0@xmath5810@xmath60s@xmath34 ( 3.35@xmath5810@xmath61gs@xmath34 ) and @xmath88.0@xmath5810@xmath62s@xmath34 ( 1.34@xmath5810@xmath59gs@xmath34 ) , respectively . under the assumption that the 100 mbar level lies near the observed transit radius , the high thermally driven hydrodynamic outflow obtained for corot-24b
is caused by the combination of the planet s high temperature and low gravity , independent of the stellar euv flux .
this happens because for hot , low - gravity planets the thermal energy @xmath35 shown in eq . [ eq : thermalenergy ] for a large fraction of the particles in the upper atmosphere is of the order of or larger than the potential energy . as a consequence ,
the atmosphere flows upwards , even without the need of an additional external energy source , such as the stellar euv flux . in this case , the stellar euv flux is not the main driver of the atmospheric escape .
this is similar to the so called `` boil - off '' evaporation regime of very young planets @xcite .
.results of the hydrodynamic simulations for corot-24b with = 5.7 , = 1070k , and = r@xmath63 ranging between 1.7 and 3.7 .
the table lists the planetary densities @xmath64 in g@xmath11 , hydrodynamic @xmath65 and energy - limited @xmath66 escape rates in s@xmath34 and their ratio ( @xmath67 ) . [
cols="^,^,^,^,^",options="header " , ] the high escape rate obtained for corot-24b could in principle last for @xmath5100myr ( see sect .
[ sec : indentifying ] ) , which is inconsistent with the old age of the system : with the exception of very young planets , the escape rate can only be less than or equal to the maximum possible euv - driven h escape rate , which can be analytically estimated with the energy - limited formula ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , @xmath68 where @xmath69 is the heating efficiency , @xmath70 is the stellar euv flux , @xmath71 is the effective radius where the euv energy is absorbed in the upper atmosphere @xcite , and @xmath72 the mass of the h atom .
the factor @xmath73 takes into account roche lobe effects @xcite . by applying this formula to corot-24b and corot-24c
, we obtain escape rates of 5.3@xmath5810@xmath74s@xmath34 ( 8.9@xmath5810@xmath59gs@xmath34 ) and 5.5@xmath5810@xmath62s@xmath34 ( 9.2@xmath5810@xmath75gs@xmath34 ) , respectively . for the hotter and less massive sub - neptune corot-24b ,
the escape rate based on the hydrodynamic model @xmath65 is @xmath840 times higher than @xmath66 obtained from eq .
[ eq : energylimited ] , while for the more massive and cooler planet , corot-24c , @xmath65 is consistent with @xmath66 , within the expected uncertainties . from these estimates
it follows that the high escape rate obtained from the hydrodynamic simulations for corot-24b can not be realistic .
the only possible solution to this discrepancy is that is not close to the 100 mbar level , but lies much higher in the atmosphere . and corresponding energy - limited escape rates @xmath66 as a function of planet radius , considered to be at the 100 mbar level , and planet mass .
the dotted horizontal line indicates @xmath65=@xmath66 .
the vertical lines mark the transit radius , and the possible range of planet ( at the 100 mbar pressure level ) and core radii @xmath76 ( see sect .
[ sec : discussion ] ) assuming a mass of 5.7 .
the dashed line indicates the results obtained assuming a mass of 5.7 and a pressure at of 1bar.,width=321 ] to find out which set of planet parameters satisfies the @xmath67@xmath771 condition for corot-24b , we run a grid of hydrodynamic models with mass and radius values in the 4.08.0 and 1.73.7 range , respectively , to directly compare @xmath65 and @xmath66 as a function of planet parameters .
table [ tab : table ] shows the results of the hydrodynamic model for corot-24b with = 5.7 and ranging between 1.7 and 3.7 .
figure [ fig : lhyvslen ] shows the @xmath67 ratio as a function of planetary radius and mass and indicates that , assuming @xmath775.7 , the condition @xmath67@xmath771 is reached for planetary radii smaller than @xmath82.2 . in this work , the transit radius of corot-24b is the main observational uncertainty , but it has however no impact on the results : the @xmath67@xmath771 relation is fulfilled for @xmath52.2 , regardless of the observed radius and its uncertainty .
our simulations at larger masses show that the transit radius lies close to the 100 mbar level for @xmath148 . according to formation and structure models of hot ( @xmath81000k ) sub - neptunes ( e.g. , * ? ? ?
* ) , the transit radius would correspond to an atmospheric mass fraction @xmath78 of @xmath80.04 , hence an atmospheric mass @xmath79 of @xmath81.36@xmath5810@xmath80 g . by assuming a core mass of 5.7 for corot-24b , from the escape rates given in table 1 , one can also roughly estimate the evolution of atmospheric mass with time .
atmospheric masses corresponding to the radii given in table 1 were estimated from the 1000k structure models of @xcite .
as one can see in fig .
[ fig : evolution ] , an atmospheric mass above 2.2 corresponding to a radius of 3.7 would be lost within 100myr due to the high thermally - driven escape rates , in agreement with the results of @xcite .
because of the short time period in which the mass - loss rates are exceptionally high and the corresponding fast shrinking of the planet radius , the 100 mbar level of corot-24b must be located much deeper in the atmosphere , compared to the position of .
evolution normalized to the atmospheric mass corresponding to @xmath81 estimated from the results shown in table 1 .
the horizontal line indicates @xmath79=@xmath82.,width=321 ] thermally driven hydrodynamical mass loss could occur at the first stages of planetary evolution @xcite , but is unlikely to occur for older planets because , as shown in fig .
[ fig : evolution ] , the atmosphere structure would be modified in a short time .
it is also not possible that corot-24b is the remnant of a more massive planet ( e.g. , hot - jupiter or hot - saturn ) , because the escape rates of such massive objects are not high enough to affect significantly the planet s mass over the main - sequence lifetime of the host star @xcite .
both escape rates , i.e. @xmath65 and @xmath66 , depend on the adopted euv flux and heating efficiency @xmath69 .
for this reason we tested the effect of a higher / lower euv flux or @xmath69 on the results .
a different euv flux affects both @xmath65 and @xmath66 rates in a similar , though not identical , way .
a variation of a factor of @xmath82 in the euv flux would lead to a variation on the @xmath67 ratio of @xmath80.3 , which in terms of planetary radius corresponds to @xmath50.1 .
variations in @xmath69 also affect both escape rates . in this case
we estimate that a factor of @xmath82 difference in the heating efficiency would lead to a variation on the planetary radius of @xmath50.1 . from the comparison of the escape rates of the hot - jupiter hd209458b described in sect .
[ sec : modelling ] we expect that the use of a different hydrodynamic code would lead to mass - loss rates consistent with those presented here within a factor of @xmath82 , which , in terms of the derived planetary radius of corot-24b , corresponds to @xmath80.1 .
we further tested the effect of varying ( 1070@xmath9140k ; * ? ? ? * ) , finding that variations in up to 200k have a negligible impact on the results .
on the basis of planetary formation models calculated for planets in the 110 range , one can estimate the core radius @xmath76 of a planet on the basis of its mass if @xmath83 as @xmath84 , depending upon the exact composition @xcite .
taking into account that , hence the 100 mbar level , can not be too close to @xmath76 ( the planet needs sufficient atmosphere to produce the observed ) , from fig .
[ fig : lhyvslen ] it follows that corot-24b should have a minimum mass of @xmath85.0 , with a core radius in the @xmath81.551.68 range and at the 100 mbar level in the @xmath81.92.2 range .
these values yield an average density of 2.94.6g@xmath11 .
the mass contained between and is @xmath85 of .
we conclude therefore that , accounting for the uncertainties on , the `` true '' radius @xmath86 of corot-24b is @xmath830 - 60% smaller than the one measured by the transit . for the solar system gas giant planets ,
is by definition set at the 1bar pressure level . by setting this pressure level at and following the same procedure described in sect .
[ sec : indentifying ] , we obtained again that would be about 2.2 ( see fig . [ fig : lhyvslen ] ) .
figure [ fig : profiles ] compares the temperature and pressure profiles of corot-24b , considering a of 2.2 and a of 5.7 , and corot-24c . in the case of corot-24b ,
the atmosphere is hydrostatic near , while it becomes hydrodynamic and cools adiabatically above 5 . for the cooler and more massive planet , corot-24c
, the atmosphere is hydrostatic above and the temperature increases up to 3000k due to the euv heating . above 10
the atmosphere cools adiabatically . for corot-24b ,
[ fig : profiles ] shows that assuming = 2.2 and = 5.7 , lies near the 110@xmath87bar pressure level .
as one can see from fig .
[ fig : tau ] , the absorption at the level of the transit radius can not be caused by h@xmath13 rayleigh scattering and the alkali lines are too narrow to produce broad - band absorption . at such a high altitude , absorption could be for example produced by clouds , though cloud formation models based on tio@xmath13 calculated for hd189733b ( similar to that of corot-24b ) suggest formation pressures of about 0.1 mbar @xcite .
by comparing the escape rates obtained from a hydrodynamic code with the maximum possible euv - driven escape rate calculated from the energy - limited formula , we discovered that for the h@xmath88-dominated close - in sub - neptune corot-24b the planetary radius , by definition set at the 100 mbar pressure level , is 30 - 60% smaller than the observed transit radius .
additional considerations on the range of possible planetary core radii allowed us to constrain the most likely values of the planetary mass and radius to lie in the range 5.05.7 and 1.92.2 , respectively .
the kepler satellite has discovered several similar , low - density `` sub - neptunes '' and it is very likely that the transit radii measured for the vast majority of these planets also differs from the true planetary radius , therefore introducing a systematic bias in the measured radii .
this has dramatic implications for example in the determination of the mass - radius relation ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and for planet population synthesis studies ( e.g. , * ? ? ?
our results may also help explaining some of the discrepancies discovered between the planet radius distribution observed by kepler and that derived from population synthesis models @xcite .
this has also profound implications for our understanding of the structure of the large number of transiting low - mass exoplanets expected to be observed and discovered in the near future by the cheops , tess , and plato space missions and characterized by jwst .
hl and ij acknowledge the funding by ffg-847963 . tl acknowledges funding by ffg - asap11 .
the authors thank the fwf nfn project s11601-n16 and the subprojects s11604-n16 and s11607-n16 .
hl , nve , and po acknowledge the fwf project p27256-n27 .
nve acknowledges support by the rfbr grant no.15 - 05 - 00879-a .
the authors thank the anonymous referee for the useful comments that led to a considerable improvement of the manuscript .
we thank r. alonso for fruitful discussions . | for the hot exoplanets corot-24b and corot-24c , observations have provided transit radii of @xmath0 and @xmath1 , and masses of @xmath2 and @xmath3 , respectively .
we study their upper atmosphere structure and escape applying an hydrodynamic model . assuming @xmath4 , where is the planetary radius at the pressure of 100 mbar
, we obtained for corot-24b unrealistically high thermally - driven hydrodynamic escape rates .
this is due to the planet s high temperature and low gravity , independent of the stellar euv flux .
such high escape rates could last only for @xmath5100myr , while shrinks till the escape rate becomes less than or equal to the maximum possible euv - driven escape rate .
for corot-24b , must be therefore located at @xmath6 and high altitude hazes / clouds possibly extinct the light at .
our analysis constraints also the planet s mass to be @xmath7 . for corot-24c , and
lie too close together to be distinguished in the same way .
similar differences between and may be present also for other hot , low - density sub - neptunes .
[ firstpage ] planets and satellites : atmospheres stars : ultraviolet hydrodynamics |
sr@xmath0yruo@xmath1 belongs to the family of double perovskite insulators , sr@xmath12ruo@xmath1 ( @xmath13 = rare earth or y ) @xcite , where the ru ions exist in the pentavalent state ( ru@xmath4 ) with a high - spin state @xmath14 and @xmath15 configuration ( @xmath16)*. * even though the structure of these compounds can be derived from the well known perovskite structure of srruo@xmath17 by replacing alternate ru ions with @xmath13 ions @xcite , these compounds do not show any similarity to their parent compound srruo@xmath17 , which is a ferromagnetic metal .
the layered structure , consisting of alternate @xmath13ruo@xmath18 and sro planes , accommodates both the ru and rare earth atoms in the same @xmath13ruo@xmath18 plane and hence both the atoms share the same site symmetry ( @xmath19-site of the perovskite structure abo@xmath17 ) .
the alternating positions of the ru and @xmath13 atoms in the unit cell result in two type of interactions between the ru atoms ; ( i ) direct interaction of ru - o - o - ru and ( ii ) indirect interaction through the rare earth atoms , ru - o-@xmath13-o - ru . since
the compounds having nonmagnetic @xmath13 ions ( y and lu ) are also found to order magnetically @xcite , the direct interaction is assumed to be stronger than the indirect interaction through the rare earth atoms . among the sr@xmath12ruo@xmath1 compounds , sr@xmath0yruo@xmath1 has captured additional interest due to the occurrence of superconductivity when ru is partially ( @xmath20 15% ) replaced by cu @xcite .
cu is found to get substituted at the ru site in the yruo@xmath18 planes and thus the structure of the substituted compounds remains the same as that of the parent compound , without creating any additional cu - o planes @xcite .
the parent compound sr@xmath0yruo@xmath1 is known to be an antiferromagnetic insulator with the ru moments ordering at @xmath21 k @xcite .
the magnetic ordering temperature ( @xmath22 ) was inferred as 26 k from the position of the peak in the magnetization measurements .
neutron diffraction measurements at 4.2 k have confirmed the magnetic ordering of the ru moments , consisting of a type i afm structure . due to the monoclinic distortion of the structure ,
the compound is expected to show canting of the ru moments resulting from the dzyaloshinsky - moriya ( d - m ) interactions @xcite among the antiferromagnetically ordered spins .
how such a compound becomes a metallic magnetic superconductor without creating cu - o planes is still a puzzling question .
there are still many unanswered questions regarding the origin of magnetism and superconductivity in the cu - substituted sr@xmath0yruo@xmath1 compounds . at the same time , there are no detailed magnetization studies available for the parent compound itself , except for one report on sr@xmath0yruo@xmath1 single crystals @xcite which confirms the magnetic ordering and weak ferromagnetism .
in addition , the resistivity of sr@xmath0yruo@xmath1 single crystals @xcite shows anomalous behaviour below @xmath22 followed by a mott - like transition at 17k whereas the magnetoresistance becomes negative below 30k .
the band structure calculations @xcite have indicated the competition between the antiferromagnetic and ferromagnetic fluctuations among the ru moments .
we present here some additional evidence for the competition between antiferromagnetic and ferromagnetic coupling in this compound .
detailed measurements of magnetization and heat capacity show some anomalous properties exhibited by sr@xmath0yruo@xmath1 .
both the measurements unfold clear evidence for two magnetic orderings ( @xmath2 k and @xmath3 k ) , even though the magnetic ordering in this compound can occur only by ru moments .
the magnetization measurements corroborate that both the magnetic ordering occurs with ferromagnetic components and these two components align opposite to each other with respect to the magnetic field direction , resulting in a magnetization reversal .
the results presented here have relevance to the magnetic properties exhibited by the cu - substituted superconducting sr@xmath0yru@xmath10cu@xmath11o@xmath1 compounds @xcite .
yruo@xmath1 in zero field - cooled ( zfc ) and field - cooled ( fc ) modes under various applied fields . ]
polycrystalline samples of sr@xmath0yruo@xmath1 were prepared by the standard solid state reaction method by mixing stoichiometric amounts of srco@xmath17 , y@xmath0o@xmath17 and ru metal powder and heating at 960@xmath23c for 12 hours .
the final sintering of the pelletized powder was carried out at 1360@xmath23c for 24 hours after several intermediate heat treatments followed by grindings .
x - ray diffraction pattern of the samples was recorded on an xpert pro diffractometer ( panalytical , holland ) .
the magnetization as a function of temperature and magnetic field was measured using a vibrating sample magnetometer ( quantum design , usa ) .
the heat capacity measurements using the relaxation method were performed using a physical property measurement system ( quantum design , usa ) in the temperature range 1.8 - 300 k.
the rietveld analyses of the x - ray diffraction patterns using fullprof software showed that the compound forms in single phase with a monoclinic structure ( space group @xmath24 ) .
the lattice parameters obtained from the analyses are , @xmath25 , @xmath26 and @xmath27 along with @xmath28 which are in good agreement with those reported earlier @xcite .
figure 1 illustrates the magnetization of sr@xmath0yruo@xmath1 as a function of temperature in zero field - cooled ( zfc ) and field - cooled ( fc ) modes . in the zfc measurements the sample
was cooled in zero applied field to 2 k , the required magnetic field was applied and then the data were taken while increasing the temperature . for the fc measurements ,
the sample was cooled from the paramagnetic state to 2 k in an applied field and the data were recorded while heating the sample . in order to minimize the remnant field in the superconducting magnet before the zfc measurements , the magnetic field was reduced to zero from a large field value in the oscillating mode .
this made sure that the remnant field was within @xmath29 oe .
the lower panel shows the zfc measurements for various applied field values . for low field values ,
the magnetization is negative at lower temperatures . as the temperature
is increased , the magnetization remains independent of temperature till @xmath30 k and then surprisingly decreases to go through a minimum . as the temperature
is further increased , the magnetization increases , goes through a positive maximum and then shows the normal paramagnetic behaviour . for @xmath31 koe ,
the zfc magnetization starts with a positive value at low temperatures , but goes through negative value at the minimum . for higher fields ,
the magnetization is always positive , even though it goes through a minimum .
the width of the peaks at the maximum and minimum as well as the temperature at which they occur depends slightly on the applied fields ; both decrease with increasing field .
the upper panel of fig .
1 shows the fc measurements at various applied fields .
the fc magnetizations show a broad peak and the temperature at which the peak occurs shows a weak temperature dependence on the applied fields . in order to ascertain that the anomalies observed in the zfc magnetization is not entirely due to effects of negative remnant magnetic field in the superconducting coils , we have carried out fc measurements in smaller field values , both positive and negative .
figure 2 shows the fc measurements for an applied field of @xmath32 oe .
it is clear that the fc magnetization remains negative whether the field is positive or negative .
such effects are seen upto 25 oe above which the fc curves switch over to the positive side .
neutron diffraction studies at 4.2 k @xcite had indicated only an afm ordering of the ru moments .
it was also proposed that the distorted monoclinic structure can give rise to a small canting of the ru moments and hence a small ferromagnetic component in this compound resulting from the d - m interactions between the antiferromagnetically ordered ru moments .
this , however , can not explain the observed magnetization behaviour in this compound .
a simple ferromagnetic component due to canting can make the magnetization negative in the zfc mode if the remnant field is negative .
but then the magnetization will monotonically decrease as the temperature is increased and will cross over to the positive side before completing the magnetic order .
a typical example for such a behaviour is shown in the inset of fig . 2 for mnco@xmath17 , which is a well known canted antiferromagnet having d - m interactions @xcite .
we have also made sure that the compound does not contain any srruo@xmath17 impurities ( not detected in x - ray diffraction patterns ) by taking the zfc and fc data for small field values in the temperature range 100 - 200 k ( even a small trace of srruo@xmath17 impurity will give a thermal hysteresis around its ferromagnetic ordering temperature ( 150 - 160 k ) between the zfc and fc measurements ) . in order to further corroborate that there exists more than a simple canting in this compound
, we have carried out detailed magnetization measurements as a function of magnetic field at different temperatures .
figure 3 depicts the low field magnetic isotherms at some selected temperatures . at low temperatures , well below the magnetic anomalies ,
the magnetization curves are almost linear with a small value for coercivity ( @xmath7 ) .
as the temperature crosses 22 k , the magnetization shows significant hysteresis and the magnetization loops open up .
the opening of the loop increases until the temperature reaches @xmath33 k and then decreases as the temperature is further increased . even at 32 k ,
the hysteresis is much more than the same at 5 k. at 35 k , we see only a linear behaviour expected for a paramagnet . even though the hysteresis loops are not closed at some temperatures ( fig .
3(d ) , 3(e ) ) , they show a normal behaviour when the applied fields are extended to higher values ( see main panel of fig . 4 ) .
no other anomalies are observed in the high field magnetization curves .
the coercivity and the remnant magnetization plotted as a function of temperature in the inset of fig .
4 show a maximum near 27 k and decrease on either side of this temperature .
this clearly demonstrates that some sort of magnetization reversal happens at 27 k. there are no reports about the heat capacity of this compound in the literature .
the result of our heat capacity measurements for sr@xmath0yruo@xmath1 is presented in fig .
two peaks are obvious , one at @xmath34 k and the other at @xmath35 k , which correspond well to the anomalies observed in the magnetization .
there is only a minor effect by the magnetic field on the heat capacity of the sample even at 50koe ( fig .
5(a ) ) , even though a small decrease in the temperature dependence of the peak positions were observed in the magnetization measurements .
in order to have an estimate of the approximate magnetic heat capacity , the phonon contribution needs to be subtracted from the total measured heat capacity .
since there are no nonmagnetic analogues available for this compound , the phonon contribution was calculated from the combined debye and einstein equations @xcite , where @xmath37 and @xmath38 are the anharmonicity coefficients , @xmath39 is the debye temperature , @xmath40 is the einstein temperature , @xmath41 and @xmath42 .
the best possible fit was obtained when the calculations were performed by using one debye and three einstein frequencies along with a single @xmath37 . the solid line in fig .
5(b ) represents the fit to the phonon contribution , which is in good agreement with the experimental data at high temperatures ( above the magnetic ordering ) .
the parameters obtained from the best fit are : @xmath43 k , @xmath44 k , @xmath45 k , @xmath46 k , @xmath47 k and @xmath48 k. the debye and einstein temperatures obtained for sr@xmath0yruo@xmath1 are comparable with those obtained for yvo@xmath17 where the phonon contribution was obtained in a similar fashion , but with only two einstein frequencies along with one debye frequency @xcite . the magnetic heat capacity , @xmath49 , obtained by subtracting the calculated phonon heat capacity from the total heat capacity , is shown in the fig .
5(c ) along with the magnetic entropy ( @xmath50 ) .
the two peaks become more obvious in the magnetic heat capacity . in sr@xmath0yruo@xmath1
, the magnetic transition can occur only due to the ordering of ru@xmath4 moments .
in that case , the exact reason for the observed double peak behaviour is not very clear at present .
magnetic entropy ( @xmath51 ) increases with temperature and saturates to a value of @xmath52 2.6 j mole@xmath53 k@xmath53 above 30 k. if we consider the ground state of ru@xmath4 ions as @xmath16 , then the expected magnetic entropy is 11.52 j mole@xmath53 k@xmath53 ( @xmath54 $ ] ) , corresponding to the four - fold degenerate ground sate .
however , the crystalline electric fields , if present , can split this ground state into two doubly degenerate states , giving rise to a ground state multiplicity of only two @xcite .
this will reduce the magnetic entropy of the compound to 5.76 j mole@xmath53 k@xmath53 ( @xmath55 ) .
the observed entropy , however , is even less than half of this value .
in fact , neutron diffraction measurements had estimated a value of 1.8 @xmath56/ru@xmath4 at 4 k ( instead of the expected value of 3 @xmath56/ru@xmath4 ) in the magnetically ordered state @xcite .
this moment value corresponds well with the doubly degenerate ground state .
if we compare the reduction in entropy of sr@xmath0yruo@xmath1 to that observed for yvo@xmath17 @xcite , frustrations of ru spins at high temperatures ( above the magnetic ordering ) can be attributed as the reason for the reduction in entropy .
the correlation between the frustrated moments at high temperatures reduces the contribution of the entropy to the magnetic ordering .
such a frustration among the ru moments is inferred in sr@xmath0yruo@xmath1 as the possible reason for the reduction in @xmath22 even though the compound possesses a large exchange integral value @xcite . ) of sr@xmath0yruo@xmath1 in applied fields @xmath57 oe and 5 koe .
( b ) heat capacity with the fit ( solid line ) for phonon contribution .
( c ) magnetic contribution to heat capacity ( @xmath49 ) ( left scale ) and calculated magnetic entropy ( @xmath51 ) ( right scale ) as a function of temperature . ]
it is clear that sr@xmath0yruo@xmath1 exhibits two anomalies , the first at @xmath58 k ( @xmath59 ) and the second at @xmath33 k ( @xmath60 ) even though the magnetic ordering in this compound can come only from the ru@xmath4 moments .
if we assume that the two anomalies are associated with the magnetic ordering of the ru moments , then the observed behaviour is very interesting . the isothermal magnetization curves at different temperatures ( fig .
3 ) clearly demonstrate that the first magnetic order starts at @xmath2 k with a ferromagnetic component resulting in the increase of hysteresis and @xmath7 as the temperature is lowered .
this ferromagnetic component is expected from the canting of the antiferromagnetically ordered ru moments because of the d - m interactions .
however , the decrease of hysteresis and @xmath7 below 27 k indicates that a second component of the magnetic order also develops with a ferromagnetic component ( @xmath60 ) , but aligns itself opposite to the first component and hence opposite to the applied field .
this component almost cancels the first component and hence the hysteresis is negligible at low temperatures ( @xmath61 k ) .
these anomalies are further confirmed in the zero field remnant magnetization measurements as shown in fig .
6(a ) . here
the sample was cooled ( fc ) in a field of 5 koe down to 10 k. the field was then removed and the remnant magnetization was measured in zero field while warming the sample .
the remnant magnetization shows a normal decrease upto @xmath30 k , but then increases , goes through a maximum at @xmath33 k and then decreases to zero above 32 k. this clearly demonstrates that the magnetic ordering consists of two components and they are aligned opposite to each other with respect to the magnetic field direction . while cooling the sample in magnetic field , the first component orders and aligns parallel to the field at @xmath58 k , but the second component aligns antiparallel to the field at @xmath33 k , decreasing the net magnetization .
however , this antiparallel component is not strong enough to make the magnetization negative as in the case of some @xmath13vo@xmath17 compounds .
as the sample is warmed up in zero field , the remnant magnetization increases when the antiparallel component relaxes and completes its disordering . in order to verify the thermodynamic reversibility of the fc state
, we have measured the remnant magnetization by field - cooling the sample down to two pre - selected temperatures ( @xmath63 and @xmath64 ) on either side of the maximum in the fc magnetization curve as shown in fig .
6(a ) . in the first case ,
the sample was field cooled to 28 k ( @xmath65 ) , where only the first magnetic component would have ordered , and the field was removed . in the second case , the sample was field cooled to 23.5 k ( @xmath66 ) , where both components would have undergone magnetic ordering , before removing the field . in both cases ,
the remnant magnetization follows exactly the original remnant curve which was obtained by switching the field off at 10 k , proving the thermodynamic reversibility of the magnetically ordered states .
further evidence for the opposite alignments of the two components of magnetic ordering is evident from the data in fig .
6(b ) . here
the sample was cooled in a negative field ( @xmath67 oe ) so that the magnetization at 5 k is negative . at 5 k
, the field was increased in the positive direction ( @xmath68 koe ) until the magnetization became positive .
magnetization shows normal behaviour upto 20 k , but shows a sudden dip to go through a negative minimum at @xmath5227 k.
thus it is clear that whether the sample is cooled in positive or negative field , the two components of magnetic order align always opposite to each other .
neutron diffraction measurements in this compound at 4.2 k * * @xcite had indicated that the nuclear structure of the compound remains unchanged at 4.2 k , ruling out the possibility of any structural changes . on the other hand if we assume that one of the anomalies is due to a possible structural change , then the changed structure should again revert back to the original structure at low temperatures .
from the position of the magnetic diffraction peaks , magnetic ordering of the ru moments was deduced to be antiferromagnetic with a type i structure @xcite .
however , the temperature variation of the intensity of magnetic peak(s ) was not reported and hence the exact temperature at which the ru moments order is not available .
the @xmath22 of @xmath69 k was assigned to this compound simply from the position of the peak in the high field magnetization measurements @xcite .
even though there are no detailed experimental observations in sr@xmath0yruo@xmath1 , many experimental results including neutron diffraction exist for cu - substituted sr@xmath0yru@xmath10cu@xmath11o@xmath1 compounds * * @xcite . assuming that the magnetic properties associated with the ordering of the ru moments are not drastically altered
, we can analyze the results of the @xmath70sr measurements in the cu - substituted compounds @xcite . both the precession frequency and relaxation rate show anomalies at @xmath71 k for muons trapped in the two possible sites ; oxygen in the yruo@xmath18 layers ( @xmath72 ) and oxygen in the sro layers ( @xmath73 ) .
the authors assigned this anomaly to the fluctuation of the ru moments which order in a spin - glass state .
this is exactly the same temperature range at which the first component of magnetic ordering starts in our studies for the parent compound ( @xmath59 ) .
the variation of the power exponent at the @xmath72 site as a function of temperature @xcite almost resembles the mirror image of our temperature variation of the coercivity and remnant magnetization ( inset of fig . 4 ) .
it was also proposed that the fluctuations of the ru moments stop at @xmath74 k and order antiferromagnetically .
we have identified this temperature in our studies as the temperature at which the second component of magnetic moments completes its ordering .
@xmath75ru mssbauer measurements in sr@xmath0yru@xmath10cu@xmath11o@xmath1 ( @xmath76 ) @xcite have confirmed the magnetic ordering of ru moments .
both the isomer shift and the hyperfine field values are consistent with the pentavalent nature of the ru moments . from the temperature variation of the mssbauer spectra @xcite ,
it was concluded that that the magnetic ordering of the ru moments persists upto 30 k , which is consistent with the magnetic ordering temperature @xmath59 in our measurements
. additional support for the ru ordering as high as 30k comes from the neutron diffraction measurements reported for sr@xmath0yru@xmath10cu@xmath11o@xmath1 ( @xmath77 ) where the intensity of the magnetic peak due to the ordering is ru moments is evident upto @xmath71 k @xcite . in zfc measurements ,
the sample is always cooled in a nominal remnant field ( @xmath78 oe in our case ) .
this may result in the negative magnetization at low temperatures .
when the magnetic field is applied , the magnetization remains negative for low fields ( fig . 1 ) but changes over to positive values for sufficiently large fields .
even then a magnetization reversal happens and the magnetization goes though a negative minimum in between the two transitions .
this behaviour may be caused by the fact that the large fields flip ( to orient parallel to the field ) some , but not all , of the spins oriented against the field .
as the field value increases , majority of the spins , oriented antiparallel to the field , align parallel to the field , resulting only in a small minimum at @xmath60 . in fc magnetization
( @xmath79 oe ) , the first component of the magnetic transition aligns along the field at @xmath59 , resulting in a positive magnetization .
however , at @xmath60 , the second component aligns antiparallel to the first component , resulting in the reduction in magnetization .
when this alignment is over , the magnetization remains constant , as in the case of fc measurements in smaller fields .
the reason for the two components of the magnetic ordering reminiscent of ferrimagnetic ordering in this compound is not clear . in
the ordered double perovskite sr@xmath12ruo@xmath1 compounds , the @xmath19-site of the perovskite structure abo@xmath17 is uniquely occupied either by ru or the rare earth metal ions due to the lower coordination number compared to sr . since ruthenium is considered to be in the oxidation state of 5 + in these compounds , the chances of ferrimagnetic ordering of ru@xmath4/ru@xmath80 moments are very rare .
we have repeated the measurements with samples annealed in air , oxygen or argon and all of them showed the same behaviour .
in fact , the isomer shift values from the @xmath81ru mssbauer measurements in sr@xmath0yruo@xmath1 had confirmed the pentavalent oxidation state of ru moments @xcite . even if the spin - glass state is assumed as suggested by harshman et al @xcite , the observed properties - the magnetization reversal , thermodynamic reversibility of magnetization and two peaks in heat capacity - can not be explained .
the magnetic interactions among the ordered ru moments can take place in two ways ; ( i ) the @xmath82-super exchange interaction between nearest neighbour ( nn ) ru@xmath4 ions via ru - o - o - ru pathway and ( ii ) the @xmath83-super exchange between the next nearest neighbour ( nnn ) ru@xmath4 ions via ru - o - y - o - ru pathway .
the relative strengths of these two interactions will determine the type of magnetic ordering at low temperatures .
since y is a nonmagnetic ion , it is not expected to take part in the exchange interaction and hence the second interaction between the nnn is assumed to be negligible .
this assumption can further be supported by the fact that the relaxation rate at the @xmath73 site for the muons is very much smaller than the relaxation rate at the @xmath72 site @xcite since the ru - o - y - o - ru pathway includes the oxygen at the o(3 ) sites only .
whether the competition between these two interactions , however , can give rise to the observed double peak behaviour needs further investigations .
it is possible that the stronger interaction orders the ru moments at @xmath59 and the weaker interaction realigns some of the ru moments in the opposite direction at @xmath60 .
it is possible that the second component of the magnetic order at @xmath60 is not strong enough to bring the magnetization down to negative values even though it gets reduced .
however , why the second component aligns the moments against the first component and hence the magnetic field is not clear now .
detailed neutron diffraction measurements are needed as a function of temperature to explain the reason for the observed anomalies . in conclusion ,
we have reported some of the anomalous properties exhibited by sryruo@xmath1 deduced from the detailed magnetic and heat capacity measurements .
two distinct magnetic orderings are evident in the compound even though only ru moments can order in this compound .
the two components of the magnetic order align always opposite to each other and to the magnetic field .
observation of hysteresis and coercivity in magnetic isotherms below the magnetic order indicates the presence of ferromagnetic component associated with the magnetic ordering .
the presence of two well defined peaks in the heat capacity indicates that the two magnetic components have large entropy change associated with the ordering .
blackstead , j. d. dow , d.r .
harshman1 , m.j .
demarco , m.k.wu , d. y. chen , f. z. chien , d. b. pulling , w. j. kossler , a. j. greer , c. e. stronach , e. koster , b. hitti , m. haka and s. toorongian , eur .
j. b * 15 * , 649 ( 2000 ) . | anomalous magnetic properties of the double perovskite ruthenate compound sr@xmath0yruo@xmath1 are reported here .
magnetization measurements as a function of temperature in low magnetic fields show clear evidence for two components of magnetic order ( @xmath2 k and @xmath3 k ) aligned opposite to each other with respect to the magnetic field direction even though only ru@xmath4 moments can order magnetically in this compound .
the second component of the magnetic order at @xmath3 k results only in a magnetization reversal , and not in the negative magnetization when the magnetization is measured in the field cooled ( fc ) mode .
isothermal magnetization ( @xmath5-@xmath6 ) measurements show hysteresis with maximum coercivity ( @xmath7 ) and remnant magnetization ( @xmath8 ) at @xmath927 k , corroborating the presence of the two oppositely aligned magnetic moments , each with a ferromagnetic component .
the two components of magnetic ordering are further confirmed by the double peak structure in the heat capacity measurements .
these anomalous properties have significance to some of the earlier results obtained for the cu - substituted superconducting sr@xmath0yru@xmath10cu@xmath11o@xmath1 compounds . |
actinide based materials possess interesting physical behaviors due to the existence of strongly - correlated 5@xmath2 electrons and have attracted extensive attentions @xcite . different from actinide oxides which have been widely studied to reveal the detailed configurations and correlation effects of the 5@xmath2 electrons @xcite , none of the actinide chalcogenides has ever received comparable concerns . here
we take us@xmath1 and use@xmath1 as two representatives to study the electron correlation and spin - orbit coupling ( soc ) effects of 5@xmath2 electrons in actinide chalcogenides .
the other reason for us to investigate their electronic structures is that they employ the layered mx@xmath1 structure ( with m to be a metal element , and x to be s , se , or te ) , which can be used as models for studying electronic behaviors in 1-dimensional ( 1d ) systems .
the mx@xmath1 structure belongs to the space group of p2@xmath3/@xmath4 , with its atomic organizations depicted in fig .
1(a ) . in the monoclinic lattice , the top two chalcogen ( s , se or te ) atoms form a tightly bound pair suggesting that the mx@xmath1 compounds may be regarded as mx(x@xmath5 ) @xcite . the top x - x pair together with an underneath m and a more underneath x atom form a triangular unit , which repeats along the @xmath6 lattice direction forming a 1d chain , and two such triangle units form a unit cell of the mx@xmath1 compounds in the ( @xmath7,@xmath8 ) plane . in this way
, the mx@xmath1 compounds are always considered as 1d materials . for example , the metallic zrte@xmath1 and nbse@xmath1 exhibiting charge - density - wave transitions have been thoroughly investigated in angle - resolved photoelectron spectroscopy ( arpes ) and optical experiments @xcite . in the present paper , by contrast , we address the electronic structures of us@xmath1 and use@xmath1 , two semiconducting members of the mx@xmath1 family . in studies of actinide based materials , one has to be careful for the electron correlation and soc effects of the actinide 5@xmath2 electrons .
as an example , for actinide dioxides conventional density functional theory ( dft ) schemes that apply the local density approximation ( lda ) or the generalized gradient approximation ( gga ) underestimate the strong on - site coulomb repulsion of the 5@xmath2 electrons and consequently fail to capture the insulating properties @xcite . several approaches , the lda / gga+@xmath0 , the hybrid density functional of ( heyd , scuseria , and enzerhof ) hse ,
the self - interaction corrected local spin - density ( sic - lsd ) , and the dynamical mean - field theory ( dmft ) , have been developed to correct the pure lda / gga failures in calculations of actinide materials . among them
the effective modification of pure dft by lda / gga+@xmath0 formalisms has been widely used in theoretical studies of uo@xmath5 @xcite and puo@xmath5 @xcite .
the obtained structural parameters as well as the electronic structure and phonon dispersion curves @xcite accord well with experiments . in our present work , the gga+@xmath0 schemes due to dudarev _
_ @xcite are employed to study the electron correlation and spin - orbit coupling effects in us@xmath1 and use@xmath1 , as well as the two materials electronic , mechanical , and raman properties .
the rest of the paper is organized as follows . in sec .
ii the computational method is briefly described . in sec .
iii we present the results of the physical properties of us@xmath1 and use@xmath1 , and discuss the electron correlation and soc effects of the uranium 5@xmath2 electrons .
finally in sec .
iv , we close our paper with a summary of our main results .
our total - energy calculations are carried out by employing the plane - wave basis pseudopotential method as implemented in vienna ab initio simulation package ( vasp ) @xcite .
the exchange and correlation effects are described with the gga approximation in the perdew - burke - ernzerhof ( pbe ) form @xcite .
the projected augmented wave ( paw ) method of blchl @xcite is employed with the frozen - core approximation .
electron wave function is expanded in plane waves up to a cutoff energy of 400 ev and all atoms are fully relaxed until the hellmann - feynman forces on them are less than 0.01 ev / .
a 7@xmath99@xmath91 monkhorst - pack @xcite k - point mesh is employed for integration over the brillouin zone of us@xmath1 and use@xmath1 .
the uranium 6@xmath106@xmath115@xmath126@xmath137@xmath10 , sulphur 3@xmath103@xmath11 , and selenium 4@xmath104@xmath11 electrons are treated as valence electrons .
noncollinear calculations are used when considering the spin - orbital coupling effects . the strong on - site coulomb repulsions among the localized uranium
5@xmath2 electrons are described by using the formalism formulated by dudarev et al .
@xcite . in this scheme ,
the total gga energy functional is of the form @xmath14,\ ] ] where @xmath15 is the density matrix of @xmath2 states with spin @xmath16 , while @xmath0 and @xmath17 are the spherically averaged screened coulomb energy and the exchange energy , respectively . in this paper , the coulomb @xmath0 is treated as a variable , while the exchange energy is set to be a constant @xmath17=0.51 ev .
this value of @xmath17 is in the ball park of the commonly accepted one for uranium compounds @xcite and close to the theoretically predicted value of 0.54 ev in uo@xmath5 @xcite .
since only the difference between @xmath0 and @xmath17 is significant , we will henceforth label them as one single parameter @xmath18=@xmath0-@xmath17 , while keeping in mind that the nonzero @xmath17 has been used during calculations .
the phonon frequencies at the gamma point for us@xmath1 and use@xmath1 are calculated by using the density functional perturbation theory ( dfpt ) . and
their raman - active frequencies are obtained through symmetry analysis on the corresponding vibration modes . before dfpt calculations
, the lattice constants and atomic positions of us@xmath1 and use@xmath1 are further optimized using a denser @xmath19-point mesh of 9@xmath913@xmath93 , and a finer force convergence criteria of 0.001 ev / .
a 2@xmath92@xmath91 supercell is subsequently used for dfpt calculations . since both us@xmath1 and use@xmath1
are stacked along the @xmath8 direction through van der waals interactions , and with relatively large lattice constants , we do not extend the supercell along the @xmath8 direction . during dfpt calculations on the 2@xmath92@xmath91 supercell ,
the @xmath19-point mesh is set to be 5@xmath97@xmath95 , and the energy convergence criteria is set to be 1e-6 ev .
experimentally , the magnetic orderings of uranium trichalcogenides were studied ever since 1961 @xcite .
the relatively large atomic distances between neighboring u atoms in us@xmath1 and use@xmath1 suggest that magnetic orderings may occur in these compounds through super - exchange interactions of uranium 5@xmath2 electrons via the s or se ions @xcite . and
different measurements indicate that both us@xmath1 and use@xmath1 crystals undergo afm transitions at very low temperatures , with the magnetic moments of the two uranium atoms within each unit cell opposite to each other @xcite . in the present gga+@xmath0 study , we consider the nonmagnetic ( nm ) , ferromagnetic ( fm ) , and antiferromagnetic ( afm ) phases for each choice of the @xmath18 value and then determine the lowest - energy state by a subsequent total - energy comparison of these three phases .
our calculated electronic energies for different magnetic phases of us@xmath1 and use@xmath1 are all listed in table i. one can see that within gga formalism or the gga+@xmath0 formalism with a too small @xmath18 value , the fm state is more stable for both us@xmath1 and use@xmath1 , in contradiction with experimental results . therefore , the electron correlation effect has to be accounted for to correctly describe ground - state us@xmath1 and use@xmath1 . in the discussions that follow ,
we will confine our reports to the afm solutions for us@xmath1 and use@xmath1 .
the experimentally determined lattice parameters of us@xmath1 and use@xmath1 are ( @xmath20=5.37 , @xmath21=3.96 , @xmath22=9.94 ) and ( @xmath20=5.65 , @xmath21=4.06 , @xmath22=10.47 ) respectively @xcite .
our calculated lattice constants with different @xmath18 values are summarized in table ii , together with the experimental results .
we can see that different from what we found for actinide dioxides @xcite , the lattice constants of actinide trichalcogenides do not change monotonically with the value of the @xmath18 parameter .
besides , the lattice constants obtained in gga calculations are obviously too small in comparison with corresponding experimental results , for both us@xmath1 and use@xmath1 . for us@xmath1 ,
the value of @xmath18 has to be as large as 6 ev to get reasonable lattice constants compared with experimental results .
differently , a @xmath23 value of 4 ev is enough to get reasonable lattice constants for use@xmath1 , and changing @xmath18 from 4 to 6 ev has negligible effects on the obtained lattice constants . besides of the prominent changes in the atomic structure parameters , the most dramatic improvement brought by the gga+@xmath0 formalism when compared to the gga results is in the description of electronic structure properties . for this , we have investigated the band structures in afm phases for us@xmath1 and use@xmath1 aiming at revealing the fundamental influences of considering the on - site coulomb interaction .
figures 2(a ) and 2(b ) show the obtained local density of states ( ldos ) for the s1 , s3 , and u atoms in us@xmath1 , and se1 , se3 , and u atoms in use@xmath1 respectively . without accounting for the on - site coulomb repulsion
, one can see that the gga calculations predict an incorrect metallic ground state by nonzero occupation of uranium electrons at the fermi energy ( @xmath24 ) .
when switching on the @xmath18 parameter , as shown both in figs .
2(a ) and 2(b ) , the uranium electronic states in us@xmath1 and use@xmath1 begin to split at @xmath24 and tend to form two peaks with the gap of @xmath25 .
the amplitude of this gap increases with increasing @xmath18 .
previous electric resistivity measurements pointed out that use@xmath1 did not conduct at room temperatures @xcite , thus a band gap at @xmath24 should be contained in its electronic structure .
we can see from fig .
2(b ) that an energy band gap appears only when @xmath18 is larger than or equal to 4 ev for use@xmath1 . for us@xmath1 ,
the value of @xmath18 has to be enlarged to 6 ev to open the band gap at the fermi energy .
these results also imply that the electron correlation strength might be different in us@xmath1 and use@xmath1 . from the last figure in figs .
2(a ) and 2(b ) , we can see that the sulphur ( selenium ) and uranium electronic states overlap with each other and contribute equally to the valence band maximum ( vbm ) of us@xmath1 ( use@xmath1 ) , while the conduction band minimum ( cbm ) is composed of uranium electronic states .
the orbital mixing between sulphur ( selenium ) and uranium electronic states below the fermi energy indicates that there are covalent interactions between the sulphur ( selenium ) and uranium atoms in us@xmath1 ( use@xmath1 ) . to further analyze the orbital - resolved electronic structures , we calculate the projected density of states ( pdos ) for the x-@xmath26 ( @xmath27=3 ( x = s ) or 4 ( x = se ) ) , u-5@xmath2 and u-6@xmath28 electronic states in ux@xmath1 . figures 3(a)-3(d ) show the obtained pdos for us@xmath1 in different calculations , while the corresponding pdos results for use@xmath1 are shown in figs .
3(e)-3(h ) .
the value of @xmath18 is chosen to be 6 ev in all gga+@xmath0 and gga+@xmath0+soc calculations .
one can see that without considering the electron correlation effect , us@xmath1 and use@xmath1 are both wrongly predicted as metallic materials , which contradicts with experimental results . after adding the @xmath0 parameter to describe the strong on - site energy of u-5@xmath2 electrons ,
the energy gaps are opened for us@xmath1 and use@xmath1 .
we can see from figs .
3(d ) and 3(h ) that further considering soc has little influences on the electronic states around the fermi energy . for both us@xmath1 and use@xmath1 ,
the soc effect lies in the deep energy level , causing an energy splitting of the u-6@xmath28 states . from the pdos results
, we can also see clear difference between the x1- and x3-@xmath26 ( n=3 ( x = s ) or 4(x = se ) ) electronic states . the pdos for the electronic states of the x2 atom is very similar to that of the x1 atom and thus is not shown here .
the different electronic state distribution of the x1(x2 ) and x3 atoms proves the theory of considering ux@xmath1 as ux(x@xmath5 ) , and indicates the strong covalent bondings between the x1 and x2 atoms . to analyze more carefully the soc effects in us@xmath1 and use@xmath1
, we recalculate the electronic structures of them with higher resolutions , with the reduced gaussian smearing width of only 0.02 ev and a pdos resolution of 0.01 ev .
figures 4(a ) and 4(b ) show the obtained pdos results for uranium 5@xmath2 electrons of us@xmath1 and use@xmath1 , by using the gga method , while figs . 5(a ) and 5(b ) show the obtained energy band structures of us@xmath1 and use@xmath1 by using the gga+@xmath0 method , respectively .
all the gga+soc and gga+@xmath0+soc calculations are noncollinear , with both soc and orbital polarizations considered . and
the results before and after considering the soc effects are shown in solid and dotted lines respectively . from the pdos results obtained by using the gga method
, we can see that pure inclusion of soc to the gga method does not lead to energy band gaps for us@xmath1 and use@xmath1 .
this discovery is different from the situation of coo , where soc can solely open a band gap @xcite . from the band structures of us@xmath1 shown in fig .
5(a ) , we can see that both the lowest unoccupied conduction and highest occupied valence bands distribute along the @xmath29-z direction . for the lowest conduction band
, the soc effect causes an energy downshift of 0.05 ev along the @xmath29-z direction .
contrarily for its highest valence band , the soc effect causes an energy upshift of 0.03 ev at the @xmath29 point and an upshift of 0.07 ev at the z point . as a result , the energy band gap of us@xmath1 changes from indirect ( @xmath30z ) into direct type ( z@xmath31z ) after considering the soc effect , and the gap value reduces from 1.59 to be 1.48 ev . different from us@xmath1 , the lowest conduction and highest valence bands of use@xmath1 distributes along the k@xmath3-b and b-@xmath29-y directions respectively . for the lowest conduction band along the k@xmath3-b direction
, the soc effect causes an energy downshift of 0.04 ev at the k@xmath3 point and a downshift of 0.02 ev at the b point .
correspondingly the cbm of use@xmath1 moves from b to the k@xmath3 point .
however , since the vbm of use@xmath1 is at some point along the b-@xmath29-y direction instead of at any high - symmetry @xmath19 points , the soc effect does not change the indirect band gap type .
overall the band gap of use@xmath1 changes from 1.30 to be 1.23 ev after considering the soc effect .
our studies reveal that the energy band gap of use@xmath1 is smaller than that of us@xmath1 .
after systematically presenting the electronic structure results for us@xmath1 and use@xmath1 , we now turn to their mechanical properties .
the elastic constants are obtained by solving the eigenvalues of their hessian matrix , which is calculated based on the hooke s law and small position changes on independent s , se and u atoms .
the obtained elastic constants of us@xmath1 and use@xmath1 in different calculations by using the gga and gga+@xmath0 methods are both listed in table iii .
we can see that the elastic constants calculated by using the gga formalism for us@xmath1 and use@xmath1 do not satisfy the stability criteria @xcite for monoclinic structures that the @xmath32 , @xmath33 , @xmath34 , @xmath35 , @xmath36 , @xmath37 should be all positive .
this result further proves that conventional gga formalism fails for describing the uranium trichalcogenides .
contrarily , the calculated elastic constants by using the gga+@xmath0 method for both us@xmath1 and use@xmath1 satisfy the above stability criteria , as well as the other six criteria for monoclinic structures @xcite , proving the stable existence of us@xmath1 and use@xmath1 , and their strong electron correlation effects .
based on the elastic constants of us@xmath1 and use@xmath1 , we further calculate the voigt and reuss bounds on their bulk ( @xmath38 , @xmath39 ) and shear moduli ( @xmath40 , @xmath41 ) @xcite . and the bulk and shear moduli of us@xmath1 and use@xmath1 can be estimated by @xmath42=(@xmath38+@xmath39)/2 , and @xmath43=(@xmath40+@xmath41)/2 . and the poisson s ratios can be calculated by @xmath44=(3@xmath42 - 2@xmath43)/(6@xmath42 + 2@xmath43 ) . the obtained mechanical properties for us@xmath1 and use@xmath1 are all listed in table iv .
we can see that the voigt and beuss bounds obviously differ from each other for bulk and shear moduli .
this result reflects the structural anisotropy of us@xmath1 and use@xmath1 . besides
, the bulk and shear moduli are both found to be larger for us@xmath1 than for use@xmath1 .
the poisson s ratio is calculated to be 0.19 and 0.22 for us@xmath1 and use@xmath1 respectively .
until now , there are no experimental reports on the mechanical properties of us@xmath1 or use@xmath1 . hence our theoretical values can be used as references for further investigations or industrial applications of them . at another side , the raman properties of the mx@xmath1 group materials have been systematically studied for a long time , and some of the characteristic peaks of us@xmath1 and use@xmath1 were successfully observed in experiments @xcite .
therefore , here we further calculate the raman properties of us@xmath1 and use@xmath1 . according to group theory and the symmetry of them , both us@xmath1 and use@xmath1
are predicted to have 12 raman - active vibrational modes @xcite : 8 a@xmath45 and 4 b@xmath45 modes respectively .
the symbols a@xmath45 and b@xmath45 represent for two different vibration symmetries respectively : vibrations inside and out of the mirror plane .
specifically , a@xmath45 corresponds to the vibrations where @xmath46 is always 0 while b@xmath45 corresponds to the vibrations where @xmath46 is not 0 .
our raman results of us@xmath1 and use@xmath1 are obtained by symmetry analysis based on the vibrational modes calculated within the gga+@xmath0 method , where soc effects are considered .
figures 6(a ) and 6(b ) show our determined raman frequencies for us@xmath1 and use@xmath1 respectively , together with the experimental results by g. nouvel _ et al .
one can see that by using the gga formalism for both us@xmath1 and use@xmath1 , the raman frequencies are totally confused compared with the experimental results . at another side , the gga+@xmath0 results on the raman frequencies are clearly reasonable in two obvious aspects : i ) there exist a highest single raman frequency for both us@xmath1 and use@xmath1 , corresponding to the molecular - like x@xmath5 mode vibrations ; ii ) the highest raman frequency of us@xmath1 is much larger than that of use@xmath1 , indicating that the s - s interaction is stronger than se - se .
when comparing all the raman frequencies , there seems to be a frequency shift of about 25@xmath4735 @xmath48 between our gga+@xmath0 and the experimental results .
by using the gga and gga+@xmath0 methods , we have systematically studied the electronic , mechanical , and raman properties of us@xmath1 and use@xmath1 , aiming to reveal the underlying electron correlation and soc effects . by comparing the calculated lattice constants of us@xmath1 and use@xmath1 with corresponding experimental results , and monitoring their electronic structures with different @xmath18 parameters ,
we conclude that a @xmath18 value of 6 ev is needed to get reasonable lattice constants , and the right insulating properties for both us@xmath1 and use@xmath1 . after considering soc effects ,
the deep u-6@xmath28 band will be split into two separate energy bands in both us@xmath1 and use@xmath1 .
the soc effects also changes the energy band shapes around the fermi energies and lead to the fact that us@xmath1 is a direct band gap while use@xmath1 is an indirect band gap insulator . because of the soc effects
, the energy band gaps of us@xmath1 and use@xmath1 reduce by 0.11 and 0.07 ev respectively .
based on the obtained antiferromagnetic ground states of us@xmath1 and use@xmath1 , we then systematically give out their electrical , mechanical , and raman properties .
the vbm and cbm of us@xmath1 ( use@xmath1 ) are found to be contributed by 3@xmath28 - 5@xmath2 ( 4@xmath28 - 5@xmath2 ) hybrid and 5@xmath2 ( 5@xmath2 ) electronic states respectively .
the elastic constants of us@xmath1 and use@xmath1 are found to satisfy the stability criteria , and their poisson s ratios are calculated to be 0.19 and 0.22 respectively . for the raman properties , we theoretically repeat the two important experimental observations that the s - s or se - se dimer vibrations have the largest frequencies in us@xmath1 and use@xmath1 , and the frequency of the s - s dimer vibration is larger than that of the se - se dimer vibration .
this work was supported by the national natural science foundation of china under grants no .
10904004 and no . 90921003 and
foundations for development of science and technology of china academy of engineering physics under grants no .
2011b0301060 , no .
2011a0301016 , and no . 2008a0301013 .
.electronic energies per unit cell of us@xmath1 and use@xmath1 in the nonmagnetic ( nm ) , ferromagnetic ( fm ) , and antioferromagnetic ( afm ) states by using different calculation methods .
all values are in units of ev . [ cols="<,^,^,^,^,^,^ " , ] * list of captions * + * fig.1 * ( color online ) .
( a ) the atomic structures of mx@xmath1 , where blue and dark green balls representing for metal ( m ) and chalcogen ( x ) atoms respectively .
the monoclinic lattices are depicted by the dashed lines .
( b ) depiction of the brillouin zone for the monoclinic mx@xmath1 lattice . + * fig.2 * ( color online ) . the local density of states ( ldos ) for the s1 , s3 , and u atoms in us@xmath1 ( a ) and se1 , se2 , and u atoms in use@xmath1 ( b ) by using the gga and gga+@xmath0 methods with @xmath18 ranging from 1 to 6 ev .
the s1 ( se1 ) and s3 ( se3 ) atoms correspond to x1 and x3 atoms depicted in fig .
the fermi energies are denoted by dashed lines . + * fig.3 * ( color online ) .
( a)-(d ) the projected density of states ( pdos ) for the 3@xmath28 electronic states of s1 and s3 atoms , and 5@xmath2 , 6@xmath28 electronic states of the u atom in us@xmath1 .
( e)-(h ) the pdos for the 4@xmath28 electronic states of se1 and se3 atoms , and 5@xmath2 , 6@xmath28 electronic states of the u atom in use@xmath1 .
the s1 ( se1 ) and s3 ( se3 ) atoms correspond to x1 and x3 atoms depicted in fig .
the fermi energies are denoted by dashed lines .
the values of @xmath18 are chosen to be 6 ev in all gga+@xmath0 and gga+@xmath0+soc calculations . + * fig.4 * ( color online ) .
the projected density of states for the uranium 5@xmath2 electronic states in us@xmath1 ( a ) and use@xmath1 ( b ) , by using the gga method .
the fermi energies are set to be zero and denoted by dashed lines .
the results before and after considering the spin - orbit coupling effects are shown in red solid and blue dotted lines respectively . +
* fig.5 * ( color online ) .
the electronic energy band structures for us@xmath1 ( a ) and use@xmath1 ( b ) by using the gga+@xmath0 method .
the values of @xmath18 are chosen to be 6 ev . the fermi energy and different @xmath19-points are denoted by the dashed lines .
the results before and after considering the spin - orbit coupling effects are shown in solid and dotted lines respectively . + * fig.6 * ( color online ) .
the calculated frequencies at the @xmath29 point for raman - active vibrational modes of us@xmath1 ( a ) and use@xmath1 ( b ) by using the gga and gga+@xmath0 methods , together with the experimental results listed in ref . .
the values of @xmath18 are chosen to be 6 ev . the symmetry for each vibration mode are also presented . | a systematic density functional theory ( dft ) + @xmath0 study is conducted to investigate the electron correlation and spin - orbit coupling ( soc ) effects in us@xmath1 and use@xmath1 .
our calculations reveal that inclusion of the @xmath0 term is essential to get energy band gaps for them , indicating the strong correlation effects for uranium 5@xmath2 electrons . taking consideration of the soc effect results in small reduction on the electronic band gaps of us@xmath1 and use@xmath1 , but
largely changes the energy band shapes around the fermi energy . as a result , us@xmath1 has a direct band gap while use@xmath1 has an indirect one .
our calculations predict that both us@xmath1 and use@xmath1 are antiferromagnetic insulators , in agreement with corresponding experimental results .
based on our dft+@xmath0 calculations , we systematically present the ground - state electronic , mechanical and raman properties for us@xmath1 and use@xmath1 . |
in the 80 p. sauer , and his group at the university of hanover , solved the faddeev equations for the three nucleon systems @xcite with the state of the art realistic nn potentials @xcite . simultaneously a vigorous experimental program was developed at saclay .
thanks to the availability of an @xmath0he and a tritium targets , the isoscalar and isovector components of the charge and the magnetic elastic form factors of the three nucleon systems were determined @xcite .
they remarkably agreed with the predictions of the hanover group .
the key to this success was not only the tour de force " in operating the radioactive target , which was only made possible by an environment of high technology , but also the close collaboration between a theoretical and an experimental team .
i met peter at this occasion and , starting from his wave functions , i extended my diagrammatic approach @xcite to the disintegration channels of the three nucleon systems @xcite .
an experimental program was ongoing not only at saclay but also at nikhef and bates , in order to determine the wave function of the few body systems .
again , the experiments confirmed that peter s wave functions were doing a good job up to nucleon momentum of 500 mev / c .
however , it became evident that the energy and the duty factor of the available electron accelerators were not sufficient to map out the wave function at higher momentum .
kinematical limits prevented to increase significantly the momentum transfer .
rescattering and interaction effects were not negligible at such a low energy . finally the low duty factor prevented to single out the very small cross sections , associated with high momentum transfers , from background .
one of the reasons to built cebaf at jefferson laboratory ( jlab ) was to overcome these drawbacks .
the energy was increased by one order of magnitude ( 600 mev to 6 gev ) and the duty factor by two orders of magnitude ( 1% to 100% ) .
about seven years of operation have produced a few jewels in the disintegration of few body systems .
this meeting in honor of p. sauer provides me with a good opportunity to revisit the field in view of recent achievements and to evaluate future developments .
the primary goal of the study of the ( e , ep ) reaction on nuclei was , and still is , the determination of the high momentum components of the nuclear wave function , in view of the study of short range correlations and possible exotic configurations . in the past the spectral functions measured at saclay or amsterdam suffered for large corrections ( about a factor two or more ) due to final state interactions ( fsi ) and meson exchange currents ( mec ) .
a survey of the state of the art at that time can be found in ref .
the corresponding experiments were performed at low values ( @xmath1 gev@xmath2 ) of the virtuality @xmath3 of the exchanged photon . when cebaf was decided , a common belief was that increasing @xmath3 was the way to suppress fsi and mec contributions .
this was partly true , since both the fsi and mec amplitudes involve a loop integral , which connects the nuclear bound and scattering states and which is expected to decrease when @xmath3 increases as form factors do .
but this was partly wrong , since the singular part of the fsi integral does not depends on @xmath3 , besides the trivial momentum dependency of the elementary operators .
it comes from unitarity , and corresponds to the propagation of an on - shell nucleon .
it involves on - shell elementary matrix elements and it is maximum when the kinematics allows for rescattering on a nucleon at rest .
this happens in the quasi - free kinematics , when @xmath4 ( being @xmath5 the energy of the virtual photon ) .
[ sing ] exhibits these features .
it shows the angular distribution , against the neutron angle @xmath6 , of the ratio between the full cross section of the d(e , ep)n reaction and the quasi free contribution , when the momentum @xmath7 of the recoiling neutron is kept constant .
fsi are maximum near @xmath8 where @xmath9 and on - shell rescattering is maximized . at low values of the recoil momentum ( @xmath10 mev / c ) ,
on - shell nucleon rescattering reduces the quasi free contribution , as expected from unitarity ( a part of the strength of the quasi elastic channel is transferred to inelastic ones ) . at high values of the recoil momentum
( @xmath11 mev / c ) the quasi free contribution strongly decreases as the nucleon momentum distribution : on - shell rescattering takes over and dominates .
an experiment @xcite recently performed at jlab confirms this behavior .
[ ulmer ] shows the recoil momentum distribution up to 600 mev / c , in the quasi elastic kinematic ( x=1 ) . above @xmath12
mev / c , fsi ( dashed line ) dominate while mec ( full line ) contribute to a lesser extent . although the virtuality ( @xmath13 gev@xmath2 ) of the exchanged photon is about twice as much than what was achieved twenty years ago , the relative kinetic energy of the two outgoing nucleons ( @xmath14 mev ) is still low enough to rely on the partial wave expansion of the np scattering amplitude .
the curves are the results of the original model @xcite , where both the on - shell and half off - shell np scattering amplitudes are solution of the lippman - schwinger equation with the same potential ( paris ) as for the bound state . at higher energies
( let s say when the relative kinetic energy of the outgiong fragment exceeds 500 mev or so ) , too many partial waves enter into the game and their growing inelasticities prevent to compute the scattering amplitude from a potential .
it is better to use a global parameterization of the nn scattering amplitude . on general ground
, it can be expanded as follows @xmath15 above 500 mev , the central part @xmath16 dominates and is almost entirely absorptive , and takes the simple form @xmath17\ ] ] in the forward direction its imaginary part is related to the total cross section @xmath18 , while the slope parameter @xmath19 is related to the angular distribution of nn scattering .
both can be determined from the experiments performed at los alamos , saturne and cosy .
the ratio @xmath20 , between the real and imaginary part of the amplitude , is small and does not exceed 0.2 .
the spin - orbit and spin - spin terms are related to polarization observables , but i have not yet implemented them in the three body codes .
such a parameterization is very convenient to compute the rescattering amplitude , but leads only to an accurate prediction of its singular part ( on - shell scattering ) .
contrary to low energy , there is unfortunately no way to constrain the half - off part of the nn scattering amplitude , and one can get only estimate of the principal part of the rescattering amplitude .
it turns out that it does not dominate at high energy .
so , the method is founded on solid grounds in the quasi - elastic kinematics ( x@xmath211 ) .
away , it tells us in which kinematics fsi are minimized .
[ he3pd ] shows how well this method reproduces the cross section of the @xmath22 reaction recently measured at jlab @xcite at @xmath23 gev@xmath2 , in the quasi - free kinematics ( x=1 ) . at such a high virtuality ,
the relative kinetic energy between the outgoing proton and deuteron is @xmath24 mev , where the nn cross section reaches its maximun and becomes flat around @xmath25 mb .
again , fsi reduces the quasi - free contribution below 300 mev / c and overwhelms it by more than a factor five around 500 mev / c .
details on the model are given in ref .
@xcite . to summarize ,
a fair agreement with the recent jlab data has been reached in the quasi - elastic regime , up to recoil momentum of the order of 1 gev / c , provided that the nn scattering amplitude relevant to the same energy range is used . in order to determine the high momentum components of the nuclear wave function ,
one has to go away the quasi - elastic kinematics : as demonstrated in fig .
[ sing ] , this occurs in parallel or anti - parallel kinematics , where on - shell nucleon rescattering is suppressed .
two body short range correlations are the primary source of high momentum components in the nuclear wave function .
since they are strongly coupled to high energy state in the continuum , two nucleon production experiments are the natural way to reveal them . however , above the pion production threshold , one must eliminate meson production channels and perform exclusive experiments .
clas ( cebaf large acceptance spectrometer ) at jlab is the ideal detector for disentangling and studying multiparticle channels .
however , clas is not an hermetic detector : there is a hole in the forward direction as well as in the backward direction ; also the coils , which provide the toroidal electromagnetic field , induce dead zones in the acceptance .
so the only fair way to compare the data to any theory is to use the same cuts . to that end , g. audit and i ,
we have developed a monte carlo programme which simulates the @xmath0he(e , e2p)n and @xmath0he(@xmath26,2p)n reactions within the clas acceptance .
the event generator is the code which i developed twenty years ago and which i updated to take into account the high energy parameterization of the nn scattering amplitude , as well as relativistic effects .
let me discuss first the real and then the virtual photon sectors .
[ dalitz ] shows the result of such a simulation within the clas acceptance .
the model @xcite has been calibrated against previous data recorded below 800 mev in restricted parts of the phase space : two magnetic spectrometers @xcite in the two body disintegration sector ( neutron at rest ) ; two magnetic spectrometers @xcite and daphne detector @xcite in the three body disintegration sector .
clas results @xcite enlarge the available data set , not only by covering the full phase space but also by extending the energy range in a single shot .
cuts can be made in various places of the dalitz plot , in order to emphazis different mechanisms and access different aspects of the three body dynamics .
for instance , the peak at bottom of the dalitz plot ( small @xmath27 ) corresponds to the disintegration of a pair of protons at rest .
since this pair has no dipole moment , mec associated with the @xmath28 resonance are suppressed and one can probe higher order mec .
the peaks on the left and right edges of the dalitz plot correspond to rescattering within a proton neutron pair , while the other proton takes almost all the available energy .
finally , in the center of the dalitz plot the energy is shared by the three nucleons : this is the place where three nucleon effects dominate .
preliminary results show a clear distinction between two and three body effects and follow the trends of the model .
i refer to ref .
@xcite for a more detailed account of the preliminary clas data .
the @xmath0he(e , e2p)n reaction was advocated @xcite as the best way to access two proton correlations in few body systems . in particular , in kinematics where the neutron is almost at rest , the transverse part of the amplitude is suppressed since the pair of proton has no dipole moment . it also turns out that three body mechanisms are suppressed , since they prefer kinematics where the momentum transfer is shared by the three nucleon . on the contrary
, the coupling of a longitudinal photon the the proton pair is not suppressed , and the only sizeable correction is due to fsi between the two outgoing protons .
clas provides us with the first comprehensive study of this channel @xcite .
[ back ] shows the momentum distribution of the pair of protons almost at rest in @xmath0he .
the plane wave calculation exhibits the characteristic node of the s - wave part of its wave function , which is the only one which survives when the pair is at rest .
of course , fsi fill in this hole ( nature does not like holes ! ) , and mec contribute moderatly at high momentum . both the theory and the experiment
are integrated within the same cuts ( for instance , the fall off of the cross section below 0.2 gev@xmath2 reflects the clas acceptance ) and there is no normalization factor ( absolute comparison ) .
the wave function of p. sauer does a good job in a configuration where it has never been checked before .
increasing the virtuality from @xmath29 gev@xmath2 up to a few gev@xmath2 will provide us with a more stringent test . since this kinematic is far from the quasi - elastic one ,
on - shell nucleon rescattering is not dominant , and fsi and mec will be suppressed . on the contrary , on - shell
nucleon rescattering dominates the cross section in the kinematics reported in fig .
[ sym ] . here
the two protons are required to be emitted in a symmetric way around the direction of the virtual photon , and the cross section is plotted against the energy of the virtual photon .
the good agreement between the theory and the experiment tells us that our description of the on - shell rescattering matrix element is fine over a wide energy range ( @xmath30 gev ) .
other cuts have been applied to clas data .
for instance , one can select a very fast neutron and look for a proton pair almost at rest in @xmath0he . in a plane wave picture , this gives access to the relative wave function of the two protons .
however , in the interesting situation where the relative momentum between the two protons is large , the three nucleons share the total energy and momentum and three body mechanisms contribute significantly .
i refer to ref .
@xcite for a more detailed accounts of such a study .
the operation of cebaf at jefferson laboratory , over the past few years , has confirmed the expectations which we had twenty years ago , but has enlarged in a considerable way the data set , both in accuracy and in energy as well as momentum range . in the real photon sector , the three body photo disintegration of @xmath0he is dominated by three body mechanisms related to the on - shell propagation of a pion between the three nucleons .
the extension to the virtual photon sector gives access to three body meson exchange currents , which may be related to three body forces . in the virtual photon sector
, the two - body electrodisintegration of @xmath0he , and of a pair of two nucleons bound in @xmath0he , opens windows on the momentum distribution at short distance .
however special care has to be taken in order to avoid on - shell nucleon scattering which dominates the electrodisintegration cross section in a well defined part of the phase space , namely near the quasi free kinematics .
the best candidate to study the short range components of the nuclear wave function is the parallel kinematics , where the outgoing nucleons are emitted along the direction of the virtual photon . in that kinematics fsi and mec
contributions are reasonably small and are expected to decrease when the virtuality @xmath3 increases in the range of several gev@xmath2 . alternately , one may take advantage of the strong on - shell scattering contribution to study the rescattering of exotic components of the nucleon ( and other hadrons ) wave function .
the study of color transparency ( ct ) is an example of such a possibility .
ct is a natural consequence of qcd .
when a photon of high virtuality @xmath3 interacts elastically with a nucleon , it selects configurations with small transverse extension which undergo less rescattering when they travel in the nuclear medium .
however it turns out that in the range of virtuality currently available ( @xmath31 gev@xmath2 ) the lifetime of such small configurations is comparable to the internucleon distance and much smaller than the radius of a nucleus . this explains why no signal of color transparency has been reported so far . a better way would be to work in kinematics where rescatterings between to nucleons are maximized , such as in fig .
[ sing ] . at low virtuality
@xmath3 the hight and the width of the rescattering peak are on solid grounds , since the matrix element depends of on - shell elementary amplitudes ( which can be borrowed from the corresponding elementary channels ) and on the low momentum part of the nuclear wave function . at higher virtuality
, ct will cause the peak to decrease and its width to increase .
i refer to my contribution in ref .
@xcite for a more detailed discussion .
while the present energy of cebaf ( 6 gev ) allows to test this conjecture at low virtually ( @xmath31 gev@xmath2 ) , only its energy upgrade to 12 gev will allows to reach high enough virtuality ( @xmath32 gev@xmath2 ) to see the effects of color transparency . | twenty years after p. sauer released the state of the art faddeev solution of the bound state three nucleon systems , i revisit photo and electrodisengration of few body systems with a special emphasis on the prospects opened at jefferson laboratory . |
see @xcite for terms used here for plane graphs .
it is well - known that the p - vector of any @xmath1-valent plane graph satisfies to @xmath8 .
some examples of applications of plane @xmath1-valent graphs are _ projections of links _ , _ rectilinear embedding _ in vlsi and _ gauss crossing problem _ for plane graphs ( see , for example , @xcite ) .
@xmath0-hedrite _ any plane @xmath3-connected @xmath1-valent graph , such that the number @xmath9 of its @xmath10-gonal faces is zero for any @xmath10 , different from @xmath11 and @xmath1 , and such that @xmath12 .
so , an @xmath2-vertex @xmath0-hedrite has @xmath13 .
clearly , @xmath14 , @xmath15 , @xmath16 , @xmath17 and @xmath18 are all possibilities .
an @xmath19-hedrite is called _ octahedrite _ in @xcite ; in fact , this paper is a follow - up of @xcite . in a way , this paper continues the program of kirkman ( @xcite p. 282 ) of classification of projections of alternating links .
see on the table below short presentation of @xmath0-hedrites and their symmetry groups . in the last column
we indicate goldberg - coxeter operation @xmath20 starting from @xmath0-hedrite @xmath21 with smallest number of vertices ( see @xcite , @xcite and @xcite ) . for @xmath4-connected plane graphs without @xmath3-gons ,
the following theorem of mani is valid : the symmetry group of the graph can be realized as point group of a convex polyhedron having this graph as skeleton .
but in the presence of @xmath3-gonal faces , one can not speak of convex polyhedra but we expect that the mani result still holds for this more general case . [ cols="^ " , ] any @xmath0-hedrite is @xmath3-connected . *
* let @xmath22 be an @xmath0-hedrite and assume that there is one vertex @xmath23 , such that @xmath24 is disconnected in two components @xmath25 and @xmath26 . then
two edges from @xmath23 will connect to a vertex @xmath27 of @xmath25 and two edges from @xmath23 will connect to a vertex @xmath28 of @xmath26 , because , otherwise , the exterior face is @xmath29-gonal with @xmath30 .
see below the corresponding drawing .
but the vertex @xmath27 will disconnect the graph and so , iterating the construction , we obtain an infinite sequence @xmath31 of vertices that disconnects @xmath22 .
this is impossible since @xmath22 is finite .
any @xmath0-hedrite ( moreover , any eulerian graph ) has at least one eulerian circuit of edges ; so , there is no cut - edge . but
a cut - vertex appears already for some eulerian ( @xmath4-vertex ) and @xmath1-valent ( @xmath1-vertex ) plane graphs .
[ 3-connectedness ] any @xmath0-hedrite , which is not @xmath4-connected , is one of @xmath32 , @xmath33 , @xmath34 , @xmath35 for some @xmath36 .
* let @xmath22 be an @xmath0-hedrite and assume that it is not @xmath4-connected .
then there are two vertices , say , @xmath23 and @xmath37 , such that @xmath38 is disconnected in two components , say , @xmath25 and @xmath26 . amongst the @xmath1 edges from @xmath23 ( respectively @xmath37 ) ,
the edges @xmath39 ( respectively @xmath40 ) go to @xmath25 .
two numbers @xmath41 and @xmath42 can takes values @xmath43 ,
@xmath3 or @xmath4 ; we will consider all possible cases .
assume that @xmath44 and @xmath45 , then the edges @xmath46 and @xmath47 must be distinct , since , otherwise , @xmath25 is the empty graph .
moreover , @xmath46 and @xmath47 have no common vertices , since , otherwise , @xmath22 would not be @xmath3-connected .
so , @xmath23 and @xmath37 are connected by @xmath46 and @xmath47 to a vertex @xmath27 and @xmath28 , respectively . since
face - size is at most @xmath1 , the vertices @xmath23 and @xmath37 , ( respectively , @xmath27 and @xmath28 ) are linked by two edges ( see figure [ fig : thethreecases ] ) .
two points @xmath27 and @xmath28 can either be connected by two edges and we are done , or disconnect the graph . in the latter case , we can iterate the construction . since the graph is finite , the construction eventually finish and we get a graph @xmath48 . if @xmath44 and @xmath49 , then by a similar reasoning , one gets again a graph @xmath48 .
assume that @xmath50 and @xmath51 .
one has @xmath52 , since , otherwise , one can attribute an edge to @xmath26 and get the case @xmath44 and @xmath45 .
so , one has , say , @xmath53 and @xmath54 , and the following two possibilities ( see figure [ fig : thethreecases ] ) : either @xmath55 ( this corresponds to @xmath56 and @xmath57 being the edges of two @xmath3-gons ) and we are done , or @xmath58 .
assume now that @xmath58 ; two points @xmath59 and @xmath60 can either be connected by two edges and we are done , or disconnect the graph . in the latter case , we can iterate the construction . since the graph is finite ,
the construction eventually finish .
if we do the same construction on the other side , then we get a similar structure and the graph is of the form @xmath61 , @xmath33 or @xmath32 with @xmath36 .
assume now that @xmath50 and @xmath45 .
the edges @xmath62 , @xmath63 and @xmath64 are all distinct , since , otherwise , the vertex @xmath23 disconnects the graph .
so , @xmath37 is connected by @xmath64 to a vertex @xmath28 . now
, either @xmath37 or @xmath28 is connected to @xmath23 , since , otherwise , we would have a @xmath65-gonal face .
if @xmath28 is connected to @xmath23 , then the pair @xmath66 disconnects the graph .
this construction is infinite ( see figure [ fig : thethreecases ] ) ; so , we get a contradiction .
\(i ) if @xmath22 is an @xmath0-hedrite with two adjacent @xmath3-gons , then this is a @xmath1-hedrite * 2 - 1 * or a @xmath48 with @xmath36 .
\(ii ) if @xmath22 is an @xmath0-hedrite with two @xmath3-gons sharing a vertex , then it is either @xmath1-hedrite * 4 - 1 * , or an @xmath61 , or @xmath65-hedrite * 3 - 1 * , or an @xmath67 with @xmath36 .
* proof . *
the proof for ( i ) , ( ii ) is similar to the cases @xmath68=@xmath69 , @xmath70 of theorem [ 3-connectedness ] .
\(i ) an @xmath2-vertex @xmath1-hedrites exists if and only if @xmath71 , even . \(ii ) an @xmath2-vertex @xmath65-hedrites exists if and only if @xmath72 , @xmath73 .
\(iii ) an @xmath2-vertex @xmath74-hedrites exists if and only if @xmath75 .
\(iv ) an @xmath2-vertex @xmath76-hedrites exists if and only if @xmath77 .
\(v ) an @xmath2-vertex @xmath19-hedrites exists if and only if @xmath78 , @xmath79 . *
proof . * the case ( v ) is proven in @xcite , page 282 . the case ( i )
is trivial ; take , for example , the serie @xmath48 .
the series @xmath32 , @xmath67 for any @xmath80 , give @xmath74-hedrites and @xmath65-hedrites with even and , respectively , odd number of vertices .
for @xmath65- , @xmath74- and @xmath76-hedrites , we get by @xmath81-inflation along a central circuit of length @xmath1 in corresponding @xmath0-hedrite * 6 - 1 * , * 5 - 1 * and * 7 - 1 * , series with @xmath82 , @xmath83 and @xmath84 vertices . by @xmath85-inflation along such central circuit in @xmath76-hedrite * 8 - 1 * , we get a serie of @xmath76-hedrites with @xmath86 vertices . by @xmath81-inflation along central circuit in @xmath74-hedrite * 11 - 4 * , we get serie of @xmath74-hedrites with @xmath87 vertices . by @xmath88-inflation along central circuit of length @xmath19 in @xmath74-hedrite * 15 - 10 * , we get serie of @xmath74-hedrites with @xmath89 vertices . inscribing consecutively @xmath1-gons in the @xmath1-gon , which is adjacent only to @xmath4-gons , in @xmath76-hedrite * 9 - 1 * and * 10 - 2 * , we get series of @xmath76-hedrites for the remaining cases of @xmath83 and @xmath82 vertices . in case of @xmath65-hedrites , it remains to prove existence for the case @xmath90 .
we obtain existence in the sub - cases @xmath91 , where @xmath92 and not divisible by @xmath1 , @xmath93 , and @xmath94 , respectively : by @xmath95-inflation of some @xmath29-vertex @xmath65-hedrite ; we showed their existence , by @xmath95-inflation of @xmath65-hedrites * 8 - 1 * , by @xmath95-inflation of any @xmath96-vertex @xmath65-hedrite ( for example , one coming from * 10 - 2 * by inflation along central circuit of length @xmath74 ) . our computation ( see the last section ) present all @xmath0-hedrites with at most @xmath6 vertices .
[ irre ] any irreducible @xmath0-hedrite has at most @xmath7 central circuits and equality is attained for each @xmath0 , @xmath97 .
* proof . * for @xmath98 ,
the theorem is proved in @xcite .
we will show , using suitable cutting , that this result implies the theorem for @xmath99 .
let us start with the simplest case of @xmath76-hedrites . consider a simple circuit @xmath100 in its curvature graph , which contains the vertex corresponding to unique @xmath3-gon .
remind that the vertices in the curvature graph correspond to @xmath3- or @xmath4-gons , while edges correspond to pseudo - roads .
so , the simple circuit @xmath100 corresponds to the circuit of faces of @xmath22 , containing our @xmath3-gon , some @xmath1-gons and , possibly , some of six @xmath4-gons ; see the picture below .
suppose that the @xmath76-hedrite @xmath22 is irreducible and has @xmath101 central circuits . by adding the central circuit @xmath102 , which is shown by dotted lines on picture above
, we produce an @xmath19-hedrite ( since , the @xmath3-gon is cut by @xmath102 in two @xmath4-gons ) , which is still irreducible and has @xmath103 central circuits .
so , @xmath104 , by theorem 3 of @xcite . for remaining cases of @xmath0-hedrites with @xmath105 ,
the proof is similar . in each case
, we consider all possible distribution of @xmath3-gons by simple circuits in the graph of curvatures and , for each such circuit , we add suitable number of new central circuits .
all possibilities are presented on figure [ the456hedritecases ] : two for @xmath106 , three for @xmath107 and two for @xmath108 . in the last case ,
there are no @xmath4-gons and so , simple circuits in the curvature graph contain only even number of @xmath3-gons by local euler formula of theorem [ local - euler - formula ] .
note that the case of @xmath1-hedrites is obvious by theorem 5 of @xcite .
let @xmath22 be an irreducible @xmath19-hedrite and @xmath102 a central circuit , which is incident to three @xmath4-gons on one side
. then one can add another central circuit to @xmath22 , so that the resulting graph is still irreducible .
* proof . * from every one of the three @xmath4-gons , say , @xmath109 , @xmath110 , @xmath111 , one can define two pseudo - roads from the sides of the @xmath4-gon , which do not belong to the central circuit .
each such pseudo - road defines an edge , say , @xmath62 , @xmath63 , @xmath112 in the graph of curvatures and so , a triangle in that graph . then , either two triangles @xmath113 and @xmath114 are linked by a path , which does not involve the edges @xmath115 , or they are not linked by such a patch . in both cases we can cut the @xmath19-hedrite according to figure [ twocasescutting ] and obtain another @xmath19-hedrite , which is still irreducible .
see below example of an irreducible @xmath76-hedrite ( its cc - vector is @xmath116 , its symmetry group is @xmath117 ) with the maximum number of central circuits . for all other @xmath0
, there is an example of irreducible @xmath0-hedrite with @xmath7 central circuits ( see theorem [ theonewithsimplecentralcircuit ] ) , which is , moreover , pure
. an irreducible @xmath0-hedrite is maximal irreducible if and only if it has @xmath7 central circuits .
the easiest case , @xmath108 , of i - hedrites admits following complete characterization : [ theorem - for-4-hedrite ] ( i ) any @xmath1-hedrite can be obtained from some @xmath1-hedrite with two central circuits by simultaneous @xmath118- and @xmath119-inflation along those circuits ; it is irreducible if and only if @xmath120 .
\(ii ) any @xmath1-hedrite with two central circuits is defined by its number of vertices @xmath2 and by _ shift _
@xmath10 , @xmath121 with @xmath122 , vertices between the pair of boundary @xmath3-gons on the horizontal circuit ( see , for example , @xmath1-hedrite * 8 - 1 * ) and the remaining pair of @xmath3-gons .
remark that several different values of shift can yield the same graph .
\(iii ) any @xmath1-hedrite is balanced
. * proof .
* ( i ) and ( ii ) are proved in @xcite , while ( iii ) is obvious for @xmath1-hedrite with two central circuits and remain true under @xmath118- and @xmath119-inflation .
the shift @xmath123 corresponds to * 2 - 1 * and its only - on - one - circuit @xmath29-inflations @xmath124 .
the shift @xmath125 corresponds to * 4 - 1 * and @xmath126 .
denote by @xmath127 , for any @xmath80 , any @xmath128-vertex @xmath1-hedrite obtained from * 4 - 1 * by @xmath29-inflation of only one its central circuit ; so , its cc - vector is @xmath129 , its symmetry is @xmath130 and it is reducible . clearly , any @xmath131 has the maximal shift @xmath132 .
[ theonewithsimplecentralcircuit ] any pure irreducible @xmath0-hedrite is , either any @xmath1-hedrite with two central circuits , or a @xmath65-hedrite * 6 - 2 * , or one of @xmath74-hedrites * 8 - 6 * , * 14 - 20 * , or one of the following eight @xmath19-hedrites : * 6 - 1 * , * 12 - 4 * , * 12 - 5 * , * 14 - 7 * , and ( see figure [ thepureirreducibleoctahedritewith56cc ] ) * 20 - 1 * , * 22 - 1 * , * 30 - 1 * , * 32 - 1*. * proof . *
let @xmath22 be a pure irreducible @xmath0-hedrite having @xmath133 central circuits .
if one deletes a central circuit , then , in general , @xmath43-gon can appear .
it does not happen for @xmath22 , since it would imply a self - intersection of a central circuit .
so , the result of deletion of a central circuit from @xmath22 produces an pure irreducible @xmath0-hedrite with @xmath134 central circuits . first , if @xmath135 , then the theorem 5 from @xcite gives that such @xmath22 are exactly @xmath1-hedrites with two central circuits ; all of them are classified in theorem [ theorem - for-4-hedrite ] .
we prove the theorem by systematic analysis of all possible ways to add to @xmath22 ( for @xmath136 ) a central circuit , in order to get a pure irreducible @xmath0-hedrite with @xmath137 central circuits .
let @xmath135 .
then @xmath22 can be only one of two smallest @xmath1-hedrites .
in fact , if @xmath22 is another @xmath1-hedrite , then , because of classification theorem [ theorem - for-4-hedrite ] , it has a form as in figure [ cutting4hedrite ] .
new central circuit should cut both @xmath3-gons on opposite edges , since , otherwise , there is a rail - road . but figure [ cutting4hedrite ] shows , on example for @xmath138 , that a self - intersection appears if two central circuits intersect in more than four vertices .
so , the only possible @xmath1-hedrites with _ two _ central circuits , which can be cut in order to produce irreducible pure @xmath0-hedrite are @xmath1-hedrites * 2 - 1 * and * 4 - 1*. all cases are indicated below . now , all irreducible pure @xmath0-hedrites with _ three _ central circuits are @xmath65-hedrite * 6 - 2 * , @xmath74-hedrite * 8 - 5 * and @xmath19-hedrite * 6 - 1 * ( i.e. the projections of links @xmath139 , @xmath140 and @xmath141 ) .
in fact , we apply the same procedure to those three @xmath0-hedrites ; see picture below : next , all irreducible pure @xmath0-hedrites with _ four _ central circuits are @xmath19-hedrites * 12 - 4 * , * 12 - 5 * , * 14 - 7 * and @xmath74-hedrite * 14 - 20*. by the same method , one can see that there are exactly two pure irreducible @xmath0-hedrites with _ five _ central circuits and two with _ six _ central circuits ( see figure [ thepureirreducibleoctahedritewith56cc ] ) .
= 2.5 cm * nr.20 - 1 * @xmath130 + @xmath142 + = 2.5 cm * nr.22 - 1 * @xmath143 + @xmath144 + = 2.3 cm * nr.30 - 1 * @xmath145 + @xmath146 + = 2.3 cm * nr.32 - 1 * @xmath147 + @xmath148 + any pure @xmath0-hedrite comes from a pure _ irreducible _ @xmath0-hedrite with , say , @xmath10 central circuits by simultaneous @xmath149-inflation along those circuits ; it is irreducible if and only if @xmath150 .
we consider below the maximal symmetry groups of plane graphs ; these groups are identified with the corresponding point groups .
we indicate here the lists of symmetry groups of @xmath0-hedrites , together with the smallest number of vertices , for which they appear : * the only symmetry groups of @xmath1-hedrites are point subgroups of @xmath147 , which contain @xmath151 as point subgroup , i.e. @xmath147(@xmath152 ) , @xmath153(@xmath154 ) , @xmath143(@xmath155 ) , @xmath130(@xmath138 ) , @xmath156(@xmath157 ) . *
the only symmetry groups of @xmath65-hedrites are : @xmath25(@xmath154 ) , @xmath26(@xmath158 ) , @xmath159(@xmath160 ) , @xmath117(@xmath161 ) , @xmath162(@xmath163 ) , @xmath164(@xmath165 ) . * the only symmetry groups of @xmath74-hedrites are : @xmath130(@xmath155 ) , @xmath143(@xmath138 ) and all their point subgroups , i.e. @xmath151(@xmath157 ) , @xmath166(@xmath154 ) , @xmath117(@xmath161 ) , @xmath167(@xmath168 ) , @xmath169(@xmath138 ) , @xmath170(@xmath171 ) , @xmath172(@xmath171 ) . *
the only symmetry groups of @xmath76-hedrites are point subgroups of @xmath117 , i.e. @xmath117(@xmath160 ) , @xmath169(@xmath173 ) , @xmath170(@xmath158 ) , @xmath172(@xmath173 ) . * the only symmetry groups of @xmath19-hedrites are : @xmath172(@xmath174 ) , @xmath159(@xmath175 ) , @xmath26(@xmath157 ) , @xmath117(@xmath173 ) , @xmath167(@xmath176 ) , @xmath166(@xmath177 ) , @xmath178(@xmath179 ) , @xmath156(@xmath154 ) , @xmath130(@xmath175 ) , @xmath143(@xmath180 ) , @xmath162(@xmath181 ) , @xmath182(@xmath157 ) , @xmath164(@xmath171 ) , @xmath153(@xmath181 ) , @xmath183(@xmath158 ) , @xmath147(@xmath154 ) , @xmath145(@xmath184 ) , @xmath185(@xmath138 ) .
* proof . * for @xmath1-hedrites , see @xcite .
any transformation stabilizing a @xmath3-gon , can interchange its two edges and two vertices .
so , the stabilizing point subgroup of a @xmath3-gon can be @xmath117 , @xmath159 , @xmath26 or @xmath25 only .
the unique @xmath3-gon of a @xmath76-hedrite has to be preserved by the symmetry group ; so , all possibilities are : @xmath117 , @xmath159 , @xmath26 , @xmath25 . every symmetry of an @xmath0-hedrite induces a symmetry on its @xmath3-gons and @xmath4-gons . since the stabilizer of a @xmath3-gon , @xmath4-gon has maximal size @xmath1 , @xmath74 , this imply that the order of the symmetry group of an @xmath0-hedrite is bounded from above by @xmath186 and @xmath187 .
so , in particular , the order of symmetry group of an @xmath74-hedrite is at most @xmath19 .
if @xmath188 is an element of order three , then it fixes each of two @xmath3-gons . since the stabilizer of a @xmath3-gon does not contain an element of order three , no such @xmath188 exists .
if @xmath188 is a rotation of order @xmath1 , then @xmath189 is a rotation of order @xmath3 stabilizing each @xmath3-gon ; so , the axis of @xmath188 goes through the two @xmath3-gons .
this is a contradiction . by a search in the tables of the groups
one can see that the only possibilities are : @xmath25 , @xmath159 , @xmath26 , @xmath167 , @xmath117 , @xmath166 , @xmath156 , @xmath143 , @xmath130 .
but there exists a @xmath74-hedrite for any of such symmetries , in figure [ special - i - hedrites ] and subsection [ subsection-6-hedrites ] below . for @xmath65-hedrites ,
since there are two @xmath4-gons , the maximal order of the group is @xmath190 .
the oddness of the number of @xmath3-gons excludes central symmetry , axis of order @xmath1 , and groups @xmath156 , @xmath143 , @xmath130 .
if @xmath22 is a @xmath65-hedrite with a @xmath4-fold axis then this axis goes through the two @xmath4-gons , say , @xmath109 and @xmath110 .
if one consider a belt of @xmath1-gons around @xmath109 , then , after a number @xmath191 of steps , one will encounter a @xmath3-gon and so , by symmetry , three @xmath3-gons .
so , we will have the following possibilities for @xmath192 : there is only one way to extend this graph to a @xmath65-hedrite and the obtained extension has symmetry at least @xmath162 .
so , the group is @xmath193 , @xmath164 or @xmath182 .
an @xmath19-hedrite @xmath22 has @xmath101-fold axis of rotation with @xmath194 , @xmath4 or @xmath1 .
if @xmath195 , then the axis of the rotation goes through two @xmath4-gons , say @xmath109 and @xmath196 .
if one consider , around @xmath4-gon @xmath109 , a belt of @xmath1-gons , then , after some number @xmath191 of steps , one will encounter a @xmath4-gon and so , by symmetry , three @xmath4-gons , say , @xmath197 .
adding , if necessary , belts of @xmath1-gons , one will encounter the last three @xmath4-gons , say , @xmath198 . the patch formed by the six triangles
@xmath199 has symmetry @xmath162 at least and so , @xmath22 has also this symmetry .
consequently , the symmetry of @xmath22 is @xmath162 , @xmath164 or @xmath182 .
if @xmath200 , then the axis of the rotation goes through a vertex or a @xmath1-gon .
assume , for simplicity , that this axis goes through a vertex ; then , by repeating above reasoning , one obtains two orbits of @xmath4-gons , say , @xmath201 and @xmath202 , under @xmath1-fold symmetry and the symmetry group is @xmath153 , @xmath147 , @xmath183 , @xmath145 or @xmath185 .
so , one obtains the above list of @xmath203 possible point groups .
all these groups appear in subsection [ subsection-8-hedrites ] ( groups @xmath185 , @xmath183 , @xmath164 , @xmath151 , @xmath147 , @xmath117 , @xmath182 , @xmath26 ) , in figure [ special - i - hedrites ] ( groups @xmath130 , @xmath159 , @xmath25 , @xmath153 , @xmath193 , @xmath166 , @xmath204 , @xmath178 ) and in figure [ thepureirreducibleoctahedritewith56cc ] ( groups @xmath143 , @xmath145 ) ; for all groups ( except , possibly , @xmath167 , @xmath166 and @xmath178 ) those examples of @xmath19-hedrites have smallest number of vertices .
@xmath65-hedrite * 15 - 1 * @xmath162 smallest @xmath205 @xmath74-hedrite * 30 - 1 * @xmath167 @xmath206 @xmath19-hedrite * 14 - 4 * @xmath130 smallest @xmath207 @xmath19-hedrite * 14 - 2 * @xmath159 smallest @xmath208 @xmath19-hedrite * 16 - 1 * @xmath25 smallest @xmath209 @xmath19-hedrite * 18 - 1 * @xmath153 smallest @xmath210 @xmath19-hedrite * 18 - 2 * @xmath162 smallest @xmath211 @xmath19-hedrite * 26 - 1 * @xmath166 @xmath212 @xmath19-hedrite * 46 - 1 * @xmath167 @xmath213 @xmath19-hedrite * 60 - 1 * @xmath178 @xmath214 the simplest case , @xmath108 , of @xmath0-hedrites admits following characterization for each of its five possible groups .
it has symmetry @xmath143 ( respectively , @xmath130 ) if and only if it is an @xmath88-inflation , for some @xmath215 , of @xmath48 or @xmath61 with even @xmath29 ( respectively , of @xmath127 or @xmath61 with odd @xmath29 ) .
any @xmath1-hedrite with group @xmath216 or @xmath147 has @xmath217 vertices for some @xmath218 ( the group @xmath147 corresponds to the case @xmath219 or @xmath220 ) ; it comes from the smallest @xmath1-hedrite * 2 - 1 * by goldberg - coxeter construction ( see @xcite , @xcite and @xcite ) .
all other 4-hedrites have symmetry @xmath156 ; in shift terms , they are exactly those , for which interchange of central circuits changes the value of shift .
we expect , that a @xmath76-hedrite with the highest symmetry @xmath117 exists for any @xmath221 .
here and below all links are given in rolfsen s notation ( see the table in @xcite and also , for example , @xcite ) for links with at most 9 crossings and knots with 10 crossings , or , otherwise , in dowker - thistlewhaite s numbering ( see @xcite ) , if any .
we write @xmath222 if the projection in the pictures and table below is different from the one given in corresponding cases above .
we give on the pictures below all @xmath0-hedrites with at most @xmath190 vertices , indicating under picture of each its symmetry , cc - vector and corresponding alternating link .
if an @xmath0-hedrite is @xmath3-connected but not @xmath4-connected , then we add a symbol @xmath223 just after the number .
if an @xmath0-hedrite is reducible , i.e. has a rail - road , then we add mention `` red . '' .
all @xmath0-hedrites with @xmath224 , @xmath225 and @xmath6 vertices are listed in table [ tab : i - hedrite13_14 ] .
+ & @xmath130 & @xmath375 & + * 14 - 2 * & @xmath156 & @xmath375 & + * 14 - 3@xmath231*&@xmath143 & @xmath376 red . & + + & @xmath117 & @xmath377 & @xmath378 + * 13 - 2 * & @xmath25 & @xmath377 & @xmath379 + * 13 - 3 * & @xmath26 & @xmath380 & + * 13 - 4 * & @xmath159 & @xmath381 & + & @xmath159 & @xmath382 & @xmath383 + * 14 - 2 * & @xmath25 & @xmath384 & + * 14 - 3 * & @xmath25 & @xmath385 & + * 14 - 4 * & @xmath117 & @xmath386 & + * 14 - 5 * & @xmath117 & @xmath387 red . &
+ * 14 - 6 * & @xmath117 & @xmath388 red . & + * 14 - 7 * & @xmath117 & @xmath389 red . & + & @xmath162 & @xmath390 & @xmath391 + * 15 - 2 * & @xmath117 & @xmath390 & @xmath392 + * 15 - 3 * & @xmath159 & @xmath390 & @xmath393 + * 15 - 4 * & @xmath26 & @xmath390 & @xmath394 + * 15 - 5 * & @xmath159 & @xmath395 & + * 15 - 6 * & @xmath159 & @xmath396 & + * 15 - 7 * & @xmath25 & @xmath397 & + * 15 - 8 * & @xmath117 & @xmath398 & + * 15 - 9 * & @xmath159 & @xmath399 red . &
+ * 15 - 10 * & @xmath159 & @xmath400 red . &
+ + & @xmath26 & @xmath377 & @xmath401 + * 13 - 2 * & @xmath26 & @xmath377 & @xmath402 + * 13 - 3 * & @xmath26 & @xmath377 & @xmath403 + * 13 - 4 * & @xmath159 & @xmath377 & @xmath404 + * 13 - 5 * & @xmath25 & @xmath377 & @xmath405 + * 13 - 6 * & @xmath25 & @xmath377 & @xmath406 + * 13 - 7 * & @xmath25 & @xmath377 & @xmath407 + * 13 - 8 * & @xmath25 & @xmath408 & + * 13 - 9 * & @xmath159 & @xmath409 & + * 13 - 10 * & @xmath159 & @xmath410 red . &
+ * 13 - 11 * & @xmath117 & @xmath411 red . &
+ * 13 - 12 * & @xmath117 & @xmath412 & + * 13 - 13 * & @xmath117 & @xmath413 red . & + * 13 - 14 * & @xmath159 & @xmath414 red . & + & @xmath166 & @xmath382 & @xmath415 + * 14 - 2 * & @xmath169 & @xmath382 & @xmath416 + * 14 - 3 * & @xmath26 & @xmath382 & @xmath417 + * 14 - 4 * & @xmath26 & @xmath382 & @xmath418 + * 14 - 5 * & @xmath26 & @xmath382 & @xmath419 + * 14 - 6 * & @xmath25 & @xmath382 & @xmath420 + * 14 - 7 * & @xmath25 & @xmath382 & @xmath421 + * 14 - 8 * & @xmath169 & @xmath384 & + * 14 - 9 * & @xmath26 & @xmath384 & + * 14 - 10 * & @xmath26 & @xmath422 & + * 14 - 11 * & @xmath26 & @xmath422 & + * 14 - 12 * & @xmath159 & @xmath423 & + * 14 - 13@xmath424 * & @xmath143 & @xmath375 & + * 14 - 14 * & @xmath117 & @xmath375 & + * 14 - 15 * & @xmath26 & @xmath425 & + * 14 - 16 * & @xmath170 & @xmath425 & + * 14 - 17 * & @xmath26 & @xmath426 red . &
+ * 14 - 18 * & @xmath26 & @xmath427 & + * 14 - 19 * & @xmath26 & @xmath428 & + * 14 - 20 * & @xmath143 & @xmath429 & + * 14 - 21 * & @xmath117 & @xmath430 red . &
+ * 14 - 23 * & @xmath117 & @xmath431 red . &
+ & @xmath26 & @xmath390 & @xmath432 + * 15 - 2 * & @xmath26 & @xmath390 & @xmath433 + * 15 - 3 * & @xmath26 & @xmath390 & @xmath434 + * 15 - 4 * & @xmath25 & @xmath390 & @xmath435 + * 15 - 5 * & @xmath25 & @xmath390 & @xmath436 + * 15 - 6 * & @xmath25 & @xmath390 & @xmath437 + * 15 - 7 * & @xmath25 & @xmath390 & @xmath438 + * 15 - 8 * & @xmath25 & @xmath439 & + * 15 - 9 * & @xmath25 & @xmath439 & + * 15 - 10 * & @xmath117 & @xmath440 & + * 15 - 11 * & @xmath25 & @xmath440 & + * 15 - 12 * & @xmath25 & @xmath440 & + * 15 - 13 * & @xmath159 & @xmath441 & + * 15 - 14 * & @xmath25 & @xmath442 & + * 15 - 15 * & @xmath117 & @xmath443 & + * 15 - 16 * & @xmath117 & @xmath444 & + * 15 - 17 * & @xmath159 & @xmath445 red . & + + & @xmath159 & @xmath377 & @xmath446 + * 13 - 2 * & @xmath25 & @xmath377 & @xmath447 + * 13 - 3 * & @xmath25 & @xmath448 & + * 13 - 4 * & @xmath25 & @xmath449 & + * 13 - 5 * & @xmath159 & @xmath449 & + * 13 - 6 * & @xmath117 & @xmath380 & + * 13 - 7 * & @xmath170 & @xmath380 & + & @xmath25 & @xmath382 & @xmath450 + * 14 - 2 * & @xmath25 & @xmath382 & @xmath451 + * 14 - 3 * & @xmath25 & @xmath382 & @xmath452 + * 14 - 4 * & @xmath25 & @xmath382 & @xmath453 + * 14 - 5 * & @xmath25 & @xmath384 & + * 14 - 6 * & @xmath159 & @xmath423 & + * 14 - 7 * & @xmath26 & @xmath375 & + * 14 - 8 * & @xmath159 & @xmath428 & + * 14 - 9 * & @xmath117 & @xmath428 & + & @xmath26 & @xmath390 & @xmath454 + * 15 - 2 * & @xmath25 & @xmath390 & @xmath455 + * 15 - 3 * & @xmath25 & @xmath390 & @xmath456 + * 15 - 4 * & @xmath159 & @xmath457 & + * 15 - 5 * & @xmath25 & @xmath457 & + * 15 - 6 * & @xmath25 & @xmath457 & + * 15 - 7 * & @xmath25 & @xmath442 & + * 15 - 8 * & @xmath25 & @xmath442 & + * 15 - 9 * & @xmath117 & @xmath396 & + * 15 - 10 * & @xmath25 & @xmath396 & + * 15 - 11 * & @xmath117 & @xmath458 & + * 15 - 12 * & @xmath117 & @xmath399 red . &
+ + & @xmath26 & @xmath377 & @xmath459 + * 13 - 2 * & @xmath117 & @xmath380 & + & @xmath26 & @xmath382 & @xmath460 + * 14 - 2 * & @xmath159 & @xmath384 & + * 14 - 3 * & @xmath156 & @xmath384 & + * 14 - 4 * & @xmath130 & @xmath375 & + * 14 - 5 * & @xmath26 & @xmath428 & + * 14 - 6 * & @xmath156 & @xmath427 & + * 14 - 7 * & @xmath147 & @xmath429 & + * 14 - 8 * & @xmath147 & @xmath389 red . &
+ & @xmath26 & @xmath390 & @xmath461 + * 15 - 2 * & @xmath159 & @xmath439 & + * 15 - 3 * & @xmath159 & @xmath457 & + * 15 - 4 * & @xmath26 & @xmath440 & + * 15 - 5 * & @xmath164 & @xmath462 & + h.s.m.coxeter , _ virus macromolecules and geodesic domes _ , in _ a spectrum of mathematics _ ; ed . by j.c.butcher , oxford university
press / auckland university press : oxford , u.k./auckland new - zealand , ( 1971 ) 98107 .
o.heidemeier , _ die erzeugung von 4-regulren , planaren , simplen , zusammenhngenden graphen mit vorgegebenen flchentypen _ , diplomarbeit , universitt bielefeld , fakultt fr wirtschaft und mathematik , 1998 .
a.kotzig , _ eulerian lines in finite 4-valent graphs and their transformations _ , in : theory of graphs , proceedings of a colloquium , tihany 1966 , ed . by p.erdos and g.katona , academic press , new york ( 1969 ) 219230 .
h.shank , _ the theory of left - right paths _ , in : combinatorial mathematics iii , proceedings of 3rd australian conference , st lucia 1974 , lecture notes in mathematics * 452 * , springer - verlag , berlin - new york ( 1975 ) , pp . 4254 . | call _ @xmath0-hedrite _ any @xmath1-valent @xmath2-vertex plane graph , whose faces are @xmath3- , @xmath4- and @xmath1-gons only and @xmath5 .
the edges of an @xmath0-hedrite , as of any eulerian plane graph , are partitioned by its _ central circuits _ ,
i.e. those , which are obtained by starting with an edge and continuing at each vertex by the edge opposite the entering one .
so , any @xmath0-hedrite is a projection of an alternating link , whose components correspond to its central circuits . call an @xmath0-hedrite _ irreducible _ , if it has no _ rail - road _ , i.e. a circuit of @xmath1-gonal faces , in which every @xmath1-gon is adjacent to two of its neighbors on opposite edges .
we present the list of all @xmath0-hedrites with at most @xmath6 vertices .
examples of other results : + * all @xmath0-hedrites , which are not @xmath4-connected , are identified . + * any irreducible @xmath0-hedrite has at most @xmath7 central circuits .
+ * all @xmath0-hedrites without self - intersecting central circuits are listed .
+ * all symmetry group of @xmath0-hedrites are listed . _ mathematics subject classification_. primary 52b05 , 52b10 ; secondary 05c30 , 05c10 . _ key words_. plane graphs , eulerian graphs , alternating links , point groups . |
k.f . is supported by kakenhi no.16h02211 , presto , jst , crest , jst and erato , jst .
mh is partially supported by fund for the promotion of joint international research ( fostering joint international research ) no .
the centre for quantum technologies is funded by the singapore ministry of education and the national research foundation as part of the research centres of excellence programme .
he is also grateful to dr .
michal hajdusek for helpful comments .
the error detection on the black vacuum qubits ( edges of the primal cubic lattice ) is executed as follows .
if there is no error on the graph state , the outcome @xmath109 of the @xmath24-basis measurements satisfies the condition : @xmath110 where @xmath111 indicates an addition modulo two over all black qubits adjacent to the vertex @xmath112 . depending on a given error @xmath113 ( @xmath114 ) on the graph state
, we can obtain the error syndrome @xmath115 at the vertices belonging to the defect region . from the error syndrome ,
the most likely location of the errors is estimated . here
we employ the minimum distance decoding , which can be done by finding a minimum path connecting pairs of vetices of @xmath116 with the minimum - weight - perfect - matching ( mwpm ) algorithm @xcite .
let @xmath117 be the estimated error location , where @xmath118 indicates the number of 1s in a bit string @xmath119 .
if a chain of edges specified by @xmath120 have a nontrivial cycle in the sense of the relative homology @xcite , the error correction fails . at the defect region
far from the singular qubits , a nontrivial cycle have at least length @xmath88 , which is the characteristic length of the defect determined from the required accuracy of quantum computation .
let @xmath92 be the size of the quantum computation that alice wants to do fault - tolerantly . to guarantee the accuracy of the output ,
it is enough to choose the distance @xmath121 .
therefore , the number of the qubits of the graph state is @xmath122 .
now we can define the correctable set of errors as follow : an error location @xmath123 belongs to the correctable set @xmath124 of the errors iff there exists a connected component of length @xmath88 in the chain of edges specified by @xmath120 .
the error detection and definition of the correctable error set @xmath125 on the white vacuum qubits are done in the same way but on the dual lattice . from the test @xmath78 , we know the error location @xmath123 .
since the mwpm algorithm works in polynomial time in the number of vertices with @xmath126 , we can decide whether or not @xmath123 belongs to the correctable error set @xmath127 .
the same argument also holds for the error location @xmath87 on the white vacuum qubits tested by @xmath79 .
therefore , we can efficiently check whether or not the errors on a given resource belong to @xmath128 .
we here , for simplicity , do not employ magic state distillaion @xcite but encodes each logical qubit into the reed - muller 15-qubit code .
then we perform a fault - tolerant logical @xmath83-basis measurement by transversal physical @xmath83-basis measurements on the singular qubits as done in ref @xcite .
thereby , alice can fix her strategy of quantum computation , which makes easy to define the correctable set of errors for the test .
let @xmath129 and @xmath104 be the number of concatenation levels and the number of the logical @xmath94-basis measurements , respectively .
then we need @xmath130 physical @xmath83-basis measurements , on the singular qubits .
note that @xmath131 is enough to reduce the logical error sufficiently . in the following , we the error on the graph state
is specified by @xmath132 by converting it into @xmath23 operators on the graph state , @xmath133 .
the logical @xmath83-basis measurement is done by physical transversal @xmath83-basis measurements by encoding each qubit into a concatenated reed - muller 15-qubit codes @xcite .
this is also the case for all pauli - basis measurements . in the vacuum region near the singular qubits
, we have a logical error of length smaller than @xmath88 as shown in fig .
[ fig2 ] , since they are not topologically protected .
correctable error for the fault - tolerant logical @xmath83-basis measurement is defined for a given error @xmath134 recursively as follows : at physical level , which we call level-@xmath135 , if @xmath136 or @xmath137 becomes a logical error for a singular qubit , the level-@xmath135 ( singular ) qubit is labeled to be faulty . at @xmath138th concatenation level , if the level-@xmath138 logical qubit consisting of 15 level-@xmath139 logical qubits encoded in the reed - muller 15-qubit code has two or more faulty level-@xmath139 logical qubits , the level-@xmath138 logical qubit is labeled to be faulty . at the highest level @xmath140
, if no level-@xmath129 logical qubit is faulty , the given error @xmath141 belongs to the correctable set @xmath142 .
and estimated one @xmath143 are denoted by solid and dotted lines , respectively . the vertices ( error syndrome ) of @xmath126
are denoted by red squares .
the 3d lattice(spatial two dimensions and time - like one dimension ) is depicted as if it is two dimensional ( one dimension for both spatial and time - like axe ) .
, width=188 ] let us first consider the pass probability of the test for topological protection .
the error @xmath123 is rejected if @xmath136 contains a connected component of length at least @xmath88 .
such a probability is calculated @xcite to be @xmath144 therefore , if @xmath99 is sufficiently smaller than a constant value , the rejection probability is exponentially suppressed .
next we consider the test for the logical @xmath83-basis measurement .
let @xmath145 be the probability that a level-0 ( singular ) qubit is faulty .
@xmath146 is evaluated in a similar way to the previous case for the topological protection but we have to count logical errors consist of the chains of length lower than @xmath88 : @xmath147 where @xmath148 is the number of chains of length @xmath149 that contribute to the logical error of length @xmath149 .
@xmath148 is counted in ref .
@xcite rigorously up to @xmath150 , which indicates that we can reduce @xmath145 by decreasing @xmath99 sufficiently .
the probability @xmath151 of obtaining the level-@xmath138 faulty qubit is given recursively by @xmath152 the we obtain @xmath153 the probability to obtain no faulty level-@xmath129 logical qubit at the highest level is given by @xmath154 . since @xmath104 , it is sufficient to chose @xmath157 and @xmath158 , which are independent of @xmath41 , the number of the samples of the graph state
. therefore , in the large @xmath88 limit for a given @xmath92 , we can reduce the logical error probability polynomially , and hence amplify the acceptance probability @xmath159 arbitrarily close to 1 . | quantum systems , in general , output data that can not be simulated efficiently by a classical computer , and hence is useful for solving certain mathematical problems and simulating quantum many - body systems .
this also implies , unfortunately , that verification of the output of the quantum systems is not so trivial , since predicting the output is exponentially hard . as another problem ,
quantum system is very delicate for noise and thus needs error correction .
here we propose a framework for verification of the output of fault - tolerant quantum computation in the measurement - based model .
contrast to existing analyses on fault - tolerance , we do not assume any noise model on the resource state , but an arbitrary resource state is tested by using only single - qubit measurements to verify whether the output of measurement - based quantum computation on it is correct or not .
the overhead for verification including classical processing is linear in the size of quantum computation . since
full characterization of quantum noise is exponentially hard for large - scale quantum computing systems , our framework provides an efficient way of practical verification of experimental quantum error correction .
moreover , the proposed verification scheme is also compatible to measurement - only blind quantum computation , where a client can accept the delegated quantum computation even when a quantum sever makes deviation , as long as the output is correct . _
introduction. _ quantum computation provides a new paradigm of information processing offering both fast and secure information processing , which could not be realized in classical computation @xcite .
recently , a lot of experimental efforts have been paid to realize quantum computation @xcite . there ,
fault - tolerant quantum computation with quantum error correction @xcite is inevitable to obtain quantum advantage using noisy quantum devices . due to the recent rapid progresses on experimental quantum error correction techniques
@xcite , there is an increasing demand on an efficient way of a performance analysis of fault - tolerant quantum computation . in particular , in the majority of performance analyses of fault - tolerant quantum computation , a specific noise model , such as independent and identical pauli error operation and some specific correlation models , is assumed apriori @xcite .
however , in an actual experiments more general noise occurs including general trace preserving completely positive ( tp - cp ) maps with various correlation between qubits .
further , to guarantee the correctness of the output of quantum computation , we need to care about all cases including unexpected types of errors . , i.e.,we should not assume any specific error model .
also , in our scenario , the full tomographic approach would not work efficiently for the increasingly many qubits .
unfortunately , existing fault - tolerant quantum computations have not equipped an efficient verification scheme yet .
the aim of this paper is to develop a fault - tolerant quantum computation equipping a verification scheme without assuming the underlying noise model . as properties of verifiable fault - tolerance , we require following two conditions .
one is _
detectability _ which means that if the error of a quantum computer is not correctable , such a faulty output of the quantum computation is detected with high probability .
the other is _ acceptability _ which means that an appropriately constructed quantum computer can pass the verification with high probability .
in other words , under a realistic noise model , the test accepts the quantum computation with high probability .
both properties are important to characterize performance of test in statistical hypothesis testing @xcite . in this paper , we develop verifiable fault - tolerance in measurement - based quantum computation ( mbqc ) @xcite , which satisfies both detectability and acceptability .
we take a rather different approach to fault - tolerance than conventional one .
we do not assume any noise model underlying , but define a correctable set of errors on a resource state of mbqc and test whether the error on a given resource state belongs to such a set or not .
to this end , we employ the stabilizer test proposed in ref .
@xcite , where an efficient verification of mbqc can be carried out by testing the graph state . however , this method is not fault - tolerant lacking acceptability ; any small amount of noise on the graph state causes rejection regardless whether or not it is correctable .
although the paper @xcite extended the stabilizer test to the self - testing for the measurement basis , it still has the same problem .
therefore , we crucially extend the stabilizer test @xcite for a noisy situation , so that we can decide whether the given resource states belong to a set of fault - tolerant resource states or not . under the condition of a successful pass of the test , the accuracy of fault - tolerant mbqc is guaranteed to be arbitrarily high ( i.e. , contraposition of detectability ) .
the total resource required for the verification is linear to the size of quantum computation . as a concrete example
, we explicitly define a set of correctable errors on the resource state for topologically protected mbqc @xcite , where we can show acceptability by calculating the acceptance probability concretely under a realistic noise model .
note that contrast to detectability , the requirement of acceptability is unique for the verification of fault - tolerant quantum computation . indeed , when we can expect no error like the previous case @xcite
, we do not need fault - tolerance .
so , we could correctly judge the no error case with probability @xmath0 , i.e. , acceptability of the test is trivially satisfied because the stabilizer test would be passed in the no error case .
on the other hand , in verification of fault - tolerant quantum computation consisting of many elementary parts , each of which can not be checked directly , we have to judge whether the output of the computation is correct or not carefully under an expected error model , which imposes the second requirement , acceptability .
that is , to discuss acceptability , we firstly fix our verification method and an expected error model .
then , we calculate the acceptance probability , which corresponds to power of test in statistical hypothesis testing @xcite .
we also discuss an application of verifiable fault - tolerance to verification of blind quantum computation @xcite under a quantum server s deviation or quantum channel noise . _ a general setup for fault - tolerant mbqc. _
let us consider a generic scenario of fault - tolerant mbqc on a two - colorable graph state composed of the black system @xmath1 and the white system @xmath2 , which are consist of @xmath3 and @xmath4 qubits , respectively .
then , we have two kinds of operators @xmath5 , @xmath6 , on @xmath7 , where @xmath8 .
when we restrict them to the black system @xmath1 ( the white system @xmath2 ) , we denote @xmath9 and @xmath10 by @xmath11 and @xmath12 ( @xmath13 and @xmath14 ) . by using the binary - valued adjacency matrix @xmath15 ( i.e.
, @xmath16 element is @xmath0 iff vertices @xmath17 and @xmath18 are connected ) corresponding to the graph , the graph state @xmath19 is characterized as @xmath20 for @xmath21 and @xmath22 .
this relation explains that any error on the @xmath23-basis can be converted to an error on the @xmath24-basis .
then , the total space @xmath25 is spanned by @xmath26 .
suppose we execute a fault - tolerant mbqc quantum computation on the two - colorable graph state .
then a set of correctable errors on the two - colorable graph state is defined such that an ideal state @xmath19 and erroneous one @xmath27 result in the same computational outcome under error correction .
such a set of errors can is specified as a subset @xmath28 of @xmath29 .
the projection to the subspace is written by @xmath30 .
we assume that the subset @xmath28 is written as @xmath31 by using two subsets @xmath32 and @xmath33 . _
test for verification of fault - tolerance. _ similar to ref .
@xcite , we employ the following sampling protocol to verify whether the error is correctable .
our protocol runs as follows : * honest bob generates @xmath34 , where @xmath19 is an @xmath35-qubit graph state on a bipartite graph @xmath36 , whose vertices are divided into two disjoint sets @xmath37 and @xmath38 .
( see fig .
[ fig1](a ) . )
bob sends each qubit of it one by one to alice .
evil bob can generate any @xmath39-qubit state @xmath40 instead of @xmath34 .
* alice divides @xmath41 blocks of @xmath35 qubits into three groups by random choice .
the first group consists of @xmath42 blocks of @xmath35 qubits each .
the second group consists of @xmath42 blocks of @xmath35 qubits each .
the third group consists of a single block of @xmath35 qubits . *
alice uses the third group for her computation .
other blocks are used for the test , which will be explained later . *
if alice passes the test , she accepts the result of the computation performed on the third group . for each block of the first and second groups
, alice performs the following test : * for each block of the first group , alice measures qubits of @xmath37 in the @xmath23 basis and qubits of @xmath38 in the @xmath24 basis .
then , she obtain @xmath43 and @xmath44 . if @xmath45 , then the test is passed .
* for each block of the second group , alice measures qubits of @xmath38 in the @xmath23 basis and qubits of @xmath37 in the @xmath24 basis .
then , she obtain @xmath46 and @xmath47 . if @xmath48 , then the test is passed .
_ detectability and acceptability. _ to show detectability , taking account into unexpected errors , we obtain the following theorem in the same way as @xcite : [ l1 ] assume that @xmath49 .
if the test is passed , with significance level @xmath50 , we can guarantee that the resultant state @xmath51 of the third group satisfies @xmath52 ( note that the significance level is the maximum passing probability when bob erroneously generates incorrect states so that the resultant state @xmath51 does not satisfy @xcite .
that is , @xmath53 expresses the minimum probability to detect such incorrect states . )
the previous study @xcite considers the case with @xmath54 , @xmath55 , and proves this special case by discussing the two kinds of binary events @xmath56 or @xmath57 and @xmath58 or @xmath57 . replacing these two events by the two kinds of events @xmath59 or @xmath60 and @xmath61 or @xmath62 in the proof given in @xcite
, we can show theorem 1 with the current general form . from the theorem and the relation between the fidelity and trace norm ( * ? ? ? * ( 6.106 ) ) , we can conclude the verifiability : if alice passes the test , she can guarantee that @xmath63 for any povm @xmath64 with the significance level @xmath50 .
that is , the property of fault - tolerant quantum computation guarantees that the probability that the obtained computation outcome is different from the true computation outcome is less than @xmath65 .
if we take @xmath66 , for example , this error probability is @xmath67 if @xmath68 , and therefore the verifiability is satisfied .
note that the lower bound , @xmath69 , of the significance level @xmath50 is tight , since if bob generates @xmath70 copies of the correct state @xmath71 and a single copy of a wrong state , bob can fool alice with probability @xmath72 , which corresponds to @xmath73 .
note that the above theorem on detectability holds without any assumption on the underlying noise .
note that noise in the measurements can also be taken as noise on the resource state , if it does not depnd on the measurement bases . even if it is not the case , we can add noise such that the amounts of noise are the same for all measurement bases .
next , we consider acceptability . to address the success probability under a realistic noise model , we assume a specific application of pauli channel on @xmath25 as an expected error model . that is , the error given as the distribution @xmath74 on the set @xmath75 of @xmath24-basis errors and @xmath23-basis errors . then , we denote the marginal distribution with respect to the pair of @xmath24-basis errors on @xmath38 and @xmath23-basis errors on @xmath37 ( @xmath23-basis errors on @xmath38 and @xmath24-basis errors on @xmath37 ) by @xmath76 ( @xmath77 ) .
hence , the probability that alice passes the test @xmath78 ( @xmath79 ) with one round is @xmath80 ( @xmath81 ) .
since we apply them @xmath70 rounds , the probability to be passed is @xmath82 .
hence , when the probabilities @xmath80 and @xmath81 are close to @xmath0 , alice can accept the correct computation result on the third group with high probability .
_ verifiable fault - tolerance for topologically protected mbqc. _ to show acceptability , below we will explain how to define a correctable set of the errors on a graph state .
then , for a concrete example , we will calculate the acceptance probability @xmath82 under a realistic noise model .
in the theory of fault - tolerant quantum computation , it is conventional that we translate fault - tolerance in the circuit model into fault - tolerance in the measurement - based model @xcite as follows . in the circuit model , we can define a set of correctable ( sparse ) fault paths so that the output of quantum computation does not damaged even if any error occurs on such a fault path @xcite . then , translating the correctable ( sparse ) fault paths in the circuit model into the measurement - based model , we can define a correctable set of the errors on the graph state in general . for example , the schemes in refs @xcite and refs @xcite can be viewed as measurement - based versions of circuit - based fault - tolerant schemes using the concatenated steane 7-qubit code @xcite and the surface code with the concatenated reed - muller 15-qubit code @xcite , respectively .
-basis .
singular qubits are located in - between two defect regions , which are measured in the @xmath83-basis for a transversal logical @xmath83-basis measurement .
other regions are vacuum , where qubits are measured in the @xmath24-basis to obtain the error syndrome . ,
width=245 ] let us see a concrete example by using topologically protected mbqc @xcite , which has been employed as a standard framework for fault - tolerant mbqc recently @xcite .
here we focus on the original scheme proposed in ref .
@xcite , where the surface code and the concatenated reed - muller code are employed to perform two - qubit clifford gate and single - qubit non - pauli - basis measurements , respectively . in the following we will briefly sketch how the correctable set are defined . a detailed description is shown in appendix [ app1 ] . in the following ,
we characterize the correctable sets of errors @xmath84 and @xmath85 .
the errors specified by the set @xmath84 , which correspond to @xmath24 basis ( the pauli-@xmath23 operator ) on black qubits and @xmath23 basis ( the pauli-@xmath24 operator ) on white qubits , are detected on the priaml cubic lattice consisting of the edges on which the black qubits are located as shown in fig .
[ fig1](b ) .
then , the error configuraion @xmath86 can be associated with a set of edges on the primal cubic lattice .
similarly , the errors in the set @xmath85 is detected on the dual cubic lattice and the error configuration @xmath87 is associated with a set of edges on the dual cubic lattice . in the following ,
all arguments are made independently of black ( primal lattice ) and white ( dual lattice ) qubits . depending on quantum computation that alice wants to do fault - tolerantly ,
a measurement pattern is determined .
specifically , from an analogy of topological quantum computation @xcite , the sets of qubits measured in @xmath24 , @xmath23 , and @xmath83-bases @xcite are called defect , vacuum , and singular qubits , respectively . as shown in fig .
[ fig2 ] , defect qubits shape tubes , which represent logical degrees of freedom ; at each time slice they corresponds to the surface code with defects . by braiding the defects ,
a clifford two - qubit gate can be performed . for the surface code
, the minimum distance decoding can be done by finding a shortest path connecting the boundary of the error chain on the cubic lattice .
then , if the minimum distance decoding results in a logical operator of a weight ( distance ) larger than the code distance @xmath88 by wrapping around a defect or connecting two different defects , such an error is uncorrectable ( see appendix [ app1 ] for the detail ) .
accordingly , we can define @xmath89 for @xmath90 as the complement of them .
the code distance @xmath88 is chosen to be @xmath91 with @xmath92 being the size of the quantum computation that alice wants to do fault - tolerantly .
therefore , the number of qubits of the graph state is given by @xmath93 .
around the singular qubits , we still have a logical error of a weight lower than @xmath88 as shown in fig .
[ fig2 ] .
such a logical error is corrected by using another code , the concatenated reed - muller code . to this end ,
the fault - tolerant clifford gates using the surface code are further employed to encode the logical qubits into concatenated reed - muller codes , on which we can implement all pauli - bases and @xmath94-basis measurements transversally .
the corresponding physical @xmath94-basis measurements , i.e. , measurements on the singular qubits , are depicted by red circles in fig .
[ fig2 ] .
then we can define the correctable set @xmath95 of the errors for the concatenated reed - muller code recursively for @xmath90 as done in ref .
@xcite ( see appendix [ app2 ] for the detail ) .
since we employ two types of error correction codes as seen above , the correctable set of the errors are defined as an intersection of the correctable sets @xmath89 and @xmath95 for the surface code and the concatenated reed - muller code , respectively , for both colors @xmath90 .
since both minimum distance decoding for the surface code and recursive decoding for the concatenated code can be done efficiently , we can efficiently decide whether a given error pattern @xmath96 ( @xmath97 ) are in @xmath84 ( @xmath85 ) or not .
_ acceptance probability under a typical error model. _ to calculate the acceptance probability , we assume , for simplicity , the errors @xmath10 ( @xmath98 ) are distributed independently and identically for each qubit with probability @xmath99 .
it is straightforward to generalize the following argument to any local cptp noise as long as the noise strength measured by the diamond norm is sufficiently smaller than a certain threshold value @xcite .
then the standard counting argument of the self - avoiding walk for the surface code @xcite tells us that @xmath100 for @xmath90 .
apparently , if @xmath99 is sufficiently smaller than a certain constant value , @xmath101 converges to 1 for @xmath90 . by considering a recursive decoding of the concatenated code ,
we obtain @xmath102^m,\end{aligned}\ ] ] for @xmath90 where @xmath103 is a logical error probability of a weight lower than @xmath88 , which occurs around the singular qubits .
such a logical probability is also calculated as a function of the physical error probaility @xmath99 by counting the number of self - avoiding walk @xcite as show in appendix [ app3 ] eq .
( [ saw ] ) .
the integer @xmath104 and @xmath105 are the numbers of the logical @xmath94-basis measurements and the number of concatenation , respectively . again by using counting the number of self - avoiding walks @xcite we can evaluate @xmath103 . by choosing @xmath99 smaller than a certain constant value
, @xmath103 becomes sufficiently small so that @xmath106 converges to 1 for @xmath90 .
since @xmath107 for @xmath90 , the probability @xmath108 also converges to 1 exponentially in the large @xmath88 limit , if the physical error probability @xmath99 is smaller than a certain constant threshold value ( see appendix [ app3 ] for the detailed calculation ) . since
@xmath88 can be chosen independently of @xmath42 , the acceptance probability @xmath82 converges to 1 .
_ verifiable blind quantum computation. _ a promising application of the proposed framework is verification of measurement - only blind quantum computation @xcite .
suppose a quantum server generates two - colorable graph states and sends them to a client who execute universal quantum computation by only single - qubit measurements , where client employ the proposed verification .
first , our protocol is a one - way quantum communication from bob to alice , and therefore , the blindness is guaranteed by the no - signaling principle as in the protocol of ref .
@xcite , which contrasts to verifiable blind quantum computation @xcite of bfk ( broadbent - fitzsimons - kashefi ) type @xcite .
according to theorem [ l1 ] ( detectability ) under the condition of acceptance the accuracy of the output is guaranteed .
contrast to the earlier verifiable blind quantum computation @xcite , by virtue of acceptability , the proposed verification scheme can accept the delegated quantum computation even under quantum server s deviation or quantum channel noise as long as they are correctable . in this way , we can verify the quantum server is honest enough to obtain a correct output by only using single - qubit measurements .
it would be interesting to apply the proposed framework to quantum interactive proof systems @xcite . |
let @xmath0 denote the random @xmath1 matrix @xmath2 , where @xmath3 are independent copies of a given complex valued random variable @xmath4 with mean zero and unit variance : @xmath5=0\quad\text{and}\quad{\mathbb{e}}\big[|{\mathbf x}|^2\big]=1.\ ] ] let @xmath6 denote the spectral radius of @xmath0 : @xmath7 the well known circular law states that , in probability , the empirical distribution of the eigenvalues of @xmath8 weakly converges to the uniform law on the unit disc of the complex plane @xcite .
in particular , it follows that with high probability @xmath9 for any @xmath10 and large enough @xmath11 . here and below
we say that a sequence of events holds with high probability if their probabilities converge to one .
the corresponding upper bound on @xmath6 has been established by bai and yin @xcite under a finite fourth moment assumption : if @xmath12<\infty$ ] , then with high probability @xmath13 , for any @xmath10 and large enough @xmath11 ; see also geman and hwang @xcite and geman @xcite for an independent proof under stronger assumptions .
together with , this says that if @xmath14<\infty$ ] then , in probability , @xmath15 , as @xmath16 .
we refer to @xcite and references therein for related estimates and more background and applications concerning the spectral radius of a random matrix .
surprisingly , there seems to be little or no discussion at all in the literature even in the recent works @xcite and @xcite about the necessity of the fourth moment assumption for the behavior @xmath17 .
we propose the following conjecture , which is illustrated by figure [ fig1 ] .
[ conju ] the convergence in probability @xmath18 holds under the sole assumptions .
another way to put this is to say that there are no outliers in the circular law .
this phenomenon reveals a striking contrast between eigenvalues and singular values of @xmath0 , the latter exhibiting poisson distributed outliers in absence of a fourth moment , see for instance @xcite .
a tentative heuristic explanation of this phenomenon may proceed as follows .
suppose @xmath4 has a heavy tail of index @xmath19 , that is @xmath20 , as @xmath21 . if @xmath22 , then with high probability in the matrix @xmath23 there are elements @xmath3 with @xmath24 , for any @xmath25 .
any such element is sufficient to produce a singular value diverging as fast as @xmath26 . on the other hand , to create a large eigenvalue , a single large entry is not sufficient .
roughly speaking one rather needs at least one sequence of indices @xmath27 with @xmath28 with a large product @xmath29 , i.e. one cycle with a large weight if we view the matrix as an adjacency matrix of an oriented and weighted graph .
it is not difficult to see that the sparse matrix consisting of all entries @xmath3 with @xmath24 is acyclic with high probability , as long as @xmath30 .
somewhat similar phenomena should be expected for heavy tails with index @xmath31 . as shown in @xcite , in that case the circular law
must be replaced by a new limiting law @xmath32 in the complex plane .
more precisely , the empirical distribution of the eigenvalues of @xmath33 tends weakly as @xmath16 to a rotationally invariant light tailed law @xmath32 , while the empirical distribution of the singular values of @xmath33 tends weakly as @xmath16 to a heavy tailed law @xmath34 . by the above reasoning
, no significant outliers should appear in the spectrum .
the precise analogue of in this case is however less obvious since the support of @xmath32 is unbounded . from the tail of @xmath32
, one might expect that the spectral radius is of order @xmath35 while typical eigenvalues are of order @xmath36 .
in this paper we prove that the conjectured behavior holds if @xmath4 is symmetric and has a finite moment of order @xmath37 for an arbitrary @xmath38 .
we say that @xmath4 is symmetric if the law of @xmath4 coincides with the law of @xmath39 .
[ main * ] suppose that @xmath4 is symmetric and that @xmath40=1 $ ] .
suppose further that @xmath41<\infty $ ] for some @xmath38 .
then , in probability , @xmath42 in view of , to prove the theorem one only needs to establish the upper bound @xmath43 with high probability , for every @xmath10 .
we shall prove the following stronger non - asymptotic estimate , covering variables @xmath4 whose law may depend on @xmath11 .
[ main ] for any @xmath44 and @xmath45 , there exists a constant @xmath46 such that for any @xmath47 , for any symmetric complex random variable @xmath4 with @xmath40{\leqslant}1 $ ] and @xmath41{\leqslant}b$ ] , we have @xmath48 the rest of this note is concerned with the proof of theorem [ main ] .
we finish this introduction with a brief overview of the main arguments involved .
the proof of theorem [ main ] combines the classical method of moments with a novel cycle weight truncation technique . for lightness of notation
, we write @xmath49 instead of @xmath0 .
the starting point is a standard general bound on @xmath50 in terms of the trace of a product of powers of @xmath49 and @xmath51 .
let @xmath52 denote the operator norm of @xmath49 , that is the maximal eigenvalue of @xmath53 , which is also the largest singular value of @xmath49 .
recall the weyl inequality @xmath54 . for any integer @xmath55
one has @xmath56 it follows that for any integer @xmath57 , setting @xmath58 , @xmath59_{i , j}[(x^*)^{k-1}]_{j , i}.\ ] ] expanding the summands in one obtains @xmath60 where the internal sum ranges over all paths @xmath61 and @xmath62 of length @xmath63 from @xmath64 to @xmath65 , the weight @xmath66 of a path @xmath67 is defined by @xmath68 and @xmath69 denotes the complex conjugate of @xmath66 .
so far we have not used any specific form of the matrix entries . as a warm up , it may be instructive to analyze the following simple special case .
assume that @xmath3 has the distribution @xmath70 where @xmath71 $ ] is a parameter that may depend on @xmath11 , while @xmath72 is a fixed small constant .
if @xmath73 , then we have a uniformly random @xmath74 matrix , while if @xmath75 , @xmath16 one has a matrix that may serve as a toy model for the sparse matrices from the heuristic discussion given above .
notice that the assumptions of theorem [ main ] are satisfied with the same parameter @xmath76 and with @xmath77 , since @xmath78=q^{{\varepsilon}}\quad\text{and}\quad{\mathbb{e}}[|{\mathbf x}|^{2+{\varepsilon}}]{\leqslant}q^{{\varepsilon}/2}.\ ] ] we can now take expectation in . using the symmetry of @xmath4
we may restrict the sum over paths @xmath79 satisfying the constraint that in the union @xmath80 each directed edge @xmath81 appears an even number of times .
we say that @xmath80 is even .
in this case @xmath82= q^{-(1-{\varepsilon})(k-1)}q^{n}$ ] , where @xmath83 is the number of edges in @xmath80 without counting multiplicities .
let @xmath84 denote the closed path obtained as follows : start at @xmath64 , follow @xmath61 , then add the edge @xmath85 , then follow @xmath62 , then end with the edge @xmath85 again .
thus , @xmath84 is an even closed path of length @xmath86 .
notice that @xmath87 { \leqslant}q^{-{\varepsilon}}{\mathbb{e}}[w(p)].\ ] ] since the map @xmath88 is injective we have obtained @xmath89{\leqslant}q^{-{\varepsilon}}\sum_{p}{\mathbb{e}}[w(p)],\ ] ] where the sum ranges over all even closed paths of length @xmath86 .
observe that @xmath90{\leqslant}q^{-(1-{\varepsilon})k}q^{\ell},\ ] ] where @xmath91 is the number of distinct vertices in @xmath84 .
therefore , letting @xmath92 denote the number of even closed paths of length @xmath86 with @xmath91 vertices , is bounded above by @xmath93 combinatorial estimates to be derived below , see lemma [ paths and graphs ] and lemma [ graphs counting ] , imply that @xmath94 .
putting all together we have found @xmath95{\leqslant}k^2 n^{k}\sum_{\ell=1}^k a(k , n , q)^{k-\ell}\ ] ] where @xmath96 .
we choose @xmath97 .
suppose that @xmath98 .
then @xmath99 and therefore @xmath100 if @xmath11 is large enough .
it follows that @xmath101{\leqslant}k^3 n^k$ ] , and by markov s inequality , for all fixed @xmath10 : @xmath102\nonumber \\ & { \leqslant}(1+\delta)^{-2k+2}k^3n.\end{aligned}\ ] ] since @xmath97 this vanishes faster than @xmath103 for any @xmath104 . on the other hand ,
if @xmath105 , then a different , simpler argument can be used . indeed ,
since an acyclic matrix is nilpotent , it follows that if @xmath106 then there must exist a cycle with nonzero entries from the matrix @xmath49 .
the probability of a given such cycle is @xmath107 where @xmath91 is the number of vertices of the cycle .
estimating by @xmath108 the number of cycles with @xmath91 vertices one has @xmath109{\leqslant}\sum_{\ell=1}^\infty ( qn)^\ell.\ ] ] thus , if @xmath105 , then @xmath110{\leqslant}2qn{\leqslant}2n^{-{\varepsilon}}$ ] .
this concludes the proof of in the special case of the model . the given argument displays , albeit in a strongly simplified form , some of the main features of the proof of theorem [ main ] : the role of symmetry , the role of combinatorics , and the fact that cycles with too high weights have to be ruled out with a separate probabilistic estimate . the latter point requires a much more careful handling in the general case .
since it represents the main technical novelty of this work , let us briefly illustrate the main idea here .
consider the collection @xmath111 of all possible oriented cycles with @xmath112 edges of the form @xmath113 with @xmath114 , and with no repeated vertex except for @xmath115 .
let @xmath116 denote the uniform distribution over the set @xmath111 .
given the matrix @xmath0 , we look at the weight @xmath117 corresponding to the cycle @xmath118 repeated @xmath119 times , where @xmath120 is defined in . since one can restrict to even closed paths , and each such path can be decomposed into cycles that are repeated an even number of times , it is crucial to estimate the empirical averages @xmath121 = \frac1{|{\mathcal{c}}_m|}\sum_{c\sim{\mathcal{c}}_m}|w(c)|^{2t},\ ] ] where the sum runs over all cycles with @xmath112 edges and @xmath122 denotes the total number of them . broadly speaking , we will define an event @xmath123 by requiring that @xmath124{\leqslant}k^2\,,\quad \text{and } \quad\nu_m[|w(c)|^{2+{\varepsilon}}]{\leqslant}k^2 b^m,\ ] ] for all @xmath125 , where as before @xmath126 .
the assumptions of theorem [ main ] ensure that @xmath123 has large probability by a first moment argument .
thus , in computing the expected values of @xmath66 we may now condition on the event @xmath123 .
actually , on the event @xmath123 we will be able to estimate deterministically the quantities @xmath127 $ ] . to see this ,
observe that if @xmath128 denotes the maximum weight for a cycle with @xmath112 edges , then @xmath129 if @xmath76 is small enough , on the event @xmath123 , from one has @xmath130 .
since @xmath131 , a simple iteration proves that for any @xmath132 : @xmath133{\leqslant}(k^2n^mb^m)^{t(1-{\varepsilon}/4 ) } { \leqslant}n^{mt(1-{\varepsilon}/8)},\ ] ] for all @xmath11 large enough .
the bound turns out to be sufficient to handle all paths @xmath84 of the form of a cycle @xmath134 repeated @xmath119 times , for all @xmath125 .
to control more general even closed paths @xmath84 one needs a more careful analysis involving the estimate of larger empirical averages corresponding to various distinct cycles at the same time .
we refer to section [ statistics ] below for the details .
the combinatorial estimates are worked out in section [ combi ] .
finally , in section [ mainproof ] we complete the proof of theorem [ main ] .
we first introduce the basic graph theoretic terminology and then prove some combinatorial estimates . for each natural @xmath11 , @xmath135 $ ] denotes the set @xmath136 . a directed graph , or simply digraph , on @xmath135 $ ] , is a pair @xmath137 , where @xmath138 $ ] is the set of vertices and @xmath139\times[n]$ ] is the set of directed edges .
we also consider multisets @xmath140 , where a directed edge @xmath141 appears with its own multiplicity @xmath142 . in this case
we say that @xmath143 is a _
multi digraph_. given a vertex @xmath144 of a multi digraph , the out - degree @xmath145 is the number of edges of the form @xmath146 , counting multiplicities .
similarly , the in - degree @xmath147 is the number of edges of the form @xmath148 , counting multiplicities .
notice that each loop of the form @xmath149 is counted once both in @xmath150 and @xmath147 .
\(1 ) 1 ; ( 2 ) [ right of=1 ] 2 ; ( 3 ) [ below of=2 ] 3 ; ( 4 ) [ right of=3 ] 4 ; ( 5 ) [ right of=4 ] 1 ; ( 6 ) [ right of=5 ] 2 ; ( 7 ) [ above of=6 ] 3 ; ( 8) [ right of=6 ] 4 ; ( 1 ) edge [ bend right=10 ] node ( 2 ) edge [ bend left=10 ] node ( 2 ) edge [ bend right=10 ] node ( 3 ) edge [ loop left ] node ( 1 ) ( 2 ) edge node ( 3 ) edge [ loop right ] node ( 2 ) edge node ( 4 ) ( 3 ) edge [ bend right=10 ] node ( 1 ) ( 4 ) edge node ( 3 ) ; ( 5 ) edge [ bend right=10 ] node ( 6 ) edge [ bend left=10 ] node ( 6 ) \(6 ) edge [ bend left=10 ] node ( 7 ) edge [ bend left=30 ] node ( 7 ) edge [ bend right=10 ] node ( 8) edge [ bend left=10 ] node ( 8) ( 7 ) edge [ bend right=10 ] node ( 5 ) edge [ bend left=10 ] node ( 5 ) edge [ bend left=10 ] node ( 6 ) edge [ bend left=30 ] node ( 6 ) ( 8) edge [ bend right=10 ] node ( 7 ) edge [ bend left=10 ] node ( 7 ) ; [ fig : fig1 ] given natural @xmath112 , a _ path _ of length @xmath112 is a sequence @xmath151^{m+1}$ ] .
the path @xmath84 is _ closed _ if the first and the last vertex coincide .
each path @xmath152 naturally generates a multi digraph @xmath153 , where @xmath154 and @xmath140 contains the edge @xmath155 with multiplicity @xmath83 if and only if the path @xmath84 contains exactly @xmath83 times the adjacent pair @xmath155 .
notice that in general there is more than one path generating the same multi digraph .
if the path @xmath84 is closed , then @xmath156 is strongly connected , that is for any @xmath157 one can travel from @xmath158 to @xmath144 by following edges from @xmath140 . a closed path without repeated vertices except for the first and last vertices
is called a _ cycle_. a loop @xmath159
is considered a cycle of length @xmath160 .
a multi digraph will be called a _ double cycle _ if it is obtained by repeating two times a given cycle . in particular ,
a double cycle is not allowed to have loops unless its vertex set consists of just one vertex .
we say that @xmath84 is an _ even path _ if it is closed and every adjacent pair @xmath155 is repeated in @xmath84 an even number of times .
a multi digraph is called an _ even digraph _ if it is generated by an even path ; see figure [ fig : fig1 ] for an example .
thus , an even digraph is always strongly connected .
the following lemma can be proved by adapting the classical theorems of euler and veblen .
[ veblen ] for a strongly connected multi digraph @xmath161 , the following are equivalent : 1 .
@xmath161 is an even digraph ; 2 .
@xmath162 is even for every vertex @xmath144 ; 3 .
@xmath161 can be partitioned into a collection of double cycles .
two multi digraphs @xmath143 and @xmath163 are called _ isomorphic _ if there is a bijection @xmath164 such that @xmath165 if and only if @xmath166 and the multiplicities of the corresponding edges coincide .
the associated equivalence classes are regarded as unlabeled multi digraphs .
given an unlabeled multi digraph @xmath167 , we will write @xmath168 for any multi digraph @xmath161 belonging to the class @xmath167 . an edge - rooted multi digraph @xmath169 , or simply a _ rooted digraph _ , is defined as a multi digraph with a distinguished directed edge @xmath170 . the definition of equivalence classes is extended to rooted digraphs as follows .
two rooted digraphs @xmath171 and @xmath172 are called isomorphic if there is a bijection @xmath164 such that @xmath165 if and only if @xmath166 , multiplicities of corresponding edges coincide , and @xmath173 . with minor abuse of notation
we will use the same terminology as above , and write @xmath168 for rooted digraphs @xmath161 belonging to the equivalence class @xmath167 .
we turn to an estimate on the number of paths generating a given even digraph .
let @xmath143 be an even digraph with @xmath174 edges .
unless otherwise specified , multiplicities are always included in the edge count @xmath175 . by lemma [ veblen ] every vertex @xmath144 has even in- and out - degrees satisfying @xmath176 thus @xmath161 has at most @xmath177 vertices .
moreover , since the number of edges in @xmath161 is @xmath86 , we have @xmath178 [ paths and graphs ] let @xmath143 be an even digraph with @xmath174 and @xmath179
. the number of paths generating @xmath161 does not exceed @xmath180 there are @xmath91 possibilities for the starting points of the path .
the path is then characterized by the order in which neighboring vertices are visited . at each vertex @xmath144 , there are @xmath145 visits , and at most @xmath181 out - neighbors .
if @xmath182 , there is only one possible choice for the next neighbor . if @xmath183 , then there are at most @xmath184 possible choices considering all visits to the vertex @xmath144 .
hence , the number of paths generating @xmath161 is bounded by @xmath185 where we have used that the product of factorials does not exceed the factorial of the sum .
now , let @xmath186 be the number of vertices @xmath144 such that @xmath183 . from , we have @xmath187 estimating the sum in from below by @xmath188 one has @xmath189 hence
, @xmath190 using in one finds @xmath191 for integers @xmath192 , let @xmath193 be the set of rooted even digraphs @xmath143 with @xmath138 $ ] such that @xmath179 and @xmath174 .
[ graphs counting ] for any @xmath194 , @xmath195 , the cardinality of @xmath193 satisfies @xmath196 we first choose @xmath91 vertices among @xmath11 .
there are @xmath197 choices . without loss of generality
we assume that the set of vertices is given by @xmath198 .
next , we assign an admissible degree to each vertex of @xmath198 .
let @xmath199 be defined as @xmath200 .
in view of and , one has @xmath201 and @xmath202 .
thus there are @xmath203 choices for the vector @xmath204 .
next , we need to count the number of multi digraphs with the given degree sequence . to this end
, we may use the configuration model .
namely , we think of every vertex @xmath65 as having @xmath205 heads and @xmath205 tails . altogether
, there will be @xmath177 heads and @xmath177 tails .
each head is thought of as a pair of loose out - edges ( without an assigned out - neighbor ) while each tail is thought of as a pair of loose in - edges ( without an assigned in - neighbor ) .
the number of multi digraphs with the given degree sequence is bounded by the number of bipartite matchings of heads and tails , which gives @xmath206 possible choices .
thus , using @xmath207 , we see that the total number of even multi digraphs with @xmath91 vertices and @xmath86 edges is bounded above by @xmath208 it remains to choose the root edge .
since there are at most @xmath177 choices , the proof is complete .
every edge @xmath209\times[n]$ ] is given the random weight @xmath3 , where @xmath3 are independent copies of a random variable @xmath4 satisfying the assumptions of theorem [ main ] .
the weight of an even digraph @xmath143 , is defined as @xmath210 where each edge @xmath165 has multiplicity @xmath211 .
note that in this formula we interpret `` @xmath165 '' without taking into account the multiplicity in the multiset @xmath140 .
given an unlabeled even graph @xmath167 , consider the equivalence class of even digraphs @xmath212 .
we are interested in estimating @xmath213 for @xmath214 moreover , we define @xmath215 we refer to @xmath216 as the _ statistics _ of the unlabeled even digraph @xmath167 .
we extend the above definitions to rooted even digraphs as follows .
the weight of a rooted even digraph @xmath169 is defined by @xmath217 note that @xmath218 is well defined even if @xmath219 since the root edge @xmath220 satisfies @xmath221 and thus @xmath222 .
if @xmath167 is an unlabeled rooted even digraph , that is an equivalence class of rooted even digraphs , then @xmath223 and @xmath224 are defined as in and , provided @xmath225 is replaced by @xmath226 in that expression .
estimates for the statistics @xmath224 will be derived from a basic estimate for double cycles .
let @xmath111 be the unlabeled double cycle with @xmath227 edges .
similarly , @xmath228 will denote the unlabeled rooted double cycle with @xmath227 edges . from the assumptions of theorem [ main ] , for any double cycle @xmath229 we have @xmath230{\leqslant}1\,,\quad \mathbb{e}[p(c)^{1+\varepsilon/2}]{\leqslant}b^m.\ ] ] note that the same bounds apply for any rooted double cycle @xmath231 , with the weights @xmath232 replaced by @xmath233 .
[ cycle stats ] for any @xmath234 , and @xmath125 , define the event @xmath235 where @xmath236
then @xmath237 for any @xmath238 one has @xmath239 take any @xmath134 .
the first inequality in yields @xmath240 taking the expectation , implies @xmath241 on the other hand , by symmetry any @xmath134 satisfies @xmath242.\ ] ] hence , from markov s inequality and a union bound over @xmath243 , one has @xmath244 for all @xmath125 .
next , as in one shows that @xmath245 then and imply @xmath246 = \sum\limits_{h=0}^\infty2^{h+h\varepsilon/2}{\mathbb{p}}(p(c){\geqslant}2^h ) { \leqslant}2\ , { \mathbb{e}}\left [ p(c)^{1+{\varepsilon}/2}\right ] { \leqslant}2b^m.\ ] ] therefore , from markov s inequality and a union bound over @xmath243 , @xmath247 finally , we observe that the same argument leading to can be repeated for rooted cycles , with no modifications .
it follows that @xmath248 from - and the union bound over @xmath249 , it follows that @xmath237 in the remainder of this section , on the event @xmath250 , we will deterministically upper bound the statistics of any unlabeled rooted even digraph ; see proposition [ exp prop ] below .
the proof will use the following induction statement .
[ induction ] fix integers @xmath251 .
let @xmath252 be an unlabeled rooted even digraph with at most @xmath177 vertices and assume that @xmath253 can be decomposed as @xmath254 for some unlabeled rooted even digraph @xmath167 and a double cycle @xmath111 of length @xmath227 having @xmath255 common vertices with @xmath167 .
suppose that @xmath250 holds
. then 1 .
@xmath256 ; 2 . if @xmath257 , then @xmath258 .
fix an even rooted digraph @xmath259 and denote by @xmath229 and @xmath168 , respectively , the double cycle with @xmath227 edges and the even rooted digraph isomorphic to @xmath167 so that @xmath260 .
further , let @xmath261 be a uniform random permutation of @xmath135 $ ] , which we assume to be defined on a different probability space .
any permutation induces a mapping on rooted digraphs via vertex relabeling , so that the rooted digraph @xmath262 $ ] is uniformly distributed on the set @xmath263 .
hence we may write @xmath264){\geqslant}2^h),\;\;h=0,1,\dots \end{gathered}\ ] ] where @xmath265 denotes the probability w.r.t . the random permutation @xmath261 . for any @xmath266 ,
@xmath267 using this and @xmath268)=p_r(\pi[g])\,p(\pi[c])$ ] , one may estimate @xmath269){\geqslant}2^h\right ) & { \leqslant}\sum\limits_{\ell=1}^h{\mathbb{p}}_\pi\left(p_r(\pi[g]){\geqslant}2^{h-\ell};\,p(\pi[c]){\geqslant}2^{\ell-1}\right)\nonumber\\ & \qquad + \,{\mathbb{p}}_\pi\left(p(\pi[c]){\geqslant}2^{h}\right)+{\mathbb{p}}_\pi\left(p_r(\pi[g]){\geqslant}2^{h}\right ) .
\end{aligned}\ ] ] let us condition on a fixed realization @xmath270 of @xmath261 restricted to the vertices @xmath271 of @xmath161 .
thus , @xmath272 represents a uniform average over all permutations that agree with the given @xmath270 on @xmath271 .
we write @xmath273 for any digraph @xmath274 that has the form @xmath275 $ ] for some @xmath261 that agrees with @xmath270 on @xmath271 .
since @xmath118 has @xmath276 free vertices ( those which do not fall into @xmath271 ) , and we can pick them among @xmath277 available vertices , the cardinality of @xmath278 is at least @xmath279 where we use that the total number of vertices satisfies @xmath280 .
since the number of double cycles of length @xmath227 is @xmath281 , we can write for any @xmath282 : @xmath283){\geqslant}\tau\,|\,r ) & = \frac{|\{c'\sim(c;r):\,p(c'){\geqslant}\tau\}|}{|\{c'\sim(c;r)\}|}\\ & { \leqslant}(n - k)^{r - m}|\{c'\sim{\mathcal{c}}_m:\,p(c'){\geqslant}\tau\}| \\&{\leqslant}(n - k)^{r - m}n^m{\mathbb{p}}_\pi(p(\pi[c]){\geqslant}\tau ) { \leqslant}en^r{\mathbb{p}}_\pi(p(\pi[c]){\geqslant}\tau),\end{aligned}\ ] ] where we use @xmath284 to bound @xmath285 . since
the above estimate is uniform over the realization @xmath270 , for any @xmath286 we have @xmath287){\geqslant}2^{h-\ell};\;p(\pi[c]){\geqslant}2^{\ell-1}\right)\\ & \qquad { \leqslant}{\mathbb{p}}_\pi\left ( p_r(\pi[g]){\geqslant}2^{h-\ell}\right)\,\sup\limits_r\,{\mathbb{p}}_\pi\left(\pi[c]{\geqslant}2^{\ell -1}\,|\,r\right ) \\ & \qquad { \leqslant}en^r{\mathbb{p}}_\pi\left(p_r(\pi[g]){\geqslant}2^{h-\ell}\right){\mathbb{p}}_\pi\left(p(\pi[c]){\geqslant}2^{\ell-1}\right ) .
\end{aligned}\ ] ] using the definition of @xmath224 and the identity applied to @xmath161 and @xmath118 we obtain , for all @xmath288 : @xmath289){\geqslant}2^{h-\ell};\;p(\pi[c]){\geqslant}2^{\ell-1}\right ) { \leqslant}en^r2^{1-h}{\mathcal{s}}({\mathcal{u}}){\mathcal{s}}_{\ell-1}({\mathcal{c}}_m ) .
\end{aligned}\ ] ] from one has @xmath290){\geqslant}2^h)&{\leqslant}en^r2^{1-h}{\mathcal{s}}({\mathcal{u}})\sum\limits_{\ell=0}^{h-1}{\mathcal{s}}_{\ell}({\mathcal{c}}_m)+
2^{-h}{\mathcal{s}}_h({\mathcal{c}}_m ) + 2^{-h}{\mathcal{s}}({\mathcal{u } } ) . \end{aligned}\ ] ] since @xmath291 , on the event @xmath250 of lemma [ cycle stats ] one can estimate @xmath292){\geqslant}2^h){\leqslant}2en^r{\mathcal{s}}({\mathcal{u}})\sum\limits_{\ell=0}^{\infty } { \mathcal{s}}_{\ell}({\mathcal{c}}_m ) + { \mathcal{s}}({\mathcal{u } } ) { \leqslant}3ek^2n^r{\mathcal{s}}({\mathcal{u}}).\ ] ] taking the supremum over @xmath293 , the above relation proves the first assertion of the lemma .
let us prove the second assertion . on the event @xmath250 of lemma [ cycle stats
] , for any @xmath294 , @xmath295 fix @xmath296 .
if @xmath257 , then @xmath297 estimating as in for all @xmath298 , we obtain @xmath299){\geqslant}2^{h-\ell};\;p(\pi[c]){\geqslant}2^{\ell-1}\right ) { \leqslant}2^{-h+1}ek^2{\mathcal{s}}({\mathcal{u}})n^{r(1-\varepsilon/8)}.\ ] ] on the other hand , using @xmath300){\geqslant}2^{h-\ell}\right){\leqslant}2^{-h+\ell}{\mathcal{s}}({\mathcal{u}})$ ] , we find @xmath301){\geqslant}2^{h-\ell};\;p(\pi[c]){\geqslant}2^{\ell-1 } ) { \leqslant}2^{-h}{\mathcal{s}}({\mathcal{u}})2^{t+1 } { \leqslant}2^{-h+2}{\mathcal{s}}({\mathcal{u}})n^{r(1-\varepsilon/8)}.\ ] ] from it follows that @xmath290){\geqslant}2^h ) & { \leqslant}2^{-h+2}ek^2{\mathcal{s}}({\mathcal{u}})n^{r(1-\varepsilon/8 ) } + 2^{-h}{\mathcal{s}}_h({\mathcal{c}}_m)+2^{-h}{\mathcal{s}}({\mathcal{u } } ) .
\end{aligned}\ ] ] on the event @xmath250 one has @xmath302 , and therefore @xmath303){\geqslant}2^h)&{\leqslant}5ek^2 n^{r(1-\varepsilon/8 ) } \mathcal{s}({\mathcal{u } } ) .
\end{aligned}\ ] ] taking the supremum over @xmath293 , we obtain the second assertion of the lemma .
we turn to the main statement of this section [ exp prop ] suppose @xmath304 , and let @xmath167 be an unlabeled rooted even graph with @xmath86 edges and @xmath305 vertices . define @xmath306 then , on the event @xmath250 we have @xmath307 by lemma [ veblen ] we may represent @xmath167 as the union of double cycles @xmath308 , such that : 1 .
@xmath309 is rooted ; 2 . for all
@xmath310 $ ] , @xmath311 has @xmath312 edges ; 3 . for @xmath313 , @xmath311 has @xmath314 common vertices with @xmath315 .
define the rooted even digraphs @xmath316 , @xmath317 .
let @xmath318 denote the associated equivalence classes .
let @xmath319 be the set of indices @xmath313 such that @xmath320 since @xmath321 for any @xmath313 , @xmath322 , using @xmath323 we see that @xmath324 since @xmath325 is a rooted double cycle with at most @xmath86 edges , and we are assuming the validity of the event @xmath250 , by lemma [ cycle stats ] we have @xmath326 .
moreover , by lemma [ induction ] , one has @xmath327 where we used the assumption @xmath328 . next , observe that @xmath329 thus , combining the above estimates one has @xmath330 where @xmath331 .
note that @xmath332 implying that @xmath333 .
the proof is complete .
let @xmath334 denote the event that @xmath335 for all @xmath336\times [ n]$ ] .
an application of markov s inequality and the assumption @xmath337{\leqslant}1 $ ] shows that @xmath338 .
thus , if we define @xmath339 , where @xmath250 is the event from lemma [ cycle stats ] , then @xmath340 we are going to choose eventually @xmath97 .
therefore , thanks to , to prove the theorem it will be sufficient to prove the conditional statement @xmath341 to prove this , we estimate the conditional moments @xmath342 $ ] . from the expansion in one
has @xmath343 { \leqslant}\sum_{i , j}\sum_{p_1,p_2:i\mapsto j } { \mathbb{e}}[w(p_1)\bar w(p_2)\mid{\mathcal{e}}_k]\,,\ ] ] where the internal sum ranges over all paths @xmath61 and @xmath62 of length @xmath63 from @xmath64 to @xmath65 , the weight @xmath66 of a path is defined by , and @xmath69 denotes the complex conjugate of @xmath66 .
notice that since @xmath344 on the event @xmath123 , all expected values appearing above are well defined . by the symmetry assumption
we can replace the variables @xmath3 by @xmath345 where @xmath346 are symmetric i.i.d .
random variables , independent from the @xmath347 .
conditioning on @xmath123 the entries @xmath348 are no longer independent .
however , since @xmath123 is measurable with respect to the absolute values @xmath349 , the signs @xmath350 are still symmetric and i.i.d . after conditioning on @xmath123 .
it follows that @xmath351=0,\ ] ] whenever there is an edge with odd multiplicity in @xmath80 .
thus , in we may restrict to @xmath79 such that each edge in @xmath352 has even multiplicity .
let @xmath84 denote the closed path obtained as follows : start at @xmath64 , follow @xmath61 , then add the edge @xmath85 , then follow @xmath62 , then end with the edge @xmath85 again .
thus , @xmath84 is an even closed path of length @xmath86 .
note that according to our definition , if @xmath156 is the rooted even digraph generated by the path @xmath84 , with root at the edge @xmath85 , then @xmath353 since the map @xmath88 is injective , allows us to estimate @xmath354 { \leqslant}\sum_{p}{\mathbb{e}}\left[p_r(g_p)\mid { \mathcal{e}}_k\right],\ ] ] where the sum ranges over all even closed paths @xmath355 of length @xmath86 and @xmath156 is defined as the rooted even digraph generated by the path @xmath84 , with root at the edge @xmath356 . by lemma [ paths and graphs ] , the sum in can be further estimated by @xmath357,\ ] ] where we used @xmath358 , and @xmath359 denotes the set of all rooted even digraphs with @xmath86 edges and @xmath305 vertices .
below we estimate @xmath360 deterministically on the set @xmath123 .
using the second inequality in one has , for any @xmath361 : @xmath362 since on the event @xmath123 all entries satisfy @xmath344 , it follows that @xmath363
. therefore the above sum can be truncated at @xmath364 let @xmath167 be a given equivalence class of rooted even digraphs with @xmath305 vertices and @xmath86 edges .
summing over all @xmath168 , and recalling , @xmath365 from proposition [ exp prop ] , on the event @xmath123 we can then estimate @xmath366 where @xmath367 .
summing over all equivalence classes @xmath167 of rooted even digraphs with @xmath305 vertices with @xmath86 edges , on the event @xmath123 one obtains @xmath368 going back to , using , and lemma [ graphs counting ] to estimate @xmath369 , one finds @xmath370 { \leqslant}3hk^4n^{k}\bigl(3ek^2\bigr)^{\frac{4k\log b } { \varepsilon\log n } } \sum_{x=1}^k ( 4k)^{6(k - x)}n^{-\varepsilon y_x/16}\ ] ] fix @xmath126 . if @xmath371 , then @xmath372 and therefore @xmath373 provided that @xmath11 is sufficiently large .
it follows that @xmath374 from , for large enough @xmath11 and @xmath97 , one has @xmath375{\leqslant}n^k ( \log
n)^{c\log n } , \ ] ] where @xmath376 is a constant depending only on @xmath377 .
the proof of is concluded by using markov s inequality : for any @xmath10 , @xmath378 \\ & { \leqslant}(1+\delta)^{-2k+2}n ( \log n)^{c\log n}. \ ] ] since @xmath97 , for fixed @xmath10 , the expression above is @xmath379 for any @xmath104 .
this ends the proof of theorem [ main ] .
bordenave and m. capitaine .
outlier eigenvalues for deformed i.i.d random matrices . to appear in _ comm .
pure appl .
( 2016 ) preprint available at http://arxiv.org/abs/1403.6001[arxiv:1403.6001 ] | consider a square matrix with independent and identically distributed entries of zero mean and unit variance .
it is well known that if the entries have a finite fourth moment , then , in high dimension , with high probability , the spectral radius is close to the square root of the dimension .
we conjecture that this holds true under the sole assumption of zero mean and unit variance , in other words that there are no outliers in the circular law . in this work
we establish the conjecture in the case of symmetrically distributed entries with a finite moment of order larger than two .
the proof uses the method of moments combined with a novel truncation technique for cycle weights that might be of independent interest . |
although trapped ions have found many applications in physics @xcite , they caused a turning point in the evolution of quantum computing when the paper entitled _ quantum computation with cold trapped ions _ was published by cirac and zoller in 1995 @xcite .
this proposal launched also an avalanche of other physical realizations of quantum computing using different physical systems , from high finesse cavities to widely manufactured semiconductors @xcite . through the years
we have learnt a lot , but also revealed many peculiarities , about the physical realization of quantum computing which has led to many discussions concerning the conditions under which we could in principle implement quantum computing in certain quantum systems . before we give the list of requirements for the physical implementation of quantum computing we will introduce the fundamental terminology to appear throughout this paper .
we will follow the definitions in ref .
@xcite . * a _ qubit _ is a quantum system in which the logical boolean states 0 and 1 are represented by a prescribed pair of normalized and mutually orthogonal quantum states labelled as @xmath0 and @xmath1 .
these two states form a computational basis and any other ( pure ) state of the qubit can be written as a superposition [ q1 ] the multi - qubit cnot on @xmath2 ions ( [ qg21 ] ) differs from the two - qubit cnot gate only in the number of laser pulses required for its realization .
the multi - qubit cnot gate corresponds to two @xmath3-pulses on the carrier [ @xmath4 , @xmath5 , a single @xmath6-pulse on the first red sideband ( @xmath7 ) and @xmath8 @xmath9-pulses also on the first red sideband ( @xmath10 , @xmath11 , @xmath12 , @xmath13 ) .
thus , it requires all together @xmath14 laser pulses , where @xmath15 refers to the number of the ions involved in the gate ( @xmath16 control ions , one target ion ) .
then , in the spirit of the previous discussion the minimal total time for the realization of the multi - qubit cnot gate on @xmath2 ions reads [ sp2 ] t=2(t_a+qt_b ) , where we assume @xmath17 and @xmath18 is given by eq .
( [ sp1 ] ) for a total number @xmath19 ( @xmath20 ) of the ions confined in the trap .
we give some estimations for the realization of multi - qubit cnot gates in table [ tab ] , where the number @xmath2 of the ions involved in the gate and the total number @xmath19 of the ions in the trap are equal ( @xmath21 ) .
we stress this point because there is a difference if we realize a two - qubit cnot gate ( @xmath22 ) having just two ions in the trap ( @xmath23 ) , then we get for the total time of the gate [ sp3 ] t_1=2(t_a+2t_b^n=2)0.5 , or having a larger register of ten ions in the trap ( @xmath24 ) , what gives [ sp4 ] t_2=2(t_a+2t_b^n=10)1.1 at the same fidelity @xmath25 . any multi - qubit gate on a register of size @xmath19
can be decomposed into a network of single - qubit rotations and two - qubit cnot gates @xcite .
however , there might be a possibility to realize this multi - qubit directly , if a given physical system allows it .
for instance , the multi - qubit cnot gate on six qubits can be decomposed into the network of 12 two - qubit cnot gates including three additional auxiliary qubits @xcite . in the case of cold trapped ions
it requires the total time for the realization of the whole network ( @xmath22 , @xmath26 ) [ sp5 ] t_3=122(t_a+2t_b^n=9)12.7 .
however , the direct implementation ( [ qg21 ] ) would take only ( @xmath27 ) [ sp6 ] t_4=2(t_a+6t_b^n=6)2.6 , which is about five - times less than the former case .
we have again assumed almost the perfect fidelity @xmath25 of the operation .
we conclude that there can be quantum systems that may support a direct implementation of multi - qubit gates , rather than their decomposition into fundamental gates , what may bring advantages at experimental realization as well as in quantum state synthesis @xcite .
throughout the paper we have discussed many aspects of cold trapped ions for quantum computing but we have not dealt with the decoherence which appears to be a main obstacle in achievements of experimental quantum computing . the reason was that our main goal has been to give a basic review and the discussion on decoherence sources and effects would refer more to an advanced study @xcite . nevertheless , for the sake of completeness we would like to mention on this place some decoherence aspects met in the lab .
we will follow a detailed study of experimental issues in quantum manipulations with trapped ions given by wineland et al . @xcite .
the decoherence will be met in a more general usage of this term .
thus , by the _ decoherence _ we mean any effect that limits the fidelity ( the match between desired and achieved realization ) .
further , we will distinguish three categories .
* _ motional state decoherence _ is the most troublesome source of the decoherence in ion trap experiments and refers to the relaxation of two vibrational states @xmath28 and @xmath29 of a given motional mode used as the quantum data bus . the ions are cooled to the ground motional state @xmath28 and the excitation to the @xmath29 state is used for the transfer of information on a distinct ion . however , this scenario is not ideal for several reasons : * * instability of trap parameters .
we are simply not able to control all voltages as they undergo fluctuations and dephasing . *
* we also have to count on ( i ) the micromotion , ( ii ) the coulomb repulsion between the ions making the motional modes ( except the com mode ) anharmonic in reality and ( iii ) stray electrode fields causing possible excitations of the ion motion . *
* we have considered just a single motional mode in our approach , but there are also other @xmath30 modes present and the cross - coupling between the modes appears . if spectator @xmath30 modes are not cooled to their ground motional states , the energy can be transferred to the mode of interest .
this happens because the trapping potential is anharmonic in real and these higher anharmonic terms are responsible for the cross - coupling .
* * we should mention also inelastic and elastic collisions with the background gas , even though experiments are carried out in an excellent environment ( @xmath31 ) . * _ internal state decoherence _ corresponds to the evolution when a pure state of the ion @xmath32 transforms into a mixture @xmath33 .
the ions demonstrate internal decoherence times of the order of seconds ( calcium ) up to minutes and hours ( beryllium ) .
the type of the decoherence discussed here can be eliminated by a proper choice of metastable excited states with long lifetimes . *
_ operational decoherence
_ refers to the precision of coherent laser - ion manipulations .
there are several aspects that we have to consider : * * when the ion is illuminated with a laser beam , one has to control the pulse duration and the phase adjustment in order to avoid the preparation of unwanted states . * * due to the laser spatial intensity profile , there is a probability ( if the ions are spaced too closely ) that the state of a neighbouring ion will be affected . * * if we consider the standing - wave configuration we have to take care of the precise position of the ion in the node or the antinode of the standing wave what seems to be very troublesome .
* * finally , off - resonant transitions are always present and we have to control the laser power very carefully to avoid their excitations
. however , there is a way to eliminate the effect of the decoherence .
we can encode information into a decoherence - free subspace whose states are invariant under coupling to the environment @xcite .
we have been discussing cold trapped ions so far .
_ cold _ refers to the fact that all motional modes have to be cooled to their ground motional states because the dynamics assumes the precise control over the motional state .
however , there have appeared other proposals referring to _ warm _ or _ hot _ trapped ions which assume an arbitrary motional state .
* poyatos et al.@xcite proposed a scheme for the realization of two - qubit cnot gates between two trapped ions using ideas from the atomic interferometry .
they split the wavepacket of the control ion into two directions depending on its internal state with a laser pulse .
then they address one of the wavepackets of the target ion changing conditionally its internal state and finally they bring together the wavepackets of the ions using another laser pulse .
the ions communicate through the coulomb repulsion and under ideal conditions the scheme is independent on the motional state of the ions . *
milburn et al.@xcite described two schemes for manipulations with warm trapped ions .
firstly , they use the adiabatic passage for the conditional phase shift , i.e. the phase of the ion is flipped if the motional mode is in the superposition of odd number states and the ion is excited . the com mode in an arbitrary vibrational state
is used for quantum logic but all other motional modes are assumed to be cooled to their ground motional states .
secondly , they apply the idea of the collective spin @xcite for faster gates .
this idea has been also used for the introduction of multi - qubit gates for quantum computing @xcite . *
srensen and mlmer @xcite proposed a novel scheme based on the idea of bichromatic light ( @xmath34 ) . realizing the two - qubit cnot gate they illuminate two ions with the bichromatic light coupling the states @xmath35 and @xmath36 .
they choose detunings far enough from the resonance with the first red and blue sidebands such that the intermediate states @xmath37 and @xmath38 are not populated in the process .
the scheme is not sensitive on fluctuations of the number of phonons in the relevant motional mode .
it is also possible to illuminate with the bichromatic light more ions and generate a multiparticle entangled state .
actually , these experiments were already realized in nist @xcite and they generated the ghz state with two ions @xmath39 with a fidelity @xmath40 and also the ghz state with four ions @xmath41 with a fidelity @xmath42 using beryllium ions and the raman scheme .
jonathan and plenio proposed light shift induced quantum gates for trapped ions insensitive on phonon number in motional modes ( thermal motion ) @xcite . *
finally , there has appeared a proposal of a scalable quantum computer with the ions in an array of microtraps by cirac and zoller @xcite detailed in ref .
the ions are placed in a 2d array of independent ion microtraps @xcite and there is another ion ( head ) that moves above this plane .
if we position the head above a particular ion from the array and switch on the laser in the perpendicular direction , we can realize a two - qubit gate .
this operation allows us to swap the state of the ion to the head which can be moved immediately above a distinct ion in the array and transfer the information onto it .
the ions oscillating in the microtraps are not assumed to be cooled to their ground motional states .
however , their motion can couple to the environment .
it becomes relevant during the time when the ion interacts with the head but not in the case when the head moves .
in this paper we have tried to review achievements accomplished in the field of cold trapped ions . in the first part we have discussed in detail the ion loading and trapping process , the collective vibrational motion of the ions .
we have also given a detailed derivation of the hamiltonian governing the dynamics of the system including the discussion of weak coupling and lamb - dicke regime .
further , we have reviewed laser cooling techniques and a detection process with experimental illustrations on calcium ions . in the second part we have discussed the implementation of quantum computing using cold trapped ions .
in particular , we have described how to realize single - qubit , two - qubit and multi - qubit quantum logic gates .
finally , we have estimated the speed of quantum gates with cold trapped ions .
the aim of this paper is to give an introduction to this field with many references on relevant papers and studies .
we would like to thank peter knight , danny segal , martin plenio , andrew steane and miloslav duek for their helpful comments and suggestions .
we are also grateful to rainer blatt , giovanna morigi and david kielpinski for sending us original files of their figures .
this work was supported by the european union projects qubits ( ist-1999 - 13021 ) and quest ( hprn - ct-2000 - 00121 ) .
we acknowledge the special support from the slovak academy of sciences .
one of us ( m. . ) is thankful for the support from the esf via the _ programme on quantum information and quantum computation_.
the expression ( [ int9 ] ) corresponds to the complete hamiltonian in the sense that it includes also off - resonant transitions . for instance , even though a sufficiently intense laser is tuned on the carrier , the off - resonant transitions on the sidebands are also present and they cause imprecisions and perturbations in the dynamics .
this can be avoided by setting the laser intensity @xmath43 , i.e. the laser coupling constant [ eq .
( [ int8 ] ) or ( [ int8.1 ] ) ] , sufficiently small . in
what follows we will determine conditions ( characterizing the _ weak coupling regime _ ) under which we can neglect off - resonant transitions .
the implementation of quantum gates on cold trapped ions requires laser pulses on the carrier [ fig.[3freq](a ) ] and on the first red sideband [ fig.[3freq](b ) ] .
therefore , we will discuss the dynamics on these two spectral lines . at first , let us assume that the laser is tuned on the carrier @xmath44 .
the closest off - resonant transitions ( detuned by the frequency @xmath45 ) are on the first blue and on the first red sideband [ fig.[off](a ) ] .
we will consider only the respective terms in the hamiltonian ( [ int9 ] ) and drop down the index @xmath46 . in the lamb - dicke limit
we can write [ off1 ] _ 1= _ + + ( i)_+e^it+ ( i)_+e^-it+ .
further , we assume the ion to be initially in the state @xmath47 , we apply the hamiltonian ( [ off1 ] ) and calculate the probability of off - resonant transitions .
the dynamics governed by a time - dependent hamiltonian @xmath48 is described to the first order by the unitary evolution operator [ off3 ] ( t , t_0)- _
t_0^t(t^)dt^. then the probability to find the ion ( initially prepared in the state @xmath47 ) in the state @xmath49 ( i.e. undergoing the off - resonant transition on the first blue sideband ) is [ off4 ] p_b=|e|n+1|_1|g|n|^2= ^2 , where @xmath50 is given by eq .
( [ off3 ] ) for the hamiltonian ( [ off1 ] ) .
the probability to find the ion in the state @xmath51 corresponding to the off - resonant transition on the first red sideband is given as [ off5 ] p_r=|e|n-1|_1|g|n|^2= ^2 .
if there is no population transferred via the off - resonant transitions to the states @xmath49 and @xmath51 , i.e. @xmath52 and @xmath53 at any time @xmath54 , we can neglect these off - resonant transitions .
then we get the conditions of the weak coupling regime for the transition on the carrier in the form [ off6 ] ||and [ off7 ] || .
we can also avoid the off - resonant transitions by setting the laser beam perpendicular to the @xmath55 axis ( @xmath56 ) . then the lamb - dicke parameter [ see eq .
( [ int5 ] ) ] is equal to zero ( @xmath57 ) and the coupling on the off - resonant transitions vanishes .
analogically , we can assume the laser to be tuned on the first red sideband ( @xmath58 ) and the closest off - resonant transitions are on the carrier and on the second red sideband [ fig.[off](b ) ] .
however , the strength of the second red sideband is of the order of @xmath59 and we can omit it in the lamb - dicke limit
. then the respective hamiltonian is given as [ off8 ] _
2= ( i)_++ _ + e^it+ .
the probability to find the ion ( initially in the state @xmath47 ) in the state @xmath60 ( after the off - resonant transition on the carrier ) can be calculated as [ off9 ] p_c=|e|n|_2|g|n|^2= ^2 , where @xmath61 is given by eq .
( [ off3 ] ) for the hamiltonian ( [ off8 ] ) .
we can neglect the off - resonant dynamics if @xmath62 at any time @xmath54 .
then for the weak coupling regime on the first red sideband applies [ off11 ] || . even though the transition on the carrier is off - resonant , it has stronger coupling in the lamb - dicke limit than the first red sideband
. therefore , it is very important to follow in the experiment the constraint given by eq .
( [ off11 ] ) . besides the population of the off - resonant levels
there is another source of imprecisions in the state manipulation .
however , it is weaker and it does nt require any special constraints on physical parameters except those for the weak coupling regime . when the laser drives a transition between two levels
there appears a frequency shift called _ stark light shift _ caused by the presence of other spectator levels .
therefore , in the experiment we have to consider the detuning ( [ int9.0 ] ) rather in the form [ off20 ] = _ l-_0- , where @xmath63 corresponds to the stark light shift .
the higher the laser intensity , the more significant the light shift is .
we can correct for this effect by shifting the laser frequency ( @xmath64 ) which tunes the transition back to the resonance .
further , we will estimate the frequency shift @xmath63 .
let us consider the hamiltonian ( [ int3 ] ) and transform it to the interaction picture ( [ int6 ] ) with @xmath65 , where @xmath66 .
then we get [ off21 ] = _ 0+= ( -_z+)+ ( _ + e^i(+)+ ) , where in the lamb - dicke limit the interaction term reduces to [ off22 ] _ + + ( i)_++ ( i)_++ with the first term corresponding to the transition on the carrier , the second term to the first blue sideband and the last one to the transition on the first red sideband . in the second order of the time - independent perturbation theory ( the first order gives no contribution ) we can write for the shift of the energy levels [ off23 ] e_|g|n&= & _ m(m - n ) and [ off24 ] e_|e|n&= & _ m(n - m ) .
for instance , for the transition on the first red sideband ( @xmath58 ) we can calculate [ off25 ] e_|g|n-,e_|e|n-1 and the corresponding light shift can be estimated as [ off26 ] = ( e_|e|n-1-e_|g|n)/. if we choose @xmath67 and @xmath68 so that the condition ( [ off11 ] ) holds , then @xmath69 .
in the relation for the coupling constant ( [ int15 ] ) we can expand the exponential function to the taylor series about the value @xmath70 and use the expression ( [ int16 ] ) for the laguerre polynomial . then we get [ ldl1 ] _
j^n , k()&= & _ j ( i_j)^|k| + & & , where @xmath71 denotes the terms proportional to the fourth and higher powers of @xmath72 . in the _ lamb - dicke regime _
we consider the dependence of @xmath73 on the parameter @xmath72 only to its lowest order , i.e. in eq .
( [ ldl1 ] ) we neglect all terms with any higher power than @xmath74 and we get the coupling constant @xmath73 given by the expression ( [ int20 ] ) .
this approximation can be done only if the conditions [ ldl2 ] 1 and [ ldl3 ] _ j^2 ( n+|k|n-1)(n+|k|n)_j^21 are satisfied .
we will refer to the condition ( @xmath75 ) [ ldl4 ] _ j1 as the _ lamb - dicke limit _
, where @xmath76 is the average number of phonons in the respective vibrational mode @xcite .
* the lamb - dicke limit corresponds physically to the situation where the spatial extent of the vibrational motion of the ion @xmath77 is much smaller than the wavelength @xmath78 of the laser , where @xmath79 and @xmath80 [ see def . in eq .
( [ int5 ] ) ] . *
the lamb - dicke limit can be physically interpreted also from a different point of view .
we may rewrite the lamb - dicke parameter of the @xmath46 ion of @xmath19 ions in the com mode to the form @xmath81 , where @xmath82 is the recoil energy .
it can be shown @xcite that the trapped ion emits spontaneously photons of the average energy @xmath83 , where @xmath84 is for the atomic frequency .
taking into account the lamb - dicke limit ( @xmath85 ) we may say that during the spontaneous emission the change in the vibrational state of the ion is very unlikely .
in other words , the trapped ion in the lamb - dicke regime decays spontaneously mostly on the carrier @xmath86 . *
the lamb - dicke parameter for a single trapped ion equals to @xmath87 and for an ion from the string of @xmath19 ions in the com mode is given as @xmath88 .
it means that we can reach the lamb - dicke limit ( [ ldl4 ] ) for @xmath19 ions even if the limit is not fulfilled for single ions @xcite . d. leibfried , c. roos , p. barton , h. rohde , s. gulde , a. mundt , g. reymond , m. lederbauer , f. schmidt - kaler , j. eschner and r. blatt , _ experiments towards quantum information with trapped calcium ions _ ,
quant - ph/0009105 ( 2000 ) h. rohde , s.t .
gulde , c.f .
roos , p.a .
barton , d. leibfried , j. eschner , f. schmidt - kaler and r. blatt , _ sympathetic ground state cooling and coherent manipulation with two - ion crystals _ , quant - ph/0009031 ( 2000 ) d.g .
enzer , m.m .
schauer , j.j .
gomez , m.s .
gulley , m.h .
holzscheiter , p.g .
kwiat , s.k .
lamoreaux , c.g .
peterson , v.d .
sandberg , d. tupa , a.g .
white , r.j . hughes and d.f.v .
james , _ observation of power - law scaling for phase transitions in linear trapped ion crystals _ ,
phys . rev .
lett . * 85 * , 2466 ( 2000 ) d.j .
wineland , c. monroe , w.m .
itano , d. leibfried , b.e .
king be and d.m .
meekhof , _ experimental issues in coherent quantum - state manipulation of trapped atomic ions _ , journal of research of the national institute of standards and technology * 103 * , 259 ( 1998 ) , ( see also quant - ph/9710025 ) f. schmidt - kaler , c. roos c , h.ch .
ngerl , h. rohde , s. gulde , a. mundt , m. lederbauer , g. thalhammer , t. zeiger , p. barton , l. hornekaer , g. reymond , d. leibfried , j. eschner and r. blatt , _ ground state cooling , quantum state engineering and study of decoherence of ions in paul traps _ ,
journal of modern optics * 47 * , 2573 ( 2000 ) r.j .
hughes , d.f.v .
james , j.j .
gomez , m.s .
gulley , m.h .
holzscheiter , p.g .
kwiat , s.k .
lamoreaux , c.g .
peterson , m.m .
sandberg , v.d .
schauer , c.m .
simmons , c.e .
thorburn , d. tupa , p.z .
wang and a.g .
white , _ the los alamos trapped ion quantum computer experiment _ , fortschritte der physik * 46 * , 329 ( 1998 ) b.e .
king , c.s .
wood , c.j .
myatt , q.a .
turchette , d. leibfried , w.m .
itano , c. monroe and d.j .
wineland , _ cooling the collective motion of trapped ions to initialize a quantum register _ , phys .
lett . * 81 * , 1525 ( 1998 ) b.e .
king , c.s .
wood , c.j .
myatt , q.a .
turchette , d. leibfried , w.m .
itano , c. monroe and d.j .
wineland , _ cooling the collective motion of trapped ions to initialize a quantum register _ , phys
lett . * 81 * , 1525 ( 1998 ) d. kielpinski , b.e .
king , c.j .
myatt , c.a .
sackett , q.a .
turchette , w.m .
itano , c. monroe , d.j .
wineland and w.h .
zurek , _ sympathetic cooling of trapped ions for quantum logic _ , phys . rev . *
a61 * , 032310 ( 2000 ) c.f .
roos , d. leibfried , a. mundt , f. schmidt - kaler , j. eschner and r. blatt , _ experimental demonstration of ground state laser cooling with electromagnetically induced transparency _ , phys .
lett . * 85 * , 5547 ( 2000 ) f. schmidt - kaler , j. eschner , g. morigi , c.f .
roos , d. leibfried , a. mundt and r. blatt , _ laser cooling with electromagnetically induced transparency : application to trapped samples of ions or neutral atoms _ , quant - ph/0107087 ( 2001 ) q.a .
turchette , c.j .
myatt , b.e .
king , c.a .
sackett , d. kielpinski , w.m .
itano , c. monroe and d.j .
wineland , _ decoherence and decay of motional quantum states of a trapped atom coupled to engineered reservoirs _ ,
phys . rev . *
a62 * , 053807 ( 2000 ) c.a .
sackett , d. kielpinski , b.e .
king , c. langer , v. meyer , c.j .
myatt , m. rowe , q.a .
turchette , w.m .
itano , d.j .
wineland and i.c .
monroe , _ experimental entanglement of four particles _ , nature * 404 * , 256 ( 2000 ) t. calarco , h.j .
briegel , d. jaksch , j.i .
cirac and p. zoller , _ quantum computing with trapped particles in microscopic potentials _
, fortschritte der physik * 48 * , 945 ( 2000 ) , ( see also quant - ph/0010105 ) | in this tutorial we review physical implementation of quantum computing using a system of cold trapped ions .
we discuss systematically all the aspects for making the implementation possible .
firstly , we go through the loading and confining of atomic ions in the linear paul trap , then we describe the collective vibrational motion of trapped ions .
further , we discuss interactions of the ions with a laser beam . we treat the interactions in the travelling - wave and standing - wave configuration for dipole and quadrupole transitions .
we review different types of laser cooling techniques associated with trapped ions .
we address doppler cooling , sideband cooling in and beyond the lamb - dicke limit , sympathetic cooling and laser cooling using electromagnetically induced transparency .
after that we discuss the problem of state detection using the electron shelving method . then quantum gates are described .
we introduce single - qubit rotations , two - qubit controlled - not and multi - qubit controlled - not gates .
we also comment on more advanced multi - qubit logic gates .
we describe how quantum logic networks may be used for the synthesis of arbitrary pure quantum states .
finally , we discuss the speed of quantum gates and we also give some numerical estimations for them .
a discussion of dynamics on off - resonant transitions associated with a qualitative estimation of the weak coupling regime and of the lamb - dicke regime is included in appendix .
+ + pacs numbers : 03.65.ud , 03.67.lx , 32.80.pj , 32.80.ys |
the gauge interaction properties of the standard model ( sm ) have been measured quite precisely in the last decade .
however , the higgs particle has not yet been discovered in spite of much effort .
the physics behind the electroweak symmetry breaking ( ewsb ) and the origin of masses of quarks and leptons are left as unresolved problems .
the idea of the top quark condensate @xcite , in which the 4-top - quark interaction is introduced to trigger the ewsb , explains naturally the large mass of the top quark at the ewsb scale .
( see also the earlier attempt @xcite . )
this model is often called the `` top mode standard model '' ( tmsm ) , because the scalar bound state of @xmath16 plays the role of the higgs boson in the sm .
it is known that the original version of the tmsm has some difficulties : the top quark mass @xmath17 is predicted about 10%30% larger than the experimental value ( @xmath18 gev ) , even if we take the ultraviolet cutoff ( or the compositeness scale ) to the planck or the gut scale @xcite .
in addition , such a huge cutoff causes a serious fine - tuning problem . in topcolor models ( tcms )
@xcite , the 4-top - quark interaction , whose origin is not specified in the original version of the tmsm , is provided by exchange of colorons which are color - octet massive gauge bosons strongly interacting with the top quark .
@xcite if we assume that the coloron mass @xmath4 is @xmath19 , we need not tune the coupling of the 4-top - quark finely to its critical value . although the mass generated by the dynamics of colorons ( the dynamical mass @xmath5 ) becomes quite large , typically @xmath20 tev , in tcms where only the top - condensate is responsible for the ewsb , the problem @xmath21 can be resolved in some classes of models such as topcolor assisted technicolor models ( tc2 ) @xcite and top - seesaw ( tss ) models @xcite . in tc2
, we introduce techni - fermions in addition to colorons .
we assume that the topcolor interaction is responsible for the top - quark mass , while the ewsb occurs mainly due to technicolor interactions .
namely , the dynamical mass itself is adjusted to @xmath22 with a small vacuum expectation value ( vev ) of the top quark condensate @xmath23 , where @xmath24 denotes the ewsb scale .
the top - yukawa coupling in tc2 is about @xmath25 times larger than the value in the sm .
this enhancement of the top - yukawa is quite important to the higgs boson production via the gluon fusion process . in tss models ,
we introduce a vector - like heavy colored fermion @xmath26 , whose mass term has nothing to do with the ewsb , in order to have a seesaw - type mass matrix .
the dynamical mass @xmath5 in tss models does not correspond to the mass eigenvalue of the top - quark .
the experimental value @xmath22 is obtained after diagonalization of the mass matrix for the top - quark via the seesaw mechanism .
namely , the large dynamical mass of the order of 0.6 tev can be realized in tss models consistently with @xmath22 . here ,
we note that the mixing between @xmath27 and @xmath28 is severely constrained from the custodial symmetry violation , i.e. the @xmath29-parameter , which usually requires @xmath26 to have a mass of the order of several - tev . although the tmsm with extra dimensions proposed by arkani - hamed et al .
@xcite is also a candidate to resolve the fine - tuning problem etc . , it is a pure bulk gauge theory without 4-fermion interactions in the bulk rather than a tcm @xcite .
the phenomenology of the tmsm with extra dimensions will be studied elsewhere @xcite . in this paper , we study the production of the higgs boson in tcms .
we regard the higgs boson as the tightly bound state of @xmath16 ( or @xmath30 ) and do not consider direct signatures of other bound states .
the higgs boson can be searched for at the tevatron @xcite or at the lhc @xcite . in the sm ,
the leading contribution to the higgs boson production at hadron colliders comes from the gluon fusion process via loops of heavy quarks @xcite .
the gluon fusion process is thus sensitive to the top - yukawa coupling and the number of heavy quarks .
in addition , it is possible that extensions to qcd affect the gluon fusion process .
especially , colorons are strongly coupled to the top quark , and the dynamical mass @xmath5 is very large for tcms with a single vev triggering the ewsb , so that the effect of colorons can be expected to be sizable . in order to evaluate roughly the size of the contribution of colorons
, we first derive the low - energy effective theory by eliminating colorons by using the equation of motion ( eom ) for colorons .
we then find that the effective gluon - gluon - higgs- ( @xmath1- ) operator is induced in the low - energy effective theory and that the coefficient is proportional to @xmath3 .
in the situation of @xmath31 and @xmath8 , the contribution of colorons becomes comparable to the top - loop effect .
next , we estimate quantitatively the size of the effective @xmath1-coupling including the contributions of the top - quark loop , other heavy - quark loops , and colorons in tss models and tc2 . we find that the contribution of colorons can be sizable in tss models with the coloron mass @xmath7 and the mass of @xmath26 , @xmath32 , and that the effect of the top - loop is almost same as in the sm .
we estimate numerically the effects of colorons and the top - loop as @xmath33 \times 10^{-2 } ( \rm tev^{-1})$ ] and @xmath34 \times 10^{-2 } ( \rm tev^{-1})$ ] for @xmath35 tev , respectively .
namely , the contribution of colorons ( the top - loop ) is dominant in the real ( imaginary ) part of the @xmath10 amplitude . in tc2 ,
the contribution of colorons becomes small , while the effect of the top - loop is strongly enhanced .
we find that signatures of the higgs boson in tc2 can be observed even at the tevatron run ii as well as at the lhc .
in particular , we evaluate @xmath13 for an integrated luminosity of 2 fb@xmath36 and @xmath15 gev at the tevatron run ii .
this paper is organized as follows : in sec .
ii we derive the low - energy effective theory by eliminating colorons by using the eom in a tcm with only the top- and bottom - quarks , for illustration .
we find that the contribution of colorons is proportional to @xmath3 . in sec .
iii , we estimate quantitatively the size of the effective @xmath1-coupling in tss models and in tc2 . we find that the contribution of colorons can be sizable in tss models , while it can not in tc2 . as for the effect of the top - loop
, it is very large in tc2 .
we show expected signals of the higgs boson in tc2 at the tevatron run ii and at the lhc .
iv is devoted to summary and discussion .
we define the amplitude of @xmath10 as @xmath38 where @xmath39 denotes a gluon with momentum @xmath40 and the suffix @xmath41 is the index of colors for the adjoint representation . in the sm
, we can obtain @xmath42 at the 1-loop level , @xmath43 with @xmath44 , where we have defined @xmath45 and @xmath46 since the factor @xmath47 in the @xmath48 limit becomes independent on @xmath49 , @xmath50 the @xmath10 amplitude can be written as the local operator @xmath51 in this sense , the factor @xmath52 corresponds to the coefficient of the local operator @xmath53 .
we also note that the @xmath10 amplitude has the imaginary part above the threshold of the top - pair production ( @xmath54 ) .
the gluon fusion process is obviously sensitive to the top - yukawa coupling and the number of heavy quarks . in this paper
, we also consider the effect of colorons strongly interacting with the top quark .
we hence classify nonstandard contributions to the @xmath10 amplitude into three categories : ( a ) : : _ loop effects of the top quark ; _ + in the sm , the lowest order amplitude for the gluon fusion arises from the triangle diagram of the top quark . on the other hand
, the top - yukawa coupling can be larger than the sm one in tcms with multi composite higgs doublets .
the enhancement of the top - yukawa coupling is quite important for the gluon fusion process .
( b ) : : _ loop effects of other heavy quarks ; _ + generally , heavy quarks other than the top are also introduced in tcms .
they contribute to the @xmath10 amplitude at the 1-loop level .
( c ) : : _ contributions of strongly interacting colorons ; _ + since colorons are strongly coupled to the top quark , effects of colorons are potentially sizable .
corresponding to the three contributions ( a ) , ( b ) and ( c ) , we split the factor @xmath52 in eq .
( [ ggh_amp ] ) into three parts , @xmath6 , @xmath55 and @xmath2 , @xmath56 in this section , we roughly estimate the size of ( c ) . for this purpose
, we derive the low - energy effective theory by eliminating colorons by using the equation of motion ( eom ) for colorons . for illustration , we take a toy model of tcms with only the top- and bottom - quarks .
this model is not realistic , but it is sufficient to understand the origin of the coloron contribution @xmath2 . in the next section ,
we evaluate quantitatively the sizes of @xmath57 , and @xmath2 in typical tcms such as tc2 and tss models .
after the topcolor symmetry is spontaneously broken down , we write down the model , @xmath58 with @xmath59 { \cal l}_{\rm int } & \equiv & g ' a^a_\mu ( j_l^{a\mu}+j_r^{a\mu } ) , \label{l_int }
\\ { \cal l}_{\rm col } & \equiv & - \frac{1}{2}{\mbox{tr } } ( a_{\mu\nu } a^{\mu\nu } ) + m^2 { \mbox{tr}}(a_\mu a^\mu ) \nonumber \\ & & + i g_{gaa } { \mbox{tr}}(g_{\mu\nu}[a^\mu , a^\nu ] ) + i g_{a^3 } { \mbox{tr}}(a_{\mu\nu}[a^\mu , a^\nu ] ) + g_{a^4 } { \mbox{tr}}([a^\mu , a^\nu]^2 ) \label{l_col}\end{aligned}\ ] ] up to dimension four operators , where @xmath60 is the coupling constant between the top quark and colorons , and @xmath61 and @xmath62 denote the coloron field and the gluon , respectively -gauge interaction , for example . instead of specifying the manner for suppression of the bottom condensation , we adjust the 4-fermion interactions so that only the top condensation takes place . ] .
the current @xmath63 for the weak doublet @xmath64 ( the weak singlet @xmath65 , @xmath66 ) is defined as usual .
we note that @xmath62 and @xmath61 are matrices defined as @xmath67 with generators @xmath68 of @xmath69 , where @xmath70 is the number of colors . in eqs .
( [ l_top])([l_col ] ) with the gauge symmetry of qcd , the covariant derivative @xmath71 , and the field strength @xmath72 and @xmath73 for the gluon and the coloron are written as @xmath74 , \quad a_{\mu\nu } \equiv \partial_\mu a_\nu - \partial_\nu a_\mu -ig_s[g_\mu , a_\nu]+ig_s[g_\nu , a_\mu ] .\ ] ] we do not incorporate higher dimensional operators in @xmath75 . if we would allow them , we could freely choose the coefficient @xmath2 of the operator @xmath53 , as we will see later on .
we note that the triple coupling @xmath76 plays an important role in the following analysis and that it is equal to the qcd coupling , @xmath77 in usual tcms .
in order to obtain the low - energy effective theory with the higgs doublet @xmath93 , we rewrite the effective lagrangian eq .
( [ ggh0 ] ) at the coloron scale as follows : @xmath94 \nonumber \\ & & -\frac{n_c^{-1}}{4}\frac{g'{}^2}{g_{t0}^2 } \frac{(g_{gaa}+g_s)^2}{m^2 } g_{\mu\nu}^a g^{a\,\mu\nu } \phi_0^\dagger \phi_0 -\frac{n_c^{-1}}{4}\frac{g'{}^2}{g_{t0}^2 } \frac{(g_{gaa}+g_s)^2}{(m^2)^2 } g_{\mu\nu}^a g^{a\,\mu\nu } \left ( g_{t0}(\bar q_l t_r)\phi_0 + \mbox{h.c . }
\right ) \label{h0}\end{aligned}\ ] ] with the bare higgs doublet @xmath95 .
we can confirm the equivalence of eq .
( [ h0 ] ) and eq .
( [ ggh0 ] ) through the eom for @xmath96 and @xmath95 .
the kinetic term of the higgs doublet develops below the coloron scale .
we then obtain the low - energy effective theory at the weak scale : @xmath97 with @xmath98 \nonumber \\ & & -\frac{n_c^{-1}}{4z_\phi}\frac{g'{}^2}{g_{t0}^2 } \frac{(g_{gaa}+g_s)^2}{m^2 } g_{\mu\nu}^a g^{a\,\mu\nu } \phi^\dagger \phi -\frac{n_c^{-1}}{4}\frac{g'{}^2}{g_{t0}^2 } \frac{(g_{gaa}+g_s)^2}{(m^2)^2 } g_{\mu\nu}^a g^{a\,\mu\nu } \left [ g_t(\bar q_l t_r)\phi + \mbox{h.c . }
\right ] , \label{h}\end{aligned}\ ] ] @xmath99 since the neutral component of @xmath93 can be written as @xmath100 , the coefficient @xmath2 of the effective @xmath1-operator is given by @xmath101 here , we comment on the reason why we did not allow higher dimensional operators in the lagrangian for colorons .
if the dimension six operator @xmath102[g^{\mu\lambda},a_\lambda])$ ] was not forbidden in @xmath75 , it would lead to the same composite operator as the third term of @xmath103 in eq .
( [ eff_lag ] ) .
this means that the coefficient @xmath2 of the operator @xmath53 at the weak scale is not constrained .
namely , tolerance of higher dimensional operators in @xmath75 is equivalent to adding @xmath53 with a free coefficient by hand .
the vev of @xmath104 has a nonzero value only when the coefficient @xmath105 of the 4-top quark interaction is larger than its critical coupling @xcite , @xmath106 the top quark then acquires the dynamical mass , @xmath107 within the usual large @xmath70-bubble approximation , we find the wave function renormalization of the higgs doublet , @xmath108 with the renormalization point @xmath109 .
substituting eq .
( [ z_phi ] ) for eq . ( [ m_t ] )
, we obtain the relation @xmath110 which is called the pagels - stokar ( ps ) formula @xcite , and can evaluate the dynamical mass as @xmath8 for @xmath7 . although the top quark mass is predicted too large ( @xmath111 ) in this toy model , we consider realistic tcms such as tss models and tc2 in the next section . now
, we estimate roughly the size of @xmath2 and compare it with the size of @xmath6 .
using @xmath112 , eq .
( [ z_phi ] ) reads @xmath113 up to a logarithmic term .
namely , the expression of @xmath2 in eq .
( [ ggh ] ) has no loop - suppression factor @xmath114 .
it is , in other words , a consequence of the large dynamical mass obtained from the ps formula , @xmath115 up to a logarithmic term .
actually , we can explicitly rewrite eq .
( [ ggh ] ) as @xmath116 in terms of @xmath5 , where @xmath117 and the suppression factor @xmath118 is canceled by @xmath115 .
we find roughly that the contribution of colorons is @xmath119 with @xmath120 , while the top - loop effect is @xmath121 $ ] from eq .
( [ sm_inf ] ) . in the toy model
, we can estimate the contribution of the top quark loop as @xmath122 by using the relation @xmath123 in the bubble approximation , we obtain @xmath124 . ] . in any case
, we obtain the ratio of @xmath2 and @xmath6 , @xmath125 where we used @xmath117 and @xmath120 .
this relation suggests that @xmath2 becomes comparable to @xmath6 in the situation of @xmath7 and @xmath8 .
the coloron mass @xmath31 is favorable in order to avoid the fine - tuning problem , while the large dynamical mass @xmath8 is obtained from a single vev triggering the ewsb .
the large dynamical mass can be realized within tss models .
however , the dynamical mass is not necessarily large in tcms with non - minimal higgs sectors such as tc2 . in the next section ,
we estimate quantitatively the effective @xmath1-coupling in tss models and tc2 .
in tss models , it is possible that the ewsb occurs via the `` top - condensate '' only . the large dynamical mass can be tuned consistently with the experimental value @xmath126 through the seesaw mechanism , which requires an additional vector - like heavy quark @xmath26 .
we hence consider the original version of tss models ( the @xmath127 tilting model ) @xcite , where the sm is embedded into a topcolor scheme , @xmath128 gauge group . the gauge group @xmath129 [ @xmath130 acts only on the third ( first and second ) generation(s ) of quarks and leptons .
the @xmath131 charges are @xmath132 ( @xmath133 ) for @xmath65 and @xmath28 , @xmath134 for the other quarks , and @xmath135 for leptons .
the charge assignment is also shown in table [ tss_org ] .
the gauge symmetry breaking @xmath136 , @xmath137 , and @xmath131 leave colorons , and two massive gauge bosons @xmath138 and @xmath139 , respectively .
we note that the mass term @xmath140 is related to the @xmath131 breaking vev in the @xmath127 tilting model
. we may take common masses @xmath141 for @xmath138 and @xmath139 of the order of a multi - tev and the coloron mass @xmath4 of the order of 1 tev .
in this situation , we can expect that the contribution of colorons becomes sizable and that the constraint on the @xmath29-parameter can be satisfied . here , we comment on the higgs boson mass @xmath11 in the @xmath127 tilting model : it is possible that the higgs boson has a mass of order 100 gev with tuning of some parameters @xcite .
however , the higgs boson mass is likely of the order of @xmath5 , i.e. @xmath142 .
we should note that the @xmath143-fusion process becomes comparable to the gluon fusion in the mass range @xmath142 in the sm .
.the representation of the third - generation and @xmath26 fermion in the @xmath127 tilting model .
[ tss_org ] [ cols="^,^,^,^,^,^,^",options="header " , ] in tc2 , the condensate of the techni - fermion @xmath29 mainly triggers the ewsb , @xmath144 where condensates of @xmath145 and @xmath16 provide @xmath146 and @xmath147 , respectively . adjusting the value of @xmath147
, we can obtain the experimental value of the top quark mass @xmath22 : the ps formula , @xmath148 which is same as eq .
( [ ps0 ] ) , leads to @xmath149 .
while the dynamical mass is small in tc2 , @xmath150 gev , the top - loop effect becomes very large due to enhancement of the top - yukawa coupling ( @xmath151 ) . as a result
, the gluon fusion process does not suffer from the contribution of colorons .
we estimate the higgs boson mass in tc2 as @xmath152 .
we easily obtain the contribution of colorons @xmath2 in tc2 from eq .
( [ ggh2 ] ) : @xmath153 with @xmath154 and @xmath155 tev . on the other hand ,
the effect of enhancement of the top - yukawa coupling is given by @xmath156 where the value of the sm is numerically found , @xmath157 for @xmath158 gev .
we emphasize that the value of @xmath159 is not changed , even if the coloron mass @xmath4 is taken to the order of a multi - tev .
unless we specify the matter content of tc2 in detail , we can not estimate contributions of techni - fermions having qcd charges , i.e. @xmath55 in our notation .
for example , techni - fermions in models of ref .
@xcite do not have qcd charges , so that they do not contribute to the gluon fusion process .
we then estimate the enhancement factor to the sm value as @xmath160 with @xmath161 and find that the coloron effect @xmath2 is negligible compared to the contribution of the top - loop @xmath6 .
now , we study the signature of higgs boson production at hadron colliders .
since masses of top - pions @xmath162 much below @xmath163 gev are phenomenologically forbidden @xcite due to the absence of the decay mode @xmath164 , the main decay mode of the higgs boson is expected to be pair production of weak bosons . in this situation
, we can apply the sm analysis at the tevatron @xcite and at the lhc @xcite directly to tc2 .
although @xmath165- and @xmath166-couplings in tc2 are suppressed , @xmath167 the effects are irrelevant under the narrow width approximation , because the branching ratio of @xmath168 is not changed from the sm in the mass range @xmath169 .
the considerable enhancement of the top - yukawa gives a chance to observe the higgs boson in tc2 even at the tevatron run ii .
( see table [ tab1 ] . ) while we can not expect to find any evidence of the sm higgs boson with @xmath12 gev for an integrated luminosity of 2 fb@xmath36 at the tevatron run ii , we can estimate @xmath13 for the higgs boson in tc2 with 2 fb@xmath36 and @xmath15 gev .
we show expected signals of the higgs boson in tc2 up to @xmath15 gev at the tevatron , since there is no background estimate for @xmath170 gev in ref . @xcite . at the lhc ,
signals of the higgs boson in tc2 are considerably enhanced .
( see table [ tab2 ] . )
the higgs boson in tc2 with the mass range @xmath169 can be discovered at the lhc much more easily than the sm higgs boson .
in this paper , we have studied the effective @xmath1-coupling in tcms .
we have considered the effect of colorons , in addition to the loop contributions of the top quark and other heavy quarks . in order to estimate the coloron effect ,
we have derived the low - energy effective theory by eliminating colorons by using the eom for colorons .
we have found that the contribution of colorons @xmath2 is proportional to @xmath3 .
thus , @xmath2 becomes sizable for the coloron mass @xmath7 and the dynamical mass @xmath8 .
an important point is that the large dynamical mass @xmath8 can be realized consistently with the experimental value of the top - quark mass @xmath9 in tss models , while the dynamical mass itself is adjusted to @xmath9 in tc2 .
we have shown that the contribution of colorons is actually sizable in tss models with @xmath171 and @xmath32 : we have evaluated the contributions of colorons and the top - loop as @xmath33\times 10^{-2 } ( \rm tev^{-1})$ ] and @xmath34\times 10^{-2 } ( \rm tev^{-1})$ ] , which is almost same as the sm value , for @xmath35 tev with some parameters being safe from the constraint on the @xmath29-parameter .
namely , @xmath2 ( @xmath6 ) is numerically dominant in the real ( imaginary ) part of the @xmath10 amplitude for @xmath172 . in tc2
, we have found that the effect of colorons is negligible compared to the contribution of the top - loop , because the dynamical mass @xmath5 itself is small and the top - yukawa coupling is enhanced by a factor of @xmath25 .
since the higgs boson production is considerably enhanced in tc2 , we can obtain signatures of the higgs boson in tc2 with @xmath12 gev even at the tevatron run ii as well as at the lhc .
in particular , we have evaluated @xmath13 for an integrated luminosity of 2 fb@xmath36 and @xmath15 gev at the tevatron run ii .
we have not studied the expected signals of the higgs boson in tss models at hadron colliders .
since the @xmath143-fusion process becomes comparable to the gluon fusion in the mass range @xmath142 , we would have also needed to study the @xmath143-fusion process for tss models .
in addition , we may investigate the higgs boson production in non - perturbative approaches , because @xmath142 is near the perturbative unitarity bound .
a detailed analysis of the expected signals of the higgs boson in tss models will be performed elsewhere @xcite .
the author is very grateful to m.tanabashi for helpful discussions .
y. nambu , enrico fermi institute report no .
89 - 08 , 1989 ; in _ proceedings of the 1989 workshop on dynamical symmetry breaking _ , edited by t. muta and k. yamawaki ( nagoya university , nagoya , japan , 1990 ) . | we discuss the production of the composite higgs boson in topcolor models via the gluon fusion process .
we consider the contribution of color - octet massive gauge bosons ( colorons ) strongly interacting with the top quark , in addition to nonstandard contributions of the top - yukawa coupling and heavy colored fermions other than the top quark . in order to estimate the contribution of colorons
, we derive the low - energy effective theory by eliminating colorons by using the equation of motion for colorons .
we replace the composite operator @xmath0 in the effective theory by the composite higgs operator .
we then obtain the effective gluon - gluon - higgs ( @xmath1- ) operator induced by colorons and find that its coefficient ( @xmath2 ) is proportional to @xmath3 , where @xmath4 and @xmath5 denote the coloron mass and the mass dynamically generated by colorons , respectively .
the contribution of colorons @xmath2 becomes comparable to the top - loop effect @xmath6 for @xmath7 and @xmath8 .
such a large dynamical mass can be realized in top - seesaw ( tss ) models consistently with the experimental value of the top quark mass ( @xmath9 ) , while the dynamical mass itself is adjusted to @xmath9 in topcolor assisted technicolor models ( tc2 ) .
we find that the coloron contribution @xmath2 can be sizable in a certain class of tss models : the contribution of colorons ( the top - loop ) is dominant in the real ( imaginary ) part of the @xmath10 amplitude for the higgs boson mass @xmath11 of the order of 1 tev . on the other hand , enhancement of the top - yukawa coupling
becomes important in tc2 .
we can observe signatures of the higgs boson in tc2 with @xmath12 gev even at the tevatron run ii as well as at the lhc .
we estimate @xmath13 for an integrated luminosity of 2 fb@xmath14 and @xmath15 gev at the tevatron run ii . |
many statistical data sets involve covariates @xmath0 that are error - contaminated versions of their true unobserved counterpart @xmath1 .
however , the measurement error often does not fit the classical error structure @xmath2 with @xmath3 independent from @xmath1 .
a common occurrence is , in fact , the opposite situation , in which @xmath4 with @xmath5 independent from @xmath0 , a situation often referred to as berkson measurement error [ @xcite , @xcite , @xcite ] .
a typical example is an epidemiological study in which an individual s true exposure @xmath1 to some contaminant is not observed , but instead , what is available is the average concentration @xmath0 of this contaminant in the region where the individual lives .
the individual - specific @xmath1 randomly fluctuate around the region average @xmath0 , resulting in berkson errors .
existing approaches to handle data with berkson measurement error [ e.g. , @xcite , @xcite ] unfortunately require the distribution of the measurement error to be known , or to be estimated via validation data , which can be costly , difficult or impossible to collect .
( in classical measurement error problems , the distribution of the error can be identified from repeated measurements via a kotlarski - type equality [ @xcite , @xcite ] .
however , such results do not yet exist for berkson - type measurement error . ) a popular approach to relax the assumption of a fully known distribution of the measurement error is to allow for some adjustable parameters in the distributions of the variables and their relationships , and solve for the parameter values that best reproduce various conditional moments of the observed variables , under the assumption that this solution is unique .
this approach has been used , in particular , for polynomial specifications [ @xcite ] and , more recently , for a very wide range of parametric models [ wang ( @xcite ) ] .
the present paper goes beyond this and provides a formal identification result and a general nonparametric regression method that is consistent in the presence of berkson errors , without requiring the distribution of the measurement error to be known a priori . instead , the method relies on the availability of a so - called instrumental variable [ e.g. , see chapter 6 in @xcite ] to recover the relationship of interest .
for instance , in the epidemiological study of the effect of particulate matter pollution on respiratory health we consider in this paper , suitable instruments could include ( i ) individual - level measurement of contaminant levels that can even be biased and error - contaminated or ( ii ) incidence rates of diseases other than the one of interest that are known to be affected by the contaminant in question .
our estimation method essentially proceeds by representing each of the unknown functions in the model by a truncated series ( or a flexible functional form ) and by numerically solving for the parameter values that best fits the observable data .
although such an approach is easy to suggest and implement , it is a challenging task to formally establish that such a method is guaranteed to work in general .
first , there is no guarantee that the solution ( i.e. , parameter values that best match the distribution of the observable data ) is unique .
second , estimation in the presence of a number of unknown parameters going to infinity with sample size is fraught with convergence questions .
can the postulated series represent the solution asymptotically ?
is the parameter space too large to obtain consistency ? is the noise associated with estimating an increasing number of parameters kept under control ?
our solution to these problems is two - fold .
first , we target the most difficult obstacle by formally establishing identification conditions under which the regression function and the distribution of all the unobserved variables of the model are uniquely determined by the distribution of the observable variables .
a second important aspect of our solution to the berkson measurement error problem is to exploit the extensive and well - developed literature on nonparametric sieve estimation [ e.g. , @xcite , @xcite , @xcite ] to formally address the potential convergence issues that arise when nonparametric unknowns are represented via truncated series with a number of terms that increases with sample size .
these theoretical findings are supported by a simulation study and the usefulness of the method is illustrated with an epidemiological application to the effect of particulate matter pollution on respiratory health .
we consider a regression model of the general form @xmath6 where the function @xmath7 is the ( unknown ) relationship of interest between @xmath8 , the observed outcome variable and @xmath1 , the _ unobserved _ true regressor , while @xmath9 is a disturbance .
information regarding @xmath1 is only available in the form of an observable proxy @xmath0 contaminated by an error @xmath5 .
equation ( [ eqz ] ) assumes the availability of an instrument @xmath10 , related to @xmath1 via an unknown function @xmath11 and a disturbance @xmath12 .
our goal is to estimate the function @xmath13 in ( [ eqy ] ) nonparametrically and without assuming that the distribution of the measurement error @xmath5 is known .
[ as by - products , we will also obtain @xmath11 and the joint distribution of all the unobserved variables . ] to this effect , we require the following assumptions , which are very common in the literature focusing on nonlinear models with measurement error [ e.g. , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ] . [
condindep]the random variables @xmath0 , @xmath5 , @xmath14 , @xmath12 are mutually independent .
note that assumption [ condindep ] implies the commonly - made `` surrogate assumption '' @xmath15 , as can be seen by the following sequence of equalities between conditional densities : @xmath16 .
[ condloc]the random variables @xmath5 , @xmath9 , @xmath17 are centered ( i.e. , the model s restrictions preclude replacing @xmath18 by @xmath19 for some nonzero constant @xmath20 , and similarly for @xmath9 and @xmath12 ; this includes either zero mean , zero mode or zero median , e.g. ) .
as our approach relies on the availability of an instrument @xmath10 to achieve identification , it is instructive to provide practical examples of suitable instruments in common settings .
although the use of instrumental variables has historically been more prevalent in the econometrics measurement error literature [ @xcite ] , instruments are gathering increasing interest in the statistics literature , especially in the context of measurement error problems [ see chapter 6 entitled `` instrumental variables '' in @xcite and the numerous references therein ] . note
that instrument equation ( [ eqz ] ) is entirely analogous to ( [ eqy ] ) , the equation generating the main dependent variable .
hence , the instrument is nothing but another observable `` effect '' caused by @xmath1 via a general nonlinear relationship @xmath11 .
let us consider a few examples , which were inspired by some of the case studies found in @xcite , @xcite and @xcite .
epidemiological studies . in these studies ,
the dependent variable @xmath8 is typically a measure of the severity of a disease or condition , while the true regressor @xmath21 is someone s true but unobserved exposure to some contaminant .
the average concentration @xmath0 of this contaminant in the region where the individual lives is , however , observed .
the error on @xmath0 is berkson - type because individual - specific @xmath1 typically randomly fluctuate around the region average @xmath0 . in this setup ,
multiple plausible instruments are available : a measurement of contaminant concentration in the individual s house ( these would be error - contaminated by classical errors , since the concentration at a given time randomly fluctuates around the time - averaged concentration which would be relevant for the impact on health ) . thanks to the flexibility introduced by the function @xmath11 in ( [ eqz ] ) , these measurements can even be biased . they can therefore be made with a inexpensive method ( that can be noisy and not even well - calibrated ) , making it practical to use at the individual level . hence , it is possible to combine ( i ) accurate , but expensive , region averages that are not individual - specific ( @xmath0 ) and ( ii ) inexpensive , inaccurate individual - specific measurements ( @xmath10 ) to obtain consistent estimates .
another plausible instrument could be a measure of the severity of another disease or condition that is _ known _ to be caused by the contaminant .
the fact that it is _ caused by _ the contaminant , introduces an error structure which is consistent with equation ( [ eqz ] ) .
other measurable effects due to the contaminant ( e.g. , the results of saliva or urine tests for the presence of contaminants ) could also serve as instruments .
clearly these measurements are not units of exposure , but the function @xmath11 can account for this . experimental studies .
researchers may wish to study how an effect @xmath8 ( e.g. , the production of some chemical ) is related to some imposed external conditions @xmath0 ( e.g. , oven or reactor temperature ) , but the true conditions @xmath21 experienced by the sample of interest may deviate randomly from the imposed conditions ( e.g. , temperature may not be completely uniform ) . in this case ,
an instrument @xmath10 could be ( i ) another `` effect '' ( e.g. , the amount of another chemical ) that is known to be caused by @xmath21 or ( ii ) a measurement of @xmath1 that is specific to the sample of interest but that may be very noisy or even biased ( e.g. , it could be an easier - to - take temperature measurement after the experiment is completed and the sample has partly cooled down ) .
self - reported data .
@xcite have argued that individuals reporting data ( e.g. , their food intake , or exercise habits ) are sometimes aware of the uncertainty in their estimates of @xmath1 and , as a result , try to report an average @xmath0 over all plausible estimates consistent with the information available to them , thus leading to berkson - type errors , because the individuals try to make their prediction error independent from their report . in this setting , an instrument @xmath10 could be another observable outcome variable @xmath10 that is also related to @xmath1 .
we now formally state conditions under which the berkson measurement error model can be identified with the help of an instrument .
let @xmath22 , @xmath23 , @xmath24 and @xmath25 denote the supports of the distributions of the random variables @xmath8 , @xmath0 , @xmath1 and @xmath10 , respectively .
we consider @xmath26 and @xmath10 to be jointly continuously distributed ( with @xmath27 , @xmath28 , @xmath29 and @xmath30 with @xmath31 ) .
accordingly , we assume the following .
[ conddens]the random variables @xmath32 admit a bounded joint density with respect to the lebesgue measure on @xmath33 . all marginal and conditional densities are also defined and bounded .
we use the notation @xmath34 and @xmath35 to denote the density of the random variable @xmath36 and the density of @xmath36 conditional on @xmath37 , respectively .
lower case letters denote specific values of the corresponding upper case random variables .
next , as in many treatments of errors - in - variables models [ @xcite , @xcite , @xcite , @xcite , schennach ( @xcite ) ] , we require various characteristic functions to be nonvanishing .
we also place regularity constraints on the two regression functions of the model .
[ condinv]for all @xmath38 , @xmath39 \neq0 $ ] and for all @xmath40 , @xmath41 \neq0 $ ] ( where @xmath42 ) . [ condnodup]@xmath43 and @xmath44 are one - to - one ( but not necessarily onto ) .
[ condcont]@xmath45 is continuous .
assumption [ condnodup ] is somewhat restrictive when @xmath1 has a dimension larger or equal to the ones of @xmath8 ( or @xmath10 ) .
fortunately , it is often possible to eliminate this problem by re - defining @xmath8 ( and @xmath10 ) to be a vector containing auxiliary variables in addition to the outcome of interest , in order to allow for enough variation in @xmath8 ( and @xmath10 ) to satisfy assumption [ condnodup ] .
each of these additional variables need not be part of the relationship of interest per se , but does need to be affected by @xmath1 is some way . in that sense , such auxiliary variables would also be a type of `` instrument . ''
our main identification result can then be stated as follows .
( note that the theorem also holds upon conditioning on an observed variable @xmath46 , so that additional , correctly measured , regressors can be straightforwardly included . )
[ thid]under assumptions [ condindep][condcont ] , given the true observed conditional density @xmath47 , the solution @xmath48 to the functional equation @xmath49 for all @xmath50 , @xmath51 , @xmath52 is unique ( up to differences on sets of null probability measure ) .
a similar uniqueness result holds for the solution @xmath53 to @xmath54\\[-8pt ] & & \qquad = f_{x } ( x ) \int f_{\delta z } \bigl ( z - h \bigl ( x^{\ast } \bigr ) \bigr ) f_{\delta y } \bigl ( y - g \bigl ( x^{\ast } \bigr ) \bigr ) f_{\delta x^{\ast } } \bigl ( x^{\ast}-x \bigr ) \,dx^{\ast}.\nonumber\end{aligned}\ ] ] establishing this result demands techniques radically different from existing treatment of berkson error models , such as the spectral decomposition of linear operators [ see @xcite for a review ] , which are emerging as powerful alternatives to the ubiquitous deconvolution techniques that are typically applied in classical measurement error problems .
the proof can be found in the and can be outlined as follows .
assumption [ condindep ] lets us obtain the following integral equation relating the joint densities of the observable variables to the joint densities of the unobservable variables : @xmath55 from which equation ( [ eqfyxz ] ) follows directly .
uniqueness of the solution is then shown as follows .
equation ( [ eqpreindep ] ) defines the following operator equivalence relationship : @xmath56 where we have introduced the following operators : @xmath57 ( z ) & = & \int f_{y , z|x } ( y , z|x ) r ( x ) \,dx,\nonumber\\ { } [ f_{z|x^{\ast}}r ] ( z ) & = & \int f_{z|x^{\ast } } \bigl ( z|x^{\ast } \bigr ) r \bigl ( x^{\ast } \bigr ) \,dx^{\ast } , \nonumber\\ { } [ f_{z|x}r ] ( z ) & = & \int f_{z|x } ( z|x ) r ( x ) \,dx,\\ { } [ d_{y;x^{\ast}}r ] \bigl ( x^{\ast } \bigr ) & = & f_{y|x^{\ast } } \bigl ( y|x^{\ast } \bigr ) r \bigl ( x^{\ast } \bigr ) , \nonumber\\ { } [ f_{x^{\ast}|x}r ] \bigl ( x^{\ast } \bigr ) & = & \int f_{x^{\ast } |x } \bigl ( x^{\ast}|x \bigr ) r ( x ) \,dx \nonumber\end{aligned}\ ] ] for some sufficiently regular but otherwise arbitrary function @xmath58 .
note that , in the above definitions , @xmath59 is viewed as a parameter ( the operators do not act on it ) and that @xmath60 is the operator equivalent of a diagonal matrix .
next , we note that the equivalence @xmath61 also holds [ e.g. , by integration of ( [ eqleqlll ] ) over all @xmath50 ] . we can then isolate @xmath63 @xmath64 and substitute the result into ( [ eqleqlll ] ) to yield , after rearrangements , @xmath65 where all inverses can be shown to exist over suitable domains under our assumptions .
equation ( [ eqdiag ] ) states that the operator @xmath66 admits a spectral decomposition .
the operator to be `` diagonalized '' is defined in terms of observable densities , while the resulting eigenvalues @xmath67 ( contained in @xmath60 ) and eigenfunctions @xmath68 ( contained in @xmath69 ) provide the unobserved densities of interest .
a few more steps are required to ensure uniqueness of this decomposition , which we now briefly outline .
one needs to ( i ) invoke a powerful uniqueness result regarding spectral decompositions [ theorem xv 4.5 in @xcite ] , ( ii ) exploit the fact that densities integrate to one to fix the scale of the eigenfunctions , ( iii ) handle degenerate eigenvalues and ( iv ) uniquely determine the ordering and indexing of the eigenvalues and eigenfunctions .
this last , and perhaps most difficult , step , addresses the issue that both @xmath68 and @xmath70 , for some one - to - one function @xmath71 , are equally valid ways to state the eigenfunctions that nevertheless result in different operators @xmath69 . to resolve this ambiguity , we note that for any possible operator @xmath72 satisfying ( [ eqdiag ] ) , there exist a unique corresponding operator @xmath63 , via equation ( [ eqgivexsx ] ) .
however , only one choice of @xmath69 leads to an operator @xmath63 whose kernel @xmath73 satisfies assumption [ condloc ] .
hence , @xmath73 , @xmath74 and @xmath75 are identified , from which the functions @xmath76 , @xmath77 , @xmath78 , @xmath45 and @xmath79 can be recovered by exploiting the centering restrictions on @xmath5 , @xmath9 and @xmath12 . an operator approach has recently been proposed to address certain types of nonclassical measurement error problems [ @xcite ] , but under assumptions that rule out berkson - type measurement errors : it should be emphasized that , despite the use of operator decomposition techniques similar to the ones found in @xcite ( hereafter hs ) , it is impossible to simply use their results to identify the berkson measurement error model considered here , for a number of reasons .
first , the key condition ( assumption 5 in hs ) that the distribution of the mismeasured regressor @xmath0 given the true regressor @xmath1 is `` centered '' around @xmath1 does not hold for berkson errors . consider the simple case where the berkson measurement error is normally distributed and so are the true and mismeasured regressors .
the distribution of @xmath0 given @xmath80 is a normal centered at @xmath81 .
hence , there is absolutely no reasonable measure of location ( mean , mode , median , etc . )
that would yield the appropriate centering at @xmath82 that is needed in assumption 5 of hs .
in addition , one can not simply replace the assumption of centering of @xmath0 given @xmath83 ( as in hs ) by a centering of @xmath1 given @xmath0 ( as would be required for berkson errors ) and hope that theorem 1 in hs remains valid .
hs exploit the fact that , in a conditional density , there is no jacobian term associated with a change of variable in a conditioning variable ( here @xmath1 ) .
however , with berkson errors , the corresponding change of variable would not take place in the conditional variables , and a jacobian term would necessarily appear , which makes the approach used in hs fundamentally inapplicable to the berkson case . solving
this problem involves ( i ) using a different operator decomposition than in hs and ( ii ) using a completely different approach for `` centering '' the mismeasured variable.=1 a referee suggested an alternative argument ( formalized in the ) that makes a more direct connection with theorem 1 in hs but under the additional assumption that @xmath10 and @xmath1 have the same dimension .
such an assumption is rather restrictive because it will often result in the assumption that @xmath11 is one - to - one ( assumption [ condnodup ] ) being violated .
for instance , if @xmath1 is scalar and we have access to two instruments @xmath84 and @xmath85 such that neither @xmath86 $ ] nor @xmath87 $ ] are strictly monotone , then @xmath11 is not one - to - one for either instrument used in isolation .
however , the mapping @xmath88,e [ z_{2}|x^{\ast } ] ) $ ] will typically be one - to - one , except for really exceptional cases .
hence , allowing for the dimensions of @xmath1 and @xmath10 to differ is important .
nevertheless , even assuming away this problem , such an approach still requires a different technique for centering @xmath1 than the one used in hs .
that said , both hs and the current paper rely on operator spectral decomposition as an alternative to conventional convolution / deconvolution techniques , and it appears likely that these new techniques will find applications in a number of other measurement error models .
observe that our identification result is also useful in a parametric and semi - parametric context , as it provides the confidence that , under simple conditions , the model is identified .
rank conditions that would need to be verified on a case - by - case basis in any given parametric model are automatically implied by our identification results in a wide class of models .
also , although @xmath0 is allowed to be random throughout , considering @xmath0 to be fixed poses no particular difficulty , since equation ( [ eqfyxz ] ) provides a valid conditional likelihood function in that case .
as discussed in @xcite , a number of extensions of the method are possible : ( i ) relaxing the independence between @xmath0 and @xmath18 to allow for some heteroskedasticity in the measurement error and ( ii ) combining classical and berkson errors , a possibility considered in , for example , @xcite , @xcite , @xcite and @xcite .
it can also be shown that some extensions are not plausible , such as assuming that both the measurement equation ( [ eqxs ] ) and the instrument equation ( [ eqz ] ) have a berkson error structure [ @xcite ] .
a natural way to obtain a nonparametric estimator of the model is to substitute truncated series approximations into ( [ eqfyxz ] ) or ( [ eqfyxz3 ] ) for each of the unknown functions and construct a log likelihood function to be maximized numerically with respect to all coefficients of the series [ e.g. , @xcite ] .
such sieve - based estimators have recently found applications in a variety of measurement error problems [ e.g. , @xcite , @xcite , @xcite , @xcite , among others ] .
below we first define our estimator before establishing its consistency .
we represent the regression functions @xmath7 and @xmath89 as @xmath90 where @xmath91 is some sequence ( indexed by the truncation parameters @xmath92 ) of progressively larger sets of basis functions indexed by @xmath93 while @xmath94 is a vector of coefficients to be determined .
the @xmath95 could be some power series , trigonometric series , orthogonal polynomials , wavelets or splines , for instance .
the double indexing by @xmath96 and @xmath92 is useful to allow for splines , where changing the number of knots modifies all the basis functions .
a similar expansion in terms of basis functions @xmath97 ( with truncation parameter @xmath98 ) is used for the density of each disturbance @xmath99 , @xmath100 where @xmath101 is a vector of coefficients to be determined , and @xmath102 is a user - specified `` baseline '' function .
the `` baseline '' function is convenient to reduce the number of terms needed in the expansion , when the approximate general shape of the density is known .
it is not strictly needed , however , and can be set to 1 .
either way , the method is fully nonparametric . a convenient choice of basis [ see @xcite ] is to take @xmath102 to be a gaussian and @xmath103 for any @xmath98 .
an important distinction with the functions @xmath7 and @xmath11 is that some constraints have to be imposed on the densities .
one constraint is needed to ensure centering ( assumption [ condloc ] ) , @xmath104 where , for some user - specified function @xmath105 , we define @xmath106 for instance , to impose zero mean on the disturbance @xmath107 , let @xmath108 . to impose zero median ,
let @xmath109 , where @xmath110 denotes an indicator function , while to impose zero mode , let @xmath111 ( a delta function derivative , in a slight abuse of notation ) .
another constraint is needed to ensure unit total probability : @xmath112 .
note that both types of constraints exhibit the computationally convenient property of being linear in the unknown coefficients .
given the above definitions , we can define an estimator of all unknown functions based on a sample @xmath113 and equation ( [ eqfyxz ] ) [ a corresponding estimator based on equation ( [ eqfyxz3 ] ) can be derived analogously ] .
let @xmath114 denote the minimizer of the sample log likelihood @xmath115 where @xmath116 @xmath117 , subject to @xmath118 for @xmath99 and subject to technical regularity constraints to be defined below .
estimators are then given by @xmath119\\[-8pt ] \hat{f}_{v } ( v ) & = & \hat{f}_{v}^ { ( k_{v } ) } \bigl ( v,\hat{\theta}_{v}^ { ( k_{v } ) } \bigr ) \qquad\mbox{for } v= \delta x^{\ast } , \delta y,\delta z. \nonumber\end{aligned}\ ] ] this type of estimator falls within the very general class of sieve nonparametric maximum likelihood estimators ( mle ) , whose asymptotic theory has received considerable attention over the last few decades [ e.g. , @xcite , @xcite , @xcite ] . here , we parallel the treatment of @xcite and @xcite to establish the consistency of the above procedure . although the consistency of sieve - type estimators has been previously established in very general settings under some high - level assumptions , our contribution is to provide very primitive sufficient conditions for consistency for the class of models considered here .
we first need to define the set in which the densities of interest reside .
the formal proof of consistency of the estimator requires this set to be compact , although this requirement appears to have little impact in practice .
in essence , compactness is helpful to rule out very extreme but rare events associated with very poor estimates .
it is a standard regularity condition ; see , for example , @xcite , @xcite , @xcite . a well - known type of infinite - dimensional but compact sets are those generated via boundedness and lipschitz constraints in an @xmath120 space . here
, we use a weighted lipschitz constraint in order to allow for densities supported on an unbounded set , while still maintaining compactness ( our treatment can be straightforwardly adapted to cover the simpler case where the variables are supported on finite intervals ) .
following @xcite , we enforce restrictions that avoid too rapid divergences in the log likelihood .
[ defsetf]let @xmath121 .
let @xmath37 be finite and strictly positive .
let @xmath122 be strictly positive and bounded function that is decreasing in @xmath123 , symmetric about @xmath124 and such that @xmath125 .
let @xmath126 $ ] such that @xmath127 .
let @xmath128 and @xmath129 be strictly positive and bounded functions with @xmath128 decreasing in @xmath123 and @xmath130 .
let @xmath131 .
we also define suitable norms and sets for the regression functions . here
, we need to allow for functions that diverge to infinity at controlled rates toward infinite values of their argument . in analogy with any existing global measure of expected error
, we also use a norm that downweights errors in the tails , which is consistent with the fact that the tails of a nonparametric regression function are always estimated with more noise , since there are fewer datapoints there .
[ defsetg]let @xmath132 by some given strictly positive , bounded and differentiable weighting function . for any function @xmath133 ,
let @xmath134 where @xmath135 .
let @xmath136 and @xmath137 where @xmath138 is a given positive function that is increasing in @xmath123 and symmetric about @xmath124 .
we can now state the regularity conditions needed .
[ condiid]the observed data @xmath139 are independent and identically distributed across @xmath140 .
[ condball]we have @xmath141 and @xmath142 . [
conddense]the set of functions representable as series ( [ expf ] ) and ( [ expg ] ) are , respectively , dense in @xmath143 ( in the norm @xmath144 ) and @xmath145 ( in the norm @xmath146 ) .
denseness results for numerous types of series are readily available in the literature [ e.g. , @xcite , @xcite ] .
although such results are sometimes phrased in a mean square - type norm rather than the sup norm used here , lemma [ lemnorm2 ] below [ proven in @xcite ] establishes that , within the sets @xmath143 and @xmath145 , denseness in a mean square norm implies denseness in the norms we use .
[ lemnorm2]let @xmath147 be a sequence in @xmath143
. then @xmath148 implies @xmath149 ( for @xmath143 and @xmath144 as in definition [ defsetf ] ) .
we also need standard boundedness and dominance conditions .
[ conddom]for any @xmath150 , @xmath151 for @xmath152 and @xmath153 as in definitions [ defsetg ] and [ defsetf ] , respectively .
[ condmomex ] there exists @xmath154 such that @xmath155 < \infty$ ] , where @xmath156 for @xmath157 and @xmath158 as in definitions [ defsetf ] and [ defsetg ] , respectively .
we can then state our consistency result [ proven in @xcite ] : [ thconsis]under assumptions [ conddens][condmomex ] , if @xmath159 , for @xmath160 , the estimators given by ( [ eqnest ] ) evaluated at the minimizer of ( [ eqnopti ] ) subject to ( [ eqnlincons ] ) , @xmath161 and @xmath162 and satisfying assumption [ conddom ] are such that @xmath163 , @xmath164 , @xmath165 , @xmath166 , @xmath167 , where the stared quantities denote the true values [ i.e. , the unique solution to ( [ eqfyxz ] ) ] .
the practical implementation of the above approach necessitates the selection of the number of terms @xmath98 in each of the approximating series .
theorem [ thconsis ] allows for a data - driven selection of the @xmath98 , since @xmath98 is allowed to be random . to select the @xmath98
, one can employ the bootstrap cross - validation model selection method based on the kullback
leibler ( kl ) criterion , shown by @xcite to be consistent even when the number of candidate models grows to infinity with sample size ( as it is here ) . in this method ,
a fraction @xmath168 of the sample is excluded at random and the remaining @xmath169 fraction is used to estimate the model parameters with given numbers @xmath170 of terms in the corresponding series .
the likelihood ( or kl criterion ) is then evaluated using the excluded fraction @xmath168 at the value of the estimated parameters found in the previous step .
the process is repeated many times with different random partitions of the sample into fractions @xmath168 and @xmath171 , to obtain an average kl criterion with a sufficiently small variance ( which can be estimated from the kl criterion of each random partitions ) .
this procedure is carried out for various trial choices of @xmath170 and the choice that yields the largest likelihood is selected .
this method is consistent asymptotically ( as sample size @xmath172 ) as @xmath173 and @xmath174 and under some mild technical regularity conditions stated in @xcite . our nonparametric approach nests parametric and semiparametric models .
these subcases can be easily implemented by replacing some , or all , of the nonparametric series approximations by suitable parametric models .
it is possible to obtain convergence rates and limiting distribution results , along the lines of @xcite or @xcite , although we do not do so here due to space limitations [ stating suitable regularity conditions , even in high - level form , is rather involved , as seen in the supplementary material of @xcite , which covers a related but different measurement error model ] .
it is , however , important to point out one important property .
sieve nonparametric mle is optimal in the following sense : under suitable regularity conditions , any sufficiently regular semiparametric functional of the nonparametric sieve mle estimates is asymptotically normal and root @xmath175 consistent and reaches the semiparametric efficiency bound for that functional ; see theorem 4 in @xcite .
this notion of optimality is a natural nonparametric generalization of the well - known efficiency of parametric maximum likelihood .
we now investigate the practical performance and feasibility of the proposed estimator via a simulation example purposely chosen to be a difficult case .
the data is generated as follows .
the distribution of @xmath0 is a uniform distribution over @xmath176 $ ] ( implying a standard deviation of @xmath177 ) .
we consider a thick - tailed @xmath178 distribution with 6 degrees of freedom scaled by @xmath179 as the distribution of @xmath18 .
the standard deviation of the error @xmath5 is almost identical to the one of the `` signal '' @xmath0 , thus making this estimation problem exceedingly difficult .
the distribution of @xmath9 is a logistic scaled by @xmath180 while the distribution of @xmath17 is a @xmath178 distribution with @xmath181 degrees of freedom scaled by @xmath182 .
the regression function has the form=1 @xmath183=0 which is only finitely many times differentiable , thus limiting the convergence rate of its series estimator in the measurement - error - robust estimator ( the naive estimator would be less affected since it would `` see '' a smoothed version of this function ) .
the instrument equation has a specification that is strictly convex and therefore tends to exacerbate the bias in many nonparametric estimators , @xmath184 a total of @xmath185 independent samples , each containing @xmath186 observations , were generated as above and fed into our estimator . for estimation purposes , the functions @xmath7 and @xmath11
are both represented by polynomials while the densities of @xmath5 , @xmath9 and @xmath12 are represented by a gaussian multiplied by a polynomial [ following @xcite , who establish that these choices satisfy a suitable denseness condition ] .
the gaussian is centered at the origin , but its width is left as a parameter to be estimated . note that the functional forms considered are not trivially nested within the space spanned by the truncated sieve approximation .
this was an intentional choice aimed at properly accounting for the nonparametric nature of the problem ( in which the researcher never has the fortune of selecting a truncated sieve fitting the true model exactly ) .
the integral in equation ( [ eqfyxz ] ) is evaluated numerically by discretizing the integral as a sum over the range @xmath187 $ ] in intervals of @xmath188 .
naive least - squares estimators ignoring measurement error ( i.e. , least - squares regressions of @xmath8 on @xmath0 and of @xmath10 on @xmath0 ) were used as a starting point for the numerical sieve optimization of the @xmath79 and @xmath45 functions , while the variances of the corresponding residuals were used to construct an initial gaussian guess for the optimization of all the error distributions . the simplex method due to @xcite ( also known as `` amoeba '' ) was used to carry out the numerical optimization of the log likelihood ( [ eqnopti ] ) with respect to all the parameters @xmath189 for @xmath190 and @xmath191 for @xmath192 simultaneously .
the constraints that the estimated densities and regression functions lie , respectively , in the sets @xmath143 and @xmath145 of the form given in definitions [ defsetf ] and [ defsetg ] are implied by bounds on the magnitude of the sieve coefficients @xmath193 and @xmath194 in ( [ expf ] ) and ( [ expg ] ) .
such constraints are easy to impose within the simplex optimization method : parameter changes that would yield violations of the bounds are simply rejected ( effectively assigned an `` infinite '' value)the simplex optimization method easily accommodates such extreme behavior in the objective function , since it does not rely on derivatives .
however , we found that these constraints are rarely binding in practice , unless the number of terms @xmath98 in the expansions is large [ @xcite reports a similar observation ] .
such large values of @xmath98 tend to be naturally ruled out via our data - driven selection method of the number of terms . to select the number of terms in the approximating series for a given sample
, we use the `` bootstrap cross - validation '' method described in section [ secest ] with a fraction @xmath195 and @xmath185 bootstrap replications .
trial values of the number of free parameters ( not counting parameters uniquely determined by zero mean and unit area constraints ) in the series representing @xmath196 each span the set @xmath197 while for @xmath79 , @xmath45 each span the set @xmath198 .
the optimal numbers of parameters ( kept constant during the replications ) were found to be @xmath199 ; @xmath200 ; @xmath201 ; @xmath202 ; @xmath203 .
figure [ figberksim ] summarizes the result of these simulations , where a naive nonparametric series least - squares estimator ignoring measurement error ( i.e. , least - square regressions of @xmath8 on @xmath0 and of @xmath10 on @xmath0 ) with the same number of sieve terms is also shown for comparison .
the reliability of the method can be appreciated by noting how closely the median of the replicated measurement - error - robust estimates matches the true model , while the naive estimator ignoring the presence of measurement error is considerably more biased , even missing the fact that the true regression function is nearly flat in the middle section and instead producing a very misleading linear shape despite the strong nonlinearity of the true model .
in fact , unlike the proposed estimator , the naive estimator is so significantly biased that any type of hypothesis test based on it would exhibit completely misleading confidence levels : the true model curves ( for @xmath79 and @xmath45 ) almost always lies beyond the 95% or 5% percentiles of the estimator distribution .
overall , the proposed measurement - error - robust estimator exhibits low variability and low bias at the reasonable sample size of @xmath186 .
the bias is not exactly zero in a finite sample because our estimator is a nonlinear functional of sample averages and because the sieve approximation necessarily has a limited accuracy in a finite sample .
nevertheless , the fact that our estimator performs so well in the presence of measurement error of such large magnitude is a strong indication of its practical usefulness .
this behavior is not specific to this model we have tested the method in other simulation settings ; see @xcite .
numerous studies have sought to quantify the effect of air pollution on respiratory health [ e.g. , @xcite ] .
specifically , there is a growing concern regarding the effect of small particulate matter [ @xcite , @xcite ] .
a key difficulty with such studies is that air quality monitors are not necessarily located near the subjects being affected by air pollution , implying that the main regressor of interest is mismeasured .
our approach to this question relies on very comprehensive country - wide data collected by environment protection agency ( epa ) and the center for disease control ( cdc ) in the united states .
pollution levels are taken from epas monitor values report criteria air pollutants database for year 2005 .
epas data provides point measurements of the particulate matter levels ( we focus on so - called 95th percentile level of pm2.5 particles , those having less than 2.5 micrometers in diameter ) at various monitoring stations throughout the united states , from which we construct state - averaged pollution levels ( our @xmath0 variable , measured in @xmath204 g of particles per m@xmath205 ) .
we do so because pollution data is only available for a small fraction of counties , and even where it is available , the nature of its measurement error is complex ( it could be a mixture of classical and berkson errors ) . by constructing state - level averages ,
we average out the randomness in monitor measurements while leaving the randomness in the individual exposure untouched , thus obtaining a valid berkson error - contaminated estimate of the pollution level experienced by individuals from each state , whether they live in a county with a monitoring station or not .
each individual faces an exposure equal to the state average plus an unknown random noise due to his / her precise geographic whereabouts and lifestyle .
health data is obtained from the publicly available `` cdc wonder '' database entitled `` mortality underlying cause of death '' for year 2005 . to measure respiratory health ,
we use data on causes of death , which offers the advantage that it is very comprehensive and accurate ( medical professionals are required to collect it and there is no reliance on voluntary surveys ) .
one limit to the completeness of the data is that , for some counties , the data is `` suppressed '' ( for privacy reasons ) or labeled as `` unreliable '' by the cdc and were therefore omitted from our sample .
our dependent variable of interest ( @xmath8 ) is the rate ( per 10,000 ) of death due to `` chronic lower respiratory diseases '' ( e.g. , asthma , bronchitis , emphysema ) , while our instrument ( @xmath10 ) is the rate ( per 10,000 ) of death resulting from `` lung diseases due to external agents '' ( e.g. , pneumoconiosis due to organic or inorganic dust , coalworker s pneumoconiosis ) .
the rationale is to use , as an instrument , a variable that is clearly expected to be affected by pollution levels .
this variable indirectly provides information regarding the true level of pollution , so that the effect of pollution ( if any ) on the variable of interest can be more accurately assessed .
we employ county - level data on causes of death because they are readily available without concerns for patient privacy issues .
moreover , the cdc provides age - corrected death rates , thus correcting for demographic differences between counties .
we construct our sample by matching mortality data via counties and matching pollution data via states , resulting in 1305 observations over as many counties and covering all 51 states .
a limitation of our approach is that it does not control for other possible confounding effects , for example , if the proportion of smokers differs between industrial and nonindustrial cities .
however , such a limitation is common in studies of this kind [ as noted in @xcite ] .
we use the same types of sieves and computational methods as in the simulation example and select the number of terms using the `` bootstrap cross - validation '' method described in section [ secest ] with a fraction @xmath195 and @xmath185 bootstrap replications .
trial values of the number of free parameters in the series representing @xmath196 span the range @xmath206 while trial values of the number of terms in the series representing @xmath79 and @xmath45 span the range @xmath207 ( increasing any one of the @xmath98 beyond that range resulted in clearly worse performances ) .
the optimal numbers of free parameters ( not counting parameters uniquely determined by zero mean and unit area constraints ) were found to be @xmath208 ; @xmath209 ; @xmath210 ; @xmath211 ; @xmath212 .
pointwise 90% confidence bands around the nonparametric estimates were obtained using the standard bootstrap [ see , e.g. , @xcite for general conditions justifying its use ] with 100 replications .
results are shown in figure [ figapp ] .
a few observations are in order .
first , our measurement error - robust estimator is perfectly able to detect a clear monotone relationship between @xmath8 and @xmath1 and between @xmath10 and @xmath1 with useful confidence bands , despite the use of a fully nonparametric approach .
second , although the distribution of the measurement error is difficult to estimate ( as reflected by the wide confidence bands ) , the impact of this uncertainty on the main function of interest [ @xmath213 is fortunately very limited .
the 90% confidence bands indicate that the presence of substantial measurement error is consistent with the data : the measurement error is of the order of 10 @xmath204g / m@xmath205 , whereas the observed @xmath0 roughly ranges from 10 to 40 @xmath204g / m@xmath205 .
third , the distribution of @xmath9 exhibits nonnegligible asymmetry , thus illustrating the drawbacks of methods merely assuming normality of all the error terms .
in contrast , the distributions of @xmath5 and @xmath12 are apparently very close to symmetric ( this is a conclusion of the formal model selection procedure , not an assumption ) . for comparison purposes
, we also naively regress the dependent variables ( @xmath8 or @xmath10 ) on the mismeasured regressor @xmath0 using a conventional least squares ( thereby neglecting measurement error ) with a polynomial specification with the same number of terms as our berkson model . a first troubling observation from this exercise
[ see figure [ figapp](b ) ] is that the naive estimate of @xmath214 is not monotone , although in the region where it is unexpectedly decreasing , the confidence bands do not rule out a constant response .
second , it is perhaps counter - intuitive that the confidence bands for the naive estimator are sometimes larger than the corresponding bands for the measurement error - robust estimator .
this is a consequence of the fact that correcting for berkson errors amounts to an operation akin to convolution ( rather than deconvolution , as in classical measurement errors ) . unlike deconvolution
, convolution is a noise - reducing operation , effectively averaging observations of @xmath8 over a wide range of values of @xmath0 to yield an estimate the expected value of @xmath8 given a specific value of @xmath0 .
this phenomenon is probably also responsible for the more reasonable ( i.e. , increasing ) behavior of the response for the measurement error - robust estimate .
finally , the measurement error - robust regression function often lies at or beyond the 95% or 5% percentiles of the naive estimator distribution ; see figure [ figapp](b ) .
this implies that the level of any statistical test would be severely biased .
for instance , the confidence bands of the naive estimator would reject our best estimate of @xmath214 obtained with the measurement - error robust procedure . in summary
, this application example serves to illustrate that ignoring berkson errors can be seriously misleading in nonlinear settings .
not only is the shape of the estimated response considerably affected , but statistical inferences based on a measurement error - blind method would be seriously biased .
this application example also shows that our fully nonparametric and measurement error - robust method works well at sample sizes typically available in real data sets , without assuming the knowledge of the distribution of the measurement error .
let @xmath215 with @xmath216 @xmath217 for some @xmath218 denote the set of all bounded functions in @xmath219 endowed with the usual @xmath220 norm . also , whenever we state an equality between functions in @xmath215 , we mean that their difference is zero in the @xmath220 norm .
we provide two proofs of theorem 1 .
the first one , suggested by a referee , relies on the additional assumptions that ( i ) @xmath10 and @xmath1 have the same dimension and ( ii ) @xmath45 and its inverse are differentiable . assumption ( i ) makes assumption [ condnodup ] unlikely to hold , but enables a somewhat direct application of theorem 1 in @xcite .
the second proof relaxes those assumptions .
it borrows some of the operator techniques from @xcite , yet requires considerable changes in the approach we focus here on the aspects of the proof that differ .
proof theorem [ thid ] ( simple special case ) let variables from @xcite be denoted by the corresponding uppercase letter with tildes and make the following assignments : @xmath221 .
we now verify the 5 assumptions of theorem 1 in @xcite . to verify assumption 1 , we observe that the densities of @xmath222 and @xmath223 are related through:@xmath224 where the density @xmath225 exists by assumption [ conddens ] , and @xmath226 exists by assumption [ condnodup ] . the jacobian @xmath227 matrix is only defined if @xmath1 and @xmath10 ( and therefore @xmath228 ) have the same dimension and is finite and nonsingular under the assumption that @xmath45 and its inverse are differentiable . a similar argument can be used for marginals and conditional distributions . to verify assumption 2 , we note that our model can be written in terms of tilded variables as @xmath229 to verify assumption 2(i ) , we write @xmath230 where we have used , in turn , ( i ) the equality @xmath231 and the fact that changes of variables in the conditioning variables do not introduce jacobian terms , ( ii ) the fact that conditioning on @xmath232 is equivalent to conditioning on @xmath233 , ( iii ) assumption [ condindep ] , ( iv ) the relationship between @xmath9 and @xmath8 via ( [ eqtilde1 ] ) and ( v ) the equality @xmath234 .
assumption 4 requires that @xmath236 for @xmath237 .
this can be verified as follows : @xmath238 by invoking ( i ) the definition of @xmath9 , ( ii ) independence of @xmath9 from @xmath1 ( and therefore @xmath228 ) , ( iii ) the fact that @xmath239 implies @xmath240 since @xmath7 and @xmath11 are one - to - one by assumption [ condnodup ] and so is @xmath241 .
theorem 1 in @xcite then allows us to conclude that the joint distribution of @xmath242 is identified .
however , in order to identify the distribution of @xmath243 , we need to identify @xmath11 . to this effect
, we note that , conditional on @xmath244 , the fluctuations in @xmath228 are entirely caused by fluctuations in @xmath18 by equation ( [ eqtilde2 ] ) .
moreover , @xmath5 is independent from @xmath0 , hence @xmath245 where the left - hand side was previously identified and where the jacobian term is well defined by assumptions [ condnodup ] and the assumed differentiability of @xmath226 .
the jacobian can be identified by integrating ( [ eqfj ] ) with respect to @xmath246 to yield @xmath247 . by varying @xmath248 while keeping @xmath249 fixed in equation ( [ eqfj ] )
, we can identify the density @xmath78 up to a shift of @xmath226 .
assumption [ condloc ] , pins down what the shift should be , so that @xmath250 is identified for any given @xmath249 .
since @xmath89 is one - to - one by assumption [ condnodup ] , @xmath251 uniquely determines @xmath11 .
hence , the joint distribution of @xmath252 is identified . finally , noting that @xmath253 ( by assumption [ condindep ] ) , then establishes the identification of @xmath214 with the help of assumption [ condloc ] .
the definition of marginal and conditional densities in combination with assumption [ condindep ] lead to the following sequence of equalities : @xmath254 or , equivalently , @xmath255 as in @xcite , this integral equation can be written more conveniently as an operator equivalence relation @xmath256 by introducing the operators defined in equation ( [ eqdefop ] ) , which are acting on an arbitrary @xmath257 [ or @xmath258 .
similarly , one can show that @xmath259 and thus @xmath260 . by assumptions
[ conddens ] , [ condindep ] , [ condinv ] , [ condnodup ] , [ condcont ] and lemma [ leminj ] below , we know that @xmath261 admits an inverse on the range of @xmath69 ( and therefore the range of @xmath262 ) , and we can write @xmath263 substituting ( [ eqop2i ] ) into ( [ eqop1 ] ) , we obtain @xmath264 by assumptions [ conddens ] , [ condindep ] , [ condinv ] , [ condnodup ] , [ condcont ] and lemma [ leminj ] below again , @xmath262 admits an inverse .
moreover , by lemma 1 in @xcite , the domain of @xmath265 is dense in @xmath266 , and we can then write @xmath267 equation ( [ eqdiagpr ] ) states that the operator @xmath66 admits a spectral decomposition , where the eigenvalues are given by the @xmath268 for @xmath269 ( for a fixed @xmath59 ) defining the operator @xmath60 while the eigenfunctions are the functions @xmath270 for @xmath269 defining the kernel of the operator @xmath69 . as usual , the knowledge of a linear operator [ e.g. , @xmath262 ] only determines the value of its kernel [ e.g. , @xmath272 everywhere except on a set of null lebesgue measure .
the resulting equivalence class exactly matches the usual equivalence class for probability densities with respect to the lebesgue measure , so identifiability of the model is not affected .
the operator to be diagonalized is entirely defined in terms of observable densities while the decomposition provides the unobserved densities of interest .
to ensure uniqueness of this decomposition , we employ four techniques .
first , a powerful result from spectral analysis [ theorem xv 4.5 in @xcite ] ensures uniqueness up to some normalizations .
second , the a priori arbitrary scale of the eigenfunctions is fixed by the requirement that densities must integrate to one .
third , to avoid any ambiguity in the definition of the eigenfunctions when degenerate eigenvalues are present , we use assumption [ condnodup ] and the fact that the eigenfunctions [ which do not depend on @xmath59 , unlike the eigenvalues @xmath273 must be consistent across different values of the dependent variable @xmath59 .
these three steps are described in detail in @xcite and are not repeated here .
the fourth step [ which differs from the approach taken in @xcite ] is to rule out that the eigenvalues @xmath274 and eigenfunctions @xmath275 could be indexed by a different variable without affecting the operator @xmath66 .
( this issue is analogous to the nonunique ordering of the eigenvalues and eigenvectors in matrix diagonalization . )
suppose that the eigenfunctions can be indexed by another value , that is , they are given by latexmath:[$f_{z|\tilde{x}^{\ast } } ( \cdot variable related to @xmath82 through @xmath277 for some one - to - one function @xmath71 .
is also measurable , for otherwise @xmath278 would not be a proper random variable . ] under this alternative indexing , all the assumptions of the original model must still hold with @xmath82 replaced by @xmath279 , so a relationship similar to ( [ eqfzxsidx ] ) would still have to hold , for the same observed @xmath280 @xmath281 or , in operator notation , @xmath282 . in order for @xmath283 to be a valid alternative density
, it must satisfy the same assumptions ( and their implications ) as @xmath75 .
in particular , the fact that @xmath69 is invertible ( established above via lemma [ leminj ] ) must also hold for @xmath284 .
hence , for any alternative @xmath284 , there is a unique corresponding @xmath285 , given by @xmath286 .
we can find a more explicit expression for @xmath287 as follows .
first note that we trivially have that @xmath288 since @xmath289 and @xmath71 is one - to - one . by performing the change of variable @xmath290 in ( [ eqfzxsidx ] )
, we obtain @xmath291 where the measure @xmath204 is defined , via @xmath292 for any measurable set @xmath293 , where @xmath294 denotes the lebesgue measure and @xmath295 . from this
we can conclude the equality between the two following measures : @xmath296 by comparison to equation ( [ eqaltfzxs ] ) and the uniqueness of the measure@xmath297 due to the injectivity of the @xmath284 operator , shown in lemma [ leminj ] in the general case where the domain of @xmath284 could include finite signed measures .
we will now show that @xmath287 necessarily violates assumption [ condloc ] ( with @xmath5 replaced by @xmath298 ) , unless @xmath299 is the identity function .
since @xmath300 with @xmath5 independent from @xmath0 , we have @xmath301 and by a similar reasoning @xmath302 with @xmath303 . equation ( [ eqds1 ] ) then becomes @xmath304 now , for a given @xmath248 , consider radom
nikodym derivative of @xmath305 with respect to the lebesgue measure @xmath306 , which is , by definition ( almost everywhere ) equal to @xmath307 , a bounded function by assumption [ conddens ] . by equation ( [ eqmueq ] ) , the existence of the radom
nikodym derivative of the left - hand side implies the existence of the same radom
nikodym derivative on the right - hand side , and we can write @xmath308 almost everywhere . integrating both sides of the equation over all @xmath309 ,
we obtain ( after noting that points where the equality may fail have null measure and therefore do not contribute to the integral ) , @xmath310 , since densities integrate to @xmath311 , which implies that @xmath312 , that is , @xmath204 is also the lebesgue measure .
it follows from ( [ eqallf ] ) that , almost everywhere @xmath313 in order for assumption [ condloc ] to hold for both @xmath314 and @xmath5 , we must have that @xmath315 , when viewed as a function of @xmath249 for any given @xmath248 , is centered at @xmath316 , and we must simultaneously have that @xmath317 , when viewed as a function of @xmath82 for any given @xmath248 , is centered at @xmath318 , that is , @xmath319 .
the two statements are only compatible if @xmath320 .
thus , there can not exist two distinct but observationally equivalent parametrization of the eigenvalues / eigenfunctions .
hence we have shown , through equation ( [ eqdiagpr ] ) , that the unobserved functions @xmath268 and @xmath321 are uniquely determined ( up to an equivalence class of functions differing at most on a set of null lebesgue measure ) by the observed function @xmath322 .
next , equation ( [ eqop2i ] ) implies that @xmath323 is uniquely determined as well .
once @xmath268 and @xmath324 are known , the functions @xmath214 and @xmath325 can be identified by exploiting the centering restrictions on @xmath9 , @xmath5 and @xmath12 , for example , @xmath326 if @xmath9 is assumed to have zero mean .
next , @xmath327 can be straightforwardly identified , for example , @xmath328 for any @xmath269 .
similar arguments yield @xmath325 and @xmath329 from @xmath330 as well as @xmath331 from @xmath73 .
it follows that equation ( [ eqfyxz ] ) has a unique solution .
the second conclusion of the theorem then follows from the fact that both @xmath322 and @xmath332 are uniquely determined ( except perhaps on a set of null lebesgue measure ) from @xmath333 .
the following lemma is closely related to proposition 2.4 in @xcite .
it is different in terms of the spaces the operators can act on and more general in terms of the possible dimensionalities of the random variables involved .
[ leminj]let @xmath334 and @xmath10 be generated by equations ( [ eqxs ] ) and ( [ eqz ] ) .
let @xmath335 be the set of finite signed measures on a given set @xmath336 or @xmath25 [ and note that @xmath335 includes @xmath337 as a special case , in the sense that for any function in @xmath338 , there is a corresponding measure @xmath339 whose radom nikodym derivative with respect to the lebesgue measure is @xmath58 ] . under assumptions [ condindep ] , [ conddens ] , [ condinv ] , [ condnodup ] and [ condcont ] , the operators @xmath341 , @xmath342 and @xmath343 , defined in ( [ eqdefop ] ) , are injective mappings .
first , one can verify that @xmath344 implies that @xmath345 and similarly for @xmath69 and @xmath262 , since the ( conditional ) densities involving variables @xmath346 and @xmath10 are bounded by assumption [ conddens ] and are absolutely integrable .
we now verify injectivity of @xmath69 . by assumptions
[ condindep ] , [ conddens ] and equation ( [ eqz ] ) , we have , for any @xmath347 , @xmath348 ( z ) = \int f_{z|x^{\ast } } \bigl ( z|x^{\ast } \bigr ) \,dr \bigl ( x^{\ast } \bigr ) = \int f_{\delta z } \bigl ( z - h \bigl ( x^{\ast } \bigr ) \bigr ) \,dr \bigl ( x^{\ast } \bigr).\ ] ] next , let @xmath349 denote the signed measure assigning , to any measurable set @xmath350 , the value@xmath351 and note that @xmath349 is a finite signed measure since @xmath352 is .
then , we can express @xmath353 as @xmath354 ( z ) = \int f_{\delta z } \bigl ( z-\tilde{x}^{\ast } \bigr ) \,d\tilde{r } \bigl ( \tilde{x}^{\ast } \bigr),\ ] ] that is , a convolution between the probability measure of @xmath12 ( represented by its lebesgue density ) and the signed measure @xmath349 ; see chapter 5 in @xcite . by the convolution theorem for signed measures [ theorem 5.1(iii ) in @xcite ] , one can convert the convolution ( [ eqconvol1 ] ) into a product of fourier transforms , @xmath355 where @xmath356 ( z ) e^{\mathbf{i}\zeta z}\,dz$ ] , @xmath357 $ ] and @xmath358 .
since @xmath359 , the characteristic function of @xmath12 , is nonvanishing by assumption [ condinv ] , we can isolate @xmath360 as @xmath361 since there is a one - to - one mapping between finite signed measures and their fourier transforms [ by theorem 5.1(i ) in @xcite ] , @xmath349 can be recovered as the unique signed measure whose fourier transform is @xmath362 .
we now show that the signed measure @xmath349 uniquely determines the measure @xmath363 .
let @xmath364 for any measurable @xmath365 , and note that @xmath366 is also measurable since @xmath45 is continuous by assumption [ condcont ] . then observe that by assumption [ condnodup ] , @xmath367 if and only if @xmath368 , and we have @xmath369 since @xmath370 is arbitrary , the knowledge of @xmath371 uniquely determines the value assigned to any measurable set by the signed measure @xmath363 .
injectivity of @xmath63 is a special case of the above derivation ( with @xmath372 replaced by @xmath346 ) , in which @xmath45 is the identity function .
finally , injectivity of @xmath262 is implied by the injectivity of @xmath69 and @xmath63 , since @xmath61 by assumption [ condindep ] and equations ( [ eqxs ] ) and ( [ eqz ] ) . | this paper establishes that so - called instrumental variables enable the identification and the estimation of a fully nonparametric regression model with berkson - type measurement error in the regressors .
an estimator is proposed and proven to be consistent .
its practical performance and feasibility are investigated via monte carlo simulations as well as through an epidemiological application investigating the effect of particulate air pollution on respiratory health .
these examples illustrate that berkson errors can clearly not be neglected in nonlinear regression models and that the proposed method represents an effective remedy . |
such classical configurations of lines as the @xmath0 lines on a complex cubic surface or the @xmath1 bitangents to a smooth quartic plane curve have been extensively studied in the 19th century ( see e.g. @xcite ) .
their automorphism groups were known , but only at the beginning of the 20th century were their close connections with the weyl groups of the root systems @xmath2 and @xmath3 , recognized , in particular through the relation with del pezzo surfaces of degree three and two , respectively @xcite .
del pezzo surfaces of degree one provide a similar identification of the diameters of the root system @xmath7 , with the @xmath8 tritangent planes to a canonical space curve of genus @xmath9 .
can we go beyond the weyl groups and find a connection with the lie groups themselves ?
the @xmath0 lines on the cubic surface are in natural correspondence with the weights of the minimal representation of @xmath2 , from which the lie group can be recovered as the stabilizer of a cubic form that already appears in elie cartan s thesis ; faulkner showed how to define this form in terms of the @xmath10 tritangent planes .
a similar phenomenon holds for the @xmath1 bitangents to the quartic plane curve , which can be put in correspondence with pairs of opposite weights of the minimal representation of @xmath3 . in both cases
the connection between the lie group and its weyl group is particularly close because of the existence of a minuscule representation . for @xmath7
the minimal representation is the adjoint one and is no longer minuscule .
the first aim of this paper is to use these connections with the lie groups , or rather the lie algebras @xmath11 , @xmath4 , @xmath5 , to shed a new light on the work of the classical geometers on line configurations .
our main idea is that each time we consider a semisimple lie subalgebra , the restriction of the minimal representation branches into a direct sum of subrepresentations , and consequently the weights split into special subsets forming interesting subconfigurations . in the case of @xmath11 and the @xmath0 lines we get the following correspondence : [ cols="^,^,^ " , ] in the case of @xmath4 and the @xmath1 bitangents the notion of steiner complexes of bitangents make a natural appearance .
they are special sets of @xmath12 bitangents which can be put in correspondence with positive roots , and also with points of a @xmath13-dimensional projective space over @xmath14 .
this space is endowed with a natural symplectic form , and indeed the weyl group of @xmath3 is closely connected with the finite symplectic group @xmath15 .
this leads to a very interesting finite symplectic geometry whose lines are known to correspond to the so called _ syzygetic triads _ of steiner sets .
we prove that planes in this geometry are in correspondence with what we call _ fano heptads _ of bitangents .
the upshot is a finite geometry modeling the symplectic geometries related to the third line of freudenthal magic square , whose last term is precisely @xmath4 @xcite .
very interestingly , this leads to a beautiful model of @xmath4 and its minimal representation which , rather unexpectedly , turns out to be closely related with the fano plane and the octonionic multiplication .
indeed , recall that @xmath16 , the cayley algebra of octonions , can be defined as the eight - dimensional algebra with a basis @xmath17 , with multiplication rule encoded in an oriented fano plane .
( 30,11)(-15,-1 ) ( -5.5,0 ) ( -3.6,4.6)@xmath18(3.3,4.6)@xmath19 ( -.3,-.8)@xmath20(-6.1,-.8)@xmath21 ( 5.5,-.8)@xmath22(-.2,9.7)@xmath23(.9,3)@xmath24 _ figure 1 . octonionic multiplication _
this means that @xmath25 if @xmath26 are three distinct points on one of the projective lines in this plane , with a plus sign if and only if @xmath27 gives the cyclic orientation fixed on the line .
we define an @xmath16-grading on a lie algebra @xmath28 to be a decomposition @xmath29 such that @xmath30\subset\fh_k$ ] if @xmath25 .
in particular @xmath31 is a subalgebra and each @xmath32 is an @xmath31-module .
more is true : for any point @xmath33 and any line @xmath34 in the fano plane , the direct sums @xmath35 are subalgebras of @xmath28 , so that we really have a configuration of lie algebras defined by a plane projective geometry .
our discussion of fano heptads lead us to discover that @xmath4 has a natural structure of an @xmath16-graded algebra , compatible with its action on the minimal representation @xmath36 .
indeed , attach to each line @xmath34 of the fano plane a two dimensional vector space @xmath37 .
then we can describe @xmath4 and @xmath36 as follows : @xmath38 going a little deeper in the lie algebra structure , we will discover a natural connection with the multiplication table of the cayley algebra .
this leads to an amusing interpretation of the isomorphism @xmath6 in terms of _ harmonic cubes _ , and a permutation representation of the group of the bitangents on the triangles of the fano plane .
a similar description of @xmath5 also exists , and the biggest two exceptional lie algebras appear as plane projective geometries whose points are copies of @xmath39 and @xmath40 , and whose lines are copies of @xmath41 and @xmath42 , respectively .
moreover , this octonionic model of @xmath5 makes obvious the existence of the multiplicative orthogonal decomposition that was a key ingredient in thompson s construction of the sporadic simple group denoted @xmath43 or @xmath44 ( see @xcite , chapters 3 and 13 ) .
it would certainly be interesting to use this octonionic model , suitably adapted , to construct forms of @xmath5 over arbitrary fields .
classically , two unifying perspectives on the line configurations we are interested in have been particularly successful .
we briefly discuss the connection with our present approach
. * theta characteristics . *
bitangents to the plane quartic curve ( a canonical curve of genus @xmath45 ) , as well as tritangent planes to the canonical curve of genus @xmath46 , can be interpreted as odd theta - characteristics .
since the theta - characteristics can be seen as points of an affine space over the half - periods of the curve , this leads to an interpretation in terms of finite symplectic geometries in dimension @xmath47 over the field @xmath14 .
this was developped in great detail by the classical geometers , in particular by coble ( @xcite , chapter ii ) .
for example the theta - characteristics can be understood as the quadrics whose associated polarity is the natural symplectic form .
isotropic linear spaces also have natural geometric interpretations . for @xmath45 one
recovers the connection that we already mentionned between @xmath48 and @xmath15 . for @xmath46 ,
the weyl group @xmath49 is the automorphism group of the lines in the del pezzo surface of degree one , whose canonical model is a double covering of a quadratic cone , branched along a canonical sextic curve .
as noticed by schottky , there is a unique even theta - characteristic vanishing at the vertex of the cone , which explains why the automorphism group of the tritangent planes is an orthogonal group @xmath50 rather than a symplectic group .
* semi - regular polytopes .
* gosset seems to have been the first , at the very beginning of the 20th century , to understand that the lines on the cubic surface can be interpreted as the vertices of a polytope , whose symmetry group is precisely the automorphism group of the configuration .
coxeter extended this observation to the 28 bitangents , and todd to the 120 tritangent planes .
du val and coxeter provided systematic ways to construct the polytopes , which are denoted @xmath51 for @xmath52 and live in @xmath53 dimensions @xcite .
they have the characteristic property of being semi - regular , which means that the automorphism group acts transitively on the vertices , and the faces are regular polytopes . in terms of lie theory
they are best understood as the polytopes in the weight lattices of the exceptional simple lie algebras @xmath54 , whose vertices are the weights of the minimal representations .
coxeter investigated in great detail their semi - regular sub - polytopes @xcite .
algebraically , this amounts to identifying certain lie subalgebras of the @xmath54 .
but coxeter does not describe how the full polytopes are organized around these special sub - polytopes . in a sense
this is what we will be doing in this paper , with the nice conclusion that it leads to a very natural , unified and easy - going description of ( at least part of ) the classical combinatorics of the line configurations , as well as new insights in the fascinating structure of the exceptional lie algebras . _
acknowledgements : _ i thank i. dolgachev and p.e .
chaput for their useful comments .
a classical elementary configuration of lines is the reye configuration below , obtained from a cube in a three dimensional projective space ( see @xcite and @xcite ) .
this configuration can be understood as a central projection of the @xmath55-cell , one of the regular polytopes in four dimensions .
the vertices of this polytope are given by the roots of the root system @xmath56 ( we use @xcite as a general reference on root systems ) .
( 30,17)(-10,-1 ) ( -5.5,0 ) _ figure 2 . the reye configuration _ as one
can easily see on figure 2 , there is a unique way to partition the points of the reye configuration into three types , in such a way that each line contains exactly one point of each type .
this decomposes the @xmath55-cell into three @xmath57-cells given by the vertices of three hypercubes .
each of these defines a root subsystem of @xmath56 of type @xmath58 . restricting the adjoint representation of @xmath59 to the corresponding subalgebra , a product of four copies of @xmath60
, we obtain the _ four - ality model _
@xcite @xmath61 whose existence is indicated by the shape of the affine dynkin diagram @xmath62 .
( here @xmath63 are two dimensional complex vector spaces .
note that the construction works on the reals to give the split form @xmath64 . )
four - ality reduces to the classical cartan triality through the morphism @xmath65 induced by the permutation of the three partitions of four objects in two pairs . in terms of representations
, this translates into the permutation of the three non equivalent eight dimensional representations of @xmath59 : @xmath66 among the models of the exceptional lie algebras that we will meet in the sequel , most will be derived from the _ triality model _
first defined , in a more general context , by allison in @xcite , and rediscovered in @xcite .
the idea is to associate to a ( complexified ) real normed algebra @xmath67 , its _ triality algebra _ @xmath68 with its three natural modules @xmath69 .
then for any pair @xmath70 of normed algebras , the direct sum @xmath71 has a natural lie algebra structure .
this leads to the famous freudenthal magic square , whose fourth line @xmath72 is the series of exceptional lie algebras @xmath73 .
the lie algebras on the second and third lines of the magic square are endowed each with a special module : @xmath74 with the cubic jordan algebra @xmath75 , and @xmath76 with the zorn algebra @xmath77 .
the natural inclusions @xmath78 then lead to the _ binary _ and _ ternary _ models for the exceptional lie algebras : @xmath79 this extends to @xmath80 , whose ternary model is the four - ality model related to the reye configuration . note also that the triality models can be interpreted as @xmath81-graded lie algebras , with a similar definition to the one we introduced for @xmath16-gradings .
the configuration of the 27 lines on a smooth cubic surface in @xmath82 have been thoroughly investigated by the classical algebraic geometers .
it has been known for a long time that the automorphism group of this configuration can be identified with the weyl group of the root system of type @xmath2 , of order @xmath83 @xcite . moreover , the minimal representation @xmath84 of the simply connected complex lie group of type @xmath2 has dimension @xmath0 .
this is a minuscule representation , meaning that the weight spaces are lines and that the weyl group @xmath85 acts transitively on the weights .
in fact one can recover the lines configuration of the cubic surface by defining two weights to be incident if they are not orthogonal with respect to the unique ( up to scale ) invariant scalar product .
conversely , one can recover the action of the lie group @xmath2 on @xmath84 from the line configuration .
faulkner defines a cubic form on @xmath84 as the sums of signed monomials associated to the tritangent planes @xcite .
the stabilizer of that cubic form in @xmath86 is precisely @xmath2 .
note that the polarization of this cubic form is a symmetric bilinear map @xmath87 .
identifying appropriately @xmath84 with @xmath88 we get an algebra structure which is known to concide with the exceptional complex jordan algebra @xmath89 .
the closed @xmath2-orbit in the projectivization @xmath90 is known as the complex _ cayley plane _ @xmath91 and should be thought of as the projective plane over the cayley algebra of octonions .
being the orbit of any weight space it is circumscribed to the schoute polytope @xmath92 , which appears as a discrete version of the cayley plane . in particular the @xmath93 lines incident to a given line correspond to the polar quadric or @xmath16-line ( whose euler characteristic is @xmath93 ) .
the property that two general @xmath16-lines on the cayley plane have a unique intersection point , thus mirrors the obvious fact that , two concurrent lines on the cubic surface being given , there exists a unique line meeting both .
it is well known that most of the interesting subgroups of @xmath85 can be realized as stabilizers of some subconfigurations .
it seems not to have been noticed before that most of them also have natural interpretations in terms of branching . by this
we mean that we can find a subalgebra of @xmath11 such that the restriction of the representation in @xmath84 splits in such a way that the relevant subconfiguration can immediately be read off .
there is a general recipe to identify semisimple subalgebras of a simple complex lie algebra , that we illustrate with the case of @xmath11 ( see @xcite , chapter 6 ) .
one begins with the affine dynkin diagram , which in the case we are interested in has a remarkable threefold symmetry : ( 30,7)(-10,-5 ) ( 0,0)(2,0)5@xmath94 ( 0.5,.4)(2,0)4(1,0)1.6 ( 4,-2)@xmath94 ( 4,-4)@xmath94 ( 4.3,-3.4)(0,2)2(0,1)1.5 then we choose a set of nodes , that we mark in black . suppressing these nodes we get the dynkin diagram ( usually disconnected ) of a semisimple lie subalgebra @xmath95 of @xmath11 which is uniquely defined up to conjugation .
the weyl group @xmath96 of this semisimple lie algebra is a subgroup of @xmath85 , also uniquely defined up to conjugation .
we get three types of data : 1 . _ combinatorial data _ :
@xmath96 can be realized as the stabilizer of a certain subconfiguration of the @xmath0 lines , encoded in the marked dynkin diagram ; 2 .
_ representation theoretic data _ : as an @xmath95-module , @xmath84 splits into a direct sum of irreducible components ; 3 .
_ geometric data _
: the @xmath95-components encode certain special subvarieties of the cayley plane . _
example 1_. we mark two of the three extreme nodes
. then @xmath97 and @xmath98 .
( 30,7)(-10,-5 ) ( 0,0)(2,0)5@xmath94 ( 0,0)(8,0)2@xmath99 ( 0.5,.4)(2,0)4(1,0)1.6 ( 4,-2)@xmath94 ( 4,-4)@xmath94 ( 4.3,-3.4)(0,2)2(0,1)1.5 the index of @xmath96 in @xmath85 is @xmath0 : this subgroup is just the stabilizer of some line in the configuration .
in fact @xmath95 is the semisimple part of the lie algebra of the stabilizer of a one - dimensional weight space @xmath34 , which defines a point on the cayley plane and can be identified with one of the lines of the configuration .
the branching , i.e. the decomposition of @xmath84 as an @xmath95-module , gives @xmath100 the @xmath57-dimensional half - spin representation @xmath101 can be identified with the tangent space to the cayley plane at @xmath34 ; combinatorially , the sixteen weight spaces generating @xmath101 give the sixteen lines which do not meet @xmath34 ; geometrically , the intersection of the cayley plane with its tangent space at @xmath34 is a cone over a ten dimensional spinor variety .
the @xmath93-dimensional natural representation @xmath102 encodes the normal space to the cayley plane at @xmath34 ; combinatorially , the ten weight spaces generating @xmath102 give the ten incident lines to @xmath34 . note that this representation is self - dual , so its weights occur in opposite pairs corresponding to incident pairs of incident lines to @xmath34 .
geometrically , the intersection of the cayley plane with @xmath103 is the polar eight - dimensional quadric , a copy of the projective line @xmath104 over the cayley algebra . _
example 2_. we mark the three extreme nodes . in this case
@xmath105 and @xmath106 .
( 30,7)(-10,-5 ) ( 0,0)(2,0)5@xmath94 ( 0,0)(8,0)2@xmath99 ( 0.5,.4)(2,0)4(1,0)1.6 ( 4,-2)@xmath94 ( 4,-4)@xmath99 ( 4.3,-3.4)(0,2)2(0,1)1.5 by restricting the previous case we get the branching @xmath107 where @xmath108 are the three eight dimensional representations of @xmath109 , which we deliberately avoid to distinguish since they are exchanged by cartan s triality .
the three lines @xmath110 are pairwise incident , hence they are the three intersection lines of the cubic surface with a _ tritangent plane_. note that the index of @xmath96 in @xmath85 is @xmath111 .
since we have a sixfold ambiguity on the order of the three lines , we recover the classical fact that the cubic surface has exactly @xmath10 tritangent planes .
once we have fixed these three lines , each of them has eight more incident lines coming into four pairs , and corresponding to the pairs of opposite weights of one of the eight - dimensional representations of @xmath59 .
note that this exhausts the @xmath0 lines .
( 20,14)(-2,-.5 ) ( 0,0 ) ( 9,12)@xmath112 ( 15,1)@xmath113 ( 4.5,12)@xmath114 _ figure 3 .
triality from the @xmath0 lines _
the sum of the weights of @xmath110 is zero , and this characterizes triples of lines on a tritangent plane .
geometrically , we have three eight - dimensional quadrics on the cayley plane , any two of them meeting exactly in one point . in terms of plane projective geometry , these three quadrics are projective lines which are the sides of a self - polar triangle . _
example 3_. now we mark a unique node , which is neither extremal nor central
. then @xmath115 and @xmath116 .
this leads to the binary model of @xmath11 .
( 30,7)(-10,-5 ) ( 0,0)(2,0)5@xmath94 ( 0.5,.4)(2,0)4(1,0)1.6 ( 4,-2)@xmath99 ( 4,-4)@xmath94 ( 4.3,-3.4)(0,2)2(0,1)1.5 the index of @xmath96 in @xmath85 is @xmath117 .
the branching gives @xmath118 where @xmath102 denotes the six - dimensional natural representation of @xmath119 , and @xmath120 the natural representation of @xmath60 .
the twelve weights of @xmath121 split into six pairs @xmath122 where @xmath123 have the same component over @xmath120 .
of course @xmath124 is a double - six , and there are exactly @xmath117 double - sixes on the cubic surface .
note that the node that we have marked is the node of the dynkin diagram of @xmath2 that defines the adjoint representation .
this explains the correspondence between double - sixes and pairs of opposite roots of @xmath2 ( @xcite , 10.1.5 ) .
geometrically , such a pair of roots defines a point of the adjoint variety , the projectivization of the minimal nilpotent orbit in the lie algebra @xmath11 .
the image of its action on @xmath84 has minimal rank , namely @xmath125 @xcite , and is a maximal linear space in the cayley plane , with a weight basis given by one half of the double - six .
the other half can be recovered from the similar action on @xmath88 , whose image is again a maximal linear space in ( a dual copy of ) the cayley plane .
explicitely , let @xmath126 denote the set of weights of @xmath84 ( the @xmath85-orbit of the fundamental weight @xmath127 , in the notation of @xcite ) .
then the double - six @xmath128 associated to a root @xmath129 ( up to sign ) , considered as a set of such weights , is @xmath130 now given two double - sixes @xmath128 and @xmath131 , they can have only two different relative positions , following that @xmath132 or not .
in the first case they are said to be _
azygetic _ , and @xmath133 .
then @xmath134 or @xmath135 is again a root and defines a third double - six @xmath136 , which is azygetic to both @xmath128 and @xmath131 .
there exist @xmath8 such _ azygetic triads of double - sixes _ , corresponding to the @xmath8 subsystems of type @xmath137 of the root system @xmath2 . in the latter case the double - sixes
are _ syzygetic _ , and @xmath138 .
as indicated by the marked dynkin diagram ( 30,7)(-10,-5 ) ( 0,0)(2,0)5@xmath94 ( 0.5,.4)(2,0)4(1,0)1.6 ( 4,-2)@xmath99(2,0)@xmath99(6,0)@xmath99 ( 4,-4)@xmath94 ( 4.3,-3.4)(0,2)2(0,1)1.5 we can complete such a pair of double - sixes into a _
syzygetic tetrad _ of double - sixes .
if we let @xmath139 be the two - dimensional natural representations of the four copies of @xmath60 corresponding to the white nodes of the diagram , the minimal representation of @xmath11 branches to @xmath140 and the associated tetrad of double - six is given by the weights of the four submodules @xmath141 the number of syzygetic tetrads is the number of root subsystems of type @xmath58 in @xmath2 , that is @xmath142 . _
example 4_. now we mark the central node . then @xmath143 and @xmath144 .
this leads to the ternary model of @xmath11 .
( 30,7)(-10,-5 ) ( 0,0)(2,0)5@xmath94 ( 0.5,.4)(2,0)4(1,0)1.6 ( 4,0)@xmath99 ( 4,-2)@xmath94 ( 4,-4)@xmath94 ( 4.3,-3.4)(0,2)2(0,1)1.5 the index of @xmath96 in @xmath85 is @xmath145 .
the branching gives @xmath146 where @xmath147 denote the natural representations of the three copies of @xmath148 in @xmath95 .
the @xmath0 weights are thus split into three bunches of nine .
consider for example the nine weights @xmath149 of @xmath150 , where the @xmath151 are the weights of @xmath120 and the @xmath152 those of @xmath153 ( note that both sets sum to zero ) .
display these weights on a @xmath154 square by putting @xmath155 in the box @xmath156 , the indices being considered modulo three . then three weights on the same line or column have they sum equal to zero , and thus define a tritangent plane .
we have obtained what is called a _
steiner set _
nine lines obtained as the intersections of two trihedra .
moreover , we have split the @xmath0 lines into three such sets forming a so - called _ steiner triple system_. steiner sets are in correspondence with root subsystems of type @xmath137 of the root system @xmath2 .
the orthogonal to such a subsystem is the orthogonal product of two other @xmath137-subsystems .
this is why , a steiner set being given , there is a unique way to complete it into a triple system ( @xcite , 10.1.7 ) .
of course this is another manifestation of triality ( see @xcite ) !
note also that the invariant cubic form on @xmath84 can be characterized , up to scalar , as the unique @xmath85-invariant cubic form whose restriction to each factor of type @xmath157 is proportional to the determinant . geometrically , three steiner sets correspond to three @xmath158 s in @xmath159 , each cutting the cayley plane along a copy of the segre variety @xmath160 .
_ define two steiner sets to be incident if they have no common tritangent plane .
each steiner set is incident to exactly @xmath161 other sets , including the two special ones which complete it into a triple system .
if we choose one of these , exactly @xmath1 steiner sets are incident to both , including the remaining set in the triple system .
contrary to what we could be tempted to believe , the configuration of the remaining @xmath0 steiner sets is not combinatorially equivalent to that of the @xmath0 lines .
indeed , one can check that each of the @xmath0 steiner sets is incident to only eight of the other ones .
nevertheless , the steiner sets define an interesting regular graph , which has the same number of edges and vertices than the graph defined by the diameters of the polytope @xmath162 , although it is not combinatorially equivalent .
coxeter already noticed in @xcite that the @xmath163 triple systems can be interpreted as hexagons on the polytope @xmath162
the @xmath1 bitangents to a smooth plane quartic curve give rise to @xmath161 lines on the del pezzo surface of degree two defined as the double cover of the projective plane , branched over the quartic @xcite .
this line configuration has for automorphism group the index two normal subgroup @xmath164 of the weyl group of @xmath3 , which has order @xmath165 .
the lie group of type @xmath3 has a minimal representation @xmath36 of dimension @xmath161 , which is again minuscule .
the invariant forms are a symplectic form so that the @xmath161 weights split into @xmath1 pairs of opposite weights and a quartic form which can not be deduced , contrary to the case of the @xmath0 lines , solely from the configuration .
the weights of this representation form the _ gosset polytope _ @xmath166 .
this polytope appears as a discrete version of the minimal @xmath3-orbit in @xmath167 , which we call the _
freudenthal variety_. the affine dynkin diagram of type @xmath3 has a two fold symmetry : ( 30,5)(-10,-3 ) ( 0,0)(2,0)7@xmath94 ( 0.5,.4)(2,0)6(1,0)1.6 ( 6,-2)@xmath94 ( 6.3,-1.4)(0,1)1.5 again we deduce classical configurations of bitangents from markings of this diagram . _ example
1_. we mark the two opposite extreme nodes .
then @xmath168 , and @xmath169 has index @xmath1 in @xmath48 .
( 30,5)(-10,-3 ) ( 2,0)(2,0)5@xmath94 ( 0,0)(12,0)2@xmath99 ( 0.5,.4)(2,0)6(1,0)1.6 ( 6,-2)@xmath94 ( 6.3,-1.4)(0,1)1.5 indeed it is well - known that the stabilizer of any bitangent is a copy of the automorphism group of the @xmath0 lines . since the action of the latter is irreducible , the branching has to give an irreducible supplement , up to the sign of the weights .
and indeed we have the decomposition into @xmath95-modules @xmath170 geometrically , the factor @xmath171 appears as the tangent space to the freudenthal variety , and @xmath88 is the first normal space .
note that the intersection of the freudenthal variety with its tangent space is a cone over the cayley plane , whose discrete skeleton is precisely given by the @xmath0 lines . _
example 2_. mark only one node next to one of the opposite extremal nodes .
then @xmath172 , and @xmath96 has index @xmath173 in @xmath48 .
this leads to the binary model of @xmath4 .
( 30,5)(-10,-3 ) ( 0,0)(2,0)7@xmath94 ( 2,0)@xmath99 ( 0.5,.4)(2,0)6(1,0)1.6 ( 6,-2)@xmath94 ( 6.3,-1.4)(0,1)1.5 let @xmath174 denote the natural representations of @xmath60 and @xmath175 . here the branching gives the very simple decomposition @xmath176 where @xmath101 is one of the half - spin representations , of dimension @xmath177 .
the factor @xmath178 corresponds to a set of twelve bitangents .
since @xmath96 has an invariant quadratic form , its weights come into pairs of opposite weights .
we thus get six pairs of bitangents forming a _ steiner complex _ ( @xcite , 6.1.2 ) . since the node that we have marked defines the adjoint representation of @xmath4 on the dynkin diagram of type @xmath3 , the @xmath173 steiner complexes are in natural correspondence with the @xmath173 pairs of opposite roots in the root system @xmath3 .
from this perspective they play the same role as the double - sixes of lines on the cubic surface . geometrically ,
the twelve weight spaces of a steiner complex generate a @xmath179 in @xmath180 whose intersection with the freudenthal variety is a ten dimensional quadric .
if we denote by @xmath181 the set of weight of @xmath36 ( the @xmath48-orbit of the fundamental weight @xmath127 , in the notation of @xcite ) , the steiner complex associated to the root @xmath129 ( up to sign ) is @xmath182 now if @xmath183 is another root , we can have @xmath132 or not , with respectively @xmath184 and @xmath177 possibilities for @xmath185 , up to sign . in the first case , the corresponding root spaces generate in @xmath4 a copy of @xmath186 , and the steiner complexes are _ syzygetic _ , which means that @xmath187 .
the roots which are orthogonal both to @xmath129 and @xmath185 form a reducible root system of type @xmath188 .
in particular , there is a third steiner complex @xmath189 canonically associated to the pair @xmath190 , and syzygetic to both of them .
the characteristic property of the triple @xmath191 is that its union is the whole set of bitangents ( see example 6 below ) .
the number of such _
syzygetic triads of steiner complexes _
is @xmath192 . in the second case ,
the root spaces generate a copy of @xmath148 , and the steiner complexes are _ azygetic _ : @xmath193 . then @xmath134 ( or @xmath135 ) is again a root , and @xmath194 is azygetic both to @xmath195 and @xmath196 : we obtain an _
azygetic triad of steiner complexes_. the number of such triads is @xmath197 . _
example 3_. we mark the lowest node .
then @xmath198 , and @xmath199 has index @xmath117 in @xmath48 .
( 30,5)(-10,-3 ) ( 0,0)(2,0)7@xmath94 ( 0.5,.4)(2,0)6(1,0)1.6 ( 6,-2)@xmath99 ( 6.3,-1.4)(0,1)1.5 if we denote by @xmath102 the natural representation of @xmath200 , the branching gives @xmath201 once we fix a basis @xmath202 of @xmath102 , we can therefore identify each bitangent with a pair @xmath203 , with @xmath204 . such a notation seems to have been first introduced by hesse .
moreover , the weyl group of @xmath3 is generated by the symmetric group @xmath205 and the symmetries associated to the roots coming from the factor @xmath206 .
these symmetries are indexed by partitions @xmath207 of @xmath208 $ ] into disjoints fourtuples .
the induced action on the bitangents is given by @xmath209 this is classicaly called a _
bifid transformation_. _ example 4_. we mark two extreme but not opposite nodes .
then @xmath210 , and @xmath211 has index @xmath212 in @xmath48 .
( 30,5)(-10,-3 ) ( 2,0)(2,0)6@xmath94 ( 0,0)@xmath99 ( 0.5,.4)(2,0)6(1,0)1.6 ( 6,-2)@xmath99 ( 6.3,-1.4)(0,1)1.5 obviously this example comes from the previous one : we have just passed from @xmath200 to @xmath213 .
if we denote by @xmath102 the natural representation of @xmath213 , the branching gives @xmath214 we have thus distinguished a set of seven bitangents forming what is called an _ aronhold set _
( @xcite , 6.1.3 ) .
geometrically , the seven weight spaces of an aronhold set generate a @xmath215 which is a maximal linear space inside the freudenthal variety .
note that the aronhold sets generate the remaining basic representation @xmath216 of @xmath4 , the one corresponding to the lowest extremal node of the dynkin diagram . by this
we mean that among the lines of @xmath217 generated by the weight vectors forming aronhold sets , one has a dominant weight and generates a copy of @xmath216 .
we know that any irreducible representation of @xmath4 can then be constructed from @xmath36 , @xmath216 and the adjoint representation by natural tensorial operations . _
example 5_. we mark the central node
. then @xmath218 , and @xmath219 has index @xmath220 in @xmath48 .
( 30,5)(-10,-3 ) ( 0,0)(2,0)7@xmath94 ( 6,0)@xmath99 ( 0.5,.4)(2,0)6(1,0)1.6 ( 6,-2)@xmath94 ( 6.3,-1.4)(0,1)1.5 denote by @xmath221 the two - dimensional natural representation , by @xmath222 the two four dimensional ones
. then the branching gives @xmath223 note that @xmath224 and @xmath225 are self - dual , as well as @xmath221 , so we have distinguished two sets of @xmath125 bitangents forming a steiner complex .
the remaining sixteen bitangents are indexed by the weights of @xmath226 .
recall that four bitangents form a _
syzygetic tetrad _ when their eight tangency points are the eight intersection points of the plane quartic with some conic ( @xcite , 6.1.1 ) .
in terms of weights , this means that the four bitangents can be represented by weights summing to zero .
here we observe a phenomenon very similar to the property of a steiner set of lines on the cubic surface : our sixteen bitangents can be split into four syzygetic tetrads in essentially twelve different ways .
indeed our tetrads must be of the form @xmath227 for some permutation @xmath228 , and we have to find four permutations @xmath229 , @xmath230 , @xmath231 , @xmath232 such that @xmath233 for each @xmath33 and each @xmath234 .
the twelve possibilities are given by the four - tuples of permutations of the form @xmath235 , @xmath236 , @xmath237 , @xmath238 or @xmath235 , @xmath236 , @xmath239 , @xmath240 . _
example 6_. now we mark a non extremal node next to the central one .
then @xmath241 , and @xmath242 has index @xmath243 in @xmath48 .
this case leads to the ternary model of @xmath4 .
( 30,5)(-10,-3 ) ( 0,0)(2,0)7@xmath94 ( 4,0)@xmath99 ( 0.5,.4)(2,0)6(1,0)1.6 ( 6,-2)@xmath94 ( 6.3,-1.4)(0,1)1.5 let @xmath244 denote the natural representations of @xmath148 and @xmath119 .
the branching gives the decomposition @xmath245 where the middle factor is self - dual .
the factor @xmath226 splits , following the @xmath120-component , into three sixers of bitangents .
aggregating them two by two we get an azygetic triad of steiner sets . _
example 7_. we mark the two nodes next to the two opposite extremal nodes .
then @xmath246 , and @xmath96 has index @xmath247 in @xmath48 .
( 30,5)(-10,-3 ) ( 0,0)(2,0)7@xmath94 ( 2,0)(8,0)2@xmath99 ( 0.5,.4)(2,0)6(1,0)1.6 ( 6,-2)@xmath94 ( 6.3,-1.4)(0,1)1.5 let @xmath244 denote the natural representations of the two copies of @xmath60 . up to triality
we may suppose that the natural ( non spinorial ) eight - dimensional representation @xmath96 of @xmath59 is given by the lowest node .
then the branching gives the decomposition @xmath248 where the last two factors are two copies of the same module , put on duality by the symplectic form .
the symmetry of the picture suggests that we should introduce a supplementary copy of @xmath60 , hence three copies with natural representations @xmath249 such that @xmath250 indeed this is precisely what the trialitarian description of @xmath4 tells us ( see @xcite , theorem 4.1 ) .
we get a partition of the @xmath1 bitangents into three groups of eight and a group of four . adding the latter to the three former we get a
_ syzygetic triad of steiner complexes_. we have exactly @xmath251 such triads , and this is also the number of syzygetic tetrads ( @xcite , corollary 6.1.4 ) . indeed
the weights of @xmath252 define such a tetrad .
we recapitulate : there are natural correspondences between : 1 .
steiner complexes of bitangents and root subsystems of type @xmath253 ; 2 .
azygetic triples of steiner complexes and subsystems of type @xmath137 ; 3 .
syzygetic triples of steiner complexes and root subsystems of type @xmath56 of the root system @xmath3 .
we have already mentionned that the weyl group of @xmath3 can ( almost ) be identified with a classical group over a finite field ( see @xcite , exercice 3 of section 4 , p. 229 ) , namely @xmath254 this means that the incidence geometry of the @xmath1 bitangents should be interpreted as a symplectic six - dimensional geometry over the field with two elements .
such symplectic geometries appear on the third line of freudenthal s magic square , and the @xmath1 bitangents give a finite model of these geometries ( see e.g. @xcite ) . recall that a symplectic five - dimensional projective geometry has three types of elements : points , isotropic lines and isotropic planes . in the @xmath3 geometry ,
points and isotropic lines are points and lines in the freudenthal variety , while isotropic planes are in correspondence with maximal , ten dimensional quadrics on the freudenthal variety . in our finite geometry
, we have seen that the points correspond to the @xmath173 steiner complexes .
the @xmath251 isotropic lines are in correspondence with the syzygetic triads of steiner complexes , where we split the @xmath1 bitangents into three sets of twelve , with four bitangents common to the three .
it is obvious from that point of view that two syzygetic complexes can be uniquely completed into a syzygetic triad : indeed , they are syzygetic if they generate an isotropic line , and there is a unique other point on that line . what are the @xmath142 planes ?
classically , they are called gpel spaces ( see @xcite , chapter ii , 22 , and @xcite , chapter ix ) and play an important role in the study of the schottky problem . but
let us rather skip to our representation theoretic point of view .
a projective plane over @xmath14 is a fano plane .
it has @xmath255 points and @xmath255 lines .
so to get an isotropic plane we should be able to partition the @xmath1 bitangents into @xmath255 quadruples , in such a way that the complement of each of them can be split into three octuples defining syzygetic steiner complexes .
this looks like a combinatorial challenge but representation theory tell us what to do !
we have already used the fact that @xmath4 has a maximal semisimple lie algebra isomorphic to @xmath256 . thanks to the four - ality model of @xmath39
we can take four two - dimensional spaces @xmath257 and decompose @xmath258 then the three eight - dimensional representations decompose as @xmath259 then we plug that in the decomposition of the @xmath161-dimensional representation of @xmath3 given in example 7 above .
we obtain , if we denote @xmath260 : @xmath261 the seven steiner complexes that we are looking for are the sets of weights of the submodules @xmath262 and the seven syzygetic triads they form are given by the weights of the three submodules @xmath263 for @xmath27 one of the seven triples in the decomposition of @xmath36 .
note that these seven triples of indices have the crucial property that any pair of integers between one and seven , appear in one and only one of them .
otherwise said , they form a _ steiner triple system _ @xmath264 ( see e.g. @xcite ) .
up to isomorphism there is only one such system , given by the lines of the fano plane , as one can see on the next picture : ( 20,8)(-10,-1 ) ( -3.1,0)(1,0)6.2 ( -3.1,0)(3,5)3.1 ( 3.1,0)(-3,5)3.1 ( -3,0)(5,3)4.44 ( 3,0)(-5,3)4.44 ( 0,0)(0,1)5.2 ( -2.3,2.8)2(1.8,2.8)6(-.3,-1)4(-3.8,-1)1 ( 3.1,-1)5(-.3,5.5)3(.2,2.4)7 ( 0,1.7 ) in particular , by projective duality these lines can be represented by points of another fano plane .
we deduce that the stabilizer of our configuration in @xmath48 is the product of @xmath255 copies of @xmath137 by the automorphism group of the fano plane , which is nothing else than the klein group @xmath265 , with @xmath266 elements .
the index of this stabilizer is @xmath142 , as expected .
we call the corresponding configurations _
fano heptads of steiner complexes_. let us recapitulate the correspondence : @xmath267 the @xmath3-module @xmath36 has two basic invariants tensors , such that @xmath3 can be described as the group of linear transformations of @xmath36 preserving these tensors : a symplectic form and a quartic form .
the existence of such invariant forms is clear on our previous decomposition of @xmath36 .
indeed , each factor @xmath268 has a symplectic form induced by the choice of two - forms on each factor @xmath269 .
moreover , @xmath268 has an invariant quartic form given by the _ cayley hyperdeterminant _ , which is an equation of the dual variety of the segre product @xmath270 @xcite .
the invariant quartic form on @xmath36 is the unique @xmath48-invariant quartic form whose restriction to each factor @xmath268 is proportional to the cayley hyperdeterminant .
since their is a unique invariant quartic form on a factor @xmath268 , up to scalar , it certainly concides with the restriction of the invariant quartic form on @xmath36 .
conversely , we know that up to the action of the weyl group , the monomials in the invariant quartic form on @xmath36 are of three types ( see @xcite ) : products of two , equal or distinct , products of two variables associated to opposite weights ; other products of four variables associated to fourtuples of weights of sum zero ( thus defining syzygetic tetrads of bitangents )
. then we must give suitable relative coefficients to the sums of monomials of each type .
this is fixed by restriction to a single factor @xmath268 since the three types of monomials appear in the hyperdeterminant ( see @xcite ) . from the trialitarian construction of @xmath4 and the four - ality for @xmath39
we deduce the model : @xmath271 where @xmath272 is the following set of @xmath255 quadruples : @xmath273 these quadruples are in natural correspondence with lines : simply associate to a line the four points of its complement .
moreover , the action on @xmath36 can be recovered geometrically : each quadruple @xmath274 in @xmath272 can be seen as a complete quadrangle in the fano plane , with three pairs of opposite sides which are sent one to the other by the @xmath4-action restricted to @xmath275 .
let us rather try to reconstruct the lie bracket .
consider two factors @xmath275 and @xmath276 : we have two indices @xmath277 in common , and the third point of the line generated by @xmath278 and @xmath279 is @xmath280 .
the restriction of the lie bracket defines a map @xmath281 such that for some non zero constant @xmath282 , @xmath283 = \theta_{\a,\b}\omega ( x_i , y_i)\omega ( x_j , y_j ) x_k\ot x_l\ot y_m\ot y_n.\ ] ] indeed this is the unique equivariant map up to scalar , and it must be non zero because of the properties of the lie bracket in a semisimple lie algebra .
the skew symmetry of the lie bracket then implies that @xmath284 the jacobi identity can be expressed in the following way : for each triangle @xmath285 in the fano plane , we have the relation @xmath286 the fano plane has @xmath1 triangles , hence @xmath161 quadratic relations .
let @xmath287 according to the following rule : the multiplication table of the canonical basis @xmath288 of the imaginary octonions is given by @xmath289 then the relations above are satisfied . denote by @xmath290 , where @xmath291 , the canonical basis of the octonions , ( see the introduction ) .
our claim amounts to the identity @xmath292 when @xmath293 are distinct and not aligned . to prove this we need to remember that the cayley algebra , although not associative , is _ alternative _ @xcite .
this means that the associator @xmath294 is an alternating function of the arguments .
using the fact that @xmath295 when @xmath296 are distinct , we deduce that @xmath297 which proves our claim .
this means that the model that we have found for @xmath4 really has a very close connection with the octonions .
we can reformulate our discussion as follows .
the exceptional complex lie algebra @xmath4 has a natural structure of an @xmath16-graded algebra , given in terms of points @xmath33 and lines @xmath34 on the fano plane by @xmath298 moreover , its minimal representation decomposes as @xmath299 note that there is a quaternionic analogue of this construction , where instead of the fano plane we consider only one of its lines .
this means that we glue three copies of @xmath39 along the product of four copies of @xmath60 .
the resulting algebra is @xmath41 .
we conclude that the lie algebra @xmath4 is the support of a finite plane projective geometry whose points represent @xmath255 copies of @xmath39 , and whose lines represent @xmath255 copies of @xmath41 . we have just seen that the octonionic multiplication table gives a solution to the problem of finding a set of constants @xmath282 satisfying the skewsymmetry condition and the @xmath161 quadratic relations associated to the @xmath1 triangles in the fano plane .
what are the other solutions such that @xmath287 ? there exist exactly sixteen such solutions , falling into two @xmath301-orbits .
each orbit can be identified , as a @xmath302-set , with a copy of the projective line @xmath303 .
we proceed as follows .
we first check that on a line the orientations must be coherent in the following sense : put an arrow from @xmath129 to @xmath185 if @xmath304 .
then the three arrows on a line , if we draw it as a circle , must go in the same direction .
in particular there are only two possible choices of signs on a line , one for each cyclic orientation .
we can switch from one to the other by changing one of the basis vectors in its opposite .
moreover , the possible solutions to our problem can now be interpreted as a coherent orientation of the seven lines in the plane .
now we choose a triangle in the fano plane .
we have eight possible orientations for the three sides .
we observe that once we choose one , the orientation of the line joining the three middle points of the sides of the triangle is fixed , and that there are only two possibilities for the three remaining lines , those passing through the center of the triangle .
moreover we pass from one to the other by changing the sign of the basis vector corresponding to the center .
finally we check that once we fix a coherent orientation , we can transform it by @xmath305 to an arbitrarily chosen orientation on the triangle of reference .
this implies that we have @xmath57 possible orientations splitting in at most two orbits .
but there can not be a single orbit since @xmath57 does not divide the order of @xmath305 . to identify the two orbits with a projective line over @xmath306 , there just remains to observe that @xmath305 , up to conjugation , has a unique subgroup of index @xmath307 ( see @xcite ) .
this suggests an interpretation of the isomorphism between @xmath301 and @xmath302 .
we obtained the following one which we could not find in the litterature .
it is conveniently expressed in terms of cubes in a projective line , by which we simply mean a graph with eight vertices , which is topologically the incidence graph of a cube . _
definition_. a cube in a projective line @xmath308 is _ harmonic _ if each of its faces @xmath309 is harmonic , that is , the opposite vertices @xmath310 and @xmath311 are in harmonic position
. there exist fourteen harmonic cubes in @xmath303 , made of @xmath312 harmonic faces .
they split uniquely into two @xmath302-orbits in such a way that each harmonic face belongs to exactly one cube of each family .
( 20,42)(-1,-1 ) ( 0,0)(0,3)2(1,0)3 ( 0,0)(3,0)2(0,1)3 ( 2,1)(0,3)2(1,0)3 ( 2,1)(3,0)2(0,1)3 ( 0,0)(0,3)2(2,1)2 ( 3,0)(0,3)2(2,1)2 ( -.8,-.7)5(3.2,-.7)3(5.3,.5)2(1.3,1.2)6 ( -.8,3.1)0(3.3,2.5)@xmath313(5.2,4.2)1(1.3,4.2)4 ( -4,2)@xmath314 ( 16,0)(0,3)2(1,0)3 ( 16,0)(3,0)2(0,1)3 ( 18,1)(0,3)2(1,0)3 ( 18,1)(3,0)2(0,1)3 ( 16,0)(0,3)2(2,1)2 ( 19,0)(0,3)2(2,1)2 ( 15.2,-.7)6(19.2,-.7)5(21.3,.5)3(17.3,1.2)2 ( 15.2,3.1)0(19.3,2.5)@xmath313(21.2,4.2)1(17.3,4.2)4 ( 10,2)@xmath315 ( 0,6)(0,3)2(1,0)3 ( 0,6)(3,0)2(0,1)3 ( 2,7)(0,3)2(1,0)3 ( 2,7)(3,0)2(0,1)3 ( 0,6)(0,3)2(2,1)2 ( 3,6)(0,3)2(2,1)2 ( -.8,5.3)2(3.2,5.3)5(5.3,6.5)@xmath313(1.3,7.2)6 ( -.8,9.1)4(3.3,8.3)3(5.2,10.2)1(1.3,10.2)0 ( -4,8)@xmath316 ( 16,6)(0,3)2(1,0)3 ( 16,6)(3,0)2(0,1)3 ( 18,7)(0,3)2(1,0)3 ( 18,7)(3,0)2(0,1)3 ( 16,6)(0,3)2(2,1)2 ( 19,6)(0,3)2(2,1)2 ( 15.2,5.3)5(19.2,5.3)6(21.3,6.5)1(17.3,7.2)2 ( 15.2,9.1)0(19.3,8.3)4(21.2,10.2)@xmath313(17.3,10.2)3 ( 10,8)@xmath317 ( 0,12)(0,3)2(1,0)3 ( 0,12)(3,0)2(0,1)3 ( 2,13)(0,3)2(1,0)3 ( 2,13)(3,0)2(0,1)3 ( 0,12)(0,3)2(2,1)2 ( 3,12)(0,3)2(2,1)2 ( -.8,11.3)1(3.2,11.3)4(5.3,12.5)@xmath313(1.3,13.2)5 ( -.8,15.1)3(3.3,14.3)2(5.2,16.2)0(1.3,16.2)6 ( -4,14)@xmath318 ( 16,12)(0,3)2(1,0)3 ( 16,12)(3,0)2(0,1)3 ( 18,13)(0,3)2(1,0)3 ( 18,13)(3,0)2(0,1)3 ( 16,12)(0,3)2(2,1)2 ( 19,12)(0,3)2(2,1)2 ( 15.2,11.3)6(19.2,11.3)2(21.3,12.5)@xmath313(17.3,13.2)3 ( 15.2,15.1)4(19.3,14.3)1(21.2,16.2)0(17.3,16.2)5 ( 10,14)@xmath319 ( 0,18)(0,3)2(1,0)3 ( 0,18)(3,0)2(0,1)3 ( 2,19)(0,3)2(1,0)3 ( 2,19)(3,0)2(0,1)3 ( 0,18)(0,3)2(2,1)2 ( 3,18)(0,3)2(2,1)2 ( -.8,17.3)5(3.2,17.3)4(5.3,18.5)@xmath313(1.3,19.2)6 ( -.8,21.1)2(3.3,20.3)0(5.2,22.2)3(1.3,22.2)1 ( -4,20)@xmath320 ( 16,18)(0,3)2(1,0)3 ( 16,18)(3,0)2(0,1)3 ( 18,19)(0,3)2(1,0)3 ( 18,19)(3,0)2(0,1)3 ( 16,18)(0,3)2(2,1)2 ( 19,18)(0,3)2(2,1)2 ( 15.2,17.3)4(19.2,17.3)5(21.3,18.5)@xmath313(17.3,19.2)3 ( 15.2,21.1)0(19.3,20.3)2(21.2,22.2)6(17.3,22.2)1 ( 10,20)@xmath321 ( 0,24)(0,3)2(1,0)3 ( 0,24)(3,0)2(0,1)3 ( 2,25)(0,3)2(1,0)3 ( 2,25)(3,0)2(0,1)3 ( 0,24)(0,3)2(2,1)2 ( 3,24)(0,3)2(2,1)2 ( -.8,23.3)4(3.2,23.3)2(5.3,24.5)@xmath313(1.3,25.2)6 ( -.8,27.1)5(3.3,26.3)1(5.2,28.2)0(1.3,28.2)3 ( -4,26)@xmath322 ( 16,24)(0,3)2(1,0)3 ( 16,24)(3,0)2(0,1)3 ( 18,25)(0,3)2(1,0)3 ( 18,25)(3,0)2(0,1)3 ( 16,24)(0,3)2(2,1)2 ( 19,24)(0,3)2(2,1)2 ( 15.2,23.3)6(19.2,23.3)1(21.3,24.5)4(17.3,25.2)3 ( 15.2,27.1)0(19.3,26.3)5(21.2,28.2)@xmath313(17.3,28.2)2 ( 10,26)@xmath323 ( 0,30)(0,3)2(1,0)3 ( 0,30)(3,0)2(0,1)3 ( 2,31)(0,3)2(1,0)3 ( 2,31)(3,0)2(0,1)3 ( 0,30)(0,3)2(2,1)2 ( 3,30)(0,3)2(2,1)2 ( -.8,29.3)4(3.2,29.3)5(5.3,30.5)@xmath313(1.3,31.2)3 ( -.8,33.1)1(3.3,32.3)0(5.2,34.2)2(1.3,34.2)6 ( -4,32)@xmath324 ( 16,30)(0,3)2(1,0)3 ( 16,30)(3,0)2(0,1)3 ( 18,31)(0,3)2(1,0)3 ( 18,31)(3,0)2(0,1)3 ( 16,30)(0,3)2(2,1)2 ( 19,30)(0,3)2(2,1)2 ( 15.2,29.3)4(19.2,29.3)2(21.3,31.5)@xmath313(17.3,31.2)6 ( 15.2,33.1)3(19.3,32.3)5(21.2,34.2)1(17.3,34.2)0 ( 10,32)@xmath325 ( 0,36)(0,3)2(1,0)3 ( 0,36)(3,0)2(0,1)3 ( 2,37)(0,3)2(1,0)3 ( 2,37)(3,0)2(0,1)3 ( 0,36)(0,3)2(2,1)2 ( 3,36)(0,3)2(2,1)2 ( -.8,35.3)3(3.2,35.3)4(5.3,36.5)@xmath313(1.3,37.2)2 ( -.8,39.1)0(3.3,38.3)6(5.2,40.2)1(1.3,40.2)5 ( -4,38)@xmath326 ( 16,36)(0,3)2(1,0)3 ( 16,36)(3,0)2(0,1)3 ( 18,37)(0,3)2(1,0)3 ( 18,37)(3,0)2(0,1)3 ( 16,36)(0,3)2(2,1)2 ( 19,36)(0,3)2(2,1)2 ( 15.2,35.3)5(19.2,35.3)4(21.3,36.5)@xmath313(17.3,37.2)6 ( 15.2,39.1)1(19.3,38.3)2(21.2,40.2)0(17.3,40.2)3 ( 10,38)@xmath327 _ figure 4 .
the fourteen harmonic cubes in @xmath303 _
we thus get an interpretation of these two sets of harmonic cubes in @xmath303 as points and lines in a fano plane .
we arbitrarily distinguish these two sets by calling them @xmath328-cubes and @xmath34-cubes , respectively .
( note an interesting analogy with the construction of the projective space over @xmath14 given in @xcite from the fano plane : there exists @xmath184 unequivalent labelings of the seven vertices up to the @xmath301 action , and they are split into two families of @xmath329 labelings by the property that each line appears only once in each family . )
the incidence relations can be defined as follows : * given a @xmath328-cube ( respectively @xmath34-cube ) , there exist exactly three @xmath34-cubes ( respectively @xmath328-cubes ) sharing a pair of opposite faces with it . *
given two @xmath328-cubes ( respectively @xmath34-cubes ) , there exists a unique @xmath328-cube ( respectively @xmath34-cube ) such that the three cubes can be split each into two pairs of opposite edges forming squares with the same fourtuples of vertices .
note that each pair @xmath203 is the diagonal of exactly one @xmath328-cube and one @xmath34-cube .
a nice feature of the correspondence is that each pair of points in @xmath303 defines an edge of exactly three cubes in each family , corresponding to the three vertices and to the three edges of a triangle in the fano plane .
therefore : there is an equivariant correspondence between the @xmath1 triangles in the fano plane , and the @xmath1 pairs of points in the projective line over @xmath306 .
explicitely , this correspondence is as follows , where the triples in boldface are the vertices of a triangle in the fano plane : @xmath330 what about the orientations of the fano planes that we were interested in ?
we can associate such an orientation , in a @xmath302-equivariant way , to each point @xmath331 as follows .
for each point @xmath332 in the fano plane , consider the point @xmath333 such that @xmath334 is a diagonal of the harmonic cube corresponding to @xmath332 .
( this defines a bijection between @xmath335 and @xmath336 . )
then the line @xmath337 will we positively oriented if the cross - ratio @xmath338 on can easily check that in @xmath306 , this condition is invariant under a cyclic permutation of @xmath339 , so that this definition really makes sense !
this makes explicit the identification that we obtained between the projective line over @xmath306 and half of the coherent orientations of the fano plane .
of course we obtain the other half by reversing all the arrows .
our correspondence has the property to transform certain special configurations of bitangents into nice sets of triangles .
we mention three instances of that . .
when we index the bitangents by pairs of points in a set with eight elements , we give a special role to the eight aronhold sets @xmath340 defined as the seven bitangents @xmath203 , @xmath341 .
indeed , this follows from the discussion of example 4 above .
we thus get eight sets of seven triangles @xmath342 in the fano plane ( corresponding to the eight points in @xmath303 ) with the following properties : given a line @xmath34 and two points on it , there is a unique triangle in @xmath342 they are vertices of which ; in particular , @xmath342 is a copy of the steiner triple system @xmath264 ; from the three pairs of points in @xmath34 we thus deduce three triangles in @xmath342 ; the three vertices of these triangles which do not belong to @xmath34 are the vertices of a fourth triangle in @xmath342 ; this defines a natural bijection between the lines of the fano plane and the triangles in @xmath342 .
@xmath343 this gives a remarkable configuration of eight steiner triple systems formed on @xmath1 triangles in such a way that each of them appears exactly twice .
moreover , each point and each line in the fano plane belongs to exactly three of the seven triangles in each system . and the centers of the seven triangles are the seven points of the plane . .
we have seen in example 5 that syzygetic tetrads of bitangents are defined by fourtuples of weights in the fundamental representation of @xmath4 summing to zero , but not in two opposite pairs .
there are two types of such tetrads in hesse s notations : @xmath344 are permutations of @xmath345 and @xmath346 are permutations of @xmath347 .
each of these tetrads defines a special configuration of four triangles , a typical one being ( 10,8)(-3,-1 ) ( 0,0)(1,0)6 ( 0,0)(1,2)3 ( 6,0)(-1,2)3 ( 1.5,3)(1,0)3 ( 1.5,3)(1,-2)1.5 ( 1.5,3)(3,-2)1.5 ( 3,0)(0,1)2 ( 4.5,3)(-1,-2)1.5 ( 4.5,3)(-3,-2)1.5 the four triangles in this picture are the great exterior triangle , plus the three triangles having the center of the picture for vertex , plus two others on the middle points of two of the sides of the first one . .
a triple of points on the projective line over @xmath306 defines three pairs , hence three triangles in the fano plane .
the following statement leads us back to the orientation problem which was the starting point of this long digression .
there is an induced equivariant correspondence between triples of points on the projective line over @xmath306 , and oriented triangles in the fano plane .
this goes as follows .
consider an oriented triangle in @xmath335 .
for each vertex , consider the middle point on the opposite side , and then go to the next vertex following the orientation .
we thus get three ( non oriented ) triangles , which can be checked to be in correspondence with three pairs @xmath348 of a unique triple @xmath349 of points on @xmath303 .
given a pair @xmath350 of points in @xmath303 , it defines a triangle in the fano plane , hence two oriented triangles , hence two triples of points in the projective line .
how can we obtain them directly ? simply by considering the unique harmonic cube in one of our two families having @xmath350 for diagonal .
then the three points on an edge of this cube passing to @xmath328 ( respectively @xmath351 ) give the two triples .
the natural inclusion @xmath352 suggests to encode the @xmath1 bitangents and their symmetry group , directly in the geometry of the fano plane .
this can indeed be done in a very natural way , once we have identified the bitangents with the @xmath1 triangles in the fano plane . recall that the group of the bitangents is generated by the transpositions @xmath353 and the bifid tranformations @xmath354 .
we have checked that they have a simple geometric interpretation in terms of triangles . _
transpositions_. let @xmath355 be the triangle associated to the pair @xmath203 .
there is a unique point in the fano plane which does not belong to a side of @xmath355 , and we call this point the center of the triangle . up to the action of @xmath356
we can draw the fano plane in such a way that the exterior of the picture is precisely @xmath355 .
then the involution @xmath357 on the set of triangles exchanges triangles as shown in the following picture : ( 20,8)(-1,-1 ) ( 0,0)(1,0)6 ( 0,0)(1,2)3 ( 6,0)(-1,2)3 ( 12,0)(1,0)6 ( 12,0)(1,2)3 ( 18,0)(-1,2)3 ( 8.6,3.6)@xmath358 ( 0,0)(1,0)3 ( 3,0)(0,1)2 ( 0,0)(3,2)3 ( 15,0)(0,1)6 ( 13.5,3)(1,2)1.5 ( 13.5,3)(1,-2)1.5 ( 7.5,3)(1,0)3 ( 10.5,3)(-1,0)3 otherwise said , a triangle whose vertices are a vertex @xmath359 of @xmath355 , the center @xmath360 of @xmath355 and the middle point @xmath328 of a side of @xmath355 , is mapped to the triangle whose vertices are @xmath328 , its symmetric point @xmath361 with respect to @xmath360 , and the middle point of the side @xmath362 and conversely , while the other triangles remain unchanged . _
bifid transformations_. one can check that the @xmath363 bifid transformations , when we interprete them as operations on the triangles , split into three types which are naturally associated to the seven points @xmath328 , the seven lines @xmath34 , and the @xmath364 pairs of incident points and lines @xmath365 in the fano plane .
we get the following transformations .
associated to a point @xmath328 is a transformation @xmath366 who takes a triangle with a side whose middle point is @xmath328 , and changes the opposite vertex to the symmetric point with respect to @xmath328 .
( 20,8)(-1,-1 ) ( 0,0)(1,0)6 ( 0,0)(1,2)3 ( 6,0)(-1,2)3 ( 12,0)(1,0)6 ( 12,0)(1,2)3 ( 18,0)(-1,2)3 ( 8.6,3.6)@xmath366 ( -.8,-.5)@xmath328 ( 11.2,-.5)@xmath328 ( 3,0)(1,0)3 ( 3,0)(1,2)1.5 ( 6,0)(-1,2)1.5 ( 15,0)(1,0)3 ( 15,0)(0,1)2 ( 15,2)(3,-2)3 ( 7.5,3)(1,0)3 ( 10.5,3)(-1,0)3 associated to a line @xmath34 is a transformation @xmath367 who takes a triangle with a unique vertex @xmath359 on @xmath34 , and changes the two other vertices to the symmetric points with respect to @xmath359 .
( 20,8)(-1,-1 ) ( 0,0)(1,0)6 ( 0,0)(1,2)3 ( 6,0)(-1,2)3 ( 12,0)(1,0)6 ( 12,0)(1,2)3 ( 18,0)(-1,2)3 ( 8.6,3.6)@xmath367 ( 2.8,-.9)@xmath34 ( 14.8,-.9)@xmath34 ( 1.5,3)(3,-2)4.5 ( 1.5,3)(1,2)1.5 ( 3,6)(1,-2)3 ( 15,2)(3,-2)3 ( 15,2)(3,2)1.5 ( 16.5,3)(1,-2)1.5 ( 7.5,3)(1,0)3 ( 10.5,3)(-1,0)3 note that these two types of transformations are exchanged by the polarity transformation on the set of triangles , which exchanges vertices and sides in the fano plane with sides and vertices in the dual fano plane . finally , associated to a pair @xmath365 is a transformation @xmath368 who takes a triangle with a unique vertex @xmath369 on @xmath34 , whose opposite side @xmath370 has @xmath328 for middle point , to the triangle with vertex the symmetric point of @xmath359 with respect to @xmath328 , and opposite side the symmetric of @xmath370 with respect to @xmath34 .
( 20,8)(-1,-1 ) ( 0,0)(1,0)6 ( 0,0)(1,2)3 ( 6,0)(-1,2)3 ( 12,0)(1,0)6 ( 12,0)(1,2)3 ( 18,0)(-1,2)3 ( 8.4,3.6)@xmath368 ( 2.8,-.9)@xmath34 ( 14.8,-.9)@xmath34 ( -.8,-.5)@xmath328 ( 11.2,-.5)@xmath328 ( 3,0)(0,1)6 ( 1.5,3)(1,2)1.5 ( 1.5,3)(1,-2)1.5 ( 15,2)(3,-2)3 ( 15,2)(3,2)1.5 ( 16.5,3)(1,-2)1.5 ( 7.5,3)(1,0)3 ( 10.5,3)(-1,0)3 the group of the bitangents is isomorphic with the group of permutations of the triangles generated by the elementary tranformations @xmath371 , @xmath372 , @xmath373 , @xmath374 .
it contains @xmath356 as the group of collinations acting on the triangles .
this makes clear the natural inclusions @xmath375
the bicanonical model of a del pezzo surface of degree one is the double cover of a quadratic cone , branched over a canonical space curve of genus @xmath9 and degree @xmath125 given by the complete intersection of the cone with a unique cubic surface ( @xcite v.1 ) .
the @xmath145 lines on the del pezzo surface arise in pairs from the @xmath8 tritangent planes to the canonical curve , which can be identified with its odd theta characteristics .
moreover , the fact that the unique quadric containing this curve is a cone distinguishes one of the @xmath376 even theta characteristic by the property that it vanishes at the vertex of that cone .
the automorphism group of the @xmath145 lines is the weyl group @xmath49 of the root system of type @xmath7 .
its order is @xmath377 .
the automorphism group of the @xmath8 tritangent planes is the quotient by the normal subgroup @xmath378 and can be identified with the orthogonal group @xmath50 which preserves the quadratic form given by half the natural norm on the root lattice mod 2 ( @xcite , exercice 1 of section 4 , p. 228 ) . among the @xmath379 points in @xmath380 , those with norm one can be identified with the odd theta - characteristics , and those with norm zero with the even theta - characteristics , including the special one which identifies with the origin .
if we consider the action of the adjoint group @xmath7 on the projectivized adjoint representation @xmath381 , the @xmath145 root spaces can be interpreted as a kind of finite skeleton of the closed orbit , the adjoint variety of @xmath7 .
this variety parametrizes the so - called symplecta in freudenthal s metasymplectic geometry ( see @xcite ) .
the branching gives the binary model @xmath383 where @xmath36 is again the minimal @xmath161-dimensional representation of @xmath4 .
in particular , this associates to each root @xmath129 ( up to sign ) of @xmath5 a complex @xmath195 of @xmath161 tritangent planes . two complexes @xmath195 and
@xmath196 have two possible relative positions , distinguished by the fact that @xmath132 or not . in the latter case , exactly as for @xmath4 the two complexes are _ azygetic _ and can be completed uniquely with a third complex @xmath189 , with @xmath384 , azygetic to both of them .
there exists @xmath385 such _ azygetic triads of complexes_. this leads to the ternary model of @xmath5 , @xmath386 the @xmath8 tritangent planes are partitioned into the triple @xmath285 , three sets @xmath387 , @xmath388 , @xmath389 of cardinality @xmath0 , and their complement of cardinality @xmath117 , with @xmath390 in the former case , the two complexes are _
syzygetic_. since their common orthogonal subsystem is a root system of type @xmath391 , we can define a _
syzygetic tetrad of complexes _ to consist in four syzygetic complexes orthogonal to a @xmath56-subsystem of the root system @xmath7 .
note that we have three syzygetic tetrads for each @xmath56-subsystem , making a total of @xmath392 such tetrads .
a pair of syzygetic complexes can be completed uniquely into a syzygetic tetrad @xmath393 .
the other @xmath394 tritangent planes are then partitioned into a set @xmath395 of cardinality @xmath307 , six sets @xmath387 , @xmath388 , @xmath396 , @xmath389 , @xmath397 , @xmath398 of cardinality @xmath57 , and their complement of cardinality @xmath12 . here
the branching gives one of the prettiest models of @xmath5 , namely @xmath406 where @xmath102 denotes the natural nine dimensional representation .
the @xmath407 factor defines what du val calls an @xmath408 configuration in the uniform polytope @xmath162 .
( note that there are @xmath409 such configurations rather than @xmath405 .
this is because of the invariance by @xmath410 , the longest element in @xmath49 : indeed its restriction to the root system @xmath411 of @xmath407 does not belong to @xmath412 but defines an order two _ outer _ automorphism ) .
it can be interpreted as a special system of then even theta - characteristics ( @xcite , p. 51 ) .
one deduces a special set of @xmath413 , and the complementary set of @xmath117 tritangent planes : the odd theta - characteristics of the former set are obtained as sums of three of the distinguished even theta - characteristics ( which is visible from the fact that they correspond to weights of a third wedge power ) , and the latter as sums of only two of them .
there is also a strong analogy with the decomposition of @xmath4 that we described in example 2 above , from which we recovered hesse s notation for the bitangents to the plane quartic , and the so called bifid transformations . here
the roots of @xmath5 , up to sign , are split in two types by the model we are discussing .
the roots that are weights of the factor @xmath415 are indexed by a triple @xmath27 of integers between @xmath416 and @xmath417 .
the roots or @xmath407 are indexed , up to sign , by a pair @xmath203 that we can identify with the triple @xmath418 .
the tritangent planes are then indexed by the @xmath8 triples of integers between @xmath416 and @xmath93 , a notation first introduced by pascal ( see @xcite )
. moreover , the group of the tritangent planes is generated by the symmetric group @xmath419 plus the symmetries associated to the triples @xmath27 , which are again called _ bifid transformations _
( but beware that it does not contain the symmetric group @xmath420 ) .
an easy computation shows that the bifid transformation associated to @xmath27 exchanges @xmath421 with @xmath422 and sends a triple @xmath349 to @xmath423 if @xmath424 is a permutation of @xmath425 while the other triples are fixed .
there is a connection with the model for @xmath11 discussed in example 4 of section 3 , as we can see by splitting the nine dimensional representation @xmath102 into three supplementary spaces of dimension three .
in fact the corresponding subroot system of type @xmath426 in @xmath7 is orthogonal to another @xmath137 subsystem , and we get another interesting model for @xmath5 , which is graded over @xmath427 : @xmath428 where @xmath429 are three dimensional and we used the notation @xmath430 . but
this grading does not seem to be supported by any interesting geometry . indeed , we get @xmath434 where @xmath435 are the three eight - dimensional irreducible representations of @xmath59 .
this models splits the @xmath8 tritangent planes into two groups of @xmath12 and three groups of @xmath177 . with which geometric characterization ? then we can split each @xmath432 using four copies of @xmath60 ; we know that each of its eight - dimensional representation has a nice decomposition , and putting them together we obtain @xmath436 where @xmath272 is the following set of @xmath437 quadruples : @xmath438 as in the case of @xmath4 these fourteen quadruples form a _ steiner quadruple system _ @xmath439 .
in fact there exists a unique such system up to isomorphism .
_ such a decomposition is induced by the choice of a root subsystem of type @xmath440 inside the root system @xmath7 .
this is equivalent to what du val calls a @xmath441 configuration in the polytope @xmath162 .
there is also a correspondence with the @xmath142 even theta - characteristics , or the @xmath142 norm one vectors in the root lattice mod 2 , which can be seen as follows .
consider such a vector , for example @xmath442 if we denote by @xmath443 the weights of @xmath340 . among the @xmath444 projective lines through @xmath445 , @xmath173
are contained in the quadric , @xmath161 are tangents and only @xmath307 are true bisecants ; on each of these bisecants there is a unique point outside the quadric and we obtain a set of eight points defining a subsystem of type @xmath440 .
let us consider our steiner quadruple system is some detail .
we first note that each quadruple @xmath446 defines a copy of @xmath447 inside @xmath5 .
we have thus constructed @xmath5 by gluing together fourteen copies of @xmath39 overlapping over eight copies of @xmath60 !
we can interprete the @xmath437 quadruples in @xmath272 as _ points _ of a configuration whose _ lines _ are triples of type @xmath448 .
a straightforward inspection shows that there are exactly @xmath1 lines .
moreover , each line has three points and each point belongs to @xmath125 lines .
in other words , we have obtained a @xmath449-configuration .
this configuration has the following interpretation .
consider @xmath450 , the three dimensional projective space over the field with two elements .
it has fifteen points .
choose one , say @xmath451 , and throw it away . since @xmath450 contains @xmath363 lines , @xmath255 of which passing through @xmath451 , we remain with @xmath437 points and @xmath1 lines whose incidence configuration is the one we are interested in . 1 .
each point @xmath328 has an antipodal point @xmath452 , the unique point to which it is not joined by a line in @xmath450 , this is the third point of the line @xmath453 ; 2 . the @xmath125 lines passing through a point @xmath328 split naturally into three pairs , in such a way that the four points different from @xmath328 on each pair , belong to a pair of lines passing through @xmath452 ; in @xmath450 , these three pairs are cut by the three planes containing the line @xmath453 ; 3 .
the configuration is made of eight copies of the fano plane , corresponding to the eight planes in @xmath450 not containing @xmath451 ; the other seven planes give sub - configurations of type @xmath454 which are pointed fano planes .
but the most satisfactory way to understand our configuration is probably to see it as a _ doubled fano plane_. by this we mean that we can associate to each point of the fano plane a pair of antipodal quadruples , in such a way that the @xmath1 lines of the configuration correspond four by four to the lines of the fano plane . in terms of lie subalgebras of @xmath5 ,
this associates to each point of the fano plane a copy of @xmath455 , and to each line a copy of @xmath42 .
moreover , there exists four copies of @xmath41 inside the @xmath42 defined by a line , meeting each @xmath40 corresponding to one of its points , along one of the two @xmath39 factors .
we have even more structure if we note that each integer @xmath33 between @xmath416 and @xmath307 determines a copy of @xmath4 inside @xmath5 .
two such copies meet along one of the @xmath41 indexed by the @xmath1 lines .
this has an interesting combinatorial interpretation .
consider the following @xmath456 array : @xmath457 it is symmetric with respect to the main diagonal , and each number @xmath458 between @xmath459 and @xmath255 appears once and only once in each line and in each column .
if we associate to @xmath458 the four pairs @xmath460 of numbers indexing the lines and columns of the boxes where @xmath458 appears , this means that we obtain a partition of @xmath208 $ ] into four pairs , for a total number of @xmath1 distinct pairs : @xmath461 ( note that this array can also be obtained from the diagonals of each family of seven harmonic cubes of figure 5 . )
these seven partitions index our seven lines in the following way : to each @xmath462 are associated the six quadruples obtained by selecting two of the four pairs .
the incidence is given by the following rule : two lines being given , they can always be indexed by partitions of type @xmath462 @xmath463 , and then their intersection point is indexed by the pair of antipodal quadruples @xmath274 and @xmath235 .
_ first application_. this construction of @xmath5 isolates eight tritangent planes corresponding to the roots of the eight copies of @xmath60 , the remaining @xmath464 tritangent planes being split into fourteen groups of eight .
syzygetic tetrad _ of tritangent planes by the property that their twelve tangency points are the twelve intersection points of the sextic canonical curve with some quadric hypersurface .
this means that they can be defined by four roots summing to zero .
we claim that the eight tritangent planes associated to each factor @xmath275 in our decomposition of @xmath5 , can be split in two syzygetic tetrads in exactly six ways .
indeed , the corresponding roots are the @xmath465 and two syzygetic tetrads are given by the following two sign tables : _ second application_. an _ orthogonal decomposition _ ( od ) of a semisimple lie algebra @xmath28 is a decomposition @xmath467 into a direct sum of cartan subalgebras .
such an od is _ multiplicative _ if for each pair @xmath156 , there exists an integer @xmath468 such that @xmath469\subset\ft_k$ ] .
a trivial example is that of @xmath470 , once we have chosen a basis @xmath471 of @xmath120 . if @xmath472 is the associated canonical basis of @xmath473 , a multiplicative od ( or mod ) is given be the three lines generated by @xmath474 and @xmath475 .
multiplicative od s have been used by thompson to construct the finite sporadic simple group denoted @xmath43 or @xmath44 . indeed , his construction relied on the existence of a multiplicative od for @xmath5 .
in fact there exists , up to isomorphism , a unique multiplicative od of @xmath5 ( @xcite , chapter 3 ) .
our construction provides it for free !
indeed we just need to notice that each component @xmath476 can be split into the direct sum of two cartan subalgebras @xmath477 by choosing a basis @xmath478 of each @xmath340 , and letting @xmath479 where @xmath480 is @xmath478 or @xmath481 .
this gives @xmath1 cartan subalgebras of @xmath5 , and we add three others , say @xmath482 , @xmath483 and @xmath484 , by putting together the od s of the @xmath485 s associated to the basis of the @xmath340 s that we have chosen .
we obtain : | the automorphism groups of the @xmath0 lines on the smooth cubic surface or the @xmath1 bitangents to the general quartic plane curve are well - known to be closely related to the weyl groups of @xmath2 and @xmath3 .
we show how classical subconfigurations of lines , such as double - sixes , triple systems or steiner sets , are easily constructed from certain models of the exceptional lie algebras . for @xmath4 and @xmath5
we are lead to beautiful models graded over the octonions , which display these algebras as plane projective geometries of subalgebras .
we also interprete the group of the bitangents as a group of transformations of the triangles in the fano plane , and show how this is related to an interpretation of the isomorphism @xmath6 in terms of harmonic cubes . |
this paper is concerned with solutions of the epdiff equations ; that is , with the euler - poincar equations associated with the diffeomorphism group in @xmath1-dimensions . in particular
, we are concerned with singular solutions that generalize the peakon solutions of the camassa - holm ( ch ) equation in one dimension .
the ch equation ( see @xcite ) for the dynamics of shallow water in a certain asymptotic regime , is @xmath4 where @xmath5 is the fluid velocity , subscripts denote partial derivatives in position @xmath6 and time @xmath7 , and @xmath8 is a positive constant .
( see @xcite for recent discussions of the derivation and asymptotic validity of the ch equation for shallow water waves , at one order beyond the korteweg - de vries equation . )
equivalently , in hamiltonian form , the ch equation reads @xmath9 where @xmath10 and @xmath11 is a positive constant . as @xcite show , in hamiltonian form
the ch equation is the lie - poisson equation associated with the lie algebra of one dimensional vector fields and with the hamiltonian @xmath12 the ch equation may be equivalently expressed in euler - poincar form by using the lagrangian associated with the @xmath13 metric for the fluid velocity .
that is , the lagrangian as a function of the fluid velocity is given by the quadratic form , @xmath14 it follows from euler - poincar theory ( see @xcite and @xcite ) that the one parameter curve of diffeomorphisms @xmath15 depending on parameter @xmath7 and defined implicitly by @xmath16 is a geodesic in the group of diffeomorphisms of @xmath17 ( or , with periodic boundary conditions , of the circle @xmath18 ) equipped with the right invariant metric equal to the @xmath13 metric at the identity .
a remarkable analytical property of the ch equation , conjectured by keeping track of derivative losses in @xcite and proved in @xcite is that the geodesic equations literally define a smooth vector field in the sobolev @xmath19 topology .
that is , in the material representation , the equations have no derivative loss .
this property is analogous to the corresponding results for the euler equations for ideal incompressible fluid flow ( discovered by @xcite ) and the lagrangian averaged euler equations ( again conjectured by @xcite and proved by @xcite ) . as we will explain in
[ smoothness_section ] , a similar statement holds for the @xmath1-dimensional epdiff equation if we use the @xmath13 metric .
this is all the more remarkable because smoothness of the geodesic flow is _ not true _ for the @xmath20 metric , at least not without assuming incompressibility .
smoothness of _ volume - preserving _ geodesic flow with respect to the @xmath20 metric does hold for the incompressible flow of an ideal euler fluid , a result proved in @xcite . before proceeding with a discussion of the general case of the @xmath1-dimensional epdiff equations
, we shall quickly review , mostly to establish notation , a few facts about the euler - poincar and lie - poisson equations , whose basic theory is explained , for example , in @xcite . [ [ review - of - euler - poincar - and - lie - poisson - equations . ] ] review of euler - poincar and lie - poisson equations .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let @xmath21 be a lie group and @xmath22 its associated lie algebra ( identified with the tangent space to @xmath21 at the identity element ) , with lie bracket denoted by @xmath23 $ ] for @xmath24 .
let @xmath25 be a given lagrangian .
let @xmath26 be the right invariant lagrangian on @xmath21 obtained by translating @xmath27 from the identity element to other points of @xmath28 via the right action of @xmath21 on @xmath29 .
a basic result of euler - poincar theory is that the euler
lagrange equations for @xmath30 on @xmath21 are equivalent to the ( right ) euler - poincar equations for @xmath27 on @xmath22 , namely to @xmath31 here , @xmath32 is the adjoint operator , the linear map given by the lie bracket @xmath33 $ ] . also , @xmath34 is its dual ; that is , @xmath35 \right\rangle$ ] , where @xmath36 is the natural pairing between @xmath37 and @xmath22 .
also , @xmath38 denotes the functional derivative of @xmath27 with respect to @xmath39 . for _
left _ invariant systems , we change the sign in ( [ epequations ] ) .
the euler - poincar equations can be written in the variational form @xmath40 for all variations of the form @xmath41 $ ] for some curve @xmath42 in @xmath22 that vanishes at the endpoints .
if the reduced legendre transformation @xmath43 is invertible , then the euler - poincar equations are equivalent to the ( right ) lie - poisson equations : @xmath44 where the reduced hamiltonian is given by , @xmath45 these equations are equivalent ( via lie - poisson reduction and reconstruction ) to hamilton s equations on @xmath46 relative to the hamiltonian @xmath47 , obtained by right translating @xmath48 from the identity element to other points via the right action of @xmath21 on @xmath49 .
the lie - poisson equations may be written in the poisson bracket form @xmath50 where @xmath51 is an arbitrary smooth function and the bracket is the ( right ) lie - poisson bracket given by @xmath52 \right\rangle .\ ] ] in the important case when @xmath27 is quadratic , the lagrangian @xmath53 is the quadratic form associated to a right invariant riemannian metric on @xmath21 . in this case , the euler
lagrange equations for @xmath30 on @xmath21 describe geodesic motion relative to this metric and these geodesics are then equivalently described by either the euler - poincar , or the lie - poisson equations . [ [ outline - of - the - paper . ] ] outline of the paper .
+ + + + + + + + + + + + + + + + + + + + + the main results of the present paper are as follows : 1 . in [ epdiff_section ]
we review some basic facts about the epdiff equations , especially the _ singular solution ansatz _ of @xcite that introduces a class of singular solutions that generalize the peakon solutions of the ch equation to higher spatial dimensions
. 2 . in [ smoothness_section ]
we give a plausibility argument that the epdiff equations possess an interesting _ smoothness property _ ; namely , they define a smooth vector field ( that is , they define ode s with no derivative loss ) in the lagrangian representation .
this means , in particular , that the epdiff equations are locally well posed for sufficiently smooth initial data .
because of the development of singularities in finite time , which the numerics suggests , the smooth solutions may not exist globally in time .
this smoothness property is similar to the corresponding smoothness property of the euler equations for ideal incompressible fluid mechanics shown in @xcite .
[ ssmommap_section ] , we show that the singular solution ansatz defines an equivariant momentum map .
we do this in a natural way by identifying the singular solutions with certain curves in the space of embeddings @xmath54 of a generally lower dimensional manifold @xmath55 into the underlying manifold @xmath0 ( or an @xmath1-manifold @xmath2 ) and letting the diffeomorphism group act on this space .
the right action of @xmath56 corresponds to the right invariance of the epdiff equations , while the left action of @xmath57 gives the desired solution ansatz .
4 . in [ geom_mommap ] we explore the geometry of the singular solution momentum map , in parallel with the corresponding work on singular solutions ( vortices , filaments , etc . ) for the euler equations of an ideal fluid in @xcite .
5 . finally , in [ future_section ]
, we discuss some of the remaining challenges and speculate on some of the many possible future directions for this work .
[ [ historical - note . ] ] historical note .
+ + + + + + + + + + + + + + + + this paper is dedicated to our friend and collaborator alan weinstein and , for us , this work parallels some of our earlier collaborations with him .
alan s basic works on reduction , poisson geometry , semidirect product theory , and stability in mechanics just to name a few areas have been , and remain incredibly influential and important to the field of geometric mechanics .
see , for instance , @xcite mechanics on lie groups was pioneered by @xcite , a reference that is a key foundation for the subject and in particular for the present paper . however , this theory was in a relatively primitive state , even by 1980 , and it has benefited greatly from alan s insights .
in fact , the clear distinction between the euler - poincar and lie - poisson equations , the former with a variational structure and the latter with its poisson structure , which took until the 1980 s to crystalize , was greatly aided by alan s work .
alan has made key contributions to many fundamental concepts in geometrical mechanics , such as lagrangian submanifolds and related structures ( @xcite ) , symplectic reduction ( @xcite ) , normal modes and periodic orbits ( @xcite ) , poisson manifolds ( @xcite ) , geometric phases ( @xcite ) , dirac structures ( @xcite ) groupoids and lagrangian reduction ( @xcite ) and the plethora of related `` oid '' structures he has been working on during the last decade ( just look over the over 200 papers on mathscinet he has written ! ) that will surely play an important role in the next generation of people working in the area of geometric mechanics .
of alan s papers , the one that is most directly relevant to the topics discussed in the present paper is @xcite .
alan himself is is still developing this area , too , as in @xcite .
in this section we review the epdiff equation ; that is , the euler - poincar ( ep ) equation associated with the diffeomorphism group in @xmath1-dimensions .
this equation coincides with a limiting case of the ch equation for shallow water waves in one and two dimensions .
it also coincides with the atme equation ( the averaged template matching equation ) in two dimensions .
the latter equation arises in computer vision ; see , for instance , @xcite or @xcite for a description and further references .
we have chosen to call this by a generic name , the epdiff equation , because it has these various interpretations in different applications .
of course these different interpretations also provide opportunities : for example , to see to what extent the singular solutions found in the epdiff equations are applicable , either for shallow water wave interactions , or for computer vision applications .
a recent combination of these ideas in which image processing imitates soliton interactions appears in @xcite .
[ [ statement - of - the - epdiff - equations . ] ] statement of the epdiff equations .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + treating analytical issues formally at this point , let @xmath58 denote the lie algebra of vector fields on an @xmath1-dimensional manifold @xmath2 ( such as @xmath0 ) .
the vector fields comprise the algebra associated with the diffeomorphism group of @xmath2 , but the usual jacobi - lie bracket is the negative of the ( standard ) lie algebra bracket .
( see @xcite for a discussion . )
let @xmath59 be a given lagrangian and let @xmath60 denote the space of one - form densities on @xmath2 , that is , the momentum densities .
the corresponding momentum density of the fluid is defined as @xmath61 which is the functional derivative of the lagrangian @xmath27 with respect to the fluid velocity @xmath62 .
if @xmath63 is the basic dynamical variable , the epdiff equations are simply the euler - poincar equations associated with this lagrangian .
equivalently , if @xmath64 is the basic dynamical variable , a legendre transformation allows one to identify the epdiff equations as the lie - poisson equations associated with the resulting hamiltonian . for the case of @xmath0
, we will use vector notation for the momentum density @xmath65 ( a bold @xmath66 instead of a lightface @xmath64 ) .
the epdiff equations are as follows ( see @xcite for additional background and for techniques for computing the euler - poincar equations for field theories ) , @xmath67 in coordinates @xmath68 , @xmath69 , using the summation convention , and writing @xmath70 ( regarding @xmath66 as a one - form density ) and @xmath71 ( regarding @xmath72 as a vector field ) , the epdiff equations read @xmath73 the epdiff equations can also be written very nicely as @xmath74 where @xmath75 denotes the lie derivative of the momentum one form density @xmath66 with respect to the velocity vector field @xmath72 . as mentioned earlier , if @xmath27 is a quadratic function of @xmath72 , then the epdiff equation ( ( [ ep - eqn - vec - intro ] ) or , equivalently ( [ epdifflie ] ) ) , is the eulerian description of geodesic motion on the diffeomorphism group of the underlying space ( in this case @xmath0 ) .
the corresponding metric is the right invariant metric on the group whose value on the lie algebra ( the group s tangent space at the identity the space of vector fields ) , is defined by @xmath27 . since the lagrangian @xmath27 is positive and quadratic in @xmath72 , the momentum density is linear in @xmath72 and so defines a positive symmetric operator @xmath76 by @xmath77 [ [ variational - formulation . ] ] variational formulation .
+ + + + + + + + + + + + + + + + + + + + + + + + following the variational formulation of ep theory , the particular ep equation ( [ ep - eqn - vec - intro ] ) may be derived from the following constrained variational principle : @xmath78 the variations are constrained to have the form @xmath79 this assertion may of course be verified directly .
these constraints are analogous to the so - called `` lin constraints '' used for a similar variational principle for fluid mechanics .
( see @xcite for a discussion and references . )
[ [ hamiltonian - formulation . ] ] hamiltonian formulation .
+ + + + + + + + + + + + + + + + + + + + + + + + a legendre transformation yields the hamiltonian , @xmath80 where @xmath81 is the natural pairing between one form densities and vector fields given by integration .
this hamiltonian is the corresponding quadratic form for the momentum , @xmath82 of course it often happens that @xmath76 is a differential operator and in this case the inverse is usually given in terms of the convolution with the green s function @xmath21 , corresponding to the appropriate solution domain and boundary conditions ; @xmath83 according to the general theory , the ep equation ( [ ep - eqn - vec - intro ] ) may be expressed in hamiltonian form by using the lie - poisson bracket on @xmath60 as @xmath84 [ [ one - dimensional - ch - peakon - solutions . ] ] one - dimensional ch peakon solutions .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we return now to the ch equation ( [ chlp ] ) , which , as we have noted , is the same as the epdiff equation ( [ ep - eqn - vec - intro ] ) for the case of one spatial dimension , when the momentum velocity relationship is defined by the helmholtz equation , @xmath85 . in one dimension ,
the ch equation has solutions whose momentum is supported at points on the real line via the following sum over dirac delta measures , @xmath86 the velocity corresponding to this measure - valued momentum is obtained by convolution with the green s function , @xmath87 for the one - dimensional helmholtz operator , @xmath88 , appearing in the ch momentum velocity relationship , @xmath89 .
consequently , the ch velocity corresponding to this momentum is given by a superposition of peaked traveling wave pulses , @xmath90 thus , the superposition of `` peakons '' in velocity arises from the delta function solution ansatz ( [ m - peakon - ansatz - intro ] ) for the momentum .
remarkably , the isospectral eigenvalue problem for the ch equation implies that _ only _ these singular solutions emerge asymptotically in the solution of the initial value problem in one dimension , @xcite .
figure [ peakon_figure ] shows the emergence of peakons from an initially gaussian velocity distribution and their subsequent elastic collisions in a periodic one - dimensional domain .
was kindly supplied by martin staley ] this figure demonstrates that singular solutions dominate the initial value problem and , thus , that it is imperative to go beyond smooth solutions for the ch equation ; the situation is similar for the epdiff equation .
remarkably , the dynamical equations for @xmath91 and @xmath92 , @xmath93 , that arise from solution ansatz ( [ m - peakon - ansatz - intro]-[u - peakon - ansatz - intro ] ) comprise an integrable system for any @xmath94 .
this system is studied in ( @xcite ) and references therein . see also @xcite for discussions of how the integrable dynamical system for @xmath94 peakons is related to the toda chain with open ends . [ [ generalizing - the - ch - peakon - solutions - to - n - dimensions ] ] generalizing the ch peakon solutions to @xmath1 dimensions + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + building on the peakon solutions for the ch equation and the pulsons for its generalization to other traveling - wave shapes ( see @xcite ) , @xcite introduced the following measure - valued ( that is , density valued ) ansatz for the @xmath95dimensional solutions of the epdiff equation ( [ ep - eqn - vec - intro ] ) : @xmath96 these solutions are vector - valued functions supported in @xmath0 on a set of @xmath94 surfaces ( or curves ) of codimension @xmath97 for @xmath98 with @xmath99 .
they may , for example , be supported on sets of points ( vector peakons , @xmath100 ) , one - dimensional filaments ( strings , @xmath101 ) , or two - dimensional surfaces ( sheets , @xmath102 ) in three dimensions .
_ one of the main results of this paper is that the singular solution ansatz is a momentum map .
this result helps to organize the theory and to suggest new avenues of exploration , as we shall explain .
_ substitution of the solution ansatz ( [ m - ansatz - intro ] ) into the epdiff equations ( [ ep - eqn - vec - intro ] ) implies the following integro - partial - differential equations ( ipdes ) for the evolution of such strings and sheets , @xmath103 importantly for the interpretation of these solutions , the coordinates @xmath98 turn out to be lagrangian coordinates .
the velocity field corresponding to the momentum solution ansatz ( [ m - ansatz - intro ] ) is given by @xmath104 when evaluated along the curve @xmath105 , the velocity satisfies , @xmath106 thus , the lower - dimensional support sets defined on @xmath105 and parameterized by coordinates @xmath107 move with the fluid velocity .
moreover , equations ( [ intdiffeqn - q ] ) for the evolution of these support sets are canonical hamiltonian equations , @xmath108 the hamiltonian function @xmath109 is , @xmath110 this is the hamiltonian for geodesic motion on the cotangent bundle of a set of curves @xmath111 with respect to the metric given by @xmath21 .
this dynamics was investigated numerically in @xcite to which we refer for more details of the solution properties .
one of our main goals is to show that the solution ansatz ( [ m - ansatz - intro ] ) can be phrased in terms of a momentum map that naturally arises in this problem .
this geometric feature underlies the remarkable reduction properties of the epdiff equation and `` explains '' why the preceding equations must be hamiltonian , namely , because momentum maps are poisson maps . as explained in general terms in @xcite , the way one
implements a coadjoint orbit reduction is through a momentum map , and this holds even for the case of singular orbits ( again ignoring functional analytic details ) .
thus , in summary , _ the reduction is the epdiff analog of the reduction in fluid mechanics ( that is , the equations ) to point ( or blob ) vortex dynamics , vortex filaments , or sheets . _
there are , however , some important differences between vortex dynamics for incompressible flows and the dynamics of the measure valued epdiff solutions . for example , the lagrangian representations of the equations of motion show that epdiff solutions have inertia , while the corresponding solutions for point ( or blob ) vortices of the epdiff@xmath112 dynamics have no inertia .
that is , the equations of motion for measure valued solutions on epdiff@xmath113 are _ first order _ in time , while the dynamical equations for measure valued solutions on epdiff are _ second order _ in time .
this difference has profound effects on the properties of the solutions , especially on their stability properties .
numerical investigations of @xcite show , for example , that the codimension - one solutions of epdiff are stable , while higher codimension solutions of epdiff are very unstable to codimension - one perturbations .
in contrast , the codimension - two solutions of epdiff@xmath112 are known to be stable .
[ [ comments - on - the - physical - meaning - of - the - equations . ] ] comments on the physical meaning of the equations .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the epdiff equations with the helmholtz relation between velocity and momentum are not quite the ch equations for surface waves in 2d .
those would take precisely the same form , but the shallow water wave relation in the 2d ch approximation would be @xmath114 rather than the helmholtz operator form , @xmath115 the corresponding lagrangians are , respectively , @xmath116 and @xmath117 this difference was noted in @xcite , which identified ( [ div - norm ] ) as the generalization of ( [ ch - lag ] ) for water waves in two dimensions .
one may also verify this by considering the limit of the green - nagdhi equations for small potential energy .
( the ch equation arises in this limit .
the lake and great lake equations of @xcite also arise in a variant of this limit . )
remarkably , the numerics in @xcite show that the solutions for a variety of initial conditions are indistinguishable in these two cases .
the initial conditions in @xcite were all spatially confined velocity distributions .
notice that this difference affects the choice of hamiltonian , but the equations are still euler - poincar equations for the diffeomorphism group and the description of the anzatz ( [ m - ansatz - intro ] ) as a momentum map is independent of this difference in the equations .
figure [ strings_waves ] shows the striking reconnection phenomenon seen in the nonlinear interaction between wave - trains , as simulated by numerical solutions of the epdiff equation and observed for internal waves in the ocean .
fig [ strings_waves](a ) shows a frame taken from simulations of the initial value problem for the epdiff equation in two dimensions , excerpted from @xcite .
( see also @xcite . )
fig [ strings_waves](b ) shows the interaction of two internal wave trains propagating at the interface of different density levels in the south china sea , as observed from the space shuttle using synthetic aperture radar , courtesy of a. liu ( 2002 ) .
importantly , both fig [ strings_waves](a ) and fig [ strings_waves](b ) show nonlinear reconnection occurring in the wave train interaction as their characteristic feature .
internal waves and these sorts of interactions are generally thought to be described by the kp equation , and so any relations among the kp equation , the epdiff equation and the 2d ch equation would be of great interest to explore ; cf .
the derivations of the kp equation and the ch equation differ in the way the transverse motions are treated in the asymptotics ; so some difference in their solution behavior is to be expected .
[ [ the - one - dimensional - case . ] ] the one - dimensional case .
+ + + + + + + + + + + + + + + + + + + + + + + + + based on a formal argument given in @xcite , it was shown in @xcite that when the ch equation ( [ chlp ] ) is transformed to lagrangian variables , it defines a smooth vector field ( that is , one obtains an evolution equation with no derivative loss ) .
this means that one can show using ode methods that the initial value problem is well - posed and one may also establish other important properties of the equations when the data is sufficiently smooth .
as above , we write the relation between @xmath64 and @xmath63 as @xmath118 , so that in the one dimensional case , @xmath76 is the operator @xmath119 .
we first recall how the equations are transformed into lagrangian variables .
introduce the one parameter curve of diffeomorphisms @xmath15 defined implicitly by @xmath120 so that @xmath42 is a geodesic in the group of diffeomorphisms of @xmath17 ( or , with periodic boundary conditions , of the circle @xmath18 ) equipped with the right invariant metric equaling the @xmath13 metric at the identity .
we compute the second time derivative of @xmath42 in a straightforward way by differentiating ( [ defeta ] ) using the chain rule : @xmath121 acting on this equation with @xmath76 and using the definition @xmath118 yields @xmath122u - 2 m u_x \\ & = - 3 { \alpha}^2 u_x u_{xx } - 2 m u_x \\ & = - { \alpha}^2 u_x u_{xx } - 2 u u_x \,,\end{aligned}\ ] ] where the third step uses the commutator relation calculated from the product rule , @xmath123u = - 3 { \alpha}^2 u_x u_{xx } \,.\ ] ] hence , the preceding equation becomes @xmath124 the important point about this equation is that the right hand side has no derivative loss .
that is , if @xmath63 is in the sobolev space @xmath125 for @xmath126 , then the right hand is also in the same space .
regarding the right hand side as a function of @xmath42 and @xmath127 , we see that it is plausible that the second order evolution equation ( [ chlag ] ) for @xmath42 defines a smooth ode on the group of @xmath19 diffeomorphisms .
( this argument requires the use of , for example , weighted sobolev spaces in the case @xmath128 ) .
the above is the essence of the argument given in @xcite , remark 3.5 , which in turn makes use of the type of arguments found in @xcite for the incompressible case and which shows , by a more careful argument , that the spray is smooth if @xmath129 .
however , one should note that the complete argument is not quite so simple ( just as in the case of incompressible fluids ) .
a subtilty arises because smoothness means as a function of @xmath130 .
hence , one must express @xmath63 in terms of @xmath42 , namely through the relation @xmath131 , where the subscript @xmath7 here denotes that this argument is held fixed , and is not a partial derivative .
doing this , one sees that , while there is clearly no derivative loss , the right hand side of ( [ chlag ] ) does involve @xmath132 and the map @xmath133 is known to _ not _ be smooth ( just as in @xcite )
. nevertheless , the _ combination _ that appears in ( [ chlag ] ) is , quite remarkably , a smooth function of @xmath134 .
[ [ the - n - dimensional - case . ] ] the @xmath1-dimensional case .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the above argument readily generalizes to the @xmath1-dimensions , which we shall present in the case of @xmath0 or the flat @xmath1-torus @xmath135 for simplicity .
namely , we still have the relation @xmath136 between @xmath42 and @xmath72 .
consequently , we may compute the second partial time derivative of @xmath42 in the usual fashion using the chain rule : @xmath137 therefore , as in the one dimensional case , we get @xmath138 \mathbf{u } + \mathbf{u } \cdot \nabla \mathbf{m } + \frac{\partial \mathbf{m } } { \partial t } \,.\ ] ] calculating the commutator relation in @xmath1-dimensions gives @xmath139 \mathbf{u } = -\ , \alpha^2{\rm div}\big ( \nabla\mathbf{u}\cdot\nabla\mathbf{u } + \nabla\mathbf{u}\cdot\nabla\mathbf{u}^{\,t } \big ) + \alpha^2(\nabla\mathbf{u})\cdot\nabla{\rm div}\,\mathbf{u}\ ] ] or , in components , @xmath140 \mathbf{u } \big)\,_i = -\ , \alpha^2\partial_k\big ( u_{i , j}\,u_{j , k } + u_{i , j}\,u_{k , j } \big ) + \alpha^2(u_{i , j})\,\partial_j{\rm div}\,\mathbf{u } \ , , \label{div - tensor - tau}\ ] ] with a sum on repeated indices . upon substituting the preceding commutator relation , the epdiff equation ( [ ep - eqn - vec - intro ] ) and the vector calculus identity @xmath141 then imply the @xmath95dimensional result @xmath142 this form of the epdiff equation is useful for interpreting some of its solution behavior . as in the one dimensional case , the crucial point is that the right hand side involves at most second derivatives of @xmath72 ; so there is no derivative loss in the overall expression .
the precise formulation of the smoothness result thus should hold , as in the one - dimensional case , although , naturally the details are a bit more complicated . we shall leave the technical details for another publication .
a consequence is that the equations are well posed _ for short time _ if the initial data is smooth enough and , hence , e.g. , two nearby smooth solutions can be joined by a unique geodesic . moreover , the solutions of the epdiff equation are automatically @xmath143 in time . in the sections that follow
, we will be interested in nonsmooth data , which is in stark contrast to the preceding discussion , which requires initial data that is at least @xmath144 .
remarkably , the same smoothness results hold for the case of the lae-@xmath145 ( lagrangian averaged euler ) equations , a set of incompressible equations in which small scale fluctuations are averaged .
one can view the lae-@xmath145 equations as the incompressible version of the epdiff equations .
this smoothness property for the lae-@xmath145 equations was shown by @xcite for regions with no boundary and for regions with boundary ( for various boundary conditions ) , it was shown in @xcite . however , unlike the incompressible case , the results apparently do not hold if @xmath145 is zero ( as also noted in @xcite ) .
this sort of smoothness result also appears not to hold for many other equations , such as the kdv equation , even though it too can be realized as euler - poincar equations on a lie algebra , or as geodesics on a group , in this case the bott - virasoro group , as explained in @xcite and references therein .
[ [ the - development - of - singularities . ] ] the development of singularities .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the smoothness property just discussed does not preclude the development in finite time of singular solutions from smooth localized initial data as was indicated in figure [ peakon_figure ] . to capture the local singularities in the epdiff solution ( either verticality in slope , or discontinuities in its spatial derivative ) that develop in finite time from arbitrarily smooth initial conditions
, one must enlarge the solution class of interest , by considering weak solutions .
there are a number of papers on weak solutions of the ch equation such as @xcite that we will not survey here .
we just mention that the theory is not yet complete , as it is still unknown in what sense one may define _ global unique _ weak solutions to the ch equations in @xmath3 . as discussed in @xcite for the ch equation , one most likely must consider weak solutions in the _ spacetime sense_. the steepening lemma of @xcite proves that in one dimension any initial velocity distribution whose spatial profile has an inflection point with negative slope ( for example , any antisymmetric smooth initial distribution of velocity on the real line ) will develop a vertical slope in finite time
. note that the peakon solution ( [ u - peakon - ansatz - intro ] ) has no inflection points , so it is not subject to the steepening lemma .
however , the steepening lemma underlies the mechanism for forming these singular solutions , which are continuous but have discontinuous spatial derivatives ; they also lie in @xmath13 and have finite energy .
we conclude that solutions with initial conditions in @xmath146 with @xmath147 go to infinity in the @xmath125 norm in finite time , but remain in @xmath3 and presumably continue to exist in a weak spacetime sense for all time in @xmath3 .
numerical evidence in higher dimensions and the inverse scattering solution for the ch equation in one dimension ( the latter has _ only _ discrete eigenvalues , corresponding to peakons ) both show that the singular solutions completely dominate the time - asymptotic dynamics of the initial value problem ( ivp ) .
this singular ivp behavior is one of the main discoveries of @xcite .
this singular behavior has drawn a great deal of mathematical interest to the ch equation and its relatives , such as epdiff .
the other properties of ch its complete integrability , inverse scattering transform , connections to algebraic geometry and elliptical billiards , bi - hamiltonian structure , etc .
are all interesting , too . however , the requirement of dealing with singularity as its main solution phenomenon is the primary aspect of ch ( and epdiff ) .
we aim to show that many of the properties of these singular solutions of ch and epdiff are captured by recognizing that the singular solution ansatz itself is a momentum map .
this momentum map property explains , for example , why the singular solutions ( [ m - ansatz - intro ] ) form an invariant manifold for any value of @xmath94 . in one dimension ,
the complete integrability of the ch equation as a hamiltonian system and its soliton paradigm completely explain the emergence of peakons in the ch dynamics .
namely , their emergence reveals the initial condition s soliton ( peakon ) content .
however , beyond one dimension , we do not have an explicit mechanism for explaining why _ only _ singular solution behavior emerges in numerical simulations .
one hopes that eventually a theory will be developed for explaining this singular solution phenomenon in higher dimensions .
such a theory might , for example , parallel the well - known explanation of the formation of shocks for hyperbolic partial differential equations .
( note , however , that epdiff is not hyperbolic , because the relation @xmath148 between its velocity and momentum is nonlocal . ) in the remainder of this work , we shall focus our attention on the momentum map properties of the invariant manifold of singular solutions ( [ m - ansatz - intro ] ) of the epdiff equation .
[ [ the - momentum - ansatz - is - a - momentum - map . ] ] the momentum ansatz is a momentum map .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the purpose of this section is to show that the solution ansatz for the momentum vector in the epdiff equation ( [ ep - eqn - vec - intro ] ) defines a momentum map for the action of the group of diffeomorphisms on the support sets of the dirac delta functions .
these support sets are the analogs of points on the real line for the ch equation in one dimension .
they are points , curves , or surfaces in @xmath0 for the epdiff equation in @xmath95dimensions .
this result , as we shall discuss in greater detail later , shows that the solution ansatz fits naturally into the scheme of clebsch , or canonical variables in the sense advocated by @xcite as well as showing that these singular solutions evolve on special coadjoint orbits for the diffeomorphism group .
one can summarize by saying that the map that implements the canonical @xmath149 variables in terms of singular solutions is a ( cotangent bundle ) momentum map .
such momentum maps are poisson maps ; so the canonical hamiltonian nature of the dynamical equations for @xmath150 fits into a general theory which also provides a framework for suggesting other avenues of investigation .
[ mom - map ] the momentum ansatz for measure - valued solutions of the equation , defines an equivariant momentum map @xmath151 that we will call the singular solution momentum map .
we shall explain the notation used in this statement in the course of the proof .
right away , however , we note that the sense of `` defines '' is that expressing @xmath66 in terms of @xmath152 ( which are , in turn , functions of @xmath153 ) can be regarded as a map from the space of @xmath154 to the space of @xmath66 s .
this will turn out to be the lagrange - to - euler map for the fluid description of the singular solutions .
we shall give two proofs of this result from two rather different points of view . the first proof below uses the formula for a momentum map for a cotangent lifted action , while the second proof focuses on a poisson bracket computation .
each proof also explains the context in which one has a momentum map .
( see @xcite for general background on momentum maps . ) * first proof .
* for simplicity and without loss of generality , let us take @xmath155 and so suppress the index @xmath156 .
that is , we shall take the case of an isolated singular solution .
as the proof will show , this is not a real restriction . to set the notation , fix a @xmath157-dimensional manifold @xmath55 with a given volume element and whose points
are denoted @xmath158 .
let @xmath159 denote the set of smooth embeddings @xmath160 .
( if the epdiff equations are taken on a manifold @xmath2 , replace @xmath161 with @xmath2 . ) under appropriate technical conditions , which we shall just treat formally here , @xmath159 is a smooth manifold .
( see , for example , @xcite and @xcite for a discussion and references . )
the tangent space @xmath162 to @xmath159 at the point @xmath163 is given by the space of material velocity fields , namely the linear space of maps @xmath164 that are vector fields over the map @xmath165 . the dual space to this space will be identified with the space of one - form densities over @xmath165 , which we shall regard as maps @xmath166 . in summary ,
the cotangent bundle @xmath167 is identified with the space of pairs of maps @xmath168 .
these give us the domain space for the singular solution momentum map .
now we consider the action of the symmetry group .
consider the group @xmath169 of diffeomorphisms of the space @xmath170 in which the epdiff equations are operating , concretely in our case @xmath0 .
let it act on @xmath170 by composition on the _
left_. namely for @xmath171 , we let @xmath172 now lift this action to the cotangent bundle @xmath167 in the standard way ( see , for instance , @xcite for this construction ) .
this lifted action is a symplectic ( and hence poisson ) action and has an equivariant momentum map .
_ we claim that this momentum map is precisely given by the ansatz . _
to see this , we just need to recall and then apply the general formula for the momentum map associated with an action of a general lie group @xmath173 on a configuration manifold @xmath174 and cotangent lifted to @xmath175 .
first let us recall the general formula .
namely , the momentum map is the map @xmath176 ( @xmath177 denotes the dual of the lie algebra @xmath22 of @xmath173 ) defined by @xmath178 where @xmath179 and @xmath180 , where @xmath181 is the infinitesimal generator of the action of @xmath173 on @xmath174 associated to the lie algebra element @xmath182 , and where @xmath183 is the natural pairing of an element of @xmath184 with an element of @xmath185 .
now we apply this formula to the special case in which the group @xmath173 is the diffeomorphism group @xmath186 , the manifold @xmath174 is @xmath159 and where the action of the group on @xmath159 is given by ( [ action ] ) .
the sense in which the lie algebra of @xmath169 is the space @xmath187 of vector fields is well - understood .
hence , its dual is naturally regarded as the space of one - form densities .
the momentum map is thus a map @xmath188 . with @xmath189 given by ( [ momentummap ] ) , we just need to work out this formula .
first , we shall work out the infinitesimal generators .
let @xmath190 be a lie algebra element . by differentiating the action ( [ action ] ) with respect to @xmath42 in the direction of @xmath191 at the identity element
we find that the infinitesimal generator is given by @xmath192 thus , taking @xmath193 to be the cotangent vector @xmath150 , equation ( [ momentummap ] ) gives @xmath194 on the other hand , note that the right hand side of ( [ m - ansatz - intro ] ) ( again with the index @xmath156 suppressed , and with @xmath7 suppressed as well ) , when paired with the lie algebra element @xmath191 is @xmath195 this shows that the expression given by ( [ m - ansatz - intro ] ) is equal to @xmath189 and so the result is proved .
* second proof . * as is standard ( see , for example , @xcite ) , one can characterize momentum maps by means of the following relation , required to hold for all functions @xmath197 on @xmath198 ; that is , functions of @xmath165 and @xmath199 : @xmath200 \,.\ ] ] in our case , we shall take @xmath189 to be given by the solution ansatz and verify that it satisfies this relation . to do so , let @xmath201 so that the left side of ( [ mom_map_poisson ] ) becomes @xmath202d\ , ^k s \,.\ ] ] on the other hand , one can directly compute from the definitions that the infinitesimal generator of the action on the space @xmath203 corresponding to the vector field @xmath204 ( a lie algebra element ) , is given by ( see @xcite , formula ( 12.1.14 ) ) : @xmath205 which verifies that ( [ mom_map_poisson ] ) holds .
an important element left out in this proof so far is that it does not make clear that the momentum map is _ equivariant _ , a condition needed for the momentum map to be poisson .
the first proof took care of this automatically since momentum maps for cotangent lifted actions are always equivariant and hence poisson .
thus , to complete the second proof , we need to check directly that the momentum map is equivariant .
actually , we shall only check that it is infinitesimally invariant by showing that it is a poisson map from @xmath206 to the space of @xmath66 s ( the dual of the lie algebra of @xmath58 ) with its lie - poisson bracket .
this sort of approach to characterize equivariant momentum maps is discussed in an interesting way in @xcite .
the following computation accomplishes this methodology by showing that the singular solution momentum map is poisson .
indeed , we use the canonical poisson brackets for @xmath207 and apply the chain rule to compute @xmath208 , with notation @xmath209 . we get @xmath210 \ , . \nonumber\end{aligned}\ ] ] substituting the canonical poisson bracket relations @xmath211 into the preceding computation yields , @xmath212 thus , @xmath213 which is readily checked to be the lie - poisson bracket on the space of @xmath64 s .
this completes the second proof of theorem .
@xmath196 each of these proofs has shown the following basic fact .
[ poisson_mom - map ] the singular solution momentum map defined by the singular solution ansatz , namely , @xmath214 is a poisson map from the canonical poisson structure on @xmath215 to the lie - poisson structure on @xmath216 .
this is perhaps the most basic property of the singular solution momentum map .
some of its more sophisticated properties are outlined in the following section .
[ [ pulling - back - the - equations .
] ] pulling back the equations .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + since the solution ansatz ( [ m - ansatz - intro ] ) has been shown in the preceding corollary to be a poisson map , the pull back of the hamiltonian from @xmath217 to @xmath218 gives equations of motion on the latter space that project to the equations on @xmath219 .
this is why the functions @xmath111 and @xmath220 satisfy canonical hamiltonian equations .
note that the coordinate @xmath107 that labels these functions is a lagrangian coordinate . in terms of the pairing @xmath221 between the lie algebra @xmath22 ( vector fields in @xmath0 ) and its dual @xmath222 ( one - form densities in @xmath0 ) , the following relation holds for measure - valued solutions under the momentum map ( [ m - ansatz - intro ] ) , @xmath223 which is the natural pairing between the points @xmath224 and @xmath225 .
the pull - back of the hamiltonian @xmath226 $ ] defined on the dual of the lie algebra @xmath222 , to @xmath227 is easily seen to be consistent with what we had before : @xmath228
\equiv \frac{1}{2}\langle\mathbf{m}\,,\,g*\mathbf{m}\rangle = \frac{1}{2}\langle\!\langle\mathbf{p}\,,\,g*\mathbf{p}\rangle\!\rangle \equiv h_n[\mathbf{p},\mathbf{q } ] \ , .
\label{geodesic - ham}\ ] ] in summary , in concert with the poisson nature of the singular solution momentum map , we see that the singular solutions in terms of @xmath165 and @xmath199 satisfy hamiltonian equations and also define an invariant solution set for the epdiff equations .
in fact , this invariant solution set is a special coadjoint orbit for the diffeomorphism group , as we shall discuss in the next section .
[ [ smoothness . ] ] smoothness .
+ + + + + + + + + + + it would be extremely interesting if the smoothness properties explored in [ smoothness_section ] were also valid on the space @xmath229 .
we hope to explore this point in future publications .
in this section we explore the geometry of the singular solution momentum map discussed in [ ssmommap_section ] in a little more detail .
the main idea may be stated as follows : _ simply apply all of the ideas given in in a systematic way to the current setting . _ [ [ coadjoint - orbits . ] ] coadjoint orbits .
+ + + + + + + + + + + + + + + + + the first claim is that _ the image of the singular solution momentum map is a coadjoint orbit in @xmath230 .
_ this means that ( perhaps modulo some issues of connectedness and smoothness , which we do not consider here ) the solution anzatz given by ( [ m - ansatz - intro ] ) defines a coadjoint orbit in the space of all one - form densities , regarded as the dual of the lie algebra of the diffeomorphism group .
these coadjoint orbits should be thought of as singular orbits that is , due to their special nature , they are not generic .
also , this explains why the singular solutions form dynamically invariant sets it is because they are coadjoint orbits , which are _ symplectic submanifolds _ of the lie - poisson manifold @xmath216 .
the idea of the proof of this is simply this : whenever one has an equivariant momentum map @xmath231 for the action of a group @xmath21 on a symplectic or poisson manifold @xmath232 , and that action is transitive , then the image of @xmath189 is an orbit ( or at least a piece of an orbit ) .
this general result , due to kostant , is stated more precisely in @xcite , theorem 14.4.5 .
roughly speaking , the reason that transitivity holds in our case is because one can `` move the manifolds @xmath55 around at will '' using diffeomorphisms .
[ [ symplectic - structure - on - orbits . ] ] symplectic structure on orbits .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + recall ( from , for example , @xcite ) , the general formula for the symplectic structure on coadjoint orbits : @xmath233 \right\rangle,\ ] ] where @xmath234 is a chosen point on an orbit and where @xmath235 are elements of @xmath22 .
we use a plus sign in this formula since we are dealing with orbits for the _
right action_. just as in @xcite , this leads to an explicit formula for the coadjoint orbit symplectic structure in the case of @xmath236 . in the present case
, it is a particularly simple and transparent formula . recall that in the case of incompressible fluid mechanics , this procedure leads naturally to the symplectic ( and poisson ) structure for many interesting singular coadjoint orbits , such as point vortices in the plane , vortex patches , vortex blobs ( closely related to the planar lae-@xmath145 equations ) and vortex filaments . for the case of the diffeomorphism group ,
let @xmath237 denote the coadjoint orbit through the point @xmath238 .
the symplectic structure @xmath239 on @xmath240 is given by @xmath241 \right\rangle \ , d^n x.\ ] ] we simply plug into the general kirillov - kostant - souriau formula ( [ coadjoint_sympl ] ) for the symplectic structure on coadjoint orbits .
( as noted above , there is a @xmath242 sign , since we are dealing with a _
right _ invariant system ) . the only thing needing explanation is that our lie algebra convention always uses the left lie bracket . for @xmath236 ,
this is the _ negative _ of the usual lie bracket as is explained in @xcite .
[ [ dual - pairs . ] ] dual pairs .
+ + + + + + + + + + + the singular solution momentum map @xmath243 forms one leg of a dual pair .
the point is that there is another group that acts on @xmath244 , namely the group @xmath56 of diffeomorphisms of @xmath55 , which acts on the _ right _ , while @xmath57 acted on it by composition on the _ left _
( and which gave rise to our singular solution momentum map )
. this momentum map will be denoted @xmath245 .
the dual pair arises for reasons that are very similar to those in @xcite .
see the following diagram .
( 150,100)(-70,0)(100,75)@xmath246 ( 78,50)@xmath247 ( 160,50)@xmath248 ( 72,15)@xmath249 ( 170,15)@xmath250 ( 130,70)(-1 , -1)40 ( 135,70)(1,-1)40 we also note that this framework shows that the parameterization of the singular solutions in terms of @xmath165 and @xmath199 are exactly clebsch variables in the sense given in @xcite . also notice that when we write the singular solutions in @xmath165-@xmath199 space , we are finding solutions that are _ collective _ and so all the properties of collectivization are valid .
see @xcite for a general discussion and references to the original work of guillemin and sternberg on this topic .
[ [ numerical - issues - geometric - integrators . ] ] numerical issues : geometric integrators . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the computations of martin staley that were illustrated in this paper and that are discussed in @xcite , make use of both mimetic differencing and reversibility in a critical way and this is important for good numerical resolution . in other words , integrators that respect the basic geometry underlying
the problem seem to play an key role
. it would be interesting to pursue this aspect further and also incorporate discrete exterior calculus and variational multisymplectic integration methods ( see @xcite as well as @xcite and @xcite ) .
the emergence in finite time of singular solutions from smooth initial data observed numerically in @xcite indicates that the diffeomorphism group with respect to the right invariant @xmath3 metric is _ geodesically incomplete _ when the diffeomorphism group has the @xmath251 topology , @xmath252 .
the degree of its geodesic incompleteness is not known , but we suspect that almost all epdiff geodesics in @xmath3 can not be extended indefinitely .
this certainly holds in one spatial dimension , where the discreteness of the ch isospectrum implies that asymptotically in time the ch solution arising from any confined initial velocity data consists _ only _ of peakons .
_ it is an important challange to find a context in which one can put the @xmath13 topology on the diffeomorphism group and reestablish geodesic
completeness_. the numerics suggests that this might be possible , while known existence theorems , even for the ch equation are not yet capable of showing this to the best of our knowledge .
epdiff is a reversible equation , and the collisions of its peakon solutions on the line @xmath253 ( or the circle @xmath254 ) are known to be reversible . in principle , the reconnections of the singular epdiff solutions observed numerically in @xcite in periodic domains @xmath255 and @xmath256 must also be reversible .
reversibility of its reconnections distinguishes the singular solutions of epdiff from vortex fluid solutions and shocks in fluids , whose reconnections apparently require dissipation and so , are not reversible . the mimetic finite differencing scheme used for the numerical computation of epdiff solutions in @xcite was indeed found to be reversible for overtaking collisions , but it was found to be only approximately reversible for head - on collisions , which are much more challenging for numerical integration schemes .
the singular epdiff solutions correspond to outlines ( or cartoons ) of images in applications of geodesic flow for the template , or pattern matching approach .
the dynamics of the singular epdiff solutions described by the momentum map introduces the paradigm of soliton collisions into the mechanics and analysis of image processing by template matching .
( see @xcite for more discussions of this new paradigm for image processing . )
the reversibility of the collisions among singular solutions and their reconnections under epdiff flow assures the preservation of the information contained in the image outlines .
in addition , the invariance of the manifold of @xmath94 singular solutions under epdiff assures that the fidelity of the image is preserved in the sense of approximation theory .
that is , an @xmath94 soliton approximation of the image outlines remains so , throughout the epdiff flow .
a natural approach for numerically simulating epdiff flows in image processing is to use multisymplectic algorithms .
the preservation of the space - time multisymplectic form by these algorithms introduces an initial - value , final - value formulation of the numerical solution procedure that is natural for template matching . in @xcite ,
the question of the ( rigorous ) poisson nature of the time @xmath7 map of the flow of the euler equations for an ideal fluid in appropriate sobolev spaces is explored .
given the smoothness properties in
[ smoothness_section ] , it seems reasonable that similar properties should also hold for the epdiff equations .
however , as mentioned earlier , these smoothness properties do not preclude the emergence of singular solutions from smooth initial data in finite time , because of the possibility for geodesic incompleteness .
the general setting of this paper suggests that perhaps one should look for similar measure valued or singular solutions associated with other problems , including geodesic flows on the group of symplectic diffeomorphisms ( relevant for plasma physics , as in @xcite ) , bott - virasoro central extensions and super - symmetry groups
. it might be interesting to explore the relation of the singular solution momentum map to integrability and scattering data .
for example , see @xcite for an interesting discussion of the poisson bracket for the scattering data of ch in 1d .
this turns out to be the atiyah - hitchin bracket , which is also related to the toda lattice , and this fascinating observation leads to an infinite - dimensional version of jacobi elliptic coordinates .
of course there are many other issues remaining to explore that are suggested by the above setting , such as convexity of the momentum map , its extension to riemannian manifolds , etc .
we shall , however , leave these issues for other publications and other researchers .
we are very grateful to alan weinstein for his collaboration , help and inspiring discussions over the years .
we also thank martin staley for letting us illustrate some important points using his computations .
ddh is grateful for support by us doe , under contract w-7405-eng-36 for los alamos national laboratory , and office of science ascar / ams / mics .
the research of jem was partially supported by the california institute of technology , the national science foundation through the nsf grant dms-0204474 and by the air force contract f49620 - 02 - 1 - 0176 .
alber , m. , r. camassa , y. fedorov , d. holm and j. e. marsden [ 2001 ] , the complex geometry of weak piecewise smooth solutions of integrable nonlinear pde s of shallow water and dym type , _ comm .
phys . _ * 221 * , 197227 .
courant , t. and a. weinstein [ 1988 ] , beyond poisson structures . in _ action hamiltoniennes de groupes .
troisime thorme de lie ( lyon , 1986 ) _ , volume 27 of _ travaux en cours _ ,
pages 3949 .
hermann , paris .
dullin , h. r. , g. a. gottwald and d. d. holm [ 2001 ] , an integrable shallow water equation with linear and nonlinear dispersion .
_ , * 87 * , 194501 - 04 , http://xxx.lanl.gov/abs/nlin.cd/0104004 .
holm , d. d. , j. e. marsden and t. s. ratiu [ 1998a ] , the euler
poincar equations and semidirect products with applications to continuum theories , _ adv . in math . _ * 137 * , 181 , http://xxx.lanl.gov/abs/chao-dyn/9801015 .
holm , d. d. , j. e. marsden and t. s. ratiu [ 2002 ] , the euler
poincar equations in geophysical fluid dynamics . in norbury ,
j. and i. roulstone , editors , _ large - scale atmosphere - ocean dynamics ii : geometric methods and models _ , pages 251300 .
cambridge univ . press .
holm , d. d. , a. trouv and l. younes [ 2003 ] , solitons in computational anatomy . to appear in _
neuroimage_. kodama , y. [ 1985 ] , on integrable systems with higher order corrections _ phys .
lett . a _ * 107 * , 245249 .
kruse , h.p . ,
j. scheurle and w. du [ 2001 ] , a two - dimensional version of the ch equation . in _
symmetry and perturbation theory : spt 2001 _ edited by d. bambusi , g. gaeta and m. cadoni .
world scientific : new york , 120127 . | we study the dynamics of measure - valued solutions of what we call the epdiff equations , standing for the _ euler - poincar equations associated with the diffeomorphism group ( of @xmath0 or an @xmath1-dimensional manifold @xmath2)_. our main focus will be on the case of quadratic lagrangians ; that is , on geodesic motion on the diffeomorphism group with respect to the right invariant sobolev @xmath3 metric .
the corresponding euler - poincar ( ep ) equations are the epdiff equations , which coincide with the averaged template matching equations ( atme ) from computer vision and agree with the camassa - holm ( ch ) equations in one dimension .
the corresponding equations for the volume preserving diffeomorphism group are the well - known lae ( lagrangian averaged euler ) equations for incompressible fluids . we first show that the epdiff equations are generated by a smooth vector field on the diffeomorphism group for sufficiently smooth solutions .
this is analogous to known results for incompressible fluids both the euler equations and the lae equations and it shows that for sufficiently smooth solutions , the equations are well - posed for short time .
in fact , numerical evidence suggests that , as time progresses , these smooth solutions break up into singular solutions which , at least in one dimension , exhibit soliton behavior . with regard to these non - smooth solutions , we study measure - valued solutions that generalize to higher dimensions the peakon solutions of the ( ch ) equation in one dimension .
one of the main purposes of this paper is to show that many of the properties of these measure - valued solutions may be understood through the fact that their solution ansatz is a momentum map .
some additional geometry is also pointed out , for example , that this momentum map is one leg of a natural dual pair . |
the driven viscous motion of an elastic interface in a medium with random pinning forces is relevant for the understanding of various problems in condensed matter physics @xcite .
examples include the ordering dynamics of an impure ising magnet after a quench below the critical temperature @xcite , wetting immiscible displacement of one fluid by another in a porous medium @xcite , and pinning of flux lines in type - ii - superconductors @xcite . in recent years , studies of fluid imbibition in paper @xcite and of flameless paper burning @xcite have been carried out , and observable interfaces allow a direct comparison with theoretical predictions .
common to all of these problems is a competition between smoothening due to the surface tension and roughening due to the interaction with the random pinning forces of the medium .
further , there is a competition between the driving force and the pinning forces , resulting in a depinning transition . on a coarse - grained level
, it is expected that the dynamics of the interface can be described by the following continuum equation of motion , @xmath0 here , @xmath1 is the normal velocity of the interface at position @xmath2 .
the surface tension generates a term proportional to the total curvature @xmath3 , where @xmath4 is the normal vector on the interface at @xmath2 .
the coefficient @xmath5 measures the stiffness of the interface .
@xmath6 is a homogeneous driving force .
the last two terms , @xmath7 and @xmath8 represent random - field and random - bond disorder , respectively . the random forces @xmath9 and the random potential @xmath10 are short - range correlated in space .
equation ( [ v_n ] ) is considerably simplified by restricting ourselves to an almost planar interface without overhangs .
a coordinate system @xmath11 can be introduced , so that the interface position is given by a single - valued function @xmath12 .
the dimension of @xmath13 is denoted by @xmath14 .
equation ( [ v_n ] ) becomes @xcite @xmath15 @xmath16,\ ] ] where @xmath17 . for sufficiently large values of the driving force @xmath6
, the interface grows continuously .
however , for smaller values of @xmath6 , growth on some regions of the interface can come to a halt , due to the interaction with the quenched disorder .
we say that these regions of the interface have become _ pinned_. as the rest of the interface continues to grow , the pinned regions can be dragged over the pinning barriers by neighboring moving regions .
then , the formerly pinned regions advance quickly , which can be considered as an avalanche @xcite .
the maximum linear size , @xmath18 , of the pinned regions diverges when @xmath6 approaches its critical value @xmath19 , @xmath20 the threshold @xmath19 is the critical point of a dynamical phase transition , and @xmath21 the corresponding correlation length .
the role of the order parameter is played by the mean velocity , @xmath22 .
( @xmath23 is the system size and the overbar denotes the spatial average over @xmath13 . )
the velocity is zero for @xmath24 , and increases as @xmath25 for @xmath26 . on length scales
@xmath27 pinning can be neglected and we can therefore replace the argument @xmath28 in the disorder terms @xmath29 and @xmath30 by @xmath31 , i.e. , the quenched disorder crosses over to thermal noise @xcite . then , the interface is governed by the kardar - parisi - zhang ( kpz ) equation @xcite . in this paper , we are interested in the critical behavior on length scales @xmath32 , especially when @xmath33 at the depinning transition .
the global interface width @xmath34 is another characteristic quantity of the interface . here and elsewhere
, @xmath35 denotes an average over the disorder distribution . choosing a flat interface as the initial condition , @xmath36
, @xmath37 scales as @xcite @xmath38 for a sufficiently large system size , @xmath39 . if @xmath40 , the correlation length @xmath18 in eq .
( [ w^2sca ] ) has to be replaced by @xmath23 .
@xmath41 and @xmath42 are scaling functions for @xmath43 and @xmath44 , respectively .
both functions scale as @xmath45 for @xmath46 , where @xmath47 .
it follows @xmath48 for @xmath43 , pinning is irrelevant on length scales @xmath27 , so we can neglect pinning also on time scales @xmath49 .
thus , @xmath50 for @xmath51 , where @xmath52 is the growth exponent of an interface subject to thermal noise . below threshold ,
the interface becomes pinned and @xmath53 using the scaling of @xmath54 for @xmath46 and eq .
( [ w^2sca ] ) we obtain @xmath55 it has been shown that the critical exponents fulfill an exact scaling relation @xcite , @xmath56
to further simplify the equation of motion ( [ ght ] ) , we assume that the typical gradients @xmath57 are small on large length scales , so that the roughness exponent @xmath58 is smaller than one .
this assumption has to be compared with the final results .
when expanding @xmath59 , nonlinear terms proportional to @xmath60 are generated in eq .
( [ ght ] ) .
a natural question is whether these terms are relevant at the depinning transition .
a term @xmath61 with a positive coefficient on the right hand side of eq .
( [ ght ] ) would give a nonzero contribution to the driving force for any rough interface .
this contribution increases when imposing a global tilt of the interface .
thus , the threshold @xmath19 becomes a function of the average orientation of the interface .
this is reasonable for anisotropic systems but not for isotropic ones . for interfaces in an isotropic environment , it can indeed be shown that the nonlinear terms generated by expanding @xmath62 in eq .
( [ ght ] ) are irrelevant close to the depinning transition @xcite .
if however , the medium is anisotropic , a term of the form @xmath63 can be relevant even for the case @xmath64 .
in fact , tang , kardar , and dhar @xcite argued that if @xmath65 and @xmath66 are differently distributed , the term @xmath63 is generated under coarse graining . more generally , if the system is anisotropic in the sense that the threshold @xmath19 depends on the average orientation of the interface , @xmath63 is the only relevant term that can change the universality class of the depinning transition @xcite . motivated by these observations , we consider the following equation of motion , @xmath67 for @xmath68 the threshold @xmath19 has a maximum for an interface without tilt , i.e. , an interface with periodic boundary conditions in eq .
( [ dht ] ) is driven along the hard direction of the anisotropic medium @xcite .
equation ( [ dht ] ) is the model we study in this paper . for simplicity ,
we restrict ourselves to interface dimension @xmath69 and consider only random - field disorder @xmath70 .
it was shown by narayan and fisher @xcite for isotropic systems that random fields and random bonds give rise to the same critical behavior .
this has been supported by numerical simulations of interfaces subject to random - bond disorder @xcite .
the random forces are assumed to have zero mean and short - range correlations , @xmath71 , where @xmath72 decreases exponentially for @xmath73 greater than a microscopic cutoff .
the case @xmath74 in eq .
( [ dht ] ) was first investigated by feigelman @xcite .
significant progress has been made by a functional renormalization - group treatment @xcite and by extensive numerical simulations @xcite .
the results for the anisotropic case @xmath68 are less well established .
it was first suggested in ref .
@xcite that eq .
( [ dht ] ) is in the same universality class as the `` directed percolation depinning '' ( dpd ) model @xcite . for @xmath69 , directed percolating paths of pinning sites stop the interface .
thus , the roughness of the pinned interface is given by the scaling of the directed paths and the corresponding roughness exponent is @xmath75 @xcite .
the other critical exponents of the dpd model can also be obtained by a mapping to directed percolation ; in @xmath69 the dynamical exponent @xmath76 , the correlation length exponent @xmath77 , and the velocity exponent @xmath78 .
amaral , barabsi , and stanley @xcite measured the tilt dependence of the velocity for several versions of the dpd model and found a behavior consistent with the relevance of the term @xmath63 at the depinning transition .
galluccio and zhang @xcite simulated a self - organized version @xcite of eq .
( [ dht ] ) and obtained @xmath79 , thereby supporting the conjecture of ref .
recently , olami , procaccia , and zeitak @xcite argued that the slopes of the pinned surfaces in eq .
( [ dht ] ) are bounded and that therefore eq .
( [ dht ] ) could belong to the directed percolation universality class . however , csahk et al .
@xcite performed the only direct simulation of the continuum equation ( [ dht ] ) and came to a different conclusion : their numerical values for the exponents @xmath80 and @xmath81 in @xmath69 are in agreement with a scaling theory for eq .
( [ dht ] ) which yields @xmath82 and @xmath83 @xcite .
yet another proposal was made by parisi @xcite .
he argued that @xmath84 and supported this value in @xmath69 by a simulation of a lattice model , which was assumed to be in the universality class of eq .
( [ dht ] ) .
problematic is also the interpretation of stepanow s renormalization - group calculation in @xmath85 dimensions @xcite .
the extrapolation to @xmath69 gives the results @xmath86 , @xmath87 , @xmath88 , and @xmath89 @xcite . to resolve these discrepancies we carefully determine in this paper the critical exponents by large - scale simulations of eq .
( [ dht ] ) for interface dimension @xmath69 . to this end
we carry out a numerical integration of the equation of motion with a continuous height variable as well as a simulation of an automaton model where the height variable takes integer values only .
in addition to the previously measured exponents @xmath80 and @xmath81 we also determine the exponents @xmath90 and @xmath91 to strengthen our conclusions .
we find that the numerical values for all critical exponents are in agreement with the suggestion that eq .
( [ dht ] ) and the dpd model belong to the same universality class . in addition , we investigate the interface roughness for @xmath43 .
it is shown that a moving interface is not self - affine and that the behavior of the roughness is very similar to that of the dpd model .
first we simulate eq . ( [ dht ] ) with a discretization of the transverse
coordinate only , @xmath92 ( with @xmath93 ) .
the random forces @xmath94 are chosen as follows : each integer position @xmath95 on a square lattice is assigned a random number @xmath96 between zero and one . for non - integer @xmath95
the forces @xmath94 are obtained by linear interpolation @xcite .
finally , the @xmath28-coordinates of @xmath94 in each column @xmath97 are shifted by a random amount @xmath98 , i.e. , @xmath99 @xcite . at @xmath100
the interface is flat , @xmath101 .
the interface configuration at @xmath102 is calculated simultaneously for all @xmath97 using the method of finite differences @xcite , @xmath103 + \ ] ] @xmath104 ^ 2 + g \eta _ i ( h_i ) \bigr \ } .\ ] ] periodic boundary conditions are used and @xmath105 is a parameter measuring the strength of the disorder . we choose the parameters @xmath106 , @xmath107 , @xmath108 @xcite and use @xmath109 , for which eq .
( [ hdt ] ) is found to be stable .
we checked that simulations with @xmath110 yield consistent results .
since the simulations of the continuum equation ( [ hdt ] ) are computationally expensive , we also study a lattice model @xcite of probabilistic cellular automata , which allows to determine the critical exponents more effectively .
the automaton model is defined on a square lattice where each cell @xmath111 $ ] ( with @xmath93 ) is assigned a random force @xmath112 which takes the value 1 with probability @xmath113 and @xmath114 with probability @xmath115 . during the motion at a given time @xmath116 the local force @xmath117 + \ ] ] @xmath118 ^ 2 + g \eta _ { i , h_i}\ ] ] is determined for all @xmath97 . the interface configuration is then updated simultaneously for all @xmath97 @xcite : @xmath119 the difference @xmath120 determines the driving force . for simplicity
, we use the density @xmath113 as the tunable parameter , and the depinning threshold is denoted by @xmath121 .
the results shown in this paper were performed with the following parameters , @xmath122 , @xmath123 , and @xmath124 @xcite .
the growth rule specified by eqs .
( [ f_i ] ) and ( [ h_i ] ) can be derived from the continuum equation ( [ dht ] ) by temporal and spatial discretizations .
in addition , a simple two - state random force @xmath125 is used .
the discretization implies that the critical slowing down close to the threshold is reduced .
this can be seen as follows . for @xmath126 , the local force for most of the interface elements @xmath127 in eq .
( [ hdt ] ) is almost zero .
nonetheless , in the finite difference approximation eq .
( [ hdt ] ) , all @xmath128 and @xmath129 have to be updated at each time step . on the other hand ,
only a subset of values @xmath130 and @xmath131 are updated at each time step in the cellular automata model .
in this section we determine the critical exponents @xmath80 and @xmath81 , defined in eqs .
( [ w^2sca ] ) and ( [ w2 t ] ) , respectively .
first we find the threshold value @xmath19 and @xmath121 from a measurement of the interface width @xmath132 for different values of the driving force . since the correlation length increases when @xmath133 , the range of the scaling regime @xmath134 also increases ( see eq .
( [ w2 t ] ) ) .
the threshold is estimated as the value where the power - law scaling holds for the longest time interval .
this method allows to determine the threshold very accurately .
= = 0.275 = = 0.275 in fig .
[ wt]a the width @xmath132 at the estimated threshold is shown in the transient regime , @xmath135 . in fig .
[ wt]b the corresponding effective exponents @xmath136/\log 4 $ ] are shown .
plotting the effective exponents @xmath137 sensitively shows the quality of the power - law scaling . from fig .
[ wt ] we conclude that @xmath138 for both the continuum equation ( [ hdt ] ) and the automaton ( eqs . ( [ f_i ] ) , ( [ h_i ] ) ) .
this is in very good agreement with the value @xmath139 , obtained by mapping the dpd model to directed percolation @xcite .
a convenient estimate of the roughness exponent @xmath80 can be obtained by measuring the equal - time correlation function @xmath140 ^ 2 } \rangle$ ] of pinned interfaces . in fig .
[ cr]a , the scaling @xmath141 for @xmath142 and @xmath143 is shown .
the corresponding effective exponents @xmath144/\log 4 $ ] are plotted in fig .
[ cr]b and [ cr]c . equation ( [ hdt ] ) and the automaton yield the same result , @xmath145 , again in agreement with the prediction of directed percolation , @xmath75 . using the numerical result for @xmath81 , we see that the simulations are also consistent with @xmath146 @xcite . in this section
we consider the steady - state behavior of the roughness for moving interfaces at driving forces @xmath147 such that @xmath148 .
we are interested in the large - time limit , @xmath149 , when the instantaneous velocity of the interface fluctuates around its mean value @xmath150 .
recently , it was proposed that the roughness of moving interfaces in the dpd model exhibits scaling with exponents different from the critical ones @xcite .
this is in disagreement with ref .
@xcite , where it was argued that on scales @xmath151 , the moving interfaces are _ not _ self - affine , because they consist of pinned regions with @xmath79 and laterally moving regions with roughly linear slopes ( @xmath152 ) . as a consequence , different moments of the equal - time correlation function , @xmath153 , yield different effective roughness exponents , @xmath154/[q \log 2]$ ] .
we measure @xmath155 for the continuum equation ( [ hdt ] ) and for the automaton and find results very similar to the dpd model @xcite ( see fig .
[ cr]b and [ cr]c ) .
the effective exponents @xmath156 increase with @xmath157 .
the reason is that the moving regions ( large slopes ) have an increasing weight with increasing @xmath157 @xcite .
another possibility to investigate the scaling of moving interfaces is a measurement of the height - height correlation function , @xmath158 ^ 2 } \rangle$ ] , for @xmath149 and @xmath148 . for @xmath159 ,
the height - height correlation function scales as @xmath160 , provided the interface is self - affine . as an illustration we consider the case @xmath74 .
= = 0.239 = = 0.239 = = 0.239 the growth exponent @xmath81 has been determined in refs .
@xcite by simulations of the automaton model eqs .
( [ f_i ] ) , ( [ h_i ] ) with @xmath74 .
the scaling of the interface width at threshold , eq .
( [ w2 t ] ) , yields @xmath161 @xcite and @xmath162 @xcite . here
, we measure the effective exponents in the steady - state regime , @xmath163 / \log 4 $ ] .
= = 0.27 = = 0.27 from fig .
[ betamfd]a we see that @xmath164 is approximately constant over three orders of magnitude of @xmath165 , with @xmath166 .
the effective exponent @xmath167 of the infinite moment of the height - height correlation function , @xmath168 , is also shown in fig .
[ betamfd]a .
the effective exponent @xmath169 increases slightly for @xmath170 and approaches a value @xmath171 which is consistent with the values of refs . @xcite and @xcite obtained by the width at threshold .
we conclude that the interface in eq .
( [ dht ] ) with @xmath74 is self - affine with the same growth exponent @xmath81 for both @xmath172 and @xmath43 .
in contrast , the same measurement for the dpd model ( see fig . [
betamfd]b ) gives decreasing effective exponents @xmath164 and @xmath167 , which shows that there is no scaling with the critical growth exponent @xmath173 . for small times
@xmath165 , the effective exponents @xmath164 and @xmath167 are roughly one , which can be understood from the lateral motion of the parts of the interface with linear slope .
= = 0.27 = = 0.27 the behavior of the automaton with @xmath174 is very similar ( see fig . [
betamca]b ) .
the effective exponent @xmath164 for the continuum equation ( fig .
[ betamca]a ) also decreases with @xmath165 but shows a plateau at a value @xmath175 .
the infinite moment @xmath176 seems to show a scaling for large @xmath165 with @xmath177 .
it is not clear whether or not these plateaus in fig .
[ betamca]a can be considered as some ( multi- ) scaling regime .
we can only conclude that the behavior is different from that of a self - affine interface which is characterized by the same critical growth exponent @xmath81 for @xmath172 and @xmath43 as long as time scales @xmath159 are considered .
we now proceed to determine the correlation length exponent @xmath90 and the velocity exponent @xmath91 , which characterize the behavior of the interface close to the depinning transition .
= = 0.27 = = 0.27 in fig .
[ nu ] , scaling plots according to eqs .
( [ xi ] ) and ( [ w^2sca ] ) are shown .
since we already determined the threshold and the critical exponents @xmath80 and @xmath178 , we can tune the correlation length exponent @xmath90 to achieve the best data collapse .
the result for eq .
( [ hdt ] ) is @xmath179 and for the automaton @xmath180 .
the corrections to the scaling @xmath181 due to finite size effects are much larger for the continuum equation
. therefore simulations very close to the threshold are not shown in fig .
[ nu]a and the error bar on the result for @xmath90 is rather large .
the results for the correlation length exponent are consistent with the prediction of directed percolation , @xmath77 .
= = 0.27 = = 0.27 the growth exponent @xmath52 defined in eq .
( [ w^2cas ] ) for driving forces @xmath43 and time scales @xmath49 is found to be @xmath182 for the continuum equation and @xmath183 for the automaton .
these values for @xmath184 are consistent with the exponent @xmath185 of the kpz equation @xcite .
this supports the picture that the quenched disorder @xmath70 crosses over to a thermal noise @xmath186 on time scales @xmath49 ( see fig .
[ nu ] ) .
the scaling eq . ( [ v ] ) of the steady - state velocity @xmath150 and the corresponding effective exponents @xmath91 are shown in fig .
[ theta ] .
the result for the continuum equation is @xmath187 , and @xmath188 for the automaton , again in agreement with the value of the dpd model , @xmath78 @xcite .
the results for the roughness of eq .
( [ dht ] ) with and without the term @xmath63 are summarized in table i. [ cols="^,^,^",options="header " , ] 0.2truecm table ii .
comparison of the critical exponents of the dpd model with our numerical results . 0.4truecm the numerical results for both , the continuum equation ( [ hdt ] ) and the automaton are in excellent agreement with the dpd model .
we therefore expect that the critical exponents of the anisotropic depinning model eq .
( [ dht ] ) are exactly given by the exponents of directed percolation .
in the dpd model , the dynamic and the static behavior is determined by directed percolation paths of pinning sites . due to the restricted solid - on - solid condition of the directed percolation paths , the pinned regions of the interface have small gradients ( bounded by a slope one ) @xcite .
in contrast , the laterally moving regions have a linear slope of about two @xcite .
this behavior is analogous to that of eq .
( [ dht ] ) ; regions of the interface with large slopes are likely to move due to the positive contribution of the term @xmath63 to the driving force
. regions of the interface with small gradients , the other hand , are easier to pin , due to the smaller contribution of @xmath63 .
the observation that the interface motion is mainly due to the gradient term @xmath63 , causes a clustering of growth sites , which can be understood as follows . a motion of an interface element @xmath189 increases the contribution of @xmath63 to the local force at @xmath190 or @xmath191 .
thus , this neighboring interface element is likely to be the next new growth site , resulting in a cluster of growth sites to be formed .
the moving regions have larger slopes than the pinned parts . as a consequence ,
a moving interface is not self - affine .
the clustering of growth sites in eq .
( [ dht ] ) does not destroy the dynamical scaling of global quantities of the interface , such as the width and the velocity ( see eqs .
( [ v][w2 t ] ) ) . the scaling relation ( [ exasca ] )
is also fulfilled by the exponents of the dpd model and by our numerical results .
the behavior of moving interfaces deserves further investigation to understand better the concepts of scaling and self - affinity ( see sec .
further , it would be very interesting to construct a rigorous proof that eq .
( [ dht ] ) and the dpd model are in the same universality class .
i would like to thank l. a. n. amaral , p. bak , s. v. buldyrev , r. cuerno , s. galluccio , s. t. harrington , s. havlin , k. b. lauritsen , h. a. makse , m. paczuski , h. e. stanley , l .- h .
tang , and s. zapperi for valuable discussions .
i acknowledge support from the deutsche forschungsgemeinschaft .
the center for polymer studies is supported by the national science foundation . for reviews
see e.g. , t. nattermann and p. rujan , int . j. mod .
b * 3 * , 1597 ( 1989 ) ; d. p. belanger and a. p.
young , j. magn .
magn . mat . *
100 * , 272 ( 1991 ) ; g. forgacs , r. lipowsky , and th . m. nieuwenhuizen , in _ phase transitions and critical phenomena _ , vol .
14 , edited by c. domb and j. l. lebowitz ( academic press , london , 1991 ) , p.135 .
m. a. rubio , c. a. edwards , a. dougherty , and j. p. gollub , phys .
* 63 * , 1685 ( 1989 ) , v. k. horvth , f. family , and t. vicsek , j. phys .
a * 24 * , l25 ( 1991 ) ; s .- j .
he , g. kahanda , and p .- z .
wong , phys .
lett . * 69 * , 3731 ( 1992 ) . for reviews
see e.g. , d. s. fisher , m. p. a. fisher , and d. a. huse , phys .
b * 43 * , 130 ( 1991 ) ; g. blatter , m. v. feigelman , v. b. geshkenbein , a. i. larkin , and v. m. vinokur , rev .
phys . * 66 * , 1125 ( 1994 ) .
barabsi , s. v. buldyrev , s. havlin , g. huber , h. e. stanley , and t. vicsek , p. 193
, in _ surface disordering : growth , roughening , and phase transitions _ , r. jullien , j. kertsz , p. meakin , and d. e. wolf ( ed . ) , nova science , new york ( 1992 ) .
s. havlin , a .-
barabsi , s. v. buldyrev , c. k. peng , m. schwartz , h. e. stanley , and t. vicsek in _ growth patterns in physical sciences and biology _ ,
e. louis , l. sander and p. meakin ( ed . ) , plenum , new york ( 1993 ) .
galluccio und zhang @xcite used a different discretization of the gradient term in eq .
( [ dht ] ) , @xmath192 ( s. galluccio , private communication ) .
we believe that this is an unsuitable discretization causing a growth instability of eq .
( [ dht ] ) : it has been observed in ref .
@xcite that an interface element @xmath95 will grow infinitely faster than the rest of the interface when the height difference of @xmath95 to its neighbors exceeds a certain amount .
it has been concluded in ref .
@xcite that the anisotropic depinning model eq .
( [ dht ] ) is ill defined due to this instability .
however , with the correct discretization in eqs .
( [ hdt ] ) and ( [ f_i ] ) no growth instability occurs and eq .
( [ dht ] ) is well defined .
the strength of the disorder @xmath105 has to be significantly larger than @xmath193 to assure that the depinning transition occurs at a critical driving force greater than zero .
the stiffness @xmath5 is chosen to have the same order of magnitude as @xmath105 , so that the interface can become rough on length scales of the lattice constant .
we checked that other parameters yield consistent results . | we study numerically a stochastic differential equation describing an interface driven along the hard direction of an anisotropic random medium .
the interface is subject to a homogeneous driving force , random pinning forces and the surface tension .
in addition , a nonlinear term due to the anisotropy of the medium is included .
the critical exponents characterizing the depinning transition are determined numerically for a one - dimensional interface .
the results are the same , within errors , as those of the `` directed percolation depinning '' ( dpd ) model .
we therefore expect that the critical exponents of the stochastic differential equation are exactly given by the exponents obtained by a mapping of the dpd model to directed percolation .
we find that a moving interface near the depinning transition is not self - affine and shows a behavior similar to the dpd model .
# 1#2#3#4 # 1#2 2 |
throughout this paper we work in the piecewise linear category .
let @xmath0 be a finite graph .
an embedding @xmath1 of @xmath0 into the @xmath2-sphere is called a _ spatial embedding _ of @xmath0 and @xmath3
is called a _ spatial graph_.
we denote the set of all spatial embeddings of @xmath0 by @xmath4 .
we call a subgraph @xmath5 of @xmath0 which is homeomorphic to the circle a _ cycle _ of @xmath0 and denote the set of all cycles of @xmath0 by @xmath6 . in particular , we call a cycle of @xmath0 which contains exactly @xmath7 edges a _ @xmath7-cycle _ of @xmath0 and denote the set of all @xmath7-cycles of @xmath0 by @xmath8 . for a positive integer @xmath9 , @xmath10 denotes the set of all cycles of @xmath0 ( @xmath11 ) if @xmath12 and the set of all unions of mutually disjoint @xmath9 cycles of @xmath0 if @xmath13 .
we denote the union of @xmath10 over all positive integer @xmath9 by @xmath14 . for an element @xmath5 in @xmath10 and an element @xmath1 in @xmath4 , @xmath15 is none other than a knot in @xmath3 if @xmath12 and an @xmath9-component link in @xmath3 if @xmath13 .
@xmath16-exchange _ is an operation to obtain a new graph @xmath17 from a graph @xmath18 by removing all edges of a @xmath2-cycle @xmath19 $ ] of @xmath18 with the edges @xmath20 and @xmath21 , and adding a new vertex @xmath22 and connecting it to each of the vertices @xmath23 and @xmath24 as illustrated in fig .
[ delta - y ] ( we often denote @xmath25 by @xmath26 ) .
@xmath27-exchange _ is the reverse of this operation .
let @xmath28 be the _ complete graph _ on @xmath9 vertices , namely the simple graph consisting of @xmath9 vertices in which every pair of distinct vertices is connected by exactly one edge .
the set of all graphs obtained from @xmath29 by a finite sequence of @xmath16-exchanges and @xmath27-exchanges is called the _
petersen family_. the petersen family consists of seven graphs @xmath29 , @xmath30 , @xmath31 , @xmath32 , @xmath33 , @xmath34 and the _ petersen graph _ @xmath35 as illustrated in fig . [ petersen ] , where an arrow between two graphs indicates the application of a single @xmath16-exchange . for spatial embeddings of a graph in the petersen family , the following is known .
[ pf ] let @xmath0 be an element in the petersen family .
for any element @xmath1 in @xmath4 , it follows that @xmath36 where @xmath37 denotes the _ linking number _ in the @xmath2-sphere .
we remark here that the case of @xmath38 in theorem [ pf ] is what is called the conway - gordon @xmath29 theorem @xcite , and the other cases were shown by sachs @xcite indirectly , and also pointed out by taniyama - yasuhara @xcite .
theorem [ pf ] implies that each element @xmath0 in the petersen family is _ intrinsically linked _
, that is , for any element @xmath1 in @xmath4 , there exists an element @xmath5 in @xmath39 such that @xmath15 is a nonsplittable @xmath40-component link .
it is known that a graph is intrinsically linked if and only if the graph contains an element in the petersen family as a minor @xcite .
namely , the petersen family plays the role of a complete obstruction for graphs not to be intrinsically linked .
our purpose in this paper is to give an integral lift of theorem [ pf ] . in the following
, @xmath41 denotes the @xmath42-th coefficient of the _ conway polynomial _ for an oriented link @xmath43 .
[ petersen_refine ] let @xmath0 be an element in the petersen family .
we give the labels for all vertices of @xmath0 as indicated in fig .
[ petersen ] .
let @xmath44 be a map from @xmath6 to the set of integers @xmath45 defined as follows : 1 .
if @xmath38 , then for an element @xmath5 in @xmath46 , we define @xmath47 2 .
if @xmath48 , then for an element @xmath5 in @xmath49 , we define @xmath50 where @xmath51 3 .
if @xmath52 , then for an element @xmath5 in @xmath53 , we define @xmath54 where @xmath55 4 . if @xmath56 , then for an element @xmath5 in @xmath57 , we define @xmath58 where @xmath59 5 . if @xmath60 , then for an element @xmath5 in @xmath61 , we define @xmath62 where @xmath63 6 . if @xmath64 , then for an element @xmath5 in @xmath65 , we define @xmath66 where @xmath67 7 . if @xmath68 , then for an element @xmath5 in @xmath69 , we define @xmath70 then for any element @xmath1 in @xmath4 , it follows that @xmath71 note that theorem [ pf ] can be obtained from theorem [ petersen_refine ] by taking the modulo two reduction .
we also should remark here that theorem [ petersen_refine ] was already known in the case of @xmath38 ( nikkuni @xcite ) , @xmath32 ( odonnol @xcite ) and @xmath30 ( nikkuni - taniyama @xcite ) .
the other cases are new .
we say that an element @xmath1 in @xmath4 is _ knotted _
if @xmath3 contains a nontrivial knot , and _ complexly algebraically linked _ if @xmath3 contains a @xmath40-component link whose linking number is not equal to @xmath72 or a pair of @xmath40-component links with nonzero linking number @xcite .
then odonnol showed the following .
[ calinked ] ( odonnol @xcite ) let @xmath0 be an element in the petersen family and @xmath1 an element in @xmath4 . if @xmath1 is complexly algebraically linked , then @xmath1 is knotted . in @xcite , the place of the cycle whose image is a nontrivial knot
was not examined .
as an application of theorem [ petersen_refine ] , we give an alternative proof of theorem [ calinked ] in more refined form as follows .
[ calinked_refine ] let @xmath0 be an element in the petersen family and @xmath44 the map from @xmath6 to @xmath45 in theorem [ petersen_refine ] .
let @xmath1 be an element in @xmath4 .
if @xmath1 is complexly algebraically linked , then it follows that @xmath73 by theorem [ pf ] , it follows that @xmath1 is complexly algebraically linked if and only if @xmath74 then by theorem [ petersen_refine ] and ( [ calk ] ) , it follows that @xmath75 thus we have the desired conclusion . by corollary [ calinked_refine ]
, there exists an element @xmath76 in @xmath6 with @xmath77 such that @xmath78 .
namely , corollary [ calinked_refine ] refines theorem [ calinked ] by identifying the cycles that might be nontrivial knots in @xmath3 . in the next section ,
we prepare general results for @xmath16-exchanges and the conway - gordon type theorems which are based on @xcite .
we give a proof of theorem [ petersen_refine ] in section @xmath2 .
let @xmath18 and @xmath17 be two graphs such that @xmath17 is obtained from @xmath18 by a single @xmath16-exchange .
we denote the set of all elements in @xmath79 containing @xmath80 by @xmath81 .
let @xmath82 be an element in @xmath79 which does not contain @xmath80 .
then there exists an element @xmath83 in @xmath84 such that @xmath85 .
it is easy to see that the correspondence from @xmath82 to @xmath83 defines a surjective map @xmath86 in particular , if @xmath82 is an element in @xmath87 then @xmath83 is an element in @xmath88 .
this implies that the restriction map of @xmath89 on @xmath87 induces a surjective map from @xmath87 to @xmath88 , and thus it is clear that if @xmath90 is an empty set , then @xmath88 is also an empty set .
the inverse image of an element @xmath5 in @xmath84 by @xmath89 contains at most two elements in @xmath91 .
[ not_inj2 ] illustrates the case that the inverse image of @xmath5 by @xmath89 consists of exactly two elements . in general , the inverse image of @xmath5 by @xmath89 consists of exactly one element if and only if @xmath5 contains @xmath92 and @xmath22 , or @xmath5 does not contain @xmath22
let @xmath93 be an additive group .
we say that an @xmath93-valued unoriented link invariant @xmath94 is _ compressible _ if @xmath95 for any unoriented link @xmath43 which has a component @xmath96 bounding a disk @xmath97 in the @xmath2-sphere with @xmath98 .
suppose that for each element @xmath82 in @xmath79 , an @xmath93-valued unoriented link invariant @xmath99 is assigned .
then for each element @xmath5 in @xmath84 , we define an @xmath93-valued unoriented link invariant @xmath100 by @xmath101 for an unoriented link @xmath43
. then the following theorem holds .
[ main2 ] ( nikkuni - taniyama @xcite ) suppose that @xmath99 is compressible for each element @xmath82 in @xmath102 .
suppose that there exists a fixed element @xmath103 in @xmath93 such that @xmath104 for any element @xmath105 in @xmath106 .
then we have @xmath107 for any element @xmath1 in @xmath108 . by an application of theorem [ main2 ] ,
the following is shown .
[ main3 ] ( nikkuni - taniyama @xcite ) let @xmath0 be an element in the petersen family .
then , there exists a map @xmath44 from @xmath6 to @xmath45 such that for any element @xmath1 in @xmath4 , it follows that @xmath109 we give a proof of theorem [ main3 ] for the reader s convenience .
let @xmath18 and @xmath17 be two elements in the petersen family such that @xmath17 is obtained from @xmath18 by a single @xmath16-exchange .
assume that there exists a map @xmath110 from @xmath79 to @xmath45 such that for any element @xmath105 in @xmath106 , it follows that @xmath111 for each element @xmath82 in @xmath79 , we define an integer - valued unoriented link invariant @xmath99 of an unoriented link @xmath43 as follows .
note that @xmath18 is obtained from @xmath29 or @xmath32 by a finite sequence of @xmath16-exchanges .
since both @xmath112 and @xmath113 are empty sets for @xmath114 , we have @xmath90 is an empty set for @xmath114 . if @xmath82 is an element in @xmath115 , then @xmath116 if @xmath43 is a knot and @xmath117 if @xmath43 is not a knot . if @xmath82 is an element in @xmath118 , then @xmath119 if @xmath43 is a @xmath40-component link and @xmath117 if @xmath43 is not a @xmath40-component link .
then by ( [ assump ] ) , we have @xmath120 note that @xmath99 is compressible for any element @xmath82 in @xmath79 .
thus by theorem [ main2 ] and ( [ const_k6 ] ) , for any element @xmath1 in @xmath108 , it follows that @xmath121 now we define a map @xmath122 from @xmath123 to @xmath45 by @xmath124 for an element @xmath5 in @xmath123 .
let @xmath5 be an element in @xmath84 .
note that @xmath88 is also an empty set for @xmath114 . if @xmath5 belongs to @xmath125 , then by ( [ c ] ) , we have @xmath126 if @xmath5 belongs to @xmath127 , then @xmath128 consists of exactly one element because each union of two mutually disjoint cycles of a graph in the petersen family contains all of the vertices of the graph
. then we have @xmath129 thus by combining ( [ const_q7 ] ) , ( [ omega ] ) and ( [ xi ] ) , we have @xmath130 namely , for any element @xmath1 in @xmath108 , it follows that @xmath131 as we remarked before , the cases of @xmath29 and @xmath32 have already shown by @xcite and @xcite , respectively .
thus by repeating the argument as above , we have the desired conclusion . in the following , we show two lemmas which are useful in proving theorem [ petersen_refine ] .
we say that two cycles of a graph are _ edge - disjoint _ if the intersection of them does not contain an edge .
let @xmath0 be a graph and @xmath132 @xmath2-cycles of @xmath0 such that @xmath133 is edge - disjoint for @xmath134 .
then we also can regard @xmath135 as a @xmath2-cycle of the graph obtained from @xmath0 by a finite sequence of @xmath16-exchanges at @xmath136 s for @xmath134 .
let @xmath137 be a graph obtained from @xmath138 by a single @xmath16-exchange at @xmath139 , where @xmath140 ( @xmath141 ) .
on the other hand , let @xmath142 be a permutation of order @xmath7 .
let @xmath143 be a graph obtained from @xmath144 by a single @xmath16-exchange at @xmath145 , where @xmath146 ( @xmath141 ) .
note that @xmath147 .
[ inverse ] for any element @xmath5 in @xmath148 , it follows that @xmath149 let @xmath82 be an element in the inverse image of @xmath5 by @xmath150 . since @xmath151 is edge - disjoint for @xmath134
, we have @xmath152 thus @xmath82 also be an element in the inverse image of @xmath5 by @xmath153 .
this implies the result .
we assign an @xmath93-valued unoriented link invariant @xmath99 for each element @xmath82 in @xmath14
. then for an element @xmath5 in @xmath154 , we define an @xmath93-valued unoriented link invariant @xmath155 by @xmath156 if @xmath157 and @xmath158 if @xmath141 with respect to the sequence of @xmath16-exchanges at @xmath132 . on the other hand
, we define an @xmath93-valued unoriented link invariant @xmath159 by @xmath160 if @xmath157 and @xmath161 if @xmath141 with respect to the sequence of @xmath16-exchanges at @xmath162 .
then we have the following .
[ ab ] for any element @xmath5 in @xmath148 , it follows that @xmath163 .
let @xmath5 be an element in @xmath148 .
then for an unoriented link @xmath43 , by lemma [ inverse ] , we have @xmath164 thus we have the result .
let @xmath0 be a graph and @xmath165 a set of mutually edge - disjoint @xmath2-cycles of @xmath0 .
we say that @xmath166 is _ stable _ if for any @xmath167-element subset @xmath168 of @xmath166 ( @xmath169 ) , @xmath16-exchanges at @xmath170 produce the same graph up to isomorphism .
we denote the @xmath2-cycles @xmath171 $ ] , @xmath172 $ ] , @xmath173 $ ] , @xmath174 $ ] and @xmath175 $ ] of @xmath29 by @xmath176 , @xmath177 , @xmath178 , @xmath179 and @xmath180 , respectively .
note that @xmath151 is edge - disjoint for @xmath181 , and @xmath182 is also edge - disjoint .
\(2 ) let @xmath0 be @xmath30 which is obtained from @xmath29 by a single @xmath16-exchange at @xmath176 . though this case has been shown in @xcite , we give it again for the reader s convenience .
let @xmath183 be the map from @xmath184 to @xmath45 as in ( 1 ) and @xmath185 the map from @xmath186 to @xmath45 defined by ( [ c ] ) with respect to the @xmath16-exchange at @xmath176 . in the following we show @xmath187 .
then the result follows from the proof of theorem [ main3 ] .
let @xmath5 be an element in @xmath49 .
if @xmath5 belongs to @xmath188 , then there uniquely exists an element @xmath82 in @xmath189 such that @xmath190 .
thus we have @xmath191 .
if @xmath5 belongs to @xmath192 , then @xmath5 does not contain exactly one of the vertices of @xmath30 .
then it is sufficient to consider the following three cases up to symmetry of @xmath30 : ( i ) if @xmath5 does not contain @xmath22 , then there uniquely exists an element @xmath82 in @xmath189 such that @xmath190 .
thus we have @xmath191 .
( ii ) if @xmath5 does not contain @xmath193 , then there exists an element @xmath194 in @xmath195 and an element @xmath196 in @xmath189 such that @xmath197 .
thus we have @xmath198 .
( iii ) if @xmath5 does not contain @xmath199 , then there uniquely exists an element @xmath82 in @xmath195 such that @xmath190 .
thus we have @xmath200 . if @xmath5 belongs to @xmath201 , then the inverse image of @xmath5 by @xmath89 consists of exactly one element in @xmath195 or a pair of an element in @xmath195 and an element in @xmath202 .
thus in any case we have @xmath203 .
if @xmath5 belongs to @xmath204 , we have @xmath205 . in conclusion
, we see that @xmath206 for an element @xmath5 in @xmath49 .
thus it follows that @xmath187 .
\(3 ) let @xmath0 be @xmath31 which is obtained from @xmath30 by a single @xmath16-exchange at @xmath180 .
let @xmath207 be the map from @xmath186 to @xmath45 as in ( 2 ) and @xmath208 the map from @xmath209 to @xmath45 defined by ( [ c ] ) with respect to the @xmath16-exchange at @xmath180 . in the following we show @xmath210 .
let @xmath5 be an element in @xmath53 .
note that @xmath211 if @xmath212 . if @xmath5 belongs to @xmath213 , then there uniquely exists an element @xmath82 in @xmath188 such that @xmath214 .
thus we have @xmath215 .
if @xmath5 belongs to @xmath216 , then @xmath5 does not contain exactly two of the vertices of @xmath31 .
then it is sufficient to consider the following three cases up to symmetry of @xmath31 : ( i ) if @xmath5 does not contain either @xmath22 or @xmath217 , then there uniquely exists an element @xmath82 in @xmath192 such that @xmath214 . since @xmath82 does not contain @xmath22 , we have @xmath215 .
( ii ) if @xmath5 contains exactly one of @xmath22 and @xmath217 , then we may assume that @xmath5 contains @xmath217 by the following reason .
note that @xmath218 is stable . actually , the graph @xmath219 which is obtained from @xmath29 by a single @xmath16-exchange at @xmath180 is isomorphic to @xmath30 , and @xmath31 is also obtained from @xmath219 by a single @xmath16-exchange at @xmath176 , see fig .
[ q8iso ] . then the map from @xmath53 to @xmath45 obtained from @xmath183 by @xmath220 coincides with @xmath208 by ( [ omega ] ) and lemma [ ab ] .
thus the case that @xmath5 contains @xmath22 and the case that @xmath5 contains @xmath217 are compatible through the isomorphism between @xmath219 and @xmath30 . from now on , to simplify the description of the proof we often use the argument of such a compatibility without any notice .
now assume that @xmath5 contains @xmath217 .
then there uniquely exists an element @xmath82 in @xmath201 such that @xmath214 .
thus we have @xmath221 .
( iii ) if @xmath5 contains both @xmath22 and @xmath217 , then we may assume that @xmath5 does not contain either @xmath193 or @xmath199 .
then there exists an element @xmath194 in @xmath201 and an element @xmath196 in @xmath192 such that @xmath222 .
since @xmath196 does not contain @xmath193 , we have @xmath223 .
if @xmath5 belongs to @xmath224 , we have @xmath225 . in conclusion
, we see that @xmath226 for an element @xmath5 in @xmath53 .
thus it follows that @xmath210 .
\(5 ) let @xmath0 be @xmath33 which is obtained from @xmath30 by a single @xmath16-exchange at @xmath177 .
note that @xmath227 is stable , see fig .
[ p8iso ] .
let @xmath207 be the map from @xmath186 to @xmath45 as in ( 2 ) and @xmath208 the map from @xmath228 to @xmath45 defined by ( [ c ] ) with respect to the @xmath16-exchange at @xmath177 . in the following we show @xmath229 .
let @xmath5 be an element in @xmath61 .
if @xmath5 belongs to @xmath230 , then there uniquely exists an element @xmath82 in @xmath188 such that @xmath231 .
thus we have @xmath215 .
if @xmath5 belongs to @xmath232 , then @xmath5 does not contain exactly one of the vertices of @xmath33 .
then it is sufficient to consider the following four cases up to symmetry of @xmath33 : ( i ) if @xmath5 does not contain one of @xmath22 and @xmath233 , then we may assume that @xmath5 contains @xmath22 .
then there uniquely exists an element @xmath82 in @xmath188 such that @xmath231 .
thus we have @xmath215 .
( ii ) if @xmath5 does not contain @xmath24 , then there exists an element @xmath194 in @xmath192 and an element @xmath196 in @xmath188 such that @xmath234 . since @xmath194
also does not contain @xmath24 , we have @xmath235 .
( iii ) if @xmath5 does not contain one of @xmath236 and @xmath237 ( in other words , @xmath5 contains all of @xmath238 and @xmath93 ) , then we may assume that @xmath5 does not contain @xmath193
. then there uniquely exists an element @xmath82 in @xmath192 such that @xmath231 . since @xmath82
also does not contain @xmath193 , we have @xmath239 .
( iv ) if @xmath5 does not contain @xmath93 , then there uniquely exists an element @xmath82 in @xmath192 such that @xmath231 .
since @xmath82 contains all of @xmath92 and @xmath22 , we have @xmath221 .
if @xmath5 belongs to @xmath240 , then @xmath5 does not contain exactly two of the vertices of @xmath33 .
then it is sufficient to consider the following four cases up to symmetry of @xmath33 : ( i ) if @xmath5 contains @xmath24 and does not contain one of @xmath22 and @xmath233 , then we may assume that @xmath5 does not contain @xmath233 .
then there uniquely exists an element @xmath82 in @xmath192 such that @xmath231 . since @xmath82 contains all of @xmath92 and @xmath22 , we have @xmath221 .
( ii ) if @xmath5 does not contain @xmath24 and one of @xmath22 and @xmath233 , then we may assume that @xmath5 does not contain @xmath233 .
then there uniquely exists an element @xmath82 in @xmath192 such that @xmath231 . since @xmath82 does not contain @xmath24 , we have @xmath239 .
( iii ) if @xmath5 does not contain either @xmath24 or @xmath93 , then there exists an element @xmath194 in @xmath201 and an element @xmath196 in @xmath192 such that @xmath234 . since @xmath196 contains all of @xmath92 and @xmath22 , we have @xmath241 .
( iv ) if @xmath5 contains all of @xmath242 and @xmath24 , then we may assume that @xmath5 does not contain either @xmath193 or @xmath199 .
then there exists an element @xmath194 in @xmath201 and an element @xmath196 in @xmath192 such that @xmath234 .
since @xmath196 does not contain @xmath193 , we have @xmath223 .
if @xmath5 belongs to @xmath243 , then we have @xmath244 in the same way as the case of @xmath30 .
if @xmath5 belongs to @xmath245 , we have @xmath225 . in conclusion
, we see that @xmath246 for an element @xmath5 in @xmath61 .
thus it follows that @xmath229 .
\(6 ) let @xmath0 be @xmath34 which is obtained from @xmath33 by a single @xmath16-exchange at @xmath178 .
note that @xmath247 is stable .
let @xmath248 be the map from @xmath228 to @xmath45 as in ( 5 ) and @xmath249 the map from @xmath250 to @xmath45 defined by ( [ c ] ) with respect to the @xmath16-exchange at @xmath178 . in the following we show @xmath251 .
let @xmath5 be an element in @xmath65 .
if @xmath5 belongs to @xmath252 , then there uniquely exists an element @xmath82 in @xmath230 such that @xmath253 .
thus we have @xmath254 .
if @xmath5 belongs to @xmath255 , then @xmath5 does not contain exactly one of the vertices of @xmath34 .
then it is sufficient to consider the following two cases up to symmetry of @xmath34 : ( i ) if @xmath5 contains all of @xmath256 and @xmath93 , then we may assume that @xmath5 does not contain one of @xmath22 and @xmath24 .
in any case , there uniquely exists an element @xmath82 in @xmath232 such that @xmath253 . since @xmath82
also does not contain one of @xmath22 and @xmath24 , we have @xmath254 .
( ii ) if @xmath5 does not contain one of @xmath256 and @xmath93 , then we may assume that @xmath5 does not contain @xmath237 .
then there uniquely exists an element @xmath82 in @xmath232 such that @xmath253 .
since @xmath82 contains all of @xmath242 and @xmath24 , we have @xmath257 .
if @xmath5 belongs to @xmath258 , then @xmath5 does not contain exactly two of the vertices of @xmath34 .
then it is sufficient to consider the following two cases up to symmetry of @xmath34 : ( i ) if @xmath5 contains all of @xmath256 and @xmath93 , then we may assume that @xmath5 does not contain either @xmath22 or @xmath24 .
then there uniquely exists an element @xmath82 in @xmath240 such that @xmath253 . since @xmath82 contains @xmath93 and does not contain @xmath24 , we have @xmath257 .
( ii ) if @xmath5 does not contain one of @xmath256 and @xmath93 , then we may assume that @xmath5 does not contain either @xmath22 or @xmath259 , or @xmath5 does not contain either @xmath199 or @xmath259 . in the former case , there uniquely exists an element @xmath82 in @xmath240 such that @xmath253 .
since @xmath82 contains @xmath24 , we have @xmath260 .
in the latter case , there exists an element @xmath194 in @xmath240 and an element @xmath196 in @xmath232 such that @xmath261 .
since @xmath194 contains @xmath24 and @xmath196 contains all of @xmath238 and @xmath93 , we have @xmath262 . if @xmath5 belongs to @xmath263 , then @xmath5 does not contain exactly three of the vertices of @xmath34 .
then it is sufficient to consider the following two cases up to symmetry of @xmath34 : ( i ) if @xmath5 contains all of @xmath256 and @xmath93 , then we may assume that @xmath5 does not contain any of @xmath242 or @xmath24 , or @xmath5 does not contain any of @xmath264 or @xmath265 .
in the former case , there uniquely exists an element @xmath82 in @xmath243 such that @xmath253 .
thus we have @xmath260 .
in the latter case , there uniquely exists an element @xmath82 in @xmath240 such that @xmath253 . since @xmath82 contains @xmath24 , we have @xmath260 .
( ii ) if @xmath5 does not contain one of @xmath256 and @xmath93 , then we may assume that @xmath5 does not contain any of @xmath266 or @xmath259 , or @xmath5 does not contain any of @xmath256 or @xmath93 .
in any case , there exists an element @xmath194 in @xmath243 and an element @xmath196 in @xmath240 such that @xmath261 .
since @xmath196 contains @xmath24 , we have @xmath267 . if @xmath5 belongs to @xmath268
, then we have @xmath269 in the same way as the case of @xmath30 .
if @xmath5 belongs to @xmath270 , we have @xmath271 . in conclusion
, we see that @xmath272 for an element @xmath5 in @xmath65 .
thus it follows that @xmath251 .
\(7 ) let @xmath0 be @xmath35 which is obtained from @xmath34 by a single @xmath16-exchange at @xmath179 .
note that @xmath273 is stable .
let @xmath274 be the map from @xmath250 to @xmath45 as in ( 6 ) and @xmath275 the map from @xmath276 to @xmath45 defined by ( [ c ] ) with respect to the @xmath16-exchange at @xmath179 .
in the following we show @xmath277 .
let @xmath5 be an element in @xmath69 .
note that @xmath278 is the empty set if @xmath279 .
if @xmath5 belongs to @xmath280 , then @xmath5 does not contain exactly one of the vertices of @xmath35 .
then it is sufficient to consider the following two cases up to symmetry of @xmath35 : ( i ) if @xmath5 does not contain @xmath281 , then there uniquely exists an element @xmath82 in @xmath252 such that @xmath282 .
thus we have @xmath283 .
( ii ) if @xmath5 contains @xmath281 , then we may assume that @xmath5 does not contain @xmath93 .
then there exists an element @xmath194 in @xmath255 and an element @xmath196 in @xmath252 such that @xmath284 .
since @xmath194 does not contain @xmath93 , we have @xmath285 .
if @xmath5 belongs to @xmath286 , then @xmath5 does not contain exactly two of the vertices of @xmath35 .
then we may assume that @xmath5 does not contain either @xmath22 or @xmath24 , or @xmath5 does not contain either @xmath93 or @xmath24 . in the former case , there uniquely exists an element @xmath82 in @xmath258 such that @xmath282 .
since @xmath82 contains all of @xmath287 and @xmath237 , we have @xmath288 .
in the latter case , there exists an element @xmath194 in @xmath258 and an element @xmath196 in @xmath255 such that @xmath284 .
since @xmath194 does not contain @xmath93 and @xmath196 contains all of @xmath287 and @xmath237 , we have @xmath289 . if @xmath5 belongs to @xmath290 , then @xmath5 does not contain exactly four of the vertices of @xmath35 .
then we may assume that @xmath5 does not contain any of @xmath291 or @xmath237 or @xmath5 does not contain any of @xmath292 or @xmath24 .
in any case , there uniquely exists an element @xmath82 in @xmath263 such that @xmath282 .
since @xmath82 does not contain one of @xmath287 and @xmath237 , we have @xmath293 . if @xmath5 belongs to @xmath294
, then we have @xmath269 in the same way as the case of @xmath30 . in conclusion
, we see that @xmath295 for an element @xmath5 in @xmath69 .
thus it follows that @xmath277 .
this completes the proof .
let us denote the @xmath2-cycle @xmath296 $ ] of @xmath32 by @xmath297 .
note that @xmath33 is obtained from @xmath32 by a single @xmath16-exchange at @xmath297 .
let @xmath298 be the map from @xmath299 to @xmath45 as in theorem [ petersen_refine ] ( 4 ) and @xmath300 the map from @xmath228 to @xmath45 defined by ( [ c ] ) with respect to the @xmath16-exchange at @xmath297 .
then it can be shown that @xmath300 coincides with @xmath301 . | for every spatial embedding of each graph in the petersen family , it is known that the sum of the linking numbers over all of the constituent 2-component links is congruent to 1 modulo 2 . in this paper
, we give an integral lift of this formula in terms of the square of the linking number and the second coefficient of the conway polynomial . |
in this note we study a class of systems whose parameters are driven by a time reversed markov chain . given a time horizon @xmath0 and a standard markov chain @xmath1
taking values in the set @xmath2 , we consider the process @xmath3 and the system @xmath4 where , as usual , @xmath5 represents the state variable of the system and @xmath6 is the control variable . these systems may be encountered in real world problems , specially when a markov chain interacts with the system parameters via a _ first in last out _ queue .
an example consists of drilling sedimentary rocks whose layers can be modelled by a markov chain from bottom to top as a consequence of their formation process .
the first drilled layer is the last formed one .
another example is a dc - motor whose brush is grind by a machine subject to failures , leaving a series of imprecisions on the brush width that can be described by a markov chain , so that the last failure will be the first to affect the motor collector depending on how the brush is installed .
one of the most remarkable features of system @xmath7 is that it provides a dual for optimal filtering of standard _
markov jump linear systems _ ( mjls ) .
in fact , if we consider a quadratic cost functional for system @xmath7 with linear state feedback , leading to an optimal control problem @xcite that we call _ time reversed markov jump linear quadratic _ problem ( trm - jlqp ) , then we show that the solution is identical to the gains of the _ linear minimum mean square estimator _ ( lmmse ) formulated in @xcite , with time - reversed gains and transposed matrices . in perspective with existing duality relations ,
the one obtained here is a direct generalization of the well known relation between control and filtering of linear time varying systems as presented for instance in ( * ? ? ?
* table 6.1 ) , or also in @xcite in different contexts . as for mjls , the duality between control and filtering
have been considered e.g. in @xcite , while purely in the context of standard mjls , thus leading to more complex relations involving certain generalized coupled riccati difference equations . here
, the duality follows naturally from the simple reversion of the markov chain given in , with no extra assumptions nor complex constructions .
another interesting feature of @xmath7 is that the variable @xmath8 , which is commonly used in the literature of mjls , @xcite , evolves along time @xmath9 according to a _ time - varying _ linear operator , as shown in remark [ rem_evolution_first_moment ] , in a marked dissimilarity with standard mjls .
this motivated us to employ @xmath10 , the conditioned second moment of @xmath11 , leading to time - homogeneous operators .
the contents of this note are as follows .
we present basic notation in section [ sec - notation ] . in section [ sec - sys ]
we give the recursive equation describing @xmath12 , which leads to a stability condition involving the spectral radius of a time - homogeneous linear operator . in section [ sec - problem ] ,
we formulate and solve the trm - jlq problem , following a proof method where we decompose @xmath12 into two components as to handle @xmath13 that are visited with zero probability . the duality with the lmmse then follows in a straightforward manner , as presented in section [ sec - duality ] .
concluding remarks are given in section [ sec - conclusions ] .
let @xmath14 be the @xmath15-dimensional euclidean space and @xmath16 be the space formed by real matrices of dimension @xmath17 by @xmath15 .
we write @xmath18 to represent the hilbert space composed of @xmath19 real matrices , that is @xmath20 , where @xmath21 , @xmath22 .
the space @xmath23 equipped with the inner product @xmath24 where @xmath25 is the trace operator and the superscript @xmath26 denotes the transpose , is a hilbert space .
the inner product induces the norm @xmath27 .
if @xmath28 , we write simply @xmath29 .
the mathematical operations involving elements of @xmath23 , are used in element - wise fashion , e.g. for @xmath30 and @xmath31 in @xmath29 we have @xmath32 , where @xmath33 is the usual matrix multiplication .
similarly , for a set of scalars @xmath34 we write @xmath35 . regarding the system setup , it is assumed throughout the paper that @xmath36 is a random variable with zero mean satisfying @xmath37 .
we have @xmath38 and @xmath39 .
the system matrices belong to given sets @xmath40 , @xmath41 , @xmath42 and @xmath43 with @xmath44 and @xmath45 for each @xmath22 .
we write @xmath46 , where @xmath47 is the probability measure ; @xmath48 is considered as an element of @xmath49 , that is , @xmath50 .
@xmath51 stands for the limiting distribution of the markov chain @xmath52 when it exists , in such a manner that @xmath53 .
also , we denote by @xmath54 $ ] , @xmath55 the transition probability matrix of the markov chain @xmath52 , so that for any @xmath56 , @xmath57 no additional assumption is made on the markov chain , yielding a rather general setup that includes periodic chains , important for the duality relation given in remark [ rema - extending - to - time - varying ] . we shall deal with linear operators @xmath58 .
we write the i - th element of @xmath59 by @xmath60 and similarly for the other operators . for each @xmath22 , we define : @xmath61
let @xmath62 be the expected value of a random variable .
we consider the _ conditioned second moment _ of @xmath63 defined by @xmath64 [ lem - def - second - moment ] consider the system @xmath7 with @xmath65 for each @xmath9 .
the conditioned second moment @xmath66 is given by @xmath67 and @xmath68 for a fixed , arbitrary @xmath69 , note that @xmath70 from , and the total probability law we obtain : @xmath71 in order to compute the right hand side of , we need the following standard markov chain property : for any function @xmath72 we have @xmath73 then by replacing @xmath74 with @xmath75 and applying the above in ( [ eq_ayuda_1 ] ) yields @xmath76 which completes the proof . [ rem_evolution_first_moment ]
let @xmath77 , @xmath78 be given by @xmath79 this variable is commonly encountered in the majority of papers dealing with ( standard ) mjls .
however , calculations similar to that in lemma [ lem - def - second - moment ] lead to @xmath80 note that the markov chain measure appears explicitly , leading to a time - varying mapping from @xmath81 to @xmath82 .
the only exception is when the markov chain is _ reversible _ , in which case the facts that @xmath83 and that the markov chain starts with the invariant measure ( by definition ) yield @xmath84 in which case @xmath85 evolves exactly as in a standard mjls .
the following notion is adapted from ( * ? ? ?
* chapter 3 ) .
we say that the system @xmath7 with @xmath65 is _ mean square stable _ ( ms - stable ) , whenever @xmath86 this is equivalent to say that the variable @xmath87 converges to zero as @xmath0 goes to infinity , leading to the following result .
the system @xmath7 with @xmath65 is ms - stable if and only if the spectral radius of @xmath88 is smaller than one .
let the output variable @xmath89 given by @xmath90 and @xmath91 the trm - jlq consists of minimizing the mean square of @xmath89 with @xmath0 stages , as usual in jump linear quadratic problems , @xmath92 regarding the information structure of the problem , we assume that @xmath93 is available to the controller , that the control is in linear state feedback form , @xmath94 where @xmath95 is the decision variable , and that one should be able to compute the sequence @xmath96 prior to the system operation , that is , @xmath97 is not a function of the observations @xmath98 , @xmath99 .
the conditioned second moment @xmath12 for the closed loop system is of much help in obtaining the solution .
the recursive formula for @xmath12 follows by a direct adaptation of lemma [ lem - def - second - moment ] , by replacing @xmath100 with its closed loop version @xmath101 [ lem - def - second - moment - controlled - version ] the conditioned second moment @xmath66 is given by @xmath102 and @xmath103 in what follows , for brevity we denote @xmath104 [ lem_cost_function ] the trm - jlq problem can be formulated as @xmath105 the mean square of the terminal cost , @xmath106 is : @xmath107 now , by a calculation similar as above leads to @xmath108 substituting and into we obtain .
let us denote the gains attaining by @xmath109 . from a dynamic programming standpoint , we introduce value functions @xmath110 by : @xmath111 and for @xmath112 , @xmath113 where @xmath114 and @xmath115 , @xmath116 , satisfies .
[ teo_optimal_gains ] define @xmath117 and @xmath118 , @xmath119 , as follows .
let @xmath120 and for each @xmath121 and @xmath22 , compute : if @xmath122 , @xmath123 else ( if @xmath124 ) , @xmath125 then , @xmath126 and @xmath127 , @xmath78 .
we apply the dynamic programming approach for the costs defined in and the system in , whose state is the variable @xmath128 .
it can be checked that @xmath129 is the adjoint operator of @xmath130 , and consequently @xmath131 let us decompose @xmath12 as follows ; let the set of states @xmath13 having zero probability of being visited at time @xmath9 be denoted by @xmath132 we write @xmath133 , where @xmath134 is such that @xmath135 for any @xmath136 , and in a similar fashion @xmath137 for @xmath138 .
we now show that the term @xmath139 is zero irrespectively of @xmath97 .
first , note that for @xmath138 we have @xmath140 .
second , for @xmath138 one can check that @xmath141 for all @xmath142 such that @xmath143 , so that @xmath144 .
this yields @xmath145 . bringing these facts together and recalling that by construction @xmath135 for all @xmath136 , we evaluate @xmath146 by substituting into we write @xmath147 by expanding some terms and after some algebra to complete the squares
, we have @xmath148 where @xmath149 as given in .
this makes clear that the minimal cost is achieved by setting @xmath150 , @xmath136 .
now , by replacing @xmath151 with @xmath152 in , @xmath153 with @xmath154 as given in , @xmath136 .
finally , by choosing @xmath155 , @xmath138 , we write @xmath156 which completes the proof .
we consider the lmmse for standard mjls as presented in @xcite .
the problem consists of finding the sequence of sets of gains @xmath157 , @xmath78 , that minimizes the covariance of the estimation error @xmath158 when the estimate is given by a luenberger observer in the form @xmath159 where @xmath160 is the output of the mjls @xmath161 and @xmath162 and @xmath163 are i.i.d .
random variables satisfying @xmath164 , @xmath165 and @xmath166 .
moreover , it is assumed that @xmath167 and @xmath168 .
we write @xmath169 , @xmath22 , so that it is the time - reverse of @xmath170 , @xmath171 .
note that we are considering the same problem as in @xcite , though our notation is slightly different : here we assume that @xmath172 is available for the filter to obtain @xmath173 and the system matrices are indexed by @xmath174 , while in the standard formulation @xmath175 are observed at time @xmath9 and the system matrices are indexed by @xmath176 .
this `` time shifting '' in @xmath52 avoids a cluttering in the duality relation . along the same line , instead of writing the filter gains as a function of the variable @xmath177 , given by the coupled riccati difference equation ( * ? ? ?
* equation 24 ) @xmath178 whenever @xmath179 and @xmath180 otherwise , in this note we use the variable @xmath181 defined in @xcite , leading to @xmath182 replacing this in the above equation , after some algebraic manipulation one obtains : @xmath183 whenever @xmath179 and @xmath184 otherwise , with initial condition @xmath185 .
the optimal gains are given for @xmath78 by @xmath186 whenever @xmath179 and @xmath187 otherwise .
the duality relations between the filtering and control problems are now evident by direct comparison between and . @xmath188 and @xmath189 are replaced with @xmath190 and @xmath191 , respectively . moreover , comparing the initial conditions of the coupled riccati difference equations , we see @xmath192 replaced with @xmath193 . also , we note that @xmath194 are equivalent to @xmath195 , with a similar relation for the gains @xmath196 and @xmath197 .
the markov chains driving the filtering and control systems are time - reversed one to each other .
[ rema - extending - to - time - varying ] time - varying parameters can be included both in standard mjls and in @xmath7 by augmenting the markov state as to describe the pair @xmath198 , @xmath199 , @xmath200 , and considering a suitable matrix @xmath201 of higher dimension @xmath202 .
although this reasoning leads to a matrix @xmath201 of high dimension , periodic and sparse , it is useful to make clear that our results are readily adaptable to plants whose matrices are in the form @xmath203 .
either by this reasoning or by re - doing all computations given in this note for time - varying plants , we obtain the following generalization of ( * ? ? ?
* table 6.1 ) .
c c + + filtering of mjls & control of @xmath7 + + + @xmath204 & @xmath205 + @xmath206 & @xmath207 + @xmath208 & @xmath209 + @xmath210 & @xmath211 + @xmath212 & @xmath213 + @xmath214&@xmath215 + @xmath216 & @xmath217 + @xmath218 & @xmath219 +
we have presented an operator theory characterization of the conditional second moment @xmath220 , an ms stability test and formulas for the optimal control of system @xmath7 .
the results have exposed some interesting relations with standard mjls .
for system @xmath7 it is fruitful to use the _ true _ conditional second moment @xmath220 whereas for standard mjls one has to resort to the variable @xmath85 given in to obtain a recursive equation similar to the ones expressed in the lemmas 3.1 and 4.1 .
moreover , these classes of systems are equivalent if and only if the markov chain is revertible , as indicated in remark 1 .
the solution of the trm - jlq problem is given in theorem 4.1 in the form of a coupled riccati equation that can be computed backwards prior to the system operation , as usual in linear quadratic problems for linear systems .
the result beautifully extends the classic duality between filtering and control into the relations expressed in table 1 . | we study a class of systems whose parameters are driven by a markov chain in reverse time .
a recursive characterization for the second moment matrix , a spectral radius test for mean square stability and the formulas for optimal control are given .
our results are determining for the question : is it possible to extend the classical duality between filtering and control of linear systems ( whose matrices are transposed in the dual problem ) by simply adding the jump variable of a markov jump linear system .
the answer is positive provided the jump process is reversed in time . |
the vela molecular ridge ( hereinafter vmr ) is a cloud complex constituted by four giant molecular clouds @xcite , hosting large regions of active star formation ( hereinafter sf ) , in which low and intermediate mass stars are forming @xcite .
its position on the galactic plane ( @xmath4 ; @xmath5 ) makes this complex a unique laboratory for studying the sf phenomenon when the environmental conditions are those typical of the galactic disk .
this is an important point to consider in extrapolating global star formation parameters to the galactic scale .
indeed , the most commonly studied sf regions are projected outside the galactic plane , in part because of the observational need to minimize the source confusion .
the drawback is that the global quantities derived for these regions could not be fully representative of the bulk of the sf in the galaxy .
our study is focused on a particular subregion of the vela cloud complex , the so - called d cloud ( * ? ? ?
* hereinafter vmr - d ) , that has already been the subject of previous papers by our group .
former studies have shown a rich phenomenology associated with this cloud , involving outflows @xcite , jets @xcite , and evidences of clustered and isolated sf @xcite .
after the co emission survey of the whole vmr carried out by @xcite , two maps of @xmath6 1 square degree region of vmr - d were obtained with higher resolution by @xcite ( @xmath7co(1 - 0 ) and @xmath8co(2 - 1 ) emission lines ) and @xcite ( mm dust continuum ) , aimed at studying the cloud morphology and properties as well as their relationships with the sf process active in this region ( see also * ? ? ?
this molecular cloud has been recently observed with both mips and irac cameras onboard the _
spitzer space telescope _ and the mips aperture photometry has already been the subject of a previous work @xcite .
in this paper we present the analysis of the _
spitzer_-irac mosaics of the vmr - d cloud , covering a region which mostly overlaps the field observed at longer wavelengths by mips and also corresponding to the sky regions already mapped in the mm spectral range and discussed in the previous papers .
these new irac observations allowed us to compile a four - band photometric catalog of point sources that , complemented by the information previously obtained at longer wavelengths , constitutes a unique data bank for studying the young stellar objects ( ysos ) in this cloud . with this as our goal we have also revisited the photometric analysis of the mips 24 @xmath1 m and 70 @xmath1 m mosaics to obtain more accurate psf photometry .
the inclusion of the longer wavelength photometry is motivated by our aim to determine , as accurately as possible , the spectral slopes of the sources , infer their evolutionary stages , and obtain a reliable statistical description of the yso population .
a particularly valuable characteristic of our irac observations is that the data were collected by splitting the total integration time into two observing epochs 4.5 months apart , thereby enabling a first exploration of source variability in vmr - d .
indeed we found a number of sources that clearly showed brightness variations ; they constitute a significant subsample deserving special attention , and are discussed in a separate paper @xcite .
this paper is organized as follows : after a brief description of the observations in [ obse ] , we give an account of the procedure followed for source extraction and photometry for both irac and mips data in [ photom ] . in the same section
we also discuss the source selection criteria as well as the completeness and reliability .
the final vmr - d point source catalog is presented in [ thepsc ] . in [ discussion ]
we focus our attention on a subsample of 8796 sources ( out of a total of 170299 ) that have been detected in all the four irac bands and for which good quality photometry was obtained . in the following
we shall refer to this subsample as the `` working sample '' . by means of color - color and color - magnitude diagrams , we select the candidate ysos obtaining a census of the yso content of the vmr - d region . in the same section a brief account of the clustering properties
our conclusions are summarized in [ concl ] .
the observations were carried out with the infrared array camera ( irac ; * ? ? ? * ) onboard the _
spitzer space telescope _
the data were obtained as part of the spitzer / irac cycle 3 gto ( pid 30335 , pi g. fazio ) .
the galactic region observed was delimited in longitude by @xmath9 and in latitude by @xmath10 , covering about 1.2 square degrees .
this field corresponds to a significant portion of the vmr - d cloud that has been mapped in four spectral bands , corresponding to the four irac camera detectors observing at the wavelengths @xmath0 = 3.6 , 4.5 , 5.8 , and 8.0 @xmath1 m , respectively .
the effective integration time per pixel , @xmath11 s in each band , has been accumulated by splitting the observations in two epochs separated by 4.5 months , a choice mainly dictated by the desire to monitor the observed region for possible variability of the sources .
the same splitting of the observation time into two epochs has also been beneficial for a better cleaning of artifacts from the final mosaics .
the irac data were obtained by mapping the whole region , in each epoch , with 12@xmath1214 tiles , each observed in two dither positions .
the observation was performed in irac 12 s hdr mode , consisting of two consecutive integrations of 0.4 and 10.4 s exposure times .
the hdr mode was chosen to limit the saturation effects around the bright regions , a problem which is clearly present in the long exposure images in the neighborhood of the brightest sources .
the data were reduced using the iracproc package ( schuster et al .
2006 ) to combine the individual tiles into a flux calibrated mosaic on a pixel grid of 0.8627" / pix .
iracproc is a wrapper for the spitzer science center mosaic software mopex and provides enhanced outlier ( cosmic rays ) rejection .
thus for each spectral band we obtained a total of four mosaics , one long and one short exposure for each of the two epochs .
the mosaics of each epoch were subsequently coadded to obtain two final mosaics in each irac band , one with the long ( @xmath13 s ) and the other with the short ( @xmath14 s ) total exposure time . however , due to the different orientation constraints in the two observing epochs , and to the fact that the four irac cameras do not see exactly the same fields simultaneously , the images obtained in the two epochs do not fully overlap . as a consequence , in the coadded mosaic the region corresponding to the full exposure time
is limited by the spatial overlap of the mosaics obtained in the two observing epochs .
the final mosaics therefore include a central region , observed for the total integration time @xmath15 s , as well as adjacent surrounding regions corresponding to the non - overlapping fields that have been imaged only in one of the two epochs for a total exposure time @xmath16 s. the geometry of the mapping and the different parts covered are shown in figure [ fig_mosaic ] . in the following we shall consider the whole region observed in the four bands , delimited in figure [ fig_mosaic ] by the yellow contour , to produce the corresponding point source catalog , recalling that the area that has been imaged in both epochs , included in the green dashed contour , is clearly the region observed with a slightly higher s / n ratio . in producing this catalog
we also take advantage of our previous observations of the vmr - d region to assess if a source falls on the mapped co - emitting cloud or not .
this evaluation is important and interesting in this context because we do not have adequate off - source observations of control fields , and the archival @xmath7co observations allow us to evaluate , at least to a first approximation , the possible contamination of the on - cloud yso population by off - cloud background / foreground objects .
operatively , we shall consider a source as projected on the molecular cloud , i.e. on - cloud , if it is contained within the @xmath7co contour line corresponding to the 5 k km s@xmath17 level in the integrated intensity map of @xcite which is shown superimposed on the irac 8 @xmath1 m mosaic in figure [ vela_yso+co ] .
clearly , this is only a necessary condition for a source to be considered for additional constraints aimed to discriminate bona fide on - cloud ysos from galactic / extragalactic contaminants .
to obtain the photometric information on the point sources detected in our mosaics , we adopted the daophot @xcite photometric package .
this offers a consolidated suite of algorithms aimed at detecting all the possible point - like objects whose signal is larger than a predefined threshold .
the photometry was derived on our final ( two epochs combined ) mosaics .
single epoch photometry was however also measured for some sources , in order to assess the reliability of our final catalog . in this section
we shall describe the methods used for source detection and photometry , as well as the selection criteria adopted for compiling the vmr - d point - source catalog .
completeness and reliability is also discussed and finally the mips mosaics are reconsidered to obtain a psf photometry to be used as a complement to the irac catalog . due to the large spatial extent of our mosaics and to the presence of extended bright nebulosity , distinct ,
local background intensity gradients are invariably present , so that it is not advisable to simply scan the original image with a given sensitivity threshold .
the output of the detection algorithms in this case would be affected by a spatially variable sensitivity , depending on the different levels of the local background , a problem that is made even worse when strong gradients are present . to ameliorate this difficulty we first evaluated the point spread function of our mosaics by using relatively isolated , bright , unsaturated stars , and then performed psf point - source subtractions to remove , to first order , the point - like objects . in this way we obtained images practically devoid of strong point sources , even if some residuals are clearly present .
we used these images to estimate the background level across the whole field , a step that was done in each observed band by evaluating the sky value around each pixel in a square box of 51-pixel in size .
this was done by using the daophot - sky procedure that estimates the modal value after clipping the outliers .
note that the relatively large size adopted for the scanning box is justified by the need to include a sufficient number of sky pixels even in crowded images , as is the case particularly for the first two irac bands centered at @xmath0 = 3.6 @xmath1 m and @xmath0=4.5@xmath2 m , respectively .
once the background in each band was obtained and subsequently subtracted from the original mosaics , we were left with four background - subtracted images in which the large scale gradients were strongly smoothed while the small scale noise was preserved . at this point ,
the compatibility of the noise with poissonian statistics is recovered by adding the original mean sky value back onto the background subtracted images that were then used in the photometry extractions .
source detection was then performed by scanning these final images with a star finding procedure which is based on a convolution with a wavelet - like filter whose spatial extent has been tuned to match the size of the point - like objects in the image .
the signal produced by this convolution is then analyzed at different detection thresholds that , in our case , were taken at 5 @xmath3 and 10 @xmath3 , respectively .
once we had the list of detections from the star finding algorithm , we used aperture photometry to obtain a first evaluation of the source magnitudes . although this is a relatively inaccurate method when crowding or background gradients are present in the image , it is nevertheless useful for a first estimate of source brightness , an information that was used later as a starting point for more accurate photometry .
the final magnitude values were indeed obtained by means of an iterative psf - fitting procedure that , in the implementation adopted by daophot , is based on an estimate of the instrumental psf obtained from stellar images appearing relatively isolated in the frame . however , because the psf is undersampled by the irac cameras , the corresponding point response function ( prf ) that we obtain from our mosaics can not be completely accurately determined ; this produces a corresponding uncertainty in the subsequent fit to the stellar images . in spite of this difficulty , the photometry resulting from this method can be considered more accurate than the aperture photometry because the differences in fitting the prf to the stellar images remain approximately balanced between positive and negative residuals . in this way
the total flux is substantially preserved , even in presence of residuals , although at the cost of an increased uncertainty in the fitting parameters ( see also @xmath18 [ completeness ] below on reliability ) .
the prf we used is actually a combination of an analytic function ( in our case a lorentzian showed better results in minimizing the residuals ) and a look - up table whose aim is to compensate for systematic differences between the pure analytic function and the actual stellar images .
an example of the residuals remaining after subtraction of the modeled sources from the irac mosaics is shown in figure [ fig_subtracted ] , which illustrates this point for the two irac mosaics obtained at 3.6 @xmath1 m ( upper panels ) and the 8.0 @xmath1 m ( lower panels ) , respectively .
note , however , that the knowledge of the prf also allowed us to take into account , and then correct for , the contribution to the flux of a source from wings in the brightness distribution of its neighboring sources .
this advantage is fully exploited in crowded regions because the photometric values are obtained only after convergence of an iterative procedure that fits simultaneously groups of nearby stars instead of single objects .
we selected the sources to be included in our catalog by applying to the photometric lists obtained from the long exposure mosaics a sequence of logical filters based on a reasonable use of the quality parameters associated with the photometric procedures .
our first choice was to consider as `` good photometry sources '' only those detected above the 5 @xmath3 threshold of the background noise . to easily distinguish faint and bright sources in the catalog
, we also flagged those detected above the 10 @xmath3 threshold .
the daophot procedure for obtaining the point source photometry , besides evaluating the brightness , also computes two additional quantities : @xmath19 and @xmath20 , related to the goodness of the fit and to the source sharpness , respectively . here
@xmath19 is evaluated as the ratio between the observed scatter in the fitting residuals and that expected on the basis of the pure image noise , and is a measure of the departure of the source image from the model prf .
the sharpness parameter @xmath20 , instead , is computed as a difference between the square of the size , in pixels , of a given source and the same quantity computed for the fitting prf so that it can be used , within certain limits , as an indicator of the source angular extent . at
first we constrained these two parameters to be @xmath21 and @xmath22 to represent _ bona fide _ point - like sources ; we subsequently applied a further criterion to the flux uncertainty to select only sources with a flux relative error @xmath23 .
as a third step we also considered the number of iterations needed for the prf fitting algorithm to converge on a given source : to avoid the worst cases , mainly corresponding to faint sources in crowded fields , we limited selections to those sources requiring fewer than eight iterations . finally , because the prf fitting procedure also implies a recentering of the source position , we filtered out all those sources in which the corresponding best fit centroid was displaced from the original position as evaluated in the previous detection phase by more than 1 . despite all our efforts to select genuine point sources , the photometric lists still included many detections that , on visual inspection , are actually artifacts in the vicinity of the brightest sources .
this problem affects in particular the mosaics at @xmath24 3.6 @xmath1 m and @xmath24 4.5 @xmath1 m , and is related to the so called _ muxbleed _ defect .
these artifacts are recognized because they appear as a sequence of equally spaced pseudo - sources aligned along the same row containing the true bright source . because their appearance is very much like that of a train of point sources , these pseudo - sources escaped all our other selection filters so that it was necessary to eliminate them by eye . in this way
more than 1000 false sources were discarded . besides these false sources the previous selection criteria , applied to the long exposure photometry , discarded an additional , but relatively small , number of sources ( @xmath25 1000 ) . these are mainly located in regions affected by crowding or saturation , the latter case occurring especially in the neighborhood of the brightest objects . to compensate for this we also considered the short exposure mosaics , obtaining the corresponding photometry with the same procedure adopted for the long exposures .
the short exposure photometry was used when , in assembling the catalog , the long exposure photometry of a given source was rejected due to one of the adopted selection criteria , as opposed to pixel replacement in the mosaic or a hard cutoff between the short and long frames . in these cases we further checked to see
if alternative , short exposure photometry existed that met all the catalog constraints .
if it did , we included the source in the catalog ; in this way we recovered a total of 276 sources .
we also reexamined an additional 77 rejected sources , most of which ( 68 cases ) are associated with 24 @xmath1 m mips sources , that escaped irac detection because they are either so bright they suffered saturation effects or were embedded in a bright nebulosity . for most of these sources
we used the short exposure mosaics to obtain the photometry even though it is more uncertain .
these objects were included in the catalog to produce a list as complete as possible , even though they have been flagged as `` bad photometry '' and are not considered further in the analysis of the yso population .
however the reader can use these flux densities with their uncetainties , but at the same time should be aware that the corresponding quality flags do not satisfy all the constraints previously discussed for the automatic source selection . in conclusion , we obtained four lists of sources , one for each irac band , that have been spatially delimited to include only objects lying in the sky region surveyed in all the four irac bands as delimited by the yellow contour line in figure [ fig_mosaic ] . an upper limit to the completeness of our catalog
is clearly set by the sensitivity detection threshold adopted in the detection phase that , we recall , was set to 5 @xmath3 .
this threshold corresponds to different limiting flux densities because the background level changes with the wavelength .
we evaluated the limiting flux densities in each spectral band by calculating the average value of the pixel fluctuation in the background images obtained in the course of the photometric procedure .
the same limiting flux densities are independently derived from the magnitude plots of figure [ hist_mag ] , showing the distribution of the catalog point sources as a function of their magnitude in the four irac bands . from these plots
it is apparent that the logarithm of the number of sources increases proportionally to the magnitude , until the linear behaviour is lost and a fall in the number of sources is clearly seen .
the vertical line drawn in each panel denotes the largest magnitude at which the catalog can be considered as complete , a value that is clearly lower than the limiting magnitude .
these values are summarized in table [ tab_lim_fluxes ] along with their corresponding flux densities . as far as the reliability is concerned
, we can exploit the fact that the vmr - d field was observed twice , allowing a check on the source detections .
the criterion we adopt for this evaluation is to consider as reliable detections in a given irac band two possible cases : i ) sources detected in both observing epochs , and ii ) sources detected only once but showing at least another detection in an adjacent band . on the other hand we consider as less reliable detections case iii ) when sources were detected only once in a given band and with no detection in the adjacent bands . to ensure that the sources we consider in estimating the reliability were observed twice , we restricted this analysis to the area delimited by the overlap of the irac mosaics from both epochs ( dashed green line in fig .
[ fig_mosaic ] ) .
applying to these sources the previous criteria , we obtain the distributions shown in figure [ hist_mag ] as a dotted ( reliable ) and dashed ( unreliable ) line . from these plots
we estimate the reliability as a function of the magnitude by ratioing the number of sources satisfying the criteria i ) and ii ) with the totality of the catalog sources i)+ii)+iii ) in a given band and magnitude interval . in this way
we derive a reliability estimate that is always larger than 99% in the whole magnitude range .
all the sources observed in both epochs show flux densities that are compatible within their uncertainities ; the very few sources ( around 0.2% ) showing a significant photometric variability have been discussed in @xcite .
as mentioned in the introduction , we also reconsidered the mips observations at 24 @xmath1 m and 70 @xmath1 m of the vmr - d region covering a slightly larger field of approximately 1.8 square degrees ( see fig .
[ fig_mosaic ] ) .
these observations have already been presented by @xcite who reported the aperture photometry for 849 and 61 sources in the 24 @xmath1 m and 70 @xmath1 m band , respectively . however , because aperture photometry tends to be less accurate in the presence of large background gradients such as are present in the mips vmr - d mosaics , in @xcite the mips sources considered were limited to those in regions where reliable aperture photometry was possible . because psf fitting photometry is a more accurate tool , we re - analyzed the regions in the two mips mosaics of vmr - d to obtain a revised list of sources with the corresponding psf photometry .
our aim was to obtain more accurate seds and then a more reliable determination of the spectral slopes characterizing the yso population .
the analysis of the mips mosaics was done with the same procedures described for the irac images .
first , we minimized the effects of the background variations by obtaining a rough subtraction of the point sources from the original image .
after this first step eliminated most of the point source signal , the mosaics were scanned with a square box of 21 pixels on a side to produce an estimated background image .
this in turn was subtracted from the original mosaic producing a flattening of the large scale gradients , thus allowing a subsequent point source detection across the field with a much more uniform sensitivity . by using a 5 @xmath3 detection criterion we obtained two preliminary lists of sources at 24 @xmath1 m and 70 @xmath1 m , respectively , constrained by selection criteria similar to those adopted for the catalog of the irac sources ( see [ auto_selected ] ) .
a comparison between the aperture and psf photometry methods shows that at 24 @xmath1 m the aperture photometry tends to systematically overestimate the flux by up to @xmath2550% with respect to the psf photometry .
similarly , the flux differences at 70 @xmath1 m are of the same order ( @xmath26 50% ) , but they appear more erratic and without a systematic trend . in the end , we obtained two more complete lists of sources measured with psf photometry , containing 1347 and 63 sources at 24 @xmath1 m and 70 @xmath1 m respectively , and we used them to supplement the irac catalog .
note that among the 24 @xmath1 m sources we include for completeness six saturated objects , corresponding to irs 16 , 17 , 18 , 19 , 20 , and 21 ( nomenclature from * ? ? ?
in the mips catalog these correspond to the entries id=146 , 300 , 813 , 814 , 991 , 1018 , respectively , that had been flagged with negative 24 @xmath1 m flux densities , whose absolute value corresponds to the saturation limit ( 4 jy ) .
we adopted angular separation as the criterion for associating the 24 @xmath1 m and 70 @xmath1 m sources , requiring that the corresponding centroids be within 20 ; in this way we produced a unique , merged set of sources with mips photometry .
the merged mips catalog is available in electronic form ( vmr - d_mips_catalog.txt ) , and it also contains information on the possible associations of the millimeter sources as reported in previous studies .
these associations are based on positional coincidences of the mips source centroid being within the beamsize of the mm observations .
a short excerpt from the mips catalog is printed in table [ tab_mips_cat ] .
the histogram of the magnitudes is presented in figure [ hist_mips ] , and it shows that the completeness flux densities are essentially the same as those quoted in @xcite , namely @xmath27 mjy at 24 @xmath1 m and @xmath28 mjy at 70 @xmath1 m , respectively .
some statistics of the mips sources are presented in table [ tab_stat_mips ] . given this new mips photometry , a re - examination of the color - magnitude diagrams in ( * ? ? ?
* their fig . 7 and 8) is in order . in figure [ k_24 + 24_70 ]
we present the resulting k vs. k-[24 ] and [ 24 ] vs. [ 24]-[70 ] diagrams while in table [ tab_class_mips - irac ] , 2nd and 3rd columns , we report the subdivision in the different classes based on the spectral indexes derived by using the 2mass k - band magnitudes as a complement of the present mips photometry .
this classification scheme is the same used in @xcite and is based on the correspondence of the colour k-[24 ] with the spectral slope of the different classes adopted by @xcite .
a comparison of the present results with those in @xcite , based on aperture photometry , is possible .
we renormalized the values in their table 5 to reflect the changes in the total number of ysos ; because the measurement of the flux density now has changed by up to 50% , the numbers of sources in each spectral class changes as well .
we note that : * the number of mips sources detected now in the 24 @xmath1 m band is increased while at 70 @xmath1 m it remains practically the same ; * the percentage of on - cloud class i sources is now @xmath25 10% , decreased from the previously estimated value of @xmath25 30% ; * the percentage of the on - cloud flat - spectrum sources is approximately the same , @xmath25 26% , as compared with the previous estimate of @xmath25 29% ; * the percentage of on - cloud class ii objects is @xmath25 62% , an increase with respect to the previous estimate of @xmath25 37% , while class iii sources are now @xmath25 2% with respect to the previous value of @xmath25 4% .
furthermore , the more accurate modeling of the point sources allows us now to obtain more reliable correspondences between the mips , irac , and 2mass objects , alleviating in this way the problem of multiple associations . in the light of the significantly lower percentage
we find here for class i spectrum sources , we conclude that our previous suggestion for an unusual excess of young objects in vmr - d does not hold up based on psf photometry , and in all future discussions of mips sources we shall refer to our current catalog .
once photometry was obtained we proceeded to the positional association of the sources detected in the different spectral bands . in this phase of band merging objects detected at different irac wavelengths
were considered to be associated if the relative distance of their centroids was less than 1 . for those cases
in which , when comparing two bands , more than one source is found within 1 , we considered only the nearest source to be associated .
the final vmr - d catalog was produced by including all of the sources detected in _ at least _
one irac band which fulfill all the constraints reported in [ sourceselection ] .
an exception to this rule are the 77 sources , mainly corresponding to very bright objects , whose addition we justified in [ additional_sources ] . to obtain a coherent list of sources , we restricted the final catalog to the field observed in all the four irac bands , namely the region delimited by the yellow contour line in figure [ fig_mosaic ] .
obviously , by construction such a catalog contains many sources lacking good photometry in some irac bands ; only 8796 sources out of the full 170,299 have accurate photometry in all four irac bands .
however , because for some purposes it is useful to have flux densisites in all bands , we choose to give a flux density value in any case .
this was done by reconsidering less accurate photometry previously discarded in the source extraction process ( see [ sourceselection ] ) or , in absence of this photometry , obtaining 5 @xmath3 upper limits for the source position . consequently those missing flux densities in the catalog were recovered from the values that had been previously rejected due to the imposed constraints when the detection fall within a radius of 1 from the source position .
note , however , that all these band filling flux densities have been flagged as `` bad photometry '' in the final lists , in the same way the 5 @xmath3 upper limits are reported in the catalog for those cases without a specific detection within 1 .
in addition , to allow a more accurate determination of the spectral slopes and a more secure classification of the ysos , the final catalog also considers the spatial coincidence between irac and mips 24 @xmath1 m sources , and includes the mips photometry when the distance between centroids is within 6 .
a larger distance , up to 20 , has been adopted for mips sources detected only at 70 @xmath1 m .
these relatively large distances are close to the diffraction radius at 24 @xmath1 m and 70 @xmath1 m , respectively , and have been adopted to include all the potentially associated sources . of course , because the probability of the irac - mips source association decreases with the centroid distance , one must pay increasing attention when the angular distance is larger than 3and 10at 24@xmath1 m and 70@xmath1 m , respectively .
note however that 70% of cases have been found within 3 . with this criterion a total of 870 mips 24 @xmath1 m sources have been associated with irac sources ; out of these , 689 are associated to a single irac source , 153 are double associations , 27 are triple , and one is associated with four irac sources . to allow the reader to evaluate these associations
we also report in the catalog the distance of the centroids for each association .
the published version of the catalog , available in electronic form ( vmr - d_irac_catalog.txt ) , summarizes the photometry arranging the information it in the following columns : * column ( 1 ) : source i d .
sources are sorted in ascending order of r.a . *
columns(2 - 3 ) : equatorial coordinates ( j2000 ) . * columns(4 - 5 ) : galactic coordinates .
* columns(6 - 7 ) : irac flux at 3.6 @xmath1 m and its uncertainty .
when the flux reported is an upper limit , the uncertainty has been put to zero .
* column(8 ) : quality flag , assigned as follows : the `` good photometry '' data are flagged with `` h '' ( high ) if they refer to a more than 10 @xmath3 detection , and with `` l '' ( low ) if the source is revealed between 5 and 10 @xmath3 , respectively ; in addition to this , a flag `` 2 '' or `` 1 '' indicates whether the source lies in the sky area covered in both observing epochs , or only in one of them , respectively .
the `` bad photometry '' data ( see above ) are flagged with `` b '' , and upper limits with `` u '' .
* columns(9 - 17 ) : same as for columns 6 - 8 , but for 4.5 @xmath1 m , 5.8 @xmath1 m , and 8.0 @xmath1 m band , respectively . *
column(18 ) : mips catalog i d number of the associated mips source : this is reported if its centroid is within a distance of 6@xmath29 from the irac source .
mips sources detected only at 70 @xmath1 m are reported when the centroid is within 20 .
notice that , due to the different beam sizes , the association of more than one irac source with the same mips source is allowed . *
column(19 ) : distance in arcsec of the associated mips source centroid . * columns(20 - 22 ) : mips flux density at 24 @xmath1 m , flux uncertainty , and quality flag similar to that used for irac flux densities but with good photometry simply flagged `` g '' .
the symbol nan reported for the flux density denotes a position falling outside the mips mosaic . *
column(23 - 25 ) : the same as for column 20 - 22 but for the mips 70 @xmath1 m band .
note that for mips sources showing only the 70 @xmath1 m detection ( 10 cases ) the association is reported when the irac sources ( 51 cases ) are within 20 from the mips source .
* columns(26 - 27 ) : a flag indicating the degree of association with @xmath7co(1 - 0 ) and @xmath8co(2 - 1 ) line emission , respectively , as mapped in @xcite .
`` c '' indicates that the irac source lies on a map pixel assigned to one or more gas clumps ; `` a '' indicates that the irac source is not associated to a clump , but lies inside the integrated intensity contour corresponding to the 5 k km s@xmath17 , adopted here as a criterion for spatial association with the gas ; `` o '' indicates a source lying inside the area observed in the given line , but not associated with the gas according to the two previous criteria ; finally , `` n '' means that the irac source is outside the sky area covered by observations in the given line . *
colum(28 ) : name of the 1.2 mm dust core , taken from the lists of @xcite and reported here when the irac source falls within the quoted core radius .
examples of the catalog entries are given in table [ tab_irac_psc ] and some statistics of the catalog content is summarized in table [ tab_stat_global ] and table [ tab_stat_perband ] .
ysos in sf regions can be identified by the presence of an infrared excess which is produced by extended circumstellar disks / envelopes surrounding these objects , and that participate in their pre - main sequence ( pms ) evolution toward the main sequence . for classification purposes it is customary to use the @xmath30 plot to derive the spectral slope , @xmath31 , that is generally evaluated in the @xmath32 m range .
this parameter is taken to be dependent on the evolutionary phase of the emitting source and its value is generally used to define four classes @xcite : class i for @xmath33 , `` flat - spectrum '' for @xmath34 , class ii for @xmath35 , and class iii for @xmath36 . a further class ,
the class 0 , has been also introduced to include sources that , while escaping any detection at @xmath37 m , show relatively strong submillimeter emission @xcite and should then correspond to very early and cold objects .
this relatively simple classification scheme has been however criticized because of its sensitivity to inclination effects to the line of sight ( see , e.g. , * ? ? ?
* ; * ? ? ?
alternative classification schemes based on the bolometric temperature @xcite or the bolometric luminosity @xcite have also been proposed in an attempt to minimize the influence of the system inclination to the line of sight .
however these alternative schemes have themselves been found to be sensitive to the system geometry , so that we have decided to adopt the most popular classification scheme based on the spectral slope .
this choice also allows us a more direct comparison with a series of studies carried out with _
spitzer _ on galactic star forming regions in the framework of the legacy project `` from molecular cores to planet - forming disks '' ( c2d , * ? ? ?
the reader interested in a more detailed discussion on the effects of adopting other classification schemes should refer to ( * ? ? ?
* hereinafter e09 ) , who also discussed the role of the extinction corrections . in the following analysis
we focus our attention on the 8796 , out of a total of 170,299 catalog sources , for which good photometry was obtained in all the four irac bands ; i.e. the working sample already defined in [ introduction ] .
in addition to this database we also consider , when available , the @xmath38-band 2mass photometry and the 24 @xmath1 m mips photometry to allow a more accurate estimate of the spectral index in the 2 24 @xmath1 m spectral interval .
our first step was the determination of this spectral index @xmath31 from the linear fit of the observed @xmath39 points for all the working sample sources .
the distributions of the @xmath31 values obtained are shown in the two panels of figure [ hist_alpha ] , representing the spectral indexes for on - cloud ( upper ) and off - cloud ( lower ) sources , respectively .
in the two panels the histograms are drawn separately for the objects in the working sample ( solid line ) and for the fraction of the candidate ysos ( dashed line ) that have been subsequently selected according to the criteria we shall discuss in the next section .
we note that the number of objects falling within the on- and off - cloud region are quite similar ( 4484 against 4312 ) .
such a fact derives from the similarity of both areas ( 0.57 against 0.60 squared deg ) and indicates that the overall population of both regions is dominated by photospheres and background / foreground objects , while the yso population possibly represents a negligible contribution .
this latter is a common feature already found in surveys of other star forming regions ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
as a consequence , stringent criteria have to be exploited to select the candidate ysos .
the ability of the irac colors to identify ysos has been studied by @xcite in the serpens cloud and by @xcite in a large number of yso clusters , and we use parts of both approaches here , augmented by criteria identifying agb contaminants from @xcite . after considering the locus of the normal reddened stars and the non - negligible extragalactic contamination , these authors suggested two statistical criteria to select candidate ysos in the [ 8 ] vs ( [ 4.5]-[8 ] ) col - mag diagram , namely : i ) ( [ 4.5]-[8 ] ) @xmath40 0.5 and ii ) [ 8 ] @xmath41 14-([4.5]-[8 ] ) .
this first criterion excludes the stellar component while the second , designed to statistically cut the 95% of the extragalactic objects , has been tuned by studying the extragalactic component observed with irac in the elais n1 swire region . applying these criteria to the vmr - d cloud we select 595 yso candidates on cloud and 244 off - cloud , respectively .
following @xcite , to include other potential colder candidates that could have been escaped in this diagram , we also considered further diagrams based on longer wavelength photometry , i.e. the [ 24 ] vs ( [ 8.0]-[24 ] ) and [ 24 ] vs ( [ 4.5]-[8.0 ] ) diagrams . adopting the appropriate selection criteria we included 25 more sources on and 17 off - cloud .
the yso candidates selected are shown in figure [ col_mag_col_on ] ( on - cloud ) and [ col_mag_col_off ] ( off - cloud ) as open circles and grey crosses , representing objects showing class i / flat and class ii / iii seds , respectively
. also indicated on these plots are the original selection criteria from the @xcite study of the serpens cloud .
however , note that the serpens cloud ( 260 pc ) is much closer than the vmr - d cloud ( 700 pc , see * ? ? ?
* ) . to reflect this ,
we have also indicated in the figure the region into which a candidate yso in serpens would move if displaced to the vmr - d distance .
the region between the two inclined parallel lines can then be regarded as potentially containing both candidate ysos and extragalactic contaminants meaning that objects in this region should be more accurately studied to ascertain their nature . here
we conservatively consider only sources delimited by the original criteria represented by the dashed lines in figure [ col_mag_col_on ] and [ col_mag_col_off ] . in the same figure the two color ( [ 3.6]-[4.5 ] )
vs ( [ 5.8]-[8.0 ] ) diagram is also shown in the upper right panel with the same meaning of the symbols , and with a square box that is drawn to show the approximate locus of the class ii sources according to @xcite .
the irregular area delimited by the dot - dashed line is the locus of the agb stars in this diagram , according to @xcite . for comparison , in a separate figure [ col_mag_col_off ]
, we show the same diagrams drawn for the off - cloud sources . with the aim to obtain a list of candidate ysos as clean as possible from contaminants we also adopted further criteria recently discussed and summarized by @xcite .
these have been derived for identifying as likely contaminants those sources satisfying a series of constraints flagging star - forming galaxies and agn , as well as shock excited knots that can mimic point sources in star forming environments .
these are based on color - color diagrams as [ 4.5]-[5.8 ] vs [ 5.8]-[8.0 ] , [ 3.6]-[5.8 ] vs [ 4.5]-[8.0 ] , and [ 3.6]-[4.5 ] vs [ 4.5]-[5.8 ] , as well as on the color - magnitude [ 4.5 ] vs [ 4.5]-[8.0 ] diagram . in figure [ col_mag_col_onexclu ]
we present these diagrams ( displayed for simplicity only on - cloud ) for the sources emerging from the previous selection step that are reported with crosses and circles in figure [ col_mag_col_on ] . considering the constraints in @xcite we flagged 64 sources , 42 on ( shown as open squares ) and 22 off - cloud , respectively , that have then been eliminated from our list of candidate ysos . as a further check we also applied to this list the irac based criteria of @xcite for finding agb contaminants , identifying in this way 115 sources on - cloud
( 112 showing a class iii sed ) and 65 off - cloud ( 62 showing a class iii sed ) . also these sources have been eliminated from the final list . at the end of this process
we remain with 637 irac selected candidate yso sources , 463 on and 174 off - cloud that we assign to a specific class simply considering their spectral slope as derived considering all the available data in the 2 24 @xmath1 m spectral interval .
these sources have been used to draw the dash - dotted histogram in figure [ hist_alpha ] .
the corresponding distribution in classes of these candidate ysos is shown in 4th and 5th column of table [ tab_class_mips - irac ] .
the analysis presented above allowed us to classify the vmr - d young population and it is interesting to compare our results to those obtained by _
spitzer _ for other sf regions .
these have been summarized and analyzed in a consistent way by and we adopted it for the vmr - d comparison . however , because our method for selecting the ysos is not exactly the same as that of , before to compare our results it is desirable to evaluate the effects produced by applying our pipeline at least to one of the five clouds considered by .
the serpens cloud is probably the most appropriate for this check : it is projected near the galactic plane and has been scrutinized with accurate spectroscopic follow - up aimed to identify agb contaminants @xcite . for this cloud
we do not attempted to select candidate ysos out of the full catalog , delivered by the c2d legacy project and made available at the _ spitzer _ science center , because our selection criteria are similar to those used to obtain the c2d yso list .
the difference in our approach is mainly in the identification of the contaminants so that we used the serpens yso catalog to analyze with our procedure the 262 sources listed . applying our pipeline , based on the @xcite criteria , to this list
we exclude a total of 14 sources .
none of these is in the list of the 78 candidate ysos scrutinized by @xcite and their distribution in classes is : two sources in class i , two are flat - spectrum , eight in class ii , and two in class iii . despite we exclude these 14 sources out of the 262 in the serpens yso catalog ,
we note that after their subtraction from the numbers quoted in and reported in table [ tab_stat_yso ] , the relative distribution in classes remains practically the same and also the star formation rate is not significantly affected . in the limit of this indication
we can compare in table [ tab_stat_yso ] the results obtained for vmr - d with those provided by , noting that in the numbers quoted for the ysos in serpens the sources identified as agb by @xcite have been already subtracted . in this table the rows 13
consider the sources selected by means of the @xcite criteria and shown in fig .
[ col_mag_col_on ] and [ col_mag_col_off ] . the subsequent rows
46 report the results obtained after filtering the sources in the first three rows with more stringent criteria for eliminating extragalactic and galactic contaminants @xcite as illustrated in fig . [ col_mag_col_onexclu ] .
further three rows , namely 79 , are used to present the results obtained after applying to the first three rows the correction for contaminants suggested by @xcite on the basis of their spectroscopic investigations of the serpens yso population selected with the @xcite criteria . indeed ,
because serpens is a sf region quite close to galactic plane ( @xmath42 5@xmath43 ) , it is reasonable to assume that its level of contamination of agb and giant stars is comparable to that of vmr - d .
the sample of ysos in serpens is contaminated by background giants to a level of about 25% so that , by applying the same correction to our vmr - d on and off - cloud sample , we obtain the statistics given in table [ tab_stat_yso ] , lines 79
. note that in this case we do not report the distribution in classes because the 25% contamination is referred to the total number of ysos .
in the subsequent lines 1014 we also report for comparison the values of the corresponding quantities quoted by for chaii , lupus , perseus , serpens , and ophiucus .
in each line we give the investigated area , the adopted cloud distance , the total number of ysos found , the surface density per solid angle and per square parsec , the number of objects of type class i , flat - spectrum , class ii , and class iii , respectively , and finally the derived star formation rate .
first of all we recall that , differently from these other sf regions , vmr - d is located in the galactic plane and it is also from 2 to 5 times farther away . as already mentioned , the latter circumstance imposes a larger value for the limiting luminosity ( typically 0.1 l@xmath44 instead of , e.g. , 0.004 l@xmath44 for the taurus ) resulting in significant contamination from stars ; conversely , it offers the opportunity to study the sf process where the mass reservoir is the largest available . as a rough first approximation
, we consider the off - cloud statistics as representative of the average population of the galactic plane and , as such , it could be subtracted from the on - cloud statistics after weighting for the different areas , in this way estimating the uncontaminated young population of vmr - d . by doing this
we obtain the lines 3 , 6 , and 9 , of table [ tab_stat_yso ] , with the caution that the off - cloud region could be active in sf as suggested by the the histograms in figure [ hist_alpha ] , albeit at a lower level .
because of this and of the larger luminosity limit , the numbers reported here for the ysos should actually be considered lower limits .
however , taken at their face value , they suggest an excess of class iii ( @xmath2522% ) and a deficit of class ii objects ( @xmath2548% ) with respect to the other sf regions . the cause of the apparent excess of class iii might be either evolution - the vela cloud is older than these others - or due to uncorrected contamination of agb stars . considering the statistics corrected for extragalactic and galactic contaminants , reported in the 4th and 6th lines , make the distribution in classes of the ysos in vmr - d resemble more that seen in serpens and ophiuchus rather than in other sf regions .
differences actually exist in the class ii / iii relative ratios , but they probably stem from an incomplete correction since vmr - d is more embedded in the plane than is serpens .
if we try to account for such discrepancies , we can reasonably rule out evolutionary effects since the relatively large fraction of class iii sources in vmr - d does not reflect a particularly small fraction of class i and flat - spectrum sources that instead appear relatively numerous .
moreover , if we calculate the sf rate per square parsec , by assuming a mean mass of 1 @xmath45 , a fraction of binary of 0.5 and a period of 2 myr for sf , we obtain for the vmr - d sf rate @xmath6 2.54.1 m@xmath44myr@xmath17pc@xmath46 , a range of values that is more consistent with that of the younger clouds like serpens , ophiucus , and perseus , than it is for the smaller corresponding values obtained for lupus and chamaleon .
it is also noteworthy that adopting the corrections suggested by @xcite we obtain ( tab .
[ tab_stat_yso ] , lines 68 ) very similar results to those obtained after considering the set of criteria discussed in [ source_selection ] to exclude likely contaminants ( lines 46 ) .
this adds some confidence in the present results
. given these observational indications it seems plausible to consider the vmr - d age as being in the range 12 myr , in good agreement with the dynamical time estimate of 1.5 myr derived by @xcite for vmr - d and comparable to the estimates for perseus @xcite and serpens @xcite . within the previous assumptions , and given the solid angle subtended by our on - cloud region , it is easy to estimate that approximately 200300 m@xmath47 myr@xmath17 are converted into stars in the investigated region . because the cloud mass has been estimated at @xmath48 m@xmath47 @xcite , we can also derive the depletion time for the cloud , that is @xmath49 4070 myr , as well as the sf efficiency given by m@xmath50/(m@xmath51m@xmath52)@xmath6 0.020.03 .
however , the uncertainty associated to the estimate of the cloud mass is at least @xmath53 so that these quantities are correspondingly uncertain .
we have seen that the irac data are an essential , additional ingredient of ysos classification , and by including them we have been able to reconcile ( as indeed was anticipated by @xcite ) the estimated excess of class i objects previously derived using only mips aperture photometry . in this section
we want to briefly discuss the young population census , as obtained by means of the present irac photometry complemented with 2mass and mips data , with the results that can be derived using only 2mass and mips photometry .
the aim is to assess the ability of the mips photometry to correctly classify ysos when no irac data are available , as is the case for the mips regions without overlap with the irac mosaics ( see fig . [ fig_mosaic ] ) .
the classification method used in these two cases is different : in the first case the 2mass / irac / mips photometric data are used to evaluate the spectral slope @xmath31 and to separate candidate ysos from contaminants , while in the second case only 2mass / mips data are used to exploit the k@xmath54 vs. k@xmath54-[24 ] diagram as in @xcite .
the first method is clearly the most appropriate involving more photometric information ( up to six wavelengths ) ; however , being the second method the alternative choice when mips observations are the only available , it is interesting to check if differences arise in selecting the yso population with these two methods .
the two lists of candidate ysos obtained with these two approaches are reported in the corresponding electronic catalogs : vmr - d_irac_yso.txt and vmr - d_mips_yso.txt .
typical entries for both are reported in table [ tab_irac_yso ] and [ tab_mips_yso ] , respectively . in discussing the differences emerging between these two approaches
we shall limit ourselves to consider only the corresponding lists of class i objects ; similar comparisons for the other classes ( flat , ii or iii ) follow the same line of thinking . before proceeding , we reiterate that the census of the different classes in this paper is based on the working sample , i.e. is limited to those sources with valid fluxes in all four irac bands .
this clearly excludes the sources with an upper limit , a large flux uncertainty , or even a not strictly point - like shape in at least one irac band .
these objects , although not considered here , are included in the general catalog and could represent interesting targets for future investigations .
the 49 sources selected as class i objects by scrutinizing the irac catalog working sample are listed in table [ tab_irac_yso_ci ] . along with the source i d ,
coordinates and spectral index , we also report the presence of a counterpart in other bands : k@xmath54 - 2mass in column 5 and mips bands in columns 6 to 8 .
this information is needed to compare the two methods since the mips classification is based on k@xmath54 - 2mass and 24 @xmath1 m magnitudes , and only irac sources associated with both these ancillary data are subject to both irac and mips classification . by looking at table [ tab_irac_yso_ci ] ( column 9 ) , we see that only five sources out of the 49 entries were classified with mips and all of these are class i ( flag i ) . while this is the expected result when the four irac points are reasonably well aligned with the two extreme spectral points at k@xmath54 and 24 @xmath1 m , it is however noteworthy that in the relatively few cases in which both methods can be applied they give the same classification . in columns 10 and 11
we also indicate whether the source is associated with gas and dust emission . in this respect , the genuine nature of the selected class i sources is testified by the fact that 44 out of the 49 sources are coincident with gas clumps or peaks .
the remaining 5 sources are located in regions not covered by the @xmath7co map ( flag n ) .
finally , the last column indicates whether we find that the source is also irac variable on a timescale of 4.5 months @xcite ; we identify six cases of variability .
table [ tab_mips_yso_ci ] is complementary to table [ tab_irac_yso_ci ] .
it provides the list of all the 20 sources classified as class i based on k@xmath54 - 2mass and 24 @xmath1 m photometry .
besides the k@xmath54 - 24 index ( column 4 ) information is also given about the presence of a 70 @xmath1 m counterpart ( column 5 ) .
the association with an irac counterpart is indicated in column 6 while column 7 reports the quality flags of the detections in the four irac bands . since this classification
is determined by mips , there are only seven sources ( see column 8) contained in the irac working sample , while 11 further sources are associated with irac objects lacking good photometry in one or more bands .
the remaining two sources do not have a counterpart in the irac catalog ( see the notes to table [ tab_mips_yso_ci ] ) .
five objects in this table are clearly in common with table [ tab_mips_yso_ci ] but there are two more cases included in the working sample and classified as flat - spectrum sources when we derive the spectral slope exploiting the irac data .
this suggests that the classification independently provided by k@xmath54 - 2mass / mips and k@xmath54 - 2mass / irac / mips is not completely equivalent even if the two methods seem to be in agreement in five out of the seven common cases encountered .
again , most of these sources are located in gas clumps .
note also that table [ tab_mips_yso_ci ] contains many sources whose irac flags do not fulfill our compelling requirements ( only h1 and h2 quality flags ) . although they are not members of the working sample , they are nonetheless good candidates for very young protostars . finally , for the sake of completeness , we present in table [ tab_mips_remark ] a list of mips selected sources that , for various reasons , can be considered as particularly red or cold .
when observed in both mips bands , these objects have a color [ 24]-[70 ] @xmath40 6.3 , a condition we imposed to select the coldest ( t @xmath41 50k ) sources ( see figure [ k_24 + 24_70 ] , right panel ) .
other sources listed in table [ tab_mips_remark ] are : _ i _ ) the 70 @xmath1 m sources with point - like appearance but without a 24 @xmath1 m counterpart ; _ ii _ ) the objects saturated at 24 @xmath1 m , but already classified as iras class i objects ( liseau et al .
note that increasing the angular distance of the association , reported in column 7 , the uncertainty also increases so that further analysis is required to confirm a genuine source association .
the spatial distribution of the young stars provides useful information about sf mechanisms in the cloud , since ysos do not move very far from their birthplaces during the pms evolution .
much of the accumulated observational evidence suggests that stars prefer to form in groups , and in particular , that the two - point correlation function of young stars shows a different behaviour at small and large spatial scales , respectively ( see , e.g. , * ? ? ?
more recently @xcite analyzed the spatial distribution of the ysos in 36 clusters by means of both the nearest neighbor and the minimum spanning tree method .
a critical length , in the range @xmath25 0.10.5 pc , could be identified in the cumulative distribution of the source spacings indicating differences in the slope at small and large spatial scales .
we use here the function @xmath55 to quantify the degree of the ysos clustering in the vmr - d cloud : @xmath56 where @xmath57 is the average number of pairs with angular separation between @xmath58 and @xmath59 around each object , and @xmath60 is the same number expected for a random distribution of the same sources . for sf regions ,
this function for pms objects typically shows a power law behaviour @xmath61 , but with clearly different exponents for small ( @xmath62 ) and large ( @xmath63 ) spatial scales , respectively ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this fact has been interpreted as an indication of a critical scale that separates the regime of binary and multiple stars , at smaller scales , from the hierarchical clustering dominating at larger scales .
@xcite also suggested that this scale could represent the jeans length of the molecular cloud , although this interpretation encounters difficulties in explaining the different length scales observed in taurus , ophiuchus and trapezium clusters @xcite .
we applied the @xmath55 statistics to our candidate ysos selected among the vmr - d working sample sources ( see [ yso_class ] ) ; that is , those sources satisfying the criteria for minimizing the presence of galactic and extragalactic contaminants illustrated in [ source_selection ] and figures [ col_mag_col_on ] , [ col_mag_col_off ] , and [ col_mag_col_onexclu ] . the spatial distribution of these sources is presented in figure [ vela_yso+co ] where the objects of class i / flat ( red circles ) and class ii / iii ( green crosses ) are shown superposed to the 8.0 @xmath1 m irac mosaic .
we find substantial differences between clustering in early and late classes as is shown in figure [ acf ] where we present the behaviour of the @xmath55 function for class i and flat - spectrum ( left panel ) , and class ii and iii sources ( right panel ) , respectively .
it is apparent that in both cases the correlation function behaves as a single power law , at least in the investigated range of scales , but with noticeably different slopes denoting a higher degree of clustering of the younger classes with respect to the later evolutionary stages .
this can be also qualitatively seen by eye from the spatial distributions in figure [ vela_yso+co ] .
note , however , that in the range of the investigated spatial scales , there is no break in the power law slope within each region ; this is significant even though a direct comparison with the cases discussed in @xcite for optically selected pms stars is not straightforward because here we deal with irac selected ysos instead of optically selected pms stars .
a more appropriate comparison can be done with @xcite who show the behavior of the same @xmath55 function computed for the distribution of the ysos in serpens ( their figure 10 ) .
the slopes they found for both class i / flat and class ii / iii sources are remarkably similar to those shown in figure [ acf ] for vmr - d sources .
a possible difference may be due to a trend in serpens toward smaller scales due to a flattening of the function at scales smaller than @xmath64 au .
a plausible justification for the different slopes of early and late ysos is simply that it is due to the different ages of the two populations .
it is generally accepted that stars are essentially formed in clusters , so that it is naturally expected that the older objects have moved from their birthplaces , producing object dispersal we measure in the shallower slope of the correlation function for class ii / iii sources .
finally in figures [ thumb_16_17 ] and [ thumb_19_20 ] we show more details of the four regions around irs16 , 17 , 19 , and 20 , including the spatial distributions of the class i / flat ( red circles ) and class ii / iii ( green crosses ) located within 4@xmath65 . to produce these figures
, we consider an enlarged sample containing all the catalog sources located around the irs objects and with good photometry in at least two adjacent irac bands .
this allows us to compensate for the relative scarcity in these small areas of candidate ysos selected from our working sample and then to better highlight the trend in the spatial distribution of the different classes .
the sources reported here have been selected from this enlarged sample by applying the same criteria discussed in [ source_selection ] and illustrated in figure [ col_mag_col_on ] , [ col_mag_col_off ] , and [ col_mag_col_onexclu ] . in doing this
we discarded the sources with upper limits and considered only those cases for which at least a bad photometry was obtained in the two complementary irac bands . in all cases
we see that around these four irs objects the class i / flat sources ( red circles ) are not uniformly distributed .
they appear more abundant around irs16 and irs17 ; class ii / iii objects ( green crosses ) are instead more numerous and more simmetrically distributed in irs16 , 19 , and 20 .
an exception is irs17 where the class ii / iii objects follow the same elongated distribution of the class i / flat sources .
we also note that this configuration in irs17 is suggestive of the presence of two separated clusters as has also been suggested by @xcite . speculating on this case ,
we suggest that a dust disk , whose plane is perpendicular to the observed ysos distribution , may be responsible for this appearance ; this is suggested also by the central dark lane visible in all the irac bands .
it is however clear that a conclusion can not be given here and further investigations are needed .
this paper presents the analysis of the _
spitzer_-irac mosaics obtained in the region of the vela molecular ridge , cloud d. in a @xmath6 1.2 square degrees area , corresponding to the overlap of the irac mosaics in the four bands , we obtained the photometry that has been used to compile a four - band point source catalog .
this has been supplemented by relevant information at longer wavelengths from revised photometry of the vmr - d mips mosaics and from previously published millimeter line and continuum observations of the same region .
this catalog has been used to outline the general characteristics of the ysos in this sf region .
the main results are : 1 .
the vmr - d irac point source catalog has been compiled with 170,299 entries , 8796 of which , constituting our working sample , have good quality fluxes in all four irac bands .
complementary information has been also added using a revised , psf - based , mips 24 @xmath1 m and 70 @xmath1 m photometry and previous millimeter observations obtained in both co lines and dust continuum .
2 . a revised catalog containing the point sources extracted from the vmr - d mips 24 @xmath1 m and
70 @xmath1 m mosaics has been also obtained following the same photometric procedure adopted for the irac bands .
the working sample has been analyzed according to the spectral slope that was obtained by also considering both the k@xmath54 - 2mass and the mips 24 @xmath1 m fluxes when available .
we selected the candidate ysos by means of their position in color - color and color - magnitude diagrams involving irac and mips photometry ; the likely contaminants due to extragalactic and galactic objects have been also estimated and excluded from the final list of ysos constituted by 637 objects .
the candidate ysos have been used to evaluate the surface density ( 3.35.5 pc@xmath46 ) , the star formation rate ( 2.54.1 m@xmath47myr@xmath17pc@xmath46 ) and efficiency ( @xmath6 0.020.03 ) , and the depletion time ( 4070 myr ) of the vmr - d cloud .
the global clustering properties of the ysos , studied by means of the two point correlation function , have been found to be very similar to those derived by @xcite for the serpens sf region ; it clearly shows that class i / flat sources are more clustered than class ii / iii sources .
the irac catalog sources around irs16 and irs17 show a larger fraction of class i / flat seds with respect to irs19 and irs20 where the class ii / iii objects are relatively more numerous . in the irs17 region
there is a tendency for the class ii / iii to trace the same elongated distribution of the class i / flat sources .
we thank joao lin yun , luca olmi and berlinda maiolo for many helpful discussions .
de has been partly supported by the european commission fp6 marie curie research training network `` constellation '' ( mrtn - ct-2006 - 035890 ) .
fs aknowledges partial support by italian space agency ( asi ) .
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 .
support for this work was provided by nasa through an award issued by jpl / caltech .
marengo , m. , reiter , m. & fazio , g.g .
2008 , in aip conf .
1001 , ixth torino workshop on evolution and nucleosynthesis in agb stars and the 2nd perugia workshop on nuclear astrophysics .
( new york , aip ) , 331 lcccccc f@xmath66 & @xmath67 & @xmath68 & @xmath69 & @xmath70 & @xmath71 & @xmath72 + m@xmath66 & 18.5 & 18.1 & 16.3 & 15.3 & 10.1 & 2.2 + f@xmath73 & @xmath70 & @xmath74 & @xmath75 & @xmath76 & @xmath77 & @xmath78 + m@xmath73 & 16.9 & 16.5 & 14.0 & 13.0 & 8.9 & 1.0 lcccccccc 1 & 130.79321 & -44.01385 & 263.47831 & -0.97396 & 8.420e-03 & 1.812e-03 & 0.000e+00 + & 0.000e+00 & 0.0 & n & n & n & - + 2 & 130.81630 & -43.97870 & 263.46093 & -0.93919 & 1.535e-03 & 1.135e-04 & 0.000e+00 + & 0.000e+00 & 0.0 & n & n & n & - + & & & & & & & + 135 & 131.38707 & -43.71376 & 263.50867 & -0.45199 & 4.314e-03 & 1.037e-04 & 0.000e+00 + & 0.000e+00 & 0.0 & i & n & n & - + 136 & 131.38847 & -43.83051 & 263.60060 & -0.52397 & 1.934e+00 & 2.598e-01 & 2.001e+01 + & 8.478e-01 & 2.1 & i & c & c & mms1 + & & & & & & & + lccccccccc class i & 16 ( 10.1 ) & 4 ( 3.4 ) & & 44 ( 9.5 ) & 5 ( 2.9 ) + flat spectrum & 41 ( 25.9 ) & 20 ( 16.9 ) & & 61 ( 13.2 ) & 10 ( 5.7 ) + class ii & 98 ( 62.0 ) & 84 ( 71.2 ) & & 230 ( 49.7 ) & 93 ( 53.4 ) + class iii & 3 ( 1.9 ) & 10 ( 8.5 ) & & 128 ( 27.6 ) & 66 ( 37.9 ) + lccccccccccccccccccccccccccc 1 & 131.031909 & -43.736558 & 263.366966 & -0.667202 & 1.89903e-05 & 2.54204e-06 & l1 + & 4.15450e-06 & 0.00000e+00 & u & 2.11600e-05 & 0.00000e+00 & u & 2.87050e-05 + & 0.00000e+00 & u & - & 0.0 & 3.084e-04 & 0.000e+00 & u + & 5.220e-04 & 0.000e+00 & u & n & n & - + & & & & & & & + 393 & 131.096329 & -43.775259 & 263.426188 & -0.654711 & 1.32000e-02 & 7.87000e-04 & b + & 8.33412e-03 & 2.59158e-04 & h1 & 5.53000e-03 & 3.59000e-04 & b & 3.56018e-03 + & 7.23856e-05 & h1 & 37 & 1.7 & 2.543e-02 & 3.492e-03 & g + & 3.907e-04 & 0.000e+00 & u & n & n & - + & & & & & & & + 2548 & 131.183726 & -43.701613 & 263.407676 & -0.559483 & 1.42000e-03 & 8.73000e-05 & b + & 9.11296e-04 & 3.42903e-05 & h2 & 6.48618e-04 & 5.45410e-05 & h1 & 3.10650e-04 + & 1.79767e-05 & l2 & 57 & 5.1 & 4.497e-02 & 1.205e-03 & g + & 1.165e-01 & 1.191e-02 & g & n & n & - + & & & & & & & + 17590 & 131.388578 & -43.868336 & 263.630227 & -0.547490 & 2.54057e-03 & 6.21729e-05 & h2 + & 3.98529e-03 & 1.06327e-04 & h2 & 3.52000e-03 & 1.78000e-04 & b & 1.29000e-03 + & 3.38000e-04 & b & 139 & 6.0 & 3.754e-01 & 4.395e-02 & g + & 8.675e-02 & 0.000e+00 & u & c & c & mms2 + & & & & & & & + 170299 & 132.967055 & -43.592335 & 264.135486 & 0.511679 & 5.86857e-05 & 5.13993e-06 & h1 + & 5.54702e-05 & 6.68016e-06 & h1 & 2.14100e-05 & 0.00000e+00 & u & 3.13800e-05 + & 0.00000e+00 & u & - & 0.0 & 2.574e-04 & 0.000e+00 & u + & nan & 0.000e+00 & u & n & n & - + lrrrr s / n @xmath7910 & 100964 & 81945 & 11651 & 6123 + & 24 @xmath1 m & 764 & 801 & 634 & 603 + & 24 @xmath1 m & @xmath7co & 489 & 504 & 387 & 356 + & 24 @xmath1 m & @xmath7co & dust & 48 & 50 & 39 & 31 + 5@xmath80 s / n @xmath4110 & 50291 & 48732 & 10442 & 5586 + & 24 @xmath1 m & 96 & 90 & 68 & 43 + & 24 @xmath1 m & @xmath7co & 59 & 63 & 39 & 22 + & 24 @xmath1 m & @xmath7co & dust & 3 & 2 & 7 & 3 + bad photometry & 19044 & 39622 & 148206 & 158590 + lcccccccccc on - cloud & 0.57 & 700 & 620 & 1088 & 7.3 & 60 ( 9.7% ) & 70 ( 11.3% ) & 247 ( 39.8% ) & 243 ( 39.2% ) & 5.5 + off - cloud & 0.60 & 700 & 261 & 435 & 2.9 & 11 ( 4.2% ) & 18 ( 6.9% ) & 99 ( 37.9% ) & 133 ( 51.0% ) & 2.2 + on - off & 0.57 & 700 & 372 & 653 & 4.4 & 50 ( 13.3% ) & 53 ( 14.2% ) & 153 ( 41.1% ) & 117 ( 31.4% ) & 3.3 + on - cloud & 0.57 & 700 & 463 & 812 & 5.4 & 44 ( 9.5% ) & 61 ( 13.2% ) & 230 ( 49.7% ) & 128 ( 27.6% ) & 4.1 + off - cloud & 0.60 & 700 & 174 & 290 & 1.9 & 5 ( 2.9% ) & 10 ( 5.7% ) & 93 ( 53.4% ) & 66 ( 37.9% ) & 1.5 + on - off & 0.57 & 700 & 278 & 522 & 3.5 & 39 ( 13.2% ) & 51 ( 17.3% ) & 142 ( 47.6% ) & 65 ( 21.9% ) & 2.6 + on & 0.57 & 700 & 465 & 816 & 5.5 & & & & & 4.1 + off & 0.60 & 700 & 196 & 326 & 2.2 & & & & & 1.6 + on - off & 0.57 & 700 & 279 & 490 & 3.3 & & & & & 2.5 +
cha ii & 1.04 & 178 & 26 & 25 & 2.6 & 2 ( 8% ) & 1 ( 4% ) & 19 ( 73% ) & 4 ( 15% ) & 0.65 + lupus & 3.10 & 150 & 9 & 30 & 3.3 & 5 ( 5% ) & 10 ( 11% ) & 52 ( 55% ) & 27 ( 29% ) & 0.83 + perseus & 3.86 & 250 & 385 & 100 & 5.2 & 87 ( 23% ) & 42 ( 11% ) & 225 ( 58% ) & 31 ( 8% ) & 1.3 + serpens & 0.85 & 260 & 227 & 267 & 13 & 36 ( 16% ) & 23 ( 10% ) & 140 ( 62% ) & 28 ( 12% ) & 3.2 + ophiucus & 6.60 & 125 & 292 & 44 & 9.3 & 35 ( 12% ) & 47 ( 16% ) & 176 ( 60% ) & 34 ( 12% ) & 2.3 + lccccccccccccccccccccccccccc 4378 & 131.224365 & -43.926624 & 263.602173 & -0.676375 & 1.02738e-03 & 2.88283e-05 & h1 + & 9.57767e-04 & 3.59507e-05 & h1 & 1.23024e-03 & 3.53127e-05 & h1 & 1.36355e-03 + & 4.95515e-05 & h1 & - & 0.0 & 2.721e-04 & 0.000e+00 & u + & 1.199e-03 & 0.000e+00 & u & -0.60 & n & n & - + & & & & & & & + 38674 & 131.603989 & -43.704887 & 263.599640 & -0.323964 & 2.85320e-03 & 7.00859e-05 & h2 + & 3.52582e-03 & 6.94163e-05 & h2 & 4.25391e-03 & 8.60992e-05 & h2 & 5.84348e-03 + & 9.94560e-05 & h2 & 261 & 1.1 & 4.539e-02 & 8.235e-04 & g + & 9.515e-04 & 0.000e+00 & u & 0.53 & c & n & umms1 + & & & & & & & + lcccccccc 37 & 131.09596 & -43.77494 & 263.42578 & -0.65472 & 2.543e-02 & 3.492e-03 & 0.000e+00 + & 0.000e+00 & 0.0 & 4.70 & 393 & n & n & - + & & & & & & & + 546 & 131.92834 & -43.73021 & 263.76633 & -0.15719 & 1.163e+00 & 8.047e-02 & 3.519e+00 + & 3.890e-02 & 1.2 & 10.30 & 74548 & c & o & umms11 + & & & & & & & + cccccccccccc 17793 & 131.390823 & -43.872124 & 0.61 & y & - & u & u & - & c & mms2 & n + 18177 & 131.395065 & -43.882664 & 0.37 & y & - & u & u & - & c & - & y + 18652 & 131.400284 & -43.891411 & 0.82 & n & - & u & u & - & c & - & n + 19225 & 131.406204 & -43.917553 & 1.16 & n & - & u & u & - & c & - & n + 19354 & 131.407394 & -43.918594 & 0.66 & y & - & b & u & - & c & - & n + 21173 & 131.426315 & -43.456017 & 0.62 & n & - & u & u & - & n & - & n + 21825 & 131.432907 & -43.451939 & 0.59 & n & 166 & g & g & - & n & - & n + 22179 & 131.436584 & -43.448711 & 0.61 & y & - & u & u & - & n & - & n + 22262 & 131.437500 & -43.447865 & 0.31 & y & - & u & u & - & n & - & n + 23011 & 131.445084 & -43.389664 & 1.29 & n & 172 & g & g & - & n & - & n + 38674 & 131.603989 & -43.704887 & 0.53 & y & 261 & g & u & i & c & umms1 & n + 39589 & 131.613586 & -43.711151 & 1.25 & n & 270 & g & g & - & c & umms1 & n + 41404 & 131.632233 & -43.910416 & 1.39 & n & - & u & u & - & c & mms4 & n + 41429 & 131.632477 & -43.549744 & 1.04 & n & 290 & g & u & - & c & - & n + 41786 & 131.636246 & -43.911308 & 0.89 & y & - & u & u & - & c & mms4 & n + 44510 & 131.664520 & -43.380684 & 0.55 & n & 316 & g & u & - & c & - & y + 47368 & 131.694290 & -43.336273 & 0.54 & y & - & u & u & - & c & - & n + 47436 & 131.694946 & -43.887310 & 1.37 & n & 347 & g & u & - & c & - & n + 48046 & 131.701172 & -43.886124 & 1.08 & n & 352 & g & u & - & c & mms5 & n + 50748 & 131.727905 & -43.876759 & 0.67 & y & 386 & g & u & i & c & - & y + 55956 & 131.776840 & -43.322323 & 0.50 & n & 426 & g & u & - & c & - & n + 65153 & 131.855423 & -43.815571 & 1.03 & n & 473 & g & g & - & c & - & n + 65198 & 131.855774 & -43.814709 & 0.73 & y & 473 & g & g & - & c & - & n + 67878 & 131.877563 & -43.472893 & 0.30 & n & 497 & g & u & - & c & - & y + 67896 & 131.877747 & -43.467525 & 0.33 & n & 498 & g & u & - & c & - & n + 68154 & 131.880020 & -43.898384 & 1.10 & n & 503 & g & g & - & c & - & n + 68231 & 131.880646 & -43.896694 & 0.43 & n & 503 & g & g & - & c & - & n + 70412 & 131.897675 & -43.356682 & 0.45 & n & 521 & g & u & - & c & - & n + 72549 & 131.913589 & -43.438141 & 0.54 & n & - & u & u & - & c & - & n + 84160 & 131.999054 & -43.653702 & 0.43 & n & 601 & g & u & - & c & - & n + 101621 & 132.129349 & -43.517223 & 0.85 & n & 717 & g & u & - & c & - & n + 102538 & 132.136902 & -43.515751 & 0.79 & n & - & u & u & - & c & - & n + 104823 & 132.155701 & -43.281319 & 0.51 & n & 746 & g & u & - & c & umms22 & n + 104849 & 132.155991 & -43.280067 & 0.36 & n & 746 & g & u & - & c & umms22 & n + 106972 & 132.173294 & -43.529209 & 1.78 & n & 775 & g & g & - & c & mms9 & n + 108306 & 132.184921 & -43.268906 & 1.44 & n & 787 & g & g & - & c & - & n + 111144 & 132.209610 & -43.519905 & 1.83 & n & 825 & g & u & - & c & mms14 & n + 112469 & 132.221695 & -43.515335 & 2.11 & n & 838 & g & g & - & c & mms16 & n + 112472 & 132.221710 & -43.636738 & 0.94 & y & 840 & g & u & i & c & - & n + 113053 & 132.227066 & -43.553127 & 0.59 & y & - & u & u & - & c & - & n + 114979 & 132.244598 & -43.639202 & 2.02 & n & 858 & g & g & - & c & mms17 & n + 120812 & 132.302841 & -43.633789 & 0.54 & y & - & b & u & - & c & - & n + 120962 & 132.304306 & -43.596691 & 0.73 & y & - & u & u & - & c & mms20 & y + 122296 & 132.318268 & -43.612213 & 0.35 & n & 947 & g & u & - & c & mms21 & n + 124358 & 132.339371 & -44.030910 & 0.73 & n & - & u & u & - & c & - & n + 124521 & 132.340988 & -44.033684 & 2.07 & n & - & u & u & - & c & - & y + 137019 & 132.466995 & -43.378731 & 0.46 & y & 1108 & g & u & i & c & - & n + 138002 & 132.477051 & -43.388687 & 0.87 & y & 1120 & g & u & i & c & - & n + 139510 & 132.491974 & -43.381367 & 0.88 & n & 1130 & g & u & - & c & umms26 & n + 93 & 131.29494 & -43.75684 & 8.50 & n & 8642 & h1 h2 h1 h2 & f & n & - + 130 & 131.38113 & -43.89581 & 8.90 & n & 16862 & b h2 b u & - & n & - + 138 & 131.38982 & -43.87410 & 9.90 & n & 17713 & b h2 h2 h2 & - & c & mms2 + 261 & 131.60376 & -43.70510 & 9.30 & n & 38674 & h2 h2 h2 h2 & i & c & umms1 + 308 & 131.65639 & -43.87086 & 10.80 & y & 43665 & h2 b b b & - & c & - + 315 & 131.66409 & -43.88627 & 8.70 & y & - & & - & c & - + 386 & 131.72771 & -43.87690 & 10.80 & n & 50748 & h2 h2 h2 h2 & i & c & - + 541 & 131.91966 & -43.43661 & 9.20 & n & 73282 & b h2 b h2 & - & c & umms10 + 546 & 131.92834 & -43.73021 & 10.30 & y & 74548 & h2 h2 h2 b & - & c & umms11 + 576 & 131.96498 & -43.42231 & 9.50 & y & 79572 & h2 h2 h2 b & - & c & - + 729 & 132.14119 & -43.51330 & 9.50 & y & 103109 & h2 h2 b h2 & - & c & - + 814 & 132.20293 & -43.54186 & 8.40 & y & 110380 & b b b b & - & c & mms12 + 840 & 132.22160 & -43.63700 & 9.30 & n & 112472 & h2 h2 h2 h2 & i & c & - + 884 & 132.26060 & -44.40168 & 8.80 & n & - & & - & n & - + 921 & 132.29868 & -43.59826 & 9.40 & n & 120425 & h2 h2 h2 h2 & f & c & mms20 + 928 & 132.30367 & -43.60686 & 10.80 & y & 120963 & b b h2 h2 & - & c & mms21 + 1108 & 132.46693 & -43.37880 & 9.30 & n & 137019 & h2 h2 h2 h2 & i & c & - + 1114 & 132.47073 & -43.68967 & 8.90 & n & 137393 & b h2 h2 h2 & - & o & - + 1120 & 132.47690 & -43.38887 & 9.30 & n & 138002 & h2 h2 h2 h2 & i & c & - + 1170 & 132.54562 & -43.28630 & 10.00 & n & 144801 & h1 h2 h1
b & - & c & mms29 + cclccccccc 31 & 131.03918 & -44.00486 & n & y & - & & & n & - + 48 & 131.13193 & -43.83388 & y & y & - & & 3.4 & n & - + 146 & 131.39792 & -43.85200 & s & y & 18476 & b b b b & 2.8 & c & - + 288 & 131.63107 & -43.92499 & y & y & 41283 & b b b b & 0.5 & c & mms4 + 294 & 131.63802 & -43.92678 & y & y & 41975 & b b b b & 0.9 & c & mms4 + 300 & 131.64540 & -43.91031 & s & y & - & & 6.5 & c & mms4 + 315 & 131.66409 & -43.88627 & y & y & - & & 16.0 & c & - + 642 & 132.04202 & -43.35051 & y & y & 90122 & h2 h2 l2 l2 & 0.8 & c & - + 657 & 132.06612 & -43.78766 & y & y & 93485 & b h2 b u & 2.2 & c & umms16 + 661 & 132.06978 & -43.78875 & y & y & 93906 & h2 h2 h2 h2 & 0.9 & c & umms16 + 813 & 132.20084 & -42.90561 & s & y & - & & 4.9 & n & - + 814 & 132.20293 & -43.54186 & s & y & 110380 & b b b b & 2.1 & c & mms12 + 917 & 132.29530 & -43.49447 & y & y & 120052 & h2 h2 u u & 3.2 & o & - + 923 & 132.30054 & -44.27600 & y & y & - & & 11.1 & o & - + 925 & 132.30151 & -43.27896 & n & y & 120280 & h2 h2 u u & 14.6 & c & - + 951 & 132.32057 & -43.93255 & y & y & 122524 & u b u l2 & 0.1 & c & - + 991 & 132.35916 & -43.28617 & s & y & 126339 & b b b b & 1.5 & c & mms22 + 998 & 132.36937 & -44.07478 & y & y & 127325 & b b b b & 1.5 & c & mms23 + 1018 & 132.38667 & -44.18056 & s & y & - & & 3.6 & c & mms26 + 1247 & 132.65999 & -43.82816 & n & y & 156273 & u b h2 h2 & 12.0 & n & - + 1306 & 132.87169 & -43.52266 & y & y & - & & 9.3 & n & - + | this paper presents the observations of the cloud d in the vela molecular ridge , obtained with the irac camera onboard the _
spitzer space telescope _ at the wavelengths
@xmath0= 3.6 , 4.5 , 5.8 , 8.0 @xmath1 m .
a photometric catalog of point sources , covering a field of approximately 1.2 square degrees , has been extracted and complemented with additional available observational data in the millimeter region .
previous observations of the same region , obtained with the _ spitzer _ mips camera in the photometric bands at 24@xmath2 m and 70 @xmath1 m , have also been reconsidered to allow an estimate of the spectral slope of the sources in a wider spectral range .
a total of 170,299 point sources , detected at the 5 @xmath3 sensitivity level in at least one of the irac bands , have been reported in the catalog . there were 8796 sources for which good quality photometry was obtained in all four irac bands . for this sample , a preliminary characterization of the young stellar population based on the determination of spectral slope
is discussed ; combining this with diagnostics in the color - magnitude and color - color diagrams , the relative population of young stellar objects in the different evolutionary classes has been estimated and a total of 637 candidate ysos have been selected .
the main differences in their relative abundances have been highlighted and a brief account for their spatial distribution is given . the star formation rate has been also estimated and compared with the values derived for other star forming regions . finally , an analysis of the spatial distribution of the sources by means of the two - point correlation function shows that the younger population , constituted by the class i and flat - spectrum sources , is significantly more clustered than the class ii and iii sources . |
particle and heat transport in disordered spin chains is actively studied last years @xcite since these models present apparently simple examples of interacting quantum systems with frozen disorder .
one particular example of a such studies is presented by our recent work @xcite where thermal conductivity of weakly disordered spin-@xmath0 anisotropic ( xxz ) chain was calculated at low temperatures @xmath6 and in the anisotropy range @xmath10 . in that study we employed jordan - wigner transformation to the fermionic chain problem and then used standard bozonization technique @xcite which is limited to the luttinger liquid approximation with a linear spectrum ( lll ) .
it was found in @xcite that such an approximation is valid for sufficiently low temperatures if @xmath11 and frozen disorder is weak , @xmath12 ; on the other hand , lll approximation breaks down completely at larger values of @xmath13 .
effects of nonlinear spectrum in 1d quantum liquids have been actively studied from different viewpoints , see recent review @xcite and papers @xcite .
there are basically two possible approaches to the problem , using either fermionic or bosonic representation . in the fermionic representation , bare spectrum of quasiparticles
is @xmath7 , with @xmath14 at small @xmath15 . for a vanishing or very weak interaction between fermions , one can easily construct bosonic correlation function as composite objects made out of pair of fermions
. then spectral function @xmath16 corresponding to any bosonic operator can be obtained as a convolution of two fermionic spectral functions and thus will have a finite width @xmath17 . however
, this approach becomes rather complicated if interaction between fermions is not weak . on the other hand ,
strong fermion - fermion interaction can be conviniently described in the bosonic representation , but then nonlinearity of the original fermionic spectrum transforms into non - linear interaction between ll bosons .
this interaction is cubic if @xmath18 ( which is a general case ) and quartic in the case of particle - hole symmetry when @xmath19 .
the problem with treatment of such an interaction within ll theory is nontrivial since for 1-dimensional particles with linear spectrum all interactions between particles with the same sign of velocity are of resonant nature as absolute value of velocity is fixed and particles stay in contact for arbitrarily long time . as a result
, straightforward second - order perturbation theory produces diverging result for the imaginary part of the self - energy at the `` light cone '' : @xmath20 . in the present paper
we develop diagrammatic approach to the calculation of @xmath21 for bosonic excitations with energy @xmath22 which is low in comparison with the temperature @xmath23 .
we show that in the limit @xmath24 quasiparticle decay rate @xmath25 can be found self - consistently near the light - cone , leading to a finite ( although non - analytic ) result .
we consider below two different situations : pure exchange spin chain hamiltonian symmetric w.r.t . to @xmath26 inversion , and
spin chain in presence of uniform magnetic field @xmath27 along @xmath28 axis , with @xmath29 . in the latter case our result is very similar to the one obtained long ago by a.f.andreev @xcite for fluctuational correction to viscosity in a 1-dimensional classical hydrodynamics and rederived in ref . @xcite .
we are not aware of any previous calculations , which accounts for spectrum nonlinearity , for the symmetric case @xmath30 .
we emphasize that the decay rate of bosonic excitations @xmath31 calculated in this paper does not coinside with inelastic relaxation rate ; rather it provides the measure of coherence for the specific type of excitations we study .
the meaning of @xmath31 is that it determines ( due to direct local relation between bosonic and spin variables ) the shape of the dynamic spin - spin correlation function @xmath8 at @xmath32 . for the possibility to develop a self - consistent diagrammatic approach
, the condition @xmath24 is crucial . in the opposite limit more involved calculations
are needed , like those developed by imambekov _
@xcite for the shape of bosonic spectral function at @xmath33 .
another piece of work we mention is related with a discussion of the presence of the drude weight in these spin chains @xcite .
this interesting issue is beyond the scope of our study since it is related to the decay rate of the uniform current @xmath34 , whereas we calculated @xmath35 for @xmath36 .
the rest of the paper is organized as follows : we formulate our model in sec .
[ sec : model ] , develop diagrammatic perturbation theory in sec.[sec : lowestorder ] and extend it to the self - consistent approach in sec.[sec : selfconsistent ] , where major results for the decay rates @xmath37 are obtained .
sec.[sec : correlationfunctions ] is devoted to the calculation of dynamic spin - spin correlation function @xmath8 at @xmath32 which describe spreading of the excitation wave - packet due to scattering on thermal excitations ; finally , sec.[sec : conclusions ] contains our conclusions .
we study spin-@xmath38 anisotropic xxz spin chain with external magnetic field applied along the @xmath28 direction , which is described by the following hamiltonian : @xmath39 the sign of exchange constant in the @xmath40 plane can be changed utilizing the canonical transformation @xmath41 , @xmath42 , and @xmath43 ; this allows us to fix the sign of coupling constant @xmath44 .
positive value of @xmath13 corresponds to ferromagnetic exchange , while negative sign corresponds to antiferromagnetic exchange . by means of the jordan - wigner transformation , the hamiltonian is equivalent to the following hamiltonian of interacting spinless fermions : @xmath45 with fermion density operator @xmath46 . for small
compared to the bandwidth magnetic field @xmath47 and values of anisotropy parameter @xmath48 , the low - energy properties the is described by the luttinger liquid model @xcite .
the model describes fermion density excitations , which is related to the bosonic field @xmath49 as @xmath50 . after introducing canonical conjugate momentum @xmath51 , so
that @xmath52 = i \delta(x - y)$ ] , the quadratic part of the hamiltonian density , including its field - dependent part , are written as follows : @xmath53 with @xmath54 being plasmon group velocity and @xmath55 being dimemsionless luttinger parameter , whose values are expressed in terms of coupling constant @xmath4 , lattice constant @xmath56 and anisotropy parameter @xmath13 as follows : @xmath57 if one considers the quadratic hamiltonian only , one immediately obtains that magnetic field term can be easily discarded using simple phase shift @xmath58 . however , it is not the case if one takes into the account irrelevant terms , that are higher order in bosonic fields .
alternatively , one can introduce densities of right- and left - moving fermions @xmath59 and @xmath60 , consisting of fourier harmonics of @xmath61 with @xmath62 and @xmath63 respectively .
for further calculations , it will be convenient to rescale them with @xmath64 factor , so that @xmath65
. explicit expression for the densities is @xmath66 , with upper sign corresponding to the right - movers and lower sign corresponding to the left - movers . in terms of these fields ,
the quadratic part of the hamiltonian density is written as follows : @xmath67 in addition to the quadratic part , there are also irrelevant in the rg sense terms in the hamiltonian , which keep information about the lattice nature of the original model @xcite .
the two most important of those irrelevant operators are so - called `` umklapp term '' and the `` band curvature term '' .
the `` umklapp term '' is written as follows : @xmath68 such term has scaling dimension @xmath69 and is irrelevant at @xmath70 , that is @xmath71 .
the effect of such term at nonzero temperatures was studied in ref .
@xcite , where it was shown that it leads to the finite decay of the quasiparticles @xmath72 , which is small at sufficiently large @xmath55 .
it is convenient to write the `` band curvature term '' in terms of @xmath59 and @xmath60 .
corresponding expression reads : @xmath73 with parameters @xmath74 @xmath75 @xmath76 these terms are also irrelevant with negative scaling dimension @xmath77 ; however , at finite temperatures they become more important compared to the umklapp term at @xmath78 , that is @xmath79 .
thus at zero magnetic field we will restrict ourselves to the region @xmath78 and consider hamiltonian @xmath80 .
magnetic field behaves as a chemical potential for the jw fermions .
shifting it away from the half - filling point with particle - hole symmetry leads to the appearing of the quadratic terms in the quasiparticle dispersion @xmath18 , which in language of the bosonization corresponds to the cubic interaction terms .
these terms naturally appear if one performs the phase shift @xmath81 , that discards the magnetic field term from the quadratic part of the hamiltonian and takes into the account the nonlinear terms ; they read as follows : @xmath82 with constants @xmath83 the effect of quadratic curvature on the transport properties was studied in refs . @xcite . at sufficiently small magnetic fields ,
cubic terms are less important compared to the quartic terms ; below it will be shown that crossover to the regime where cubic terms dominate in the quasiparticle decay happens at @xmath84 .
the total hamiltonian at nonzero magnetic fields is thus @xmath85 , while the latter term can be neglected when the magnetic field is large enough .
below we will study only the effect of `` band curvature term '' on the decay of the quasiparticles , restricting ourselves to the region @xmath78 where it is definitely more important at low @xmath23 , compared to the `` umklapp term '' .
in this section we will develop a perturbation theory with respect to nonlinear terms in the hamiltonian . to begin with ,
we introduce retarded correlation functions ( here we use short notation @xmath86 and @xmath87 : @xmath88\right\rangle \\
g_{ret}^{(l)}(\mathbf{x}_{1}-\mathbf{x}_{2})&=-i\theta(t_1-t_2)\left\langle \left[l(\mathbf{x}_{1}),l(\mathbf{x}_{2})\right]\right\rangle\end{aligned}\ ] ] with following unperturbed values calculated with respect to @xmath89 : @xmath90 in which upper sign corresponds to @xmath91 and lower sign corresponds to @xmath92 . to sum up reducible feynman diagrams in perturbation theory series
we will use standard dyson equation .
the equation and its solution are written as follows ( for both right and left movers ) : @xmath93 @xmath94 we are interested in the decay rate of the quasiparticles , that is governed by the imaginary part of the self - energy . neglecting the renormalization of the quasiparticle spectra and wavefunction amplitude , that correspond to its real part , one immediately obtains the following form of the greens function : @xmath95 @xmath96 in the calculations below we will use keldysh diagram technique for nonzero temperatures .
this technique introduces additional keldysh space .
green functions are 2 by 2 matrices in keldysh space with following structure : @xmath97 the keldysh component of the green functions is expressed via the probability function @xmath98 as follows : @xmath99 interaction vertices in keldysh technique has symmetric tensor structure with following components ( we use the standard notation of `` classical '' and `` quantum '' fields as it is introduced e.g. in ref .
@xcite , so indices @xmath100 ) : @xmath101 @xmath102 @xmath103 @xmath104 below we will focus only on the calculation of retarded self - energy part for density of right - moving particles ; expressions for left - movers can be extracted straightforwardly by replacing @xmath105 in @xmath106 due to l - r symmetry . in the second order of perturbation theory
, there are two diagrams for retarded self - energy of right - movers coming from terms @xmath107 and @xmath108 in , see fig .
[ fig : selfenergy4 ] .
we will focus on the imaginary part of the self - energy , since it governs the decay of the quasiparticles .
it is possible to express the contribution to the imaginary part of self - energy coming from the first diagram via the spectral weights @xmath109 : @xmath110 the unperturbed spectral weights are delta - peaked on the mass shell @xmath111 .
this fact together with energy and momentum conservation laws lead to singular behavior of self - energy , which is also shows delta - peak behavior on the mass shell : @xmath112\left[q^{2}+4\left(\frac{2\pi t}{u}\right)^{2}\right].\end{gathered}\ ] ] singularity on the mass shell in is an artifact of the lowest - order perturbation theory calculation ; below we will see that its proper regularization leads to a behavior that is finite everywhere and non - analytic as function of the coupling constant @xmath113 . in order to obtain meaningful result for this diagram
, one should perform the calculation of self - consistently by putting `` dressed '' greens functions instead of bare ones : it is known that the exact spectral weight calculated with `` dressed '' green functions have nonzero width and finite height @xcite .
such procedure is equivalent to resummation of an infinite series of feynman diagrams that are most singular near the mass shell .
below , in sec .
[ sec : selfconsistent ] , we will perform such self - consistent procedure . the analytic expression for the second diagram from fig .
[ fig : selfenergy4 ] calculated with bare greens functions can only be expressed via the polylogarithm function ; however , there are three cases of interest , namely zero temperature @xmath114 , large temperature @xmath115 and on the mass shell @xmath116 , where the asymptotic behavior can be obtained . at zero temperature
it yields : @xmath117 while both for cases @xmath118 and @xmath119 it yields : @xmath120 according to the equation , this corresponds to quasiparticle decay rate @xmath121 . as we will see later ,
such decay rate is negligible compared to the singular contribution coming from first diagram .
here we perform the same calculations as above , but with respect to the perturbation .
there are two feynman diagrams giving contribution to the imaginary part of self - energy ( see fig .
[ fig : selfenergy3 ] ) . the expression for the first diagram in terms of the spectral weights @xmath122 is following : @xmath123 direct calculation
( see appendix [ sec : appendiximsigma ] ) immediately recovers the same delta - peaked behavior on the mass shell due to the very same reason as in previous case : @xmath124 similarly , the proper way to deal with it is to solve the equation self - consistently utilizing the `` dressed '' greens functions instead of bare ones .
the second diagram yields : @xmath125 the same asymptotic analysis of cases @xmath114 , @xmath118 and @xmath116 as for quartic interaction can be performed , yielding result for zero temperature : @xmath126 and both for @xmath118 and @xmath116 : @xmath127 according to , this leads to the decay rate @xmath128 .
however , it is to be stressed again that such decay rate is negligible compared to the singular one coming from the first diagram and equation .
cubic vertices @xmath129 and
@xmath130 also lead to the appearance of the offdiagonal green functions , such as @xmath131 @xcite .
they will manifest themselves e.g. when calculating spin correlation functions , see sec .
[ sec : correlationfunctions ] ; however , they contain second power of `` coupling constant '' @xmath132 and thus are parametrically smaller compared to the contribution coming from diagonal terms .
we now switch to the self - consistent calculation as it was outlined above in sec .
[ sec : lowestorder ] .
the main contribution for linewidth @xmath37 near the mass shell @xmath119 comes from and ; for now we will neglect contributions coming from other processes .
we will also neglect the renormalization of quasiparticle spectra due to temperature effects and assume the dressed green functions taking the form .
the cases of zero and nonzero magnetic fields will be studied separately below .
we substitute green functions of the form to the eq . , and introduce new variable : instead of @xmath15 we will use deviation from mass shell defined as @xmath133 . since we are interested in well - defined quasiparticles @xmath134 , we will also replace @xmath135 prefactor in by @xmath136 .
after doing so , we arrive at the following integral equation for @xmath137 : @xmath138 @xmath139 integration is performed over all @xmath140 and @xmath141 taking into account energy - momentum conservation @xmath142 and @xmath143 .
all the @xmath141 in the @xmath144 integral are typically of the order of @xmath145 .
to start our analysis , we will assume that @xmath146 , which will allow us to neglect @xmath147-dependence and perform integration over @xmath141 explicitly ( this assumption will be checked _ a posteriori _ ) .
similarly , we will put @xmath148 in the right - hand side of the equation and obtain a closed equation for @xmath31 .
this calculation yields : @xmath149 next we introduce dimensionless linewidth @xmath150 .
the corresponding equation for @xmath151 reads as follows : @xmath152 if one assumes slow dependence of @xmath151 on @xmath28 for small @xmath28 in the right - hand side of the equation , one immediately obtains that integral is logarithmic in @xmath28 , and the main contribution comes from @xmath153 .
it means that typically @xmath154 .
since we are interested in @xmath155 , then @xmath147-dependence in @xmath144 and thus in @xmath156 is indeed negligible , and our assumption is justified .
we recover most singular term for logarithmic integral by expanding @xmath157 and arrive at following equation : @xmath158 since for constant @xmath151 the integral is logarithmic , we substitute asymptotic behavior of the form @xmath159 and arrive at following expression for constant @xmath160 .
this gives us final result for decay rate of particles near the mass shell @xmath119 : @xmath161
this is one of our major results in this paper .
let us now switch to the self - consistent calculation of decay rate of quasiparticles in the presence of magnetic field .
corresponding self - consistent dyson equation for @xmath137 is written as follows : @xmath162 @xmath163\left[\epsilon_{2}^{2}+\gamma^{2}(\omega_{2},\epsilon_{2})\right]}\ ] ] here integration is performed over all @xmath140 and @xmath141 with constraints @xmath164 and @xmath165 . making the assumption of negligible @xmath147-dependence of @xmath156 , we perform @xmath141 integration @xmath166 and switch to dimensionless decay rate , which is defined as follows : @xmath167 .
the self - consistent dimensionless integral equation for @xmath151 reads as follows : @xmath168 similarly to the previous discussion , we first assume that @xmath151 is slowly varying function of @xmath28 for small @xmath28 ; after making this assumption for the r.h.s . of the equation
, we immediately recover @xmath169 behavior of the integral , and main contribution now comes from the area @xmath170 . that again allows us to expand distribution functions @xmath171 and to write the following equation for @xmath172 : @xmath173 finally ,
substituting @xmath174 and performing integration , we evaluate the value of @xmath175 : @xmath176 which yields final result for decay rate for this case : @xmath177 let us now switch to the discussion of the obtained result . the main contribution to the decay rate
comes from the the modes with @xmath178 .
surprisingly , these modes correspond to the classical limit ; indeed , the result was originally obtained for _ classical one - dimensional liquid _ in ref .
@xcite ; the same result was more recently rederived within luttinger liquid framework @xcite .
however , we obtained that main contribution now comes from @xmath179 , which is inconsistent with original assumption of negligible @xmath147-dependence .
we argue below that taking @xmath147-dependence into the account will lead to the same behavior of the linewidth @xmath180 , but with modified constant @xmath175 .
we have calculated decay rate of quasiparticles for zero and relatively large magnetic fields , eqs . and respectively .
for the validity of self - consistent procedure performed above , either one or another decay mechanism should be parametrically stronger than another . for a particle with an arbitrary energy @xmath24 , there is a crossover magnetic field @xmath181 where two contributions for the decay rate are almost equal ; namely , comparing the results , we immediately obtain @xmath182 .
the crossover between answers and happens at that threshold magnetic field @xmath181 , below which the first answer can be applied , and above which the second one is correct . at @xmath183
all the excitations `` feel '' the magnetic field . in the previous section
we have developed a self - consistent approach , which corresponded to the resummation of the infinite series of feynman diagrams .
however , a priori it is not known which diagrams mostly contribute to the width of the spectral weight , thus one needs to perform a sanity check . however all the virtual particles participating in the processes contributing to the most singular part of of the self - energy of right - moving plasmon are right - movers as well . in this section we will consider diagrams coming from the next order of perturbation theory in order to see if they contain smallness compared to ones already taken into account . contributing to the @xmath31 for the zero magnetic field case .
bold lines correspond to dressed keldysh green functions for right - moving plasmons ( with retarded component given by ) . ]
the next diagram comes from third order of perturbation theory and is shown at fig .
[ fig : sigmanextorder:4 ] .
all the particles are right - moving , and below we will drop the @xmath91 superscript for all the green functions .
this diagram yields the following analytic expression : @xmath184+\\ + g_{ret}(\mathbf{q}_{1})\left[g_{k}(\mathbf{q}_{2})g_{ret}(\mathbf{q}_{3})+g_{ret}(\mathbf{q}_{2})g_{k}(\mathbf{q}_{3})\right]g_{ret}(\mathbf{q}_{4})g_{ret}(\mathbf{q}_{5})+\\ + g_{ret}(\mathbf{q}_{1})\left[g_{k}(\mathbf{q}_{2})g_{ret}(\mathbf{q}_{3})+g_{ret}(\mathbf{q}_{2})g_{k}(\mathbf{q}_{3})\right]g_{adv}(\mathbf{q}_{4})g_{adv}(\mathbf{q}_{5})+\\ + g_{ret}(\mathbf{q}_{1})\left[g_{k}(\mathbf{q}_{2})g_{ret}(\mathbf{q}_{3})+g_{ret}(\mathbf{q}_{2})g_{k}(\mathbf{q}_{3})\right]g_{k}(\mathbf{q}_{4})g_{k}(\mathbf{q}_{5})+\\ + g_{ret}(\mathbf{q}_{1})g_{ret}(\mathbf{q}_{2})g_{ret}(\mathbf{q}_{3})\left[g_{adv}(\mathbf{q}_{4})g_{k}(\mathbf{q}_{5})+g_{k}(\mathbf{q}_{4})g_{adv}(\mathbf{q}_{5})\right]+\\ + g_{k}(\mathbf{q}_{1})\left[g_{k}(\mathbf{q}_{2})g_{ret}(\mathbf{q}_{3})+g_{ret}(\mathbf{q}_{2})g_{k}(\mathbf{q}_{3})\right]\left[g_{k}(\mathbf{q}_{4})g_{ret}(\mathbf{q}_{5})+g_{ret}(\mathbf{q}_{4})g_{k}(\mathbf{q}_{5})\right]\big)\end{gathered}\ ] ] here we integrate over all @xmath185 with constraints @xmath186 . substituting equilibrium relation for keldysh green function , and discarding some terms due to causality , we arrive at following expression ( here @xmath187 ) : @xmath188 assuming again that major contribution comes from classical modes with @xmath24 , we replace @xmath189 .
the first term cancels out in the leading order in @xmath190 . finally ,
substituting dressed green functions and replacing @xmath191 by @xmath192 in the prefactors , we arrive at ( here @xmath193 ) @xmath194 the above result is purely real and corresponds to the renormalization of the spectrum instead of the linewidth which we are interested in .
it happens that the _ most singular _ contribution to the diagram with three vertices does not contain imaginary part .
in other terms , contribution to @xmath195 from that diagram is parametrically smaller than the value .
we now proceed with the calculation .
simple power counting shows that this integral is again logarithmic ( we integrate over 3 different @xmath140 out of 5 , since there are two independent conservation laws ) .
however , the logarithmically divergent part cancels out because it is odd in @xmath22 .
this cancellation means that contribution to @xmath31 comes from modes @xmath179 , which allows us to replace logarithms in @xmath196 by a constant , @xmath197 . finally , substituting the result
, we arrive at : @xmath198 with numerical constant @xmath199 the result given by eqs .
, of the same order in the `` coupling constant '' @xmath200 as @xmath31 obtained from the lowest - order self - consistent solution , eq . , which is what one should expect . however , it is still parametrically smaller by the factor @xmath201 .
higher - order diagrams ( for example , coming from fourth order of perturbation theory ) still contain singular contributions to the linewidth , that are of the same order in `` coupling constant '' @xmath200 as @xmath31 coming from the first diagram , eq . .
explicit calculation of the higher order diagrams is cumbersome , but we argue that ( similarly to the third order diagram discussed above ) it wo nt contain logarithmically large factors and thus will be parametrically smaller by some negative power of @xmath202 .
contributing to the @xmath31 for nonzero magnetic field case . ]
first diagram which was nt taken into account by self - consistent procedure comes from fourth order of perturbation theory and is shown on fig .
[ fig : sigmanextorder:4 ] . the corresponding expression for retarded self - energy yields : @xmath203 here
we integrate over all the @xmath185 , that are not fixed by the energy - momentum conservation laws , which are as follows : @xmath204 , @xmath205 ( thus only two energy - momentum pairs are independent ) . using equilibrium relation and noticing that some terms vanish due to retarded structure of corresponding diagram
, we regroup these terms as follows ( here again @xmath187 ) : @xmath206 we again apply assumption that main contribution comes from classical modes and thus @xmath189 and immediately obtain that first two terms vanishes in the leading order .
next we substitute dressed green functions and integrate over @xmath191 , neglecting the @xmath147-dependence of @xmath156 , which was discussed above for the sake of comparison : @xmath207 substitution of yields following expression for the correction : @xmath208 with numerical constant estimated as @xmath209 .
one finds that the correction is of the same order as the lowest - order result itself , has the different sign and numerical constant that is almost 4 times smaller .
contrary to the case @xmath30 studied before , here all relevant excitation energies @xmath210 and all diagrams are of the same order of magnitude .
it was noticed in ref .
@xcite that this problem appears to be asymptotically equivalent to the kardar - parisi - zhang nonlinear model @xcite of noisy classical dynamics .
we have calculated the decay rates of the quasiparticles on the mass shell . in this section
we will discuss how this decay rates manifest themselves in the real - space properties of the spin - spin correlation functions , i.e. @xmath211 .
this correlation is equivalent ( via the jordan - wigner transformation ) to the fermionic density - density correlation function , which in turn is related to the plasmon propagator . to be more specific : @xmath212 with @xmath213 and @xmath56 being lattice constant . in the above expression we discarded the offdiagonal green functions @xmath214 , which appear in the presence of the magnetic field ; as it was discussed in sec .
[ sec : lowestorder ] , they are parametrically smaller and thus negligible . correlation functions for left- and right - moving plasmons are related via plasmon velocity sign change @xmath215 . below we will focus on the calculating @xmath216 . without take into account the effects of spectrum nonlinearity , one arrives at @xmath217 .
inverting fourier transform give us well - known pure ll result @xcite : @xmath218 this result describes a wave packet centered at @xmath219 , with a constant width @xmath220 .
above we have evaluated the decay rate for @xmath24 .
this leads to the modification of .
direct calculation ( see appendix [ sec : appendixcorrelation ] for details ) shows that modified answer behaves as quasi - diffusive wave packet , which is still centered at @xmath219 , but its width now depends on time as @xmath221 which can be estimated using the original model parameters as @xmath222 it is also convenient to introduce dimensionless distance to the `` light cone '' @xmath223 . for the `` height '' of the corresponding wave packet , one finds the following expression : @xmath224 while its tails behave in a following manner : @xmath225 these asymptotics work only while @xmath226 .
this leads to the following answers for the spin - spin correlation functions . at the `` light cone '' ( that is @xmath227 ) , for zero magnetic field one has : @xmath228 while for nonzero magnetic field one has : @xmath229 away from light cone ( that is @xmath230 ) , for zero magnetic field one has : @xmath231 while for nonzero magnetic field one has : @xmath232
in this paper we calculated dynamic spin - spin correlation @xmath233 for the xxz spin chain with ferromagnetic @xmath1 coupling , @xmath234 , in the semiclassical regime where relevant frequency @xmath22 is much smaller than the temperature @xmath23 .
the results are given by eqs .
, for the limiting cases of small @xmath235 and large @xmath236 values of magnetic field @xmath27 .
physically these results describe motion and spreading of a `` wave packet '' of a spin excitation due to scattering by thermal excitations of the same spin chain .
the key quantity needed to derive these results is a decay rate @xmath31 of bosonic quasiparticle with energy @xmath22 , which appears in the luttinger liquid description of the interacting fermionic problem equivalent to the original xxz model via jordan - wigner transformation .
lifetime of bosonic modes is finite due to nonlinear coupling between them . in turn
, this coupling is due to fermionic spectrum nonlinearity at the energies close to the fermi level .
if original problem spin is symmetric with respect to @xmath237 reflection , the particle - hole symmetry is preserved in the fermionic representation ; as a result , the lowest - order nonlinearity in the spectrum of fermions is cubic , @xmath19 .
after bozonization , it translates into the 4-order nonlinear coupling between bosonic modes .
we have calculated the resulting @xmath31 at @xmath24 by means of self - consistent summation of the main set of diagrams , the result is provided in eq.([eq : decayzerofield ] ) .
it was shown that contributions from higher diagrams is smaller as negative powers of @xmath238 . in presense of nonzero magnetic field
@xmath27 the above - mentioned symmetry is broken and cubic nonlinear vertices are present in the bosonic representation
. relative importance of these cubic terms and 4-order terms for the decay rate @xmath31 depends on all three parameters @xmath22 , @xmath23 and @xmath27 .
namely , cubic nonlinearity dominates at @xmath239 .
our results for @xmath31 in this region coinside ( up to numeric coefficient of order 1 ) with those previously obtained in @xcite .
applicability range of these results was questioned recently in ref .
@xcite , where the statement was made that `` hydrodynamic '' scaling behavior ( [ eq : decayfinitefield ] ) is limited to a very narrow frequency range @xmath240 , while at higher frequencies @xmath241 , see fig.1 of the paper @xcite .
we doubt this statement is correct since direct substitution of @xmath242 into eq.([eq : decayfinitefield ] ) leads to the estimate for @xmath243 .
thus we tend to believe that the issue of applicability of the result ( [ eq : decayfinitefield ] ) is still open .
we are grateful to i. v. gornyi , a. d. mirlin , i. v. protopopov , m. a. skvortsov for illuminating discussions .
this research was supported by the russian science foundation grant # 14 - 42 - 00044 .
the research was also partially supported by the rf presidential grant no .
in this appendix we will demonstrate how to obtain the expressions for the imaginary parts of the self - energy .
[ [ zero - magnetic - field-3 ] ] zero magnetic field + + + + + + + + + + + + + + + + + + + we start from the analytical expression for the two diagrams on the fig . [
fig : selfenergy4 ] , that are given by the second order of perturbation theory ( here @xmath244 ) : @xmath245\end{gathered}\ ] ] @xmath246\end{gathered}\ ] ] we proceed by expressing retarded greens functions through their imaginary parts using kramers - kronig relations @xmath247 ; we also express keldysh greens functions using equilibrium relation .
next we perform integration over @xmath140 using known positions of residues ( namely , @xmath248 ) . finally ,
taking the imaginary part of the obtained expression using @xmath249 , we arrive at following expressions : @xmath250 @xmath251 for the unperturbed greens functions , the imaginary part is proportional to the delta - function at mass shell @xmath252 , which for the leads to the notorious singular behaviour of the self - energy on the mass shell : @xmath253 with @xmath254 at zero temperature @xmath255 and the integration happens only in the region where all the @xmath140 of the same sign .
the evaluation is then straightforward and yields @xmath256 .
the derivative w.r.t .
temperature can be , however , expressed via the another integral : @xmath257 and @xmath258 = \\ = \frac{1}{3}\omega(\omega^{2}+(2\pi t)^{2})\end{gathered}\ ] ] the second integral can be evaluated using the same trick .
namely , at zero temperature it yields @xmath259 ; and the derivative w.r.t .
temperature can also be evaluated exactly .
combining all together , we arrive at the expression for the first integral : @xmath260\left[\omega^{2}+4(2\pi t)^{2}\right]\ ] ] which finally yields the result .
the expression for the second diagram , eq . is not singular at the mass shell and thus is less interesting for us .
it can be expressed via another integral : @xmath261 and @xmath262 @xmath263 the latter integral can be evaluated exactly only using polylogarithm functions ; however , its asymptotic behavior is easily obtainable : @xmath264 combining all together , we arrive at results and . [
[ nonzero - magnetic - field-3 ] ] nonzero magnetic field + + + + + + + + + + + + + + + + + + + + + + the results for cubic interaction between plasmons was studied in ref.@xcite . here
we will demonstrate how to reproduce it in the same fashion .
we start from the expressions for two diagrams [ fig : selfenergy3 ] ( here @xmath265 ) : @xmath266\end{gathered}\ ] ] @xmath267 expressions for its imaginary parts obtained using krammers - kronig relations read as follows : @xmath268 @xmath269 substituting unperturbed greens functions in the first expression , we immediately obtain that it is again singular at the mass shell , and it can expressed via the same @xmath270 integral we have evaluated above as follows : @xmath271 while the second expression contains double integration over two delta - functions and is thus trivial : @xmath272 which are precisely the results and .
in this appendix we will derive the results outlined in sec . [
sec : correlationfunctions ] . we need to calculate the fourier transform of the lesser component of keldysh green function , which is expressed using the equilibrium relation : @xmath273 using the general form of dressed green functions , and neglecting the @xmath274 dependence of decay rates @xmath137 , we can perform momentum integration and arrive at following general expression : @xmath275 after substituting the result , it is convenient to make the integral dimensionless by introducing the lengthscale @xmath276 , see eq . , @xmath277 and switching to dimensionless variable @xmath278 .
we arrive at following integral : @xmath279\ ] ] for @xmath280 ( that is the light cone @xmath219 ) the expression in the brackets can be replaced by unity .
the integral is then immediately yields @xmath281 ( euler gamma - function ) , which gives the first part of the result . for @xmath282 one can rotate the integration contour by @xmath283 arriving at exponentially decaying integrals @xmath284 , and then taylor - expand the corresponding expression , arriving at : @xmath285\approx\\ \approx\frac{1}{\sqrt{2}}\int_{0}^{\infty}dz\cdot z^{3/2}e^{-z\delta}=\frac{\gamma(5/2)}{\sqrt{2}}\frac{1}{\delta^{5/2}},\end{gathered}\ ] ] @xmath286\approx\\ \approx\frac{1}{\sqrt{2}}\int_{0}^{\infty}z^{1/2}e^{-z\delta}dz=\frac{\gamma(3/2)}{\sqrt{2}}\frac{1}{\delta^{3/2}}.\end{gathered}\ ] ] substitution of these two integrals yields the first part of the result .
we again introduce lengthscale @xmath276 as in eq . , and arrive at dimensionless integral : @xmath287\end{gathered}\ ] ] for @xmath280 one can again immediately replace the expression in the square brackets by unity , then replace the slowly - varying logarithm by constant @xmath288 , and then perform the gaussian integration , which yields the value @xmath289 .
this leads us to the second part of the answer .
for @xmath282 , the integrals are again almost gaussian ( after the slowly varying logarithm being replaced by its typical value ) : @xmath290 @xmath291 combining these two expression yields the second part of the answer .
let us now discuss the applicability of the obtained results for both cases of zero and nonzero magnetic fields .
the results for @xmath31 used here are obtained under the condition @xmath24 , and thus one can use it only if the typical values of @xmath22 in all the integrals are also small compared with temperature .
this criterion is equivalent to @xmath292 . since typical values of @xmath28 in the integrals in both cases are @xmath293 ( for @xmath280 ) and @xmath294 ( for @xmath282 ) , we immediately obtain that our results are applicable when @xmath295 .
this criterion corresponds to sufficiently large times .
basko , i.l .
aleiner and b.l .
altshuler , ann . phys . * 321 * , 1126 ( 2006 ) i.v .
gornyi , a.d .
mirlin and d.g .
polyakov phys .
* 95 * , 206603 ( 2005 ) r. nandkishore and d.a.huse , annual review of condensed matter physics , * 6 * , 15 ( 2015 ) a. pal and d. a.huse , phys.rev.b * 82 * , 174411 ( 2010 ) m. serbyn , z. papic , and d. a. abanin phys .
lett . * 111 * , 127201 ( 2013 ) m. serbyn , z. papic , and d. a. abanin , phys .
lett . * 110 * , 260601 ( 2013 ) r. berkovits , phys .
b * 89 * 205137 ( 2014 ) y. bar lev , g. cohen , and d.r .
reichman , phys .
lett . * 114 * , 100601 ( 2015 ) m. kulkarni and a. lamacraft phys .
a 88 , 021603(r ) , ( 2013 ) alex kamenev and alex levchenko , advances in physics , vol .
58 , issue 3 , pp .
197 - 319 ( 2009 ) m. kardar , g. parisi , and y .- c .
zhang , phys .
56 , 889 ( 1986 ) ; t. kriecherbauer and j. krug , j. phys . a 43 , 403001 ( 2010 ) ; t. sasamoto and h. spohn , j. stat .
( 2010 ) p11013 . | we study one - dimensional anisotropic xxz spin-@xmath0 model with ferromagnetic sign of the coupling and @xmath1 exchange constant @xmath2 , where @xmath3 , and @xmath4 is the coupling within xy spin plane .
we calculate damping of low - energy excitations with @xmath5 due to their scattering from thermal excitation bath with temperature @xmath6 , taking into account nonzero curvature of the excitation spectrum , @xmath7 .
we calculate also longitudinal spin - spin correlation function @xmath8 at @xmath9 and find the shape of the spreading `` wave packet '' . |
here we consider the case of @xmath38 and discuss the stability of the integer quantum hall ( iqh ) state found at @xmath57 .
figure [ fig : ener_g ] presents the spectra for @xmath137 on a torus geometry as a function of the coupling ratio @xmath88 or its inverse .
we find that an appreciable gap above a non - degenerate ground state remains in the regions @xmath138 ( left panel ) and @xmath139 ( right panel ) .
this indicates that the iqh state is stable against moderate changes in the ratio @xmath88 around the @xmath16-symmetric case .
( left ) and its inverse ( right ) for @xmath137 .
diamonds ( @xmath140 ) indicate two lowest - energy states at @xmath141 in the equal - population case @xmath142 ; the ground - state energy of this sector is subtracted from the whole spectrum .
circles ( @xmath143 ) indicate other eigenstates in the equal - population case .
crosses ( @xmath144 ) show eigenstates for the minimally imbalanced case of @xmath145 and @xmath146 .
, scaledwidth=50.0% ] pair distribution functions provide a useful probe of particle correlations present in the ground - state wave functions . for a uniform system of area @xmath90 ,
they are defined as @xmath147 we have calculated these functions using the ground states obtained by exact diagonalization on a sphere geometry . the results for the halperin @xmath78 state at @xmath148 and the bosonic iqh state at @xmath57 are presented in fig .
[ fig : pdf_all ] . on a disc geometry ,
the halperin @xmath78 state is given by @xcite @xmath149 where @xmath150 is a complex coordinate .
in this wave function , the pair distribution functions obey peculiar power laws @xmath151 and @xmath152 as @xmath153 , which are also found in the data in fig .
[ fig : pdf_all](a ) .
by contrast , the wave function for the @xmath57 iqh state has been predicted to be @xcite @xmath154 where @xmath155 is the projection onto the lowest landau - level manifold . while the unprojected part of this wave function has the same correlation properties as @xmath156 , the result presented in fig .
[ fig : pdf_all](b ) is very different from fig .
[ fig : pdf_all](a ) .
in particular , we find a hump in @xmath157 for small @xmath158 .
these results suggest that the correlation properties of the @xmath57 state are very different from what is expected from the unprojected part of eq . . for ( a ) the halperin @xmath78 state at @xmath23 and ( b ) the bosonic iqh state at @xmath57
, title="fig:",scaledwidth=50.0% ] for ( a ) the halperin @xmath78 state at @xmath23 and ( b ) the bosonic iqh state at @xmath57 .
, title="fig:",scaledwidth=50.0% ] figure [ fig : es_all ] presents the real - space entanglement spectra ( rses ) of candidate incompressible states on a sphere geometry . here
we divide the sphere into two hemispheres , calculate the reduced density matrix @xmath89 on the northern hemisphere @xmath90 , and plot the entanglement energies @xmath92 defined from the eigenvalues @xmath93 of @xmath89 .
the case of @xmath159 in fig .
[ fig : es_all](a ) corresponds to the incompressible state with @xmath63 and @xmath22 , whose nature has been elusive .
similarly to the @xmath57 case in fig . 3 of the main text , the low - lying entanglement levels appear in both positive and negative directions of @xmath98 , indicating the counter - propagating nature of edge modes .
this suggests that the @xmath63 state has certain similarities with the @xmath57 iqh state . for comparison ,
[ fig : es_all](b ) presents the rses of the halperin @xmath78 state [ or @xmath79 state ] at @xmath23 .
the low - lying levels appear only in the positive direction of @xmath98 , indicating the chiral ( right - moving ) nature of edge modes of this state . and
@xmath160 on the sphere geometry .
we plot the entanglement energies @xmath92 defined from the eigenvalues @xmath93 of @xmath89 , where @xmath90 is the northern hemisphere .
the entanglement energies are classified by the numbers of @xmath94 and @xmath95 atoms , @xmath96 and @xmath97 , and by the @xmath27-component of the total angular momentum , @xmath98 , on the northern hemisphere .
, title="fig:",scaledwidth=50.0% ] and @xmath160 on the sphere geometry .
we plot the entanglement energies @xmath92 defined from the eigenvalues @xmath93 of @xmath89 , where @xmath90 is the northern hemisphere .
the entanglement energies are classified by the numbers of @xmath94 and @xmath95 atoms , @xmath96 and @xmath97 , and by the @xmath27-component of the total angular momentum , @xmath98 , on the northern hemisphere .
, title="fig:",scaledwidth=50.0% ] here we derive the edge - mode spectrum of the @xmath57 bosonic iqh state , starting from the effective lagrangian density @xcite @xmath103 with @xmath161 . here , @xmath104 gives the density of bosons of spin @xmath25 and @xmath105 is the velocity matrix . introducing the charge and spin modes @xmath107
, the above lagrangian density is diagonalized as @xmath162 where @xmath163 are the velocities of these modes , which are obtained as the eigenvalues of the velocity matrix .
it follows from the euler - lagrange equations that the charge mode is right - moving , while the spin mode is left - moving @xcite : @xmath164 from eq .
, the hamiltonian and the total momentum are calculated as we perform mode expansions for the bosonic fields : @xmath166 where @xmath111 ( @xmath25 ) is a change in the number of spin-@xmath34 bosons relative to the gs and @xmath167 are bosonic operators describing oscillator modes . using these , each term in eq .
can be diagonalized as @xmath168 , \end{aligned}\ ] ] with @xmath169 , where we have ignored unimportant constant terms .
suppose that we have a large synthetic magnetic field with @xmath170m@xmath131 for a two - component bose gas .
this gives the landau - level spacing of @xmath171 nk@xmath172 . using a one - dimensional optical lattice along the @xmath27 direction , we split the gas into an array of 2d systems .
for @xmath129rb , the coupling constants for the contact interactions are almost pseudospin - independent : @xmath173 with the atomic mass @xmath174 kg and the @xmath175-wave scattering length @xmath176 nm .
the effective coupling constants in the 2d plane are given by @xmath177 where @xmath178 is the width of each 2d system along the @xmath27 direction . with @xmath179 nm and the above flux density , the energy gap of the @xmath57 iqh state in fig .
2(b ) of the main text gives @xmath180 nk@xmath133 , which is a reasonable scale for observing the predicted state in ultracold atom experiments .
here we comment on two related works @xcite , which also study the ground state at @xmath57 . these works and
the present one complement one another to some extent , as we describe below . wu and
jain @xcite have calculated the rses on the sphere and the edge spectrum on the disk , and found the presence of counter - propagating modes .
figure 3 in the main text agrees with this work .
they have also constructed a trial wave function , which has a large overlap with the @xmath57 ground state .
regnault and senthil @xcite have analyzed the scaling of the energy gap on the torus and furthermore discussed the phase diagram as a function of @xmath88 .
figure 2(b ) in the main text agrees with this work .
our work employs both the sphere and torus geometries and present solid evidences for the iqh state through the consistency of the two geometries .
reference has also constructed a composite fermion wave function for @xmath148 , and found a large overlap with the ground state on a sphere with the appropriate shift @xmath181 .
for the same filling factor , refs . and
have discussed the appearance of the @xmath182 state on the basis of spectra on a torus geometry .
since the composite fermion state and the @xmath182 state have different shifts , they appear at different @xmath62 on the sphere geometry . as seen in fig .
2(a ) of the main text , the @xmath182 state at @xmath183 has slightly larger gaps than the composite fermion state at @xmath184 .
a similar result is also obtained for @xmath185 ( not shown ) .
this suggests that the @xmath182 state is likely to be more stable ; however , the gap values of the two states are very close , and calculations for larger system sizes are required to obtain a conclusive result .
99 b. halperin , helv .
acta * 56 * , 75 ( 1983 ) .
t. senthil and m. levin , phys .
* 110 * , http://link.aps.org/abstract/prl/v110/e046801[046801 ] ( 2013 ) . y .-
wu and j. k. jain , phys .
b * 87 * , http://link.aps.org/abstract/prb/v87/e245123[245123 ] ( 2013 ) . n. regnault and t. senthil , arxiv : http://arxiv.org / abs/1305.0298[1305.0298 ]
. t. gra , b. juli - daz , n. barbern , and m. lewenstein , phys .
a * 86 * , http://link.aps.org/abstract/pra/v86/e021603[021603 ] ( r ) ( 2012 ) .
s. furukawa and m. ueda , phys .
a * 86 * , http://link.aps.org/abstract/pra/v86/e031604[031604 ] ( r ) ( 2012 ) . | we study two - component ( or pseudospin-@xmath0 ) bose gases in a strong synthetic magnetic field . using exact diagonalization
, we show that a bosonic analogue of an integer quantum hall state with no intrinsic topological order appears at the total filling factor @xmath1 when the strengths of intracomponent and intercomponent interactions are comparable with each other .
this provides a prime example of a symmetry - protected topological phase in a controlled setting of quantum gases .
the real - space entanglement spectrum of this state is found to be comprised of counter - propagating chiral modes consistent with the edge theory derived from an effective chern - simons theory .
there has been an ever growing interest in artificially created gauge fields in ultracold atomic gases .
gauge fields can be induced in neutral atoms by rotating gases @xcite or by optically dressing atoms @xcite . a sufficiently strong synthetic magnetic field for ultracold atoms
is expected to offer interesting analogues of quantum hall states with a rich variety of statistics and internal states of constituent particles .
the relevant parameter for describing a high magnetic - field regime is the filling factor @xmath2 , where @xmath3 is the number of atoms and @xmath4 is the number of magnetic flux quanta piercing the system . for scalar bosons , quantum hall states
have been predicted to appear at various integer and fractional @xmath5 @xcite such as a bosonic laughlin state at @xmath6 @xcite and non - abelian read - rezayi states @xcite at @xmath7 with integer @xmath8 @xcite . from a general perspective , laughlin and read - rezayi states are examples of topologically ordered states
in the sense that they host fractional excitations in the bulk and exhibit ground - state degeneracies that depend on the system s spatial topology @xcite .
recently , there has been an intense interest in a new type of phases which are _ not _ topologically ordered but are distinct from trivial phases as long as certain symmetries are imposed .
examples include topological insulators @xcite and haldane chains @xcite . such phases are termed symmetry - protected topological ( spt ) phases . very recently , a systematic classification of spt phases for interacting bosons has been proposed @xcite , opening up possibilities of hitherto unknown various spt phases .
it has been predicted that a system of bosons subject to the particle - number conservation can exhibit an integer quantum hall ( iqh ) effect with the hall conductivity quantized to _ even _
integers @xcite ; the resulting states are analogous to the fermionic iqh states in the sense that they both host elementary excitations of unit ( rather than fractional ) charge .
it is interesting to ask whether such new spt phases can be realized in quantum gases with tunable interactions and synthetic gauge fields .
quite recently , senthil and levin @xcite have proposed that two - component bose gases in a magnetic field may realize one of such phases a bosonic iqh state with its hall conductivity quantized to @xmath9 .
this intriguing proposal was based on an effective field theory , aiming at discussing a general possibility without specifying a microscopic hamiltonian .
however , an important issue of the competition among the bosonic iqh state and other candidate states in a realistic hamiltonian has remained elusive . in this letter , we numerically study a simple realistic system of two - component ( or pseudospin-@xmath0 ) bose gases in a high magnetic field .
recent exact diagonalization studies of this system using a torus geometry @xcite have identified a series of incompressible states at @xmath10 ( @xmath11 filling for each component ) with integer @xmath12 for the case of pseudospin - independent interactions .
possible candidate quantum hall states for these filling factors include non - abelian spin - singlet states with an @xmath13 symmetry @xcite ; such states have been found to appear for @xmath14 and @xmath9 @xcite . the nature of incompressible ground states ( gs ) at higher @xmath15 has remained to be identified . here , by combining calculations on torus and sphere geometries , we analyze the competition among candidate quantum hall states .
we show that the bosonic iqh state of ref
. does appear at @xmath1 when the strengths of intracomponent and intercomponent interactions are comparable with each other .
this result establishes the appearance of an spt phase in a simple microscopic model , and opens up a unique avenue for its realization in a controlled setting of quantum gases .
we discuss possible experiments for observing this state in light of the current status of creating synthetic gauge fields in ultracold atomic systems . plane , calculated on a sphere geometry in the @xmath16-symmetric case @xmath17 . filled circles indicate gss with the total angular momentum @xmath18 , where incompressible states can appear ; the area of each symbol is proportional to the neutral gap @xmath19 .
empty circles indicate the gss with @xmath20 .
solid and broken lines show the relation @xmath21 for @xmath22 and @xmath9 , respectively .
data points are missing for large @xmath4 and @xmath3 due to an exponentially long computation time . below the dotted line with @xmath23 ,
gss are highly degenerate @xcite .
, scaledwidth=50.0% ] we consider a system of a two - dimensional ( 2d ) bose gas ( in the @xmath24 plane ) having two hyperfine spin states ( labeled by @xmath25 ) and subject to a synthetic magnetic field @xmath26 along the @xmath27 axis . in the case of a rotating system ,
the magnetic field @xmath28 is induced in the rotating frame of reference , where @xmath29 is the particle s mass , @xmath30 is the rotation frequency , and @xmath31 is a fictitious charge for neutral atoms .
unique vortex structures for moderate rotation frequencies have been investigated both theoretically @xcite and experimentally @xcite .
quantum hall states are expected to emerge as the number of flux quanta ( or vortices ) , @xmath4 , becomes comparable with the number of atoms , @xmath3 , for sufficiently strong @xmath26 . for such large @xmath26 ,
it is natural to assume that the atomic motion in the @xmath24 plane is restricted to the lowest landau level ( lll ) . within this subspace
, we consider the interaction hamiltonian consisting of intracomponent and intercomponent contact interactions : @xmath32 where @xmath33 is the number of particles in state @xmath34 , @xmath35 s are the 2d positions of the particles , and @xmath36 and @xmath37 are the effective intracomponent and intercomponent coupling constants in the 2d plane .
we mainly focus on the case of @xmath16-symmetric interactions @xmath17 ; the case of @xmath38 will also be discussed .
to study bulk properties , it is useful to work on closed uniform manifolds having no edge . in refs . and , a periodic rectangular geometry ( torus ) of sides @xmath39 and @xmath40 is used . here
, we also use a sphere geometry and compare the two geometries to identify geometry - independent robust features . for a sphere geometry @xcite , a magnetic monopole of charge @xmath41 with integer @xmath42 is placed at the center , and it produces a uniform magnetic field @xmath26 on the sphere of radius @xmath43 , where @xmath44 is the magnetic length . introducing the spherical coordinates @xmath45 , single - particle orbits in the lll are given by @xmath46 with @xmath47 and @xmath48 , where @xmath49 is the @xmath27-component of the angular momentum .
on both the sphere and the torus , the filling factor in the thermodynamic limit is given by @xmath2 with @xmath50 .
on a finite sphere , however , the relation between @xmath3 and @xmath4 involves a characteristic shift @xmath51 for incompressible states : @xmath21 , where @xmath51 depends on an individual candidate wave function .
therefore , on a sphere , competing incompressible states leading to the same @xmath15 in the thermodynamic limit can be studied separately with different @xmath52 if they have different shifts . on a torus
, there is no shift , and all candidates for the same @xmath15 compete in the same finite - size calculation .
while a torus can give less biased results , a sphere is more useful in discussing relative stabilities of given candidates . henceforth , we take the units of @xmath53 and @xmath54 . in fig .
[ fig : gsl ] , we set @xmath55 and look for candidates of quantum hall states in the @xmath52 plane using a sphere geometry .
because of the rotational symmetry of a sphere , many - body eigenstates can be classified by the total angular momentum @xmath56 .
it is known that incompressible states in general appear as unique gss with @xmath18 , which are indicated by filled circles in fig .
[ fig : gsl ] .
the area of each filled circle is proportional to the neutral gap @xmath19 to be defined later .
solid and broken lines indicate the relation @xmath21 for @xmath22 and @xmath9 , respectively .
the bosonic iqh state with @xmath57 has @xmath22 @xcite ; the @xmath13 states with @xmath58 have @xmath59 and can appear only if @xmath60 is a multiple of @xmath61 @xcite .
we find that the @xmath18 gss appear for all @xmath62 corresponding to the iqh and @xmath13 states .
furthermore , we also find @xmath18 gss with appreciable energy gaps along the @xmath22 lines with @xmath63 and @xmath64 , whose nature will be discussed later . below we analyze the stabilities of these candidate states and , in particular , the competition between the iqh state and @xmath65 state at @xmath57 based on the data of energy gaps . and
@xmath19 , calculated on a sphere with @xmath66 .
( b , c ) size dependences of the energy gaps at @xmath57 and @xmath67 on the sphere and torus geometries . for a torus , we set @xmath68 ( periodic square ) . for a sphere
, we set @xmath22 ; namely , calculations are performed along the solid lines in fig .
[ fig : gsl ] .
the dotted lines indicate linear extrapolation of the charge gap to the thermodynamic limit , using the data with @xmath69 .
, title="fig:",scaledwidth=50.0% ] + and @xmath19 , calculated on a sphere with @xmath66 .
( b , c ) size dependences of the energy gaps at @xmath57 and @xmath67 on the sphere and torus geometries . for a torus , we set @xmath68 ( periodic square ) . for a sphere ,
we set @xmath22 ; namely , calculations are performed along the solid lines in fig .
[ fig : gsl ] .
the dotted lines indicate linear extrapolation of the charge gap to the thermodynamic limit , using the data with @xmath69 . , title="fig:",scaledwidth=50.0% ] we consider two types of energy gaps , namely , charge and neutral gaps .
the charge gap is defined as @xmath70 for even integer @xmath71 , where @xmath72 is the ground - state energy for given @xmath73 .
the neutral gap @xmath19 is defined as the gap to the first excited state for @xmath74 .
figure [ fig : gap](a ) presents the result for a sphere with @xmath66 and varying @xmath3 .
both the iqh state and the @xmath13 states can appear in this case .
the two types of gaps , @xmath75 and @xmath19 , show essentially similar behaviors [ the former is divided by @xmath9 because of a doubled energy scale involved in the definition of @xmath76 .
the most prominent gap is found at @xmath77 , where the halperin @xmath78 state @xcite [ or the @xmath79 state ] gives the exact gs @xcite .
the second largest gap ( comparable with the first ) is found at @xmath80 , the latter corresponding to the @xmath57 iqh state .
this gap is ( by a factor of about @xmath81 ) larger than the gap at @xmath82 , which corresponds to the @xmath65 state .
this indicates that , of the two candidates at @xmath57 , the iqh state is likely to be more stable . by comparing fig .
[ fig : gap](a ) with the previous result on a torus @xcite , where comparable gaps at @xmath23 and @xmath9 were found , we find that interpreting the @xmath57 gs as the iqh state gives a better consistency between the sphere and torus data . in fig . [
fig : gap](a ) , we also find a large gap at @xmath83 ( comparable with the two largest ) , whose physical origin is unclear . in fig .
[ fig : gap](b ) and ( c ) , we examine size dependences of the gaps at @xmath57 and @xmath67 on the sphere and torus geometries .
a sphere tends to give larger gaps than a torus for finite sizes . for @xmath57 in fig .
[ fig : gap](b ) , we find that the difference between the sphere and torus data decreases as @xmath3 increases .
extrapolation to the thermodynamic limit ( dotted lines ) yields @xmath84 and @xmath85 for the sphere and torus cases ; they are in good agreement , given the residual oscillations in finite - size data used for the extrapolation .
furthermore , the torus data give an additional support for excluding the appearance of the @xmath65 state .
if the @xmath65 state is stabilized in the thermodynamic limit , @xmath19 in the present definition should drop because of the @xmath86-fold topological gs degeneracy expected on a torus @xcite ( a real neutral gap should be found above the degenerate gss in this case ) .
we find no appreciable drops of @xmath19 for @xmath87 ( system sizes where the @xmath65 state is allowed ) in fig .
[ fig : gap](b ) . for @xmath63 in fig .
[ fig : gap](c ) , the system size is not large enough to check the consistency between the torus and sphere data ; yet , simple linear extrapolation of the charge gap ( dotted lines ) indicates a tendency toward a nonzero value in the thermodynamic limit on both geometries . for @xmath57
, we have also analyzed the case of @xmath38 by calculating the spectra on a torus @xcite .
we find that the iqh state with an appreciable energy gap survives against moderate changes in the ratio @xmath88 .
we have so far discussed the appearance of the iqh state at @xmath57 using only spectral properties .
a useful way to unveil the topological features of this state is to investigate the entanglement spectrum @xcite .
this is an eigenvalue spectrum of a reduced density matrix @xmath89 on a subsystem @xmath90 , and has been demonstrated to serve as a useful probe of the edge spectrum .
here we use the gs on a sphere , split it into two hemispheres , and calculate the real - space entanglement spectrum ( rses ) @xcite for the northern hemisphere @xmath90 . the result
is presented in fig .
[ fig : es ] .
below we analyze this result based on an effective field theory .
iqh state with @xmath91 on a sphere .
we plot the entanglement energies @xmath92 defined from the eigenvalues @xmath93 of @xmath89 , where @xmath90 is the northern hemisphere ; the lower the entanglement energy is , the larger weight it has in @xmath89 .
the entanglement energies are classified by the number of @xmath94 and @xmath95 atoms , @xmath96 and @xmath97 , and by the @xmath27-component of the total angular momentum , @xmath98 , on @xmath90 .
, scaledwidth=50.0% ] the @xmath57 bosonic iqh state is described by the chern - simons theory with the @xmath99 @xmath100 matrix with @xmath101 and @xmath102 @xcite . through the standard procedure @xcite , the effective lagrangian for one - dimensional edge modes
is obtained as @xmath103 where @xmath104 gives the density of bosons of spin @xmath25 and @xmath105 is the velocity matrix . diagonalizing @xmath106 via charge and spin modes @xmath107 , one finds that the charge ( spin ) mode is right - moving ( left - moving ) @xcite . introducing the velocities @xmath108 for these modes and performing mode expansions ,
the hamiltonian and the total momentum for the edge of length @xmath39 are diagonalized as @xcite @xmath109 with @xmath110 here , @xmath111 ( @xmath25 ) is a change in the number of spin-@xmath34 bosons relative to the gs ; @xmath112 and @xmath113 are sets of non - negative integers describing oscillator modes . the second term in eq . exhibits the well - known counting of degeneracies associated with a @xmath114 free boson : @xmath115 . in fig
.
[ fig : es ] , the lowest entanglement energy is found at @xmath116 with @xmath117 , which can be identified with the `` gs '' of virtual edge modes probed by the rses . `` excited states '' in the rses are found in both positive and negative directions of @xmath98 relative to the `` gs '' , which indicates the counter - propagating nature of the edge modes . this is in sharp contrast to the case of a single chiral edge mode @xcite , where low - lying `` excited states '' are found only in one direction of @xmath98 .
the excitations along the left envelope of the spectrum can be interpreted as the left - moving spin mode with @xmath118 , as spin excitations with @xmath119 and @xmath120 relative to the `` gs '' are found there ; the shifts @xmath121 and @xmath122 of these excitations are consistent with eqs . and .
similarly , the excitations along the right envelope can be interpreted as the right - moving charge mode with @xmath123 , as charge excitations with @xmath124 and @xmath125 are found there .
the levels appearing between the two modes can be identified with combinations of these modes .
let us briefly discuss the nature of the gapped states at @xmath126 found in fig .
[ fig : gsl ] .
the rses of the @xmath63 state reveals a counter - propagating nature of edge modes @xcite , as in the @xmath57 case in fig .
[ fig : es ] .
this suggests that the @xmath63 state has certain similarities with the @xmath57 iqh state .
however , the bosonic iqh states with @xmath127 predicted in ref .
@xcite are _ not _ good candidates for the present system , as their corresponding simplest @xmath100 matrices are not symmetric with respect to the interchange of the two components .
a more detailed characterization of the states at @xmath126 is left for future studies .
finally , we discuss possible experiments for observing the iqh state found in this letter . in ultracold atom experiments , a clear signature of a quantum hall state would be a density plateau in an _ in situ _ image of a gas .
a challenge here is how to realize a large synthetic field required for observing this state . in ref .
@xcite , by using an optical coupling between @xmath128 internal states of @xmath129rb atoms with space - dependent detuning , a flux density @xmath130m@xmath131 has been realized .
this produces the landau - level spacing @xmath132 nk@xmath133 .
it is an important technical challenge to increase this field by a factor of a few tens so that @xmath134 dominates the experimental temperature scales .
this could be achieved by further increasing the detuning gradient in the scheme of ref . or by using spatially shifted laser beams @xcite .
we hope that the appearance of exotic phases demonstrated in this letter and refs . and stimulate experimental attempts to create such large synthetic magnetic field in two - component bose gases , by , for example , separately creating synthetic fields in the @xmath128 and @xmath9 states of @xmath129rb .
once such a large field is created , a quantum hall regime with @xmath135 is expected to be realized by splitting the gas into an array of 2d systems by using a one - dimensional optical lattice along the @xmath27 direction as suggested in ref .
@xcite . by strongly squeezing each 2d system along the @xmath27 direction ,
the iqh state exhibits a reasonable gap for its observation @xcite .
_ note added .
_ after the submission of this letter , we became aware of two independent works @xcite , which also study the ground state at @xmath57 and have some overlap with this letter @xcite .
this work was supported by kakenhi grant no .
25800225 from japan society for the promotion of science , and by a grant - in - aid for scientific research on innovative areas `` topological quantum phenomena '' ( kakenhi no .
22103005 ) , kakenhi grant no .
22340114 , a global coe program `` the physical science frontier '' , and the photon frontier network program from mext of japan .
99 for a review , see n. r. cooper , adv .
phys . * 57 * , 539 ( 2008 ) .
y .- j .
lin , r. l. compton , k. jinmnez - garca , j. v. porto , and i. b. spielman , nature * 462 * , 628 ( 2009 ) .
j. dalibard , f. gerbier , g. juzeli@xmath136nas , and p. hberg , rev .
mod .
phys . * 83 * , http://link.aps.org/abstract/rmp/v83/e1523[1523 ] ( 2011 ) .
n. k. wilkin , j. m. f. gunn , and r. a. smith , phys .
rev .
lett . * 80 * , http://link.aps.org/abstract/prl/v80/p2265[2265 ] ( 1998 ) .
n. read and e. h. rezayi , phys . rev .
b * 59 * , http://link.aps.org/abstract/prb/v59/p8084[8084 ] ( 1999 ) . n. r. cooper , n. k. wilkin , and j. m. f. gunn , phys . rev
. lett .
* 87 * , http://link.aps.org/abstract/prl/v87/e120405[120405 ] ( 2001 ) .
x .- g .
wen , _ quantum field theory of many - body systems : from the origin of sound to an origin of light and electrons _
( oxford university press , new york , 2004 ) .
m. z. hasan and c. l. kane , rev .
mod .
phys . *
82 * , http://link.aps.org/abstract/rmp/v82/e3045[3045 ] ( 2010 ) ; x .-
l .
qi and s .- c .
zhang , rev .
mod .
phys . * 83 * , http://link.aps.org/abstract/rmp/v83/e1057[1057 ] ( 2011 ) .
f. d. m. haldane , phys .
lett * 93a * , 464 ( 1983 ) .
x. chen , z .- c .
gu , z .- x .
liu , and x
.- g .
wen , science * 338 * , 1604 ( 2012 ) ; phys .
rev .
b * 87 * , http://link.aps.org/abstract/prb/v87/e155114[155114 ] ( 2013 ) . y .-
m .
lu and a. vishwanath , phys .
rev .
b * 86 * , http://link.aps.org/abstract/prb/v86/e125119[125119 ] ( 2012 ) . t. senthil and m. levin , phys
. rev .
lett . * 110 * , http://link.aps.org/abstract/prl/v110/e046801[046801 ] ( 2013 ) .
t. gra , b. juli - daz , n. barbern , and m. lewenstein , phys .
rev .
a * 86 * , http://link.aps.org/abstract/pra/v86/e021603[021603 ] ( r ) ( 2012 ) .
s. furukawa and m. ueda , phys .
rev .
a * 86 * , http://link.aps.org/abstract/pra/v86/e031604[031604 ] ( r ) ( 2012 ) .
e. ardonne and k. schoutens , phys .
rev .
lett . * 82 * , http://link.aps.org/abstract/prl/v82/p5096[5096 ] ( 1999 ) ; e. ardonne , n. read , and e. rezayi , and k. schoutens , nucl .
phys .
b * 607 * , 549 ( 2001 ) .
e. j. mueller and t .- l
. ho , phys . rev .
lett . * 88 * , http://link.aps.org/abstract/prl/v88/e180403[180403 ] ( 2002 ) .
k. kasamatsu , m. tsubota , and m. ueda , phys .
rev .
lett . * 91 * , http://link.aps.org/abstract/prl/v91/e150406[150406 ] ( 2003 ) ; int . j. mod
. phys .
b * 19 * , 1835 ( 2005 ) .
v. schweikhard , i. coddington , p. engels , s. tung , and e. a. cornell , phys .
rev .
lett . * 93 * , http://link.aps.org/abstract/prl/v93/e210403[210403 ] ( 2004 ) .
f. d. m. haldane , phys .
rev .
lett . * 51 * , http://link.aps.org/abstract/prl/v51/p605[605 ] ( 1983 ) ; g. fano , f. ortolani , and e. colombo , phys . rev .
b * 34 * , http://link.aps.org/abstract/prb/v34/p2670[2670 ] ( 1986 ) .
b. halperin , helv .
phys .
acta * 56 * , 75 ( 1983 ) .
b. paredes , p. zoller , and j. i. cirac , phys .
rev . a * 66
* , http://link.aps.org/abstract/pra/v66/e033609[033609 ] ( 2002 ) .
see the supplemental material for further numerical results , derivations of eqs . and ,
evaluation of the excitation gap , and comments on refs . and .
h. li and f. d. m. haldane , phys .
rev .
lett .
* 101 * , http://link.aps.org/abstract/prl/v101/e010504[010504 ] ( 2008 ) .
a. sterdyniak , a. chandran , n. regnault , b. a. bernevig , and p. bonderson , phys .
rev .
b * 85 * , http://link.aps.org/abstract/prb/v85/e125308[125308 ] ( 2012 ) ; j. dubail , n. read , and e. h. rezayi , phys .
rev .
b * 85 * , http://link.aps.org/abstract/prb/v85/e115321[115321 ] ( 2012 ) ; i. d. rodrguez , s. h. simon , and j. k. slingerland , phys . rev . lett .
* 108 * , http://link.aps.org/abstract/prl/v108/e256806[256806 ] ( 2012 ) .
y .- h . wu and j. k. jain , phys .
rev .
b * 87 * , http://link.aps.org/abstract/prb/v87/e245123[245123 ] ( 2013 ) . n. regnault and t. senthil , arxiv : http://arxiv.org / abs/1305.0298[1305.0298 ] . |
most studies of micro - instabilities consider those driven by gradients of density or ion and electron temperatures .
however the toroidal current gradient is also a potential source of instability .
indeed , this is the instability drive for tearing modes and it was even proposed @xcite that it could drive high mode number ideal mhd kink instability .
this analysis was based on the periodic , cylindrical , ideal mhd equation for the perturbed magnetic flux function , @xmath4 $ ] , in the tokamak limit , @xmath5 : @xmath6 where @xmath7 is the radial coordinate with @xmath8 denoting a radial derivative , @xmath9 the azimuthal angle and @xmath10 the axial co - ordinate , @xmath11 being the periodicity length in the @xmath10-direction , and we have introduced poloidal and toroidal mode numbers , @xmath12 and @xmath13 , respectively .
the magnetic field in the @xmath10-direction is @xmath14 , the safety factor @xmath15 and the normalised current density , @xmath16 , is defined by @xmath17 expanding the denominator of ( [ eq1 ] ) , @xmath18 , in the vicinity of the resonant position , @xmath19 , where , @xmath20 , so that @xmath21 with @xmath22 denoting a new ( dimensionless ) local radial variable , this equation takes the local form : @xmath23 where @xmath24 with @xmath25 being the magnetic shear at @xmath26 . equation ( [ eq5 ] ) has been the basis for a number of studies at high @xmath12 .
thus kadomtsev and pogutse @xcite used it to claim that ideal mhd instability is possible , for steep current gradients or low shear , if @xmath27 while wesson @xcite and strauss @xcite investigated the dependence of the tearing mode index , @xmath0 @xcite on @xmath1 . hegna and
callen @xcite used a generalisation of ( [ eq5 ] ) that took account of the effect of toroidal and plasma shaping on metric coefficients , to estimate @xmath28 in more general , toroidal , devices .
furthermore , the local gyrokinetic code gs2 has recently @xcite been used to study current gradient driven instabilities in collisionless plasmas , again finding instability above a critical value of @xmath1 .
however this code is based on the ballooning transformation and this also relies on a linear expansion of @xmath29 about the rational surface .
unfortunately such treatments are not entirely consistent because , for @xmath30 , the truncated expansion of @xmath31 employed is inadequate .
to see this is the case recall @xmath32 and @xmath33 are related through amp@xmath34re s equation , @xmath35 which can be written @xmath36 , so that @xmath37 with @xmath38 denoting the quantity @xmath39 . now if we restrict attention to positive values of the current density , @xmath32 , then @xmath40 and @xmath41
the condition @xmath42 then implies @xmath43 so that terms in @xmath44 must be retained in the expansion of @xmath29 around @xmath19 , giving @xmath45 .
consequently , if the parameter @xmath1 is of order unity , as in @xcite , the tearing equation must now take the form : @xmath46 \psi=0 .
\label{eq10}\ ] ] equation ( [ eq10 ] ) , however , of necessity describes a scenario in which there are two resonant surfaces ; a situation which can not be the case for a monotonic @xmath33 even when @xmath47 .
we conclude that both ( [ eq5 ] ) and ( [ eq10 ] ) must give a flawed description of stability of monotonic @xmath33 profiles .
clearly , higher order terms in the expansion of @xmath48 are required .
thus , remarkably , there is no truly local theory for such high-@xmath12 instabilities .
we note that a comparison of analytic resistive tearing mode growth rates with those obtained from a numerical code at low@xmath49 also required the inclusion of more derivatives of @xmath33 to obtain agreement when the resistivity and @xmath0 were relatively high @xcite . in this paper
we reconsider the solution of ( [ eq1 ] ) at high @xmath12 , taking full account of the structure of @xmath32 and @xmath33 .
two scenarios are studied : ( i ) a monotonic @xmath32 with a steep gradient , modelled by a `` tanh function '' ; and ( ii ) the effect of a positive current `` spike '' , possibly arising in the pedestal region of a tokamak in h - mode due to bootstrap currents .
this latter case can lead to a region of greatly reduced shear , or even non - monotonic @xmath33 and multiple resonances .
the outcomes of our calculations are self - consistent forms for @xmath0 which can be used to investigate high-@xmath12 tearing mode instability .
a value of @xmath0 that accounts for key nonlocal features of the global structure of the self - consistent @xmath29 and current profiles is required to properly interpret the gs2 results alluded to above .
the quantity @xmath0 also plays a role in determining the saturated amplitude of magnetic islands arising from tearing mode instability .
however other parameters may play a role in determining this amplitude @xcite .
in this section we consider a specific example for the current density @xmath32 which has a steep gradient , and exploit the property @xmath50 to expand ( [ eq1 ] ) close to the resonant surface at @xmath19 .
the current density employed is @xmath51 where @xmath52 is the point of steepest current gradient and we will be interested in large values of the parameter @xmath53 . the current profile of ( [ eq11 ] ) is shown in figure [ fig1 ] for @xmath54 , @xmath55 and @xmath56 . of ( [ eq11 ] ) , with @xmath54 , @xmath55 and @xmath56.,width=302 ] integrating the amp@xmath34re equation , ( [ eq8 ] ) ,
yields an expression for the resonant denominator in the tearing equation , ( [ eq1 ] ) , viz . : @xmath57,\nonumber\end{aligned}\ ] ] where @xmath19 denotes the location of the resonance , i.e. @xmath58 the numerator of the resonant term in ( [ eq1 ] ) is readily evaluated to give @xmath59 . \label{eq14}\ ] ] we next introduce the local radial variable @xmath60 , where we have now chosen @xmath52 rather than @xmath19 as origin , and corresponding resonance position , @xmath61 , and express @xmath62 and @xmath63 as functions of @xmath64 around @xmath65 , noting that these expressions are not confined to small values of @xmath64 , since we only require @xmath66 .
therefore , using @xmath67 and ignoring @xmath68 corrections in @xmath69 , we find @xmath70 -\log[\cosh(p x_s ) ] \right\},\label{eq16}\\ r\hat{j}^\prime=&-\frac{\lambda j_0}{2}\operatorname{sech}^2(px),\label{eq17}\\ p\equiv&\frac{\lambda}{m},\label{eq18}\end{aligned}\ ] ] where integrating ( [ eq8 ] ) from @xmath71 to @xmath52 gives @xmath72 for @xmath73 .
when @xmath74 , we may write the kink / tearing equation in the form of ( [ eq5 ] ) , with @xmath75}. \label{eq19}\ ] ] since we can identify @xmath76 there is again a single parameter , @xmath77 , determining stability , with @xmath78,\qquad f ( y)=\frac{y \operatorname{sech}^2(y)}{y+\log[\cosh(y)]}. \label{eq20}\ ] ] thus we can consider ( [ eq19 ] ) as a semi - localized generalisation of ( [ eq6 ] ) in which @xmath1 acquires a form factor , @xmath79 .
the function @xmath80 is shown in figure [ fig2 ] for @xmath81 , from which it is clear that @xmath80 is very far from being constant , as assumed in the local treatments of @xcite and @xcite . for the @xmath16 of figure [ fig1 ] , @xmath82 and @xmath74.,width=302 ] furthermore , if we assume that the mode numbers , @xmath12 and @xmath13 are such that the resonant surface lies at @xmath83 , i.e. that @xmath84 rather than @xmath85 , the tearing equation can be written in what we will refer to as the semi - local form : @xmath86-\log[\cosh(p x_s)]}\right\}=0 , \label{eq21}\ ] ] with dependence on two parameters , @xmath87 and @xmath2 .
in this section we consider reconnecting instabilities which might be driven by the energy source associated with a steep monotonic current gradient , i.e. instabilities which are driven by positive values of the tearing index , @xmath0 , defined as the jump in @xmath88 as in @xcite .
however we first discuss the ideal mhd stability properties of ( [ eq21 ] ) , noting that these can be determined by computing the stability properties of the two newcomb @xcite sub - intervals , @xmath89 $ ] and @xmath90 $ ] , or , in terms of the `` local '' variable , @xmath64 , @xmath91 $ ] and @xmath92 $ ] .
this analysis has been carried out numerically and no instability found at any finite values of the parameters @xmath93 and @xmath2 ( i.e. , in newcomb terms , in shooting a solution which is regular at one end - point of a sub - interval , no zero of @xmath94 is encountered before the other end of the sub - interval is reached ) . in particular , the value @xmath95 is not a marginal point for ideal mhd instability , contrary to ref .
@xcite . returning to the issue of tearing mode stability
, we calculate @xmath96 , defined as the jump in @xmath97 at @xmath2 for the solution of ( [ eq21 ] ) which vanishes at @xmath98 and at @xmath99 . the global value of @xmath0
is then related to @xmath100 by : @xmath101 , for @xmath74 as a function of @xmath76 calculated from ( [ eq21 ] ) .
the transition to unstable tearing occurs at @xmath102,width=302 ] , calculated from ( [ eq21 ] ) for @xmath103 ( solid curve ) and @xmath104 ( dashed curve).,width=302 ] figure 3 shows @xmath100 as a function of @xmath93 when @xmath74 , and figure [ fig4 ] shows @xmath100 as a function of @xmath2 for @xmath103 ( solid line ) and @xmath104 ( dashed line ) .
the somewhat surprising content of figure [ fig4 ] , namely that the most unstable location for the resonant surface is at large negative values of @xmath2 , is an artefact of our semi - local approximation , as discussed below . as a function of @xmath2 , similar to figure [ fig4 ] but calculated from solution of the global kink / tearing equation , ( [ eq1 ] ) , with @xmath32 given by ( [ eq11 ] ) . @xmath56 and @xmath105 ( solid curve ) and @xmath106 ,
( dashed curve ) , parameters which correspond to @xmath103 in the local approximation . as described in the text , ( [ eq22 ] ) has been used to present the results in terms of the `` local '' variables , @xmath100 and @xmath2.,width=302 ] we have also performed global calculations of @xmath107/m$ ] , with @xmath32 given by ( [ eq11 ] ) .
the results of such global calculations are shown in figure [ fig5 ] where , for the purpose of comparing with figure [ fig4 ] , ( [ eq22 ] ) has been used to transform `` global '' data into `` local '' variables , as in figure [ fig4 ] . in figure [ fig5 ]
parameters are @xmath56 and @xmath108 , ( solid curve ) , @xmath106 , ( dashed curve ) .
figure 5 shows that the most unstable value of @xmath100 occurs at a finite negative value of @xmath2 .
a more careful inspection of the derivation of ( [ eq21 ] ) reveals that , at large negative values of @xmath109 the shear , @xmath110 , becomes exponentially small , @xmath111 , so that terms of order @xmath112 , which have been neglected , can compete , leading to the behaviour seen in figure [ fig5 ] , for large , but finite , @xmath12 .
not surprisingly , the @xmath28 values obtained , and therefore stability , are sensitive to the global structure of the ideal mhd region since equation ( [ eq21 ] ) contains less global information than equation ( [ eq11 ] ) , thereby resulting in the @xmath100 differences shown in figures [ fig4 ] and [ fig5 ] .
for comparison with the previous results of kadomtsev and pogutse @xcite and wesson @xcite , we have also calculated the tearing stability index @xmath113 obtained by computing solutions of ( [ eq5 ] ) with @xmath114 . with figure [ fig3 ] .
the solid curve was computed with @xmath80 given by ( [ eq19 ] ) , the dashed curve was computed with constant @xmath115 .
@xmath116 passes through zero at @xmath103 , as found in @xcite and asymptotes to @xmath117 at @xmath118 as in @xcite .
the thin dotted curve represents a rational approximation to the dashed curve , used in section [ sec6].,width=302 ] figure [ fig6 ] shows ( dashed curve ) the resulting @xmath113 as a function of @xmath93 from ( [ eq5 ] ) . for comparison
the solid line shows @xmath119 of figure [ fig3 ] , computed with @xmath120 of ( [ eq19 ] ) .
ideal mhd marginality , where the value of @xmath113 becomes infinite , apparently occurs at @xmath118 which corresponds to the prediction @xcite of ideal kink instability beyond this value . at the value @xmath118 , an exact solution of ( [ eq5 ] ) in the inner newcomb sub - interval is @xmath121 , which vanishes at both end - points , again demonstrating ideal marginality
however , as noted in the foregoing discussion , this is an incorrect prediction and is not found in full cylindrical solutions , or in our semi - localised version ( [ eq19 ] ) with the correct @xmath80 dependence , which yields the solid curve in figure [ fig6 ] . for small values of @xmath93 ,
@xmath116 is negative and changes sign at @xmath122 , as reported in @xcite .
in this section we investigate a different , but possibly important , scenario in which a fairly localised positive current spike occurs relatively near the plasma edge @xcite , where resistivity is high and the inductive current density is small .
the non - linear theory of external kink modes in the presence of such a current spike has been studied by eriksson and wahlberg @xcite .
bootstrap currents in the region of a steep pedestal , or localized current drive , may produce just such a situation .
figure [ fig7 ] shows an example for the current density given by , @xmath123 ^ 3+j_1 e^{-\mu ( r / r_1 - 1)^2 } , \label{eq23}\end{aligned}\ ] ] with @xmath124 , @xmath125 , @xmath126 and the peak of the current spike at @xmath127 . in figure [ fig7 ]
the dashed curve represents the resulting safety factor , @xmath33 , indicating that the magnetic shear becomes small in the region of the current spike . .
the dashed curve shows the @xmath33 profile displaying reduced magnetic shear in the vicinity of the current spike , @xmath128 .
parameters are @xmath129 and @xmath127.,width=302 ] it might be thought that the most unstable location for a tearing mode resonance would be at the minimum of the current density ( @xmath130 in figure [ fig7 ] ) , since the gradient of the current density is destabilising on both sides of the resonance in this case , that is , @xmath131 < 0 $ ] so its sign is the opposite of the stabilizing @xmath132 line bending term in ( [ eq1 ] ) .
surprisingly , however , this is not the case , as can be seen from figure [ fig8 ] which displays the value of @xmath133 for an @xmath134 mode , calculated from ( [ eq1 ] ) , as @xmath19 is moved across the @xmath32 profile , regarding @xmath13 as a continuous variable . for an @xmath134 mode , from solution of ( [ eq1 ] ) with @xmath32 given by ( [ eq23 ] ) .
, width=302 ] figure [ fig8 ] does show that when @xmath19 is located at the maximum of the current spike , @xmath135 in figure [ fig7 ] , the tearing index , @xmath0 is strongly stabilising ( i.e. negative ) as one would expect since the current gradients on both sides of @xmath19 are stabilising in this case , that is , @xmath136 > 0 $ ] , enhancing the stabilizing @xmath132 line bending term in ( [ eq1 ] ) .
however , values of @xmath19 quite close to the maximum of the current spike are very unstable .
to model this in a high-@xmath12 , localised analysis , we expand @xmath32 and @xmath33 locally around the maximum of the current density , approximately at @xmath137 , the maximum of the current spike . around this point , @xmath32 is dominated by the current spike , @xmath138 , and is a local function , @xmath139 , if we order the parameter @xmath140 .
however , the @xmath29 profile in this region is determined by both the extended `` inductive '' current profile @xmath141 and by the current spike .
it therefore contains both a slowly varying part , due to @xmath141 and a rapidly varying part , due to @xmath138 and the result can be a greatly reduced magnetic shear , as seen in figure [ fig7 ] .
the resulting local tearing equation has been derived in the appendix , for the current profile of ( [ eq23 ] ) .
however a simpler derivation expands @xmath33 locally around the point , @xmath137 , at the maximum of @xmath32 , and orders the weakened shear , @xmath142 and @xmath143 .
thus : @xmath144 then , constructing @xmath145 from ( [ eq23 ] ) , the high-@xmath12 tearing mode equation , ( [ eq1 ] ) , can be written in the form ; @xmath146 where @xmath2 is again the location of the resonance .
we note from ( [ eq24 ] ) that a monotonic @xmath29 profile requires @xmath147 .
+ @xmath148as in section [ sec3 ] , stability depends on two parameters , @xmath3 and @xmath2 . in order to reduce the parameter space in ( [ eq25 ] ) , we have focused on three cases : @xmath149 equation ( [ eq25 ] ) has then been solved to obtain values of @xmath150 as the resonant location , @xmath2 , is moved across the local @xmath63 structure . results for cases ( a ) and ( b ) , equations ( [ eq26 ] ) and ( [ eq27 ] ) respectively , are shown in figure [ fig9 ] .
the solid curve corresponds to case ( a ) and the dashed one to case ( b ) . in case ( c ) , ( [ eq28 ] ) , we exclude consideration of the region of triple resonance , i.e. @xmath151 and figure [ fig10 ] shows the value of @xmath100 when @xmath2 falls outside this range . consideration of the triple resonances in case ( c ) raises issues involving the different characteristic frequencies associated with tearing at each of the three resonant surfaces , these frequencies being determined by diamagnetic terms and by sheared equilibrium rotation . in addition , case ( c ) is likely to arise only after an equilibrium current profile has evolved through the very unstable weak shear scenario , case ( b ) .
it is therefore sufficient to note that as @xmath2 approaches the location of @xmath152 or @xmath153 , the value of @xmath100 becomes very large . , from ( [ eq22 ] ) with @xmath154 ( solid curve ) and @xmath155 ( dashed curve ) , corresponding to monotonically increasing @xmath29 profiles.,width=302 ] but with @xmath156 corresponding to reverse shear at @xmath157 , i.e to a locally non - monotonic @xmath29 profile .
@xmath100 values are only calculated when there is a single resonant surface , i.e. in the range @xmath158.,width=302 ] positive values of @xmath100 , with energy available to drive reconnection , are predicted for modes which are resonant close to , but not at , the local maximum of the current density .
comparison of figures [ fig8 ] and [ fig9 ] demonstrates the validity of the high-@xmath12 equation , ( [ eq25 ] ) .
gs2 is a radially local gyrokinetic code modeling small scale instabilities ( @xmath159 ) in a periodic flux - tube domain in toroidal geometry . here , @xmath160 is the binormal wave number , where @xmath13 is the toroidal mode number , and the subscript @xmath161 of a quantity refers to its value at the reference radius .
the radial variation of the metric is not retained and all plasma and magnetic geometry parameters are linearized around their value at @xmath162 . in particular , @xmath163 $ ] is used ; an approximation equivalent to ( [ eq3 ] ) . for brevity ,
henceforth we will drop the @xmath161 subscripts .
we use the low - flow version of gs2 @xcite , similarly as described in @xcite , and model the current as a parallel velocity shift in the non - fluctuating maxwellian electron distribution .
we consider a collisionless , pure deuterium plasma where both species are gyrokinetic .
we use a large aspect ratio circular cross section geometry with no finite pressure corrections to the flux surfaces , and we neglect compressional magnetic perturbations .
the following parameters are used for the simulations : @xmath164 , @xmath165 , @xmath166 , @xmath167 , @xmath168 , @xmath169 , @xmath170 , @xmath171 . here
, @xmath172 is the electron flow speed , @xmath173 is the ion thermal speed with @xmath174 and @xmath175 the temperature and the mass of ions , @xmath176 is the normalized ion pressure , @xmath177 , @xmath178 , @xmath179 , and @xmath180 .
furthermore , @xmath181 is the ion thermal larmor radius with the gyro - frequency @xmath182 , @xmath183 is the major radius at the centroid of the flux surface .
we choose to set the temperature and density gradients to zero , thus there are no diamagnetic corrections to the mode frequency and we can avoid the pollution of the results with pressure gradient driven instabilities . therefore the current gradient is purely due to a gradient in the electron flow speed .
we scan the safety factor @xmath29 , which changes the current gradient drive parameter @xmath184 through the shear length @xmath185 . here
, we introduced the unperturbed poloidal flux @xmath186 . in cylindrical geometry , this definition of @xmath187 is equivalent to @xmath1 in ( [ eq6 ] ) . when we set all plasma and geometry parameters to their values specified in the previous paragraph , and let @xmath29 vary , we find @xmath188 .
for the same set of parameters the spurious ideal kink instabilities are found above @xmath189 ( @xmath190 ) , see figure 2e in @xcite . here
we will concentrate on the region @xmath191 , where destabilization of current - gradient driven tearing modes is possible ( @xmath192 ) , as indicated by the dashed line in figure [ fig6 ] .
range where the ideal kink mode is stable .
blue curve and symbols denote gs2 results , dotted line represents an analytical estimate based on @xcite .
( b ) typical radial structures of tearing modes inferred from gs2 simulations .
blue lines : real part , red lines ( mostly overlapping , with values close to zero ) : imaginary part . dashed line : analytical solution of marginally stable ideal kink ) .
darker curves : @xmath193 , lighter curves : @xmath194 . ]
figure [ growthrate]a shows the growth rate of the tearing mode as a function of @xmath195 .
the symbols represent the gs2 simulation results .
the real part of the frequency ( not shown here ) is very small in magnitude , consistent with being zero within the numerical accuracy of the calculations .
we did not perform simulations below @xmath196 since the decreasing growth rates lead to unreasonably long simulation times to reach convergence .
the simulations barely resolve the collisionless electron skin depth @xmath197 by covering an unusually large range of ballooning angles ; the number of @xmath198 segments along the field line is 40 in the simulations ( we note that convergence of the results with respect to resolution has been checked ) .
the growth rate of the collisionless tearing mode , within a constant , is given by @xmath199 @xcite , where @xmath200 is the width of the inner layer , with @xmath201 , @xmath202 is the electron plasma frequency , and @xmath28 is the jump in @xmath203 across the inner layer . introducing the dimensionless @xmath204 , we find that @xmath205\simeq\delta ( k_y\rho_i)^2\beta_i^{-1}(sa)/(qr)[(m_et_e)/(m_it_i)]^{1/2}$ ] .
we may use the following rational approximation to describe the @xmath93-dependence of @xmath100 found from local ideal mhd calculations : @xmath206 , shown as the dotted curve in figure [ fig6 ] . the dotted line in figure
[ growthrate]a represents the growth rate as estimated by the above expressions for @xmath207 and @xmath119 .
approximations in the model break down close to the ideal mhd instability limit , @xmath118 , where @xmath207 and @xmath208 diverge .
typical radial mode structures are shown in figure [ growthrate]b ; the @xmath93 values shown here are @xmath209 and @xmath210 . since gs2 solves the problem in ballooning angle @xmath9 and not in @xmath211 , the plotted functions are obtained from the appropriate fourier transform of @xmath212 .
the eigenfunctions @xmath213 are normalized so that their value is @xmath214 at the maximum of @xmath215 appearing close to @xmath216 ; then the blue ( red ) curves represent the real ( imaginary ) part of @xmath213 .
the eigenfunctions do not change appreciably as @xmath93 is varied .
in fact they very much resemble the well known marginally stable ideal mhd result which is of the form @xmath217 for @xmath218 and @xmath219 for @xmath220 ( indicated with black dotted curve in the figure ) .
taking these @xmath221 eigenfunctions , we can estimate @xmath100 from the gs2 simulations .
as also seen in figure [ growthrate]b , the eigenfunctions are affected by numerical error . in the calculation of @xmath28 a division by @xmath222
needs to be made , which amplifies small errors as @xmath222 approaches @xmath219 , which happens close to the spurious ideal stability limit @xmath118 .
therefore , we are unable to determine @xmath100 quantitatively .
figure [ deltaprime ] shows the estimated values of @xmath100 from gs2 simulations for a range of @xmath93 values ( circle symbols and solid curve ) .
the confidence intervals of the results are indicated with the shaded area , obtained by perturbing the eigenfunction within numerical uncertainties . as @xmath93 approaches @xmath223 ,
the uncertainties diverge ; accordingly , we do not show values of @xmath100 above @xmath224 . as a reference , we show the rational approximation of the ideal mhd result by the dotted line ( this is the same as the dotted curve of figure [ fig6 ] ) . as a function of @xmath93 .
solid line with circle markers : estimated value from gs2 simulations .
light blue shaded area : uncertainty in gs2 results . dashed line : rational approximation of the local ideal mhd result .
, width=302 ]
this investigation was stimulated , in part , by simulation results from gs2 @xcite , with a possible interpretation of an observed instability as an ideal current driven kink .
such a `` ballooning space '' calculation assumes the neglect of @xmath44 and all higher derivatives : i.e. it is equivalent to the approximations which lead to ( [ eq5 ] ) in configuration space , an equation which , incorrectly , predicts ideal instability for values of @xmath225 , where @xmath1 is related to the ratio of current gradient to magnetic shear .
it therefore appears that the `` ideal '' instability seen in gs2 is spurious .
however , the calculations presented here show that the consequence of correctly retaining the full functional dependence of @xmath80 in , for example , ( [ eq21 ] ) , is to exclude the possibility of ideal kink instability , while still permitting unstable values of the tearing index , @xmath226 , when @xmath77 exceeds a value around unity .
hence the mode identification in terms of a `` tearing / kink '' drive looks entirely justified , but its identification as an `` ideal kink '' as proposed by kadomtsev and pogutse @xcite , should be modified , regarding it rather as a high-@xmath12 tearing mode , driven unstable by a large value of @xmath226 , with collisionless reconnection provided by electron physics in the resonant layer around @xmath19 . as a consequence , it will be inappropriate to run gs2 with values of @xmath225 .
it is also of interest to note that hegna and callen @xcite used a modification of ( [ eq5 ] ) to describe general geometry , and as a simple way to derive a convenient formula for @xmath0 in toroidal systems .
although this may give a good approximation for values of @xmath227 , where @xmath0 is negative , it overestimates the instability drive for @xmath228 and predicts ideal instabilities for @xmath225 , where none exist .
we have also investigated a situation with a non - monotonic profile of the current density , @xmath32 , as might result from bootstrap currents near a tokamak pedestal .
stability of high-@xmath12 modes is again governed by a local equation , ( [ eq25 ] ) , depending on two parameters , @xmath3 and @xmath2 , where @xmath3 is a measure of the ratio @xmath229 and @xmath2 is the location of the resonance relative to the point of maximum @xmath230 . as for the previous case
, ideal instability does not occur for any values of the @xmath3 and @xmath2 parameters , but positive values of the tearing index , @xmath0 can be found for sufficiently large values of @xmath3 .
such values of @xmath3 can arise from the low shear resulting from the near cancellation of contributions from the background current and the current spike .
these situations can , typically , lead to @xmath231 , implying @xmath232 .
since such a current spike can result from the bootstrap current occurring naturally in the pedestal region of a tokamak h - mode plasma , these observations have possible relevance for the interpretation of elms in terms of surface `` peeling '' modes associated with tearing modes resonant within the pedestal .
they also suggest the possibility of influencing elm behaviour by driving reverse currents within the pedestal region .
we note that the potentially large values of @xmath0 could overcome stabilising effects , such as the glasser effect in a torus @xcite arising from a pressure gradient at the resonant surface in resistive mhd , or from diamagnetic effects associated with the steep gradients in the pedestal in hotter plasmas @xcite . to ameliorate the deleterious effects of large elms on divertor target plates , resonant magnetic perturbations ( rmps )
have been applied to produce magnetic islands with the intention of driving pedestal gradients below the mhd stability limit .
the tearing stability of the resulting non - symmetric equilibria is beyond the scope of this work but it is worth noting that the amplitude of such a rmp driven island depends on the value of @xmath0 calculated in this work @xcite . at first sight , it is perhaps surprising that such high-@xmath12 calculations can not always be reduced to a purely local calculation involving only the current gradient and magnetic shear at the rational surface , but instead requires that the complete structure of the @xmath29 profile be taken into account , albeit in a narrow region for a sharply localised gradient in the current profile .
consideration of this problem is beyond the scope of local gyro - kinetic codes .
consequently the extension to a toroidal calculation must inevitably become two - dimensional , unlike problems amenable to the ballooning transformation as in , e.g. the local gyrokinetic code gs2 .
this work was funded by the rcuk energy programme [ grant number ep / i501045 ] , and us department of energy grant at de - fg02 - 91er-54109 at mit .
ip is supported by the international postdoc grant of vetenskapsrdet . to obtain further information on the data and models underlying this paper
please contact [email protected] .
in this appendix we demonstrate that , using the current distribution of ( [ eq23 ] ) with the parameter @xmath233 , the global kink / tearing equation , ( [ eq1 ] ) , can be reduced , in the limit of high-@xmath12 , to the form of ( [ eq25 ] ) , and that , for the case of the @xmath134 mode investigated in figure [ fig8 ] , the equivalent @xmath3 value is @xmath234 .
we first introduce the notation , @xmath235 and treat @xmath236 as @xmath237 parameter .
then , in leading order of an expansion in @xmath112 , @xmath238 now with @xmath239 .
we next construct an expression for @xmath240 appearing in the denominator of the current drive term of ( [ eq1 ] ) , noting that the contributions of the inductive current , @xmath141 , and the localised current spike , @xmath138 , are simply additive , so that : @xmath241 where @xmath242 . in equation ( [ a3 ] ) we retained an @xmath243 correction because the leading order term is small at low shear due to near cancellation of the contributions from @xmath244 and @xmath245 to @xmath243 .
finally , transforming the radial variable in ( [ eq1 ] ) to @xmath64 , the kink / tearing equation takes the form , @xmath246 } \right\}=0 , \label{a4}\ ] ] which is precisely of the same form as ( [ eq25 ] ) when @xmath74 , with the parameter , @xmath3 given by : @xmath247 the above is easily generalised for non zero values of @xmath2 . note that the above mentioned cancellation of terms in @xmath244 and @xmath245 means that @xmath3 is formally @xmath237 but can become very large and even change sign if the shear at @xmath137 reverses .
for the parameters of figure [ fig7 ] , @xmath124 , @xmath125 , @xmath126 , @xmath248 , and for @xmath134 , the equivalent value of @xmath3 is @xmath234 .
10 kadomtsev b b and pogutse o p 1970 _ reviews of plasma physics _
vol.5 ( consultants bureau , new york , ed . by leontovich
m a ) p 249 wesson j a 1978 _ nucl .
fusion _ * 18 * 87 strauss h r 1981 _ phys . fluids _ * 24 * 2004 furth h p , killeen j and rosenbluth m n 1963 _ phys . fluids _ * 6 * 459 hegna c c and callen j d 1994 _ phys
. plasmas _ * 1 * 2308 pusztai i , catto p j , parra f i and barnes m 2014 _ plasma phys .
control .
fusion _ * 56 * 035011 militello f , huysmans g , ottaviani m and porcelli f 2004 _ phys .
plasmas _ * 11 * 125 poy a , agullo o , smolyakov a , benkadda s and garbet x 2013 _ phys .
plasmas _ * 20 * 020702 newcomb w a 1960 _ ann .
* 10 * 232 mccarthy p j and asdex upgrade team 2012 _ plasma phys .
control .
fusion _ * 54 * 015010 citrin j , saarelma s , temple d , conway n j , kirk a , meyer h , michael c a and the mast team 2012 _ plasma phys .
control .
fusion _ * 54 * 025001 eriksson h
g and wahlberg c 1997 _ plasma phys .
control .
fusion _ * 39 * 943 barnes m , parra f i , lee j p , belli e a , nave m f f and white a e 2013 _ phys .
lett . _ * 111 * 055005 drake j f and lee y c 1977 _ phys .
fluids _ * 20 * 1341 glasser a h , greene j m and johnson j l 1975 _ phys . fluids _ * 18 * 875 drake j f , antonsen t m , hassam a b and gladd n t 1983 _ phys . fluids _ * 26 * 2509 cowley s c , kulsrud r m and hahm t s 1986 _ phys . fluids _ * 29 * 3230 fitzpatrick r and hender t c 1991 _ phys .
fluids b _ * 3 * 644 | the global ideal kink equation , for cylindrical geometry and zero beta , is simplified in the high poloidal mode number limit and used to determine the tearing stability parameter , @xmath0 . in the presence of a steep monotonic current gradient ,
@xmath0 becomes a function of a parameter , @xmath1 , characterising the ratio of the maximum current gradient to magnetic shear , and @xmath2 , characterising the separation of the resonant surface from the maximum of the current gradient . in equilibria containing a current `` spike '' , so that there is a non - monotonic current profile , @xmath0 also depends on two parameters : @xmath3 , related to the ratio of the curvature of the current density at its maximum to the magnetic shear , and @xmath2 , which now represents the separation of the resonance from the point of maximum current density .
the relation of our results to earlier studies of tearing modes and to recent gyro - kinetic calculations of current driven instabilities , is discussed , together with potential implications for the stability of the tokamak pedestal . |
in our quantum protocol , the average @xmath13 of the meter observable for a given pre- and post - selected sub - ensemble of the probe system is : @xmath14 \:\operatorname{\im \mathfrak{m } } a_{m } } { \left(1+p_{m}\,\overrightarrow{r}.\overrightarrow{m}\right)\:+\left(1-p_{m}\ , \overrightarrow{r } .
\overrightarrow{m}\right)\ : \left|a_{m}\right|^{2}}\:.\label{eq : qsmeanvalue}\ ] ] for a given vector @xmath22 controlling the quantum gate action , the quantum eraser condition @xmath129 constrains @xmath21 to the red plane in figure [ fig : bloch representation].a .
we choose now particular final vectors @xmath21 of the meter system in relationship to the initial vector @xmath92 ( characterizing the meter initial state ) , in order to determine the real and imaginary parts of the modular value from the average meter observable @xmath13 .
we pick the real part of @xmath27 when the three vectors @xmath130 are coplanar ( @xmath32 in blue plane in figure [ fig : bloch representation].b ) , so that @xmath131 in equation ( [ eq : qsmeanvalue ] ) .
we isolate the imaginary part with orthogonal initial and final states of the meter ( @xmath33 orthogonal to blue plane in figure [ fig : bloch representation].c ) , so that @xmath132 in equation ( [ eq : qsmeanvalue ] ) . .
it contains all final meter states @xmath133 and @xmath134 implementing the quantum eraser condition .
( b - c ) the blue plane contains the initial meter state @xmath135 and the control state @xmath136 .
the final meter states ( b ) @xmath137 in the blue plane and ( c ) @xmath138 perpendicular to it measure the real and imaginary parts of the modular value , respectively.,scaledwidth=48.0% ] the first equality results immediately from the definition of the argument of the weak value , considering the following expression : @xmath140 in which @xmath141 can be expressed as a function of directions on the bloch sphere : @xmath142 where @xmath143 is the imaginary unit , @xmath64 and @xmath60 describe the pure initial and final qubit states , respectively , and @xmath62 gives the rotation axis associated to the observable @xmath144 .
for the second equality in ( [ eq : gmphase ] ) , our approach is inspired by the work of martinez et al .
@xcite , where they studied the geometrical characteristics ( amplitude and phase ) of polarization modulation optical devices on the poincar sphere .
the trajectory of a pure qubit state on the bloch sphere corresponding to the unitary transformation @xmath80 is a non - geodesic opened arc .
consequently , the resulting state @xmath145 is no longer in phase with the initial state @xmath146 . to yield nonetheless an expression in terms of a solid angle for the accumulated phase ,
let us express the resulting state as @xmath147 , where the phase @xmath148 is due to the non - geodesic movement of @xmath146 to the output state @xmath149 .
the vector @xmath150 is entirely defined by @xmath151 ( essentially , @xmath150 is the mirror image of @xmath152 with respect to the @xmath62 axis ) .
the additional phase @xmath148 is determined by projecting @xmath153 onto the orthogonal eigenvectors @xmath154 and @xmath155 of the operator @xmath80 . by considering : @xmath156
we conclude that the projection of @xmath153 onto the eigenvector @xmath154 yields the following two relations : @xmath157 the moduli @xmath158 are equal since @xmath64 and @xmath159 are mirror images with respect to @xmath62 .
consequently , only the accumulated total phase of the open loops , known as pancharatnam connection , remain in ( [ connectionrel ] ) : @xmath160 of the four vertices @xmath161 , @xmath162 , @xmath163 and @xmath164 represented in the bloch sphere ( the red surface ) .
the states @xmath165 and @xmath166 lying on the equator are horizontal lifts of the associated states @xmath167 and @xmath168 , respectively .
the solid angle of the counterclockwise sequence of states @xmath169 is in direct relationship to the pancharatnam connection @xmath170.[fig : sphericalquadrangles ] ] in practice , the pancharatnam connection @xmath171 relating arbitrary states @xmath167 and @xmath168 is determined by calculating the spherical quadrangle @xmath172 in the bloch sphere ( figure [ fig : sphericalquadrangles ] ) , where the supplemental vertices @xmath165 and @xmath166 are well - defined vectors . to understand , how they are determined , we must express their position in the spherical coordinate system . by convention
, @xmath173 corresponds to the azimuth angle and @xmath174 to the polar angle . in this representation , a pure state on the bloch sphere
is defined by : @xmath175 and the pancharatnam connection is given by : @xmath176 the connection is in phase , i.e. @xmath177 , for transports with the same azimuth angles @xmath178 and for transformations happening around the equator of the bloch sphere , i.e. for the polar angle @xmath179 .
these two kinds of transports are known as horizontal lifts along the geodesic connecting the states @xmath167 and @xmath168 on the bloch sphere .
the states @xmath180 and @xmath181 are fixed with the same azimuth angle as @xmath167 and @xmath168 , respectively , and with a polar angle @xmath179 .
the closed loop @xmath169 along the geodesic arcs determines the spherical quadrangle @xmath172 , which is equivalent to @xmath171 : @xmath182 note that the sign present in front of the solid angle for a given sequence of states is positive when the sequence is followed counterclockwise and is negative when the sequence is followed clockwise .
the sign of the solid angle changes when the sequence of projections is inversed , @xmath183 .
it is possible to express a solid angle linking three vertices as a sum of three spherical quadrangles @xcite : @xmath184 where each spherical quadrangle contains two vertices of the initial solid angle .
we use the decomposition property of eq .
( [ decomp ] ) to rewrite the expression giving @xmath185 ( [ phiexpr ] ) according to : @xmath186 where we made use of eq .
( [ eq : argsolidabba ] ) to express the connexions appearing in ( [ phiexpr ] ) . following
the indices may be tedious but , essentially , eq . ( [ phiexpr ] ) and ( [ eq : argsolidabba ] )
show together that the expression of the phase @xmath185 includes a sum of two spherical quadrangles ; then we used eq .
( [ decomp ] ) to express the sum of these two spherical quadrangles as a function of the third spherical quadrangle and of the spherical triangle appearing in eq .
( [ decomp ] ) . furthermore , expression ( [ triquadforphi ] )
points out that the non - geodesic phase @xmath185 is the sum of the geometric phase of the closed loop @xmath187 ( first term ) and the phase of the pancharatnam connection @xmath188 ( second term ) . using the last results , the argument of the weak value of @xmath80
is : @xmath189 equality ( a ) results from the definition of the states @xmath190 and @xmath191 .
( b ) expresses the pancharatnam connexions in terms of solid angles using eq .
( [ eq : argsolidabba ] ) .
( c ) takes the argument of the previous expression .
( d ) exploits the decomposition property of eq .
( [ decomp ] ) .
( e ) follows from the expression of @xmath185 in ( [ triquadforphi ] ) .
( f ) is due to canceling terms .
( g ) combines the two spherical triangles in one spherical quadrangle ( as the paths @xmath192 and @xmath193 present in the triangles cancel each other ) .
experimentally , we implement the polarization - entangled biphoton state after the conceptual cnot gate using two orthogonal nonlinear bbo crystals in the sandwich configuration " @xcite .
each crystal is @xmath25 mm thick and cut for type - i phase matching with @xmath194 and @xmath195 . via the nonlinear interaction of spontaneous parametric down - conversion ( spdc ) , the pump laser ( blue diode dl-7146 - 101s from sanyo electric co. ) centered at @xmath196 nm generates two polarization - entangled photons at @xmath197 nm . the laser diode is controlled by temperature ( thorlabs ted 200c ) and current ( thorlabs ldc202c ) controllers .
it produces a continuous laser output power of @xmath198 mw .
we use four single photon counting modules ( spcm - aq4c from perkin - elmerfor ) for the joint polarization measurement of the meter or probe photons . the polarization basis
are selected by half- and quarter - wave plates followed by a polarizing beam - splitter ( rchp-15.0-ca-670 - 1064 from cvi melles griot ) . before detection , the photons
are coupled into multimode fibers and filtered using low - pass filters ( fgl780 from thorlabs ) .
about @xmath199 total coincidence counts per second are acquired by using a homemade fpga ( se3board from xilinx ) coincidence counter . for given pre- and post - selected probe states
, we consider a final meter outcome which follows a binomial distribution : the meter qubit is measured either by detector @xmath200 ( and contributes to @xmath107 ) or by detector @xmath201 ( and contributes to @xmath108 ) .
the expectation value of @xmath108 is therefore @xmath202=p\left(2|3\right)n$ ] and its variance @xmath203 .
@xmath204 is the total number of pre- and post - selected detector events and @xmath205 is the conditional probability to trigger the meter detector @xmath201 for a given probe detection by @xmath206 .
consequently , the expectation value of the measured visibility is : @xmath207\nonumber \\ & { = } & 1 - \frac{2}{n } e\left[n^{min}_{23}\right]\nonumber\\ & { = } & 1 - 2 p\left(2|3\right)\nonumber \\ & { = } & v\:,\end{aligned}\ ] ] where the last equality follows directly from the definition of the conditional probability : @xmath208 , with @xmath48 and @xmath49 the maximum and the minimum of the joint probability of the measurement protocol . the corresponding variance is : @xmath209 where we used the relationship @xmath210 .
this leads to the standard deviation : @xmath211 finally , the signal - to - noise ratio of the presented measurement scheme is : @xmath212 the standard protocol determines the real ( or the imaginary ) part of the modular value by measuring the meter observable @xmath213 ( or @xmath214 ) . in this case , the visibility @xmath46 in the signal - to - noise relation ( [ snr ] ) is replaced by the absolute value of the meter average @xmath215 ( or @xmath216 ) . in the weak measurement limit , the latter is related to the modular value by the approximation @xmath217 ( or @xmath218 ) .
this expression is similar to the one obtained for our scheme in the weak measurement limit , relating the visibility to the modulus of the modular value : @xmath219 .
because the modulus of a complex number is always larger or equal than its real and imaginary parts ( @xmath220 and @xmath221 ) , our scheme improves the signal - to - noise ratio of the weak measurement compared to the standard protocol .
( and , additionaly , it works when the weak measurement approximation fails . ) as described in the main letter , the proposed measurement scheme involves the determinations of two parameters : the measurement strength @xmath93 and the purity @xmath222 .
these requirements are not specific to our scheme . in all weak value measurement protocols ,
the determination of the measurement strength @xmath93 is a necessary and inevitable process which requires a separate acquisition .
our scheme does not perform better or worse than other schemes in the literature in that respect .
the additional determination of the purity @xmath222 is only required because we considered the most general case of an initial incoherent state of the qubit meter system .
most of the literature assumes the meter to be in a known pure state to avoid this supplementary step . to determine the initial meter state in our protocol
, it is only necessary to perform the quantum tomography of a single qubit . in pratice , in our experimental implementation , we simulated the cnot gate by using spontaneous parametric down - conversion ( instead of using a true cnot gate with two separate entries that could be characterized independently ) .
for this reason , in our experiment , we determined the purity by performing quantum tomography of the two - qubit state since the input meter state could not be measured directly .
this two - qubit tomography is not at all required by the proposed protocol but appears only as a side - effect of the practical implementation of our demonstrative experiment .
y. aharonov , and l. vaidman , _ the two - state vector formalism of quantum mechanics : an updated review _ , edited by j. g. muga , r. sala mayato , and .
l. egusquiza , time in quantum mechanics , ( springer , 2008 ) , isbn 978 - 3 - 540 - 73472 - 7 . | in this letter , we present a new procedure to determine completely the complex modular values of arbitrary observables of pre- and post - selected ensembles , which works experimentally for all measurement strengths and all post - selected states .
this procedure allows us to discuss the physics of modular and weak values in interferometric experiments involving a qubit meter .
we determine both the modulus and the argument of the modular value for any measurement strength in a single step , by controlling simultaneously the visibility and the phase in a quantum eraser interference experiment .
modular and weak values are closely related . using entangled qubits for the probed and meter systems ,
we show that the phase of the modular and weak values has a topological origin .
this phase is completely defined by the intrinsic physical properties of the probed system and its time evolution .
the physical significance of this phase can thus be used to evaluate the quantumness of weak values . in 1988 ,
aharonov , albert , and vaidman ( aav ) introduced the weak value of a quantum observable @xmath0 from an extension of the von neumann measurement scheme @xcite .
they pointed out that the result of a measurement involving a weak coupling between a meter and the observable @xmath0 of a system with a pre - selected initial state @xmath1 , and a post - selected final state @xmath2 depends directly on the weak value : @xmath3 an unbounded complex number .
in particular , they showed that the shift of the average detected position due to post - selection is proportional to the real part of the weak value .
since for weak measurements in the absence of post - selection , this shift is proportional to the average of the observable @xmath4 , a direct but bold physical interpretation of the weak value assumes it represents somehow the average of @xmath0 in the pre- and post - selected ensemble .
they also related the imaginary part of the weak value to the shift of the average impulsion . beside the aav approach
, weak values may also appear using a meter strongly coupled to the observable @xmath0 @xcite .
in these instances , the effective weak interaction is achieved by selecting particular initial states of the meter system , so that the probability of actually measuring @xmath0 is low and the probed system is left unperturbed most of the time .
therefore , both methods transform the standard von neumann procedure to a weak measurement with a high incertitude . weak values and weak measurements proved useful in many fields of physics and chemistry @xcite . nevertheless , the proper physical interpretation of weak values remains highly debated .
for example , weak values were used to develop a time - symmetrized approach to standard quantum theory , the two - state vector formalism @xcite , where they appear as purely quantum objects .
oppositely , a purely classical view of the occurrence of unbounded , real weak values and possibly of complex ones was proposed recently @xcite ( which is criticizable though @xcite ) . in this letter , we uncover a physical interpretation of complex weak values in terms of their polar representation ( modulus and argument ) , which provides evidence for their quantumness .
we devised an interferometric procedure to measure and discuss complex weak values in their polar representation instead of the usually determined real or imaginary part .
our procedure relies essentially on a joint phase and visibility measurement in a quantum interferometer where the meter system acts as a quantum eraser . using simple cases exploiting entangled qubits ,
we relate the argument of the weak value to topological phases defined completely by the probed system states involved in the weak measurement .
additionally , our procedure works in conditions where the usual weak measurement procedure fails completely : ( ) for arbitrary measurement strengths ( i. e. including strong measurements ) and ( ) for orthogonal and nearly orthogonal initial and final probe states .
it proceeds by optimizing the interference phase to measure simultaneously the modulus and the argument of the weak value in a single step . formally , our procedure implements a quantum controlled evolution , in which an arbitrary quantum system @xmath1 , the probe , interacts with a qubit meter via the quantum gate ( fig .
[ fig : bloch representation].a ) : @xmath5where @xmath6 are projectors acting on the meter and @xmath7 is a phase factor first supposed to be null . the transformation @xmath8 is expressed in terms of a time independent hermitian operator @xmath0 and an arbitrary coupling strength @xmath9 , defined by the integral @xmath10 @xcite . after the gate interaction , the spin observable @xmath11 of the meter is measured .
according to the final meter state , the information about whether the transformation @xmath12 was applied on the probe can be preserved or erased , completely or partially .
finally , a projective measurement of the probe system post - selects the vector state @xmath2 .
the average @xmath13 of the meter observable for a given pre- and post - selected sub - ensemble of the probe system is then : @xmath14 \:\operatorname{\im \mathfrak{m } } a_{m } } { \left(1+p_{m}\,\overrightarrow{r}.\overrightarrow{m}\right)\:+\left(1-p_{m}\ , \overrightarrow{r } .
\overrightarrow{m}\right)\ : \left|a_{m}\right|^{2}}\:.\label{eq : qsmeanvalue}\ ] ] in this expression , the normalized vectors @xmath15 , @xmath16 and @xmath17 are the directions on the meter bloch sphere determining the initial @xmath18 and final @xmath19 meter states as well as the projector state @xmath20 controlling the interaction , respectively .
the direction of @xmath21 was chosen orthogonal to @xmath22 to select maximally interfering pathways through the meter measurement : then , the gate action appears as a superposition of having applied both @xmath12 and @xmath23 , and all information about the gate action is lost ( quantum eraser condition ) .
the parameter @xmath24 characterizes the purity of the initial meter state , ranging from @xmath25 for pure states to @xmath26 for a maximally mixed state .
@xmath27 is defined as the modular value of the probe observable @xmath0 for the pre- and post - selected sub - ensemble : @xmath28 it appears from the action of projectors in ( [ eq : controlled measurement interaction ] ) .
modular values were not often reported as such in the literature because they are directly related to weak values in the usual weak approximation limit for small coupling strengths , through a first order polynomial development in @xmath9 : @xmath29 .
nevertheless , they characterize all projective couplings between the probe and meter systems , where they generalize weak values in a non - perturbative way .
they typically describe quantum - gate type interactions @xcite and quantum interference experiments @xcite , but appear also in photon trajectory measurements @xcite for example . in the following ,
we relate the physical interpretation of modular values to the visibility and the phase of interferometric experiments . in our procedure ,
the interaction strength is not determined by @xmath9 . instead , it reflects the probability of the application of @xmath12 by the quantum gate , which is controlled by the measurement strength @xmath30 , with @xmath31 $ ] . by choosing particular final meter states
@xmath32 and @xmath33 ( constrained by @xmath34 ) , the real and imaginary parts of the modular value are determined from the average meter observable @xmath13 , respectively @xcite . for small measurement strengths @xmath35 when the purity @xmath24 is close to one , we obtain the modular value according to the standard approximations of weak measurements : @xmath36 where the weak measurement approximation effectively removes the nonlinear dependence of equation ( [ eq : qsmeanvalue ] ) on the modular value modulus ( see denominator ) . for an arbitrary measurement strength ,
we seek instead to measure the modular value in its polar form to assess directly its modulus @xmath37 and argument @xmath38 .
we introduce an additional unitary transformation @xmath39 in the meter path .
it creates a relative phase shift @xmath40 between the orthogonal states @xmath41 and @xmath42 that is effectively equivalent to a rotation of the modular value in the complex plane .
when the phase shift compensates precisely the argument of the modular value ( i. e. when @xmath43 ) , this rotation aligns the modular value with the real axis . choosing the meter configuration @xmath32 that picks
the real part of the modular value provides now its full modulus , while its argument is equal to the introduced phase shift . in practice , our procedure implements a quantum interferometer exploiting entanglement to measure the two quantities concurrently .
indeed , the expression for the joint probability outcome @xmath44 of the meter and the probe measurements is proportional to : @xmath45 typical of an interference phenomenon , where @xmath46 represents the visibility and @xmath47 the phase .
experimentally , the visibility is determined by measuring the maximum and the minimum of the joint probability , denoted by @xmath48 and @xmath49 , respectively : @xmath50 when the phase introduced by @xmath39 equals the argument of the modular value , the maximum of the joint probability is obtained for the meter vector @xmath32 , while its minimum is obtained for the orthogonal state @xmath51 .
the two situations correspond to constructive and destructive interferences in the joint measurement , respectively .
the visibility depends on the coupling strength and the modular value modulus : @xmath52 with coefficients @xmath53 defined by : @xmath54 this quadratic equation provides two solutions for the modular value modulus : @xmath55 the solution @xmath56 corresponds to @xmath37 , if the condition : @xmath57 is verified , and @xmath58 otherwise .
together , they provide the characterization of the modular value modulus for an arbitrary coupling strength .
it is directly related to the visibility . in particular , the weak measurement approximation gives @xmath59 , similarly to equation ( [ eq : realpartmodular variable ] ) .
in this expression of the modular value , the visibility plays the same role than the pointer shift in weak values .
this shows a strong connection between modular values and weak interferometric experiments . and final @xmath60 states , @xmath61 observable rotation axis @xmath62 , and the mirror image @xmath63 of @xmath64 with respect to the @xmath62 axis .
the solid angle @xmath65 associated to the geometric phase in eq .
( [ eq : gmphase ] ) is obtained by following consecutively the three great circle arcs @xmath66 ( blue ) , @xmath67 ( red ) , and @xmath68 ( red ) .
( c ) the set - up comprises three areas : the state preparation with the two qubit generation ( ) , the meter measurement by detectors @xmath69 and @xmath70 ( ) and the final probe post - selection by @xmath71 ( ) .
the coincidence counts @xmath72 and @xmath73 are acquired by four single photon counting modules ( spcm ) placed in the meter and probe paths.,scaledwidth=48.0% ] now we consider the connection between modular and weak values to gain insight into the physics of weak values .
the previously arbitrary probe system becomes a qubit and the probe transformation @xmath74 is a rotation operator involving the pauli observable @xmath75 ( @xmath62 a unit vector ) .
we set a strong aav coupling strength @xmath76 .
then , @xmath77 and the quantum gate acting on the two qubits becomes : @xmath78 where the phase factor @xmath7 in ( [ eq : controlled measurement interaction ] ) was set to @xmath79 .
this shows the equivalence of modular and weak values of @xmath80 ( see also @xcite ) .
we can thus apply our scheme to determine an arbitrary weak value of the pauli operator in its polar representation .
interestingly , the argument of the weak value depends only on the probe evolution from its initial to final state as defined by the operator @xmath81 ( a related result is mentioned in @xcite ) .
this phase has a topological component , similar to the pancharatnam geometric phase .
its value is proportional to the solid angle @xmath65 delimited by four vectors on the bloch sphere ( see fig . [
fig : bloch representation].b and derivation @xcite for details ) : @xmath82 this geometric phase is completely defined by intrinsic properties of the probe system .
it is observed directly in the interferometric experiment but does not depend on the meter properties , contrary to the pointer shift in usual weak measurement ( which depends on @xmath9 ) .
it emphasizes the quantum origin of the argument of the complex weak value and its relationship to physical properties of the probe system .
experimentally , we implement a conceptual cnot gate @xmath83 .
the initial meter state @xmath84 , with @xmath85 , controls the application of the unitary observable @xmath86 on the target probe state pre - selected in the @xmath87 state .
the meter projective measurement is then performed in the @xmath86 basis ( @xmath21 ) .
it erases the information about the application of @xmath86 on the target since it was controlled by the meter basis vectors @xmath88 and @xmath89 of @xmath90 ( @xmath22 ) .
it is followed by the probe measurement of the final post - selected state @xmath2 .
finally , we obtain the weak value @xmath91 as a function of our chosen initial meter state ( @xmath92 ) , which defines the measurement strength ( @xmath93 ) , and post - selected probe state @xmath2 . in practice
, the two - qubit state after the cnot gate is simulated by entangled photon pairs produced by type i spontaneous parametric down - conversion in two orthogonal nonlinear bbo - crystals @xcite ( see fig . [
fig : bloch representation].c and @xcite for set - up details ) .
one photon is assimilated to the meter and the other to the probe . a half - wave plate and a third bbo placed before the source control the photon pair state
conceptually , they select the initial meter polarization of the cnot gate while the initial probed state always has horizontal polarization . with pre - selected @xmath94 polarization for three initial meter states : @xmath95 and @xmath96 ( red squares ) , @xmath97 and @xmath98 ( blue circles ) , @xmath99 and @xmath100 ( black triangles ) .
( c ) positive / negative solution criterion .
values larger / smaller than unity ( violet , solid horizontal line ) admit the positive / negative solution @xmath101 , respectively .
final meter states are @xmath102 and @xmath103 for measurements ( a - b ) and @xmath94 and @xmath104 for ( c ) .
grey , solid lines represent theoretical curves .
[ fig : visibilityd ] ] we post - select the probe polarization @xmath105 at detector @xmath71 .
detectors @xmath69 and @xmath70 measure the meter polarization ( diagonal @xmath102 and anti - diagonal @xmath103 states , respectively ) .
we adjust the phase @xmath40 by tilting a birefringent @xmath106-cut quartz plate in the meter path to obtain the interference visibility @xmath46 from the coincidence counts .
when coincidence counts @xmath107 are maximal for detectors @xmath69 and @xmath71 ( constructive interference ) , coincidence counts @xmath108 for detectors @xmath70 and @xmath71 are minimal ( destructive interference ) .
then , the phase @xmath40 equals the argument @xmath109 of the weak value @xmath110 , while the visibility is given by equation ( [ eq : visibility ] ) : @xmath111 .
this scheme improves the signal - to - noise ratio of a weak measurement @xcite .
fig .
[ fig : visibilityd ] presents the visibility and the phase for three different initial meter states inconciliable with weak measurement approximations .
the corresponding strengths @xmath93 were determined from the density operator of the biphoton using quantum state tomography @xcite .
the purities were rather estimated by fitting the theoretical visibility to the data ( @xmath112 minimization method ) because @xmath46 is highly sensitive to @xmath24 .
the latter step may be skipped if the meter is supposed in a pure initial state @xcite , as usually done in the literature .
we chose pre- and post - selected states for which the argument of @xmath113 ( fig .
[ fig : visibilityd].b ) takes only the two values @xmath26 or @xmath114 for simplicity .
it determines the sign of the weak value .
the solid angle @xmath65 related to the geometric phase ( [ eq : gmphase ] ) is defined in the @xmath115 plane of the bloch sphere ( see fig .
[ fig : visibilityd].b ) : the great circle arcs @xmath116 make a full circle or compensate each other depending on the post - selected state @xmath117 . the visibility ( fig .
[ fig : visibilityd].a ) provides the modulus @xmath118 , using the two solutions obtained in relation ( [ eq : modular value relation ] ) .
the switch between them occurs at the maximum of the visibility .
the criterion ( [ eq : weakmeasurcond ] ) determining this switch is measured from the coincidence count ratio @xmath119 with horizontal @xmath94 ( detector @xmath69 ) and vertical @xmath104 ( detector @xmath70 ) meter polarizations . in this case @xmath120 , and the meter measurement reveals completely the probe state after the quantum gate interaction ( no information erasure ) .
the theoretical switch angle and the measured criterion agree strongly except for strengths approaching the range of weak measurements ( @xmath121 ) , where a difference of @xmath122 is observed due to increasing experimental noise ( fig .
[ fig : visibilityd].c ) .
( red squares ) , @xmath123 ( blue circles ) , and @xmath121 ( black triangles ) , and from the standard weak measurement technique ( violet diamonds ) using relation ( [ eq : realpartmodular variable ] ) with the weak strength @xmath124 and purity @xmath125 . all set - ups use a @xmath126 basis for the final meter measurement ( the additional measurement in the @xmath127 meter basis required by standard weak measurements was not performed since @xmath128 here ) . [
fig : wv ] ] the full weak values determined using a strong ( @xmath123 ) or a weaker ( @xmath121 ) strength are compared in fig .
[ fig : wv].a .
both set - ups provide excellent agreement with the theoretical curve , except at the solution switch , where the accuracy of the set - up using weaker measurement strengths decreases ( see insets a.1 and a.2 ) . in fig .
[ fig : wv].b , we compare our method to the standard weak measurement technique . for a small modulus of the weak value @xmath113 ,
both techniques provide results close to theoretical predictions .
however , for large moduli , the weak measurement approximation breaks down completely ( zoom b.1 ) for a wide range of post - selected states approaching orthogonality to the pre - selected state .
weak measurement results are useless there and only our method works . in conclusion ,
the presented quantum eraser procedure exploits a qubit meter to measure directly the modulus and the argument of complex modular and weak values for arbitrary measurement strengths .
the connection between modular and weak values allowed us to investigate directly weak values of qubit systems in their polar representation by performing a one - step visibility and phase measurement . in this case , the argument of the weak value is associated to a quantum geometric phase , that has a non - controversial physical meaning .
this direct relevance of the polar form of the weak value to the intrinsic physical properties of the system evolution shows that the interpretation of past and present experiments involving weak values ought not be limited to the consideration of their real and imaginary parts .
our method to determine weak values requires fewer measurements and does not suffer the limitations of the standard weak measurement technique for large weak values , while it is applicable for both weak and strong measurement conditions . experimentally , this opens the way to exploiting with greater accuracy the measure of weak values , particularly for nearly orthogonal pre- and post - selected states .
y. c. is a research associate of the belgian fund for scientific research f.r.s .- fnrs .
b. k. acknowledges financial support from the action de recherche concerte ( biostruct project ) of the university of namur ( unamur ) and the support from the nanoscale quantum optics cost - mp1403 action . the authors would like to thank profs .
b. hespel and p. a. thiry for fruitful discussions and support , without which this work would not have been possible .
we also thank s. mouchet for the careful reading of the paper and the insightful remarks as well as j .-
p .
van roy for the upgrade and the automation of the whole photon detection system . |
[ sec : intro ] in this article , we extend the analysis initiated in @xcite concerning blow - up , and blow - up properties , in solutions to the initial boundary value problem for the generalized , inviscid proudman - johnson equation ( @xcite , @xcite , @xcite ) @xmath0 , \\ i(t)=-(\lambda+1)\int_0 ^ 1{u_x^2\,dx } , \end{cases}\ ] ] where @xmath1 and solutions are subject to periodic boundary conditions @xmath2 we note that the equation arises in several important applications , in the presence or absence of the nonlocal term @xmath3 . for @xmath4 ( [ eq : nonhomo])i ) , iii ) reduces to the inviscid burgers equation of gas dynamics differentiated once in space . if @xmath5 the hunter saxton equation ( hs ) describes the orientation of waves in a massive director field of a nematic liquid crystal ( @xcite , @xcite , @xcite , @xcite ) . for periodic functions , the hs equation also has a deep geometric meaning as it describes geodesics on a group of orientation preserving diffeomorphisms on the unit circle modulo rigid rotations with respect to a right - invariant metric ( @xcite , @xcite , @xcite , @xcite ) .
if @xmath6 ( [ eq : nonhomo ] ) i ) , iii ) can be obtained directly from the @xmath7dimensional incompressible euler equations @xmath8 using stagnation point - form velocities @xmath9 , @xmath10 or through the cylindrical coordinate representation @xmath11 @xmath12 and @xmath13 , where @xmath14 ( @xcite , @xcite , @xcite , @xcite , @xcite ) .
finally , in the local case @xmath15 , the equation appears as a special case of calogero s equation @xmath16 for arbitrary functions @xmath17 ( @xcite ) .
in @xcite we derived representation formulae for periodic solutions to ( [ eq : nonhomo])-([eq : pbc ] ) and , for several classes of mean - zero initial data , examined their @xmath18 regularity for @xmath19 $ ] .
for convenience of the reader , the main results established in @xcite are summarized in theorems [ thm : sarria1]-[thm : sarria2 ] below .
[ thm : sarria1 ] consider the initial boundary value problem ( [ eq : nonhomo])-([eq : pbc ] ) .
there exist smooth , mean zero initial data such that : 1 .
[ it:1 ] for @xmath20\cup(1,+\infty)$ ] , there is a finite @xmath21 such that @xmath22 for every @xmath23 $ ] .
additionally , the blow - up is two - sided ( two - sided , everywhere blow - up ) .
[ it:2 ] for @xmath24 there is a finite time @xmath21 and a finite number of @xmath25 $ ] , @xmath26 , such that @xmath27 ( one - sided , discrete blow - up ) .
[ it:3 ] for @xmath28,$ ] solutions persist globally in time .
more particularly , these vanish as @xmath29 for @xmath30 but converge to a nontrivial steady state for @xmath31 for @xmath21 as in theorem [ thm : sarria1 ] above , theorem [ thm : lpintro ] below examines @xmath32 regularity of @xmath33 for @xmath34 and @xmath35 .
[ thm : lpintro ] let @xmath36 in theorem [ thm : sarria1 ] be a solution to the initial boundary value problem ( [ eq : nonhomo])-([eq : pbc ] ) defined for @xmath34
. then 1 .
[ it : lp2 ] for @xmath37 and @xmath38 2 .
[ it : lp1 ] for @xmath39 and @xmath40\cup(1,+\infty)$ ] , @xmath41 .
[ it : ener ] the energy @xmath42 diverges if @xmath43 $ ] as @xmath44 but remains finite for @xmath45 $ ] otherwise .
moreover , @xmath46 blows up to @xmath47 as @xmath44 when @xmath48 $ ] and @xmath49 for @xmath50 ; whereas , @xmath51 if @xmath52 $ ] but remains bounded when @xmath53 $ ] for all @xmath45 $ ] . see [ sec : preliminaries ] for details on the class of initial data used to establish theorems [ thm : sarria1 ] and [ thm : lpintro ] .
lastly , let @xmath54 denote the family of piecewise constant functions with zero mean in @xmath55 $ ] .
then , in @xcite we proved the following : [ thm : sarria2 ] for the initial boundary value problem ( [ eq : nonhomo])-([eq : pbc ] ) , 1 . [ it:4 ]
suppose @xmath56 and @xmath57 . then , there exist solutions and a finite @xmath21 for which @xmath33 undergoes a two - sided , everywhere blow - up as @xmath44 . if @xmath58 a one - sided discrete blow - up may occur instead .
in contrast , for @xmath59 $ ] , solutions may persist globally in time .
more particularly , these either vanish as @xmath29 if @xmath60 , or converge to a nontrivial steady - state for @xmath61 .
[ it:5 ] suppose @xmath62 and assume solutions are defined for all @xmath63,\ , t>0.$ ] then no @xmath64 solution may exist for @xmath65 , where @xmath66 if @xmath67 , and @xmath68 for @xmath69 .
further , @xmath70 when @xmath71 , while @xmath72 where the constants @xmath73 depend on the choice of @xmath74 and @xmath75 the reader may refer to @xcite for details , and the works @xcite , @xcite , @xcite , @xcite , @xcite , @xcite for additional background .
the purpose of this work is to extend the above results to initial data which belongs to classes of functions with varying concavity profile near certain points in the domain .
more particularly , we suppose throughout that @xmath76 is bounded and , at least , @xmath77 @xmath78 then , for @xmath79 , we will assume there are constants @xmath80 and @xmath81 , and a finite number of points @xmath82 $ ] such that , near @xmath83 , @xmath84 analogously , for @xmath67 , we suppose there are constants @xmath85 , @xmath86 , and a finite number of locations @xmath87 in @xmath55 $ ] such that , in a neighbourhood of @xmath88 , @xmath89 we refer to [ subsec : dataclass ] for specifics of the above .
it is worth mentioning that , for @xmath90 , the above local estimates may lead to cusps in the graph of @xmath91 , therefore possible jump discontinuities in @xmath92 of _ infinite _ magnitude across @xmath83 and/or @xmath88 .
in contrast , a jump discontinuity of _ finite _ magnitude in @xmath92 may occur if @xmath93 . as we will see in the coming sections ,
the finite or infinite character in the size of this jump plays a decisive role , particularly in the formation of spontaneous singularities for the special case of stagnation point - form solutions to the three dimensional incompressible euler equations .
the remaining of the paper is organized as follows . in [ sec : preliminaries ] , we provide an outline for the derivation of the representation formulae established in @xcite and provide further details on the class of initial data to be considered in this article .
then , new blow - up results are stated and proved in [ sec : blowup ] , while specific examples are to be found in
[ sec : examples ] .
[ subsec : sol ] in @xcite , we used the method of characteristics to derive a representation formula for periodic solutions to ( [ eq : nonhomo ] ) .
for convenience of the reader , below we give a brief outline of the derivation .
define the characteristics , @xmath94 as the solution to the initial value problem @xmath95,\ ] ] so that @xmath96 then , using ( [ eq : nonhomo])i ) , iii ) and the above , we obtain @xmath97 which for @xmath98 , @xmath99 , and @xmath100 , can be written as @xmath101 assume we have two linearly independent solutions @xmath102 and @xmath103 to ( [ eq : nonhomo2 ] ) satisfying @xmath104 and @xmath105 . then , since @xmath106 and @xmath107 , we deduce that @xmath108 now , uniqueness of solution to ( [ eq : cha ] ) and periodicity implies that @xmath109 for as long as @xmath36 is defined .
consequently , simplifying and integrating ( [ eq : compat])i ) with respect to @xmath110 gives @xmath111 where we define @xmath112 for @xmath113 , and @xmath114 as a result , ( [ eq : jacid ] ) and ( [ eq : j])i ) yield , after further simplification , @xmath115 the strictly increasing function @xmath116 satisfies the initial value problem @xmath117 from which the existence of an eventual finite blow - up time @xmath21 for ( [ eq : mainsolu ] ) will depend , in turn , upon the existence of a finite , positive limit @xmath118 for @xmath119 to be defined .
moreover , assuming sufficient smoothness , ( [ eq : sum ] ) and ( [ eq : mainsolu ] ) imply that @xmath120 so that , for as long as it exists , @xmath36 maintains its initial concavity profile .
[ subsec : dataclass ] suppose solutions exist for @xmath34 , @xmath121 .
define @xmath122}\{u_x(\gamma(\alpha , t),t)\},\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,m(0)=m_0 \end{split}\ ] ] and @xmath123}\{u_x(\gamma(\alpha , t),t)\},\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,m(0)=m_0 , \end{split}\ ] ] where @xmath83 , @xmath124 , and @xmath88 , @xmath125 , denote the finite number of locations in @xmath55 $ ] where @xmath126 attains its greatest and least values @xmath127 , respectively .
then , it follows from ( [ eq : mainsolu ] ) ( @xcite ) that @xmath128 for @xmath129 .
now , the results of theorems [ thm : sarria1]-[thm : sarria2 ] suggest that the curvature of @xmath91 near @xmath83 and/or @xmath88 plays a decisive role in the regularity of solutions to ( [ eq : nonhomo ] ) . therefore ,
in the following sections , we further examine this interaction by considering a large class of functions in which @xmath76 is assumed to be bounded , at least @xmath130 , and has arbitrary curvature near the location(s ) in question .
more particularly , for @xmath79 , we will assume there are constants @xmath131 and @xmath81 such that @xmath132 for @xmath133 , and small enough @xmath134 , @xmath135 .
similarly , for @xmath67 , we suppose there is @xmath85 such that @xmath136 for @xmath137 and @xmath138 , @xmath139 .
see figure [ fig : data0 ] below .
now , for @xmath140 and @xmath141 as above , define @xmath142,\,\,\,\,\,\,\,\,\,\,\,\,\,\mathcal{d}_j\equiv[\underline\alpha_j - s,\underline\alpha_j+s].\ ] ] then , below we list some of the data classes that admit the asymptotic behaviour ( [ eq : expnew0 ] ) and/or ( [ eq : expnew00 ] ) for particular values of @xmath143 .
* @xmath144 for @xmath145 and @xmath146 ( see definition [ def : order ] ) . * if @xmath147 @xmath148 for @xmath79 , or @xmath149 if @xmath67 . * in the limit as @xmath150 , @xmath151 for @xmath79 , or @xmath152 if @xmath67 . * from ( [ eq : expnew0 ] ) , we see that the quantity + @xmath153_{_{q;\overline\alpha_i}}=\sup_{\alpha\in\mathcal{d}_i}\frac{\lvert u_0'(\alpha)-u_0'(\overline\alpha_i)\rvert}{\lvert\alpha-\overline\alpha_i\rvert^q}\ ] ] + is finite . as a result , for @xmath154 and @xmath79 , @xmath91 is h@xmath155lder continuous at @xmath83 .
analogously for @xmath67 , since + @xmath156_{_{q;\underline\alpha_j}}=\sup_{\alpha\in\mathcal{d}_j}\frac{\lvert u_0'(\alpha)-u_0'(\underline\alpha_j)\rvert}{\lvert\alpha-\underline\alpha_j\rvert^q}\ ] ] + is defined by ( [ eq : expnew00 ] ) . * for @xmath79 and either @xmath157 , @xmath158 , or @xmath143 odd , @xmath159 .
similarly for @xmath67 . satisfying ( [ eq : expnew0 ] ) for several values of @xmath143 , @xmath160 , @xmath161 and @xmath162 . ]
[ sec : blowup ] in this section , we study regularity properties in solutions to ( [ eq : nonhomo])-([eq : pbc ] ) which , according to the sign of @xmath74 , arise from initial data satisfying ( [ eq : expnew0 ] ) and/or ( [ eq : expnew00 ] ) .
more particularly , finite - time blow - up and global existence in time are examined using @xmath32 banach spaces for @xmath19 $ ] .
set @xmath163 then , as @xmath164 the space - dependent term in ( [ eq : mainsolu ] ) will diverge for certain choices of @xmath110 and not at all for others .
specifically , for @xmath165 @xmath166 blows up earliest as @xmath167 at @xmath168 since @xmath169 similarly for @xmath170 diverges first at @xmath171 and @xmath172 however , blow - up of ( [ eq : mainsolu ] ) does not necessarily follow from this ; we will need to estimate the behaviour of the time - dependent integrals @xmath173 as @xmath174 to this end , in some of the proofs we find convenient the use of the gauss hypergeometric series ( @xcite , @xcite , @xcite ) @xmath175\equiv\sum_{k=0}^{\infty}\frac{\left(a\right)_k(b)_k}{\left(c\right)_k\,k!}z^k,\,\,\,\,\,\,\,\,\,\,\,\lvert z\rvert < 1 , \end{split}\ ] ] for @xmath176 and @xmath177 , the pochhammer symbol @xmath178 , @xmath179 also , we will make use of the following results : [ lem : analcont ] suppose @xmath180 and @xmath181 then the analytic continuation for @xmath182 of the series ( [ eq:2f1 ] ) is given by @xmath183=&\frac{\gamma(c)\gamma(a - b)(-z)^{-b}{}_2f_1[b,1+b - c;1+b - a;z^{-1}]}{\gamma(a)\gamma(c - b ) } \\ & + \frac{\gamma(c)\gamma(b - a)(-z)^{-a}{}_2f_1[a,1+a - c;1+a - b;z^{-1}]}{\gamma(b)\gamma(c - a ) } \end{split}\ ] ] where @xmath184 denotes the standard gamma function .
see for instance @xcite , @xcite .
[ lem : diff ] suppose @xmath185 , @xmath186 and @xmath187 for some @xmath188 then @xmath189\right)=(\epsilon+c_0\left|\beta-\beta_0\right|^q)^{-b } \end{split}\ ] ] for all @xmath131 and @xmath190 lemma [ lem : diff ] above is a generalization of lemma 4.5 in @xcite .
its proof follows similar reasoning .
finally , the next lemma provides us with additional tools for estimating the behaviour , as @xmath191 , of time - dependent integrals of the type @xmath192 .
its proof is deferred to
[ subsec : generalcase ] .
[ lem : general ] for some @xmath131 , suppose @xmath126 satisfies ( [ eq : expnew0 ] ) when @xmath193 , or ( [ eq : expnew00 ] ) if @xmath194 .
it holds : \1 .
if @xmath193 and @xmath195 , @xmath196 for @xmath197 small and positive constants @xmath198 given by @xmath199 here , @xmath200 denotes the finite number of locations @xmath83 in @xmath55 $ ] .
if @xmath194 and @xmath195 , @xmath201 for @xmath197 small and positive constants @xmath198 determined by @xmath202 above , @xmath203 represents the finite number of points @xmath88 in @xmath55 $ ] .
suppose @xmath204 and @xmath205 , or @xmath206 and @xmath207 , satisfy @xmath208 , @xmath209 , @xmath210 .
then for @xmath98 and @xmath211 as defined in ( [ eq : defeta * ] ) , @xmath212 for @xmath197 small and positive constants @xmath198 that depend on the choice of @xmath74 , @xmath209 and @xmath213 .
similarly , the integral remains bounded , and positive , for all @xmath214 $ ] and @xmath98 when @xmath215 and @xmath131 .
the outline of this section is as follows . in
[ subsec : lin ] , we examine @xmath18 , @xmath19 $ ] regularity of solutions arising from initial data satisfying ( [ eq : expnew0 ] ) and/or ( [ eq : expnew00 ] ) for @xmath93
. then , in [ subsec : generalcase ] the case of arbitrary @xmath131 is studied .
also , regularity results concerning a class of smooth initial data larger than the one studied in @xcite are discussed .
we remark that the case @xmath93 is considered separately from the more general argument in [ subsec : generalcase ] , due to the assumptions in lemma [ lem : general ] .
[ subsec : lin ] in @xcite , we showed that for a particular choice of piecewise linear @xmath126 , a special class of solutions to the 2d euler equations ( @xmath216 ) could develop a singularity in finite - time , whereas , for the corresponding 3d problem @xmath217 , solutions may converge to a nontrivial steady state as @xmath218 . therefore , it is of particular interest to determine how these results generalize to initial data satisfying ( [ eq : expnew0 ] ) for @xmath93 .
in fact , in this section we will examine @xmath18 regularity in @xmath33 for @xmath1 and @xmath19 $ ] .
[ thm : p=1 ] consider the initial boundary value problem ( [ eq : nonhomo])-([eq : pbc ] ) with @xmath126 satisfying , for @xmath93 , either ( [ eq : expnew0 ] ) when @xmath79 , or ( [ eq : expnew00 ] ) if @xmath67 .
it holds , 1 .
[ it : two ] for @xmath57 , there exists a finite @xmath21 such that both the maximum @xmath220 and the minimum @xmath221 diverge to @xmath47 and respectively to @xmath222 as @xmath44 . moreover , for every @xmath223 @xmath224 ( two - sided , everywhere blow - up ) .
[ it : one ] for @xmath59 $ ] , solutions exist globally in time .
more particularly , these vanish as @xmath29 for @xmath60 but converge to a nontrivial steady - state if @xmath225 3 .
[ it : three ] for @xmath67 , there is a finite @xmath21 such that only the minimum diverges , @xmath226 as @xmath44 ( one - sided , discrete blow - up ) .
let @xmath198 denote a positive constant which may depend on @xmath98 .
* proofs of statements * ( [ it : two ] ) * and * ( [ it : one ] ) for simplicity , we prove ( [ it : two ] ) and ( [ it : one ] ) for the case where @xmath227 occurs at a single location @xmath228 . by ( [ eq : expnew0 ] )
, there is @xmath134 small enough such that @xmath229 for @xmath230 , @xmath231 and @xmath232 .
then @xmath233 for @xmath234 .
consequently , setting @xmath235 in ( [ eq : app ] ) gives @xmath236 for @xmath197 small , @xmath237 and @xmath238 following a similar argument , or using lemma [ lem : general](1 ) with @xmath239 and @xmath93 , we estimate @xmath240 for any @xmath79 .
suppose @xmath241 then , ( [ eq : mainsolu ] ) , ( [ eq : phiest1])i ) and ( [ eq : phi12 ] ) give @xmath242 for @xmath197 small . setting @xmath243 into ( [ eq : est21 ] ) and using ( [ eq :
maxmin])i ) implies that @xmath244 as @xmath191
. however , if @xmath245 the second term in ( [ eq : est21 ] ) dominates and @xmath246 the existence of a finite @xmath21 for all @xmath247 follows from ( [ eq : etaivp ] ) and ( [ eq : phiest1])i ) , which imply @xmath248 now let @xmath249 using ( [ eq : phiest1])ii ) and ( [ eq : phi12 ] ) on ( [ eq : mainsolu ] ) , yields @xmath250 for @xmath197 small . setting @xmath243 in ( [ eq : u1 ] ) implies @xmath251 as @xmath252 if instead @xmath245 @xmath253 as @xmath174 for the threshold parameter @xmath254 we keep track of the constants and find that , as @xmath255 @xmath256 finally , ( [ eq : etaivp ] ) and ( [ eq : phiest1])ii ) imply that @xmath257 so that @xmath258 as a result , @xmath68 if @xmath259 $ ] but @xmath66 for @xmath260 .
lastly , @xmath261 for @xmath262 small and @xmath216 . then , two - sided , everywhere blow - up in finite - time follows just as above from ( [ eq : mainsolu ] ) , ( [ eq : etaivp ] ) , ( [ eq : phi12 ] ) and ( [ eq : l1 ] ) .
finally , the case @xmath263 follows from the results in @xcite .
* proof of statement * ( [ it : three ] ) for @xmath67 , set @xmath264 .
then @xmath265 remains finite , and positive , for all @xmath214 $ ] .
in fact , one can easily show that @xmath266 if @xmath267 , while @xmath268 for @xmath71 .
similarly , when @xmath267 and @xmath214 $ ] , @xmath269 however , if @xmath270 we need to estimate @xmath271 for @xmath197 small . to do so , we proceed analogously to the derivation of ( [ eq : phiest1 ] ) . for simplicity ,
assume @xmath126 achieves its least value @xmath272 at a single point @xmath273 .
then ( [ eq : expnew00 ] ) with @xmath93 implies that @xmath274 for @xmath275 , @xmath276 and @xmath138 .
it follows that @xmath277 for @xmath232 . by substituting @xmath278 into ( [ eq : l- ] ) , we find that @xmath271 has a finite , positive limit as @xmath191 for @xmath71 .
this implies that for @xmath67 , both time - dependent integrals in ( [ eq : mainsolu ] ) remain bounded and positive for all @xmath214 $ ] .
consequently , blow - up of ( [ eq : mainsolu ] ) , as @xmath191 , will follow from the space - dependent term , @xmath166 , evaluated at @xmath279 . in this way
, we set @xmath279 into ( [ eq : mainsolu ] ) and use ( [ eq : maxmin])ii ) to obtain @xmath280 as @xmath191 .
in contrast , for @xmath281 the definition of @xmath282 implies that the space - dependent term now remains bounded for @xmath214 $ ] .
finally , the existence of a finite blow - up time @xmath21 for the minimum follows from ( [ eq : etaivp ] ) and the estimates on @xmath265 .
in fact , by ( [ eq : etaivp ] ) , @xmath283 for @xmath284 , while ( @xcite ) @xmath285 see [ sec : examples ] for examples . in preparation for the next section ,
we recall some formulas , as well as upper and lower bounds , derived in @xcite for the @xmath18 norm of @xmath33 . for
as long as a solution exists , ( [ eq : sum ] ) and ( [ eq : mainsolu ] ) imply that @xmath286 for @xmath287 and @xmath35 . using the above and some standard inequalities yields @xmath288 and @xmath289 moreover , the energy function @xmath290 is explicitly given by @xmath291
lastly , multiplying ( [ eq : nonhomo])i ) by @xmath33 , integrating by parts , and using ( [ eq : pbc ] ) , ( [ eq : sum ] ) and ( [ eq : mainsolu ] ) , gives @xmath292\frac{\bar{\mathcal{k}}_1(t)}{\bar{\mathcal{k}}_0(t)^{1 + 6\lambda}}. \end{split}\ ] ] the reader may refer to @xcite for details on the above .
[ subsubsec : linlp ] in the previous section , we established the existence of a finite @xmath21 such that @xmath293 diverges as @xmath44 for all @xmath294 $ ] and initial data satisfying ( [ eq : expnew0 ] ) and/or ( [ eq : expnew00 ] ) for @xmath93 relative to the sign of @xmath74 .
if instead , @xmath59 $ ] , we proved that solutions remain in @xmath219 for all time . in this section , we examine further @xmath18 regularity of @xmath33 , as @xmath44 , for @xmath294 $ ] and @xmath35 .
[ thm : q=1,p>=1 ] for the initial boundary value problem ( [ eq : nonhomo])-([eq : pbc ] ) , let @xmath21 denote the finite @xmath219 blow - up time for @xmath33 in theorem [ thm : p=1 ] .
further , for @xmath93 , suppose @xmath126 satisfies ( [ eq : expnew0 ] ) when @xmath79 , or ( [ eq : expnew00 ] ) if @xmath67 . 1 .
[ it : onep ] for @xmath57 and @xmath295 , @xmath41 .
[ it : twop ] for @xmath67 and @xmath45 $ ] , @xmath33 remains integrable ; moreover , if @xmath296 and @xmath295 , then @xmath297 for all @xmath45 $ ] .
[ it : threep ] the energy @xmath42 diverges if @xmath298\cup(1/2,+\infty)$ ] as @xmath44 but remains finite for @xmath45 $ ] if @xmath299 .
also , @xmath300 when
@xmath301 , whereas , @xmath49 if @xmath50 while @xmath46 stays bounded for @xmath45 $ ] if @xmath302 .
let @xmath198 denote a positive constant that may depend on the choice of @xmath74 and @xmath35 .
* proof of statement * ( [ it : onep ] ) first , suppose @xmath79 and set @xmath237 . for simplicity , we prove part ( [ it : onep ] ) under the assumption that @xmath303 occurs at a single point @xmath228 . using lemma [ lem : general](1 ) with @xmath304 , @xmath93 and @xmath37 , yields @xmath305 for @xmath197 small .
similarly , taking @xmath306 we find that @xmath307 moreover , following the argument that led to estimate ( [ eq : phiest1 ] ) , with @xmath308 instead of @xmath309 , gives @xmath310 for @xmath37 and @xmath197 small .
suppose @xmath311 so that @xmath312
. then , using ( [ eq : phiest1])i ) , ( [ eq : phi12 ] ) , ( [ eq : p13 ] ) and ( [ eq : p14])ii ) in ( [ eq : lower ] ) , implies that @xmath313 as @xmath191 .
next , let @xmath314 and @xmath315 . then
, using ( [ eq : phiest1])ii ) , ( [ eq : phi12 ] ) , ( [ eq : p13 ] ) and ( [ eq : p14])i ) in ( [ eq : lower ] ) , gives @xmath316 for @xmath197 small and @xmath317 .
however , for @xmath74 and @xmath318 as prescribed , we see that @xmath319 for @xmath320 and @xmath321 .
therefore , for any @xmath260 there is @xmath322 arbitrarily small such that @xmath323 as @xmath191 . finally , if @xmath216 we have @xmath312 for @xmath295 , as a result , ( [ eq : phi12 ] ) , ( [ eq : l1 ] ) , ( [ eq : p13 ] ) and ( [ eq : p14])iii ) imply that @xmath324 for @xmath262 small , and so , @xmath323 as @xmath191 .
the existence of a finite blow - up time @xmath21 follows from theorem [ thm : p=1 ] .
* proof of statement * ( [ it : twop ] ) suppose @xmath67 and set @xmath264 .
first , recall from the proof of theorem [ thm : p=1 ] that @xmath325 remain finite and positive for all @xmath214 $ ] .
furthermore , in theorem [ thm : p=1 ] we established the existence of a finite blow - up time @xmath21 for the minimum @xmath221 .
consequently , the upper bound ( [ eq : upper ] ) implies that @xmath326 for @xmath67 and @xmath37 .
however , if @xmath327 , ( [ eq : ref2])ii ) is just @xmath271 , which remains finite as @xmath44 . as a result , @xmath328 for all
@xmath45 $ ] and @xmath67 .
if @xmath295 , we recreate the argument in ( [ eq : l- ] ) , with @xmath329 instead of @xmath330 , and find that for @xmath296 and @xmath295 , the integral remains finite and positive as @xmath191 .
consequently , ( [ eq : ref2 ] ) implies that @xmath331 for all @xmath296 and @xmath295 .
we remark that the lower bound ( [ eq : lower ] ) yields no information regarding @xmath18 blow - up of @xmath33 , as @xmath44 , for parameter values @xmath332 , @xmath295 .
nonetheless , we can use ( [ eq : energy ] ) and ( [ eq : derenergy ] ) to obtain additional blow - up information on energy - related quantities .
* proof of statement * ( [ it : threep ] ) for @xmath57 , blow - up of @xmath333 and @xmath46 to @xmath47 as @xmath44 is a consequence of part ( [ it : onep ] ) above .
further , setting @xmath334 in part ( [ it : twop ] ) implies that @xmath333 remains bounded for all @xmath299 and @xmath45 $ ] .
now , ( [ eq : derenergy])i ) yields @xmath335 and so setting @xmath336 in part ( [ it : twop ] ) implies that @xmath46 remains finite for @xmath337 and @xmath45 $ ] . according to these results , we have yet to determine the behaviour of @xmath333 as @xmath44 for @xmath338 and @xmath46 when @xmath339 .
to do so , we will use formulas ( [ eq : energy ] ) and ( [ eq : derenergy ] ) . from lemma [ lem : general](2 ) with @xmath340 , @xmath93 and @xmath339 ,
we find that @xmath341 for @xmath197 small .
also , following the argument in ( [ eq : l- ] ) , with @xmath342 instead of @xmath330 , we derive @xmath343 since both @xmath192 , @xmath344 stay finite and positive for all @xmath214 $ ] and @xmath67 , ( [ eq : energy ] ) tells us that blow - up in @xmath345 leads to a diverging @xmath333 .
then , ( [ eq : k2])i ) implies that for @xmath71 , @xmath346 as @xmath191 .
similarly for @xmath284 by using ( [ eq : k2])ii ) instead .
clearly , this also implies blow - up of @xmath46 to @xmath47 as @xmath44 for all @xmath338 .
finally , from ( [ eq : derenergy])ii ) , ( [ eq : k3 ] ) and ( [ eq : k2])iii ) , @xmath347 as @xmath191 for all @xmath348 .
the existence of a finite @xmath21 follows from theorem [ thm : p=1]([it : three ] ) . from the results
established thus far , we are able to obtain a complete description of the @xmath349 regularity for @xmath33 : if @xmath59 $ ] , @xmath350 where @xmath73 for @xmath61 but @xmath351 if @xmath60 , while , for @xmath21 the finite @xmath219 blow - up time for @xmath33 in theorem [ thm : p=1 ] , @xmath352\cup(1/2,+\infty ) ,
\\ c\in\mathbb{r}^+,\,\,\,\,&\lambda\in(-1/2,0 ) .
\end{cases}\ ] ] for @xmath21 the finite @xmath219 blow - up time for @xmath33 in theorem [ thm : p=1 ] , we may use ( [ eq : derenergy ] ) , ( [ eq : k3 ] ) and ( [ eq : k2 ] ) , as well as theorem [ thm : q=1,p>=1 ] , to establish a global bound on @xmath353 if @xmath59 $ ] , or for @xmath45 $ ] when @xmath302 , whereas @xmath354 we also note that , unlike the result in theorem [ thm : lpintro]([it : ener ] ) of [ sec : intro ] , ( [ eq:3integral ] ) and the change in sign through @xmath50 of the term @xmath355 in ( [ eq : derenergy ] ) , prevent the possibility of blow - up of @xmath46 towards @xmath222 , which might otherwise have played a role in the study of weak solutions from the point of view of energy dissipation .
notice that the two - sided , everywhere blow - up found in theorem [ thm : p=1 ] for @xmath57 corresponds , in theorem [ thm : q=1,p>=1 ] , to @xmath18 blow - up of @xmath33 for any @xmath295 . on the other hand , @xmath33 remains integrable for all @xmath67 and @xmath45 $ ] but , as @xmath44 , undergoes an @xmath219 blow - up of the one - sided , discrete type for @xmath67 .
then , as the magnitude of @xmath67 decreases , @xmath33 is guaranteed to remain , for @xmath45 $ ] , in smaller @xmath18 spaces with @xmath39 . in the coming sections , we will find that a similar correspondence between the `` strengths '' of the @xmath219 and @xmath18 , @xmath35 , blow - up in @xmath33 , as @xmath44 , also holds for other @xmath143 .
[ subsec : generalcase ] in this section , we study the case of arbitrary @xmath143 .
as in the previous sections , @xmath356 regularity of @xmath33 for @xmath1 and @xmath19 $ ] is examined .
in addition , the behaviour of the jacobian ( [ eq : sum ] ) is considered .
particularly , we will show that if @xmath357 , no blow - up occurs in stagnation point - form solutions to the 3d incompressible euler equations , whereas , for the corresponding 2d case , no spontaneous singularity forms when @xmath358 .
finally , a class of smooth , periodic initial data larger than the one considered in @xcite is studied . before stating and proving our results , we first establish lemma [ lem : general ] and obtain estimates on @xmath265 and @xmath271 .
* proof of lemma * [ lem : general](1 ) for simplicity , we prove statement ( 1 ) for functions @xmath91 that attain their greatest value @xmath303 at a single location @xmath228 . the case of several @xmath82 $ ] follows similarly . from ( [ eq : expnew0 ] ) , there is @xmath134 such that @xmath359 for @xmath131 , @xmath232 and @xmath230 .
therefore @xmath360 \end{split}\ ] ] for @xmath361 . making the change of variables @xmath362 in the first and second integrals inside the bracket , respectively , we find that @xmath363 for small @xmath232 .
suppose @xmath195 , then setting @xmath235 in ( [ eq : general2 ] ) implies @xmath364 for @xmath197 small , @xmath237 and @xmath365 now , since the beta function satisfies ( see for instance @xcite ) : @xmath366 for @xmath367 , then , letting @xmath368 , @xmath369 and @xmath370 into ( [ eq : gammarel])i ) , and using ( [ eq : gammarel])ii ) , one has @xmath371 the result follows from ( [ eq : general3 ] ) , ( [ eq : generalcst ] ) and ( [ eq : gammarel2 ] )
. * proof of lemma * [ lem : general](2 ) follows from an analogous argument using ( [ eq : expnew00 ] ) and @xmath264 instead .
* proof of lemma * [ lem : general](3 ) the last claim in ( 3 ) follows trivially if @xmath215 and @xmath131 due to the `` almost everywhere '' continuity and boundedness of @xmath91 . to establish the remaining claims , we make use of lemmas [ lem : analcont ] and [ lem : diff ] . however , in order to use the latter , we require that @xmath207 and @xmath372 .
since the case @xmath373 was established in parts ( 1 ) and ( 2 ) above , suppose that @xmath205 and @xmath207 , or equivalently @xmath204 and @xmath205 , or @xmath206 and @xmath207 .
first , for @xmath213 and @xmath209 as prescribed , consider @xmath79 and , for simplicity , assume @xmath227 occurs at a single point @xmath228 .
then , ( [ eq : expnew0 ] ) and lemma [ lem : diff ] imply that @xmath374 \end{split}\ ] ] for @xmath375 and @xmath230 .
now , the restriction on @xmath376 implies that @xmath377 .
however , our ultimate goal is to let @xmath376 vanish , so that , eventually , the argument @xmath378 of the series in ( [ eq : lastgen1])ii ) will leave the unit circle , particularly @xmath379 . at that point ,
definition [ eq:2f1 ] for the series no longer holds and we turn to its analytic continuation in lemma [ lem : analcont ] .
accordingly , taking @xmath232 small enough such that @xmath380 , we apply lemma [ lem : analcont ] to ( [ eq : lastgen1 ] ) and obtain @xmath381=\frac{2r^{1-qb}}{(1-bq)\left|c_1\right|^b}+\frac{2\gamma\left(1+\frac{1}{q}\right)\gamma\left(b-\frac{1}{q}\right)}{\gamma(b)\left|c_1\right|^{\frac{1}{q}}\epsilon^{b-\frac{1}{q}}}+\psi(\epsilon ) \end{split}\ ] ] for @xmath382 as @xmath383 , and either @xmath204 and @xmath205 , or @xmath206 and @xmath207 .
in addition , due to the assumptions in lemma [ lem : analcont ] we require that @xmath208 , @xmath209 , @xmath210 .
finally , since @xmath384 , substituting @xmath235 into ( [ eq : lastgen1 ] ) and ( [ eq : lastgen2 ] ) , implies that @xmath385 for @xmath197 small , @xmath237 , and positive constants @xmath198 that depend on @xmath79 , @xmath209 and @xmath213 .
an analogous argument follows for @xmath67 by using ( [ eq : expnew00 ] ) instead of ( [ eq : expnew0]).@xmath386 using lemma [ lem : general ] , we now derive estimates for @xmath192 , @xmath344 , which will be used in subsequent regularity theorems .
[ subsubsec : integralestimates ] * for parameters @xmath387 * for @xmath79 , we set @xmath388 into lemma [ lem : general](1)-(3 ) to obtain @xmath389 for @xmath197 small and positive constants @xmath390 given by @xmath391 also , in ( [ eq : firstint1])i ) we assume that @xmath74 and @xmath213 satisfy , whenever applicable , @xmath392 we note that corresponding estimates for the missing values may be obtained via a simple continuity argument . similarly , taking @xmath239 we find @xmath393 with positive constants @xmath394 determined by @xmath395 additionally , for ( [ eq : secint1])i )
we assume that @xmath74 and @xmath213 satisfy ( [ eq : lemmaass1 ] ) .
* for parameters @xmath396 * for @xmath67 and @xmath388 , lemma [ lem : general](3 ) implies that @xmath397 for @xmath197 small .
similarly , parts ( 2 ) and ( 3 ) , now with @xmath239 , yield @xmath398 for either @xmath399 whereas @xmath400 for @xmath401 , @xmath402 and positive constants @xmath403 determined by @xmath404 [ subsubsec : generalcaselambdapospinf ] in this section , we use the estimates in [ subsubsec : integralestimates ] to examine the @xmath219 regularity of @xmath33 for @xmath406 and @xmath91 satisfying ( [ eq : expnew0 ] ) for some @xmath131 .
furthermore , the behaviour of the jacobian ( [ eq : sum ] ) is also studied .
[ thm : lambdapos ] consider the initial boundary value problem ( [ eq : nonhomo])-([eq : pbc ] ) for @xmath126 satisfying ( [ eq : expnew0 ] ) . 1 .
[ it : global ] if @xmath131 and @xmath407,$ ] solutions exist globally in time .
more particularly , these vanish as @xmath29 for @xmath408 but converge to a nontrivial steady state if @xmath409 .
[ it : blow1 ] if @xmath131 and @xmath410 , there exists a finite @xmath21 such that both the maximum @xmath220 and the minimum @xmath221 diverge to @xmath47 and respectively to @xmath222 as @xmath44 .
moreover , @xmath224 for @xmath411 ( two - sided , everywhere blow - up ) .
[ it : blow2 ] for @xmath206 and @xmath247 such that @xmath412 and @xmath413 for all @xmath414 there is a finite @xmath21 such that only the maximum blows up , @xmath415 as @xmath44 ( one - sided , discrete blow - up ) .
further , if @xmath416 for @xmath417 , a two - sided , everywhere blow - up ( as described in ( [ it : blow1 ] ) above ) occurs at a finite @xmath21 .
[ it : blow3 ] suppose @xmath418 .
then for @xmath419 , there exists a finite @xmath21 such that , as @xmath44 , two - sided , everywhere blow - up develops .
if instead @xmath420 , only the maximum diverges , @xmath421 , as @xmath422 . 5 .
[ it : blow4 ] for @xmath423 , there is a finite @xmath21 such that @xmath33 undergoes a two - sided , everywhere blow - up as @xmath44 .
suppose @xmath424 , let @xmath198 denote a positive constant which may depend on @xmath74 and @xmath213 , and set @xmath425 * proof of statements * ( [ it : global ] ) * and * ( [ it : blow1 ] ) suppose @xmath426 for some @xmath143 .
then , for @xmath197 small @xmath265 satisfies ( [ eq : firstint1])ii ) while @xmath271 obeys ( [ eq : secint1])ii ) .
consequently , ( [ eq : mainsolu ] ) implies that @xmath427 for positive constants @xmath390 and @xmath394 given by ( [ eq : cst1 ] ) and ( [ eq : const2 ] ) .
but for @xmath428 and @xmath429 , ( [ eq : gammarel])ii ) , ( [ eq : cst1 ] ) and ( [ eq : const2 ] ) yield @xmath430 as a result , setting @xmath431 in ( [ eq : ok3 ] ) and using ( [ eq : maxmin])i ) implies that @xmath432 for @xmath197 small , whereas , if @xmath433 , @xmath434 clearly , when @xmath409 , @xmath435 as @xmath191 , while , for @xmath433 , @xmath436 if @xmath408 , ( [ eq : ux5 ] ) now implies that @xmath437 as @xmath191 , whereas , using ( [ eq : ux6 ] ) for @xmath433 , @xmath438 in contrast , if @xmath410 , @xmath439 . then ( [ eq : ux5 ] ) and ( [ eq : ux6 ] ) yield @xmath440 as @xmath191 , but @xmath441 for @xmath433 .
lastly , rewriting ( [ eq : etaivp ] ) as @xmath442 and using ( [ eq : firstint1])ii ) , we obtain @xmath443 or equivalently @xmath444 consequently , @xmath68 for @xmath445 $ ] , while @xmath66 if @xmath410 .
lastly , the case @xmath263 follows from the results in @xcite .
* proof of statement * ( [ it : blow2 ] ) first , suppose @xmath206 and @xmath247 satisfy ( [ eq : lemmaass1 ] )
. then @xmath265 and @xmath271 satisfy ( [ eq : firstint1])i ) and ( [ eq : secint1])i ) , respectively .
therefore , ( [ eq : mainsolu ] ) implies that @xmath446 for @xmath197 small .
setting @xmath431 into ( [ eq : blow3case ] ) and using ( [ eq : maxmin])i ) gives @xmath447 as @xmath191 , while , if @xmath433 , @xmath448 remains finite for all @xmath214 $ ] due to the definition of @xmath227 .
the existence of a finite blow - up time @xmath21 for the maximum is guaranteed by ( [ eq : firstint1])i ) and ( [ eq : time ] ) , which lead to @xmath449 next , suppose @xmath416 for @xmath417 , so that @xmath450 .
then , using ( [ eq : firstint1])i ) and ( [ eq : secint1])ii ) in ( [ eq : mainsolu ] ) , we find that @xmath451 for @xmath197 small .
set @xmath431 into the above and use @xmath452 to obtain @xmath453 as @xmath191 . on the other hand , for @xmath433 , the space - dependent in ( [ eq : blow4case ] ) now remains finite for all @xmath214 $ ] . as a result
, the second term dominates and @xmath454 as @xmath191 .
the existence of a finite blow - up time @xmath21 , follows , as in the previous case , from ( [ eq : time ] ) and ( [ eq : firstint1])i ) .
* proof of statement * ( [ it : blow3 ] ) part ( [ it : blow3 ] ) follows from an argument analogous to the one above .
briefly , if @xmath419 for @xmath418 , we use estimates ( [ eq : firstint1])i ) and ( [ eq : secint1])ii ) on ( [ eq : mainsolu ] ) to get ( [ eq : blow4case ] ) , with different positive constants @xmath198 .
two - sided , everywhere blow - up in finite - time then follows just as above .
if instead @xmath420 for @xmath418 , then ( [ eq : firstint1])i ) still holds but @xmath271 now remains bounded for all @xmath214 $ ] ; it satisfies ( [ eq : secint1])i ) .
therefore , up to different positive constants @xmath198 , ( [ eq : mainsolu ] ) leads to ( [ eq : blow3case ] ) , and so only the maximum diverges , @xmath421 , as @xmath455 approaches some finite @xmath21 whose existence is guaranteed by ( [ eq : timeblow3case ] ) .
* proof of statement * ( [ it : blow4 ] ) for @xmath423 , ( [ eq : firstint1])i ) , ( [ eq : secint1])ii ) and ( [ eq : mainsolu ] ) imply ( [ eq : blow4case ] ) .
then , we follow the argument used to establish the second part of ( [ it : blow2 ] ) to show that two - sided , everywhere finite - time blow - up occurs . see [ sec : examples ] for examples .
theorems [ thm : p=1 ] and [ thm : lambdapos ] allow us to predict the regularity of stagnation point - form ( spf ) solutions to the two ( @xmath216 ) and three ( @xmath61 ) dimensional incompressible euler equations assuming we know something about the curvature of the initial data @xmath456 near @xmath83 . setting @xmath216 into theorem [ thm : lambdapos]([it : global ] )
implies that spf solutions in the 2d setting persist for all time if @xmath91 satisfies ( [ eq : expnew0 ] ) for arbitrary @xmath358 . on the contrary , theorems [ thm : p=1 ] and [ thm : lambdapos]([it : blow1])-([it : blow3 ] ) ,
tell us that if @xmath457 , two - sided , everywhere finite - time blow - up occurs .
analogously , solutions to the corresponding 3d problem exist globally in time for @xmath357 , whereas , two - sided , everywhere blow - up develops when @xmath418 .
see table [ table : thecase ] below .
finally , we remark that finite - time blow - up in @xmath33 is expected for both the two and three dimensional equations if @xmath458 $ ] .
see for instance [ sec : examples ] for a blow - up example in the 3d case with @xmath459 .
| m2.0 cm | m3.4 cm | m3.4 cm | & & @xmath460 + & @xmath461 & @xmath461 + & @xmath461 & @xmath462 + & @xmath463 & @xmath462 + [ table : thecase ] corollary [ coro : lambda>1/2 ] below briefly examines the behaviour , as @xmath44 , of the jacobian ( [ eq : sum ] ) for @xmath21 as in theorem [ thm : lambdapos ] .
[ coro : lambda>1/2 ] consider the initial boundary value problem ( [ eq : nonhomo])-([eq : pbc ] ) with @xmath126 satisfying ( [ eq : expnew0 ] ) for @xmath131 , and let @xmath21 be as in theorem [ thm : lambdapos ] .
it follows , 1 .
[ it : item1 ] for @xmath143 and @xmath426 , @xmath464 where @xmath68 for @xmath445 $ ] , while @xmath66 if @xmath410 .
[ it : item1.5 ] suppose @xmath465 , or @xmath206 and @xmath57 , satisfy ( [ eq : lemmaass1 ] ) . then , there exists a finite @xmath21 such that @xmath466 where @xmath467 depends on the choice of @xmath74 , @xmath213 and @xmath433 . the limits ( [ eq : jacone ] ) and ( [ eq : jacone22 ] ) follow straightforwardly from ( [ eq : sum ] ) and estimates ( [ eq : firstint1])ii ) and ( [ eq : firstint1])i ) , respectively ; whereas , the finite or infinite character of @xmath21 is a consequence of theorem [ thm : lambdapos ] . [
subsubsec : generalcaselambdapospfin ] from theorem [ thm : lambdapos ] , if @xmath407 $ ] for @xmath131 , solutions remain in @xmath219 for all time ; otherwise , @xmath293 diverges as @xmath455 approaches some finite @xmath21 . in this section ,
we study further properties of @xmath18 regularity in @xmath33 , as @xmath44 , for @xmath469 , @xmath35 and initial data @xmath126 satisfying ( [ eq : expnew0 ] ) .
to do so , we will use the upper and lower bounds ( [ eq : upper ] ) and ( [ eq : lower ] ) .
consequently , for @xmath197 small and @xmath237 , estimates on the behaviour of the time - dependent integrals @xmath470 are required . since these may be obtained directly from lemma [ lem : general](1)-(3 ) , we omit the details and state our findings below . for @xmath35 , @xmath471 with positive constants
@xmath472 also @xmath473 for either @xmath474 whereas @xmath475 for @xmath476 the positive constants @xmath477 in ( [ eq : lpos6 ] ) are obtained by replacing every @xmath308 term in ( [ eq : c6 ] ) by @xmath478 . also , due to lemma [ lem : general ] , ( [ eq : lpos3])i ) and ( [ eq : lpos4 ] )
are valid for @xmath479 where a simple continuity argument may again be used ( see ( [ eq : lemmaass1 ] ) ) to obtain estimates for the missing values .
finally @xmath480 for either @xmath481 while @xmath482 if @xmath483,\,\,\,\,\,\ , p\in[1,1/q),\,\,\,\ , 0<\lambda<\frac{q}{1-pq } , \\ & q\in(0,1],\,\,\,\,\,\ , p\geq\frac{1}{q},\,\,\,\,\,\ , \lambda>0 , \\ & q>1,\,\,\,\,\,\,\,\,\,\,\,\,\,\ , p\geq1,\,\,\,\,\,\,\ , \lambda>0 .
\end{cases}\ ] ] estimate ( [ eq : lpos8 ] ) is in turn valid for @xmath484 in what follows , @xmath21 will denote the @xmath219 blow - up time for @xmath33 in theorem [ thm : lambdapos ] .
also , we will assume that ( [ eq : lemmaass1 ] ) , ( [ eq : req1lambdapos1 ] ) and ( [ eq : req1lambdapos2 ] ) hold whenever their corresponding estimates are used .
we begin by considering the lower bound ( [ eq : lower ] ) . in particular
, we will show that two - sided , everywhere blow - up in theorem [ thm : lambdapos ] corresponds to a diverging @xmath485 for all @xmath295 .
then , by studying the upper bound ( [ eq : upper ] ) , we will find that if @xmath131 and @xmath452 are such that only the maximum diverges at a finite @xmath21 , then @xmath33 remains integrable for all @xmath45 $ ] , whereas , its regularity in smaller @xmath18 spaces for @xmath45 $ ] will vary according to the value of the parameter @xmath74 as a function of either @xmath318 , @xmath213 , or both .
suppose @xmath486 for @xmath131 and @xmath314 .
then ( [ eq : lpos3])ii ) holds as well as ( [ eq : firstint1])ii ) , since @xmath487 .
now , if @xmath90 then @xmath488 and so ( [ eq : secint1])ii ) applies , otherwise , ( [ eq : secint1])ii ) also holds for @xmath357 and @xmath79 .
similarly for @xmath90 , we have that @xmath489 so that ( [ eq : lpos6 ] ) is valid .
alternatively , this last estimate also holds if @xmath357 for @xmath79 .
accordingly , using these estimates in ( [ eq : lower ] ) yields , after simplification , @xmath490 for @xmath197 small and @xmath491 consequently , @xmath485 will diverge as @xmath191 if @xmath492 , or equivalently for @xmath493 .
since @xmath486 for @xmath143 and @xmath314 , we find this to be the case as long as @xmath494 therefore , by taking @xmath495 arbitrarily small , we find that @xmath496 for @xmath410 and @xmath143 .
the existence of a finite blow - up time @xmath21 follows from theorem [ thm : lambdapos]([it : blow1 ] ) , while the embedding @xmath497 yields @xmath18 blow - up for any @xmath295 .
next , for @xmath417 we consider values of @xmath74 lying between stagnation point - form solutions to the 2d ( @xmath216 ) and 3d ( @xmath61 ) incompressible euler equations .
suppose @xmath498 for @xmath499 and @xmath417 .
the condition on @xmath318 simply guarantees that @xmath500 for @xmath213 as specified .
furthermore , we have that @xmath501 so that relative to our choice of @xmath74 and @xmath213 , @xmath502 using the above , we find that ( [ eq : firstint1])i ) , ( [ eq : secint1])ii ) , ( [ eq : lpos3])i ) and ( [ eq : lpos6 ] ) hold , and so ( [ eq : lower ] ) leads to @xmath503 for @xmath197 small . therefore , as @xmath191
, @xmath485 will diverge for all @xmath498 , @xmath417 and @xmath499 . here
, we can take @xmath495 arbitrarily small and use ( [ eq : embedding ] ) to conclude the finite - time blow - up , as @xmath44 , of @xmath485 for all @xmath416 , @xmath417 and @xmath295 .
the existence of a finite blow - up time @xmath21 is guaranteed by the second part of theorem [ thm : lambdapos]([it : blow2 ] ) .
now suppose @xmath418 and @xmath504 for @xmath505 .
this means that @xmath465 and @xmath506 consequently , using ( [ eq : firstint1])i ) , ( [ eq : secint1])ii ) , ( [ eq : lpos3])i ) and ( [ eq : lpos6 ] ) in ( [ eq : lower ] ) , implies ( [ eq : touse1 ] ) , possibly with distinct positive constants @xmath198 .
then , as @xmath191 , @xmath507 for all @xmath504 , @xmath418 and @xmath505 .
similarly , if @xmath213 and @xmath318 are as above , but @xmath508 , ( [ eq : firstint1])i ) , ( [ eq : secint1])ii ) , ( [ eq : lpos3])i ) and ( [ eq : lpos4 ] ) imply @xmath509 as @xmath191 . from these last two results and ( [ eq : embedding ] ) , we see that , as @xmath191 , @xmath323 for all @xmath419 , @xmath418 and @xmath295 .
the existence of a finite @xmath21 follows from theorem [ thm : lambdapos]([it : blow3 ] ) .
lastly , suppose @xmath423 and @xmath295 .
then , estimates ( [ eq : firstint1])i ) , ( [ eq : secint1])ii ) , ( [ eq : lpos3])i ) and ( [ eq : lpos6 ] ) hold for @xmath197 small . as a result , ( [ eq : lower ] ) implies ( [ eq : touse1 ] ) , which in turn leads to @xmath18 blow - up of @xmath33 for any @xmath423 and @xmath295 , as @xmath191 .
the existence of a finite @xmath21 is due to theorem [ thm : lambdapos]([it : blow4 ] ) .
notice from the results established so far , that some values of @xmath469 for @xmath143 are missing .
these are precisely the cases for which the lower bound ( [ eq : lower ] ) yields inconclusive information about the @xmath18 regularity of @xmath33 for @xmath39 . to examine some aspects of the @xmath18 regularity of @xmath33 for @xmath45 $ ] and @xmath35 in these particular cases
, we consider the upper bound ( [ eq : upper ] ) .
first , suppose @xmath206 and @xmath510 for @xmath511 .
then @xmath512 so that ( [ eq : firstint1])i ) , ( [ eq : secint1])i ) and ( [ eq : lpos8 ] ) , imply that the integral terms in ( [ eq : upper ] ) remain bounded , and nonzero , for @xmath214 $ ] .
we conclude that @xmath513 for all @xmath510 , @xmath206 and @xmath511 . here
, @xmath21 denotes the finite @xmath219 blow - up time for @xmath33 established in the first part of theorem [ thm : lambdapos]([it : blow2 ] ) .
particularly , this result implies that even though @xmath514 for all @xmath247 when @xmath206 , @xmath33 remains integrable for @xmath45 $ ] .
finally , suppose @xmath418 and @xmath515 for @xmath516 .
then @xmath517 , and so ( [ eq : firstint1])i ) , ( [ eq : secint1])i ) and ( [ eq : lpos8 ] ) hold .
consequently , ( [ eq : upper ] ) implies that @xmath518 for all @xmath515 , @xmath418 and @xmath516 .
this time , @xmath21 stands as the finite @xmath219 blow - up time for @xmath33 established in the second part of theorem [ thm : lambdapos]([it : blow3 ] ) .
furthermore , this result tells us that even though @xmath514 for @xmath420 and @xmath418 , @xmath33 stays integrable for all @xmath45 $ ] .
these last two results on the integrability of @xmath33 , for @xmath45 $ ] , become more apparent if we set @xmath327 in ( [ eq : upper ] ) to obtain @xmath519 the result then follows from the above inequality and estimates ( [ eq : firstint1])i ) and ( [ eq : secint1])i ) .
theorem [ thm : lplambdapos ] below summarizes the above results .
[ thm : lplambdapos ] consider the initial boundary value problem ( [ eq : nonhomo])-([eq : pbc ] ) for @xmath126 satisfying ( [ eq : expnew0 ] ) , and let @xmath21 be as in theorem [ thm : lambdapos ] . 1 . [ it : lplambdapos1 ] for @xmath143 and @xmath407 $ ] , @xmath520 for all @xmath37 .
more particularly , @xmath521 for @xmath408 , while , as @xmath218 , @xmath33 converges to a nontrivial , @xmath219 function when @xmath409 .
[ it : lplambdapos2 ] let @xmath295 .
then , there exists a finite @xmath21 such that for all @xmath143 and @xmath410 , @xmath41 .
similarly for @xmath423 , or @xmath416 , @xmath417 .
[ it : lplambdapos3 ] for all @xmath206 , @xmath510 and @xmath511 , there exists a finite @xmath21 such that @xmath518 ( see theorem [ thm : lambdapos]([it : blow2 ] ) ) .
[ it : lplambdapos4 ] suppose @xmath418 .
then , there exists a finite @xmath21 such that @xmath41 for @xmath419 and @xmath295 , whereas , if @xmath515 and @xmath516 , @xmath518 ( see theorem [ thm : lambdapos]([it : blow3 ] ) ) .
[ subsubsec : generalcaselambdanegpinf ] we now examine the @xmath219 regularity of @xmath33 for parameters @xmath67 and initial data satisfying ( [ eq : expnew00 ] ) for arbitrary @xmath131 .
we prove theorem [ thm : lambdaneg ] below .
[ thm : lambdaneg ] consider the initial boundary value problem ( [ eq : nonhomo])-([eq : pbc ] ) for @xmath126 satisfying ( [ eq : expnew00 ] ) .
furthermore , 1 .
[ it : lambdaneg1 ] suppose @xmath267 and @xmath143 .
then , there exists a finite @xmath21 such that only the minimum diverges , @xmath226 as @xmath44 ( one - sided , discrete blow - up ) .
[ it : lambdaneg2 ] suppose @xmath71 and @xmath90 satisfy @xmath413 and @xmath412 @xmath522 @xmath203
. then , a one - sided discrete blow - up , as described in ( [ it : lambdaneg1 ] ) , occurs in finite - time .
similarly for @xmath523 and @xmath401 .
[ it : lambdaneg3 ] suppose @xmath402 and @xmath401 .
then , there is a finite @xmath21 such that both the maximum @xmath220 and the minimum @xmath221 diverge to @xmath47 and respectively to @xmath222 as @xmath44 .
moreover , @xmath524 for @xmath411 ( two - sided , everywhere blow - up ) .
finally , for @xmath67 , @xmath143 and @xmath21 as above , the jacobian ( [ eq : sum ] ) satisfies @xmath525 where @xmath467 depends on the choice of @xmath74 , @xmath213 and @xmath526 . throughout ,
let @xmath198 denote a positive constant that may depend on @xmath67 , @xmath143 and recall that @xmath527 * proof of statement * ( [ it : lambdaneg1 ] ) suppose @xmath267 and assume @xmath126 satisfies ( [ eq : expnew00 ] ) for some @xmath143 .
then ( [ eq : thirdint1 ] ) and ( [ eq : lastest ] ) imply that both integral terms in ( [ eq : mainsolu ] ) remain finite and nonzero as @xmath191 .
more particularly , one can show that ( [ eq : quick3p ] ) and ( [ eq : quick4p ] ) hold for all @xmath214.$ ] therefore , blow - up of ( [ eq : mainsolu ] ) depends , solely , on the behaviour of the space - dependent term @xmath166 .
accordingly , we set @xmath171 into ( [ eq : mainsolu ] ) and use ( [ eq : maxmin])ii ) to find that the minimum diverges , @xmath528 , as @xmath191 .
however , if @xmath526 , the definition of @xmath282 implies that the space - dependent term now remains bounded , and positive , for @xmath214 $ ] .
the existence of a finite blow - up time @xmath21 for the minimum follows from ( [ eq : etaivp ] ) and ( [ eq : thirdint1 ] ) .
in fact , we may use ( [ eq : etaivp ] ) and ( [ eq : quick3p ] ) to obtain the estimate @xmath529 * proof of statements * ( [ it : lambdaneg2 ] ) and ( [ it : lambdaneg3 ] ) now suppose @xmath71 . as in the previous case ,
the term @xmath265 remains finite , and positive , for all @xmath214 $ ] .
particularly , @xmath265 satisfies ( [ eq : newestq ] ) for all @xmath214 $ ] . on the other hand , @xmath271 now either converges or diverges , as @xmath191 , according to ( [ eq : lastest ] ) or ( [ eq : lastest1 ] ) , respectively . if @xmath71 and @xmath143 are such that ( [ eq : lastest ] ) holds , then part ( [ it : lambdaneg2 ] ) follows just as part ( [ it : lambdaneg1 ] ) .
however , if @xmath401 and @xmath402 , we use ( [ eq : thirdint1 ] ) and ( [ eq : lastest1 ] ) on ( [ eq : mainsolu ] ) , to obtain @xmath530 for @xmath197 small . setting @xmath171 into the above implies that @xmath531 as @xmath191 , whereas , for @xmath532 , the space - dependent term now remains bounded , as a result , the second term dominates and @xmath533 as @xmath191 .
the existence of a finite blow - up time @xmath21 follows as in the case @xmath267 .
in fact , ( [ eq : etaivp ] ) and ( [ eq : newestq ] ) yield the lower bound @xmath534 .
finally , ( [ eq : lastrealjac ] ) is derived straightforwardly from ( [ eq : sum ] ) and ( [ eq : thirdint1 ] ) .
[ sec : examples ] for examples .
[ subsubsec : generalcaselambdanegpfin ] let @xmath21 denote the finite @xmath219 blow - up time for @xmath33 in theorem [ thm : lambdaneg ] above .
in this last section , we briefly examine the @xmath18 regularity of @xmath33 , as @xmath44 , for @xmath194 , @xmath35 and @xmath91 satisfying ( [ eq : expnew00 ] ) for some @xmath131 . as in [ subsubsec : generalcaselambdapospfin ] , we will make use of ( [ eq : upper ] ) and ( [ eq : lower ] ) .
first of all , by the last part of lemma [ lem : general](3 ) , we have that for @xmath143 and @xmath37 , @xmath535 for @xmath197 small , @xmath264 and @xmath67 .
similarly @xmath536 for @xmath537 .
moreover , due to the first part of @xmath538 in the lemma , estimate ( [ eq : fin1 ] ) is also seen to hold , with different positive constants @xmath198 , for @xmath539 , @xmath37 and @xmath143 satisfying either of the following @xmath540,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\ , \lambda<-\frac{1}{p } , \\ q\in(0,1/2),\,\,\ , & p>2,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\ , \frac{1}{2-p}<\lambda<-\frac{1}{p } , \\
q\in(1/2,1),\,\,\ , & p\in[1,1/q],\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\ , \lambda<-\frac{1}{p } , \\ q\in(1/2,1),\,\,\ , & p>\frac{1}{q},\,\,\,\,\,\,\,\,\,\,\,\,\ , \frac{q}{1-pq}<\lambda<-\frac{1}{p } , \\ q>1,\ , & p\geq1,\,\,\,\,\,\,\,\,\,\,\,\,\ , \frac{q}{1-pq}<\lambda<-\frac{1}{p } , \end{cases}\ ] ] as well as @xmath541 we remark that in the cases where ( [ eq : fin1 ] ) diverges , it dominates the other terms in ( [ eq : upper ] ) , regardless of whether these converge or diverge , and so no information on the behaviour of @xmath485 is obtained .
consequently , we will omit those instances . finally , using lemma [ lem : general](2 )
, one finds that @xmath542 for @xmath401 , @xmath37 and @xmath543 .
analogously , if ( [ eq : fin2 ] ) converges , the lower bound ( [ eq : lower ] ) yields no information on the @xmath18 regularity of @xmath33 . for the remaining of this section
, we will assume that ( [ eq : lemmaass1 ] ) holds whenever ( [ eq : lastest ] ) is used for @xmath71 and @xmath90 . also , ( [ eq : req2lambdaneg1 ] ) will be valid in those cases where estimate ( [ eq : fin1 ] ) is considered for @xmath74 , @xmath318 and @xmath213 as in ( [ eq : laste0 ] ) .
suppose @xmath544 for @xmath401 and @xmath295 .
then , using ( [ eq : thirdint1 ] ) , ( [ eq : lastest ] ) , ( [ eq : reallast ] ) and ( [ eq : fin2 ] ) , in the lower bound ( [ eq : lower ] ) , implies that @xmath545 if instead , @xmath402 for @xmath401 and @xmath295 , then ( [ eq : thirdint1 ] ) , ( [ eq : lastest1 ] ) , ( [ eq : reallast ] ) and ( [ eq : fin2 ] ) give @xmath546 as @xmath191 . for the upper bound ( [ eq : upper ] ) , we simply mention that estimates ( [ eq : thirdint1 ] ) , ( [ eq : lastest ] ) and ( [ eq : fin1 ] ) lead to several instances where @xmath485 remains finite for all @xmath45 $ ] .
this can be shown , just as above , by using the appropriate estimates .
for simplicity , we omit the details and summarize the results in theorem [ thm : genlplambdaneg ] below .
[ thm : genlplambdaneg ] consider the initial boundary value problem ( [ eq : nonhomo])-([eq : pbc ] ) for @xmath126 satisfying ( [ eq : expnew00 ] ) , and let @xmath21 denote the finite @xmath219 blow - up time for @xmath33 as described in theorem [ thm : lambdaneg ] . 1 . [ it : plambdaneg1 ] let @xmath206
. then , @xmath518 for either @xmath67 and @xmath547 $ ] , or @xmath548 and @xmath549 .
[ it : plambdaneg2 ] let @xmath418 .
then , @xmath518 for either @xmath67 and @xmath550 $ ] , or @xmath551 and @xmath552 .
[ it : plambdaneg3 ] let @xmath401 .
then @xmath518 for @xmath551 and @xmath37 , whereas @xmath41 for @xmath543 and @xmath295 . when applicable , ( [ eq : lemmaass1 ] ) and ( [ eq : req2lambdaneg1 ] ) apply to ( [ it : plambdaneg1 ] ) and ( [ it : plambdaneg2 ] ) above .
[ def : order ] suppose a smooth function @xmath553 satisfies @xmath554 but @xmath555 is not identically zero .
we say @xmath555 has a zero of order @xmath556 at @xmath557 if @xmath558 in @xcite , we examined a class of smooth , mean - zero , periodic initial data characterized by @xmath559 having zeroes of order @xmath560 at the finite number of locations @xmath83 for @xmath79 , or at @xmath88 if @xmath67 , that is , @xmath561 or @xmath562 .
consequently , in each case , we were able to use an appropriate taylor expansion up to quadratic order to account for the local behaviour of @xmath91 near these points .
this approach , in turn , led to the results summarized in theorems [ thm : sarria1 ] and [ thm : lpintro ] of [ sec : intro ] . assuming the order @xmath563 of these particular zeroes , @xmath83 or @xmath564 , of @xmath92 is the same regardless of location , and noticing that @xmath565 must be odd due to @xmath91 being even in a small neighbourhood of these points
, we may use definition [ def : order ] to generalize the results in @xcite to a larger class of smooth , mean - zero , periodic initial data characterized by @xmath92 having zeroes of higher orders , @xmath566 , at every @xmath83 if @xmath79 , or @xmath88 for @xmath67 . since this corresponds to replacing @xmath213 in ( [ eq : expnew0 ] ) or ( [ eq : expnew00 ] ) by @xmath567
, we obtain our results simply by substituting @xmath213 in theorems [ thm : lambdapos ] , [ thm : lplambdapos ] , [ thm : lambdaneg ] and [ thm : genlplambdaneg ] , by @xmath568 in those cases where @xmath358 .
the results are summarized in corollary [ coro : gensmooth ] below .
[ coro : gensmooth ] consider the initial boundary value problem ( [ eq : nonhomo])-([eq : pbc ] ) for smooth , mean - zero , periodic initial data .
furthermore , 1 .
suppose @xmath559 has a zero of order @xmath565 at every @xmath83 , @xmath124 .
then * [ it : globalgen ] for @xmath569 , solutions exist globally in time . more particularly , these vanish as @xmath29 for @xmath570 but converge to a nontrivial steady state if @xmath571 . *
[ it : blow1gen ] for @xmath572 , there exists a finite @xmath21 such that both the maximum @xmath220 and the minimum @xmath221 diverge to @xmath47 and respectively to @xmath222 as @xmath44 .
furthermore , @xmath224 if @xmath411 and @xmath573 for all @xmath295 .
2 . suppose @xmath559 has a zero of order @xmath565 at each @xmath88 , @xmath125 .
then * [ it : lambdaneg1gen ] for @xmath574 , there exists a finite @xmath21 such that only the minimum diverges , @xmath226 as @xmath44 , whereas , for @xmath575 and @xmath37 , @xmath518 . *
[ it : lambdaneg3gen ] for @xmath576 , there is a finite @xmath21 such that both @xmath220 and @xmath221 diverge to @xmath47 and respectively to @xmath222 as @xmath44 .
additionally , @xmath524 for @xmath411 and @xmath41 if @xmath577 and @xmath295 .
[ rem : exception ] it turns out that , unless the initial data is smooth , the results established in this paper for periodic boundary conditions extend to a dirichlet setting . for smooth initial data , if there are @xmath578 when @xmath79 , or @xmath579 for @xmath67 , then the results in the periodic setting will extend to dirichlet boundary conditions as long as @xmath92 vanishes at those end - points .
this last condition prevents a lipschitz - type behaviour of @xmath91 at the boundary , which could otherwise lead to finite - time blow - up from smooth initial data under dirichlet boundary conditions .
details on this will be presented in an upcoming paper .
also , notice that letting @xmath150 in either ( [ eq : expnew0 ] ) or ( [ eq : expnew00 ] ) implies that @xmath580 near @xmath83 , or @xmath581 for @xmath582 , respectively .
then , letting @xmath583 in corollary [ coro : gensmooth]([it : globalgen ] ) implies that , for this particular class of locally constant @xmath91 , a solution that exists locally in time for any @xmath1 , will persist for all time .
examples for theorems [ thm : p=1 ] , [ thm : lambdapos ] and [ thm : lambdaneg ] are now presented . for simplicity
, we consider initial data satisfying dirichlet boundary conditions @xmath584 and we note that ( [ eq : mainsolu ] ) is equivalent to the representation formula ( see @xcite ) @xmath585 for several choices of @xmath1 , the time - dependent integrals in ( [ eq : finalsolu ] ) are evaluated and pointwise plots are generated using mathematica . whenever possible , plots in the eulerian variable @xmath586 instead of the lagrangian coordinate @xmath587 are provided . for practical reasons , details of the computations in most examples
are omitted . also , due to the difficulty in solving for the time variable @xmath455 through the ivp ( [ eq : etaivp ] ) for @xmath116 , most plots for @xmath448 are against the variable @xmath588 rather than @xmath455 .
example 1 below applies to stagnation point - form ( spf ) solutions to the incompressible 3d euler equations ( @xmath61 ) .
we consider two types of data , one satisfying ( [ eq : expnew0 ] ) for @xmath90 , and other having @xmath401 .
recall from table [ table : thecase ] that if @xmath357 , global existence in time follows , while , for @xmath418 , finite - time blow - up occurs .
below , we see that a spontaneous singularity may also form if @xmath459 .
first , for @xmath61 and @xmath590 $ ] , let @xmath591 then @xmath592 achieves its maximum @xmath161 at @xmath593 . also , @xmath459 , @xmath594 , and @xmath595 , i.e. @xmath596 ; a jump discontinuity of infinite magnitude in @xmath92 .
evaluating the integrals in ( [ eq : finalsolu ] ) , we obtain @xmath597 and @xmath598 for @xmath599 . furthermore , in the limit as @xmath600 , @xmath601 whereas @xmath602 also , ( [ eq : etaivp ] ) and ( [ eq:3d1 ] ) yield @xmath603 so that @xmath604 using ( [ eq:3d1 ] ) and ( [ eq:3d2 ] ) on ( [ eq : finalsolu ] ) , we find that @xmath448 undergoes a two - sided , everywhere blow - up as @xmath605 .
next , replace @xmath459 in ( [ eq : data3d ] ) by @xmath589 .
then , @xmath606 so that @xmath92 is now defined as @xmath607 . also ,
for this data , both integrals now diverge to @xmath47 as @xmath608 .
particularly , this causes a balancing effect amongst the terms in ( [ eq : finalsolu ] ) that was previously absent when @xmath459 .
ultimately , we find that as @xmath609 , @xmath610 for every @xmath590 $ ] . see figure [ fig : ex7 ] below . and @xmath611 .
figure @xmath612 depicts two - sided , everywhere blow - up of @xmath448 for @xmath459 as @xmath613 ( @xmath614 ) , whereas , for @xmath589 , figure @xmath615 represents its vanishing as @xmath613 ( @xmath218).,title="fig : " ] and @xmath611 .
figure @xmath612 depicts two - sided , everywhere blow - up of @xmath448 for @xmath459 as @xmath613 ( @xmath614 ) , whereas , for @xmath589 , figure @xmath615 represents its vanishing as @xmath613 ( @xmath218).,title="fig : " ] + in @xcite ( see theorem [ thm : sarria1 ] in [ sec : intro ] ) , we showed that for a class of smooth , periodic initial data @xmath616 , finite - time blow - up occurs for all @xmath247 .
example 2 below is an instance of theorem [ thm : lambdapos]([it : global ] ) . for @xmath617 ,
we consider initial data satisfying ( [ eq : expnew0 ] ) for @xmath618 , respectively , and find that solutions persist globally in time .
also , the example illustrates the two possible global behaviours : convergence of solutions , as @xmath218 , to nontrivial or trivial steady states .
first , let @xmath619 and @xmath623 then @xmath624 achieves its greatest value @xmath161 at @xmath593 and @xmath625 . since @xmath626 , theorem [ thm : lambdapos]([it : global ] ) implies global existence in time .
particularly , @xmath610 as @xmath627 see figure [ fig : ex34]@xmath628
. now , suppose @xmath621 and replace @xmath620 in ( [ eq : dat00 ] ) by @xmath622 .
then , @xmath629 attains @xmath161 at @xmath593 and @xmath630 . because @xmath631 , theorem [ thm : lambdapos]([it : global ] ) implies that @xmath33 converges to a nontrivial steady - state as @xmath218 .
see figure [ fig : ex34]@xmath632
. represents the vanishing of @xmath448 as @xmath633 ( @xmath218 ) for @xmath619 and @xmath620 , whereas , figure @xmath615 illustrates its convergence to a nontrivial steady state as @xmath634 ( @xmath218 ) if @xmath622 and @xmath631.,title="fig : " ] represents the vanishing of @xmath448 as @xmath633 ( @xmath218 ) for @xmath619 and @xmath620 , whereas , figure @xmath615 illustrates its convergence to a nontrivial steady state as @xmath634 ( @xmath218 ) if @xmath622 and @xmath631.,title="fig : " ] suppose @xmath637 and @xmath638 .
then , @xmath639 attains its greatest value @xmath640 at @xmath641 also , @xmath594 and @xmath642 . according to theorem [ thm : lambdapos]([it : blow1 ] ) , two - sided , everywhere finite - time blow - up occurs .
the estimated blow - up time is @xmath643 see figure [ fig : ex340]@xmath628 .
let @xmath644 and @xmath646 .
then @xmath91 attains its minimum @xmath647 at @xmath648 and @xmath649 since @xmath650 theorem [ thm : lambdaneg](2 ) implies one - sided , discrete finite - time blow - up and @xmath651 .
see figure [ fig : ex340]@xmath632 .
we remark that in @xcite , the same value for @xmath74 with smooth , periodic initial data , and @xmath652 led to two - sided , everywhere blow - up instead . for example 3
depicts two - sided , everywhere blow - up of @xmath448 as @xmath608 @xmath653 for @xmath637 and @xmath636 , while , figure @xmath615 for example 4 illustrates one - sided , discrete blow - up , @xmath654 , as @xmath655 ( @xmath656 ) for @xmath644 and @xmath645.,title="fig : " ] for example 3 depicts two - sided , everywhere blow - up of @xmath448 as @xmath608 @xmath653 for @xmath637 and @xmath636 , while , figure @xmath615 for example 4 illustrates one - sided , discrete blow - up , @xmath654 , as @xmath655 ( @xmath656 ) for @xmath644 and @xmath645.,title="fig : " ] for @xmath657 , let @xmath659 then @xmath660 occurs at both @xmath661 and @xmath662 now , near @xmath663 , @xmath91 behaves quadratically ( @xmath652 ) , whereas , for @xmath664 small , it behaves linearly ( @xmath93 ) .
the quadratic behaviour is due to @xmath92 having zero of order one at @xmath665 , thus , corollary [ coro : gensmooth ] implies a discrete , one - sided blow - up .
similarly in the case of linear behaviour according to theorem [ thm : p=1 ] .
after evaluating the integrals , we find that @xmath528 as @xmath666 . due to the dirichlet boundary conditions
, we have that @xmath667 and @xmath668 for a s long a s @xmath36 is defined . then , one blow - up location is given by the boundary @xmath669 , while the interior blow - up location , @xmath670 , is obtained by integrating ( [ eq : sum ] ) .
this yields the characteristics : @xmath671 setting @xmath672 and letting @xmath673 , we find that @xmath674
. see figure [ fig : weird]@xmath628 . for @xmath216 ,
let @xmath675 then , @xmath676 occurs at both endpoints @xmath677 .
also @xmath594 and since @xmath678 the local behaviour of @xmath91 near both endpoints is linear ( @xmath93 ) .
the integrals in ( [ eq : finalsolu ] ) evaluate to @xmath679 for @xmath599 and @xmath680 . using the above on ( [ eq : finalsolu ] ) , we find that @xmath681 as @xmath608 , while @xmath682 for all @xmath683 .
the blow - up time is estimated from ( [ eq : etaivp ] ) and @xmath265 above as @xmath684 . see figure [ fig : weird]@xmath632 . for example 5 with @xmath657 and @xmath658 , depicts one - sided , discrete blow - up , @xmath685 , as @xmath686 .
the blow - up locations are @xmath669 and @xmath674 . then , figure @xmath615 for example 6 with @xmath216 and @xmath93 , represents two - sided , everywhere blow - up of @xmath687 , as @xmath688.,title="fig : " ] for example 5 with @xmath657 and @xmath658 , depicts one - sided , discrete blow - up , @xmath685 , as @xmath686 .
the blow - up locations are @xmath669 and @xmath674 .
then , figure @xmath615 for example 6 with @xmath216 and @xmath93 , represents two - sided , everywhere blow - up of @xmath687 , as @xmath688.,title="fig : " ] e.w .
barnes , `` _ _ a new development of the theory of hypergeometric functions _ _ '' , proc .
london math .
( 2 ) * 6 * ( 1908 ) , 141 - 177 .
a. bressan and a. constantin , global solutions of the hunter - saxton equation , siam j. math .
anal 37(3 ) ( 2005 ) , 996 - 1026 .
f. calogero , a solvable nonlinear wave equation , stud .
appl . math .
, 70 , 3 , ( 1984 ) , 189 - 199 . s. childress , g.r .
ierley , e.a .
spiegel and w.r .
young , blow - up of unsteady two - dimensional euler and navier - stokes equations having stagnation - point form , j. fluid mech .
203 ( 1989 ) , 1 - 22 . c. h. cho and m. wunsch , global and singular solutions to the generalized proudman - johnson equation , j. diff .
249 , ( 2010 ) , 392 - 413 .
cho and m. wunsch , global weak solutions to the generalized proudman - johnson equation , commun .
pure appl .
vol 11 , no .
4 , ( 2012 ) , 1387 - 1396 .
a. constantin and m. wunsch , on the inviscid proudman - johnson equation , proc .
japan acad .
ser . a math .
, 85 , 7 , ( 2009 ) , 81 - 83 .
dafermos , generalized characteristics and the hunter - saxton equation , j. hyperbol .
, vol . 8 , no . 1 ( 2011 ) , 159 - 168 .
a. erdlyi , w. magnus , f. oberhettinger and f.g .
tricomi , `` _ _ higher transcendental functions , vol .
i _ _ '' , mcgraw - hill , ( 1981 ) , 56 - 119 .
j. escher and m. wunsch , restrictions on the geometry of the periodic vorticity equation , pre - print , arxiv:1009.1029 , 2010 .
gamelin , `` _ _ complex analysis _
_ '' , undergraduate texts in mathematics , springer ( 2000 ) , 361 - 365 .
g. gasper and m. rahman , `` _ _ basic hypergeometric series _ _ '' , encyclopedia of mathematics and its applications , vol .
96 , second ed . , cambridge university press ( 2004 ) , 113 - 119 .
j.k . hunter and r.
saxton , dynamics of director fields , siam j. appl .
51(6 ) ( 1991 ) , 1498 - 1521 . j.k .
hunter and y. zheng , on a completely integrable nonlinear hyperbolic variational equation : i. global existence of weak solutions , arch . rat .
anal . 129 ( 1995 )
305 - 353 .
b. khesin and g. misiolek , euler equations on homogeneous spaces and virasoro orbits , adv .
( 2003 ) no . 1 , 116 - 144 . j. lenells , weak geodesic flow and global solutions of the hunter - saxton equation , discrete contin .
18 ( 4 ) ( 2007 ) , 643 - 656 .
h. okamoto and j. zhu , some similarity solutions of the navier - stokes equations and related topics , taiwanese j. math .
4 ( 2000 ) , 65 - 103 . h. okamoto , well - posedness of the generalized proudman - johnson equation without viscosity , j. math .
fluid mech . 11 ( 2009 ) , 46 - 59 . i. proudman and k. johnson , boundary - layer growth near a rear stagnation point , j. fluid mech .
12 ( 1962 ) , 161 - 168 .
a. sarria and r. saxton , blow - up of solutions to the generalized inviscid proudman - johnson equation , j. math .
fluid mech .
, online first , ( 2012 ) , doi 10.1007/s00021 - 012 - 0126-x .
r. saxton and f. tiglay , global existence of some infinite energy solutions for a perfect incompressible fluid , siam j. math .
4 ( 2008 ) , 1499 - 1515 .
f. tiglay , the periodic cauchy problem of the modified hunter - saxton equation , j. evol .
eq . 5 ( 4 ) ( 2005 ) , 509 - 527
. h. weyl , on the differential equations of the simplest boundary - layer problems , ann .
43 ( 1942 ) , 381 - 407 .
m. wunsch , the generalized proudman - johnson equation revisited , j. math .
fluid mech . 13 ( 1 ) ( 2009 ) , 147 - 154 .
z. yin , on the structure of solutions to the periodic hunter - saxton equation , siam j. math .
1 , ( 2004 ) , 272 - 283 . | in @xcite , we derived representation formulae for spatially periodic solutions to the generalized , inviscid proudman - johnson equation and studied their regularity for several classes of initial data .
the purpose of this paper is to extend these results to larger classes of functions including those having arbitrary local curvature near particular points in the domain . |
a significant fraction of type 1 active galactic nuclei ( agn ) show ` soft x - ray excess emission ' above a power - law continuum , usually identified as the steepening of the x - ray continuum below @xmath11 .
this soft excess emission was first observed by _ heao-1 _ ( pravdo et al .
1981 ) , and _ exosat _ ( arnaud et al . 1985 ; singh et al . 1985 ) .
the number of agn with soft x - ray excess emission or ` ultra soft agn ' increased dramatically with the launch of _
rosat_. boller , brandt , & fink ( 1996 ) showed that agn with steepest soft x - ray spectra in the _ rosat _ band tend to lie at the lower end of the h@xmath12 line width distribution .
these agn with fwhm@xmath13 are classified as the narrow - line seyfert 1 ( nls1 ) galaxies ( osterbrock & pogge 1985 ) , and are distinguished from the bulk of the seyfert 1 galaxies ( ` broad - line seyfert 1s ' or bls1s ) .
nls1s also have strong optical fe emission .
detailed x - ray study of nls1s was carried out with _ asca _ which confirmed the soft x - ray emission and , in addition , showed that the @xmath14 continuum slope too is steeper and anti - correlated with the fwhm of the h@xmath12 line ( brandt , mathur , & elvis 1997 ; leighly 1999b ; vaughan et al.1999b ) .
_ asca _ observations also showed that many nls1s have fe k@xmath6 line arising from highly ionized iron , and hence the accretion disks of nls1s must be ionized ( dewangan 2002 , and references therein ; see ballantyne , iwasawa , & fabian 2001 for ionized reflection model for nls1s ) . during the _ rosat _ and _ asca _ era
, nls1s drew attention of most workers due to the extreme variability of their strong soft x - ray emission .
nls1s are the most extreme x - ray variable objects among the radio - quiet agn .
nls1s frequently exhibit rapid ( doubling time scales of a few hundred seconds ) and/or large ( up to a factor of 100 ) amplitude x - ray variability ( e.g. , boller , brandt , & fink 1996 ; dewangan et al .
2001,2002 ) .
the excess variance for nls1s is typically an order of magnitude higher than that observed for bls1s .
( leighly 1999a ; turner et al 1999 ) .
the strong soft x - ray emission of nls1s is usually attributed to the intrinsic emission from viscous accretion disks ( pounds et al .
however , the variability of the soft x - ray emission has been found to be faster than the characteristic time scales for an accretion disk ( boller et al .
the fastest time scales for an accretion disk are the thermal time scale , @xmath15 , on which thermal instabilities grow , and the sonic time scale , @xmath16 , on which the sound waves cross the disk . in the above
, @xmath6 is the viscosity parameter ( shakura & synyaev 1973 ) , @xmath17 is the angular velocity and @xmath18 is the height scale of the disk .
these time scales are about a factor of two larger than the observed variability time scale for the nls1 galaxy iras 13224 - 3809 ( boller et al .
the same is true for many nls1 galaxies .
this apparently argues against the accretion disk origin for the strong soft x - ray emission and the nature of the soft x - ray excess emission remains uncertain .
_ chandra _ and _ xmm - newton _ grating spectroscopy of nls1s have revealed that the soft x - ray excess emission is a smooth continuum component rather than a blend of emission / absorption features ( turner et al .
2001 ; collinge et al 2001 ; gondoin et al .
2002 ; page et al .
it is not clear whether the rapid variability of the smooth soft x - ray emission is due to the variability of the soft excess emission above and over the power - law or it is the underlying power - law component which is varying rapidly .
based on a 10 day _ asca _ observation of iras 13224 - 3809 , dewangan et al .
( 2002 ) suggested that the rapid variability in the soft x - ray band is likely to be due to the rapid variability of the power - law emission and there may not be intrinsically rapid variability in the soft excess emission above the power - law .
this observation , if confirmed further , will have profound effects on constraining models for x - ray emission of nls1s .
however , this requires time resolved x - ray spectroscopy with short time scales which was not possible with the earlier x - ray observatories .
_ xmm - newton _ having the largest collecting area and broad bandpass is the most suitable for the above purpose . in this paper
, we present results based on an _ xmm - newton _ observation of i zw 1 , including time resolved spectroscopy on time scales of @xmath19 , comparable to the dynamical time scale for a keplerian accretion disk @xmath20 .
i zw 1 is the prototype nls1 galaxy .
its optical spectrum shows exceptionally narrow balmer emission lines ( fwhm@xmath21 ; boroson & green 1992 ) .
i zw 1 is well known for its strongest optical and uv fe emission ( boroson & green 1992 ; laor et al .
its high absolute luminosity ( m@xmath22 for h@xmath23 , @xmath24 ) makes it the nearest quasar ( @xmath25 ) and is also called as the ` narrow - line quasar ' a high luminosity version of the nls1 galaxies .
i zw 1 was observed by _ xmm - newton _ observatory ( jansen et al . 2001 ) on 22 june 2002 using the european photon imaging camera ( epic ) , the reflection grating spectrometer ( rgs ) , and the optical monitor ( om ) instruments .
the epic mos @xcite , and pn @xcite cameras were operated in small , and large window modes , respectively , to alleviate pileup of photons from the source .
the epic cameras used the medium optical blocking filter .
the raw events were processed and filtered using the most recent updated calibration database and analysis software ( sas v5.4.1 ) available in october 2003 .
events in the bad pixels and those adjacent pixels were discarded .
only events with pattern @xmath26 ( single and double ) for the pn and @xmath27 for the mos were selected .
examination of pn light curve extracted from the full field but above @xmath28 showed flaring particle background in two observation intervals of about @xmath29 .
these intervals were excluded from the rest of the analysis .
this resulted in ` good ' exposure time of @xmath30 for the pn and @xmath31 for the mos cameras .
source photons were extracted from the pn and mos data using a circular region of radius @xmath32 centered at the source position .
associated background photons were extracted from source free regions near the source .
the total number of net source photons is @xmath33 detected with the pn and @xmath34 detected with each of the mos cameras .
x - ray light curves of i zw 1 were extracted from the pn and mos data using a circular region of @xmath32 centered at the source position and in the soft ( @xmath35 ) , hard ( @xmath14 ) , and full ( @xmath36 ) bands .
the soft and hard bands were chosen to separate approximately the two spectral components a power - law and a soft excess component generally observed from nls1 galaxies ( see e.g. , leighly 1999b ) .
the time bins with sizes of @xmath37 are at least @xmath38 exposed in all the light curves .
background light curves were extracted from source free regions with the same bin sizes , exposure requirements , and in the same energy bands as for the source light curves .
the background light curves were subtracted from the respective source light curves after appropriate scaling to compensate for the different areas of the extraction regions .
the three light curves extracted from the pn and mos data , were combined to improve the signal - to - noise ratio .
the observed counts correspond to a mean observed flux of @xmath39 , and luminosity of @xmath40 in the @xmath36 band .
figure [ f1 ] shows the combined pn and mos light curves of i zw 1 in the full , soft , and hard bands . also shown in fig [ f1 ]
is the hardness ratio defined as the ratio of count rates in the @xmath14 and @xmath35 bands .
it is evident that x - ray emission from i zw 1 varied strongly during the _ xmm - newton _ observation .
the light curves show trough - to - peak variations by a factor of @xmath41 in the full band , @xmath42 in the soft band , and @xmath43 in the hard band .
the most remarkable variable event is a large x - ray flare which started at @xmath44 after the beginning of the observation and lasted for about @xmath45 . during this flare
the full band count rate increased from @xmath46 to @xmath47 in just @xmath48 .
the count rates quoted above were calculated as the mean of at least three data points .
the soft and hard band light curves show similar variability pattern and appear to be correlated . to quantify this correlation and to search for any possible time lag , we have calculated cross - correlation function ( ccf ) of the hard band light curve with respect to the soft band light curve . for this purpose we utilized pn and mos light curves with time bins of @xmath49 created in a similar way as described above .
the ccf was calculated using the ftools program crosscor .
the ccf is plotted in figure [ f3 ] with time bins of @xmath50 , which shows strong correlation between the soft and hard band light curves without any significant delay .
we also examined ccfs with smaller time bins and found no time lag between the two light curves . despite the strong correlation between the soft and hard band light curves
, the x - ray flare has different amplitude of variability in the soft and hard bands .
the flare is more intense in the hard band ( an increase in count rate by a factor of @xmath51 ) than in the soft band where the count rate increased by a factor of @xmath52 .
this is reflected in an increased hardness ratio during the flare ( see fig .
it appears that the hardness ratio generally follows the variations in the count rate in any band .
this is obvious in figure [ f2 ] where the hardness ratio is plotted against the count rate in the full band .
figure [ f2 ] shows that the x - ray emission from i zw 1 hardens with x - ray intensity .
the trend in the hardness ratio with count rate tells us that the total counts in the @xmath14 band increases more than that in the @xmath35 and nothing about the @xmath14 or @xmath35 spectral change .
photon energy spectra of i zw 1 and associated background spectra were accumulated from the epic pn and mos data using the same source and background extraction region described above .
the pulse invariant channels were grouped such that each spectral bin contained at least 80 and 30 counts in the case of pn and mos spectra , respectively .
the epic responses were generated using the sas tasks rmfgen and arfgen .
the first order rgs1 and rgs2 spectra of i zw 1 were extracted using source regions in dispersion / spatial coordinates that contained @xmath53 of the point source events .
the corresponding background spectra were generated using the entire regions of ccd arrays but excluding @xmath54 of the point spread function along the spatial direction .
rgs response matrices were calculated using the sas task rgsrmfgen .
both the rgs spectra were binned to 5 channels , a further grouping ( a minimum of 5 counts per spectral channel ) was applied to avoid zero counts in any channel .
all the spectral fits were performed with the xspec package ( version 11.2 ; arnaud 1996 ) and using the @xmath55-statistic for the epic spectra and c - statistic for the rgs spectra .
unless otherwise specified , the quoted errors on the best - fit model parameters were calculated for the @xmath53 confidence level for one interesting parameter i.e. , @xmath56 or @xmath57 . at first the mos and pn spectra
were fitted separately to check for the possible uncertainties due to cross - calibration problems .
we found generally good agreement between mos2 and pn cameras ( @xmath58 ) in the @xmath59 bands , however , mos1 results in a flatter ( @xmath60 ) spectrum .
these discrepancies are consistent with that found by molendi & sembay ( 2003 ) . in the @xmath61 band , mos1 spectrum is flatter ( @xmath62 ) than pn spectrum , while mos2 data agrees well with the pn data ( @xmath63 ) . therefore we present the spectral results obtained by fitting the same model jointly to the pn and mos2 data while allowing the relative normalizations to vary . however , pn and mos data do not agree below @xmath64 .
similar discrepancies at low energies have been found for pg 1211 + 143 ( pounds et al .
2003 ) and akn 564 ( vignali et al .
therefore , we did not use epic data below @xmath65 for out spectral analysis . in order to characterize the continuum shape of i zw 1 ,
first we fitted a redshifted and absorbed power - law model jointly to the pn and mos2 data . for this fit we chose the energy bands @xmath66 and @xmath67 bands in the observer s frame .
these bands are relatively free of any spectral features in the agn spectra .
this model resulted in a good fit ( @xmath68 for 292 degrees of freedom ( dof ) ) .
the best - fit power - law photon index is very steep ( @xmath69 ) .
a comparison of the observed data with the extrapolated best - fit power - law to lower energies is shown in figure [ f4 ] in terms of deviations @xmath70 , where @xmath71 is the observed counts in energy channel @xmath72 , @xmath73 is the standard error on @xmath71 , and @xmath74 is the best - fit model counts in channel @xmath72 .
[ f4 ] shows a weak soft x - ray excess emission above the power - law at energies below @xmath11 , a weak line - like emission at @xmath75 , and an edge - like feature at @xmath76 .
the soft x - ray excess emission is a general feature of nls1 galaxies . to model this soft component we added a blackbody component to the power - law model and carried out the fitting in the @xmath10 band .
this model provided a good description of the spectra ( @xmath77 for @xmath78 dof ) .
the best - fit absorption column density is @xmath79 , which is significantly higher than the galactic column density of @xmath80 ( dickey & lockman 1990 ) along the line of sight to i zw 1 .
this suggests an additional absorber local to the source .
therefore , we fixed the column density to the galactic value and added an additional photoelectric absorption model at the source redshift to our two component model .
this model provided a good fit ( @xmath81 for @xmath78 dof ) resulted in an excess absorption of @xmath82 .
as also noted earlier , there are signatures of possible presence of an iron k@xmath6 line and k - edge .
addition of a redshifted absorption edge to the model consisting of a blackbody and a power - law modified by galactic and intrinsic absorption improves the fit ( @xmath83 for 2 additional parameters ) . adding a gaussian line
further improves the fit ( @xmath84 for @xmath85 dof ) . the iron k - edge and the iron k@xmath6 line
are detected at significance levels of @xmath86 and @xmath87 , respectively , according to an f - test .
figure [ f5 ] shows the observed pn and mos2 data , the best - fit model , and the deviations @xmath88 ( see above ) of the data from the best - fit model . due to its excellent spectral resolution
, the rgs instrument allows a more detailed examination of the soft x - ray spectrum .
the rgs spectra of i zw 1 are shown in figure [ f6 ] .
the soft x - ray spectrum of i zw 1 does not show strong emission or absorption features .
it is a smooth continuum except for a few weak features .
therefore , we fitted an absorbed power - law model jointly to the rgs1 and rgs2 spectra in the @xmath89 band .
the relative normalizations were allowed to vary .
the model resulted in c - statistic @xmath90 for 805 dof with significant intrinsic ( excess over the galactic column ) absorption of @xmath91 and a steep power law with @xmath92 .
the excess absorption is similar to that derived from the pn and mos data .
however , the soft x - ray power - law appears to be slightly steeper ( @xmath93 ) than the hard x - ray ( @xmath59 ) power - law derived from the pn and mos data . the best - fit model and ratio of the observed data and model
are shown in fig .
there are no strong sharp features .
the only obvious features are in the @xmath94 band , probably associated with the detector response of the oxygen k - edge , and emission line - like feature at a rest - frame energy of @xmath95 . adding a narrow gaussian line at @xmath96 results in @xmath97 , and an equivalent width of @xmath98 .
the rest - frame line centroid energy is not well constrained but appears to be higher than the centroid energy of si k@xmath6 line ( @xmath99 for neutral si ) .
the blackbody and power - law model described the data adequately , but the temperature of the blackbody ( @xmath100 ) is unrealistically high for a standard accretion disk ( see section 8.4 ) .
appropriate accretion disk models in place of a simple blackbody result in similar temperatures .
the multicolor disk blackbody model diskbb provided equally good fit ( @xmath101 for 780 dof ) and resulted in @xmath102 , while the more realistic disk model diskpn ( gierli ' nski et al .
1999 ) provided equally good fit and resulted in @xmath103 .
the presence of an iron k@xmath6 line at @xmath104 and an iron k - edge at @xmath105 both suggest that these features arise from highly ionized iron ( fe fe ) .
the strength ( ew @xmath106 ) and width of the iron line ( fwhm @xmath107 ) suggests that the line is produced in the accretion disk .
these observations are supportive of compton reflection emission from a highly ionized accretion disk .
the reflection emission from ionized material can be strong below @xmath11 and contribute significantly to the soft x - ray excess emission . with this in mind
we adopt an x - ray continuum model that includes a power - law and compton reflection from ionized accretion disk .
we used the compton reflection model _
pexriv _ of magdziarz & zdziarski ( 1995 ) which calculates compton reflection of illuminated x - ray power - law continuum by an ionized accretion disk .
the pexriv model does not calculate the iron k@xmath6 line , therefore we also included a gaussian line for the iron k@xmath6 line .
as before , we used the photoelectric absorption due to galactic and intrinsic columns .
this model resulted in a good fit ( @xmath108 for 781 dof ) .
the best - fit parameters were @xmath109 , @xmath110 ( where @xmath111 corresponds to a solid angle of @xmath112 subtended by the reflector ) , @xmath113 .
slightly higher value of @xmath114 may suggest an additional spectral component for the soft x - ray excess emission .
addition of a blackbody component further improves the fit . the model consisting of a blackbody , power - law , reflection and a gaussian , and modified by galactic as well as intrinsic absorption ( model 2 ) improves the fit ( @xmath115 for 779 dof ) .
the best - fit parameters were @xmath116 , @xmath117 , @xmath118 , and parameters of the gaussian line were @xmath119 , @xmath120 , and @xmath121 ( see table [ tab1 ] ) .
the best - fit model and the unfolded spectrum are shown in figure [ f6 ] .
due to the contribution of the ionized reflection in the soft x - ray band , the strength of the blackbody component is weaker in model 2 compared to that derived in model 1 .
inverse compton scattering of soft photons by hot electrons , possibly located in a corona above the disk , can produce approximate power - law continuum .
the primary continuum emission of agn is a comptonized spectrum not an exact power law .
the departure at low energies of a thermally comptonized spectrum from a power - law continuum may have profound effects on the shape of the soft x - ray excess emission of nls1 galaxies . in model 2 ,
the soft x - ray excess component is modeled by ionized reflection and a weak blackbody component . since the thermal comptonized spectrum is an approximate power - law , we further investigate the significance of the weak blackbody component after replacing the power - law component in model 2 by a comptonization model .
we used the thermal comptonization model comptt ( titarchuk 1994 ) .
the free parameters of the model are the temperature of the seed photons ( @xmath122 ) , assumed to follow a wien law , the optical depth of the scattering region ( @xmath123 ) , the electron plasma temperature ( @xmath124 ) , and a parameter defining the geometry ( a disk or a sphere ) . in the following , we consider the disk geometry .
we fixed the parameters of the gaussian line and the illuminating spectrum in the pexriv model to be the same as that of model 2 but varied the reflection fraction @xmath114 .
this model provided a good fit ( @xmath125 ) .
the best - fit blackbody temperature is now @xmath126 .
the best - fit relative reflection is @xmath127 .
the best - fit parameters of the comptt component are seed photon temperature @xmath128 , electron temperature @xmath129 , and @xmath130 .
a corona with two electron populations with distinct temperatures can give rise to both a soft x - ray excess and a power - law .
it is likely that the surface of an accretion disk is ionized , the outermost layers may contain thermal electrons in addition to the hotter electron population above the disk in the corona . under such conditions
, the cool disk photons will be first comptonized by the cooler of the electron population at the surface of the disk , giving rise to the soft excess emission below @xmath11 .
this soft excess emission may be the seed photons for the comptonization by the hotter electrons in the corona ( page et al .
we investigated if such a process is a viable model for the x - ray spectrum of i zw 1 .
first , we tried to fit the power - law component with the comptt model . a combination of a blackbody , comptt , a gaussian line , and an edge model provided a good fit ( @xmath131 for 779 dof ) .
no intrinsic absorption is required in this model .
thus the power - law component is well described by comptonization of a soft thermal spectrum with temperature @xmath132 by hot electrons with @xmath133 and the soft x - ray excess component is described by a blackbody component with @xmath134 .
the contribution of this soft excess component to the @xmath5 band unabsorbed emission is @xmath135 . as the temperature of the soft excess component is much higher than that expected from an accretion disk .
this component is unlikely to be the intrinsic disk emission . replacing the blackbody compoent by an additional comptonization component the fit slightly ( @xmath136 for 777 dof ) .
the best - fit parameters are comptt 1 : soft photon @xmath137 , electron temperature @xmath138 and @xmath139 ; comptt 2 : soft photon @xmath140 , electron temperature @xmath141 and @xmath142 . in this model ,
soft thermal photon from the disk with @xmath143 are comptonized by hot thermal ( @xmath138 ) electrons in the corona .
this process gives rise to the soft x - ray excess emission .
the second comptonization model produces the power - law emission .
to examine the spectral variations of i zw 1 , we extracted 4 time - selected spectra across the @xmath144 _ xmm - newton _ observation .
we chose the sampling time scale of @xmath45 , which separates the flaring interval nicely .
thus we extracted four spectra spectrum 1 ( @xmath145 ) , spectrum 2(@xmath146 ) , spectrum 3 ( @xmath147 ) , and spectrum 4 ( @xmath148 ) .
the resulting `` on - source '' average exposure time was @xmath149 per epic spectrum .
we set the ancillary response , and response matrix files to be those of the mean spectrum .
since there is no variation in the background level across the observation , we used the same background spectra extracted across the full observation for the mean spectra .
again , the energy channels were appropriately grouped to achieve a good signal - to - noise ( a minimum of 30 and 20 counts per energy channel for the pn and mos , respectively ) .
as before , we carried out the spectral fitting of the time - selected pn and mos2 data jointly . to illustrate the flux and spectral variability of i zw 1 , we compared each of the time - selected spectra with the mean spectrum .
figure [ f9 ] shows deviations of the time - selected spectra from the best - fit model 1 for the mean spectrum .
these plots were constructed by using the best - fit model 1 for the mean spectrum but without actually fitting the time - selected spectra .
strong variation is present only for the flux . to quantify these variations we performed spectral fitting of the time selected spectra .
we used the model 1 ( a combination of a blackbody + power law + gaussian line multiplied by galactic , intrinsic absorption and an iron k - edge models ) and fitted the 4 time - selected epic spectra .
it was not possible to constrain both the gaussian line and iron k - edge for individual spectra .
therefore , we fixed the line energy and width of the gaussian line and the parameters of the edge model to be the same as the best - fit values derived for the mean spectrum .
model 1 provided good fit to all the the four time - selected spectra of i zw 1 .
the best - fit parameters are listed in table [ tab2 ] .
the results of the time - resolved spectroscopy are plotted in figure [ f10 ] , which shows that all the parameters except for the power - law normalization changed significantly across the @xmath144 _ xmm - newton _ observation .
the decrease in the ratio of the flux of the blackbody and power - law components in the @xmath5 band during the large flare appears to be due to increase in the power - law flux alone .
to investigate further , we fit the model 1 to the time - selected pn spectra jointly .
we excluded the mos2 data from this fit to avoid large number of model parameters .
as before , the width and the centroid energy of the iron k@xmath6 line , and the parameters of the edge model were fixed to the best - fit values derived from the mean spectrum . in this fit , except for the normalization of the power - law component , all the parameters for a data group were tied to the corresponding parameters in other data groups .
the four normalizations of the power - law components for the four time - selected were varied independently .
this type of parameterization allowed us to check if the spectral variability of i zw 1 is solely due to variations in the power - law normalization .
the joint fit to the time - selected spectra resulted in a good fit ( @xmath150 for @xmath151 dof ) .
varying the photon indices independently improves the fit ( @xmath152 for 3 less dof ) .
there is a marginal evidence for flattening of the power - law during the large flare .
the best - fit photon indices are @xmath153 , @xmath154 , @xmath155 , and @xmath156 for spectrum 1 , 2 , 3 , and 4 , respectively .
varying blackbody temperatures independently does not improve the fit ( @xmath157 for @xmath158 dof ) .
the same is true for the blackbody normalization ( @xmath159 for 1897 dof ) .
the soft and hard band x - ray emission are correlated without any significant delay . based on our time - resolved spectroscopy in the preceding section
, this correlation appears to be due to the variability of the power - law component alone without any significant changes in the soft x - ray excess component . in order to further explore and to improve the dependence of the spectral parameters
, we have carried spectral analysis at low and high flux levels .
we extracted two averaged spectra covering the large flare between the elapsed time interval of @xmath160 ( high flux or flare state ) and excluding the large flare ( low flux or quiescent state ) .
the mean pn count rates are @xmath161 ( high flux state ) and @xmath162 ( low flux state ) in the @xmath10 band .
we analyzed both the pn spectra jointly .
we used the model 1 ( absorbed and redshifted blackbody and power - law modified by an absorption edge at @xmath105 and iron k@xmath6 line at @xmath75 .
we fixed the parameters of the iron line and edge to be the same as that derived for the mean spectrum . in order to investigate which model parameter or parameters differ most significantly between the two flux states , we tied all the model parameters for the high flux state to that of low flux state and untie the parameters step - by - step . the best - fit model with the same model parameters for the two flux states is not a good fit ( @xmath163 for @xmath164 dof ) .
varying the power - law normalizations independently resulted in an acceptable fit ( @xmath165 for 1280 dof ) . untying the photon indices further improved the fit ( @xmath166 for @xmath167 dof ) , with significantly different best - fit photon indices of @xmath168 ( high flux state ) and @xmath169 ( low flux state ) .
we also varied the blackbody normalization independently , the fit did not improve further ( @xmath170 for @xmath171 dof ) .
thus we conclude that the blackbody normalization did not vary between the two flux states and tied the parameter again in the following analysis .
we calculated the confidence contours for the photon indices and the power - law normalizations in the low and high flux states .
figure [ f11 ] shows these contours at @xmath172 , @xmath53 , and @xmath173 confidence levels .
the brightening and flattening of the power - law during the x - ray flare suggests that the flare could have resulted due to an additional power - law component . to test such a scenario ,
first we tied the photon index and normalization for the two flux states and introduced a second power - law component for the high flux state .
this model resulted in a good fit ( @xmath174 for 1279 dof ) .
the best - fit photon index and the normalization of the second power - law are @xmath175 and @xmath176 , respectively .
the first power - law is steeper and stronger ( @xmath177 ; @xmath178 ) .
before we discuss the results below , we estimate the mass of the black hole for i zw 1 .
calculations of the black hole mass rely on the assumption that the dynamics of the optical broad - line region gas is dominated by the gravitational potential of the central super - massive black hole . to calculate the black hole mass for i zw 1
, we used the results of kaspi et al .
( 2000 ) , based on a reverberation study of 17 quasars .
kaspi et al .
( 2000 ) have determined an empirical relationship between the size of the broad - line region ( @xmath179 ) and the monochromatic luminosity ( @xmath180 ) as follows .
@xmath181^{0.700 } { \rm~lt - days}\ ] ] we use the flux taken from a linear interpolation between photometric data points obtained from the _ third reference catalog of bright galaxies _ ( rc3 ; de vaucouleurs et al .
1991 ) at the rest wavelength of @xmath182 ( @xmath183 ) .
the mass of the black hole is given by @xmath184 . to determine @xmath185 , the velocity
, we correct @xmath186 of the h@xmath12 emission line by a factor of @xmath187 to account for velocities in three dimensions we used @xmath188 for the h@xmath12 line from boroson & green ( 1992 ) .
the mass is then @xmath189 using the cosmology @xmath190 and @xmath191 km s@xmath192mpc@xmath192 , this gives us a value for the central black hole mass of @xmath193 .
this paper presents an analysis of @xmath194 _ xmm - newton _ observation of the prototype narrow - line seyfert 1 galaxy i zw 1 .
the time - averaged x - ray spectrum of i zw 1 shows four spectral components a power - law continuum , a soft x - ray excess component below @xmath11 , weak intrinsic absorption ( @xmath195 ) , and spectral features due to iron ( the k@xmath6 line and k - edge ) . above @xmath196 , the power - law component has a slope of @xmath197 which is steeper than that of most bls1 galaxies ( @xmath198 ) but is similar to other nls1 galaxies e.g , ton s180 ( @xmath199 ; vaughan et al .
2002 ) , mrk 335 ( @xmath200 ; gondoin et al.2002 ) ; akn 564 ( @xmath201 ; vignali et al .
the x - ray continuum slightly steepens at lower energies , resembling to the soft x - ray excess emission generally observed from nls1 galaxies .
the strength of this soft x - ray emission is weaker in i zw 1 than that in many nls1s .
the contribution of the soft x - ray excess emission to the @xmath5 band x - ray emission is only @xmath202 in i zw 1 .
_ asca _ observation of i zw 1 in 1995 also showed intrinsic absoption ( @xmath203 ) , steep power - law continuum ( @xmath204 ) , and an iron k@xmath6 line at @xmath205 but did not detect any soft excess component ( leighly 1999 ) .
i zw 1 was fainter by a factor of @xmath206 during the _ asca _ observation .
true to the nature of nls1 galaxies , i zw 1 showed rapid and large amplitude x - ray variability during the _ xmm - newton _ observation .
x - ray emission from i zw 1 increased by @xmath207 in @xmath48 only , corresponding to a change in the @xmath10 luminosity of @xmath208 .
this is a remarkable variability for a radio - quiet quasar as the time scale is a factor of @xmath209 shorter than the dynamical time scale for a @xmath210 schwarzschild black hole .
such variability is rare among radio - quiet quasars .
other radio - quiet quasars to show such rapid and large amplitude variability are phl 1092 ( brandt et al .
1999 ) , rx j1334.9 + 3759 ( dewangan et al .
2002 ) .
the major consequence of rapid and large amplitude variability is that the source has to be emit very efficiently .
it is possible to calculate the efficiency of conversion of rest mass into energy using simple arguments originally due to fabian ( 1979 ) .
if the x - ray emitting region is an accreting spherical cloud of ionized hydrogen , then there is a limit on the radiative efficiency , @xmath211 , given by @xmath212 , where @xmath213 is the change in luminosity in the rest - frame time interval @xmath214 .
first , we convert the change in the combined pn@xmath215mos count rate into change in the luminosity by using a constant conversion factor of @xmath216 , derived from the time averaged pn@xmath215mos count rate and the intrinsic luminosity .
the change in the count rate of @xmath217 over @xmath218 ( @xmath219 in the rest frame ) corresponds to a rate of change of luminosity , @xmath220 .
this gives a lower limit to the efficiency , @xmath221 .
this efficiency can be compared with the maximum efficiency possible for a schwarzschild black hole ( @xmath222 ; shapiro & teukolsky 1983 ) and a kerr black hole ( @xmath223 ; thorne 1974 ) .
thus the rapid flare observed from i zw 1 is consistent with that expected for a schwarzschild black hole .
our time resolved spectroscopy has revealed that the soft x - ray excess emission , presumably arising from an accretion disk , did not vary appreciably during the giant flare seen in fig .
the rapid and large amplitude variability is consistent with a flaring of the power - law component alone .
the primary power - law emission of radio - quiet agn is thought to be the result of comptonization of soft photons from the disk by hot , relativistic electrons in a region , the geometry of which is not clear at present .
the possibilities are ( @xmath72 ) an extended corona above a cold accretion disk ( e.g , haardt & maraschi 1991 ) , ( @xmath224 ) non - thermal electrons at the base of a jet , and ( @xmath225 ) thermal electrons in the innermost regions of an accretion disk . in the first case
, rapid and large amplitude variability can be produced by magnetic flares arising from the surface of the disk . in the second case , ejection of a relativistic plasma blob as a jet would produce the giant flare with power - law spectral shape . in the third scenario , it is difficult to produce giant and rapid flares .
it is difficult to distinguish between the first two possibilities in the absence of wide band x - ray spectrum up to several hundred kev .
however , a kerr black hole is not required in any of the two possible scenarios .
magnetic flares can be produced on very short time scales as the reconnection time scale can be very short ( merloni & fabian 2001 ) .
there is a clear evidence for a change in the hardness ratio with total count rate ( see fig . [ f2 ] ) .
the full band x - ray spectrum of i zw 1 hardens with increasing flux .
a detailed time - selected spectral analysis revealed that hardening of the full band spectrum as suggested by the hardness ratio is due to the increase in the power - law normalization and a steady soft excess component .
our flux - selected spectroscopy has revealed that the power - law component became stronger and slightly flatter in the flaring or high flux state .
thus the change in the hardness ratio is not only due to the variations in the @xmath14 band flux but also due to hardening of the primary power - law .
the spectral hardening of i zw 1 with increasing flux is in sharp contrast with that observed from many seyfert galaxies , which generally show spectral steepening with increasing flux ( turner 1987 ; singh et al . 1991 ; turner et al .
1990 ; mchardy , papadakis & uttley 1998 ; lee et al .
2000 ; chiang et al .
2000 ; done et al .
2000 ; zdziarski & grandi 2001 ; petrucci et al .
2001 ; shih , iwasawa & fabian 2002 ; vaughan and edelson 2001 ; nandra 2001 ; romano et al .
2002 ; dewangan et al .
2002 ; gallo et al . 2003 ; lamer et al .
georgantopoulos & papadakis ( 2001 ) found similar behavior in a number of seyfert 2 galaxies .
to explain such spectral variability , mchardy et al.(1998 ) and shih et al .
( 2002 ) proposed a two component model , consisting of a soft component with constant spectral shape but variable flux and a hard component with constant flux , so that the soft component dominates the spectrum at high fluxes and overall spectrum becomes steeper . using a long _ xmm - newton _ observation , fabian & vaughan ( 2003 ) showed that the spectral variability of mcg-6 - 30 - 15 , on time scales of @xmath226 , is characterized by a soft varying component which is a power - law with constant slope and a hard constant component produced by very strong reflection from a relativistic disk .
in contrast , in the case of i zw 1 , it is the soft component with constant flux and the hard component with variable flux resulting in spectral hardening .
the observed @xmath14 spectral steepening with flux has been attributed to the change in the shape of the comptonized spectrum possible due to compton cooling by increased seed or soft photon flux from the disk ( haardt , maraschi , & ghisellini 1997 ) or due to increasing covering fraction of the corona resulting from correlated magnetic flare or reconnection events ( thunder - cloud model of merloni & fabian 2001 ) . obviously both the scenarios are not applicable to the spectral behavior of i zw 1 .
the soft excess emission did not increase with flux during the flare , implying that there was no change in either the seed photons from the disk or reprocessed emission by the disk .
the soft excess emission attributed to the intrinsic disk emission and reprocessed emission simply did not appear to respond the large x - ray flare .
this is possible if the flare emission was beamed away from the accretion disk .
the x - ray flare and accompanied spectral hardening of i zw 1 is suggestive of a generation of a new population of hot or relativistic electrons without an increase in the covering fraction of the corona
. this new population must be short - lived and beamed away from the disk .
this is possible if the new population is ejected from the base of a jet and may be part of a discrete jet . the x - ray flaring
then could be due to synchrotron self - compton ( ssc ) process as in the case of radio - loud agn .
it is interesting to note the similarities of the differences between the x - ray spectra of radio - loud and radio - quiet agn and that of i zw 1 in the low flux and high flux states .
the high resolution soft x - ray spectrum of i zw 1 , obtained with the rgs , is featureless and smooth .
we did not detect any emission or absorption features below @xmath227 , the region of the spectrum usually affected by the warm absorbers .
the rgs spectrum of i zw 1 supports the earlier observations with _
letg and _ xmm - newton _ rgs that the soft x - ray excess emission of nls1 galaxies is featureless and smooth ( turner et al .
2001 ; collinge et al 2001 ; gondoin et al . 2002 ; page et al .
although the soft x - ray excess emission above a power - law continuum is well described by a blackbody component , its temperature is too high for the intrinsic emission from an accretion disk .
the temperature of a standard thin accretion disk is @xmath228 where @xmath229 is the schwarzschild radius , and @xmath230 is the black hole mass in units of @xmath231 ( peterson 1997 ) .
the temperature at the last stable orbit ( @xmath232 ) for a schwarzschild black hole is @xmath233 for a black hole mass of @xmath234 accreting at the eddington rate .
this is an upper limit to the disk temperature , because @xmath235 , and outer regions of the disk will have lower temperatures .
a more realistic model involving ionized reflection in addition to the blackbody and power - law components describes the @xmath0 spectrum well .
the reflection component fits the iron edge and contributes significantly to the soft excess emission thereby reducing the strength of the blackbody component .
however , the blackbody temperature is still too high for an accretion disk .
the primary x - ray continuum of seyfert type agn is thought to arise from thermal comptonization of soft photons from the disk by the hot electrons in the corona ( haardt & maraschi 1991 ) .
such a comptonized spectrum deviates from the power - law form at low energies , in addition to the high energy cut - off , due to the maxwellian distribution of hot electrons ( see e.g. , titarchuk 1994 ) .
therefore , power - law is not an appropriate continuum to characterize the soft x - ray excess component . in the case of i
zw 1 , the soft x - ray excess emission above a ` thermally comptonized continuum ' is well described by ionized reflection and blackbody .
most importantly the temperature of the blackbody ( @xmath236 ) is consistent with that expected for the inner regions of an accretion disk around a @xmath237 black hole .
thus the soft x - ray emission of the proto - type nls1 galaxy i zw 1 is not unusual in terms of strength , temperature or variability time scale .
we detected an iron k@xmath6 line at @xmath75 and an iron k - edge at @xmath238 .
both the features arise from he - like iron and suggest that the accretion disk of i zw 1 is ionized .
evidence for ionized accretion disks have been found in many nls1 galaxies e.g. , ton s180 ( vaughan et al . 2002 ; comastri et al . 1998 ) mrk 335 ( gondoin et al . 2002 ) , akn 564 ( vaughan et al . 1999 ) and are consistent with the general picture that accretion disks of agn with steep x - ray spectrum are ionized ( dewangan 2002 ) .
the ionization stage of photo - ionized material is determined by the ionization parameter , @xmath239 , where @xmath240 is the luminosity of ionizing photons , @xmath241 is the density of the material , and @xmath114 is the distance of the material from the ionizing source ( kallman & mccray 1982 ) . for standard thin disks , matt et al .
( 1993 ) showed that the ionization parameter strongly depends on the accretion rate relative to the eddington rate , @xmath242 . in order to ionize he - like iron ,
@xmath243 should be at least @xmath244 or @xmath245 ( matt et al .
this result is consistent with the best - fit value of the ionization parameter derived from ionized reflection model .
these observations have important implications to the nature of the soft x - ray excess emission as reflection from ionized material contributes substantially in the soft x - ray band , as revealed in the case of i zw 1 .
we presented timing and spectral characteristics of the proto - type nls1 galaxy i zw 1 based on an _ xmm - newton _ observation .
the main results are as follows . 1 . the @xmath246 spectrum of i zw 1 , obtained with _ xmm - newton _ , consists of four intrinsic spectral components , namely a steep ( @xmath247 ) primary continuum , a soft x - ray excess component described by a blackbody ( @xmath236 ) and ionized reflection from the disk , iron features k@xmath6 line and k - edge both from he - like
iron arising due to reflection of the primary continuum , and an intrinsic photoelectric absorption component .
i zw 1 showed a large x - ray flare with changes in the luminosity @xmath8 on a time scale of @xmath248 . the radiative efficiency , @xmath249 , inferred from the flare is consistent with x - ray emission outside the last stable orbit around a schwarzschild black hole .
x - ray spectrum of i zw 1 hardened with flux during the observation .
this behavior is uncommon among seyfert galaxies .
the hardening of the @xmath10 is not only due to increase in the hard ( @xmath14 ) band flux but also due to hardening of the power - law component at higher fluxes .
the soft excess emission did not respond to the changes in the power - law continuum .
the spectral variability of i zw 1 is difficult to understand in the framework of disk - corona models but appears to be related with ejection of relativistic plasma at the base of a jet .
simultaneous x - ray and radio observations would be required to understand such spectral variability during large x - ray flares .
this work is based on observations obtained with _ xmm - newton _ , an esa science mission with instruments and contributions directly funded by esa member states and the usa ( nasa ) .
reg acknowledges nasa award nag5 - 9902 in support of his mission scientist position on _ xmm - newton_. arnaud , k. a. et al .
1985 , , 217 , 105 ballantyne , d. r. , iwasawa , k. , & fabian , a. c. 2001 , , 323 , 506 boller , t. , brandt , w. n. , & fink , h. 1996 , , 305 , 53 boroson , t. a. & green , r. f. 1992 , , 80 , 109 brandt , w. n. , mathur , s. , & elvis , m. 1997 , , 285 , l25 de vaucouleurs , g. , de vaucouleurs , a. , corwin , h.g . ,
buta , r.j . , paturel , g. & fouqu/e , p. , 1991
, third reference catalogue of bright galaxies ( new york : springer - verlag ) dewangan , g. c. , singh , k. p. , jones , l. r. , mchardy , i. m. , mason , k. o. , & newsam , a. m. 2001 , , 325 , 1616 dewangan , g. c. , boller , t. , singh , k. p. , & leighly , k. m. 2002 , , 390 , 65 dickey , j. m. & lockman , f. j. 1990 , , 28 , 215 brandt , w. n. , boller , t. , fabian , a. c. , & ruszkowski , m. 1999 , , 303 , l53 collinge , m. j. et al .
2001 , , 557 , 2 done , c. , madejski , g. m. , & zdotycki , p. t. 2000 , , 536 , 213 fabian , a. c. 1979 , royal society of london proceedings series a , 366 , 449 fabian , a. c. & vaughan , s. 2003 , , 340 , l28 gallo , l. c. , boller , t. , tanaka , y. , fabian , a. c. , brandt , w. n. , welsh , w. f. , anabuki , n. , & haba , y. 2003 , arxiv astrophysics e - prints , 9169 georgantopoulos , i. & papadakis , i. e. 2001 , , 322 , 218 gierliski , m. , zdziarski , a. a. , poutanen , j. , coppi , p. s. , ebisawa , k. , & johnson , w. n. 1999 , , 309 , 496 gondoin , p. , orr , a. , lumb , d. , & santos - lleo , m. 2002 , , 388 , 74 haardt , f. & maraschi , l. 1991 , , 380 , l51 jansen , f. et al .
2001 , , 365 , l1 kaspi , s. , smith , p. s. , netzer , h. , maoz , d. , jannuzi , b. t. , & giveon , u. 2000 , , 533 , 631 lamer , g. , mchardy , i. m. , uttley , p. , & jahoda , k. 2003 , , 338 , 323 laor , a. , jannuzi , b. t. , green , r. f. , & boroson , t. a. 1997 , , 489 , 656 leighly , k. m. 1999a , , 125 , 297 leighly , k. m. 1999b , , 125 , 317 magdziarz , p. & zdziarski , a. a. 1995 , , 273 , 837 mchardy i. , papadakis i. , uttley p. , 1998
b ( proc . suppl . ) , 69/1 - 3 , 509 merloni , a. & fabian , a. c. 2001 , , 328 , 958 molendi , s. , & sembay , s. 2003 , xmm - soc - cal - tn-0036 .
nandra , k. & papadakis , i. e. 2001 , , 554 , 710 osterbrock , d. e. & pogge , r. w. 1985 , , 297 , 166 pounds , k. a. , done , c. , & osborne , j. p. 1995
, , 277 , l5 pravdo , s. h. , nugent , j. j. , nousek , j. a. , jensen , k. , wilson , a. s. , & becker , r. h. 1981 , , 251 , 50 romano , p. , turner , t. j. , mathur , s. , & george , i. m. 2002 , , 564 , 162 shakura , n. i. & sunyaev , r. a. 1973 , , 24 , 337 shapiro , s. l. & teukolsky , s. a. 1983 , research supported by the national science foundation .
new york , wiley - interscience , 1983 , 663 p. shih , d. c. , iwasawa , k. , & fabian , a. c. 2002 , , 333 , 687 singh , k. p. , garmire , g. p. , & nousek , j. 1985 , , 297 , 633 singh , k. p. , rao , a. r. , & vahia , m. n. 1991 , , 248 , 37 str " uder , l. et al .
2001 , , 365 , l18 thorne , k. s. 1974 , , 191 , 507 titarchuk , l. 1994 , , 434 , 570 turner , m. j. l. et al.2001a , , 365 , l27 turner , t. j. et al.2001b , , 548 , l13 turner , t. j. , george , i. m. , nandra , k. , & turcan , d. 1999 , , 524 , 667 vaughan , s. , boller , t. , fabian , a. c. , ballantyne , d. r. , brandt , w. n. , & tr " umper , j. 2002 , , 337 , 247 vaughan , s. , reeves , j. , warwick , r. , & edelson , r. 1999 , , 309 , 113 vaughan , s. & edelson , r. 2001 , , 548 , 694 vignali , c. , brandt , w. n. , boller , t. , fabian , a. c. , & vaughan , s. 2003 , arxiv astrophysics e - prints , 10278 zdziarski , a. a. , lubi ' nski , p. , gilfanov , m. , & revnivtsev , m. 2003 , , 342 , 355 zdziarski , a. a. & grandi , p. 2001
, , 551 , 186 lcccc & & + comp . &
parameter & bb+pl+g+edge & bb+pl+g+refl & bb+g+comptt+refl + & & ( model 1 ) & ( model 2 ) & ( model 3 ) + excess absorption & @xmath250 ( @xmath251 & @xmath252 & @xmath253 & @xmath254 + blackbody & @xmath255 ( @xmath256 ) & @xmath257 & @xmath258 & @xmath259 + & @xmath260 & @xmath261 & @xmath262 & @xmath263 + & @xmath264 & @xmath265 & @xmath266 & @xmath267 + power law & @xmath268 & @xmath269 & @xmath270 & + & @xmath271 & @xmath272 & @xmath273 & + & @xmath274 & @xmath275 & @xmath276 & + & @xmath277 & @xmath278 & @xmath279 & + comptt & @xmath280 & & & @xmath281 + & @xmath282 & & & @xmath283 + & @xmath123 & & & @xmath284 + & @xmath285 & & & @xmath286 + & @xmath287 & & & @xmath288 + & @xmath289 & & @xmath290 + gaussian & @xmath291 & @xmath292 & @xmath293 & @xmath294 + & @xmath295 ( ev ) & @xmath296 & @xmath297 & @xmath298 + & @xmath299 & @xmath300 & @xmath301 & @xmath302 + & ew ( ev ) & @xmath303 & @xmath304 & @xmath305 + edge & @xmath306 & @xmath307 & & + & @xmath123 & @xmath308 & & + pexriv & @xmath114 & & @xmath309 & @xmath310 + & @xmath243 ( @xmath311 ) & & @xmath312 & @xmath313 + total & @xmath314 & 1.6 & 1.6 & 1.6 + & @xmath315 & 1.9 & 2.0 & 1.9 + & @xmath316 & 822/780 & 820.8/779 & 815.2/783 + lccccc & & + comp . & parameter & spectrum 1 & spectrum 2 & spectrum 3 & spectrum 4 + excess absorption & @xmath250 ( @xmath251 & @xmath317 & @xmath318 & @xmath319 & @xmath320 + blackbody & @xmath255 ( @xmath256 ) & @xmath321 & @xmath322 & @xmath323 & @xmath324 + & @xmath325 & @xmath326 & @xmath327 & @xmath328 & @xmath329 + & @xmath330 & @xmath331 & @xmath331 & @xmath265 & @xmath331 + power law & @xmath268 & @xmath332 & @xmath333 & @xmath334 & @xmath335 + & @xmath271 & @xmath336 & @xmath337 & @xmath338 & @xmath339 + & @xmath340 & @xmath341 & @xmath279 & @xmath342 & @xmath343 + & @xmath344 & @xmath345 & @xmath346 & @xmath347 & @xmath346 + gaussian & @xmath348 & @xmath349 & @xmath349 & @xmath349 & @xmath349 + & @xmath350 ( ev ) & @xmath351 & @xmath351 & @xmath351 & @xmath351 + & @xmath352 & @xmath353 & @xmath354 & @xmath355 & @xmath356 + & ew ( ev ) & @xmath357 & @xmath358 & @xmath359 & @xmath360 + edge & @xmath306 & @xmath361 & @xmath361 & @xmath361 & @xmath361 + & @xmath123 & @xmath362 & @xmath362 & @xmath362 & @xmath362 + total & @xmath363 & @xmath364 & 1.50 & @xmath365 & @xmath366 + & @xmath367 & @xmath368 & 1.73 & @xmath369 & @xmath370 + & @xmath316 & 588.2/612 & 556.2/600 & 726.6/731 & 709.3/648 + | we present results based on an _ xmm - newton _ observation of the prototype narrow - line seyfert 1 galaxy i zw 1 performed in june 2002 . the @xmath0 spectrum is well described by a steep power - law ( @xmath1 ) and a weak soft blackbody component ( @xmath2 ) below @xmath3 .
the soft x - ray excess emission is featureless and contributes only @xmath4 to the total x - ray emission in the @xmath5 band .
there are tentative evidences for an iron iron k@xmath6 line and an iron k edge , both from high ionized he - like iron .
the @xmath0 continuum of i zw 1 can be explained in terms of a blackbody ( @xmath7 ) , thermal comptonization and its reflection off the surface of an ionized accretion disk .
we note that characterizing the soft excess copmonent ` above a power law continuum ' may not be appropriate due to the low energy curvature of a comptonized spectrum .
x - ray emission from i zw 1 was highly variabile during the observation .
i zw 1 showed a large , symmetric x - ray flare during which the x - ray luminosity increased by @xmath8 in an interval of @xmath9 only .
the x - ray spectral variability was unusual during the flare .
in contrast to the general observation of spectral steepening with flux from seyfert galaxies , there is a clear evidence for hardening of the @xmath10 spectrum of i zw 1 with increasing flux .
this hardening is due to a stronger and slightly flatter power - law component at higher flux without any change in the soft x - ray excess component .
the large flare , the accompanying spectral flattening , and the steady nature of the soft excess component may be due to an additional flatter power - law component during the flare possibly arising as a result of ejection of electron plasma and the ssc process . |
entanglement of formation ( @xmath0 ) and entanglement of distillation ( @xmath1 ) were invented by bennett et al in ref.@xcite and satellite papers . in a series of previous papers@xcite
, we showed how to express @xmath0 in terms of conditional mutual information ( cmi ) , but we said nothing about @xmath1 . in this brief letter , we will show how to express @xmath1 in terms of cmi .
recently , other researchers have expressed some of their entanglement ideas in terms of unconditional mutual information .
see , for example , ref.@xcite .
two reasons why cmi is useful for quantifying entanglement are the following .
first , entanglement is an
exclusively quantum " effect .
cmi satisfies this requirement .
it vanishes in the classical regime , but not in the quantum regime , for a fiducial experiment .
second , entanglement is associated with a correlation between two events @xmath2 and @xmath3 .
but there must be something to distinguish entanglement correlations from classical correlations .
cmi satisfies this requirement too .
it measures more than just the correlation of @xmath2 and @xmath3 .
those two events are assumed to have a common ancestor event ( or cause , or antecedent ) in their past , call it @xmath4 , and we condition on that common ancestor .
( see fig.[fig : ent - form ] ) for example , in bohm s version of the epr experiment , @xmath4 might correspond to the event of a spin - zero particle breaking up into two spin - half particles with opposite spins .
we will try to make this paper as self contained as we can for such a short document .
if the reader has any questions concerning notation or definitions , we refer him to ref.@xcite a much longer , tutorial paper that uses the same notation as this paper .
we will represent random variables by underlined letters .
@xmath5 will be the set of all possible values that @xmath2 can assume , and @xmath6 will be the number of elements in @xmath5 .
@xmath7 will represent the cartesian product of sets @xmath5 and @xmath8 . in the quantum case
, @xmath9 will represent a hilbert space of dimension @xmath6 .
@xmath10 will represent the tensor product of @xmath9 and @xmath11 .
red indices should be summed over ( e.g. @xmath12 ) .
@xmath13 will denote the set of all probability distributions on @xmath5 , @xmath14 such that @xmath15 .
@xmath16 will denote the set of all density matrices acting on @xmath9 .
as usual@xcite , for any three random variables , @xmath17 , we define the _ mutual information_(mi ) by h ( : ) = h ( ) + h ( ) - h ( , ) , and the _ conditional mutual information_(cmi ) by h ( : | ) & = & h(| ) + h(| ) - h ( , | ) + & = & h ( ) -h ( , ) -h ( , ) + h ( , , ) .
since @xmath18 , one might be tempted to assume that also @xmath19 , but this is not generally true .
one can construct examples for which cmi is greater or smaller than mi , a fact well known since the early days of classical information theory@xcite , @xcite .
one can define analogous quantities for quantum physics .
suppose @xmath20 , with partial traces @xmath21 , etc .
then we define s ( : ) = s ( _ ) + s ( _ ) -s ( _ , ) , and s(:| ) = s ( _ ) -s ( _ , ) -s ( _ , ) + s ( _ , , ) .
before racing off at full speed , let s warm up with a brief review of the cmi definition of @xmath0 .
consider the classical bayesian net shown in fig.[fig : ent - form ] .
it represents a probability distribution of the form : p(a , b , ) = p(a| ) p(b | ) p ( ) .
[ eq : prob - ent - form]one can easily check that for this probability distribution , @xmath22 is identically zero . in the classical case , we define @xmath0 by e_f(p _ , ) = _ p _ , , k h ( : | ) , [ eq : clas - ent - form]where @xmath23 is the set of all probability distributions @xmath24 with a fixed marginal @xmath25 .
thus , @xmath0 is a function of @xmath25 . if @xmath23 contains a @xmath24 of the form given by eq.([eq : prob - ent - form ] ) , then @xmath26 .
this is always true if @xmath23 is defined to contain all probability distributions with arbitrary positive values of @xmath27 .
but it may not be true if @xmath23 contains only probability distributions with a fixed @xmath27 value .
the fact that the right hand side of eq.([eq : clas - ent - form ] ) vanishes in the classical case ( if @xmath23 includes all @xmath27 values ) is an important motivation for defining @xmath0 this way .
we want a measure of entanglement that is exclusively quantum .
in the quantum case , suppose @xmath28 is a probability distribution for @xmath29 , and @xmath30 is an orthonormal basis for @xmath31 .
for all @xmath32 , suppose @xmath33 , and @xmath34 .
consider a separable " density matrix @xmath35 of the form _ , , = _
[ eq : rho - ent - form]one can easily check that for this density matrix , @xmath36 . in the quantum case
, we define @xmath0 by e_f ( _ , ) = _ _ , , ks ( : | ) , [ eq : quan - ent - form]where @xmath23 equals the set @xmath37 arbitrary @xmath27 , fixed marginal @xmath38 .
thus , @xmath0 is a function of @xmath39 . if @xmath23 contains a @xmath35 of the form given by eq.([eq : rho - ent - form ] ) , then @xmath0 is zero .
the quantum @xmath0 can be nonzero even if @xmath23 contains all density matrices with arbitrary @xmath27 values . in eq.([eq :
quan - ent - form ] ) , we could set @xmath40 , where @xmath41 is the subset of @xmath42 which restricts @xmath35 to be of the form _ , , = _ w _ , where @xmath43 .
one can show that eq.([eq : quan - ent - form ] ) with @xmath40 is identical ( up to a factor of 2 ) to the definition of @xmath0 originally given by bennett et al in ref.@xcite .
other possible @xmath23 choices come to mind .
for example , one could set @xmath23 equal to @xmath44 , where @xmath44 is that subset of @xmath42 which restricts @xmath35 to be of the form _ , , = _
w _ , where @xmath45 need not be pure .
@xmath46 represent different degrees of information about how @xmath35 was created .
@xmath42 represents total ignorance .
in this section , we will define a classical @xmath1 . in the next section
, we will find a quantum counterpart for it . consider the classical bayesian net of fig.[fig : ent - dist ] .
the arrow from @xmath47 to @xmath48 allows what is often referred to as classical communication from alice to bob " .
let @xmath49 , @xmath50 .
let @xmath51 and @xmath52 . the net of fig.[fig : ent - dist ] satisfies : p(a , b , a , b ) = _ x , x p(a , a| a , a ) p(b , b |b , b , a , a ) p(x ) p(x ) , [ eq : prob - ent - dist]where [ eq : prob - sepa ] p(x ) = _ p(a|)p(b|)p ( ) , and p(x ) = _ p(a|)p(b|)p( ) .
we wish to consider only those experiment in which @xmath53 and @xmath54 are both fixed at a known value , call it 0 for definiteness . for such experiments ,
one considers : p(a , b | a=b=0 ) = .
henceforth we will use @xmath55 as a short - hand for the string @xmath56 " .
we will also use @xmath57 to denote @xmath58 and @xmath59 to denote @xmath60 . in the classical case
, we define @xmath1 by e_d(p _ , p_ ) = _ u , v _ p _ , , | k h ( : | , ) , where @xmath23 is the set of all probability distributions @xmath61 with a fixed marginal @xmath62 that satisfies eq.([eq : prob - ent - dist ] ) .
@xmath62 depends on @xmath63 . since we maximize over @xmath64
, @xmath1 is a function of @xmath65 and @xmath66 .
next we will show that the net of fig.[fig : ent - dist ] , * without the classical communication arrow * , satisfies : e_d(p _ , p_)e_f(p _ ) + e_f(p_ ) .
[ eq : ed - lt - ef]suppose we could show that h ( : | , , )
h ( ( , ) : ( , )| , ) .
[ eq : cmis - goal]after taking limits on the left hand side , this gives e_d h ( , : , | , ) .
[ eq : cmis - left]note that by the independence of the prime and unprimed variables h ( , : , | , ) = h(:| ) + h(:| ) .
[ eq : cmis - right]eqs.([eq : cmis - left ] ) and ( [ eq : cmis - right ] ) imply eq.([eq : ed - lt - ef ] ) .
so let us concentrate on establishing eq.([eq : cmis - goal ] ) .
events @xmath67 all occur before @xmath55 so they are independent of @xmath55 . therefore , we can write : h ( , : , | , ) = h ( , : , | , , ) .
[ eq : gamma - indep]because of eq.([eq : gamma - indep ] ) , eqs.([eq : cmis - goal ] ) is equivalent to : h ( : | , , )
h ( , : , | , , ) .
[ eq : cmis - goal2]eq.([eq : cmis - goal2 ] ) follows easily from the following lemma , which is proven in appendix[app : dpis ] .
lemma : the net of fig.[fig : data - pro ] satisfies h ( : | ) h ( : | ) . [ eq : dat - pro - cmi ]
in this section , we will give a quantum counterpart of the classical @xmath1 defined in the previous section .
as in the classical case , let @xmath49 , @xmath50 .
let @xmath51 and @xmath52 .
suppose @xmath68 and @xmath69 are given .
suppose @xmath70 is a unitary transformation mapping @xmath71 onto @xmath72 : u^ _ , | , u _
, | , = 1 .
likewise , suppose that for each @xmath73 , @xmath74is a unitary transformation mapping @xmath75 onto @xmath76 : v^a _ , | , v _ , | , ^a = 1 .
define the following projector on @xmath9 : _ ^a= _ _ . now consider the following density matrix _ , | = _ a _
^a u _ , | , v _ , | , ^a _ _ v^a _ , | , u^ _ , | , _
^a , [ eq : rho1-ent - dist]where @xmath77 is defined so that @xmath78 .
the previous equation can also be expressed in index notation as : & = & + & & + & & + & & + & & + & & .
finally , we define @xmath1 by e_d ( _ , _ ) = _ u , v _ _ , , |k s ( : | , ) , [ eq : quan - ent - dist]where @xmath23 contains all density matrices @xmath79 with a fixed marginal @xmath80 that satisfies eq.([eq : rho1-ent - dist ] ) .
in this appendix , we will prove two well known data processing inequalities .
[ lemma : dpi - re ] ( data processing inequality for relative entropy , see ref.@xcite ) if @xmath81 and @xmath84 is a matrix of non - negative numbers such that @xmath85 , then d(p//q)d(tp//tq ) , where @xmath86 should be understood as the matrix product of the column vector @xmath82 times the matrix @xmath87 . for any two random variables
@xmath88 , let @xmath89 be shorthand for @xmath90 . in other words , @xmath91 for all @xmath92 .
the two cmi we are dealing with can be rewritten in terms of relative entropy as follows : h ( : | ) = _ p ( ) d ( q^ _ , //
q^_q^ _ ) , and h ( : | ) = _ p ( ) d ( q^ _ , // q^_q^ _ ) . thus , if we can show that d ( q^ _ , //
q^_q^ _ ) d ( q^ _ , //
q^_q^ _ ) , then the present lemma will be proven .
the last inequality will follow from lemma [ lemma : dpi - re ] if we can find a transition probability matrix @xmath93 such that * r.r .
tucci , quantum entanglement and conditional information transmission " , quant - ph/9909041 * r.r .
tucci , separability of density matrices and conditional information transmission " , quant - ph/0005119 * r.r .
tucci , entanglement of formation and conditional information transmission " , quant - ph/0010041 * r.r .
relaxation method for calculating quantum entanglement " , quant - ph/0101123 * r.r .
tucci , entanglement of bell mixtures of two qubits " , quant - ph/0103040 this data processing inequality for relative entropy is well known .
ref.@xcite mentions it on page 300 .
our proof of the inequality comes from page 55 of the book by i. csiszar and j. korner , information theory - coding theorems for discrete memoryless systems " , ( academic press , 1981 ) . | in previous papers , we expressed the entanglement of formation in terms of conditional mutual information ( cmi ) . in this brief paper , we express the entanglement of distillation in terms of cmi . |
for a spherical silicon nanoparticle ( sn ) , the contributions of spherical electric dipole ( ed ) and magnetic dipole ( md ) to the total scattering can be read as : @xmath45 @xmath46 @xmath47\psi_{1}(kr)-\psi_{0}(kr)}{[d_{1}(mkr)/m+1/(kr)]\xi_{1}(kr)-\xi_{0}(kr)}\end{aligned}\ ] ] @xmath48\psi_{1}(kr)-\psi_{0}(kr)}{[md_{1}(mkr)+1/(kr)]\xi_{1}(kr)-\xi_{0}(kr)}\end{aligned}\ ] ] where @xmath1 = 2@xmath2 is the wavenumber , @xmath3 is the wavelength in vacuum , @xmath4 is the radius of the nanoparticle and @xmath5 is the ratio between the refractive indices of si and background medium ( air in this paper ) .
@xmath49 , @xmath7 and @xmath50 are the riccati - bessel functions , @xmath51 = @xmath52 or @xmath53 and @xmath54 and @xmath8 are the mie scattering coefficients .
the contributions of high order electric and magnetic multipoles to the total scattering are assumed to be negligible . therefore , the conditions to realize an _ ideal magnetic dipole scatter _ are the solutions ( @xmath55 ) of this equation set : @xmath56 where the prime means differentiation with respect to the argument in parentheses . in realistic case , the _ ideal md scattering _
condition can be slightly released as the solutions of the following equation set : @xmath57 unfortunately , this equation set also has no solution in the visible to near infrared wavelength range .
alternatively , a core / shell geometry offers additional tunability , where the contributions of spherical ed and md to the total scattering also can be calculated by eqs .
( 1 ) and ( 2 ) with @xmath58\psi_{1}(kr)-\psi_{0}(kr)}{[\widetilde{d}_{1}/m_{2}+1/(kr)]\xi_{1}(kr)-\xi_{0}(kr)}\end{aligned}\ ] ] @xmath59\psi_{1}(kr)-\psi_{0}(kr)}{[\widetilde{g}_{1}m_{2}+1/(kr)]\xi_{1}(kr)-\xi_{0}(kr)}\end{aligned}\ ] ] the realization of _ ideal magnetic dipole scattering _ requires the following equation set has solutions in the optical spectrum : @xmath60 where @xmath61 @xmath62 @xmath63 @xmath64 where @xmath1 = 2@xmath2 wavenumber , @xmath3 is the wavelength in vacuum , @xmath17 and @xmath4 are the inner and outer radius of the core / shell nanoparticles , respectively , which are independent variables of eq . ( 9 ) and @xmath27 with @xmath28 and @xmath29 are relative refractive indices of inner and outer mediums .
this equation set can be solved numerically .
we truncate this equation set use @xmath65 = 2 here .
as can be seen from the plane wave scattering results shown in fig .
s1 , the ed and md modes in an au(core)/silicon(shell ) nanoparticle might overlap in spectrum , leading to the unidirectional scattering around 1.34 @xmath66 .
( color online ) ( a ) scattering efficiency of a au(core)/silicon(shell ) nanoparticle embedded in air with an outer radius @xmath4 = 180 nm and an inner radius @xmath17 = 52 nm .
the contributions of spherical ed , md , electric quadrupole ( eq ) and magnetic quadrupole ( mq ) to the total scattering are also presented .
sketch map shows the geometry of the core / shell nanoparticle .
( b ) far field scattering pattern at the wavelength marked by the dashed line in ( a ) .
a plane wave incidents along x - axis with its electric polarization along y - axis .
( c ) cross section views of the far field scattering pattern shown in ( b ) . ]
_ ideal md scattering _ with tunable resonant wavelengths can be realized by changing the sizes of the au core and silicon shell accordingly . as can be seen from fig .
s2 , the resonant frequencies of pure md mode can be tuned from @xmath67 1000 nm to @xmath67 1700 nm . at such resonant regions , the contributions from high order modes
are all negligible , validating the _ ideal md scattering _ in the near infrared wavelength spectrum .
( color online ) ( a ) scattering efficiency of a au(core)/silicon(shell ) nanoparticle embedded in air with an outer radius @xmath4 = 148 nm and an inner radius @xmath17 = 40 nm .
the contributions of spherical ed , md , eq and mq to the total scattering are also presented .
( b ) and ( c ) are similar to ( a ) except for @xmath4 = 180 nm , @xmath17 = 62 nm and @xmath4 = 280 nm , @xmath17 = 140 nm , respectively .
vertical dashed lines indicate the resonant frequencies where the _ ideal md scattering _ appears . ]
in order to explore the contribution of anapole mode to the _ ideal magnetic dipole scattering _ , we present the contributions of cartesian electric and toroidal dipole ( td ) modes to the total scattering , as shown in fig .
similar to the results of spherical nanoparticle presented in fig .
4 , the results in the lossless case indicate that the destructive interference between the cartesian ed and td results in zero response of spherical ed in the far - field scattering . the overlap of anapole mode and md resonance facilitates the realization of _ ideal md scattering _ in core / shell nanodisk .
( color online ) spherical ed ( red ) and md ( blue ) , cartesian ed ( p , purple ) and td ( green ) dipole moments contributions to the total scattering of core / shell nanoparticle shown in fig . 5 for lossy ( a ) and lossless ( b ) cases .
vertical dashed line indicates the resonant frequency where the ideal md scattering appears . ]
we also consider the effects of structure imperfections on the performance of the core / shell ideal md scatter based on the nanodisk . as shown in fig .
s4 , small deviations of the positions of the au core have only a minor effect on the ideal md scattering .
special attention should be paid to the fabrication process of this kind of nanostructure to preserve the exotic property of the ideal md scatter . | we introduce the concept of tunable _ ideal magnetic dipole scattering _ , where a nonmagnetic nanoparticle scatters lights as a pure magnetic dipole .
high refractive index subwavelength nanoparticles usually support both electric and magnetic dipole responses .
thus , to achieve ideal magnetic dipole scattering one has to suppress the electric dipole response .
such a possibility was recently demonstrated for the so - called anapole mode , which is associated with zero electric dipole scattering . by overlapping magnetic dipole resonance with the anapole mode
we achieve ideal magnetic dipole scattering in the far - field with tunable high scattering resonances in near infrared spectrum .
we demonstrate that such condition can be realized for two subwavelength geometries .
one of them is core - shell nanosphere consisting of au core and silicon shell .
it can be also achieved in other geometries , including nanodisks , which are compatible with current nanofabrication technology .
pacs numbers : : 42.25.-p , 78.67.-n , 42.25.fx light - matter interaction mainly relies on the electric component of light , especially for nonmagnetic materials @xcite .
the responses of such materials to the magnetic component of light is usually negligible at frequencies above a few terahertz @xcite .
therefore , realization of effective optical magnetism is very crucial for designing functional metamaterials working at this spectrum range @xcite .
optical mie resonances in dielectric nanoparticles with high refractive indices provide a platform for studying magnetic light - matter interaction mediated by their optical induced magnetic resonances @xcite .
such a nanoparticle supports a magnetic dipole ( md ) mode which resembles a circular displacement current distribution in the near - field and therefore manifests itself as a md scatter in the far - field . to date , various promising functions have been demonstrated via involving md modes , such as directional scattering of light @xcite , fano resonances @xcite , enhanced optical nonlinearity @xcite , nonradiate anapole @xcite , subwavelength topological edge states @xcite , purcell factor enhancements @xcite , and metasurface @xcite etc . in general , the resonant frequencies of electric dipole ( ed ) and md modes in silicon nanoparticles ( sns ) are shifted with respect to each other .
thus , they spectraly overlap in quite significant region , making sn as a dressed md scatter @xcite . as a result
, they can exhibit various interference effects in the far - field , leading to the unidirectional fano resonant scattering of light @xcite .
therefore , in order to unambiguously study the md scattering , it is necessary to suppress the electric dipole and all higher - order multipoles .
such kind of exotic ideal md scatterer would be extremely useful for the study of _ magnetic light - matter interaction_. regarding a sn , the conditions to realize an ideal md scatter are the solutions of the following equation set ( details can be found in the supplemental material @xcite ) :
@xmath0 where @xmath1 = 2@xmath2 is the free space wavenumber , @xmath3 is the wavelength in vacuum , @xmath4 is the radius of the sn , @xmath5 is the relative refractive indices of si with respect to the ambient medium , @xmath6,@xmath7 is the riccati - bessel function and @xmath8 is the mie scattering coefficient .
thus , we have just two free parameters - incident wavelength @xmath3 and radius @xmath4
. however , one can easily check that this set of equations does not have a solution for @xmath3 , and @xmath4 in the visible and near infrared wavelength range .
it can be understood due to the fact that these two parameters are related to each other via so - called size parameter @xmath9 and by simply changing the radius of sn will simultaneously shift the resonances of the ed and md along the same spectrum direction and the spectrum overlap between ed and md is always preserved @xcite .
but , in general , there is a possibility to achieve an ideal md scattering via overlap the md mode with the nonradiated anapole mode which had been demonstrated very recently in sns @xcite .
it has also been proved that such anapole can be effectively tuned in ag core / silicon shell nanoparticle to realize scattering transparency @xcite .
therefore , it is expected that core / shell nanoparticles could provide the possibility to suppress the ed mode at the md resonance . in this letter
, we introduce the concept of _ ideal md scatter scattering _ whose ed and high order electric and magnetic multipoles are negligible at the resonant frequency of md .
such suppression is based on the excitation of the anapole mode , which is a result of the destructive interference between cartesian ed and toroidal dipole ( td ) modes @xcite , in an au core / silicon shell nanoparticle .
we show that there are two ed modes in this hybrid system and their interaction can be used to shift the resonant frequency of the anapole mode to overlap with the md mode .
we demonstrate both analytically and numerically that the total scattering of an au core / silicon shell nanoparticle is completely dominated by its md mode which manifests itself as an ideal md scatter based on nonmagnetic materials .
such system can facilitate the study of magnetic light - matter interaction in nanoscale .
( color online ) ( a ) scattering efficiency of a silicon sphere embedded in air with a radius @xmath4 of 180 nm .
the contributions of spherical ed , md , electric quadrupole ( eq ) and magnetic quadrupole ( mq ) to the total scattering are presented .
( b ) far - field scattering pattern at the md resonant wavelength marked by a dashed line in ( a ) .
a plane wave is impinged along x - axis with its electric polarization along y - axis .
( c ) cross - section views on the x - y plane ( e plane ) and x - z plane ( h plane ) of the far - field scattering pattern shown in ( b ) . ] as mentioned above , the spectral overlap of ed and md modes is always present . in fig .
1 ( a ) we show the results based on the mie theory @xcite for a sn of radius @xmath10 = 180 nm . the md resonance occurs when the size of sn satisfies the condition @xmath11 @xcite , where @xmath12 is the radius of the sn , @xmath13 is the refractive index of silicon and @xmath14 is the free space wavelength .
considerable spectrum overlap between ed and md modes in the range 900 - 1600 nm is observed .
it can be seen from figs . 1 ( b ) and 1(c ) that the far - field scattering pattern at the resonant frequency of md is asymmetric and deviates from an ideal dipole type scattering pattern with non - zero scattering intensities along magnetic dipole axis at @xmath15 and @xmath16 .
it means that the ed component contributes to the far - field scattering and the sn can not be regarded as an ideal md scatter .
( color online ) ( a ) scattering efficiency of a au(core)/silicon(shell ) nanoparticle embedded in air with an outer radius @xmath4 = 180 nm and an inner radius @xmath17 = 62 nm .
the contributions of spherical ed , md , eq and mq to the total scattering are also presented .
a plane wave is incident along x - axis with its electric polarization along y - axis .
the far - field scattering patterns at the md and ed resonant wavelengths indicated by dashed lines are presented in the insets .
the induced electric @xmath18 and magnetic @xmath19 near - field distributions , which are obtained by subtracting the incident field from the total field , are shown for md ( b ) and ed ( c ) modes .
the corresponding cross section views of the far - field scattering patterns in ( a ) are also shown in ( b ) and ( c ) . ]
( color online ) the first column : sketch of various nanoparticle geometries , which show the au nanosphere embedded in silicon , silicon shell with a void and au(core)/silicon(shell ) nanoparticle embedded in air .
the outer radius @xmath10 of silicon shells are 180 nm ; the second column : the corresponding spherical md contributions to the total scattering referring to the structure on the left ; the third column : the corresponding spherical ed contributions to the total scattering for the lossy case ; the fourth column : the corresponding spherical ed contributions to the total scattering for the lossless case . ] for a typical au(core)/silicon(shell ) nanoparticle , the degree of spectrum overlap between the ed and md resonances can be reduced ( see fig .
s1 in the supplemental material @xcite ) . compared with the sn of the same outer radius shown in fig .
1 , the md resonant wavelength is almost unchanged and the md contribution to total scattering is only slightly reduced as the au core is located at the node of electric field for the md mode .
most importantly , the anapole mode appears around @xmath20 = 1.25 @xmath21 m ( see fig .
s1 in the supplemental material @xcite ) .
these results suggest that there is a possibility to overlap the anapole and md modes in a core / shell nanoparticle .
analytically , the conditions to obtain _ ideal md scattering _ of a core / shell nanoparticle are the solutions of the following equation set : @xmath22 where @xmath23 @xmath24 @xmath25 @xmath26 where @xmath1 = 2@xmath2 is the free space wavenumber , @xmath3 is the wavelength in vacuum , @xmath17 and @xmath4 are the inner and outer radius of the core / shell nanoparticle and @xmath27 with @xmath28 and @xmath29 are the relative refractive indices of inner and outer materials relative to the ambient medium . one set of the typical solutions of eq .
( 2 ) in the optical spectrum range , 600 nm @xmath30 @xmath3 @xmath30 1600 nm , are @xmath3 = 1280 nm , @xmath17 = 62 nm and @xmath4 = 180 nm ( see fig .
s2 in the supplemental material @xcite for other solutions ) .
fig .
2 ( a ) presents the results for this case .
we can see that the wavelength where the peak of the spherical md contribution to the total scattering locates is exactly equal to the zero response of spherical ed . meanwhile , the contributions from eq and mq modes are negligible .
the far - field scattering results shown in figs .
2(a ) and 2(b ) also validate a pure md response , exhibiting as a doughnut shape and zero scattering intensities along the magnetic dipole axis ( @xmath15 and @xmath31 ) . for comparison
, we also provide the far - field scattering results at the peak of the ed mode in figs .
2(a ) and 2(c ) .
they obviously deviate from the ideal ed scattering because of the spectrum overlap between ed and md modes . in order to further explore the underlying physics for the overlapping of the anapole and md modes in a core / shell nanoparticle
, we inspect the scattering of an au nanoparticle embedded in a silicon background first , as shown in figs .
3(a)-(c ) .
as expected , there is no md response for the au particle ( see fig .
3(a ) ) .
the dependence of the contribution of spherical ed to the total scattering on the size of the au nanoparticle is shown in fig .
3(b ) .
one can see that there is a lower limit for the resonant wavelength of the ed mode ( @xmath3 = 600 nm ) .
we further consider the case of a silicon shell ( @xmath10 = 180 nm ) embedded in vacuum with a varied size of air core ( void ) in figs .
3(d)-(f ) .
as can be seen from these figures , ed and md resonances are blue shifted together with the increasing of the void size due to the reduction of effective permittivity .
finally , we inspect the case of an au core / silicon shell nanoparticle shown in figs .
3(g)-(i ) .
it is shown that there is a bifurcation point in the spherical ed decomposition results when we increase the radius @xmath32 of au core from 10 nm to 25 nm [ see figs .
3(h ) and 3(i ) ] , indicating strong interaction between two ed modes and the effective excitation of the anapole mode ( destructive interference between cartesian ed and td modes ) @xcite .
most importantly , we can conclude from figs .
3(h ) and 3(i ) that the anapole mode can be fine - tuned from 600 - 1600 nm which can resonantly overlap with the md mode at the near infrared spectrum shown in fig . 3 ( g ) .
( color online ) spherical ed ( red ) and md ( blue ) , cartesian ed ( p , purple ) and td ( green ) moments contributions to the total scattering of the core / shell nanoparticle shown in fig . 3 for lossy ( a ) and lossless ( b ) cases .
vertical dashed line indicates the resonant frequency where the ideal md scattering appears . ] in order to validate the effective excitation of the anapole mode in this core / shell nanoparticle , we calculate the scattering efficiency of cartesian ed and td as @xcite @xmath33 where @xmath34,\end{aligned}\ ] ] @xmath35 is the permeability of vacuum , @xmath36 is the angular frequency , @xmath37 is the speed of light and @xmath38 specifies the location where the induced current @xmath39 is evaluated . as can be seen in fig .
4 , there is a cross point for the scattering efficiency of cartesian ed and td contributions , implying that the condition ( @xmath40 ) when the cartesian ed and td can cancel the scattering of each other has been fulfilled @xcite .
the slight difference between the wavelengths where the contribution of cartesian ed is equal to td and the minimum response of spherical ed is due to the material losses .
figure 4 ( b ) presents the lossless case in which the cross point of cartesian ed and td are exactly the same with the zero response of spherical ed , indicating the resonant excitation of the anapole mode at the resonance of the md mode , i.e. _ ideal md scattering_. it should be noted here that we only mimic the pure md scattering in the far - field , since the near - field distribution of electromagnetic fields shown in fig .
2 ( b ) is a superposition of md and anapole modes .
since a planar nanostructure is more compatible with the state - of - the - art nanofabrication technologies for silicon photonics , a planar ideal magnetic scatter is of great demand compared with the spherical core / shell nanoparticle .
we further provide a design for the ideal md scattering based on a nanodisk .
the plane wave scattering results of the au(core)/silicon(shell ) nanodisk shown in fig . 5 confirm that such structure can also realize the _
ideal md scattering _ in the near infrared wavelength range ( see fig .
s3 in the supplemental material @xcite for the decomposition results ) .
we also investigate the effect of imperfections on the ideal md scattering .
we model the imperfections in the fabrication process with 5 nm position variation of the au core according to the state - of - the - art fabrication technology .
the calculated results demonstrate that the proposed _
ideal md scattering _ is still robust when moderate fabrication imperfections are taken into account ( see fig .
s4 in the supplemental material @xcite ) .
( color online ) ( a ) scattering efficiency of a au(core)/silicon(shell ) nanodisk embedded in air with the outer radius @xmath4 = 160 nm , outer height @xmath41 = 350 nm and an inner radius @xmath17 = 50 nm , inner height @xmath42 = 125 nm .
the contributions of spherical ed , md , eq and mq to the total scattering are also presented based on the decomposition utilizing spherical harmonics @xcite .
schematic shows the geometry of the nanodisk .
a plane wave is incident along the axis of the cylinder ( x - axis ) with its electric polarization along y - axis .
( b ) the far - field scattering pattern at the ideal md scattering wavelength indicated by the dashed line in ( a ) .
( c ) cross section views on the x - y plane ( e plane ) and x - z plane ( h plane ) of the far - field scattering pattern shown in ( b ) . ]
the upper results demonstrate that structured nonmagnetic nanoparticles can be used to realize the _
ideal md scattering_. it should be pointed out that structured light can also be used to realize ideal md scattering as the md mode in silicon nanoparticle resembles the electromagnetic field distributions for an azimuthally polarized beam @xcite . in summary , we demonstrate both analytically and numerically that au(core)/si(shell ) nanoparticles , including spherical and disk geometries , can be used to realized the _ ideal md scattering _ , where other multipole modes ( including the ed mode ) are suppressed , with tunable resonant wavelengths at the near infrared spectrum .
these types of nanoparticles provide an ideal platform to study the magnetic light - matter interaction in nanoscale . based on our results
, we expect that functional magnetic light based devices can be realized , such as pure magnetic mirror and ideal directional magnetic scattering .
y. xu and t. feng thank the financial support from national natural science foundation of china ( grant nos . 11674130 and 11304047 ) ; natural science foundation of guangdong province , china ( grant nos .
2016a030306016 , 2014a030313376 ) ; 99 a. f. koenderink , m. kafesaki , b. c. buchler , and v. sandoghdar , phys . rev . lett . * 95 * , 153904 ( 2005 ) .
m. burresi , d. van oosten , t. kampfrath , h. schoenmaker , r. heideman , a. leinse , l. kuipers , science , * 326 * , 550 - 553 ( 2009 ) .
z. c. ruan and s. h. fan , phys .
rev .
lett . * 105 * , 013901 ( 2010 ) .
l. verslegers , z. f. yu , z. c. ruan , p. b. catrysse , and s. f. fan phys .
rev .
lett . * 108 * , 083902 ( 2012 ) n. rotenberg , and l. kuipers , nat . photon .
* 8 * , 919 - 926 ( 2010 ) .
l. d. landau , e. m. lifshitz .
electrodynamics of continuous media ( wiley , new york , 1984 ) .
m. burresi , t. kampfrath , d. van oosten , j. c. prangsma , b. s. song , s. noda , and l. kuipers , phys . rev
. lett . *
105 * , 123901 ( 2010 ) .
s. vignolini , f. intonti , f. riboli , l. balet , l. h. li , m. francardi , a. gerardino , a. fiore , d. s. wiersma , and m. gurioli phys .
rev .
lett . *
105 * , 123902 ( 2010 ) .
j. c. ginn , i. brener , d. w. peters , j. r. wendt , j. o. stevens , p. f. hines , l. i. basilio , l. k. warne , j. f. ihlefeld , p. g. clem , and m. b. sinclair , phys .
rev .
lett . * 108 * , 097402 ( 2012 ) .
s. jahani and z. jacob , nat . nanotechnol . *
11 * , 2336 ( 2016 ) .
m. kasperczyk , s. person , d. ananias , l. d. carlos , and l. novotny phys .
rev .
lett . * 114 * , 163903 ( 2015 ) .
a. i. kuznetsov , a. e. miroshnichenko , m. l. brongersma , y. s. kivshar , and b. lukyanchuk , science * 354 * , aag2472 ( 2016 ) .
a. garca - etxarri , r. gmez - medina , l. s. froufe - prez , c. lpez , l. chantada , f. scheffold , j. aizpurua , m. nieto - vesperinas , and j. j. senz , opt .
express * 19 * , 4815 - 4826 ( 2011 ) .
a. i. kuznetsov , a. e. miroshnichenko , y. h. fu , j. zhang , and b. lukyanchuk , sci . rep . * 2 * , 492 ( 2012 ) .
a. b. evlyukhin , s. m. novikov , u. zywietz , r. l. eriksen , c. reinhardt , s. i. bozhevolnyi , and b. n. chichkov , nano lett .
* 12 * , 3749 - 3755 ( 2012 ) .
y. g. liu , w. c. h. choy , w. e. i. sha , and w. cho .
chew , opt .
lett . * 37 * , 2112 - 2114 ( 2012 ) .
w. liu , a. e. miroshnichenko , d. n. neshev , and y. s. kivshar , phys .
rev .
b * 86 * , 081407 ( 2012 ) .
y. h. fu , a. i. kuznetsov , a. e. miroshnichenko , y. f. yu , and b. lukyanchuk , nat . commun . * 4 * , 1527 ( 2013 ) . s. person , m. jain , z. lapin , j. j. s@xmath43enz , g. wicks , and l. novotny , nano lett .
* 13 * , 1806 - 1809 ( 2013 ) .
i. staude , a. e. miroshnichenko , m. decker , n. t. fofang , s. liu , e. gonzales , j. dominguez , t. s. luk , d. n. neshev , i. brener , and y. kivshar , acs nano * 7 * , 7824 - 7832 ( 2013 ) .
s. v. li , d. g. baranov , a. e. krasnok , and p. a. belov , appl .
. lett . * 107 * , 171101 ( 2015 ) .
y. yang , q. li , and m. qiu , appl .
phys . lett . *
107 * , 031109 ( 2015 ) .
h. wang , p. liu , y. ke , y. su , l. zhang , n. xu , s. deng , and h. chen , acs nano * 9 * , 436 - 448 ( 2015 ) .
j. h. yan , p. liu , z. y. lin , h. wang , h. j. chen , c. x. wang and g. w. yang , nat . commun . * 6 * , 7042 ( 2015 ) . a. e. miroshnichenko and y. s. kivshar , nano lett .
* 12 * , 6459 - 6463 ( 2012 ) .
k. e. chong , b. hopkins , i. staude , a. e. miroshnichenko , j. dominguez , m. decker , d. n. neshev , i. brener , and y. s. kivshar , small * 10 * , 1985 - 1990 ( 2014 ) .
j. yan , p. liu , z. lin , h. wang , h. chen , c. wang , and g. yang , acs nano * 9 * , 2968 - 2980 ( 2015 ) .
y. tsuchimoto , t. yano , t. hayashi , and m. hara , opt .
express , * 24 * , 14451 - 14462 ( 2016 ) .
m. shcherbakov , d. n. neshev , b. hopkins , a. shorokhov , i staude , e. melik - gaykazyan , m. decker , a. ezhov , a. e. miroshnichenko , i. brener , and y. s. kivshar , nano lett . * 14 * , 6488 - 6492 ( 2014 ) . a. e. miroshnichenko , a. b. evlyukhin , y. f. yu , r. m. bakker , a. chipouline , a. i. kuznetsov , b. lukyanchuk , b. n. chichkov , and y. s. kivshar , nat . commun .
* 6 * , 8069 ( 2015 ) .
a. p. slobozhanyuk , a. n. poddubny , a. e. miroshnichenko , p. a. belov , and y. s. kivshar , phys .
rev .
lett . * 114 * , 123901 ( 2015 ) .
m. k. schmidt , r. esteban , j. j. s@xmath44enz , i. su@xmath44rez - lacalle , s. mackowski , and j. aizpurua , opt .
express * 20 * , 13636 - 13650 ( 2012 ) .
m. decker , i. staude , i. i. shishkin , k. b. samusev , p. parkinson , v. k. a. sreenivasan , a. minovich , a. e. miroshnichenko , a. zvyagin , c. jagadish , d. n. neshev , and y. s. kivshar , nat .
commun . * 4 * , 2949 ( 2013 ) .
x. zambrana - puyalto and n. bonod , phys . rev .
b * 91 * , 195422 ( 2015 ) . t. h. feng , y. xu , z. x. liang , and w. zhang , opt . lett . * 41 * , 5011 - 5014 ( 2016 ) y. m. yang , i. i. kravchenko , d. p. briggs and j. valentine , nat . commun . * 5 * , 5753 ( 2014 ) .
j. xiang , j. x. li , h. li , c. y. zhang , q. f. dai , s. l. tie , and s. lan , opt .
express * 24 * , 11420 - 11434 ( 2016 ) .
supplemental material at http://link.aps.org/supplemental/ ... , which includes the conditions for pure magnetic dipole scattering , plane wave scattering of au(core)/silicon(shell ) nanoparticle , tunable ideal md scattering in nanospheres , origin of anapole mode in au(core)/silicon(shell ) nanodisk and role of imperfections .
w. liu , j. f. zhang , a. e. miroshnichenko , laser & photon . rev .
* 9 * , 564570 ( 2015 ) . c. f. bohren and d. r. huffman , _ absorption and scattering of light by small particles _ ( john wiley and sons , 1983 ) .
e. e. radescu and g. vaman , phys .
rev .
e * 65 * , 046609 ( 2002 ) .
t. kaelberer , v. fedotov , n. papasimakis , d. tsai , and n. zheludev , science * 330 * , 1510 ( 2010 ) .
v. a. fedotov , a. rogacheva , v. savinov , d. tsai , and n. i. zheludev , sci .
rep . * 3 * , 2967 ( 2013 ) .
n. papasimakis , v. a. fedotov , v. savinov , t. a. raybould , and n. i. zheludev , nature mater . * 15 * , 263271 ( 2016 ) .
p. grahn , a. shevchenko , and m. kaivola , new .
j. phys .
* 14 * , 093033 ( 2012 ) . |
the purpose of this paper is to present a new sufficient condition ( which is almost necessary ) on a pointwise precompact sequence @xmath12 of functions @xmath13 mapping a subset @xmath4 of the real line @xmath14 into a metric space @xmath15 , under which the sequence admits a pointwise convergent subsequence . the historically first result in this direction is the classical _ helly selection principle _ , in which the assumptions are as follows : @xmath16 $ ] is a closed interval , @xmath17 , and @xmath5 is uniformly bounded and consists of monotone functions ( @xcite , ( * ? ? ? *
ii.8.910 ) , ( * ? ? ?
* viii.4.2 ) , and ( * ? ? ?
* theorem 1.3 ) if @xmath0 is arbitrary ) .
since a real function on @xmath4 of bounded ( jordan ) variation is the difference of two nondecreasing bounded functions , helly s theorem extends to uniformly bounded sequences of functions , whose jordan s variations are uniformly bounded .
further generalizations of the latter pointwise selection principle are concerned with replacement of jordan s variation by more general notions of variation ( @xcite , @xcite@xcite , @xcite ) . in all these papers ,
the pointwise limit of the extracted subsequence of @xmath5 is a function of bounded generalized variation ( in the corresponding sense ) , and so , it is a regulated function ( with finite one - sided limits at all points of the domain ) . note that pointwise selection principles ( or the sequential compactness in the topology of pointwise convergence ) and regulated functions are of importance in real analysis ( @xcite ) , stochastic analysis and generalized integration ( @xcite ) , optimization ( @xcite ) , set - valued analysis ( @xcite ) , and other fields .
a unified approach to the diverse selection principles mentioned above was proposed in @xcite .
it is based on the notion of _ modulus of variation _ of a function introduced in @xcite ( see also @xcite ) and does not refer to the uniform boundedness of variations of any kind , and so , can be applied to sequences of non - regulated functions .
however , the pointwise limit of the extracted subsequence of @xmath5 is _ again _ a regulated function . in order to clarify this situation and expand the amount of sequences having pointwise convergent subsequences , we define the notion of the _ joint modulus of variation _ for metric space valued functions : this is a certain sequence of pseudometrics @xmath2 on the product set @xmath3 ( of all functions from @xmath4 into @xmath1 ) . making use of @xmath2
, we obtain a powerful pointwise selection principle ( see theorem [ t : first ] in section [ s : main ] ) .
putting @xmath18 for all @xmath19 , where @xmath20 is a constant function , we get the selection principle from @xcite , which already contains all selection principles alluded to above as particular cases . in contrast to results from @xcite , the pointwise limit @xmath21 from theorem [ t : first ] may not be regulated in general this depends on the limit function @xmath22 , namely , since @xmath23 , the function @xmath21 is only conditionally regulated with respect to @xmath22 ( in short , _ @xmath22-regulated _ ) . in particular , if @xmath24 , finally , we point out that following the ideas from @xcite , theorem [ t : first ] may be extended to sequences of functions with values in a _ uniform space _ @xmath1 . the paper is organized as follows . in section [
s : main ] , we present necessary definitions and our main result , theorem [ t : first ] . in section [ s : joint ] , we establish essential properties of the joint modulus of variation , which are needed in the proof of theorem [ t : first ] in section [ s : proof ] .
section [ s : reg ] is devoted to the study of @xmath22-regulated ( and , in particular , regulated ) functions . in the final section [ s : ext ] ,
we extend the helly - type selection theorems from @xcite and @xcite by exploiting theorem [ t : first ] .
let @xmath25 , @xmath15 be a metric space with metric @xmath26 , and @xmath3 denote the set of all functions @xmath27 mapping @xmath4 into @xmath1 .
the letter @xmath28 stands for a _ constant _ function @xmath20 ( i.e. , @xmath29 in @xmath1 for all @xmath30 ) .
the _ joint oscillation _ of two functions @xmath31 is the quantity @xmath32,\ ] ] where latexmath:[\[\label{e : join } is the _ joint increment _ of @xmath21 and @xmath22 on the two - point set @xmath34 , for which the following two inequalities hold : @xmath35 since @xmath36 ( @xmath37the increment of @xmath21 on @xmath34 ) is independent of @xmath28 , the quantity @xmath38 is the usual _ oscillation _ of @xmath21 on @xmath4 , also known as the _ diameter of the image _ @xmath39 . clearly , by , @xmath40 .
we denote by @xmath41 the family of all _ bounded _ functions on @xmath4 equipped with the _ uniform metric _
@xmath42 given by @xmath43 ( @xmath42 is an _ extended _ metric on @xmath3 , i.e. , may assume the value @xmath44 ) .
we have @xmath45 and , by virtue of , @xmath46 . if @xmath47 , we write @xmath48 to denote a collection of @xmath49 two - point subsets @xmath50 of @xmath4 ( @xmath51 ) such that @xmath52 ( so that the intervals @xmath53,\dots,[s_n , t_n]$ ] with end - points in @xmath4 are non - overlapping ) .
we say that a collection @xmath48 with @xmath50 is a _ partition _ of @xmath4 if ( setting @xmath54 ) @xmath55 for all @xmath51 , which is written as @xmath56 .
the _ joint modulus of variation _ of two functions @xmath31 is the sequence @xmath57 $ ] defined by @xmath58 where @xmath59 is the quantity from if @xmath50 ( for finite @xmath4 with the number of elements @xmath60 , we make use of for @xmath61 , and set @xmath62 for all @xmath63 ) . note that , given @xmath31 , we have @xmath64 and @xmath65 further properties of the joint modulus of variation are presented in section [ s : joint ] . for a sequence of functions @xmath66 and @xmath67 , we write : ( a ) @xmath68 on @xmath4 to denote the _ pointwise _ ( or _ everywhere _ ) _ convergence _ of @xmath5 to @xmath21 ( that is , @xmath69 for all @xmath70 ) ; ( b ) @xmath71 on @xmath4 to denote the _ uniform convergence _ of @xmath5 to @xmath21 meaning , as usual , that @xmath72 .
the uniform convergence implies the pointwise convergence , but not vice versa . recall that a sequence @xmath66 is said to be _ pointwise precompact _ on @xmath4 if the closure in @xmath1 of the set @xmath73 is compact for all @xmath70 . making use of e. landau s notation , given a sequence , @xmath74
, we write @xmath75 to denote the condition @xmath76 . our main result , a _ pointwise selection principle _ for metric space valued functions in terms of the joint modulus of variation , is as follows .
[ t : first ] let @xmath25 and @xmath15 be a metric space .
suppose @xmath77 are two sequences of functions such that @xmath5 is pointwise precompact on @xmath4 , @xmath6 is pointwise convergent on @xmath4 to a function @xmath78 , and @xmath79 then , there is a subsequence of @xmath5 , which converges pointwise on @xmath4 to a function @xmath67 such that @xmath80 for all @xmath47 .
this theorem will be proved in section [ s : proof ] .
now , a few remarks are in order . given @xmath67 and a constant function @xmath20 , the quantity @xmath81 ( with @xmath50 ) is independent of @xmath28 , and the sequence is known as the _ modulus of variation _ of @xmath21 in the sense of chanturiya ( @xcite ) .
it characterizes regulated ( or proper ) functions on @xmath16 $ ] as follows .
we say that @xmath82\to m$ ] is _ regulated _ and write @xmath83;m)$ ] if @xmath84 as @xmath85 for every @xmath86 , and @xmath84 as @xmath87 for every @xmath88 ( and so , by cauchy s criterion , one - sided limits @xmath89 exist provided @xmath1 is complete ) .
we have @xmath90;m)=\{f\in m^{[a , b]}:\nu_n(f)=o(n)\}\ ] ] ( more general characterizations for dense subsets @xmath4 of @xmath91 $ ] can be found in @xcite ) . a certain relationship between characterizations of regulated functions and pointwise selection principles
is exhibited in @xcite .
we begin by studying the joint increment , whose properties are gathered in the following lemma .
[ l : pjin ] given @xmath92 and @xmath30 , we have : @xmath93 @xmath94 @xmath95 @xmath96 @xmath97 . properties ( a ) , ( b ) , and ( c ) , showing that @xmath98 is a _ pseudometric _ on @xmath3 , are straightforward . to establish ( d ) and ( e ) ,
take into account equality @xmath99 .
[ r:1 ] ( a ) if @xmath100 , then ( d ) , ( e ) , and ( b ) imply equalities @xmath101 and @xmath102 .
in addition to lemma [ l : pjin ] , the function @xmath103 is a pseudometric on @xmath4 .
\(b ) if @xmath104 denotes the absolute value under the supremum sign in , then @xmath105 and @xmath106 for all @xmath107 .
\(c ) by lemma [ l : pjin](d ) , @xmath108 .
so , @xmath109 moreover , it follows from lemma [ l : pjin](e ) that @xmath110 \(d ) suppose the triple @xmath111 is a _ metric semigroup _ ( ( * ? ? ?
* section 4 ) ) , i.e. , @xmath15 is a metric space , @xmath112 is an abelian semigroup with the operation of addition @xmath113 , and @xmath114 for all @xmath115 .
then , the joint increment may be alternatively replaced by @xmath116 is a normed linear space ( over @xmath14 or @xmath117 ) , we may set @xmath118 and remark [ r:1](a ) . in the sequel , we make use of more general quantity . if @xmath31 , @xmath47 and @xmath119 , we set @xmath120 , where @xmath121 is the restriction of @xmath21 to @xmath122 ( i.e. , @xmath123 for all @xmath124 ) .
accordingly , @xmath125 .
the following properties of the joint modulus of variation are immediate .
the sequence @xmath126 is nondecreasing , @xmath127 for all @xmath128 , and @xmath129 provided @xmath47 and @xmath130 .
it follows from and lemma [ l : pjin](a)(c ) that , for every @xmath47 , the function @xmath131 is a _ pseudometric _ on @xmath3 ( possibly assuming infinite values ) and , in particular ( cf . and ) , @xmath132 for all @xmath47 and @xmath31 .
furthemore , if @xmath133 , then , by , the sequence @xmath134 is bounded in @xmath135
. essential properties of the joint modulus of variation are presented in [ l : ess ] given @xmath47 , @xmath31 , and @xmath119 , we have : @xmath136 for all @xmath137 with @xmath138 , where @xmath139\cap e$ ] for @xmath140 @xmath141 if @xmath77 are such that @xmath68 and @xmath142 on @xmath122 , then @xmath143 if @xmath77 are such that @xmath71 and @xmath144 on @xmath122 , then @xmath145 \(a ) we may assume that @xmath146 .
let @xmath147 .
setting @xmath148 , we find @xmath149 , and so , @xmath150 the inequality in ( a ) follows by taking the supremum over all @xmath147 .
\(b ) we may assume that @xmath151 is finite , and apply the idea from ( * ? ? ?
* lemma ) . given @xmath152
, there is @xmath153 ( depending on @xmath154 ) such that @xmath155 setting @xmath156 , the left - hand side inequality implies .
the right - hand side inequality gives @xmath157 from which our inequality follows due to the arbitrariness of @xmath152 .
\(c ) first , we note that , given @xmath19 and @xmath30 , we have @xmath158 & \quad\,\,+d(g_j(s),g(s))+d(g_j(t),g(t ) ) .
\label{e : ten } \end{aligned}\ ] ] in fact , lemma [ l : pjin](c ) and inequality imply @xmath159 & \le d(f_j(s),f(s))+d(f_j(t),f(t))+|(f , g)(\{s , t\})|\nonumber\\[2pt ] & \quad\,\,+d(g(s),g_j(s))+d(g(t),g_j(t ) ) .
\label{e : sstt } \end{aligned}\ ] ] exchanging @xmath13 and @xmath21 as well as @xmath160 and @xmath22 , we obtain . from the pointwise convergence of @xmath5 and @xmath6 and
, we find @xmath161 by definition , given @xmath162 , we have @xmath163 passing to the limit inferior as @xmath8 , we get @xmath164 since @xmath162 is arbitrary , it remains to take into account .
\(d ) it follows from that , for any @xmath137 and @xmath19 , @xmath165 and so , definition implies @xmath166 for all @xmath19 . passing to the limit superior as @xmath8 , we get @xmath167 now , the equality in ( d ) is a consequence of lemma [ l : ess](c ) .
[ r:2 ] if the value @xmath168 ( see ) is finite for an @xmath130 ( e.g. , when @xmath169 ) , inequality in lemma [ l : ess](b ) is equivalent to @xmath170 thus , the limit @xmath171 always exists in @xmath135 .
since @xmath172 is a ( extended ) pseudometric on @xmath3 , we may introduce an equivalence relation @xmath173 on @xmath3 as follows : given @xmath31 , we set @xmath174 the equivalence class @xmath175 of a function @xmath78 is called the _ regularity class _ of @xmath22 , and any representative @xmath176 is called a _ conditionally regulated _ or , more precisely , _ @xmath22-regulated _ function .
this terminology is justified by : in the framework of the product set @xmath177}$ ] , we have @xmath178;m)=\mbox{r}(c)$ ] for any constant function @xmath179\to m$ ] . note
that , in theorem [ t : first ] , condition ` @xmath80 for all @xmath47 ' means that @xmath176 , and so , the class @xmath180 is worth studying in more detail .
[ t : greg ] given a function @xmath78 , we have : @xmath181 if and only if @xmath182 @xmath180 is closed with respect to the uniform convergence , but not closed with respect to the pointwise convergence in general ; if @xmath15 is a complete metric space , then the pair @xmath183 is also a complete metric space .
\(a ) the sufficiency is clear , because @xmath184 . now , suppose that @xmath181 , so that , by , @xmath185 . given @xmath176 , @xmath23 , and so , @xmath186 for some @xmath187 .
it follows from and that @xmath188 which implies @xmath189 .
\(b ) it is to be shown that if @xmath190 and @xmath71 on @xmath4 with @xmath67 , then @xmath176 .
we will prove a little bit more : namely , if @xmath77 are such that @xmath191 for all @xmath19 , @xmath71 and @xmath144 on @xmath4 with @xmath31 , then @xmath176 ( the previous assertion follows if @xmath192 for all @xmath19 ) .
in fact , exchanging @xmath13 and @xmath21 , and @xmath160 and @xmath22 in , we get @xmath193 by the uniform convergence of @xmath5 and @xmath6 , given @xmath152 , there is a number @xmath194 such that @xmath195 and @xmath196 . since @xmath197 is in @xmath198 , we have @xmath199 , and so , there exists @xmath200 such that @xmath201 for all @xmath202 .
the estimate above with @xmath203 implies @xmath204 , @xmath202 , which means that @xmath23 and @xmath176 . as for the pointwise convergence , consider a sequence of real step functions converging pointwise to the dirichlet function ( @xmath37the characteristic function of the rationals @xmath205 ) on @xmath206 $ ] ( see ( * ? ? ?
* examples 4 , 5 ) and example [ ex : di](a ) in section [ s : proof ] ) .
\(c ) first , we show that @xmath207 for all @xmath208 .
in fact , since @xmath209 , we have @xmath210 , and so , @xmath211 for some @xmath187 .
given @xmath212 , it follows from remark [ r:1](c ) and that @xmath213 the metric axioms for @xmath42 on @xmath180 are verified in a standard way . in order to prove that @xmath180 is complete ,
suppose @xmath190 is a cauchy sequence , i.e. , @xmath214 as @xmath215 .
since @xmath216 for all @xmath70 and @xmath15 is complete , there exists @xmath67 such that @xmath68 on @xmath4 .
noting that @xmath217 and @xmath218 on @xmath4 as @xmath219 ( and arguing as in ) , we get @xmath220 since the sequence @xmath5 is @xmath42-cauchy , we find @xmath221 thus , @xmath72 , and so , @xmath71 on @xmath4 .
applying item ( b ) of this theorem , we conclude that @xmath176 .
a traditionally important class of regulated functions is the space of functions of bounded jordan variation , @xmath222 , which is introduced by means of the joint modulus of variation as follows .
since the sequence @xmath126 is nondecreasing for all @xmath31 , the quantity ( finite or not ) @xmath223 is called the _ joint variation _ of functions @xmath21 and @xmath22 on @xmath4 .
the value @xmath224 is independent of a constant function @xmath20 and is the usual jordan _ variation _ of @xmath21 on @xmath4 : @xmath225 the supremum being taken over all partitions @xmath226 of @xmath4 ( cf.section [ s : main ] ) .
the set @xmath227 is contained in @xmath228 ( in fact , @xmath229 and @xmath230 for all @xmath231 ) . the following notion of _ @xmath154-variation _
@xmath232 , due to frakov ( * ? ? ?
* section 3 ) , provides an alternative characterization ( cf . ) of regulated functions : given @xmath67 and @xmath152 , set @xmath233 it was shown in ( * ? ? ?
* proposition 3.4 ) ( for @xmath16 $ ] ) that @xmath234;m)=\{f\in m^{[a , b]}:\mbox{$v_{\varepsilon}(f)<\infty$ for all $ { \varepsilon}>0$}\}\ ] ] ( although it was assumed in @xcite that @xmath235 , the proof of the last assertion carries over to any metric space @xmath1 , cf .
* lemma 3 ) ) .
the notion of @xmath154-variation will be needed in section [ s : ext ] .
[ ex : dir - type ] given @xmath236 with @xmath237 , let @xmath238\to m$ ] be the dirichlet - type function of the form : @xmath239 $ is rational,}\\[2pt ] y & \mbox{if } & \mbox{$t\in[0,1]$ is irrational . }
\end{array}\right.\ ] ] clearly , @xmath240;m)$ ] .
moreover ( cf . )
, we have @xmath241 to see this , first note that , given @xmath242}$ ] , inequality @xmath243 is equivalent to the following two conditions : @xmath244\cap\mathbb{q}$,\quad and \ , $ d(y , g(t))\le{\varepsilon}$ \,$\forall\,t\in[0,1]\setminus\mathbb{q}$.}\ ] ] suppose @xmath245 . to show that @xmath243 implies @xmath246 , we let @xmath47 , and @xmath247 $ ]
be a partition of @xmath248 $ ] such that points @xmath249 are rational and points @xmath250 are irrational . by the triangle inequality for @xmath26 and , we get @xmath251 if @xmath252 , we set @xmath253 ( or @xmath254 ) for all @xmath255 $ ] , so that is satisfied and @xmath256 .
thus , @xmath257
. the second assertion in can be refined , provided @xmath258 in fact , we may set @xmath259 for all @xmath255 $ ] , so that holds whenever @xmath260 , and @xmath256 .
this implies @xmath257 for all @xmath261 .
a few remarks concerning condition are in order . since @xmath262 condition refers to a certain form of ` convexity ' of @xmath1 ( which is not restrictive for our purposes ) .
for instance , if @xmath263 is a normed linear space with @xmath264 , we may set @xmath265 .
more generally , by menger s theorem ( @xcite , ( * ? ? ?
* example 2.7 ) ) , if a metric space @xmath15 is complete and _ metrically convex _
( i.e. , given @xmath266 with @xmath267 , there is @xmath268 such that @xmath269 and @xmath270 ) , then , for any @xmath266 , there is an isometry @xmath271\to m$ ] such that @xmath272 and @xmath273 . in this case , we set @xmath274 .
more examples of metrically convex metric spaces can be found in @xcite . finally ,
if @xmath275 , then condition is not satisfied , and we have @xmath276 for all @xmath277 , which is a consequence of .
with no loss of generality we may assume that @xmath4 is uncountable ; otherwise , by virtue of assumption ( a ) and the standard cantor diagonal procedure , we extract a pointwise convergent subsequence of @xmath5 and apply lemma [ l : ess](c ) .
note that @xmath278 is finite for all @xmath47 : in fact , @xmath279 whenever @xmath202 for some @xmath187 and , since @xmath280 is nondecreasing for all @xmath19 , we have @xmath281 for all @xmath282 .
we divide the rest of the proof into four steps for clarity . _ step 1 . _
let us show that there is a subsequence of @xmath283 , again denoted by @xmath284 , and a nondecreasing sequence @xmath285 such that @xmath286 we set @xmath287 .
the definition of @xmath288 implies that there is an increasing sequence @xmath289 ( i.e. , a subsequence of @xmath283 ) such that @xmath290 as @xmath8 .
setting @xmath291 , we find @xmath292 , and there is a subsequence @xmath293 of @xmath294 such that @xmath295 as @xmath8 .
inductively , if @xmath296 and a subsequence @xmath297 of @xmath283 is already chosen , we define @xmath298 as the limit superior of @xmath299 as @xmath8 , so that @xmath300 .
now , we pick a subsequence @xmath301 of @xmath297 such that @xmath302 as @xmath8 . noting that the sequence @xmath303 is a subsequence of @xmath301 ( for all @xmath47 ) and denoting the diagonal sequences @xmath304 and @xmath305 again by @xmath5 and @xmath6 , respectively
, we obtain . in the sequel ,
the set of all nondecreasing bounded functions mapping @xmath4 into @xmath306 is denoted by @xmath307
. _ step 2 . _ in this step , we prove that there are subsequences of @xmath5 and @xmath6 from , again denoted by @xmath5 and @xmath6 , respectively , and a sequence of functions @xmath308 such that @xmath309 where @xmath310 for @xmath70 . note that , for each @xmath47 , the function @xmath311 is nondecreasing on @xmath4 , and @xmath312 for all @xmath70 and @xmath47 . by virtue of
, there is a sequence @xmath313 such that @xmath314 for all @xmath315 . in what follows ,
we apply the diagonal procedure once again .
the sequence @xmath316 is uniformly bounded by constant @xmath317 , and so , by helly s selection principle , there are an increasing sequence @xmath318 ( i.e. , a subsequence of @xmath283 ) and a function @xmath319 such that @xmath320 as @xmath8 for all @xmath70 .
the sequence @xmath321 is uniformly bounded on @xmath4 by constant @xmath322 , and so , again by helly s theorem , there are a subsequence @xmath323 of @xmath324 and a function @xmath325 such that @xmath326 as @xmath8 for all @xmath70 .
inductively , if @xmath296 and a subsequence @xmath327 of @xmath283 and a function @xmath328 are already chosen , we apply the helly theorem to the sequence of functions @xmath329 , which is uniformly bounded on @xmath4 by constant @xmath330 : there are a subsequence @xmath331 of @xmath327 and a function @xmath332 such that @xmath333 as @xmath8 for all @xmath70 . since @xmath334 is a subsequence of @xmath331 ( for all @xmath47 ) , the diagonal sequences @xmath335 and @xmath336 , again denoted by @xmath5 and @xmath6 , respectively , satisfy condition .
_ step 3 .
_ let @xmath337 be an at most countable dense subset of @xmath4 .
note that @xmath337 contains all isolated ( @xmath37non - limit ) points of @xmath4 ( i.e. , points @xmath70 such that the intervals @xmath338 and @xmath339 lie in @xmath340 for some @xmath341 ) .
the set @xmath342 of discontinuity points of nondecreasing function @xmath343 is at most countable . setting @xmath344
, we find that @xmath345 is an at most countable dense subset of @xmath4 and @xmath346 since the set @xmath73 is precompact in @xmath1 for all @xmath70 , and @xmath347 is at most countable , we may assume ( applying the diagonal procedure again and passing to a subsequence of @xmath5 if necessary ) that , given @xmath348 , there is a point @xmath349 such that @xmath350 as @xmath8 . in this way , we obtain a function @xmath351
. _ step 4 .
_ now , we show that , for every @xmath352 , the sequence @xmath353 converges in @xmath1 . for this
, we prove that this sequence is cauchy in @xmath1 , i.e. , @xmath354 as @xmath215 .
let @xmath152 be arbitrarily fixed . by assumption , @xmath355 as @xmath10 , so we choose and fix a number @xmath356 such that @xmath357 by property , there is a number @xmath358 such that @xmath359 the definition of the set @xmath345 and imply that the point @xmath360 is a limit point for @xmath4 and , at the same time , a point of continuity of the function @xmath343 . by the density of @xmath345 in @xmath4 ,
there is @xmath361 such that @xmath362 it follows from that there is @xmath363 such that @xmath364 assuming that @xmath146 ( the arguments are similar if @xmath365 ) and applying lemma [ l : ess](a),(b ) , we get , for all @xmath366 : @xmath367 & \quad\,+|\nu_n(f_j , g_j;t_t^-)-\beta_n(t)|+|\beta_n(t)-\beta_n(s)|\\[2pt ] & \quad\,+|\beta_n(s)-\nu_n(f_j , g_j;t_s^-)|\\ & \le\frac{\nu_{n+1}(f_j , g_j;t_t^-)}{n+1}+{\varepsilon}+{\varepsilon}+{\varepsilon}\\ & \le\frac{\nu_{n+1}(f_j , g_j)}{n+1}+3{\varepsilon}\\ & \le\frac{\mu_{n+1}}{n+1}+\frac{\varepsilon}{n+1}+3{\varepsilon}\le5{\varepsilon}. \end{aligned}\ ] ] being convergent ( see condition ( b ) ) , sequences @xmath368 , @xmath369 and @xmath370 are cauchy in @xmath1 , and so , there is a number @xmath371 such that , for all @xmath372 , we have @xmath373 by virtue of inequality , we get @xmath374 putting @xmath375 and applying lemma [ l : pjin](e),(c),(b ) , we find @xmath376 & \le d(f_j(s),f_k(s))+|(f_j , g_j)(\{s , t\})|+|(g_j , g_k)(\{s , t\})|\\ & \quad\,+|(g_k , f_k)(\{s , t\})|\\[2pt ] & \le{\varepsilon}+5{\varepsilon}+2{\varepsilon}+5{\varepsilon}=13{\varepsilon}\quad\mbox{for all}\quad j , k\ge j_4 .
\end{aligned}\ ] ] since @xmath377 depends only on @xmath154 ( and @xmath360 ) , this proves that @xmath353 is a cauchy sequence in @xmath1 , which together with assumption ( a ) establishes its convergence in @xmath1 to an element denoted by @xmath378 . here and
at the end of step 3 , we have shown that @xmath379 is a pointwise limit on @xmath4 of a subsequence @xmath380 of the original sequence @xmath381 . since @xmath382 pointwise on @xmath4 as @xmath219
as well , we conclude from lemma [ l : ess](c ) that @xmath383 and so , @xmath384 , or @xmath385 .
this completes the proof of theorem [ t : first].@xmath386 \(a ) condition ( b ) in theorem [ t : first ] may be replaced by the following one : @xmath370 is a _ cauchy sequence _ in @xmath1 for every @xmath70 .
however , if @xmath15 is not complete , we may no longer infer the property @xmath80 , @xmath47 , of the pointwise limit @xmath21 ( as there may be no @xmath22 ) .
\(b ) condition is _ necessary _ for the uniformly convergent sequences @xmath5 and @xmath6 : in fact , if @xmath71 and @xmath144 on @xmath4 , and @xmath23 , then it follows from lemma [ l : ess](d ) that @xmath387 \(c ) condition is ` almost necessary ' in the following sense .
suppose @xmath0 is a measurable set with lebesgue measure @xmath388 , @xmath5 , @xmath389 are two sequences of measurable functions , which converge pointwise ( or almost everywhere ) on @xmath4 to functions @xmath31 , respectively , such that @xmath23 .
by egorov s theorem , given @xmath152 , there exists a measurable set @xmath390 such that @xmath391 , @xmath71 and @xmath144 on @xmath392 .
so , as in the previous remark ( b ) , we have @xmath393 [ ex : di ] ( a ) condition is _ not necessary _ for the pointwise convergence even if @xmath18 for all @xmath19 . to see this ,
let @xmath206 $ ] and @xmath266 with @xmath267 .
given @xmath19 , define @xmath394 by : @xmath395 if @xmath396 is integer , and @xmath397 otherwise , @xmath255 $ ] .
the pointwise precompact sequence @xmath66 consists of bounded regulated functions ( in fact , @xmath398 , and so , @xmath399;m)=\mbox{\rm r}(c)$ ] for all @xmath19 ) .
it converges pointwise on @xmath4 to the dirichlet - type function @xmath400 from .
note that @xmath401 , and so , @xmath402 .
since the usual jordan variation @xmath403 of @xmath13 on @xmath206 $ ] is equal to @xmath404 , we find @xmath405 \!2\cdot j ! & \mbox{if } & n\ge 2\cdot j ! , \end{array}\right.\quad\,\,n , j\in{\mathbb{n}}.\ ] ] thus , @xmath406 , i.e. , condition does not hold .
\(b ) under the assumptions of theorem [ t : first ] , condition does not in general imply @xmath407 . to see this ,
let @xmath408 be as in example ( a ) above , so that @xmath409 .
given @xmath315 , choose a collection @xmath410 with @xmath50 such that @xmath411 is rational and @xmath412 is irrational for all @xmath51 .
noting that , by virtue of , @xmath413 we get @xmath414 \(c ) the choice of an appropriate sequence @xmath6 is essential in theorem [ t : first ] .
let @xmath415 , @xmath416 be two sequences , which converge in @xmath1 to @xmath266 , respectively , @xmath267 .
define @xmath417\to m$ ] by @xmath418 , @xmath19 ( cf . ) .
clearly , @xmath5 converges uniformly on @xmath4 to @xmath400 ( so , @xmath5 is pointwise precompact on @xmath4 ) , and @xmath419 for all @xmath315 . since @xmath420
we find @xmath421 , condition is not satisfied , and theorem [ t : first ] is inapplicable with @xmath18 , @xmath19 .
on the other hand , set @xmath422 for all @xmath19 . given @xmath34 , we have , by virtue of , @xmath423 where @xmath424 as @xmath8 .
this implies the inequality @xmath425 , and so , condition is fulfilled .
in this section , we consider extensions of two selection theorems from @xcite and @xcite .
the other selection theorems from the references in the introduction were shown to be particular cases of @xcite@xcite ( see remark [ r:5 ] ) . by virtue of the inequality @xmath426 , instead of condition in theorem [ t : first ] we may assume that @xmath427 or @xmath428 , in which cases the resulting pointwise limit @xmath21 of a subsequence of @xmath5 satisfies the regularity condition of the form @xmath429 .
making use of the notion of @xmath154-variation ( section [ s : reg ] ) , we get the following [ t : ev ] given @xmath25 and a metric space @xmath15 , let @xmath66 be a pointwise precompact sequence of functions such that @xmath430 then , there is a subsequence @xmath431 of @xmath5 , which converges pointwise on @xmath4 to a regulated function @xmath432 . taking into account theorem [ t : first ] , it suffices to verify that condition implies @xmath433 , which is with @xmath434 for all @xmath435 . in fact , by , for every @xmath152 there is @xmath194 and @xmath436 such that @xmath437 for all @xmath438 .
definition yields the existence of such that @xmath439 and @xmath440 for all @xmath438 . by ( where we replace @xmath160 and @xmath22 by @xmath28 , and @xmath21by @xmath160 ) , @xmath441 for all @xmath438 and @xmath47 .
consequently , @xmath442 this implies that the left - hand side in this inequality tends to zero as : given @xmath443 , we set @xmath444 and choose a number @xmath445 such that @xmath446 , which yields @xmath447 for all @xmath202 .
[ r:4 ] ( a ) if @xmath235 in theorem [ t : ev ] , we may infer that @xmath232 does not exceed the limit superior from : in fact , it follows from ( * ? ? ?
* proposition 3.6 ) that @xmath448 for all @xmath152 .
\(b ) theorem [ t : ev ] extends theorem 3.8 from @xcite , which has been established for @xmath16 $ ] and @xmath235 under the assumption that @xmath449 for every @xmath152 .
the last assumption on the uniform boundedness of @xmath154-variations is more restrictive than condition as the following example shows .
[ ex : fn ] let @xmath415 and @xmath450 be two sequences from @xmath1 such that @xmath451 for all @xmath452 and , for some @xmath453 , @xmath454 and @xmath455 in @xmath1 as @xmath8 .
we set @xmath418 , @xmath19 , and @xmath456 for all @xmath457 $ ] .
the sequence @xmath66 converges uniformly on @xmath4 to the constant function @xmath21 : @xmath458 given @xmath152 , there is @xmath194 such that @xmath459 for all @xmath438 , and so , by , @xmath460 for all @xmath438 , which implies condition : @xmath461 on the other hand , if @xmath462 is fixed and @xmath463 , then gives @xmath464 , and so , @xmath465 .
now , we are going to present an extension of a helly - type selection theorem from ( * ? ? ?
* section 4 , theorem 1 ) and ( * ? ? ?
* theorem 2 ) .
let @xmath466\to[0,1]$ ] be a continuous , increasing and concave function such that @xmath467 , @xmath468 , and @xmath469 as @xmath470 ( e.g. , @xmath471 , @xmath472 with @xmath473 , or @xmath474 , see @xcite ) .
let @xmath16 $ ] be a closed interval in @xmath14 , @xmath475 .
we set @xmath476 , and if @xmath477 $ ] is a partition of @xmath4 ( i.e. , @xmath478 ) , we also set @xmath479 and @xmath480 , @xmath51 .
the _ joint @xmath481-variation _ of functions @xmath482}$ ] is defined by @xmath483\biggr\},\ ] ] where @xmath484 is given by . since @xmath485 is independent of a constant function @xmath179\to m$ ]
, the quantity @xmath486 is the _
korenblum @xmath481-variation _ of @xmath487}$ ] , introduced in @xcite and ( * ? ? ?
* section 5 ) for @xmath17 .
the following theorem is a generalization of theorem 2 from @xcite , established for real functions of bounded @xmath481-variation under the assumption that @xmath488 and based on the decomposition of any @xmath489}$ ] with @xmath490 into the difference of two real @xmath481-decreasing functions .
[ t : ko ] under the assumptions of theorem [ t : first ] , suppose that condition is replaced by the following one : @xmath491 .
then , there is a subsequence of @xmath5 , which converges pointwise on @xmath16 $ ] to a function @xmath176 such that @xmath492 . in order to show that is satisfied ,
let @xmath47 , @xmath493 $ ] with @xmath50 , and set @xmath494 and @xmath495 , @xmath496 . by the definition of @xmath497 and the concavity of function @xmath481 , we have @xmath498 & \qquad+\sum_{i=1}^{n-1}\!|(f_j , g_j)(i_i')|+|(f_j , g_j)(\{t_n , b\})|\\[-2pt ] & \le\!\biggl[\!\kappa\biggl(\!\frac{s_1\!-\!a}{|t|}\!\biggr ) \!+\!\sum_{i=1}^n\!\!\kappa\biggl(\frac{|i_i|}{|t|}\biggr ) \!+\!\sum_{i=1}^{n-1}\!\!\kappa\biggl(\frac{|i_i'|}{|t|}\biggr ) \!+\!\kappa\biggl(\!\frac{b\!-\!t_n}{|t|}\!\biggr)\!\biggr]v_\kappa(f_j , g_j)\\[-2pt ] & \le(2n\!+\!1)\kappa\biggl(\frac{1}{(2n\!+\!1)(b - a)}\biggl[(s_1\!-\!a ) + \sum_{i=1}^n(t_i\!-\!s_i)\\[-4pt ] & \qquad+\sum_{i=1}^{n-1}(s_{i+1}\!-\!t_i ) + ( b - t_n)\biggr]\biggr)v_\kappa(f_j , g_j)\\[-4pt ] & \le(2n\!+\!1)\kappa\biggl(\frac1{2n\!+\!1}\biggr)v_\kappa(f_j , g_j ) .
\end{aligned}\ ] ] thus , @xmath499 and so , condition in theorem [ t : first ] is satisfied .
let @xmath176 be the pointwise limit of a subsequence @xmath500 of @xmath5 .
arguing as in the proof of lemma [ l : ess](c ) , we get @xmath501 [ r:5 ] since theorem [ t : first ] is an extension of results from @xcite , it also contains as particular cases all pointwise selection principles based on various notions of generalized variation .
these principles may be further generalized in the spirit of theorem [ t : ko ] replacing the increment @xmath502 applied in @xcite@xcite by the joint increment @xmath503 from .
finally , it is worth mentioning that the joint modulus of variation , defined by means of , plays an important role in the extension of a result from @xcite to metric space valued functions : [ t : chma ] given @xmath25 and a metric space @xmath15 , let @xmath66 be a pointwise precompact sequence of functions such that @xmath504 then , there is a subsequence of @xmath5 , which converges pointwise on @xmath4 .
taking into account lemmas [ l : pjin ] and [ l : ess ] , the proof of this theorem follows the same lines as the ones given in ( * ? ? ?
* theorem 1 ) ( where @xmath111 is a metric semigroup and @xmath505 is defined via ) , and so , it is omitted .
note that the limit of a pointwise convergent subsequence of @xmath5 in theorem [ t : chma ] may be a non - regulated function . for more details , examples and relations with previously known ` regular ' and ` irregular ' versions of pointwise selection principles from @xcite we refer to (
* section 5.2 ) , @xcite .
in the context of functions of one real variable taking values in a metric space , we have presented a pointwise selection principle , which can be applied to arbitrary ( regulated and non - regulated ) sequences of functions .
it is based on notions of joint increment and joint modulus of variation , the latter being a nondecreasing sequence of pseudometrics on the appropriate product space .
our selection principle extends the classical helly theorem and contains as particular cases many selection theorems based on various notions of generalized variation of functions .
in contrast to previously established selection principles , the main assumption in our theorem on the limit superior is ` almost necessary ' , and it is shown by examples that in a certain sense this assumption is sharp .
the notion of conditionally regulated functions explains the limitations of previously known selection results .
belov , v.v .
chistyakov , a selection principle for mappings of bounded variation , j. math .
appl . 249 ( 2 ) ( 2000 ) 351366 ; and comments on this paper : j. math .
appl . 278 ( 1 ) ( 2003 ) 250251 .
chistyakov , mappings of bounded variation with values in a metric space : generalizations , in : pontryagin conference , 2 , nonsmooth analysis and optimization ( moscow , 1998 ) .
j. math .
( new york ) 100 ( 6 ) ( 2000 ) 27002715 .
chistyakov , c. maniscalco , yu.v .
tretyachenko , variants of a selection principle for sequences of regulated and non - regulated functions , in : l. de carli , k. kazarian and m. milman ( eds . ) , topics in classical analysis and applications in honor of daniel waterman , world scientific publishing , hackensack , nj , 2008 , pp . | given @xmath0 and a metric space @xmath1 , we introduce a nondecreasing sequence of pseudometrics @xmath2 on @xmath3 ( the set of all functions from @xmath4 into @xmath1 ) , called the joint modulus of variation .
we prove that if two sequences of functions @xmath5 and @xmath6 from @xmath3 are such that @xmath5 is pointwise precompact , @xmath6 is pointwise convergent , and the limit superior of @xmath7 as @xmath8 is @xmath9 as @xmath10 , then @xmath5 admits a pointwise convergent subsequence whose limit is a conditionally regulated function .
we illustrate the sharpness of this result by examples ( in particular , the assumption on the @xmath11 is necessary for uniformly convergent sequences @xmath5 and @xmath6 , and ` almost necessary ' when they converge pointwise ) and show that most of the known helly - type pointwise selection theorems are its particular cases .
joint modulus of variation , metric space , regulated function , pointwise convergence , selection principle , generalized variation _ msc2000 : _
26a45 , 28a20 , 54c35 , 54e50 |
studying the circumgalactic medium ( cgm ) is an emerging frontier of galaxy evolution that connects galaxies to the intergalactic medium . in this tenuous multi - phase gas that surrounds
galaxies lie clues to the processes that drive galaxy evolution such as gaseous inflows and outflows ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
recent observations with _
hubble s _ cosmic origin spectrograph ( cos ) have characterised the cgm in unprecedented detail .
they have shown that there is probably more oxygen in the halos around star - forming galaxies than within them @xcite , that the cgm may contain the missing baryons and missing metals within galaxy halos @xcite , and that star - forming and passive galaxies have different signatures @xcite but similar cool gas content @xcite .
recent work by @xcite and @xcite has also used cos to probe the cgm out to several hundred kiloparsecs , and observations by @xcite and @xcite have established a characteristic extent of the metal - enriched cgm to be roughly @xmath2 300 proper kpc , regardless of redshift .
in addition to these observations that characterise the cgm , there is a wealth of observational data detecting outflows at a variety of redshifts @xcite , indicating that most if not all galaxies drive winds at some point in their evolution , presumably enriching the cgm and igm with mass and metals in the process . the physical mechanisms that drive these outflows are not well characterised .
while observations can now probe individual sight lines around individual galaxies , there is not yet a robust understanding of the metallicity , covering factor , and thermodynamic phase of the outflowing material .
the specifics of how these quantities vary with the galaxy s mass , star formation rate , environment , and redshift are poorly characterised .
the fate of the ejected material after it leaves the galaxy is even less well constrained .
simulations suggest the presence of halo fountains " , where baryons are ejected from the galactic disk only to rain back down again at later times @xcite .
the contribution of this so - called recycled accretion relative to pristine inflows from the igm depends at minimum on galaxy mass @xcite and impact parameter @xcite , and plausibly also on star formation rate ( sfr ) , sfr surface density , and environment .
simulations by @xcite suggested that the igm is enriched
outside - in " , where winds propagate further from galaxies at earlier epochs and metals re - collect towards halos at late times , but this conclusion may depend on the poorly - understood mechanisms by which winds are launched .
hydrodynamic simulations that directly model inflows and outflows provide a crucial complement to emerging cgm observations by plausibly elucidating how the physical conditions and dynamical state of the cgm gas can be manifest in absorption line data .
@xcite showed that metal ions with low ionisation potential ( e.g. , ) trace dense gas close to galaxies , while metals with high ionisation potentials ( e.g. , ) trace more diffuse gas .
this work also showed how different prescriptions for outflows can result in significant differences in the temperature , density , and metallicity of the cgm gas .
@xcite followed and showed , in detail , how the baryon cycle operates in a plausible wind model . at low redshift , inflowing material is dominated by recycled gas that was previously ejected in an outflow ; hence inflows are enriched and can be traced by metal lines . in general ,
low metal ions like trace this recycled accretion , while traces hotter , less dense material built up by prior epochs of outflows . while these simulations have provided some general guidelines for interpreting cgm observations , they have yet to be tested in detail against existing cgm data .
the best existing absorption line data set that targets a range of impact parameters close to galaxies in the cgm is currently from the cos - halos survey @xcite .
this survey targets 39 quasars around 44 @xmath3 galaxies with a carefully - selected sampling of colour and stellar mass , and constitutes the first galaxy - selected absorption line survey of low - redshift cgm gas .
these galaxies have stellar masses ranging from @xmath4 , with spectroscopic redshifts @xmath5 , and are located @xmath6 kpc ( projected ) from a background uv - bright quasar . while cos - halos covers a wide range of ions , for this work we focus on 1216 , 1260 , 1335 , 1206 , 1394 , 977 , and , since they span a representative range of ionisation energies ( see table [ table : thresh ] ) .
typical cos - halos detection thresholds for the various ions are given in table [ table : thresh ] , and are generally around rest equivalent widths of 0.1 .
the cos - halos sight lines are supplemented by keck / hires echelle spectra for 35 quasars to include coverage for ( 2796 , 2803 ) .
these observations represent the largest sample of cgm absorption line data at these impact parameters , and hence provide a stringent test for models of cgm gas and their relationship to galaxies .
in this paper we build on our previous works using hydrodynamic simulations to ( i ) determine if our hydrodynamic simulations quantitatively reproduce the cos - halos absorption line observations ; ( ii ) test whether such observations can distinguish between plausible but substantively different models for galactic outflows ; and ( iii ) constrain the physical and dynamical state of cgm gas around @xmath7 galaxies as probed by cos - halos data . as observations improve with ongoing surveys such as cos - dwarfs @xcite , the cos - gto survey @xcite , and other efforts ( e.g. * ? ? ?
* ) to study the cgm of selected galaxy samples , this work aims to serve as a template for how cgm absorption line observations can constrain the physical processes driving the baryon cycle .
our paper is organised as follows : in 2 we describe our simulations , in 3 we compare to direct observables , in 4 we describe variations with wind model including variations in baryon cycling , and in 5 we summarise our conclusions .
for this paper we use our modified version of the n - body+entropy - conserving smooth particle hydrodynamic ( ec - sph ) code gadget-2 @xcite , fully described in @xcite .
our main simulation is identical to that used in @xcite , and further details can be found there .
briefly , we assume a wmap-9 concordant @xmath8cdm cosmology @xcite : @xmath9 , @xmath10 , @xmath11 , a primordial power spectrum index @xmath12 = 0.96 , an amplitude of the mass fluctuations scaled to @xmath13=0.82 , and @xmath14 .
the volume is 32@xmath15 mpc on a side with @xmath16 gas particles and @xmath16 dark matter particles .
this results in a gas particle mass of @xmath17 and a dark matter particle mass of @xmath18 , allowing us to resolve galaxies down to stellar masses approaching @xmath19 .
the main simulation output we use for this work is at @xmath20 , which is the closest snapshot we have to the typical redshift of cos - halos .
this simulation includes galactic outflows , which are implemented using a monte carlo approach and tied to the star formation rate ( sfr ) via @xmath21sfr , where @xmath22 is the mass loading factor . for this work we select two wind models to study in detail .
the first is our hybrid energy / momentum driven winds or ezw " model .
this model is our favoured wind model because it most accurately represents the relevant small - scale physics associated with the transition from momentum - driven winds to energy - driven winds at low masses @xcite , as well as reproducing observations of the stellar and mass function @xcite . in the ezw model ,
the wind speed @xmath23 and mass loading factor @xmath22 depend on the galaxy velocity dispersion @xmath24 : @xmath25 here , @xmath26=[1.05,2 ] is the luminosity in units of the eddington luminosity required to expel gas from a galaxy potential , @xmath27 , and @xmath24 is the galaxy s internal velocity dispersion , broadly constrained to match igm enrichment at high redshift @xcite .
particles are ejected individually , with velocities chosen to lie within the range determined by this range of @xmath26 .
the range of @xmath26 values is taken from observations of local starburst galaxies by @xcite , and is also consistent with observations by @xcite that find @xmath28 . in this model , as in its earlier cousin vzw which did not include the steeper @xmath22 scaling at low @xmath24 , we add an extra kick of @xmath29 to simulate continuous pumping of gas from radiation pressure , as described in @xcite . ; the one shown here is verified to be the one used in our code . ]
this simulation also includes a prescription for quenching massive galaxies , which involves super - heating particles with increasing probability above a threshold quenching stellar mass as described in @xcite .
our second wind model is a constant wind ( cw " ) model , where the mass loading factors and wind speeds are the same for each galaxy , regardless of mass . in the cw model , @xmath30 and @xmath31 .
the latter is similar to that used in the overwhelmingly large simulations ( owls ) reference model of @xcite .
the cw simulation used here has the same physics described in @xcite , including @xcite metal - line cooling , but now with a box size and resolution to match our main ezw simulation used here , namely 32@xmath15 mpc and @xmath32 particles .
this allows for consistency with the ezw " model above , and easily resolves galaxies similar in size to the smallest ones in the cos - halos survey .
the constant wind model does less well reproducing the observed measurements of at high redshift @xcite , the mass - metallicity relation @xcite , and the evolution of stellar growth in galaxies @xcite .
nonetheless , it viably reproduces the evolution of global star formation @xcite .
however , it has not yet been tested against the cos - halos survey at low - z .
we include this wind model because it serves as a useful contrast to the variable mass loading and wind speed wind model ezw . through this comparison
we aim to comment on not just _ whether _ winds are necessary to match cgm observations , but also _ what type _ of winds are necessary .
we can also gain insight into how baryons cycle in the different wind models , how they affect the gas around galaxies , and how such processes are manifested in cgm observations .
once we run our simulations , we use specexbin to calculate physical properties of the gas along lines of sight @xcite .
briefly , specexbin averages physical properties of the gas along a given sight line and then uses look - up tables calculated with cloudy @xcite to find the ionisation fraction for the relevant ionic species .
we use the same version of specexbin as in @xcite . see figure 1 of @xcite for an example of a simulated spectrum .
this version of specexbin includes a prescription for self - shielding from the ionisation background , which uses a density threshold of @xmath33 . above this density threshold
we assume that all the hydrogen is fully neutral , i.e. it is all , and that all the magnesium is in .
an hicolumn density above @xmath34 shields against most of the 15ev photons that could ionize to , but it does not shield the 7ev photons that ionize to . we choose a signal - to - noise ( s / n ) value of 30 , for comparison with our earlier work @xcite .
this is higher than the s / n value of the observations , which averages roughly 12 . however , since we will use the same equivalent width detection thresholds as observations , as listed in table [ table : thresh ] , our results are not very sensitive to our exact choice of s / n . once the model lines of sight are selected ( see the next section ) and the artificial spectra are generated using specexbin , then we use the voigt profile fitter autovp @xcite to identify the absorption lines ( see @xcite for an example ) and to fit column densities and equivalent widths .
we choose a large suite of ions to examine for this work to get the fullest possible information on the physical conditions of the cgm . as found in earlier work @xcite , a range of ionisation potentials
are necessary to probe different phases of the cgm .
generally , low ionisation potential metal lines probe high density , low temperature regions close to galaxies , while high ionisation potential metal lines probe warmer , more diffuse gas that can be detected further from galaxies .
in addition to 1216 , we generate spectra for 2796 , 1260 , 1335 , 1206 , 1394 , 977 , and 1032 .
these metal lines are all detected in cos - halos , and probe a range of ionisation energies ( see table [ table : thresh ] ) .
lccc ion & ionisation energy ( ev ) & detection threshold ( ) + + hi & 13.6 & 0.2 + & 15.04 & 0.1 + & 16.35 & 0.15 + & 24.38 & 0.08 + & 33.49 & 0.1 + & 45.14 & 0.1 + & 47.89 & 0.25 + & 138.1 & 0.1 + [ table : thresh ] we choose our lines of sight ( los ) in our simulations to best mimic the los in the cos - halos survey . in cos - halos , galaxies were selected first and the impact parameter to the galaxy is the projected distance to the observed quasar . as a result
, each cos - halos los corresponds to a specific impact parameter @xmath35 around a galaxy of a known stellar mass @xmath36 , as shown in figure [ los ] .
moreover , cos - halos galaxies were generally selected to be isolated systems ( preferring galaxies without photometric coincidences within 1 mpc , with some exceptions ) , and hence are likely to be central galaxies in their halos ( * ? ? ? * section 2.5 ) . to mimic this process in the simulations
, we select all the central galaxies in our simulations that are within @xmath370.125 dex in stellar mass of the cos - halos survey galaxies , and we generate four los at the particular impact parameter for the selected cos - halos galaxy , x+@xmath35 , x-@xmath35 , y+@xmath35 , and y-@xmath35 . in this way
we produce a suite of model los for each cos - halos los that matches its unique ( @xmath38 ) value .
for example , for the cos - halos los corresponding to @xmath39 and @xmath40 , we select all galaxies in our simulation with @xmath41 , and produce four los around each at @xmath42 kpc in @xmath43 and @xmath44 . in the end , for each of the 44 cos - halos ( @xmath38 ) los , we have between 50 - 200 simulated los .
the smaller mass galaxies tend to have larger numbers of los since there are more galaxies within @xmath370.125 dex in @xmath36 . for consistency with the cos - halos analysis
, we consider all absorption components within @xmath37 600 @xmath45 to be associated with the galaxy , and sum them for a total equivalent width , unless otherwise stated .
we note that this differs from our approach in earlier work @xcite where we only considered components within @xmath37 300 @xmath45 to be associated with a galaxy .
however , as shown in figures 8 and 15 of that work , the absorption drops off substantially after @xmath37 300 @xmath45 , so there is little difference between these two limits . this same drop - off in absorption after @xmath37 300 @xmath45 is seen in the cos - halos data themselves
@xcite .
we note that our simulation naturally produces many satellite galaxies around the chosen central galaxy , so in some cases our los will impact closer to satellites .
this will presumably contribute to the variance of absorption properties around a particular galaxy . in this paper
, we will focus on mean and median absorption statistics without examining the origin of the scatter , which we leave for future work .
we also subdivide our sample into passive and star - forming galaxies ( sfgs ) , using a specific star formation rate ( ssfr@xmath46sfr/@xmath36 ) criterion of @xmath47yr , as done for cos - halos ( see figure [ los ] ) .
figure [ sfr ] shows ssfr vs. @xmath36 for ezw ( blue crosses ) and cw ( red triangles ) models , only showing those galaxies selected to mimic the cos - halos los .
we show as the black line the median trend from a compilation of observations of star - forming galaxies by @xcite .
the observed cos - halos galaxies are generally actively star forming for the lower - mass galaxies , but are more likely to be passive with increasing stellar mass .
many of the massive galaxies in ezw have zero sfr , and those quenched galaxies are plotted at the bottom as downward facing arrows . of the 48 @xmath48 galaxies in the ezw sample ,
all but 9 are quenched , with ssfr=0 . unlike the ezw model
, cw does not include a quenching prescription , so produces no quenched galaxies at high masses .
the simulated galaxies broadly cover the range of ssfrs seen in cos - halos , albeit that our quenched galaxies have zero sfr by construction .
figure [ sfr ] only shows galaxies identified as central galaxies within their halos , but at low masses , many of the quenched galaxies shown here are likely former satellites that have been ejected from halos @xcite ; our quenching prescription is not applied to such small systems , but they have been quenched as a satellite likely by gas stripping processes ( e.g. * ? ? ?
such satellite quenching processes are self - consistently handled in the simulations , albeit the version of sph we use here does not necessarily handle them optimally ( e.g. * ? ? ?
* ; * ? ? ?
importantly , we note that the ezw model better matches the observed distribution of ssfr vs. @xmath36 at low masses .
this is one reason why we favour our ezw model relative to cw , in addition to those arguments discussed in
[ sec : code ] .
we also favour the ezw model because it more accurately encapsulates the scalings for outflows predicted from recent high - resolution individual galaxy simulations @xcite , and matches many other observations better than cw @xcite .
we now show a direct comparison between our simulations and the cos - halos survey .
this work represents the first direct comparison of our simulations against observations of absorption lines in targeted los at low redshift .
each of the following subsections represents a different way of comparing models to observations ; all must agree before one can claim agreement .
we begin by focusing on the strength of _ detected _ lines , to see if the models match the observed sample .
then we examine the total absorption along all los , as well as the incidence rate of all our ions along all los .
we then provide a preliminary glimpse into how ion ratios can constrain models , and examine the kinematics of absorbers relative to their host galaxies .
matches and mismatches between all these statistics provide insights into exactly how our simulations are succeeding and failing in reproducing the cos - halos data , and they points toward required improvements in cgm modelling .
the most direct absorption line observable is the equivalent width .
figure [ ew ] shows the rest equivalent width versus impact parameter for the cos - halos survey compared to our simulations , for our set of probed ions . for this figure we focus on _ detected _
lines only , and ask if models reproduce the same absorption strength as the detected observations . note that the detection thresholds for each ion are given in table [ table : thresh ] , and shown as solid black horizontal lines to guide the eye .
we note that for the observations , the detection limits quoted are a rough guide , as they can vary vary for individual los . for the larger sample of los in the models ,
however , we use the detection limits quoted in table [ table : thresh ] for all model los .
observed points below the detection limits are included in this figure for completeness , however in this work we only compare observed and model points _ above the detection limits_. here we have placed the ions in order of increasing ionisation potential from low to high .
the black points are from the cos - halos survey .
the circles represent detected values while the up and down facing arrows are lower and upper limits , respectively .
error bars are included in black for detected values only , errors on limits are not included . in many cases the errors are smaller than the size of the symbols .
the blue crosses and red triangles show the results for the ezw and cw models , respectively .
these points represent the median equivalent width of all los _ with detections above the threshold _ listed in table [ table : thresh ] at a given impact parameter .
the bars on the model points span the 16 - 84% range of ew values at that impact parameter , to illustrate the scatter in ew at a given impact parameter . to avoid clutter
, we only show the model range bars for points @xmath0 every 25 kpc .
the impact of non - detections will be better illustrated by the statistics shown in figures [ dedz ] and [ cf ] .
both models and observations show a weak trend of decreasing equivalent widths with impact parameter . as discussed in @xcite , the basic driver of this
is that gas is denser near galaxies .
however , the trend is weaker for higher ions , and is essentially absent for , because the density gradient is counteracted by the tendency for high ions to have increasing ionisation fractions in lower - density gas . in @xcite
we found that the ion strength dropped with both halo mass ( and thus stellar mass ) and impact parameter .
but it is important to note that figure [ ew ] conflates these two trends , owing to the way that we have done our los selection ( [ sec : los ] ) , by matching the observed galaxy sample in _ both _ stellar mass and impact parameter . as shown in figure [ los ]
, los at either low or high impact parameters can be from either low or high mass galaxies .
as such , the model points do not decrease as strongly or smoothly with impact parameter , as was seen in @xcite where we selected narrow bins in halo mass .
additionally , this degeneracy washes away some trends with ionisation potential : the equivalent widths of the low ionisation species do not drop off as steeply as found in @xcite .
this will similarly impact figure [ dedz ] .
the scatter owing to the conflation of these trends is exacerbated for the observed points , since the simulation points are averages from a larger sample . for
, both simulations are in general agreement with the observations across the 0 - 150 kpc range , showing strong absorption in every line of sight .
the typical ew is much higher than seen in random lines of sight through the igm @xcite , showing that there is substantial cool gas within these halos .
there is little sensitivity to winds , at least as can be discriminated from the cos - halos survey , which is reminiscent of the lack of sensitivity to winds in the ly@xmath49 forest @xcite .
there are , however , a few very strong absorbers with equivalent widths above 2 that lie above all the simulation median points .
these potentially provide an interesting challenge to models , hinting at a larger covering fraction of very strong than predicted .
however , current uncertainties prevent any robust constraints emerging from these few systems .
the metal lines provide a more direct test of pollution from galactic outflows . for the low and mid ions , when detections are made , they are generally in the ew range expected from the models , with a few observed points being much higher than the model points ( as with hi ) .
is also commonly detected , but the typical equivalent widths are generally @xmath50 higher than predicted ; we discuss this below .
nonetheless , the broad agreement is a non - trivial success of these galactic outflow models in enriching the cgm to roughly the observed levels .
though we do not show it , a model without winds dramatically fails this basic test since it does not yield appreciable metal absorption beyond the smallest impact parameters , as discussed in @xcite . for the low ions ( , , , and )
, the model points tend to lie generally in the middle of the observed points .
in many cases the scatter in the observed points is larger than the 16 - 84% range bars that the models would predict .
this could reflect variations in the ionisation conditions or metallicities that are not properly captured within our simulations . recall that we assume a spatially - uniform metagalactic ionising background with no local contribution from the nearby galaxy and a simple density criterion for self - shielding .
these assumptions may need to be relaxed if we wish to capture detailed variations properly
. doing so would require full radiative transfer simulations with much higher resolution to capture small , dense condensations . as discussed in @xcite ,
if such clouds are indeed the dominant contribution to low - ionisation cgm absorption , they are far from being resolved in any current cosmologically - based simulation .
the mid and high ions tend to show a smaller observed scatter than the low ions , more comparable to that seen in the simulations .
since low ions tend to arise closer to galaxies and in denser gas @xcite , this suggests that locally - varying density and ionisation conditions , and not locally - varying metallicity , is responsible for the larger scatter in the low ions .
alternatively , local variations in metallicity could be much stronger closer to galaxies , which may be reasonable if the absorbing gas arises from recent outflows .
returning to the discrepancy for , this is tricky to interpret because can arise in both photo - ionised and collisionally - ionised gas @xcite . as shown in (
* figure 6 ) , the median temperature for absorption is @xmath51k , which is dominated by photo - ionisation , and indeed the phase space diagram in their figure 5 does not suggest a distinct peak of collisionally - ionised absorption at @xmath52k .
one way to explain the predicted deficit is thus to invoke a more significant hot halo of gas .
however , since the typical virial temperature of a cos - halos galaxy is a few times @xmath53k , this means that such collisionally - ionised would have to be a transient phase through which the gas is cooling .
modelling such non - equilibrium processes is a substantial challenge for simulations , and may need to be improved if more robust and accurate predictions are desired . alternatively , a significantly higher photo - ionising background around 114 ev than what we have assumed could also result in substantially more ; the metagalactic flux at these energies is currently poorly known .
the details of the wind implementation can strongly impact the phase of cgm gas .
for instance , the models of @xcite evolved using the adaptive mesh refinement hydrodynamic code enzo show substantially hotter cgm gas , since their models drive winds by super - heating the ism .
they tended to find that the resulting cgm was too hot , and thus also a poor match to observations , at least for the most plausible assumptions about the energetic input into winds .
this highlights that the proper treatment of wind propagation , heating , and mixing is a challenging problem in current simulations , and shows that detailed comparisons to observations can provide interesting constraints on these processes . in light of this , it is somewhat surprising that the two wind models we consider here are in such good agreement with each other , and that these cos - halos observations can not discriminate between them .
@xcite showed large differences in absorption line profiles in every ion between a cw and a momentum - driven wind model that is similar to ezw .
moreover , ezw and cw are well - discriminated by other diagnostics generally related to galaxy properties @xcite .
the differences between wind models will be slightly more prominent in other cgm statistics that we explore later , but for the equivalent widths they are small .
we will discuss potential reasons for this in 4 .
in the previous section we compared individual absorber equivalent widths between the models and the observations where detections occurred , and found broad agreement in some ions but notable differences in others . that comparison , however , does not characterise the _ total _ absorption strength ; for instance , it may be that absorbers are far more or less common in the models compared to the observations , so that while individual equivalent widths are comparable when detected , the overall absorption is discrepant . to examine the overall absorption , here we present equivalent widths per unit redshift close to galaxies , including all los , both detections and non - detections .
figure [ dedz ] shows the equivalent width per unit redshift , dew / dz , versus impact parameter .
we calculate dew / dz by summing the equivalent width ( above the detection limits given in table [ table : thresh ] ) in all los probed in that impact parameter bin , and dividing by the total redshift path length within @xmath54 km / s , for all los with that impact parameter .
we bin both observed and model points in 25 kpc bins .
the observed cos - halos points are shown in black .
for the purposes of this plot , observed points that are upper or lower limits are assumed to be at their limit values ; the results do not change significantly if all upper limits were assumed to be zero .
the model values are plotted as dotted blue lines for the ezw model and as dashed red lines for the cw model .
the bars on the model points are range bars , showing 16%-84% of the model values for all los in that impact parameter bin ( and is likely an overestimate of the actual range at any given impact parameter in that bin ) .
the errors on the observed points represent the average errors on all constrained detections in the 25 kpc bins .
these errors do not include errors on los that are upper limits , lower limits , or non - detections ; hence the errors shown here are a lower limit on the actual error . the errors on the model points have been slightly offset horizontally for ease of viewing .
dew / dz serves to diagnose whether our simulations are producing the observed amount of and metal ion absorption in the cgm around these galaxies .
the absorption is in excellent agreement at all impact parameters . for comparison ,
a random los through the universe would typically yield dew / [email protected] , which is much lower than the cgm typical value of several hundreds .
hence the agreement with is non - trivial , and shows that there is both the correct amount and the correct radial distribution of neutral gas in the simulations . turning to the metals , and are well - reproduced in the model , with the observed points being within the @xmath55 range of the simulation predictions .
the models produce drastically too little and ; the models are off by as much as an order or magnitude .
conversely , the models produce too much at low impact parameters , even outside the model ranges .
the simplest interpretation is that the models do not have as much cold , dense gas in the cgm as observed , while having an over - abundance of warmer gas .
a physical interpretation might be that our assumption of optically - thin gas fails to capture the dense gas phase structure in the cgm .
qualitatively , self - shielding would be expected to increase and reduce .
however , we note that substantial self - shielding would greatly increase the amount of and , which at present are in good agreement with the observations .
clearly there is some disagreement in the phase structure predicted in the models compared to that observed , but the nature of this remains difficult to ascertain .
small features in the cgm which may give rise to absorption , such as interface regions and small , possible multi - phase substructures , may be below the resolution limit of the models presented here . for the mid - ion ,
the models produce slightly too little absorption but generally the observed points are within the model range values . for , however , the disagreement is much more stark . while the difference in the _ strength _ of _ detected _ lines ( as shown in figure [ ew ] ) between the models and the observations is roughly a factor of 2 , the difference in the _ total _ absorption over _ all _ los is more like a factor of 5 .
this difficulty in producing enough to match the @xcite observations was also found in the enzo simulations of @xcite .
the agreement between the models and the observations for , , , and to a lesser extent is generally quite good given that the models have not been tuned in any way to reproduce these data , and suggests that the amount of cgm enrichment in both these simulations is roughly in accord with observations .
however , the mismatch in the silicon ions , as well as , is quite large .
this is a significant failure , given both the degree of the mismatch , and the high quality of the observed and data .
the similarity in the predictions of the two wind models is striking , particularly since the amount of cgm metals and the phase space distribution of the gas in these two models is rather different @xcite , as we will discuss later .
the trend in is quite flat with impact parameter .
as we discussed in the previous section , this partly owes to the fact that a given impact parameter bin has a substantial range of stellar masses .
some of this likely owes to saturation , as at these equivalent widths are very much on the linear part of the curve of growth and hence fairly insensitive to variations in column density .
nonetheless , the trend in with @xmath35 is even much flatter than in any of the low - ionisation metals . at face value
, this suggests that the dropoff in metal ions at large radii owes either to gas becoming more ionised at higher impact parameter or to a dropoff in metallicity , rather than a dropoff in the amount of neutral gas . in summary ,
the predicted total equivalent width per unit redshift shows broad agreement with the observations , but with various interesting discrepancies .
, , , and are in fair agreement with the observations , and , , and are significantly underproduced .
it is difficult to identify a single physical variation that could even qualitatively reconcile all these discrepancies while preserving the good agreements , suggesting that further work is needed to discern their detailed cause .
radial dependencies are stronger for lower ions , except for that shows almost no decline with radius possibly owing to saturation . the covering fraction quantifies how patchy the cgm absorption is in a particular ion .
observations and simulations suggest that the cgm has a complex , multi - phase structure , potentially with small clouds embedded in warmer gas @xcite .
the size and fraction of material in these various phases remains poorly constrained .
@xcite , based on observed absorption around galaxies , finds that this absorption generally arises in small clouds , i.e. @xmath56 @xmath57 , line - of - sight sizes @xmath58 pc , and masses between 10 and 1000 @xmath59 .
in contrast , @xcite , from low - ionisation absorbers in cos - halos , find a median cloud mass of @xmath60 .
covering fractions thus provide another test of models , and potentially a very stringent one if the cloud sizes are well below what can be resolved in our ( or most ) simulations @xcite .
figure [ cf ] shows the covering fraction as a function of impact parameter for our set of ions . as in figure
[ ew ] , we plot ions from low to high ionisation potential .
we define a covering fraction as the fraction of los with a detection above a given ion - specific ew threshold , corresponding approximately to the cos - halos survey detection limits , as listed in table [ table : thresh ] .
the observed points are binned from 0 - 50 kpc,50 - 100 kpc , and 100 - 160 kpc .
the solid black line represents the median covering fraction , and the dotted black lines represent the uncertainties ( wilson intervals ) .
the models are plotted as individual points ( crosses for ezw , triangles for cw , as before ) , because in the simulations we have multiple los at each impact parameter and are able to define a unique covering fraction at each impact parameter . in this manner
one can get a sense of the variability in covering fraction as a function of impact parameter , albeit with the same degeneracy in stellar mass noted earlier . for , the simulations almost always have unity covering fraction above its detection threshold of 0.2 .
the covering fraction in the cos - halos survey is also very close to unity , but formally slightly below .
this high fraction was also emphasised in @xcite , and is independent of both galaxy stellar mass and galactic star formation rate ( not shown here ) , for the range of these values covered by the cos - halos survey .
in fact for all ions shown here , the covering fraction depends more strongly on the impact parameter than either the star formation rate or the stellar mass . the high covering fraction of
is noted in the cos - halos survey for both star - forming and passive galaxies @xcite .
we note that we are able to reproduce these high hi covering fractions in passive systems with our heuristic quenching prescription ( present in the ezw model but not the cw model ) . further exploring figure [ cf ]
, we see that for low ions ( , , , and ) the models generally are either at , or slightly above , the observations at impact parameters of 0 - 50 kpc but drop below the observations at larger impact parameters . for the mid ion ,
the model values are below the observed values in the 0 - 50 kpc bin , but within the errors . at higher impact parameters the model covering fractions drop below the observed errors .
for , there are model points below the observed errors at all impact parameters . for the high ion however ,
the model covering fractions are already too low at all impact parameters , with the match worsening at higher impact parameters .
we note that the covering fraction of the high ion does not decline as steeply with impact parameter as the lower ionisation species , either in the observations or in the models .
we also emphasise the similarity of the ezw and cw model points . while there is scatter in the model points ,
there is not a single impact parameter bin in a single ion where one model is consistently different from the other .
rather , both models match ( or mismatch ) the observations in similar ways .
the mismatch between the models and the observations has three possible origins : there could be not enough metals in the simulation , the metals could be in other ionisation states not shown here , or the metals could be in the wrong spatial distribution , which could happen if the metals were not mixed properly .
we note that in both figure [ dedz ] and [ cf ] the radial gradients of , , , and are steeper in the models than in the observations .
this suggests the models do not produce enough metals in the outskirts of halos to match the cos - halo observations .
it is also possible that some of the metals are not in the correct ionisation states : note our overproduction of at small impact parameter but an underprediction of and .
we also slightly overpredict but underpredict at low impact parameters , although the discrepancies are generally within the error bars of the observations .
it is likely a combination of these factors that account for the mismatch , as we discuss in further detail in
[ sec : ionratios ] and [ sec : baryonfractions ] . in our simulations , we do not mix metals into cgm gas from the particles that are ejected from the ism , and hence it is perhaps not surprising that there is too little metal mixing in our models .
it is not obvious how much mixing should occur ; again , it is challenging to model mixing of outflowing gas properly . nonetheless , it is interesting that the covering fraction can potentially provide a good constraint on this crucial physical process affecting how the outflows and the cgm interact .
a key feature of cos - halos is that it includes a significant sample of passive galaxies , and hence can examine trends of cgm absorption with star formation rate .
current models broadly suggest that passive galaxies are surrounded by hot gaseous halos ( e.g. * ? ? ?
* ) , unlike star - forming galaxies that are located in
cold mode " halos ( e.g. * ? ? ?
* ) . hence one might expect trends in the strength of various ions as a function of specific star formation rate .
we examine the dependencies on ssfr by investigating equivalent widths versus impact parameter , separated by ssfr .
as noted earlier , @xcite found that blue galaxies generally have higher equivalent widths of than red galaxies , regardless of impact parameter or galaxy mass .
in contrast , @xcite found that the hi properties in the cgm of red and blue galaxies are only marginally different , and @xcite similarly found that low - ionisation absorption is also not greatly dissimilar .
we use our ezw model , which includes our heuristic quenching prescription , to examine how well our simulation reproduces these observations . in figure [ redblue ]
we divide our model los which are initially selected based only on impact parameter and stellar mass into los around high- and low - ssfr galaxies ( above and below @xmath61yr ) .
to focus the discussion , we examine only hi , our lowest metal ion , and our highest ion ; the trends for intermediate ions are interpolatable from these .
we are able to reproduce the observed high hi covering fractions in passive systems @xcite despite our heuristic quenching prescription .
in fact there is little difference in the equivalent width of in model red and blue galaxies ( top panel ) , which is consistent with what is seen in the observations ( blue stars and red crosses ) .
hence our quenching model does not remove hi from halo gas , even though it adds substantial energy to the halo by heating the ism gas to temperatures that are well in excess of the virial temperature .
we are not claiming that this lends physical validity to our simple quenching model , but rather that such observations could provide interesting constraints on more physically - motivated models of quenching ( e.g. * ? ? ? * ) .
a similar trend is seen in ( middle panel ) , where our simulated red and blue galaxies have similar absorption strengths . in the observations , if anything there seems to be more around red galaxies .
our results suggest that whatever quenches the central galaxy in more massive halos should have a minimal effect on the cool cgm gas that gives rise to hi and . for
, the situation is different .
the observed strength around blue galaxies are higher than around red galaxies @xcite , three of which are upper limits off the bottom of the plot .
in contrast , this difference is not present in the simulations : there is no clear difference between the equivalent widths of in red and blue galaxies .
hence , the underprediction of in the models seen earlier is specific to blue galaxies , while the models greatly overpredict for passive galaxies .
this shows that , despite ejecting gas at high temperatures from the ism , our quenching model probably does not properly heat the surrounding cgm gas to prevent cooling through the temperature regime .
our implemented quenching model does not prevent accretion by keeping surrounding gas hot , but only ejects the gas once it gets in the galaxy .
@xcite argued that keeping the hot halo gas near the virial temperature , well above where it would be in the phase , provides a successful model to produce the quenched galaxy population as observed .
implementing such a model into our simulations would likely have quite a different impact on absorption , as it would explicitly prevent gas from cooling .
it remains to be seen whether such a model would concurrently over - suppress and low ion metal absorption ; we leave this investigation for future work . for now , we conclude that our implemented form of quenching is inconsistent with cgm metal observations . clearly , matching both the low ions and in the cgm of passive galaxies
represents an important constraint on the thermodynamics of quenching .
although passive galaxies in our simulations are not currently ejecting winds , we show in 4.3 that the cgm absorption in our models is dominated by gas that was ejected well before the epoch of observation , largely decoupling the cgm predictions from the current star formation rate .
ion ratios provide a complementary test for the physical conditions of the absorbing gas .
ratios between metal ions and hydrogen provide an estimate of the metallicity in the gas traced by that ion , while ratios between ions of the same element provide a constraint on the ionisation conditions of the gas .
given our current analysis , we can only study ratios of ion equivalent widths summed over 600 km / s intervals , which likely erases some of the detailed information about physical conditions through the cgm ; trying to subdivide these into components introduces great sensitivity to the exact noise level and fitting procedure , which is beyond the scope of this work to consider .
nonetheless , we will see that summed ion ratios still provide interesting constraints on cgm properties in our simulations .
selected ion ratios are shown in figure [ ionratio ] .
as before , cos - halos survey data is plotted in black , and model points ( ezw : blue crosses , cw : red triangles ) represent the median value of the ratio at each impact parameter . for the observed points , downward facing arrows
indicate that the ratio is an upper limit ( meaning an upper limit divided by an actual value , or an actual value divided by a lower limit ) , upward facing arrows indicate the ratio is a lower limit , and circles indicate an actual value .
we do not plot values that are poorly constrained , such as a limit divided by a limit .
for the model points , coloured downward arrows show where the ratio is zero , as little or none of the top ion is present . in the left panels , we plot the log of the ratio of equivalent widths of a low and high ion relative to , as a probe of the metallicity of the gas traced by that range of ionisation potentials . in the right panels ,
we show the ratio of two si lines , as well as a ratio of a low ( ) to high ( ) line .
the low ion ratio ( / ) shows generally good agreement with the observations , albeit with a scatter in the observations that is much larger than in the models . the scatter might arise owing to local variations in ionisation conditions .
there is a slight hint that the models produce slightly too high a ratio , particularly at large impact parameter , indicating that the metallicity in the low ionisation gas might be a bit too high .
there is also a non - trivial offset between the wind models , ezw is slightly higher than cw , although the scatter in the observations is far too large to discriminate between the models .
still , this is one of the few aspects in which any statistically significant difference can be seen between the ezw and the cw models .
the predicted / is considerably lower than most of the observed points .
the discrepancy is comparable to that seen in the individual ew measurements of vs. impact parameter in figure [ ew ] .
these discrepancies suggest that the hot diffuse phase in our models is either insufficiently enriched with metals or has an incorrect temperature structure that shifts oxygen to higher or lower ionisation states .
these trends hold true for both wind models ( though ezw produces slightly higher ratios than cw ) , indicating that it is not specific to which wind model we use , but rather may be indicative of an overall failure in the way we implement winds , namely by ejecting cool , unmixed ism particles . ejecting hotter or over - enriched gas that mixes
would qualitatively be expected to move towards better agreement with the observations .
nonetheless , we emphasise that the results are within a factor of @xmath0 3 of the observed ratios , which is already a non - trivial success . these ion ratio comparisons
further refine and highlight issues with our models , and demonstrate the ability of the cos - halos survey to provide important insights into galactic outflow and ionisation processes in the cgm .
the right panels of figure [ ionratio ] plot different metal ions against each other , mitigating the dependence on metallicity and allowing us to test physical conditions of the gas more directly
. however , the observational data are more limited , as it is not common to find absorbers with multiple ions detected .
the ratio of / at low impact parameters shows an overall trend of increasing ratio with impact parameter , and both models and are in good agreement with the observations from @xmath0 0 - 50 kpc . beyond 100 kpc ,
the ezw model has a higher ratio than the observations , indicating cooler gas than observed .
meanwhile , the cw model shows a very low / ratios , typically having essentially no and hence dropping off this plot at these large impact parameters ( shown as coloured downward arrows ) . the scatter is less than in the ratios versus , suggesting that the ionisation conditions giving rise to particular ions are more uniform in the cgm .
the / ratio shows a large scatter in the observations , indicating that these two ions arise in different phases that do not trace each other well .
there is , perhaps surprisingly , no clear trend with impact parameter , as one might expect for a simple model in which the low ion arises only near the central galaxy and the high ion arises throughout the halo @xcite .
this suggests that the is occurring throughout the halo , perhaps in satellite galaxies or cold dense clouds interspersed with warmer diffuse gas .
meanwhile , the models show a rising trend at small impact parameter that flattens at larger impact parameter .
the median ratio does not show a large scatter , suggesting that and are appearing together more often in these models than in observations . in summary , by examining ion ratios we can isolate trends with metallicity and ionisation state in the cgm as traced by various ions . while there is broad agreement between the models and the observations for and some low ionisation metal species , there is also significant disagreement for higher potential ions , particularly and .
clear trends emerge that suggest that the metallicity within certain phases may be too high ( for warm diffuse gas traced by mid ions ) or too low ( for hotter diffuse gas traced by high ions ) .
ion ratios generally show moderate agreement with the observations , albeit with smaller scatter and perhaps some discrepancies at large impact parameter ( @xmath62 kpc ) where the low ions disappear in the observations .
the differences between our two wind models are small compared to the differences between the models and the observations , which may owe to the similarly kinetic and cool ejection of material .
nonetheless , the differences between wind models seen here are not seen in most of the other statistics we have considered so far .
ion ratios thus provide new and interesting challenges to models of cgm gas , and obtaining a larger and higher - s / n sample could potentially provide the best constraints on the physical conditions in the cgm . with spectroscopic redshifts for all its galaxies ,
the cos - halos survey provides complementary information on the kinematics of absorbing gas relative to the systemic velocity of the host galaxy . in principle
, kinematics could distinguish outflowing gas from inflowing , but in practice this division is not so clean @xcite .
nonetheless , the kinematics provide an additional constraint on cgm models . in this analysis
we will no longer sum absorption within @xmath37600 km / s of the galaxy , but instead consider individual components to elucidate the kinematics .
we note that this introduces additional sensitivity to how our profile fitter decomposes the absorption into components , but here we really focus only on kinematics and not line strengths , hence such a sensitivity should not greatly impact our results . in figure [ velblock ] we quantify from where in velocity space the absorption originates in the observed data , and how this compares to the model predictions .
we focus on the zw model ; the cw model gives very similar results .
to do so , we lay out a coloured chequerboard " , where we bin our model data into 25 kpc spatial bins and 100 @xmath45 velocity bins .
for each of those bins , we calculate the total column density for all los with impact parameters in that bin , for all absorption within @xmath37 600 @xmath45 .
we shade each block by the fraction of the column density in that impact parameter range that comes from a given velocity interval .
dark blue blocks indicate that more than 30% of the total column between @xmath37 600 @xmath45 comes from that given velocity interval at that impact parameter , light blue blocks indicate 20 - 30% , green 10 - 20% , and orange that less than 10% comes from that velocity interval .
we note that these percentages are for that specific impact parameter bin only , and not for the entire 0 - 150 kpc range , so each vertical column of shaded blocks is independent .
we also plot the location of each individual observed absorber ( not summed over velocity , as elsewhere in this work ) as open circles whose sizes indicate the strength of the absorption . for the metals ,
the large circles indicate @xmath63 , the medium circles @xmath64 , and the small circles @xmath65 . for , the scale is shifted up by one in the @xmath66 .
generally , all the observed absorbers lie in the range where the models predict a significant fraction of the absorption should lie .
the observed absorption also clusters strongly around the systemic velocities of the host galaxies , which is in line with the model predictions .
the exception to this is that at very low impact parameters there is an excess of absorbers with positive velocities , and for and that absorption is stronger than what is expected from the models .
this owes to a single sightline that happens to have strong absorption in multiple ions and lies at a positive velocity with respect to its host galaxy .
overall , both the simulations and the observations suggest that most of the absorption occurs roughly within the virial velocity of the galaxy s halo .
only absorbers with the smallest impact parameters show any sign of an outflow , and it is difficult to assess its significance because it only occurs in one system ( with several absorbers within it ) , and there could be systematic uncertainties identifying the kinematic centre of any given system .
otherwise , the simulations broadly match the observed kinematics , and do not indicate an abundance of strongly outflowing gas at the present epoch , in general agreement with the outside - in enrichment scenario of @xcite .
in general , we have found rather modest differences in observational tracers of absorption as probed by cos - halos between our two wind models .
both models give similar predictions for the equivalent width of detected lines , total equivalent width , and covering fraction of and some of the low ionisation metal lines , which broadly match data .
both models fail , most notably , to reproduce the total equivalent widths of and , and do so in similar ways . while we favour the ezw model for independent reasons , in the cgm observations presented here it typically does no better or worse than the constant wind model in matching the observations .
this is somewhat curious , since earlier work ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) showed significant observable differences between these wind models in the physical conditions and enrichment of cgm gas .
we examine the reasons and implications of this surprising agreement here .
a key difference relative to our previous works is that here we are obtaining a galaxy sample that is matched in stellar mass , not halo mass . in our previous work we chose to match ezw and cw in halo mass since we reasoned that the cgm is more tied to the halo , e.g. the chance of metals escaping the cgm is tied to the full halo potential not just that of the galaxy , and the total amount of cgm baryons should scale more with the halo than with the galaxy .
however , cos - halos has chosen galaxies based on their stellar masses , and hence in this observational comparison we have focused on matching this quantity . as we argue here , it turns out that the better than expected " agreement between the two models presented here owes primarily to this choice , which has interesting implications .
figure [ hmsm ] shows the stellar mass halo mass ( smhm ) relation for our two wind models , with a running median for each ( blue dotted for ezw , red dashed for cw ) .
we include galaxies down to our stellar mass resolution limit of @xmath67 ( 32 gas particle masses ) .
solid black lines delineate the stellar mass range of cos - halos .
the broken coloured lines show results from abundance matching analyses for the closest available redshift : @xcite as the green dash - dot line , and @xcite as the purple dash - dot - dot line .
the ezw model is clearly a better match .
this simply reflects the fact that ezw ( or its cousin , vzw that does not have the energy - driven scalings at low masses or quenching prescription at high masses ) provides a better match to the galaxy stellar mass function than cw @xcite .
the stellar mass range relevant to this work is @xmath68 ( see figure [ hmsm ] ) , which from @xcite corresponds to a halo mass range of @xmath69 . in much of this range ,
the cw model s smhm relation is lower than ezw s .
this mainly occurs because the cw model ejects material at higher velocities in this mass range , which reduces the amount of wind recycling that dominates the inflow at these masses @xcite . in the present work
, we compare the wind models at fixed stellar mass .
hence for a given @xmath36 , the cw model will predict a higher @xmath70 . we can use this to qualitatively reconcile why @xcite found large differences between the two wind models .
for example , figure 9 of @xcite shows that for a momentum - driven wind model ( similar to the ezw model used here ) , higher mass halos have _ higher _ column density and hence ews at a fixed impact parameter .
figure 14 of that same work shows that for a fixed halo mass , the constant wind model has a _ lower _ column density at fixed impact parameter .
therefore , at fixed stellar mass , given that the constant wind model has higher halo masses than ezw , this would result in a higher column density at fixed impact parameter .
the differences in absorption strength owing to the wind model are thus offset by variations owing to halo mass . as a result , for the observables presented above
, the two wind models match much better at fixed stellar mass than at fixed halo mass .
this then propagates into all the various other statistics , such as covering fractions and dew / dz .
the interesting physical implication of this is that our models predict the cgm metal absorption properties are more tied to the stellar mass than to the halo mass . on the one hand , this is surprising since the cgm represents halo gas , which one might think would be more closely tied to the halo mass .
that being said , the metals in the cgm are formed by the stars in the galaxies and then ejected , so the closer relationship is perhaps not unexpected . in @xcite and @xcite
it was argued that the metals retained within the ism , as traced by the mass - metallicity relation , are strongly governed by outflows , whose properties must be tied most closely to stellar mass in order to give rise to a tight stellar mass - metallicity relation ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
evidently , in these simulations , the metals deposited into the cgm are likewise strongly governed by outflows that are more closely tied to the stellar mass .
we emphasise that this similarity in predictions for ezw and cw does _ not _ imply that current cgm observations are unable to rule out any outflow models . while we have varied the free parameters in our particular implementation of outflows ,
there are many other implementations of outflows in simulations that may result in quite different cgm properties .
these include models that over pressurise the ism by local supernova heating @xcite , drive winds through a combination of radiation pressure and supernovae @xcite , and include early stellar feedback driven by stellar winds and radiation @xcite . compared to these
, our simulations make some rather extreme assumptions , namely that the ejected particles are not initially heated ( though they can be shock heated by interactions with the ambient halo gas ) , and that our metals do not mix with the ambient gas .
it appears that these assumptions may be more critical to setting the cgm enrichment and phase structure than the details of the assumed wind speed and mass loading factor .
for instance , @xcite found that their amr - based models did not fare well against cgm observations , except with rather extreme assumptions about outflow energetics .
we look forward to detailed comparisons of other simulations , similar to what we have done here , in order to better assess how the physics driving outflows manifests in cgm absorption line data .
the two wind models considered here predict similar cgm absorption line properties , despite differences in the amount and physical state of cgm gas . we examine those differences here .
figure [ halofrac ] shows the _ cumulative _ enclosed baryonic fraction ( top panel ) and metal fraction ( lower panel ) within halos as a function of radius , for our cos - halos matched sample of galaxies .
the upper panel shows the total baryonic mass fraction divided by the cosmic baryon fraction ( i.e. @xmath71 ) .
unity indicates a cosmic mean baryon fraction , and shows approximately what models would predict in the absence of outflows ( in detail , it would be slightly higher than unity at these masses ; see e.g. * ? ? ?
the solid black line shows the total fraction of baryons ( stars+gas ) in halos for the ezw model , as a function of enclosed radius from the galaxy . even at the virial radius , this is less than unity , showing the efficiency of outflows in driving out baryons .
the dotted black line shows the baryon fraction for the cw model , which is substantially lower than that of the ezw model . owing to higher wind speeds in this halo mass range ,
the constant wind model carries more of its baryons out of halos into the igm than ezw .
both models push baryons out preferentially from the central regions , where most of the star formation ( and hence outflow generation ) occurs .
this leaves a large component of gas - phase baryons in the cgm . for the ezw model , cgm gas ( hot+cool phases ) makes up 65% of all baryons inside the halo .
this is broadly consistent with estimates of baryonic mass derived from observations @xcite .
the cw model predicts many fewer baryons , particularly in the cool phase .
however , it is important to note that in converting observable quantities into estimates of mass , assumptions must be made about the physical conditions in order to obtain ionisation corrections .
the consistency of ezw here , then , may simply reflect the similarity in physical conditions in the ezw model to those assumed in @xcite .
an assumption of a higher proportion of hot gas , as present in the cw model , would alter the conversion of observable quantities to mass . hence this should not be regarded as a success of ezw relative to cw , but may provide guidance on what assumptions should be made in order to obtain cgm mass estimates from absorption line data .
the ism of galaxies in the cw model contains an order of magnitude less baryons than in the ezw model , and about one - third as many stars .
the reasons for this are two - fold : more gas is driven out of the ism , making less available for star formation and strong outflows can actually heat the surrounding region thereby preventing the accretion of fresh gas that would otherwise replenish the loss @xcite .
the radial profile of the ism and stellar fractions are both quite flat , since these phases are concentrated in the galaxy s central regions .
more directly relevant to cgm absorption lines is the cool cgm baryon fraction , which is about three times higher in the ezw model than in the cw model and shows a modest increase with radius .
in contrast , the hot ( @xmath72k ) baryon fraction is not very different in the two models , and increases strongly with radius .
the fact that the hi equivalent widths are much less than a factor of three different in these two models ( see figure [ ew ] ) is , therefore , curious and suggests that the global cgm cool gas content is not straightforwardly traced by hi absorption .
however , this again may owe largely to saturation effects in hi .
the lower panel of figure [ halofrac ] shows the metal fraction in various halo gas phases .
here we have plotted the metal mass fraction enclosed within a radius @xmath73 , divided by the cosmic baryon fraction ( i.e. @xmath71 ) for ease of comparison with the upper panel .
the constant wind model has a lower enclosed metal fraction , since more of the metals have been driven out of the halo by outflows and since it made less stars .
of the metals that are present in the cw model , more are hot . for a sense of the metal content derived from observations of cos - halos galaxies , we refer the reader to @xcite , but again we note that conversions between observables and metal mass require an assumption of physical conditions . hence it is not straightforward to use such inferred measures to discriminate between wind models . in summary , we see that our two wind models distribute baryons and metals differently , even though they generally give similar observational absorption line properties .
this suggests that the total amount of cgm gas and metals is not sufficiently constrained by the current observations to distinguish between competing outflow models .
a key goal of cgm absorption line work is to be able to distinguish whether a given absorber arises from inflowing , outflowing , or ambient material ( e.g. * ? ? ?
* ) . in @xcite , we studied general observational diagnostics of the dynamical state of the gas in the ezw model and found , for example , that low metal ions tended to trace recycled accretion while high metal ions traced ancient outflows ( both defined below ) . here
, we expand on this past study by examining and contrasting the ezw and cw wind models , while focusing on our simulated los that mimic the cos - halos data set . in this way we aim to develop intuition about how cos - halos and similar cgm absorption line data sets are able to trace the baryon cycle .
the first step is to characterise inflowing and outflowing gas in the simulations , for which we follow @xcite .
briefly , we separate cgm gas particles into five different kinematic categories , utilising the past and future information we have within the simulations .
since we are interested in cgm and not ism , each of these categories excludes gas that is in the ism at @xmath20 .
the categories are :
* pristine accretion*. this is gas that is accreting , meaning it is not in the ism of a galaxy at @xmath20 but will either end up a the galaxy ( as an ism particle or star particle ) or pass through a galaxy and be ejected as a wind particle by @xmath74 .
pristine accretion is accreting gas that has not previously been in a wind .
* recycled accretion*. this is gas that is accreting , as defined above , but has been in a wind prior to @xmath20 .
3 . * young outflows*. this is gas that was ejected in a wind recently " , relative to @xmath20 .
we define recently as 1 gyr prior to @xmath20 ( @xmath75 for our cosmology ) , as this is roughly the time a particle would take to leave the halo if it were simply launched in a wind , and was not slowed down by hydrodynamic interactions . 4 .
* ancient outflows*. this is gas that was ejected in a wind longer than 1 gyr ago ( before @xmath75 ) .
* ambient*. this is gas that will not accrete onto a galaxy by @xmath74 , and has never been in a wind by @xmath20 .
one can think of ambient material as gas that is neither inflowing nor outflowing , and hence not participating in the baryon cycle . in figure [ ewcat ]
we plot the ew versus impact parameter , similar to figure [ ew ] , except now we decompose the simulated equivalent widths into contributions from the five dynamical categories above .
we calculate these in the manner described in @xcite , by making new simulation outputs containing _ only _ those gas particles that fit the definition of the various categories , rerunning los through them , generating spectra , and fitting column densities .
the ions are ordered from low to high ionisation energies , as in earlier figures .
we also plot as solid black lines results obtained including particles from all categories and the ism .
the contributions from all the categories ( coloured lines ) do not necessarily exactly add to the total ( black line ) , as they have been fitted separately .
we also show the cos - halos data for reference , as in figure [ ew ] , including detections and limits . to avoid clutter
we do not include the observed error bars , which are the same as in figure [ ew ] . finally , we bin both the model and the observed points in 25 kpc bins .
each of these new simulation snapshots is by definition sparser than the complete simulation , so it is no longer appropriate to use the medians of los with detections above a certain limit , as in figure [ ew ] . to account for the sparser simulation boxes , in figure [ ewcat ] we now plot the mean of all the model los at a given impact parameter .
as it is not appropriate to compare the mean of all los ( as plotted here ) to only detections ( as in the observations ) , we caution against comparing the solid black line to the observed black points .
the observed points are shown here to simply guide the eye and provide a rough comparison . for a more detailed , fair comparison ,
see figures [ ew ] , [ dedz ] , and [ cf ] .
we will now examine the behaviour of these ions in the models , with the understanding that these models can have significant discrepancies with the observed values , particularly for silicon and .
the top panels show hi , in ezw ( left ) and cw ( right ) .
the general trend is the same as that identified in @xcite : at low impact parameters , hi predominantly arises from recycled accretion , tracing dense gas that is about to re - accrete within several gyr , while at high impact parameters it arises mostly in pristine ambient gas . at all impact parameters
there is a non - trivial contribution from ancient outflows , which follows the ambient curve at a level that is set predominantly by the ratio of the masses in the ambient and ancient outflow categories .
these trends are qualitatively true in both wind models .
there are , however , non - trivial differences between these wind models . relative to the ezw model , in the cw model
the recycled accretion only dominates at the very smallest impact parameters . also at smaller impact parameters in the cw model ,
the contribution from ambient gas is 2 - 3 times more than than that from ancient outflows , whereas in the ezw model it is comparable . despite these variations ,
the total ( black ) line is very similar in both cases , showing that the hi equivalent width is not able to , on its own , discriminate the dynamical state of the cgm gas .
our lowest ionisation ion is and , again in accord with @xcite , we find that recycled accretion is the dominant contributor at all impact parameters . beyond 100 kpc
, there is an increasing fractional contribution from ancient outflows , but this is still sub - dominant .
this is likely because absorption ( and low ions generally ) drops off very quickly with impact parameter @xcite ; at large impact parameters the absorption is likely coming from satellites , not the central galaxy .
this trend is true in both wind models , though cw at small impact parameters shows a small contribution from ambient gas and young outflows .
the trends for are mimicked by the mid - ion ; this ion is still predominantly tracing halo gas that will accrete onto the galaxy .
in contrast , shows a dominant contribution from ancient outflows at all but the smallest impact parameters . as argued in @xcite , this is consistent with the idea that comes from a diffuse halo around the galaxy , enriched over long periods of star formation @xcite .
these trends are also fairly similar in the cw versus the ezw models , which indicates that these interpretations are robust at least to the range of wind models that we probe here .
note that , as discussed earlier , ours are not representative of all possible outflow mechanisms in current cgm simulations . in summary ,
the most clear trend is that at small impact parameters , recycled accretion dominates the absorption of low ions ( including ) , and that ancient outflows dominates absorption .
most , meanwhile , arises from ambient gas at larger impact parameters .
these trends mimic the expectations from @xcite .
hence the broad intuition developed in our previous work can be used to roughly interpret absorption line observations from cos - halos and similar studies , even though they may be less clear owing to a conflation of impact parameter and halo mass trends .
our two wind models show qualitatively similar behaviour , indicating that at least within the class of outflow models considered here these results are robust . the dominance of recycled accretion for low ions and ancient outflows for explains why our model predictions for star - forming and passive galaxies can be so similar : the absorbing metals are predominantly ones that were ejected long ago , and they are not coupled to the current star formation rate of the central galaxy .
thus , a quenching scheme that shuts off star formation but does not directly influence gas already in the halo has relatively little impact on cgm observables .
similarly , since the ambient gas dominates the hi absorption , one expects little difference between star forming and passive galaxies , except perhaps at small impact parameters where recycled accretion becomes important .
we have compared our cosmological hydrodynamic simulations including two galactic outflow prescriptions with hi and metal absorption line observations in the cgm from the cos - halos survey , to ( i ) assess how well current models that self - consistently enrich the cgm and igm fare against state of the art cgm observations ; ( ii ) determine if available observations are able to distinguish between outflow models ; and ( iii ) provide interpretations about the physical and dynamical state of cgm gas as observed by cos - halos . to accomplish this , we extracted los around simulated galaxies that match the cos - halos sample closely in stellar mass and impact parameter , analysed them in a manner consistent with the observations , and made comparisons to direct observables , namely covering fractions and equivalent widths as a function of impact parameter and specific star formation rate and the kinematics of the absorbers relative to the host galaxy .
in particular , we compared to cos - halos data tracing , , , , , , , and in 44 sightlines out to an impact parameter of 160 kpc around galaxies with stellar masses between @xmath76 at @xmath20 .
our main conclusions are as follows : 1 .
our favoured wind model , ezw , is in broad agreement with key absorption line observables from cos - halos for and some low ionisation metal lines . in particular , for these ions we find good agreement in the observed equivalent widths versus impact parameter , the total equivalent width per unit path length , and ion covering fractions above the cos - halos detection limit .
this model is also in broad agreement with ion ratios and absorber kinematics relative to the host galaxy .
however , there are numerous discrepancies that hint at missing or poorly represented physics in the simulations .
these include : * both wind models underpredict the two most commonly observed ions from cos - halos , namely the mid - ion and high - ion . for , the mismatch occurs at all impact parameters , while for the match is increasingly poor towards larger impact parameters .
this mismatch occurs in both equivalent width per unit redshift and the covering fraction . in general , all the ions but have a steeper gradient of absorption versus impact parameter in the models than what is observed , likely suggesting that there are not enough metals in the outskirts of halos in the models .
we also note that both wind model simulations have the same resolution , and hence fail equally to reproduce any cgm structure below the resolution limit . *
the total amount of absorption in , a mid ion , is generally too high compared to observations , while the amount in high ion is too low .
the ion ratios / and / echo these trends .
this is likely to be an issue related to how much metals are deposited into different phases of the cgm , in the sense that our model puts too many metals in @xmath77k gas and not enough in @xmath78k gas .
this occurs in both wind models , hinting that the underlying cause may be the way in which we eject winds as cold , unmixing gas from the ism .
* the large scatter in the equivalent widths at a given impact parameter for low ionisation lines is not well reproduced in the simulations .
this suggests that local effects perhaps owing to local ionisation sources and self - shielding might be necessary to fully explain the distribution of these low ions . *
while observations show a substantial difference in the incidence of ovi absorption around star - forming and passive galaxies , our simulations predict similar levels of absorption .
this discrepancy may arise because our phenomenological quenching scheme suppresses star formation in massive galaxies but does not directly affect halo gas .
conversely , our models successfully explain the observed similarity of low - ion absorption around star - forming and passive galaxies .
the constant wind model produces cgm absorption in all ions that is surprisingly similar to that from the ezw model , despite substantially different input wind scalings , which results in higher halo gas temperatures and fewer cool baryons and metals .
it is not possible to distinguish the cw model from the ezw model using the available equivalent width , covering fraction , or kinematic data , although ion ratios do reveal some differences .
the lack of discrimination between the ezw and the cw models is also surprising given our previous results in @xcite that showed larger differences between similar models .
we show that this owes largely to the fact that previously we compared models at a fixed halo mass , whereas here we compare at a fixed _ stellar _ mass .
the cgm metal absorption properties in these models are thus more closely correlated with stellar mass than halo mass .
both the ezw and cw models show large contribution from recycled accretion at low impact parameters ( @xmath1 kpc ) in low ions including , and from ancient outflows in at most impact parameters , in agreement with our previous results @xcite .
these trends are robust for our simulations , suggesting that they can be used to roughly infer the dynamical state of cgm absorbing gas at least within the context of the probed outflow models . the agreement in , , and between hydrodynamic simulations including galactic outflows and the most recent observations of cgm gas offers encouraging support for the baryon cycling paradigm .
the disagreement between both wind models and observations in and suggests further work is necessary in the simulations to get outflows to place metals in the proper location and phase in the cgm .
while the cos - halos observations of equivalent widths and covering fractions can not discriminate between the two wind models we consider here , it is important to recognise that these span a relatively narrow range of possible outflow models .
the ezw and cw models have different mass loading factors and ejection velocities , but in both cases our implementation ejects gas at the ism temperature ( usually @xmath79 ) and does not mix metals from sph wind particles into the surrounding cgm .
these assumptions are a limiting case , since some mixing must certainly occur , and if supernovae are involved in driving outflows it is likely that the ejected gas will be at least somewhat heated .
other groups have utilised fundamentally different algorithms for ejecting outflows , for example by super - heating the ism gas , which may give much larger variations in cgm properties than either of the models we consider here .
we plan to consider a wider range of outflow driving mechanisms in the future , so this work should be regarded as an initial investigation that we hope will set the stage for a more comprehensive comparison amongst all the groups working on models of the cgm and the baryon cycle .
the discrepancies between our models and current observations already suggest that we must improve our models of outflow ejection , propagation , and mixing . within the context of the models we consider here , our interpretations of the cos - halos observations are relatively insensitive to the details of how we implement winds .
for instance , the conclusion that low ion absorption is dominated by enriched gas that will shortly fall into the galaxy , while high ions arise in gas that was enriched at earlier times and is stably located in hot gaseous halos , is quite generic for our models .
of course , these interpretations may be different if we fundamentally changed our wind driving mechanisms , as discussed above .
nonetheless , the overall consistency of our models provides at least a plausible interpretation of how to connect the baryon cycle to cgm absorption line observations , which will help develop intuition and testable predictions for future observational probes of cgm gas .
support for program go 11598 was provided by nasa through a grant from the space telescope science institute , operated by the association of universities for research in astronomy , inc .
, under nasa contract nas 5 - 26555 .
much of the data presented herein were obtained at the w.m .
keck observatory , which operates as a scientific partnership among the california institute of technology , the university of california and the national aeronautics and space administration .
the observatory was made possible by the generous financial support of the w.m .
keck foundation .
the authors wish to recognise and acknowledge the very significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community .
we are most fortunate to have the opportunity to conduct observations from this mountain .
additionally , partial support for this work came from nasa atp nnx12ah86 g , hst grants hst - go-11598 and hst - go-12248 , nasa adp grant nnx08aj44 g , nsf grants ast-0847667 and ast-133514 , and the south african national research foundation s research chairs program .
the simulations used here were run on computing facilities owned by the carnegie observatories .
computing resources used for this work were made possible by a grant from the the ahmanson foundation , and through grant dms-0619881 from the national science foundation .
_ facilities : _ keck : i ( hires ) , _ hubble _ ( cos ) , b. j. 2009 , in american institute of physics conference series , vol . 1201 , american institute of physics conference series , ed .
s. heinz & e. wilcots , 142145 , j. k. , prochaska , j. x. , thom , c. , et al .
2012 , apjs , 198 , 3 | we analyze the low - redshift ( [email protected] ) circumgalactic medium by comparing absorption - line data from the cos - halos survey to absorption around a matched galaxy sample from two cosmological hydrodynamic simulations .
the models include different prescriptions for galactic outflows , namely hybrid energy / momentum driven wind ( ezw ) , and constant winds ( cw ) .
we extract for comparison direct observables including equivalent widths , covering factors , ion ratios , and kinematics .
both wind models are generally in good agreement with these observations for and certain low ionization metal lines , but show poorer agreement with higher ionization metal lines including and that are well - observed by cos - halos .
these discrepancies suggest that both wind models predict too much cool , metal - enriched gas and not enough hot gas , and/or that the metals are not sufficiently well - mixed .
this may reflect our model assumption of ejecting outflows as cool and unmixing gas .
our ezw simulation includes a heuristic prescription to quench massive galaxies by super - heating its ism gas , which we show yields sufficient low ionisation absorption to be broadly consistent with observations , but also substantial absorption that is inconsistent with data , suggesting that gas around quenched galaxies in the real universe does not cool . at impact parameters of @xmath1 kpc ,
recycling winds dominate the absorption of low ions and even , while almost always arises from metals ejected longer than 1 gyr ago .
the similarity between the wind models is surprising , since we show that they differ substantially in their predicted amount and phase distribution of halo gas .
we show that this similarity owes mainly to our comparison here at fixed stellar mass rather than at fixed halo mass in our previous works , which suggests that cgm properties are more closely tied to the stellar mass of galaxies rather than halo mass .
[ firstpage ] |
the study of the properties of hadrons under extreme conditions is one of the main research activities of several experimental programs and , in particular , of the forthcoming fair ( germany ) project .
the aim is to move from the light - quark sector to the heavy - quark domain and to face new challenges where charm and new symmetries , such as heavy - quark symmetries , will play a significant role .
one of the primary goals is to understand the nature of newly discovered states and , in particular , baryonic states with charm and bottom degrees of freedom . in that respect , approaches based on unitarized coupled - channel dynamics
have shown a tremendous success in the past .
recently , a unitarized coupled - channel scheme that incorporates heavy - quark spin symmetry ( hqss ) @xcite constraints has been developed @xcite .
hqss is a proper symmetry of the strong interaction that appears when the quark masses become larger than the typical confinement scale .
furthermore , nuclear medium corrections have been implemented @xcite to study the properties of the newly discovered heavy baryonic states in dense matter and their influence on heavy mesons in nuclear matter and nuclei . in this work
we study the properties of heavy hadrons in dense matter .
we aim at investigating nuclear medium effects on dynamically - generated heavy baryonic resonances and the consequences for the formation of charmed mesic nuclei as well as the propagation of heavy mesons under extreme conditions .
heavy baryonic states are dynamically generated by the scattering of mesons and baryons within a unitarized coupled - channel approach . in this work we employ a model that explicitly incorporates hqss @xcite .
hqss predicts that all types of spin interactions involving heavy quarks vanish for infinitely massive quarks , thus , connecting vector and pseudoscalar mesons containing heavy quarks .
furthermore , chiral symmetry fixes the lowest order interaction between goldstone bosons and other hadrons by means of the weinberg - tomozawa ( wt ) term .
this predictive model includes all basic hadrons ( pseudoscalar and vector mesons , and @xmath1 and @xmath2 baryons ) and it reduces to the wt interaction in the sector where goldstone bosons are involved while incorporating hqss in the sector where heavy quarks participate .
this scheme is justified in view of the reasonable outcome of the su(6 ) extension in the three - flavor sector @xcite and on a formal plausibleness in the vector - meson exchange picture of the interaction in the heavy pseudoscalar meson - baryon sector .
the extended wt model with hqss constraints is used as the kernel of the on - shell bethe - salpeter equation in coupled channels so as to calculate the scattering amplitudes .
the poles of the scattering amplitudes are the dynamically - generated heavy baryonic resonances . in this work we present results in the sector with heavy ( charm / bottom ) ( h ) , strange ( s ) and isospin ( i ) content such as @xmath3 , where the @xmath0 states are found @xcite .
we study the generalized weinberg s sum rule @xcite to estimate the importance of the different meson - baryon channels for the generation of the @xmath0 states . in table
[ fig : reso ] we show the spin - parity ( @xmath4 ) , mass ( @xmath5 ) and width ( @xmath6 ) of the different charmed and bottomed @xmath0 states , together with the absolute value of the coupling to the dominant meson - baryon channels ( @xmath7 ) .
the quantity @xmath8 represents the _ compositeness _ of the hadronic state in terms of all the considered channels , and @xmath9 is referred to as its _
elementariness_. a small value of @xmath9 indicates that the state is well described by the contributions explicitly considered , namely , @xmath10-wave meson - baryon channels , while a larger value of @xmath9 indicates that , for that state , significant pieces of information are missing in the model .
we obtain two @xmath11 and one @xmath12 @xmath0 states .
we find that the @xmath0 states which are bound states ( the three @xmath13 ) or narrow resonances ( one @xmath14 ) are well described as molecular states composed of @xmath10-wave meson - baryon pairs .
the @xmath15 wide @xmath14 as well as the @xmath16 @xmath17 states display smaller compositeness . with respect to the detailed composition of the states , we find that the first @xmath18 states of each flavor couple strongly to pseudoscalar-@xmath19 and vector-@xmath19 channels . for the second @xmath18 states ,
the main observation is its sizable coupling to the lightest channels @xmath20 .
another observation is the similar structure of the second @xmath18 and @xmath21 states , which appear as hqss or spin - flavor partners .
the in - medium modifications of the dynamically - generated @xmath0 states ( and others ) have important consequences on the properties of heavy mesons in matter . in particular , the properties of open - charm mesons in matter have been object of recent analysis due to the consequences for charmonium suppression . moreover , the existence of charmed meson bound states in nuclei have been predicted in @xmath22pb @xcite , using an attractive @xmath23 and @xmath24 -meson potential in the nuclear medium within a quark - meson coupling ( qmc ) model .
the experimental observation of these bound states , though , can be problematic since their widths could be very large compared to the separation of the levels . within our model , we obtain that the @xmath25-nucleus states are weakly bound ( see fig .
[ fig : medio ] ) , in contrast to previous results using the qmc model .
moreover , those states have significant widths @xcite , in particular , for @xmath22pb .
only @xmath25-nucleus bound states are possible since the coulomb interaction forbids the formation of bound states for @xmath26 mesons . as for @xmath24 mesons in nuclei , not only @xmath27 but
also @xmath28 bind in nuclei @xcite .
the information on bound states is very valuable for gaining insight into the meson - nucleus interaction at the panda experiment at fair . nevertheless , the experimental observation of @xmath23 and @xmath24-meson bound states is a difficult task .
open - charm mesons with high momenta would be produced in antiproton - nucleus collisions at panda and it is a challenge to bind them in nuclei @xcite .
information on the properties of heavy mesons in matter can be also achieved by analyzing the heavy - meson propagation in the hot and dense medium created in heavy - ion collisions @xcite .
the heavy - meson propagation can be studied by means of solving the corresponding fokker - planck equation .
the two relevant quantities to be determined are the drag ( @xmath29 ) and diffusion coefficients ( @xmath30 ) of heavy mesons in hot dense matter .
these are obtained from an effective field theory that incorporates both the chiral and hqss in the meson @xcite and baryon sectors @xcite .
one interesting observable is the behaviour of the spatial diffusion coefficient @xmath31 that appears in fick s diffusion law in medium , given in terms of the scalar @xmath32 and @xmath33 coefficients @xcite . in fig .
[ fig : diffusion ] we show @xmath34 for @xmath23 and @xmath35 mesons following isentropic trajectories ( @xmath36=ct ) from rhic to fair energies . for the @xmath23 mesons ,
we observe that the dependence of the @xmath34 on the entropy per baryon is similar in the hadronic and quark phase ( below and above the transition temperature of @xmath37 , respectively ) .
the possible matching between curves in both phases for a given @xmath36 seems to indicate the possible existence of a minimum in the @xmath34 at the phase transition @xcite . for @xmath23 meson ( left plot ) and @xmath35 meson ( right plot ) ( taken from ref .
@xcite ) . for recent updates in the high - temperature phase for @xmath23 meson ,
see refs .
@xcite . , title="fig:",scaledwidth=40.0%,height=188 ] for @xmath23 meson ( left plot ) and @xmath35 meson ( right plot ) ( taken from ref .
@xcite ) . for recent updates in the high - temperature phase for @xmath23 meson ,
see refs .
, title="fig:",scaledwidth=38.0%,height=188 ]
c. h .- d .
thanks the support of the jae - csic program . l.t . acknowledges support from the ramn y cajal research programme from ministerio de ciencia e innovacin and from fp7-people-2011-cig under contract no . pcig09-ga-2011 - 291679 .
this research was supported by spanish ministerio de economa y competitividad and european feder funds under contracts fpa2010 - 16963 , fis2011 - 28853-c02 - 02 , fpa2013 - 43425-p , fis2014 - 59386-p , fis2014 - 51948-c2 - 1-p and fis2014 - 57026-redt , and junta de andaluca grant fqm-225 .
9 isgur n and wise m b 1989 _ phys .
b _ * 232 * 113 ; neubert m 1994 _ phys . rept . _ * 245 * 259 ; manohar a v and wise m b 2000 _ camb .
monogr . part .
cosmol . _ * 10 * 1 garcia - recio c , magas v k , mizutani t , nieves j , ramos a , salcedo l l and tolos l 2009 _ phys .
d _ * 79 * 054004 gamermann d , garcia - recio c , nieves j , salcedo l l and tolos l 2010 _ phys . rev .
d _ * 81 * 094016 romanets o , tolos l , garcia - recio c , nieves j , salcedo l l and timmermans r g e 2012 _ phys .
d _ * 85 * 114032 garcia - recio c , nieves j , romanets o , salcedo l l and tolos l 2013 _ phys .
d _ * 87 * 034032 garcia - recio c , nieves j , romanets o , salcedo l l and tolos l 2013 _ phys . rev .
d _ * 87 * 074034 tolos l 2013 _ int .
j. mod .
e _ * 22 * 1330027 garcia - recio c , hidalgo - duque c , nieves j , salcedo l l and tolos l 2015 _ phys . rev
. d _ * 92 *
034011 tolos l , garcia - recio c and nieves j 2009 _ phys .
c _ * 80 * 065202 weinberg s 1963 _ phys .
_ * 130 * 776 tsushima k , lu d h , thomas a w , saito k and landau r h 1999 _ phys .
c _ * 59 * 2824 ozvenchuk v , torres - rincon j m , gossiaux p b , tolos l and aichelin j 2014 _ phys .
c _ * 90 * 054909 song t , berrehrah h , cabrera d , torres - rincon j m , tolos l , cassing w and bratkovskaya e 2015 _ phys . rev .
c _ * 92 * 014910 | we study the behavior of dynamically - generated baryon resonances with heavy - quark content within a unitarized coupled - channel theory in matter that fulfills heavy - quark spin symmetry constraints .
we analyze the implications for the formation of charmed mesic nuclei and the propagation of heavy mesons in heavy - ion collisions from rhic to fair . |
in nonequilibrium statistical mechanics there exist two annihilation reactions that have become prominent model systems .
these are the pair annihilation processes @xmath12 and @xmath13 .
interest in fluctuation effects on the first reaction process dates back to the seminal work by smoluchowski @xcite , and in more recent times at least to bramson and griffeath @xcite ; and for the second process fluctuations were first studied by ovchinnikov and zeldovich @xcite .
the final states of both processes are trivial : all particles ( save perhaps a single one ) disappear in the case of @xmath14 , and only the excess number of particles of the initial majority species is left in the case of @xmath11 .
the nature of the approach to the final state is , however , nontrivial and has been the subject of many investigations .
the @xmath14 problem applies , for example , to domain walls in a one - dimensional system like in the zero - temperature ising model @xcite .
its asymptotic behavior is the same as that of the diffusion - limited coagulation process @xmath15 @xcite .
the ensuing asymptotic density decay @xmath16 has been experimentally observed ( in an intermediate time window ) in the fusion kinetics of laser - induced excitons in quasi one - dimensional n(ch@xmath17)@xmath18mncl@xmath17 ( tmmc ) polymer chains @xcite .
the study of the two - species annihilation reaction @xmath11 was motivated originally by the cosmological problem of matter - antimatter annihilation .
ovchinnikov and zeldovich @xcite , and independently a few years later toussaint and wilczek @xcite , asked whether in such a simple annihilation model it would be possible that locally in space only , say , matter would be left .
the answer rapidly turned out to be positive : the @xmath11 system exhibits the phenomenon of species `` segregation '' , that is , the emergence of ever growing single - species domains ( either @xmath19 or @xmath20 ) . as a result , after a short initial period , annihilation takes place only in `` reaction zones '' where the domains border , and the decay of the total density is slowed down markedly .
indeed , a non - classical decay @xmath21 has been experimentally confirmed in three dimensions in a calcium / fluorophore reaction system @xcite .
the condition for segregation to occur is that the spatial dimension @xmath22 be less than the critical _ segregation dimension _ @xmath8 , which in the case of @xmath11 is equal to @xmath23 .
early work on this two - species problem was done by kang and redner @xcite ; clever heuristic reasoning and numerical work on the spatial structure of the domains is due to leyvraz and redner @xcite ; and rigorous results were derived by bramson and lebowitz @xcite .
the renormalization group approach to the reaction @xmath14 was pursued by peliti @xcite and lee @xcite , and to the process @xmath11 by lee and cardy @xcite , and oerding @xcite .
the strong interest in these two reactions has sparked off research on many related reaction - diffusion problems by various techniques .
here we will refer only to general introductions and overviews , _
e.g. _ refs .
@xcite-@xcite . in 1986 ben - avraham and redner @xcite , crediting kang , introduced a @xmath0-species pair annihilation problem that interpolates between these two models , and hence puts them in a common perspective .
they considered a system of @xmath0 distinct particle species @xmath24 , all propagating with the same difffusion constant , and reacting according to @xmath25 with a single fixed rate . for @xmath26 and @xmath27 this @xmath0-species
_ mutual annihilation model _ ( @xmath0-mam ) reduces to the well - understood paradigmatic cases @xmath11 and @xmath14 , respectively .
( for the latter case , note that the probability of identical species to meet becomes vanishingly small in the limit of @xmath28 ; hence any particle encounter will lead to a reaction , and the distinction of different species becomes meaningless . )
the general situation has recently been studied analytically and by monte carlo simulations in refs .
@xcite . for equal initial densities of all species the asymptotic decay with time @xmath29 of the total particle density @xmath30 follows a power law @xmath31 . in table
i we summarize the values of the decay exponent @xmath32 , known or believed to be exact , as a function of @xmath0 and of the spatial dimension @xmath33 @xcite .
we also indicate whether or not the density decay is accompanied by particle species segregation .
the results for @xmath34 and @xmath35 are based on the exact solution of a modified model argued by the authors of refs .
@xcite to be equivalent to the one - dimensional @xmath0-mam in the large - time limit . [ cols="^,^,^,^ " , ] table i. the density exponent @xmath32 of the asymptotic decay law @xmath36 for the total partical density in the @xmath0-species mutual annihilation model , as a function of the number of species @xmath0 and the spatial dimension @xmath33 .
an asterisk ( * ) indicates that segregation occurs and ( @xmath37 ) signifies the appearance of logarithmic corrections at the upper critical dimension @xmath38 .
the @xmath0-species mam is characterized by an
_ upper critical dimension _
@xmath38 , below which mean - field theory breaks down and renormalization is needed .
physically , in dimensions @xmath39 the recurrence properties of random walkers lead to depletion zones about each surviving particle , and to particle anticorrelations induced by the annihilation kinetics .
a systematic renormalization group treatment , wherein the spatial dimension @xmath33 can be treated as a continuous variable , has only been carried through for the cases @xmath27 ( equivalent to the reaction @xmath14 ) @xcite , and for @xmath26 ( the two - species reaction @xmath11 ) @xcite , which explains the empty entries in the center of table i. the derivation of the decay laws listed in table i relies heavily on the full permutational symmetry of the particle species in the @xmath0-mam .
it is therefore natural to ask what happens to these asymptotic results when the symmetry is lowered .
a case of lower symmetry occurs , for example , when the particle species are ordered cyclically and annihilation , with rate @xmath40 , is possible only between particles of two neighboring species along the cycle .
the reaction constants are then given by @xmath41 \qquad ( 1\leq i , j\leq q ) \ .
\label{kijcam}\ ] ] we will refer to this particular system as the @xmath0-species _ cyclic annihilation model _
( @xmath0-cam ) .
other examples that display lower than permutational symmetry may easily be constructed . in this work
we develop , within the mean - field framework ( _ i.e. _ , in dimensions @xmath42 ) , a general method for finding the exponent of the total density decay law @xmath36 , and determining whether or not segregation occurs in a @xmath0-species model where the reaction rates @xmath3 lower the full permutational symmetry . in sec .
[ secmodel ] we present our model , which is fully defined by the @xmath43 matrix @xmath7 of annihilation rates . in sec .
[ secmeanfield ] we study how the particle numbers of the individual species fluctuate around their average densities at any given time . in sec .
[ secfinitevolume ] we show , by comparing the average decay law with the fluctuations around it , how the reactions come to an end in a finite @xmath33-dimensional volume @xmath44 , and we determine the particle densities in the final state . in sec .
[ secfinitedimension ] we treat the full @xmath33-dimensional space as built up from blocks of size @xmath44 that do not communicate for times less than the diffusion time , which is of order @xmath45 .
this allows us to derive our general results concerning the occurrence of segregation and the exponents in the density decay law . in sec .
[ secexamples ] we consider a specific example which , by means of a suitable parameter @xmath46 , interpolates between the @xmath0-cam ( for @xmath47 ) and the @xmath0-mam ( for @xmath48 ) . in sec .
[ secsimulations ] we present monte carlo simulations which , although preliminary and of limited statistical accuracy , are compatible with our analytical findings . in sec .
[ secdiffeq ] we comment on an alternative approach , namely via rate equations with an additional particle diffusion term , before we conclude in sec .
[ secconclusion ] .
our considerations allow us to answer another question as well : in the two - species @xmath11 system the appearance of segregation is usually explained heuristically through arguments invoking the local conservation of the difference between the number of @xmath19 and @xmath20 particles .
this conservation law thus seemed to be at the origin of the segregation phenomenon .
however , the discovery @xcite that the one - dimensional mam exhibits segregation for all @xmath49 , even though it is not subject to any conservation law , cast doubt upon this apparent direct link between conservation laws and segregation .
this doubt was subsequently reenforced by a heuristic argument according to which the mam should exhibit segregation in the entire region of the @xmath0-@xmath33 plane delimited by @xmath50 and @xmath51 @xcite .
( the only weakness of this argument is that the region concerned by it does not contain any points with integer `` physical '' @xmath0 and @xmath33 , except for @xmath26 , where the conservation law applies . )
this present work clarifies this issue : we show that the existence of a conservation law , although a _ sufficient _ condition for segregation in dimensions @xmath52 , does _ not _ constitute a _ necessary _ condition : segregation may in fact occur in the absence of any conservation law .
we consider a system of @xmath0 distinct diffusing particle species @xmath53 subject to the pair annihilation reactions @xmath54 here the @xmath55 represent the reaction rates per unit of density ; they are constrained only by the requirements that they be non - negative and that there exist a symmetry operation under which all particle species are equivalent .
we will set @xmath4 , implying that no self - annihilation is possible . without loss of generality we may consider the matrix @xmath7 to be irreducible , _
i.e. _ , the system of reactants can not be decomposed into mutually noninteracting subsystems .
the @xmath0-species mam and the @xmath0-species cam discussed in the introduction are obtained as special cases when all reaction rates in eq .
( [ reactioneqns ] ) are equal and when they are given by eq .
( [ kijcam ] ) , respectively .
we shall therefore proceed to analyze the mean - field behavior of the system ( [ reactioneqns ] ) for a general non - negative traceless symmetric matrix @xmath7 .
our focus will be on the power law for the density decay , @xmath36 , and on the segregation properties of this system .
straightforward scaling analysis tells us that for pair annihilation processes such as considered here , very generally mean - field theory is applicable in spatial dimensions @xmath56 , and , likely with logarithmic corrections , at the upper critical dimension @xmath57 .
we shall operate entirely at the mean - field level , and hence our results will concern dimensions @xmath58 . in our discussion
we will briefly touch upon implications for lower dimensions @xmath59 .
note that the system of reactions ( [ reactioneqns ] ) may be visualized as a graph @xmath60 with @xmath0 vertices representing the @xmath0 species , in which bonds carrying weights @xmath3 connect the vertices of pairs of species that may react with each other . at certain points in our discussion
it will be convenient to refer to this graph representation .
let the stochastic variable @xmath61 denote the particle number of species @xmath62 present at time @xmath29 in a system of volume @xmath44 , and let @xmath63 be the average density of the @xmath62 for @xmath64 . here
@xmath65 is an average with respect to ( i ) the initial distribution of the @xmath66 at time @xmath67 and ( ii ) all realizations of the stochastic time evolution . in order to express the mean - field rate equations @xmath68 in terms of dimensionless variables , we set @xmath69 ( which because of the symmetry between all species is independent of @xmath70 ) and @xmath71 .
we furthermore define the quantities @xmath72 , @xmath73 , and @xmath74 .
the mean - field equations then become @xmath75 now with the normalization @xmath76 . if all initial densities @xmath77 and hence @xmath78 for all @xmath70 , then the solution of eq .
( [ meanfieldeqns ] ) reads @xmath79 which tends to zero as a power law with decay exponent @xmath80 . for unequal initial densities
the asymptotic decay will generically be exponential ( readily generalized to stretched exponential for @xmath39 as a consequence of the reaction rate renormalization @xcite ) , with one or several of the species tending toward a positive limit density @xmath81 ; this case will not be considered any further below . in principle , the fluctuations around this mean - field average are encoded in the master equation for the probability distribution @xmath82 , @xmath83 \ . \label{mastereq}\ ] ] a convenient means to extract them proceeds via van kampen s @xmath84-expansion @xcite , as was demonstrated for the @xmath0-mam by ben - avraham and redner @xcite , and more recently for a cyclic three - species trapping reaction by ben - naim and krapivsky @xcite , and for a zero - dimensional population dynamics model by newman and mckane @xcite . anticipating that the fluctuations should be of the order of the square root of the total particle number @xmath85 , we set @xmath86 where , due to previous definitions , @xmath87 , and transform the probability distribution @xmath88 on the extensive variables @xmath66 into an equivalent one , to be called @xmath89 , on the intensive variables @xmath90 , according to @xmath91 expanding the master equation ( [ mastereq ] ) for @xmath88 to second order in powers of @xmath92 , as in refs.@xcite , @xcite , and @xcite , and exploiting the rate equations ( [ meanfieldeqns ] ) for @xmath93 then yields a fokker - planck equation with time - dependent coefficients , @xmath94 f \ , \label{fpeqn}\ ] ] where @xmath95 .
this equation is valid provided the second term on the r.h.s . of eq .
( [ defalphaitau ] ) remains much smaller than the first one . from the fokker - planck equation ( [ fpeqn ] ) we may readily deduce equations for the time evolution of the averages @xmath96 , the variances @xmath97 , and the covariances @xmath98 . since @xmath99 , and given that the @xmath100 are zero initially , we see that they vanish for all times .
the covariance matrix @xmath101 satisfies @xmath102 now let @xmath103 be the real unitary matrix with elements @xmath104 that renders @xmath105 diagonal , and denote its eigenvalues by @xmath106 .
in the same manner , we define @xmath107 . upon transforming eq .
( [ eqnsglmt ] ) to these new variables we obtain @xmath108 \ , ( \rho^2 \ , \delta_{\mu\nu } - 2 \rho \ , \hat\gamma_{\mu\nu } ) \ . \label{eqngmunut}\ ] ] let us suppose that the particle numbers initially have independent and identical fluctuations , _
i.e. , _ @xmath109 . then also @xmath110 , and it follows from eq.([eqngmunut ] ) that @xmath111 for @xmath112 at all @xmath113 .
the remaining equation of interest is @xmath114 notice that since @xmath115 depends on time , it is not useful to scale the time @xmath116 in ( [ eqngmumutau ] ) with the factor @xmath117 .
the normalization relation @xmath76 implies that @xmath118 is an eigenvalue of @xmath119 with eigenvector @xmath120 . by the perron - frobenius theorem , and since the @xmath106 are necessarily real , we have @xmath121 for all @xmath122 . in case @xmath123 , the r.h.s . of eq.([eqngmumutau ] ) vanishes and the initial fluctuation @xmath124 does not decay , _
i.e. , _ an eigenvalue @xmath125 signals the presence of a conservation law . the solution of eq . ( [ eqngmumutau ] ) , except when @xmath126 , is a sum of two power laws , @xmath127 where we have introduced the coefficient @xmath128 .
note that the amplitude of ( only ) the first term in eq.([solnflucts ] ) depends on the initial fluctuation strength @xmath129 .
its exponent @xmath130 varies for the different modes @xmath122 , which shows that this problem is characterized by a spectrum of power laws . the second term in eq .
( [ solnflucts ] ) is just proportional to the density @xmath131 and hence decays as @xmath132 .
therefore , if @xmath133 , the first term decays more slowly than the second one ( and vice versa for @xmath134 ) .
when it so happens that @xmath135 , then eq . ( [ solnflucts ] ) should be replaced with the special logarithmic solution @xmath136 the variance of the particle number for species @xmath70 is given by @xmath137 let us denote the algebraically smallest eigenvalue of @xmath119 by @xmath138 , its degeneracy by @xmath139 , and the corresponding value of the coefficient @xmath140 by @xmath141 . for asymptotically large times
we then deduce from eqs .
( [ solnflucts])-([expravvar ] ) the behavior @xmath142 \tfrac{1}{2 } \tau^{-1 } \log\tau & \\[2 mm ] c_{\rm m } \ , q^{-1 } ( \gamma_0-k_{\rm m } ) \ , \tau^{-2(1+\hat{\kappa}_{\rm m } ) } & \end{array } \right .
\quad \big ( \hat{\kappa}_{\rm m } \gtreqqless -\tfrac{1}{2 } \ , , \quad \tau \to \infty \big ) \ , \label{fluctsllaspt}\ ] ] with the abbreviation @xmath143 .
we are now ready to exploit the results of the preceding sections .
we begin by posing the question whether there exists a characteristic time @xmath144 at which the particle number @xmath66 in the volume @xmath44 displays a root mean - square deviation from the average that is of the order of the average itself , _
@xmath145 yet the fokker - planck equation ( [ fpeqn ] ) was derived under the hypothesis of small fluctuations ( the second term in eq.([defalphaitau ] ) was supposed to small compared to the first one ) , and hence ceases to be valid for @xmath146 .
the issue then is what happens at and beyond this time scale .
fluctuations satisfying ( [ relavvar ] ) indicate that the different particles species have widely varying numbers . as we argued in the introduction , in this situation the annihilation processes come to an end epxonentially , on the prevailing time scale , through the successive extinction of one or several particle species ; this final decay is no longer described by the fokker - planck equation ( [ fpeqn ] ) . in the finite volume @xmath44 ,
the final state therefore consists of a collection of particles that are not subject to pair annihilations anymore .
if all @xmath147 for @xmath148 , then at most a single species can survive .
but if some of the rates @xmath149 vanish , a broader variety of final states is possible ( we shall return to this point in sec .
[ secexamples ] ) . in any of these cases , whereas initially all species were equivalent , the final state has this symmetry broken .
in fact , this symmetry breaking may be traced back to the first term in eq .
( [ solnflucts ] ) , which includes the amplitude @xmath150 .
thus , this symmetry breaking is enhanced by initial fluctuations , as represented by @xmath129 , yet it actually _ persists even for the case @xmath151 , i.e. if we initially take all particle numbers @xmath152 to be strictly equal_. we shall discuss several examples of such broken symmetry states in sec . [ secexamples ] .
we turn now to the determination of the characteristic time @xmath153 . utilizing in eq .
( [ relavvar ] ) that @xmath154 for @xmath155 as well as inserting the asymptotic results of eqs .
( [ expravvar ] ) and ( [ fluctsllaspt ] ) , and then solving for @xmath153 , we find @xmath156 2 n_0 l^d / \log ( n_0 l^d ) & \\[2 mm ] c^{-1 } ( n_0 l^d ) ^{1/|2\hat{\kappa}_{\rm m}| } & \end{array } \right .
\quad \big ( \hat{\kappa}_{\rm m } \gtreqqless -\tfrac{1}{2 } \ , , \quad \tau \to \infty \big ) \ , \label{solnt*}\ ] ] where @xmath157^{1/|2\hat{\kappa}_{\rm m}|}$ ] .
the order of magnitude @xmath158 of the particle number for a surviving species will be @xmath159 , and hence is given by @xmath160 \tfrac{1}{2 } \log(n_0 l^d ) & \\[2 mm ] c ( n_0 l^d)^{1 - 1/|2 \hat{\kappa}_{\rm m}| } & \end{array } \right .
\quad \big ( \hat{\kappa}_{\rm m } \gtreqqless -\tfrac{1}{2 } \ , , \quad \tau \to \infty \big ) \ .
\label{exprrhot*}\ ] ] when @xmath161 this surviving number is of order unity , but when @xmath162 it grows logarithmically with the volume ; and when @xmath163 it grows with a positive power of @xmath164 .
we now consider an infinite volume in which the @xmath0 annihilating species propagate with ( uniform ) diffusion constant @xmath165 , and apply a heuristic analysis : we imagine this infinite volume divided into hypercubic subvolumes of size @xmath44 . for times less than the characteristic diffusion time , _
i.e. , _ for @xmath166 , the subvolumes can be treated as effectively independent , and hence the results of eqs .
( [ solnt * ] ) and ( [ exprrhot * ] ) apply to each of them .
we will discuss the four distinct cases separately .
\(i ) _ case @xmath167_.from eq . ( [ exprrhot * ] ) we infer in this situation that in an isolated volume the particle number fluctuations become of the order of the average itself only at times when the total particle numbers @xmath168 have decreased to order unity themselves . in any case , in dimensions @xmath5 the time scale at which this happens , @xmath169 , is much larger than the diffusion time @xmath170 .
hence in an infinite system the subvolumes begin to mix diffusively long before the fluctuations of their particle numbers become of the order of the averages .
this is tantamount to stating that for @xmath171 there is no particle segregation .
\(ii ) _ case @xmath172_.in this case eq .
( [ solnt * ] ) states that the fluctuations become of the order of the average value @xmath131 at a characteristic time @xmath173 , and in an isolated volume the annihilation processes will come to an end on this same time scale . similarly in an infinite system , subject to the condition that @xmath174 each subvolume of size @xmath44 will reach a quasi - final state , to be referred to as a `` domain '' , and further reactions are possible only at the `` reaction zones '' separating the domains .
the quasi - final state in each subvolume is reached independently of its neighbors , and the emergence of these different disjoint quasi - final states constitutes the phenomenon of segregation into domains .
the condition ( [ reltimes ] ) implies that segregation occurs in dimensions below a _
critical segregation dimension _
@xmath175 @xmath176 it therefore appears that in the present case the critical segregation dimension for diffusion - limited multi - species pair annihilation processes is always constrained by the interval @xmath177 .
the dimensionless particle density in a given domain will be of the order @xmath178 and can not decrease further until diffusion between neighboring domains permits new reactions to take place . hence we conclude that at any time @xmath116 the particle density in an infinite system equals @xmath179 , where @xmath180 . by combining these relations
we find that the density decays asymptotically as @xmath181 where @xmath182^{-2/d } n_0^{2/d-1 } d / k_0 $ ] is a dimensionless constant .
thus we obtain @xmath183 for @xmath10 .
\(iii ) _ case @xmath184_.in this case @xmath185 , which is also the upper critical dimension @xmath186 where mean - field theory is only marginally applicable .
nevertheless , upon following the same reasoning as above we find the asymptotic density decay law @xmath187 in two dimensions .
based on renormalization group arguments and monte carlo simulations , this result was predicted in particular for the @xmath188-mam @xcite , for which indeed @xmath189 ( see subsec .
[ subsecqmam ] below ) .
it is remarkable that the logarithmic corrections ( [ rhotaudseg ] ) take the same form as quite generically predicted for pair annihilation reactions in two dimensions as consequence of the required reaction rate renormalizations induced by the appearance of depletion zones in low dimensions .
although segregation effects are very difficult to capture within the framework of renormalized field theory ( compare the discussion in ref .
@xcite ) , we therefore hypothesize , but can not prove , that no additional logarithmic factors beyond those exhibited in eq .
( [ rhotaudseg ] ) appear in the general case .
\(iv ) _ case @xmath190_.this special case , for which @xmath191 , occurs when the graph @xmath60 is bipartite , _
i.e. _ , when it is comprised of two subsets of vertices such that all bonds are between vertices of different subsets
. since these subsets are necessarily equivalent under symmetry , this obviously requires that @xmath0 be even , and we may relabel the sets to @xmath192 and @xmath193 .
the particle numbers then obey the _ local conservation law _ @xmath194 and it is readily verified that @xmath119 has an eigenvector @xmath195 with eigenvalue @xmath125 . for @xmath196
we moreover infer from eq .
( [ solnflucts ] ) that @xmath197 , which establishes that as a direct consequence of the conservation law the initial fluctuations in the total particle numbers of the two subsets do not relax .
examples where this happens are the @xmath198-mam ( the two - species pair annihilation reaction @xmath11 ) and the @xmath0-cam with even @xmath0 ( see subsec . [ subsecqcam ] ) .
we proceed to investigate a specific @xmath0-species pair annihilation system that depends on a parameter @xmath46 , interpolating for @xmath199 between the @xmath0-cam and the @xmath0-mam that were defined in the introduction .
we imagine the @xmath0 species arranged in a cycle . the annihilation rate is set equal to @xmath40 for any pair of species that are nearest neighbors along the cycle , and to @xmath200 for any other pair .
then the ratio @xmath201 is the only intervening parameter in the normalized matrix @xmath119 . using the abbreviation @xmath202
, one has @xmath203 / \ , 2 { \cal n}_{q\lambda } \qquad ( i \neq j ) \label{defkappaijgeneral}\ ] ] and @xmath204 . for @xmath205
this model is symmetric only under cyclic translation of the species , and the special case @xmath206 yields the @xmath0-cam . for @xmath48 the model displays full permutation symmetry and we recover the @xmath0-mam .
it is straightforward to diagonalize @xmath119 by fourier transforming in the variable @xmath207 , and hence the mode labels @xmath208 belong to the set of wavenumbers @xmath209 with @xmath210 .
the corresponding eigenvalues are @xmath211 / \,{\cal n}_{q\lambda } \ . \label{evkappapgeneral}\ ] ] we shall first discuss the two limiting models , the mam and the cam , whose segregation properties turn out to be quite different .
next , our study as a function of @xmath46 will characterize the crossover between these two cases , thereby shedding light on the nature of the segregation mechanism .
the parameter value @xmath48 defines the mutual annihilation model ( @xmath0-mam ) . in this case
the graph @xmath60 with @xmath0 vertices is fully connected and every bond carries the same weight .
this model , introduced in ref .
@xcite , has been extensively studied in refs .
we may reproduce the known results in dimensions @xmath58 as follows .
( [ evkappapgeneral ] ) shows that for @xmath48 the matrix @xmath119 has a single eigenvalue @xmath212 ( namely for the mode with @xmath213 ) and a @xmath214-fold degenerate eigenvalue @xmath215 ( which occurs for all @xmath216 ) . in the two - species model with @xmath26 ,
this latter eigenvalue equals @xmath125 , whence we encounter the special case ( iv ) of sec .
[ secfinitedimension ] .
there is a conservation law , we find @xmath23 , and for @xmath217 the density decays as @xmath218 @xcite . for @xmath219 ,
the eigenvalue @xmath220 lies in the interval @xmath221 $ ] , whence we are confronted with case ( ii ) of sec .
[ secfinitedimension ] . according to eq .
( [ exprdseg ] ) , there then exists a @xmath0-dependent segregation dimension @xmath222 a relation first derived in ref .
although this case does not encompass any integer values of @xmath0 , it does suggest that segregation may occur even in the absence of a conservation law .
such a scenario will indeed be confirmed below .
the marginal value @xmath223 corresponds to the special case ( iii ) of sec .
[ secfinitedimension ] .
indeed , it was concluded in refs .
@xcite on the basis of renormalization group arguments that in two dimensions the total density in the @xmath188-mam decays as @xmath224 , in agreement with eq .
( [ rhotaudseg ] ) . for @xmath225
we obtain @xmath226 , and our present theory is not applicable anymore . the analytical and numerical studies reported in refs .
@xcite however yield species segregation in one dimension , with a @xmath0-dependent density decay exponent @xmath227 .
for @xmath47 the cyclic graph @xmath60 only contains nearest - neighbor bonds , _
i.e. _ , a particle of a given species can annihilate only with particles of one of its two neighboring species .
this model has been called the _ cyclic annihilation model _ or @xmath0-cam . in this case
the eigenvalues given in eq .
( [ evkappapgeneral ] ) reduce to @xmath228 . for _
@xmath0 , the mode with @xmath229 yields the smallest eigenvalue @xmath230 .
but for _ odd _ @xmath0 the smallest eigenvalue is acquired for @xmath231 , namely @xmath232 .
it then follows from eq .
( [ exprdseg ] ) that @xmath233 4\cos\frac{\pi}{q } & \qquad ( q=3,5,7,\ldots ) \end{array } \right . \ . \label{dsegcam}\ ] ] for @xmath234 these results for @xmath235 were previously known , since in these cases the cam actually coincides with a mam : the @xmath188-cam coincides with the @xmath188-mam , and the @xmath198-cam and @xmath236-cam are both equivalent to the @xmath198-mam @xcite .
the first novel case is therefore @xmath237 , for which eq .
( [ dsegcam ] ) gives @xmath238 it follows that the @xmath239-cam exhibits segregation in spatial dimensions @xmath240 and @xmath241 , and eq .
( [ densdecay ] ) implies that @xmath242 with @xmath243 \tfrac{3}{4}(\sqrt{5}-1)=0.927 ... \qquad & ( d=3 ) \end{array } \right .
\qquad ( q=5 ) \ .
\label{goldenmean}\ ] ] ( curiously , the decay exponent in two dimensions is just the golden mean . ) in simulations , aside from the observation of species segregation , direct measurement of the exponent values @xmath32 provides a straightforward means to verify the present theory ( see sec . [ secsimulations ] below ) . the domain structure in the case of the @xmath0-cam needs to be discussed .
whereas the mam is necessarily characterized by single - species domains , this is no longer true for the cam .
let us first consider @xmath237 .
for the @xmath239-cam one can certainly conceive the possibility of single - species domains .
any given particle species , however , does not react with two other species , and hence is able to coexist with either of those .
it appears evident , therefore , that a single - species domain is not stable against penetration by either of the two species with which its particles can not annihilate .
thus we must suppose that a typical segregated domain will always contain two species , say @xmath244 and @xmath245 , out of the five , _ viz .
_ any of the cyclically equivalent combinations @xmath246 @xmath247 .
these domains are `` stable against penetration '' , in the following sense : any particle of a different species intruding into such a domain would annihilate with at least one of the two domain species . for @xmath248 new questions appear that we will not fully address here . as for @xmath237 , we may list the subsets of particles that can coexist in a domain where no interactions take place anymore .
there are two subsets with three species , \{135 } and \{246 } , and three subsets with only two species , namely \{14 } , \{25 } , and \{36}. all these five subsets are `` impenetrable '' to an intruder species .
it is therefore an interesting issue which type of domains are going to be formed in the segregation process .
one important conclusion from the @xmath0-cam is that for @xmath249 we encounter `` physical '' situations ( @xmath0 and @xmath33 are integers ) , where , in spite of the absence of a conservation law , species segregation into domains emerges .
note that the segregation mechanism does not even require initial fluctuations in the species numbers : even if initially absent , such fluctuations are dynamically generated by the annihilation reactions , and they decay more slowly than the average particle numbers themselves .
we now allow an arbitrary value for @xmath46 in the interval @xmath250 $ ] , _ i.e. _ , we consider the more general interpolating model . for even @xmath0 , eq .
( [ evkappapgeneral ] ) tells us that the eigenvalue @xmath251 assumes its minimum for @xmath229 ; for odd @xmath0 one may verify that , just as for @xmath48 , it becomes minimal for @xmath252 as close as possible to @xmath253 , that is , for @xmath254 . upon substituting these minimal eigenvalues in eq .
( [ exprdseg ] ) , we obtain @xmath255 \big [ 2\lambda + 4(1-\lambda ) \cos\frac{\pi}{q } \big ] / \ , { \cal n}_{q\lambda } & \ q \mbox { odd } \end{array } \right . \ . \label{eqndsegqeven}\ ] ] for @xmath256 , this expression reduces to the limiting values found above .
it is consistent with our mean - field assumptions only as long as it leads to a @xmath257 .
that is easily seen to imply that at given @xmath46 , it is meaningful only for @xmath0 below a maximum value @xmath258 ; for @xmath0 above that value , there is no segregation in any @xmath259 .
we have performed monte carlo simulations in one , two , and three dimensions for the @xmath0-cam with @xmath260 with the goal to test the theoretical predictions for the decay exponent @xmath32 . for more particle species ,
our computing resources unfortunately can not provide sufficiently reliable data statistics .
the simulation algorithm proceeds as described in ref .
@xcite : starting from an initially random distribution of particles on a @xmath33-dimensional ( @xmath261 ) hypercubic lattice with periodic boundary conditions , the system is evolved as follows : a particle is randomly picked .
next , one of its @xmath262 nearest - neighbor sites is selected randomly ; if it is occupied by a particle of a different species , both particles are removed , otherwise the particle hops to the empty site .
the monte carlo time is scaled with the total number of particles present ( asynchronous time update ) .
effective decay exponent @xmath263 for the three - dimensional @xmath0-cam with @xmath26 , @xmath188 , @xmath236 , and @xmath239 ( random initial conditions , equal initial particle numbers for each species ) .
the data were obtained on a @xmath264 cubic lattice , averaged over @xmath265 runs . ]
beginning with the data in three dimensions , obtained from simulations on a @xmath264 cubic lattice by averaging over @xmath265 runs with random initial conditions , we plot the effective decay exponent @xmath266 in fig .
[ 3dcam ] . in @xmath240 ,
naturally our statistics is worst , and the data cease to be reliable at around @xmath267 .
the results for even and odd @xmath0 are clearly distinct ; and as anticipated , the processes for @xmath26 and @xmath268 are equivalent . whereas eqs .
( [ dsegcam ] ) and ( [ densdecay ] ) predict @xmath269 asymptotically for @xmath0 even , the data show that the effective exponent @xmath263 rather reaches a plateau at @xmath270 . as discussed in ref .
@xcite , one should however expect the asymptotic decay law to be somewhat masked by the mean - field behavior @xmath271 , which could explain the combined effective decay exponent closer to @xmath212 .
indeed , we found the same deviation from the asymptotic value @xmath272 in our simulations for the @xmath198-mam ( see fig . 8 in ref .
the data for odd @xmath0 are similarly plagued by crossover effects .
for @xmath223 , for which the mam and cam are equivalent , we see a slow convergence towards @xmath273 . the data for @xmath237 might indicate that @xmath263 reaches a plateau at a smaller value , in line with the prediction @xmath274 of eq .
( [ goldenmean ] ) , but the deteriorating statistics preclude a definite conclusion .
effective decay exponent @xmath263 for the two - dimensional @xmath0-cam with @xmath26 , @xmath188 , @xmath236 , and @xmath239 ( random initial conditions , equal initial particle numbers for each species ) .
the data were obtained on a @xmath275 square lattice , averaged over @xmath265 runs . ] our two - dimensional results , from @xmath265 runs on a @xmath275 square lattice , are depicted in fig .
[ 2dcam ] .
the data become unreliable at about @xmath276 . once more
, the @xmath0-cam for @xmath26 and @xmath268 are seen to be equivalent , with @xmath263 settling towards the asymptotic value @xmath277 , albeit masked again by the competing mean - field power law , just as for the @xmath198-mam ( see fig . 6 in ref .
@xcite ) . yet
now the runs for @xmath223 and @xmath237 yield manifestly different power laws . in the three - species cam
, the effective exponent is still changing in the time window accessible to our simulations , running towards @xmath273 , perhaps with logarithmic corrections as predicted by eq .
( [ rhotaudseg ] ) . for @xmath237
, however , we find a plateau value @xmath278 , perhaps with a slowly decreasing tendency .
this may be interpreted as a combination of the predicted asymptotic decay law ( [ goldenmean ] ) with the mean - field result @xmath271 .
effective decay exponent @xmath263 for the one - dimensional @xmath0-cam with @xmath26 , @xmath188 , @xmath236 , and @xmath239 ( random initial conditions , equal initial particle numbers for each species ) .
the data were obtained on a @xmath279 lattice , averaged over @xmath265 runs . ] the simulation data in @xmath34 , from averaging over again @xmath265 runs on @xmath279 lattice sites , are reliable at least up to @xmath280 , and again confirm the equivalence of the @xmath0-cam with even @xmath0 . for both @xmath26 and @xmath268
we obtain a slow but definite approach towards the predicted @xmath281 , as for the @xmath198-mam ( see fig . 2 in ref .
@xcite ) . for the @xmath188-cam ,
equivalent to the @xmath188-mam , we find the expected slow convergence towards @xmath282 from above @xcite
. remarkably , the data for @xmath237 , for which no analytical prediction is available in one dimension , appear to reach @xmath283 as well , but faster and from below .
this is remarkably close to what eq .
( [ densdecay ] ) would predict if we applied it ( without justification ) in dimension @xmath34 , namely @xmath284 , with the segregation dimension given in eq .
( [ eqndseg5 ] ) . in summary ,
our simulation results for the @xmath0-cam , within the accuracy and statistical errors of our data , are consistent with the analytical predictions of the previous sections , if we allow for crossover effects induced by the presence of the competing mean - field decay law .
the equivalence of the @xmath236-cam and @xmath198-cam , which are both identical to the @xmath198-mam ( @xmath285 ) , is clearly demonstrated . since the @xmath188-cam and @xmath188-mam are identical , only our data for @xmath237 yield novel results .
unambiguous identification of the density exponents @xmath32 would ideally require better statistics , that is , simulations on considerably larger systems .
before concluding we wish to make a methodological remark . one may attempt to address the issue of segregation by considering space - dependent particle densities @xmath286 that satisfy the reaction - diffusion equations ( with diffusivity @xmath165 ) @xmath287 for spatially uniform ( well - mixed ) systems these reduce to the ordinary differential equations ( [ meanfieldeqns ] ) .
let us now apply to the system ( [ meanfielddiffeqns ] ) what is essentially a linear stability analysis . to this end
, we write @xmath288 $ ] and linearize with respect to the perturbations @xmath289 . with eqs .
( [ meanfieldeqns ] ) this yields @xmath290 where the neglected nonlinear terms are of order @xmath291 . in terms of the linearly transformed variables
@xmath292 with @xmath103 as defined in subsec .
[ sec2secondmoments ] , the time evolution equation ( [ eqnepsi ] ) becomes diagonal and reads @xmath293 \ , \epsilon_\mu(p,\tau ) \ .\ ] ] for given initial @xmath294 this is solved by @xmath295 irrespective of the value of @xmath106 this solution exists and tends to zero as @xmath155 . nevertheless , for @xmath296 it will increase initially . in the event that @xmath297 grows with time such as to be no longer negligible with respect to unity at some instant of time , this signals that the neglected nonlinear terms in the differential equation begin to play a role ; in particular , they will prevent @xmath289 from decreasing beyond @xmath125 and hence @xmath298 from turning negative .
we will argue , however , that in this case the physical justification for the differential equations ( [ meanfielddiffeqns ] ) breaks down , and that in fact segregation sets in .
this goal requires going beyond the mathematics and invoking the interpretation of the density as an average over discrete particles . for particles initially randomly and uniformly distributed in a volume @xmath44
, we typically have that @xmath299 . if this initial fluctuation is negative , and if in the course of time @xmath300 approaches @xmath125 , this should be interpreted properly as there being an appreciable probability for species extinction .
a necessary condition for the r.h.s . of eq .
( [ epsmusol ] ) to become of order unity reads explicitly @xmath301 , or , since @xmath106 must be negative , @xmath302 .
but on these time scales @xmath303 must still be much less than unity .
taking the smallest allowed valued for @xmath252 in the volume @xmath44 , that is , @xmath304 , the additional condition @xmath305 yields @xmath306 , which will happen when @xmath307 .
hence the differential equation approach , combined with appropriate considerations in which the particle discreteness intervenes , leads to the identical expression for the critical segregation dimension as found in eq .
( [ exprdseg ] ) of sec .
[ secfinitedimension ] .
the differential equation approach was followed , essentially , in ref .
@xcite . to our opinion , however , the fokker - planck method used in refs .
@xcite and in the present work is preferable , since it is based directly on the more fundamental description of an interacting many - body problem in terms of a master equation .
more specifically , the stochasticity taken into account in the differential equation approach is due only to the random fluctuations present in the initial state . as a consequence
, this approach wrongly suggests that the initially dominant species ( or , more precisely , their initially dominant mode @xmath122 ) is also the surviving one .
it follows from the fokker - planck equation that there is actually an interplay between initial and dynamically generated fluctuations , brought out by the solution ( [ solnflucts ] ) of the second moment equations .
this solution contains a dynamically generated contribution to the fluctuations ( _ viz . _ the terms @xmath308 ) , and a contribution due to the initial fluctuations ( the @xmath129 term ) . in systems
_ without _ segregation ( _ i.e. _ , when @xmath309 ) the dynamically generated fluctuations become _ larger _ than those due to the initial conditions , which are eventually forgotten . in systems _ with _ segregation ( for @xmath310 ) the initial and dynamical contributions are of the same order and the initial conditions co - determine the final state . in this weaker sense ,
the segregation phenomenon that appears in the cases with @xmath310 is still linked to the persistence of initial fluctuations , even in the absence of conservation laws .
we have studied a wide class of @xmath0-species reaction - diffusion systems with pair annihilation processes between distinct species .
this class includes the two well - studied paradigmatic cases @xmath14 ( in the limit @xmath311 and @xmath11 ( @xmath312 ) . within mean - field theory , _
i.e. _ for spatial dimensions @xmath58 , we have determined for each member of the class ( i ) whether or not segregation occurs , and ( ii ) the value of the decay exponent @xmath32 in the asymptotic power law for the total particle density .
our findings represent a considerable extension of previous work on segregation in diffusion - limited annihilation reactions .
our preliminary simulation data are compatible with the analytical results .
our method builds on ideas that were applied earlier in the context of @xmath0-species models by ben - avraham and redner @xcite , and more recently also by ben - naim and krapivsky @xcite and newman and mckane @xcite .
this approach is obviously not limited to the special cases studied here , but we have not aimed at being exhaustive .
it is straightforward , for example , to drop the restriction of no self - annihilation ( @xmath313 ) , and redo the analysis .
the method may also be employed to analyze more involved situations , such as models with species belonging to two distinct equivalence classes .
various open problems remain , in particular concerning the nature of the domain structure in several important cases .
furthermore , the theory presented here is unable to address the issue of what happens in low dimensions , @xmath314 , where particle anticorrelations become manifest and render the mean - field treatment invalid .
we expect that more elaborate simulations will be carried out in the future to test our predictions as well as to study questions beyond reach of the present theory .
this material is based upon work supported by the national science foundation , division of materials research , under grant no .
the laboratoire de physique thorique of the universit de paris - sud is associated with the centre national de la recherche scientifique as research unit umr 8627 . | we consider a system of @xmath0 diffusing particle species @xmath1 that are all equivalent under a symmetry operation .
pairs of particles may annihilate according to @xmath2 with reaction rates @xmath3 that respect the symmetry , and without self - annihilation ( @xmath4 ) . in spatial dimensions
@xmath5 mean - field theory predicts that the total particle density decays as @xmath6 , _ provided _ the system remains spatially uniform .
we determine the conditions on the matrix @xmath7 under which there exists a critical segregation dimension @xmath8 below which this uniformity condition is violated ; the symmetry between the species is then locally broken .
we argue that in those cases the density decay slows down to @xmath9 for @xmath10 .
we show that when @xmath8 exists , its value can be expressed in terms of the ratio of the smallest to the largest eigenvalue of @xmath7 .
the existence of a conservation law ( as in the special two - species annihilation @xmath11 ) , although sufficient for segregation , is shown not to be a _ necessary _ condition for this phenomenon to occur .
we work out specific examples and present monte carlo simulations compatible with our analytical results .
+ * pacs 05.40.-a , 82.20.-w * |
the standard model ( sm ) contains and unifies interactions of three different types of elementary particles , with respect to spin : scalars , spin one - half fermions and vector gauge bosons .
the quantum field theory for such types of particles has been successfully constructed and has passed through severe experimental examinations .
almost all phenomenological attempts to extend the sm consist in introducing of a number of new particles of the same type , like extra higgs multiplets , fourth fermion generation and various @xmath0 gauge bosons .
even susy phenomenology does not go further simple doubling of known types of particles . in this letter
we would like to discuss again the possibility of the presence in nature of new types of spin-1 chiral bosons , which are complementary to the gauge ones .
while the latter transform according to the _ real _ representation ( 1/2,1/2 ) of the lorentz group , transformation properties of the former ones are associated with the _ inequivalent chiral _ representations ( 1,0 ) and ( 0,1 ) .
the chiral spin-1 bosons are described by rank - two antisymmetric tensor fields @xmath1 , instead of vector potential @xmath2 , which describes the gauge spin-1 bosons . on the mass shell
both fields describe spin-1 particles and an equivalence relation between them can be set in the massive case @xmath3 where @xmath4 is the mass of spin-1 particles .
the tensor field describes simultaneously two three - component vector @xmath2 and axial - vector @xmath5 fields .
they can be expressed through six - component tensor field @xmath1 , using the on - shell operator equation @xmath6 , @xmath7 however , many years ago it was noted @xcite already that different yukawa interactions for different types of spin-1 bosons lead to inequivalent models .
the first successful phenomenological application of this idea has been realized for hadron meson states with spin one @xcite .
it was pointed out that _ three different _ quantum numbers @xmath8 of existing spin-1 mesons , @xmath9 , @xmath10 and @xmath11 , can not be assigned just to _ two _ vector @xmath12 and axial - vector @xmath13 quark states .
so , the rank - two antisymmetric tensor current @xmath14 was introduced , which also describes vector and axial - vector boson states , but with different transformation properties with respect to lorentz group and with different quantum numbers @xmath9 and @xmath11 , respectively .
this example demonstrates that both the pure tensor states , @xmath15 bosons , and mixed combinations of vector and tensor states , due to the same quantum numbers , @xmath16 and @xmath17 bosons , may exist .
these considerations , in general , support the idea of extension of the sm with a new type of spin-1 chiral particles @xcite , which has been proposed on pure theoretical grounds , in order to describe the anomalies in the weak charged meson decays .
assuming @xmath18 symmetry of electroweak interactions and that very heavy right - handed neutrinos @xmath19 are decoupled at the weak scale , most general yukawa interactions were proposed @xmath20\left ( \begin{array}{c } t^+_{mn } \\
t^0_{mn } \end{array } \right ) \nonumber \\ & & \hspace{-0.2cm}+\frac{u^q_a}{4 } \left(\bar{q}_a\sigma^{mn}u_{ra}\right ) \left ( \begin{array}{c } u^0_{mn } \\ u^-_{mn } \end{array } \right)+{\rm h.c.},\end{aligned}\ ] ] where the two doublets @xmath1 and @xmath21 of the new tensor chiral fields with opposite hypercharges @xmath22 , @xmath23 were introduced .
the fermion sector is the same as in the sm and consists of three generations ( noted by the index @xmath24 ) of the two component weyl spinors @xmath25 : the left - handed lepton @xmath26 and the color quark @xmath27 doublets and the right - handed lepton @xmath28 and the color quark @xmath29 , @xmath30 singlets . here and further on we omit quark color indexes as far as the tensor fields are color blind . the local gauge electroweak symmetry , maintained by the triplet @xmath31@xmath32 and singlet @xmath33 bosons , and @xmath34 color symmetry rest untouched .
it was also argued @xcite that in order to avoid new chiral anomalies connected with the antisymmetric chiral tensor fields , two higgs doublets with opposite hypercharges should be introduced instead of the one in the sm .
topologically the interactions ( [ y ] ) are the same as the electroweak gauge interactions @xmath35 where @xmath36 , @xmath37 are the gauge coupling constants , and @xmath38 , @xmath39 are the generators of @xmath40 and @xmath41 , correspondingly .
the only principal difference between them concerns fermion chiralities .
while the gauge field interactions preserve the helicities of incoming and outcoming fermions , the tensor field couplings lead to a helicity flip .
this feature should be the main signature in their experimental search .
in this section we will obtain relations among the gauge coupling constants @xmath36 , @xmath37 and the tensor yukawa couplings @xmath42 , @xmath43 and @xmath44 . in order to do this
we will apply hypothesis about dynamical generation of kinetic terms for the bosons .
this idea comes from nambu and jona - lasinio model @xcite with nonlinear four - fermion interactions .
it has been developed in @xcite for the more appropriate for quantization linear form of interactions , introducing auxiliary boson fields ( without kinetic terms ) with dimensionless yukawa coupling constants .
let us first apply this method for the gauge fields @xmath31@xmath32 and @xmath33 , assuming that they do not posses initial kinetic terms .
the kinetic terms are generated by self - energy quantum corrections from fermionic loops .
subsequent renormalization procedure leads to proper normalization of the kinetic terms and sets relations among various yukawa coupling constants .
so , in the lowest one - loop approximation the self - energy quantum corrections to the gauge fields @xmath31@xmath32 read @xmath45 \nonumber \\ & & \hspace{1.2cm}=-4g^2 i_0\left(g_{mn}q^2-q_m q_n\right)\delta^{ij } + { \cal o}(q^4),\end{aligned}\ ] ] where @xmath46 is the regularized value of the logarithmically divergent integral .
analogous calculations for the abelian gauge field @xmath33 give @xmath47 the comparison of the two previous equations leads to a relation between the gauge coupling constants @xmath48 and , hence , to a prediction of the electroweak mixing angle @xmath49 . for a first time
this result was obtained in @xcite .
this value coincides with the prediction of the gut theories ( which include the simple gauge group @xmath50 ) at the unification scale @xcite . due to the different behavior of the evolution of gauge coupling constants
@xmath36 and @xmath37 with energy , the renormalization effects lead to a decreasing of @xmath51 at low energies @xcite , bringing it closer to the experimental value @xcite @xmath52 a lack of a gauge symmetry for the chiral tensor interactions ( [ y ] ) does not allow to fix the yukawa coupling constants @xmath42 , @xmath43 and @xmath44 . therefore , to go further we must make model dependent assumptions
let us assume lepton - quark universality for the tensor interactions , which are also blind to the generations @xmath53 , @xmath54 . before proceeding with calculations of self - energy quantum corrections to the tensor fields , it is convenient to rewrite them through the more common notations of the vector potential
. this is possible using extension of the relation ( [ equivalence ] ) off - shell @xcite @xmath55 where @xmath56 .
( [ y ] ) takes the form @xmath57 now it is easy to calculate the self - energy quantum corrections to doublets @xmath58 @xmath59 and @xmath60 @xmath61 the comparison of ( [ w ] ) , ( [ t ] ) and ( [ u ] ) leads to the relation @xmath62 an asymmetry in the coupling constants for the doublets of the tensor fields arises due to their obviously non - symmetric interactions to the leptons . at high energy scale ,
when the right - handed neutrinos become active , the symmetry between both interactions should be restored .
it may be expected that the evolution of the coupling constants towards low - energy scale should not drastically change the relation ( [ relation ] ) , since both the gauge coupling constant @xmath36 and the yukawa coupling constants @xmath63 and @xmath64 have asymptotically free behavior @xcite
. we will use this relation below for a quantitative determination of the masses of the chiral particles .
the fermi theory of the weak interaction @xcite gives excellent example of effective field theory which perfectly describes low - energy physics with only one dimensional coupling constant @xmath65 gev@xmath66 .
it arises from the exchange of the charged massive intermediate bosons @xmath67 at small momentum transfer , which at the tree level is expressed as @xmath68 where @xmath69 is the boson mass . in the sm
the gauge coupling constant @xmath70 is related to the electromagnetic fine - structure constant @xmath71 and @xmath51 .
it allows to estimate @xmath72 gev at the tree level with relatively good precision .
therefore , in order to get a constraint on the mass of the new intermediate boson one can analyze the precision low - energy experiments , looking for an eventual admixture of new forces in the ordinary weak processes .
particularly , the chiral bosons can enhance the rates of the chirally suppressed decay modes
. one may think that @xmath73 decay is the ideal process to put constraints on the presence of new currents chirally different from the sm ones .
however , for the tensor current the hadron matrix element @xmath74 is zero by kinematic reasons , and this decay can restrict only the presence of pseudoscalar currents @xcite , but not tensor ones .
nevertheless , the quark tensor currents can contribute to the rare three - body pion decay @xmath75 , where huge experimental anomalies have been detected @xcite . in order to explain these anomalies ,
the following effective tensor interactions have been derived @xcite @xmath76 where @xmath77 is the momentum transfer between quark and lepton pairs and @xmath78 , @xmath79 are dimensionless coupling constants , which determine the strength of the new tensor interactions relative to the ordinary weak interactions .
these interactions arise from the exchange of the charged chiral @xmath80 and @xmath81 bosons at low energies . the first term in ( [ eff ] ) is generated due to the non - diagonal elements of mass mixing matrix @xcite @xmath82 it can be rewritten identically in a more common form without momentum dependence @xmath83 this local tensor interaction has been considered early in @xmath84 decay phenomenology on the same footing as other local four - fermion interactions . however , to my knowledge , the second non - local term in ( [ eff ] ) has never been used before 1993 . owing to the identity
@xmath85 it is impossible to construct _ local _ tensor interaction with an opposite to ( [ identity ] ) chirality of quark current .
it is worth noting , that the two terms of the tensor interactions ( [ eff ] ) arise as a result of the chiral boson exchanges and should be considered simultaneously .
the requirement of positivity of the determinant @xmath86 of the square mass matrix ( [ mtu ] ) leads to the positive effective coupling constants @xmath87 and the pseudotensor quark currents @xmath88 can completely cancel out in ( [ eff ] ) in the case @xmath89 .
this cancellation helps to avoid another constraint on the tensor interaction ( [ identity ] ) following from the pion decay due to generation of the pseudoscalar quark current @xmath90 through the electromagnetic radiative corrections @xcite . in the case of equality of the effective tensor
coupling constants , the diagonalization of the mass matrix ( [ mtu ] ) gives two mass states @xmath91 the maximum value of the lightest mass state is reached at @xmath92 and defines the physical mass @xmath93 , which corresponds to an energetically favored exchange by this particle .
then , the heavy state has a mass @xmath94 .
the centi - weak tensor interaction @xmath95 can explain the deficit of events in the radiative pion decay and , assuming @xmath96 universality , the excess in @xmath97 decay in comparison with cvc prediction @xcite .
this value of the effective tensor coupling constant allows to predict the masses of the charged chiral bosons @xmath98 which will be used in the next section for quantitative estimations of their production cross - sections .
these physical massive states are represented by two orthogonal combinations @xmath99 and @xmath100 , which correspond to light and heavy massive particles , correspondingly .
their yukawa interactions take the form @xmath101 besides the mixing ( [ mtu ] ) between the charged chiral particles themselves , they can mix with the @xmath102 boson leading to anomalous moment for the charged weak transitions @xcite .
the first experimental constraints on this parameter have been obtained in @xcite from @xmath103-lepton decays .
the mixings between the gauge and the chiral bosons are in a complete analogy with @xmath104 mixing in hadron physics @xcite . however , in contrast to the latter , where the mixing is maximal , experimental data point out that the weak mixing is rather small .
let us consider the neutral spin-1 sector of the model . here
besides electroweak bosons @xmath105 and @xmath106 we have introduced _ four _ degrees of freedom @xmath107 , @xmath108 , @xmath109 , @xmath110 for the neutral chiral bosons . due to their quantum numbers @xmath111 and @xmath112 states
do not mix and decouple from the others in the case of @xmath113 invariant interactions .
although @xmath114 and @xmath115 states do not mix between each other , because they couple to the different types @xmath116 and @xmath117 fermions , they can mix with the photon and @xmath106 boson leading to additional contributions to anomalous magnetic and neutral weak moments for the fermions . we will assume , in agreement with the experimental data , that these mixings are very small .
therefore , one can consider these states as physical with the following yukawa interactions @xmath118 in order to estimate the masses of the neutral chiral bosons we can use the mass matrix ( [ mtu ] ) . since the neutral chiral bosons belong to the same multiplets as their charged partners and they do not mix , their masses can be estimated as @xmath119 degeneracies in the masses of @xmath112 , @xmath115 and @xmath111 , @xmath114 states are removed as a result of mixings of @xmath115 and @xmath114 states with the photon and @xmath106 boson , leading to the inequalities @xmath120 and @xmath121 .
it is a hard task to detect neutral tensor interactions on the background electromagnetic and neutral weak processes in low energy experiments . up to
now there are no positive indications , besides , may be , the deviation in the muon anomalous magnetic moment , which can be assigned to mixings between the photon and the neutral chiral bosons .
nevertheless , as we will show in the next section the chiral bosons have a unique signature for their detection at hadron colliders due to their anomalous couplings with fermions .
the main interest represents the production cross - sections of the lightest chiral bosons . in order to estimate them
, we will calculate basic partial decay widths of the chiral bosons into leptons and quarks . using the yukawa interactions ( [ yc ] )
we can evaluate the lepton @xmath122 and the quark @xmath123 widths of the charged chiral bosons @xmath124 . assuming the presence of only three fermion generations with masses lighter than @xmath125 we can estimate the total width as @xmath126 gev .
then it is possible to evaluate the production cross - sections at the tevatron cross section is mainly determined by a valence quark from the proton and an anti - quark from the anti - proton , we neglect here the contribution of the sea quarks . ] @xmath127 and the lhc @xmath128 using , for example , cteq6 m parton distribution functions @xcite and the appropriated parameter @xmath129 , where @xmath130 is the square of the center - of - mass energy .
such a big production cross - sections would immediately contradict to the present tevatron data , unless the chiral bosons have unusual properties .
it is noteworthy , that the lepton width ( [ gl ] ) is suppressed in favor of the quark one ( [ gq ] ) .
hence , the lightest charged chiral particle shows leptophobic property that leads to a reduced lepton production @xmath131 which is the main channel of charged boson detection at hadron colliders . however , the cross sections ( [ stev_l ] ) are still big to remain undetected at the tevatron .
this fact can be explain by another unusual and unexpected feature of the chiral bosons connected to their anomalous interactions ( [ yva ] ) with fermions .
let us compare the normalized angular distributions of the lepton from the decays of @xmath132 @xmath133 and the @xmath124 @xmath134 where @xmath135 is the boson helicity .
for example , the left - handed quark @xmath136 ( from the proton ) interacting with the right - handed anti - quark @xmath137 ( from the anti - proton ) can produce the @xmath132 with spin projection on the proton beam direction @xmath138 .
hence , the decay leptons are distributed as @xmath139 . while chiral particle production arises from the interaction of a quark and an anti - quark with the same helicities , that leads to zero helicity of the produced chiral boson and @xmath140 lepton distribution @xcite .
indeed , the cross section for @xmath141 process @xmath142 is expressed through the relevant differential cross section of the parton subprocess @xmath143 @xmath144 where @xmath145 , @xmath146 and @xmath147 are the mandelstam variables . in the center - of - mass parton system
the differential cross section shows the following distribution @xmath148 here @xmath149 is the angle between the lepton and the parton direction and @xmath150 is the square of the transverse lepton momentum . since the latter is invariant under longitudinal boosts along the beam direction , the distribution ( [ cos2 ] ) versus @xmath151 holds also in the lab frame @xmath152 .
changing variables in the differential cross section from @xmath153 to @xmath154 @xmath155 leads to a kinematical singularity at the endpoint @xmath156 , which gives the prominent jacobian peak in the @xmath31 decay distribution .
in contrast to this , the pole in the decay distribution of the chiral bosons is cancelled out and , moreover , the distribution reaches zero at the endpoint @xmath156 .
the chiral boson decay distribution has a broad smooth bump with the maximum below the kinematical endpoint , instead of sharp jacobian peak ( fig .
[ fig:1 ] ) . in the case
when the chiral boson is produced with no transverse momentum , the transverse mass of the lepton pair is related to @xmath157 as @xmath158 and the jacobian peak is absent in @xmath159 distribution as well .
therefore , the decay distribution of the chiral bosons differs drastically from the distribution of the gauge bosons .
even relatively small decay width of the chiral bosons leads to a wide distribution , that obscures their identification as resonances at hadron colliders . at the present time
one can speak only about eventual excess of events in the region @xmath160 gev @xmath161 gev , where the background from the tail of the @xmath31 decays is considerably small . indeed , such excess , although not statistically significant , has been pointed out recently by the cdf collaboraion @xcite in the same region .
the form of the decay distribution for the chiral bosons resembles the bump anomalies in the inclusive jet @xmath162 distribution , reported by the cdf collaboration @xcite .
it has been suggested that these anomalies could be described by a new neutral leptophobic gauge boson with huge decay width @xcite .
although this problem has been solved in the framework of the sm by changing the gluon distribution functions @xcite , it could be reconsidered in the light of the new form of the decay distribution as a real physical signal from decays of different chiral bosons , both charged and neutral .
analysing the bumps in the jet transverse energy distribution in fig . 1 of ref .
@xcite , we can find the endpoint of the first bump at 250 gev and guess about the second bump endpoint from the minimum around 350 gev .
if we assign the first bump to the hadron decay products of the lightest charged bosons , which exactly corresponds to the estimated mass from eq .
( [ mhl ] ) , the second endpoint hints to a mass for the lightest neutral boson around 700 gev , which is also in a qualitative agreement with our estimations ( [ m0 ] ) .
however , taking into account the large systematic uncertainties in jet production , these conclusions may be premature .
nevertheless , the lightest neutral chiral bosons @xmath115 and @xmath112 can fairly play the role of the hadrophilic boson , suggested in @xcite in order to explain tevatron and lep anomalies , with nearly the same mass .
indeed , these bosons do not decay into charged leptons and can not be seen in lepton channel . in the best case they have invisible decay widths in neutrinos , if right - handed neutrinos are not too heavy .
however , in contrast to @xcite these bosons are non - gauge bosons , but chiral ones .
there exist also additional neutral chiral bosons @xmath114 and @xmath111 , which couple only to the _ down _ fermions with approximately the same yukawa coupling constants as @xmath115 and @xmath112 bosons .
they influence the low - energy hadron physics two times weaker than the @xmath115 and @xmath112 bosons due to their higher masses .
the production of the heaviest neutral chiral bosons @xmath114 , @xmath111 and the detection of their leptonic decays at the hadron colliders will be crucial test of the discussed model .
unfortunately , due to their big masses they can not be produced at the tevatron at present .
let us calculate their lepton and quark widths in order to estimate their production cross - section at the lhc . using the yukawa interactions ( [ yn ] ) we obtain the following decay widths @xmath163 for the lepton and @xmath164 for the quark channels .
this gives @xmath165 gev for the total fermion decay width .
since the lepton branching ratio @xmath166 is not negligible one expects essential signal in the dilepton channels with a cross section @xmath167 the most conclusive experimental evidence for the presence of anomalous interactions and new type of spin-1 non - gauge bosons will be the detection of wide bumps in leptonic distribution at high @xmath157 at tevatron and lhc and the dilepton production with high invariant mass around 1 tev . considering @xmath168 and @xmath169 channels for the decays of the lightest charged chiral bosons , one may expect around 1 event per day at tevatron ( average luminosity @xmath170 cm@xmath66s@xmath172 ) and around 1 event every 5 seconds at lhc ( designed peak @xmath173 cm@xmath66s@xmath172 ) .
taking into account also that in @xmath174-annihilation two different species of the neutral chiral bosons @xmath114 and @xmath111 are produced with approximately the same masses , one may expect one electron or muon pair with invariant mass around 1 tev every 25 second at lhc peak luminosity . in conclusion
, we would like to note that our numerical results essentially depend on the magnitude of the low - energy effective tensor coupling constant @xmath78 . according to our estimations
@xcite its value can be up to 30% larger , which effectively decreases all the calculated masses of the chiral bosons by 15% . however , the mass ratios of the chiral bosons @xmath175 are fixed by the model .
really , the masses of the charged @xmath176 , @xmath177 and neutral @xmath115 , @xmath114 chiral bosons can get an additional positive contributions due to the small mixings between the gauge and the chiral bosons , which have been neglected here
. taking into account the unusual lepton distribution ( [ cos2 ] ) , the main attention should be paid to events with large pseudorapidities , where the dominant lepton production occurs as a result of decays of the new chiral particles . to support these qualitative statements and to be ready to meet eventual new physics ,
detailed quantative calculations should be fulfilled on the basis of the interactions ( [ yc ] ) and ( [ yn ] ) , taking into account also all possible known effects and specifics of concrete detector .
i am grateful to d. kirilova for the overall help and appreciate the stimulating ictp environment .
99 n. kemmer , proc .
roy . soc .
a * 166 * , 127 ( 1938 ) .
m. v. chizhov , hep - ph/9610220 .
m. v. chizhov , mod .
lett . a * 8 * , 2753 ( 1993 ) .
y. nambu and g. jona - lasinio , phys . rev . *
122 * , 345 ( 1961 ) .
t. eguchi , phys .
d * 14 * , 2755 ( 1976 ) .
h. terazawa , k. akama and y. chikashige , prog .
. phys . * 56 * , 1935 ( 1976 ) ; t. saito and k. shigemoto , prog .
. phys . * 57 * , 643 ( 1977 ) . h. georgi and s. l. glashow , phys .
lett . * 32 * , 438 ( 1974 ) .
h. georgi , h. r. quinn and s. weinberg , phys .
* 33 * , 451(1974 ) .
et al . _ ,
journal of physics g * 1 * , 1 ( 2006 ) .
l. v. avdeev and m. v. chizhov , phys .
b * 321 * , 212 ( 1994 ) .
e. fermi , z. phys .
* 88 * , 161 ( 1934 ) ; nuovo cim .
* 11 * , 1 ( 1934 ) . o. shankar , nucl .
b * 204 * , 375 ( 1982 ) . v. n. bolotov _
et al . _ , phys .
b * 243 * , 308 ( 1990 ) ; e. frle _ et al._. phys . rev . lett .
* 93 * , 181804 ( 2004 ) .
m. b. voloshin , phys .
b * 283 * , 120 ( 1992 ) .
m. chizhov , hep - ph/0311360 .
m. chizhov , hep - ph/9612399 , hep - ph/0310203 .
p. abreu _ et al .
j. c * 16 * , 229 ( 2000 ) .
j. pumplin _
et al . _ ,
jhep * 07 * , 012 ( 2002 )
. m. v. chizhov , hep - ph/0008187 .
j. kim , ph.d .
thesis , fermilab - thesis-2005 - 75 , june 2005 ; the cdf collaboration , note 7459 , october 25 , 2005 . f. abe _
et al . _ ,
lett . * 77 * , 438 ( 1996 ) .
p. chiappetta _
et al._. phys .
d * 54 * , 789 ( 1996 ) ; g. altarelli _ et al._. phys .
b * 375 * , 292 ( 1996 ) .
j. huston _
et al . _ ,
lett . * 77 * , 444 ( 1996 ) ; h. l. lai _
et al._. phys .
d * 55 * , 1280 ( 1997 ) .
m. v. chizhov , phys .
* 2 * , 193 ( 2005 ) . | the possibility of production of new spin-1 chiral particles at the hadron colliders the tevatron and the lhc is considered .
their coupling constants to ordinary fermions can be fixed using the hypothesis about a dynamical generation of kinetic terms for the bosons and assuming an universality of interactions of these particles .
the masses of the chiral particles can be predicted from experimental data of precise low - energy experiments , which already indicate indirectly the existence of such new particles .
quantitative estimations of the production cross - sections of the chiral particles are made .
it is noteworthy that , due to their non - gauge interactions with fermions , their decays do not lead to the jacobian peak in the transverse momentum / mass distribution .
moreover , the lightest chiral particles show leptophobic character , which makes their detection at hadron colliders challenging . |
when the beam position change takes place during the conventional global orbit correction processes , the photon beam through the beamline is affected , and it results in the alignment of mirrors and monochromators .
this is particularly severe for a long beamline such as the undulator beamlines .
this problem can be overcome by introducing a local bump at the particular beamline .
however , for some light sources , there are not enough corrector magnets to generate local bumps as much as needed .
this difficulty can be overcome when we correct the cod under the condition where the beam positions at particular points are not changed .
this is our main objective to develop a method of the closed orbit correction under constraint conditions .
the beam position is normally described as a vector @xmath0 measured by @xmath1 beam position monitors ( bpm ) . in order to correct the cod ,
we need @xmath2 corrector magnets with their strengths described as a vector @xmath3 .
when the corrector magnets kick a beam , the new beam positions @xmath4 can be described as follows.@xmath5 here , @xmath6 is called the response matrix of @xmath7 dimensions whose components are given by @xmath8 where @xmath9 is the betatron tune of the storage ring , and @xmath10 and @xmath11 are the beta function and the phase function for the @xmath12 bpm and @xmath13 corrector magnet , respectively . in order to reduce the cod , we have to choose the kick of each corrector magnet satisfying @xmath14 it is called the psinom algorithm@xcite .
the cod correction is actually a minimization procedure of @xmath15 defined as @xmath16 by using the relations of vector operators which are hyper - dimensional gradient operators , we can get the same result of psinom algorithm as shown in eq .
( [ coc_eqn_ordinary_cod_relation ] ) .
this algorithm includes an inversion procedure of the matrix @xmath17 . in some cases , we can get unacceptable corrections due to the ill - posedness of @xmath17 . a regularization method is introduced to avoid this problem . in this case , @xmath15 is written as @xmath18 where @xmath19 is the regularization parameter . then the minimum corrector kicks can be determined by @xmath20 this equation represents the modified psinom algorithm , and we can relax the inversion problem of the singular matrix by using the diagonal matrix @xmath21 when @xmath17 is singular@xcite . after having the new closed orbit with the minimum distortion from eq .
( [ coc_eqn_modified_psinom ] ) , the new orbit is generally different from the original orbit . sometimes , this difference can be taken place at very sensitive locations such as the entrance and the exit of an undulator .
if the beamline is well aligned for this undulator , a cod correction should be avoided in this region .
a constraint condition can be described in terms of the beam position at @xmath22 bpm such as @xmath23 here , @xmath24 is the @xmath22 row of the response matrix @xmath6 .
also , @xmath25 and @xmath26 are the beam positions before and after the correction , respectively .
since we want to keep this position unchanged , @xmath27 should be zero .
if there are @xmath28 bpms involved in the constraint condition , we can write the constraint condition as follows .
@xmath29 here , @xmath30 is the ( @xmath31 ) sub - matrix of the response matrix .
each component of @xmath30 corresponds to the bpm involved in the constraint condition .
we also assume that @xmath3 has a non - trivial solution .
we now add this constraint condition to the modified psinom algorithm to obtain the new @xmath15 such as , @xmath32 \cr + \frac{1}{2 } \left < { \rm\bf k } \right| \alpha \left| { \rm\bf k } \right > + \left
< { \rm\bf \gamma } \right| { \rm\bf c}^t \left| { \rm\bf k } \right > .
\end{aligned}\ ] ] here , @xmath33 is the lagrangian multiplier , and it is an @xmath28 dimensional vector . by following the derivative to the corrector strength , we can get the vector @xmath3 which minimizes the closed orbit distortion outside the constraint region such as , @xmath34 here , we define the square matrices of @xmath35 dimensional @xmath36 and @xmath37 dimensional @xmath38 as follows .
@xmath39 eq .
( [ coc_eqn_new_correction ] ) can be rewritten as @xmath40 now , we can remove the lagrangian multiplier @xmath41 in eq .
( [ coc_eqn_new_correction ] ) by using the above equation .
then , we can finally get the kick values of the corrector magnets as follows .
the algorithm developed in the previous section has been successfully tested for the pohang light source ( pls ) operation .
the new correction code is written in c language and it is installed in one of the operator consoles which is a sun workstation . although the pls control system is not using experimental physics and industrial control system ( epics ) which is used in many accelerator laboratories world - widely , there is a plan to upgrade the control system based on epics technology@xcite . as a part of such upgrade activity
, we have started the development orbit correction algorithm with epics .
since we do not have any epics - based control system yet , we decide to develop the orbit correction simulator using epics . in order to seek the way to adopt our orbit correction algorithm into epics , we have considered two ways : one is based on the subroutine record and the other is using the state notation language ( snl ) program .
the first method needs new record support that runs the orbit correction algorithm . to do this
, the correction code is required to be written upon the protocol required by the record support such as the entry structure and the callback structure .
this approach gives relatively fast response because this method uses database access and can access the record by process passive mode@xcite .
this is a good feature for the realtime system but the large portion of the orbit correction code we have already tested must be rewritten according to the protocol of the record support .
one of important tasks of epics input / output controller ( ioc ) is the sequencer that runs programs written in snl .
the snl considers the control object as the state machine and treats transitions between states .
the sequencer monitors the transitions for the snl and runs callback functions using the entry table the corresponding program written in snl .
since the sequencer accesses to the record via channel access ( ca ) and the record access is only possible through non - process passive mode , there are some restrictions in access time or the treatment of records .
however , this method can directly imbed the program written in c language . also ,
unlike the subroutine record , this method can remove program tasks without rebooting the system , which gives the code debugging very easy@xcite .
thus , the second method gives more benefits when the system does not require heavy realtime demands . upon reviewing the two methods , we have decided to use the latter method .
the orbit correction simulator is developed in ioc level and has the snl program imbedding c codes and several database records , as shown in fig .
[ fig_schemetic ] .
there are two parts in the snl program : one for the feedback including the orbit correction algorithm and the other for the simulator which emulates the pls storage ring .
the latter calculates orbit changes from _ ai : kick$(kick_no ) _ records which store the corrector strengths obtained from the response matrix measured at the pls .
this process is actually a product of matrix and column vectors .
the calculated orbit changes are stored at _ ai : corrorbit$(bpm_no ) _ records . in the next step ,
the corrector strengths s calculated by the algorithm described in the previous section and the results are stored at _ al : kick$(kick_no)_. the snl program we have developed consists of four state sets . since they are linked together and interacted dynamically via epics database , we need to understand the database and its records as well as the linkage between them . the _ ai : kick$(kick_no ) _ record uses the analog input record to represent the corrector strength . there are 70 records altogether as representing 70 correctors in the pls storage ring .
they are linking the simulator and the feedback in a parallel manner .
the index _ $ ( kick_no ) _ is the integer value between one and 70 . the records of _ ai : orgorbit$(bpm_no ) _ and _ ai : corrorbit$(bpm_no )
_ are the orbit distortion before the feedback is applied and the orbit change after the feedback , respectively .
they are using analog input records . since these records represents the changes at the pls bpm , _ $ ( bpm_no ) _ is the interger value between one and 108 .
the superposition of these two records gives new orbit .
this can be done by the calculation record _ calc : orbit$(bpm_no)_. since this calculation record is linked forwardly from _ ai : orgorbit$(bpm_no ) _ and _ ai : corrorbit$(bpm_no ) _ , the superposition
is newly calculated whenever the values of two records are changed .
on the other hand , _ ai : numbpm _ , _ ai : numcorrector _ , _ ai : startconstr _ , and _ ai : endconstr _ are representing the bpms and correctors used in the correction algorithm , start and end point of the constraint region , respectively .
they are using analog input records and notifying the snl program if necessary .
the other records are binary input records and they are representing necessary status flags .
they are used as the mediator of state transitions between state sets in the snl program .
we have developed the cod correction algorithm under the constraint condition where the beam position at particular point is not changed .
the new algorithm is based on the modified psinom algorithm which includes the regularization process in order to avoid the inversion problem of the ill - posed response matrices .
we have confirmed that this algorithm is working well and is in good agreement with the experimental results @xcite . even though the pls is planning to upgrade their control system with epics ,
there is no working epics based control system at pls . due to this , we have developed the orbit correction simulator using c - code embedded snl program based on epics technology .
this simulator part can be replaced by the real control system with minor changes after the completion of the upgrade .
1 w. herr : " algorithms and procedures used in the orbit correction package cocu , " cern sl/95 - 07 ( ap ) , 1995 . y. n. tang and s. krinsky : proc
. aip conf .
* 315 * ( aip press , 1993 ) 87 .
see url"_http://www.aps.anl.gov / epics _ " martin r. kraimer : " epics ioc application developer s guide , " aps / anl , 1998 andy kozubal : " state notation language and sequencer user guide , " lanl , 1995 philip stanley , _ et al . _ : " epics record reference manual , " lanl , aps / anl , 1995 kukhee kim , jinhyuk choi , tae - yeon lee , guinyun kim , moohyun cho , won namkung , in soo ko : jpn . j. appl
* 40 * ( 2001 ) 4233 . | we have carried out the basic research for the accelerator and tokamak control system based on the experimental physics and industrial control system ( epics ) .
we have used the process database and the state notation language ( snl ) in the epics to develop the simulator which represents as a virtual machine . in this paper . in this paper
, we introduce the simulator of the global orbit feedback system as an example .
this simulates the global orbit feedback system under the constraint conditions for pohang light source ( pls ) storage ring .
we describe the details of the feedback algorithm and the realization of the simulator . |
p2-na@xmath7coo@xmath2 ( also called @xmath8-phase ) is composed of alternating layers of na and coo@xmath2 with a hexagonal in - plane network that can be described by space group p6@xmath9/mmc.@xcite na vacancy ordering has been an important subject both experimentally and theoretically following the discovery of rich phase diagram of na@xmath7coo@xmath2,@xcite from the a - type antiferromagnetic ( af ) ordering for [email protected] , to af and metal - to - insulator transitions near x=0.5 , and the superconductivity found in the hydrated sample of x @xmath11 1/3.@xcite in particular , the discovery of na superlattice described by the multi - vacancy cluster model proposed by roger _
_ has generated additional interest in the exploration of the origin of na vacancy ordering and its impact on the title compound.@xcite we have demonstrated that the particularly stable phases found near 0.71 , 0.50 , 0.33 , and 0.25 also show sharp drop of the na atom diffusion coefficients at room temperature.@xcite however , it is quite surprising that the commonly predicted stable phases near x = 3/4 and 2/3 do not exist , as revealed by our single crystal electrochemical de - intercalation studies at room temperature .
the na superlattice as a result of na ion ordering has been solved satisfactorily for x @xmath11 0.833 , 0.820 , 0.77 and 0.71 so far , based on na multi - vacancy model of di- , tri- , and quadri - vacancy cluster ordering.@xcite for example , x = 0.71 is composed of alternating na layers with tri- and quadri - vacancy clusters ordered in a large simple hexagonal superlattice of size @xmath5a.@xcite the implication of such na vacancy ordering has helped us understand the origin of small cross sections of the fermi surface hole pockets observed by shubnikov - de haas oscillation and the intriguing coexistence of metallic conductivity with curie - weiss behavior .
this led us to propose that the na vacancy ordering creates one special partial localization environment for the doped holes , i.e. , part of the charge is localized near the vacancy center and the rest are itinerant throughout the vacancy - free region .
while na vacancy is unavoidable for na@xmath7coo@xmath2 samples prepared under ambient pressure , even under pure oxygen environment , the most probable na content for samples prepared using solid state reaction is usually in the range of 0.75 - 0.85.@xcite different vacancy ordering patterns for x = 2/3 have been proposed based on density functional theory calculations.@xcite however , single crystal diffraction studies on na@xmath7coo@xmath2 have not revealed any signature of na ion ordering for x = 2/3,@xcite on the other hand , platova _ et al .
_ have proposed based on nmr and x - ray diffraction results , that a simple hexagonal superlattice of @xmath5a exists in na@xmath0coo@xmath2 where powder sample were carefully prepared under oxygen environment at 700@xmath12 c and quenched.@xcite the proposed microscopic structure is very different from the tri- and quadra - vacancy layer model we proposed for na@xmath4coo@xmath2 .
it is very surprising that the physical properties of their samples are very similar to ours despite the difference in na concentration and their proposed superlattice size is identical to that found in na@xmath4coo@xmath2 based synchrotron x - ray laue and electron diffraction studies on single crystal sample with electrochemically fine - tuned x.@xcite we have pointed out previously that the controversy could have resulted from differences in oxygen content,@xcite i.e. , the identical @xmath5a simple hexagonal superlattice is a result of identical average co valence between na@xmath4coo@xmath2 and na@xmath0coo@xmath3 .
herein , we provide conclusive evidence to resolve the controversy regarding the @xmath5a superlattice ordering observed in x = 2/3 .
it turns out that oxygen content plays an important factor that can not be ignored when discussing all related physical properties in the na@xmath7coo@xmath1 system .
the nature of immobile and randomly distributed oxygen vacancy at room temperature renders the na ion ordering more complicated , especially when the ordering along the c- axis ( staging ) is considered as well.@xcite
single crystal samples have been prepared from floating zone method growth , following an additional electrochemical de - intercalation technique as reported previously.@xcite oxygen environment with pressure higher than 8 atm has been kept in the optical floating - zone growth chamber to reduce na loss due to evaporation and the generation of co@xmath9o@xmath13 impurity phase in particular , to prevent the occurrence of oxygen deficiency .
high quality single crystal na@xmath7coo@xmath2 of x @xmath11 0.84 could be obtained reproducibly starting from polycrystalline feed rod prepared from na@xmath2co@xmath9 and co@xmath9o@xmath13 of na : co=0.85:1 ratio .
samples with na content of 0.71 and 2/3 were prepared from electrochemical de - intercalation as reported previously,@xcite where the as - grown crystal sample of na@xmath14coo@xmath2 was used as working electrode , pt as counter electrode , and ag / agcl with sieved liquid junction as reference electrode .
the applied voltages relative to ag / agcl refer ence electrode were 0.05v and 0.20v for the preparation of x=0.71 and 2/3 , respectively .
electron probe microanalysis ( epma ) has been performed on freshly cleaved inner surfaces to verify the chemical composition of each element .
perkin - elmer tg / dta was used for thermal gravimetric analysis .
synchrotron x - ray laue diffraction for superstructure investigation was performed at the nsrrc , taiwan and magnetic susceptibility measurement was done using quantum design squid - vsm .
( color online ) tga scan for na@xmath4coo@xmath2 and na@xmath0coo@xmath2 single crystal prepared by electrochemical technique .
the scan has been done with warming / cooling rate of 0.5k / min in an oxygen environment.,width=336 ] na@xmath4coo@xmath2 sample can be prepared through electrochemical de - intercalation method precisely , as indicated by the diverging diffusion coefficient derived previously.@xcite however , the target na concentration for na@xmath0coo@xmath2 is very hard to obtain precisely , partly because of its metastable nature .
we have prepared sample with x = [email protected] and verified by both the x - ray lattice parameter analysis of linear c(x ) extrapolation reported previously and epma to confirm this crucial concentration for this study .
additional oxygen control has been performed using tga scan under various gas flows , including oxygen and air , up to 950k .
the as - prepared single crystal samples of nearly oxygen deficiency - free samples for x = 0.71 and x = 2/3 have been ground into powder and go through additional annealing under tga as shown in fig . [
fig : fig - tga ] .
we find that a significant weight loss occurs for x = 2/3 only near 700k and this weight loss is recoverable under oxygen . on the other hand , x = 0.71 sample is stable at the same temperature range , whether the flowing gas is oxygen or air . the reversible weight loss for x = 2/3 sample near 700k is about 0.30 wt@xmath16 , which corresponds to an oxygen loss of 0.02 .
interestingly , the co valence of na@xmath4coo@xmath2 and na@xmath0coo@xmath3 are identical , and this thermal analysis suggests that the most stable average co valence is near + 3.29 . to prepare oxygen deficient na@xmath0coo@xmath3 single crystal sample for further study , we annealed the as - prepared na@xmath0coo@xmath2 single crystal sample at 700 @xmath12c for 12 hours and quenched , similar to the method used by platova _
et al . _ on their powder sample.@xcite we believe that the powder sample studied by platova _
_ has an oxygen deficiency level about 0.02 as described above .
( color online ) the laue patterns for ( a ) na@xmath4coo@xmath2 , ( b ) na@xmath0coo@xmath2 , and ( c ) na@xmath0coo@xmath3 taken along [ 001 ] direction at room temperature .
the strong diffraction peaks are indexed as [ 100 ] and [ 010 ] of the primitive cell with size * a * constructed by co ions , and the six superlattice peaks in hexagonal shape surrounding each primitive index correspond to a real lattice of size @xmath5a . , width=336 ] it is instructive to compare the laue diffraction patterns of na@xmath4coo@xmath2 , na@xmath0coo@xmath2 , and na@xmath0coo@xmath3 as shown in fig . [ fig : fig - laue ] .
the laue pattern of na@xmath4coo@xmath2 shows perfect superlattice of @xmath5a as reported previously.@xcite the superstructure of na@xmath4coo@xmath2 has been modeled to be composed of alternating layers of sodium tri- ( t ) and quadri - vacancy ( q ) clusters ordered in simple hexagonal shape.@xcite additional structure refinement suggests that these alternating layers of tqtqtq stacking has a 3c periodicity described by a right - hand screw symmetry advanced na - trimers along the c - axis.@xcite while na@xmath0coo@xmath2 shows no superlattice formation , as indicated in fig .
[ fig : fig - laue](b ) , na@xmath0coo@xmath3 reveals a superstructure nearly identical to that of na@xmath4coo@xmath2 in the reciprocal space .
we note however , that detailed comparison of laue patterns in fig .
[ fig : fig - laue ] shows some systematic missing of superlattice diffraction planes for na@xmath0coo@xmath3 as comparing with that of na@xmath4coo@xmath2 , which suggests that even though these two samples possess identical superlattice size of @xmath5a , subtle difference must exists as suggested by the nmr studies.@xcite these results seem to suggest that similar na ion ordering can be achieved as long as the average co valence is close to + 3.29 .
experimentally na@xmath4coo@xmath2 has been shown to be the most stable phase as indicated by its significant sharp drop of the diffusion coefficient and the convincingly refined crystal structure of na tri- and quadri - vacancy cluster stacking of @xmath5a@xmath6@xmath5a@xmath63c superlattice ordering,@xcite the observed identical superlattice size suggests that na@xmath0coo@xmath3 can very well be a defect type variation of the ideal na@xmath4coo@xmath2 . to provide a more quantitative description to the current finding ,
let us take as an empirical fact that na@xmath4coo@xmath2 with the tri- and quadri - vacancy ordering is most stable .
the na@xmath0coo@xmath3 with oxygen vacancy means that for every 48 co atoms , one oxygen is missing , i.e. , 1 out of 4 @xmath5a@xmath6@xmath5a supercell contains an oxygen vacancy .
in order not to pay too much coulomb energy for the o@xmath17 vacancy two additional na vacancies will appear .
then on average the two tri - vacancy layers become quadra - vacancies resulting in 8 na per 12 cobalt atoms , i.e. , we are led to a defect type na@xmath4coo@xmath2 structure of na@xmath0coo@xmath3 stoichiometry .
another way of saying is that the oxygen vacancy is frozen in place randomly but the na is mobile at room temperature .
the additional missing na ( relative to 0.71 ) will be bound close to the missing oxygen in order to maintain the @xmath5a superlattice .
the reason for this stability presumably has large contribution from madelung energy , so it is at least partly electron driven .
in fact nicolaou _
have also proposed that the stacking layer potential could create a special charge ordering within the coo@xmath2 layer based on their arpes experimental results.@xcite the superlattice found in na@xmath0coo@xmath3 must be robust because it is prepared from a high temperature quenching procedure of easy disorder introduction , while na@xmath0coo@xmath2 is prepared from a slow cooling process under oxygen atmosphere .
we must note that oxygen vacancy is not mobile at room temperature while na ion self diffusion is active above @xmath11150 - 200k .
the randomly distributed oxygen vacancy at the level of @xmath110.02 per formula unit for na@xmath0coo@xmath3 must be tolerated in order to allow na layer to form a superstructure similar to that of na@xmath4coo@xmath2 , it is difficult to understand within the proposal of platova _
et al_. for na@xmath0coo@xmath2 why oxygen vacancies are needed to stabilize the superlattice structure . in any case , based on our findings it is clear that the role of oxygen deficiencies need to be accounted for in the modeling of the @xmath18 material .
.[tab : tablei]summary of magnetic susceptibility data analysis for na@xmath0coo@xmath2 , na@xmath0coo@xmath3 , and na@xmath4coo@xmath2 fitted using equation @xmath19+c/(t-@xmath20 ) in the range of @xmath11100 - 300k . [ cols="^,^,^,^,^,^",options="header " , ] magnetic susceptibility measurement results are summarized in fig .
[ fig : fig - chi ] .
the displayed data are powder average of the single crystal measurement , i.e. , 2/3@xmath21 + 1/3@xmath22 without considering the minor correction of in - plane anisotropy .
there is no magnetic anomaly found down to 2k for all three samples .
the curie - weiss behavior for a system with metallic property has long been a puzzle,@xcite and it transforms from curie - weiss to pauli - like behavior only when x is reduced to the level of nearly half - filled na layer.@xcite based on our proposed na vacancy cluster model published previously,@xcite na tri- and quadri - vacancy cluster superlattice ordering could be responsible for the curie - weiss metal behavior , i.e. , part of the doped holes are localized near the na vacancy cluster center while the rest are itinerant .
both localized and itinerant electrons should contribute to the magnetic susceptibility .
we propose that the total magnetic susceptibility should include contributions from temperature independent core diamagnetic @xmath23 , van vleck paramagnetic @xmath24 , pauli paramagnetic @xmath25 , and the spin susceptibility from localized spins which are tentatively described by a curie - weiss like behavior of c/(t-@xmath20 ) .
the curie - weiss behavior for all three samples has been fitted following an equation of @xmath19+c/(t-@xmath20 ) in the range of 100 - 300k quite satisfactorily as seen in the inset of fig .
[ fig : fig - chi ] , where c corresponds to the number of localized spins , weiss temperature @xmath20 indicates the mean field average of the magnetic coupling strength among isolated spins , and @xmath19 = @xmath23+@xmath24+@xmath25 .
the fitted parameters for @xmath26(t ) described by the curie - weiss law are shown in table [ tab : tablei ] .
the fitted @xmath19 values do not provide reliable physical meaning before each individual contribution can be analyzed .
it is not surprising to find that c and @xmath20 are about the same for na@xmath4coo@xmath2 and na@xmath0coo@xmath3 , which implies that the quantity and the coupling strength of isolated spins are nearly identical .
na@xmath0coo@xmath2 shows less isolated spins even though more holes are introduced compared with na@xmath0coo@xmath3 .
clearly , formation of long - range na multi - vacancy cluster ordering such as that demonstrated by the ideal na@xmath4coo@xmath2 , is prohibited in na@xmath0coo@xmath2 , unless oxygen vacancy is introduced .
the co - existence of local moments and itinerant electrons is a complicated many - body problem which is not fully understood .
the data fitting using curie - weiss law plus pauli is only a crude beginning .
for instance , the @xmath27 coefficient found for na@xmath4coo@xmath2 corresponds to a @xmath28 local moment per doped hole , which is clearly inconsistent with metallic behavior .
we also acknowledge that another point of view has been proposed in the literature , i.e. , the curie - weiss behavior signifies an almost ferromagnetic metal@xcite , as described , for instance , by the self consistent renormalized ( scr ) spin fluctuation theory.@xcite however , in the standard scr theory , the spin susceptibility saturates at low temperatures leading to fermi liquid behavior whereas in the present system @xmath26 continues to rise down to 4 k ( as seen in fig.3 ) .
such a low fermi liquid scale is not expected without fine tuning .
furthermore , the resistivity exhibits a linear @xmath29 dependence below 100 k and cross - over to @xmath30 below 4 k , while the specific heat exhibits @xmath31 dependence.@xcite at the moment , neither the itinerant spin fluctuation model nor the partial local moment picture is capable of explaining these anomalies .
we leave this complicated and interesting issue open for future discussions .
the main point of the present analysis is simply that the magnetic properties of na@xmath4coo@xmath2 and na@xmath0coo@xmath3 are very similar , while that of na@xmath0coo@xmath2 has distinctly different @xmath27 and @xmath20 .
we have also performed 4-probe resistivity measurements on the na@xmath4coo@xmath2 , na@xmath0coo@xmath2 , and na@xmath0coo@xmath3 crystals as shown in fig [ fig : fig - resistivity ] .
for na@xmath4coo@xmath2 , the resistivity shows metallic behavior and drops with decreasing temperature . there is a crossover at about 100 k ,
below which the resistivity becomes practically linear in t. similar behavior has also been observed in na@xmath0coo@xmath2 and na@xmath0coo@xmath3 .
however , such deviation from the t - linear dependence at low temperature is evident and more significant in na@xmath0coo@xmath2 .
in addition , we note that the resistivity value is more than twice as large in na@xmath0coo@xmath2 at low temperature , which suggests a shorter carrier lifetime due to a more disordered phase in na@xmath0coo@xmath2 .
these observations are in accord with the absence of na superstructure ordering as mentioned earlier and also imply an intimate correlation between the na ordering and the co la yer electronic structure . longitudinal magnetoresistance ( lmr ) and transverse magnetoresistance ( tmr ) were measured for the three samples , where lmr and tmr were obtained under a magnetic field up to 14 tesla applied parallel and perpendicular to the coo@xmath2 plane , respectively .
the orbital magnetoresistance ( omr ) was determined by subtracting tmr with lmr in order to exclude the spin related contribution to the tmr .
a comparison of lmr and omr for the three samples is shown in fig .
[ fig : fig - mr ] .
the omr is positive for all three samples with quadratic magnetic field dependence as expected from orbital contribution , which is simply proportional to the square of carrier mobility .
na@xmath0coo@xmath2 shows smaller omr than that of na@xmath4coo@xmath2 and na@xmath0coo@xmath3 , which infers a shorter carrier lifetime in a more disordered na@xmath0coo@xmath2 as expected . on the other hand ,
lmr is typically derived from spin effect , where the negative lmr is most likely associated with the suppression of spin scattering due to the field alignment of localized moment .
we find that the lmr in na@xmath0coo@xmath2 is nearly half of that in na@xmath4coo@xmath2 and na@xmath0coo@xmath3 at low temperature , which suggests less spin scatterers in the system .
again , the main point is the similarity between na@xmath4coo@xmath2 and na@xmath0coo@xmath3 and their contrast with na@xmath0coo@xmath2 .
we also remark that the extraction of carrier density from hall coefficient data is not possible because of its strong temperature dependence due to strongly correlated nature .
in summary , using single crystal samples of well controlled oxygen and sodium content , we have provided strong evidence to resolve the controversy of superlattice ordering found in na@xmath7coo@xmath2 of x=0.71 and 2/3 .
synchrotron x - ray laue diffraction results indicate that there is no sodium ordering in na@xmath0coo@xmath2 at room temperature , while na@xmath0coo@xmath3 , which has identical co valence to that of na@xmath4coo@xmath2 , shows identical @xmath5a simple hexagonal superstructure to that of na@xmath4coo@xmath2 , also with nearly identical magnetic and transport properties .
these results imply that na@xmath4coo@xmath2 with @xmath5a@xmath6@xmath5a@xmath63c superlattice is the most stable structure at this particular concentration , while na@xmath0coo@xmath3 of identical average co valence possesses a nearly identical na superstructure and magnetic properties with the help of oxygen defect generation .
this proposed model is supported in part by the higher resistivity found in na@xmath0coo@xmath2 without na superstructure .
studies on na ion ordering based on screened inter - sodium interaction without considering the involved oxygen deficiency could result in misleading prediction or interpretation .
gjs and fcc acknowledge the support from national science council of taiwan under project number nsc-98 - 2119-m-002 - 021 .
we also thank chung - yuan kao in ntu for the help on epma measurement .
pal acknowledges the support of doe grant de - fg02 - 03er46076 .
+ claude fouassier , guy matejka , jean - maurice reau , and paul hagenmuller , j. solid state chem 6 , 532 ( 1973 ) .
maw lin foo , yayu wang , satoshi watauchi , h.w .
zandbergen , tao he , r. j. cava , and n. p. ong , phys .
92 , 247001 ( 2004 ) .
kazunori takada , hiroya sakurai , eiji takayama - muromachi , fujio izumi , ruben a. dilanian , and takayoshi sasaki , nature 422 , 53 ( 2003 ) .
m. roger , d. j. p. morris , d. a. tennant , m. j. gutmann , j. p. goff , j .- u .
hoffmann , r. feyerherm , e. dudzik , d. prabhakaran , a. t. boothroyd , n. shannon , b. lake , and p. p. deen , nature 445 , 631 ( 2007 ) .
f. c. chou , m. -w .
chu , g. j. shu , f .-
huang , woei wu pai , h. s. sheu , and patrick a. lee , phys .
lett . 101 , 127404 ( 2008 ) .
g. j. shu and f. c. chou , phys .
b 78 , 052101 ( 2008 ) .
g. j. shu , andrea prodi , s. y. chu , y. s. lee , h. s. sheu , and f. c. chou , phys .
b 76 , 184115 ( 2007 ) . f .-
huang , m .- w .
chu , g. j. shu , h. s. sheu , c. h. chen , l .- k .
liu , patrick a. lee , and f. c. chou , phys .
b 79 , 014413 ( 2009 ) .
g. j. shu , f .-
huang , m .- w .
chu , j .- y .
lin , patrick a. lee , and f. c. chou , phys .
b 80 , 014117 ( 2009 ) .
l. balicas , y. j. jo , g. j. shu , f. c. chou , and p. a. lee , phys .
lett 100 , 126405 ( 2008 ) .
t. motohashi , m. karppinen , and h. yamauchi , arxiv : cond - mat/0304479v1 , ( 2003 ) .
ying s. meng , yoyo hinuma , and gerbrand ceder , j. chem .
128 , 104708 ( 2008 ) .
peihong zhang , rodrigo b. capaz , marvin l. cohen , and steven g. louie , phys
b 71 , 153102 ( 2005 ) .
t. a. platova , i. r. mukhamedshin , h. alloul , a. v. dooglav , and g. collin , phys .
b 80 , 224106 ( 2009 ) .
h. alloul , i. r. mukhamedshin , t. a. platova , and a. v. dooglav , euro .
85 , 47006 ( 2009 ) .
a. nicolaou , v. brouet , m. zacchigna , i. vobornik , a. tejeda , a. taleb - ibrahimi , p. le fevre , f. bertran , c. chambon , s. kubsky , s. hebert , h. muguerra , and d. grebill , eur .
lett . , 89 , 37010 ( 2010 ) m. h. julien , c. de vaulx , h. mayaffre , c. berthier , m. horvatic , v. simonet , j. wooldridge , g. balakrishnan , m. r. lees , d. p. chen , c. t. lin , and p. lejay , phys .
lett . , 100 , 096405 ( 2008 ) .
y. takahashi , j. phys .
, 55 , 3553 ( 1986 ) .
h. alloul , i.r .
mukhamedshin , g. collin , and n.blanchard , euro .
82 , 17002 ( 2008 ) .
t.morita , _ spin fluctuations in itinerant electron magnetism _ , springer - verlag , 1985 . | the impact of oxygen deficiency on physical properties of na@xmath0coo@xmath1 has been investigated . from the combined thermogravimetric , magnetic susceptibility and synchrotron x - ray laue diffraction studies , it is demonstrated that na@xmath0coo@xmath2 shows no superlattice ordering due to na ions ; however na@xmath0coo@xmath3 , which has the same co valence as that of na@xmath4coo@xmath2 , shows nearly identical magnetic and transport properties and the same simple hexagonal superlattice ordering of @xmath5a .
it is proposed that the na ion ordering found in na@xmath0coo@xmath3 is identical to the ideal na@xmath4coo@xmath2 of large @xmath5a@xmath6@xmath5a@xmath63c superlattice but with additional na vacancies which are bound to the oxygen defects at room temperature .
we conclude that oxygen vacancies play a key role in stabilizing the superlattice structure and must be accounted for in its modeling . |
lattice regularization of chiral gauge theories has remained a long standing problem of nonperturbative investigation of quantum field theory .
lack of chiral gauge invariance in @xmath3 proposals is responsible for the longitudinal gauge degrees of freedom ( _ dof _ ) coupling to fermionic _
dof _ and eventually spoiling the chiral nature of the theory . the obvious remedy to control the longitudinal gauge _
dof _ is to gauge fix with a target theory in mind @xcite .
the formal problem is that for compact gauge - fixing a brst - invariant partition function as well as ( unnormalized ) expectation values of brst invariant operators vanish as a consequence of lattice gribov copies @xcite .
shamir and golterman @xcite have proposed to keep the gauge - fixing part of the action brst non - invariant and tune counterterms to recover brst in the continuum . in their formalism , the continuum limit is to be taken from within the broken ferromagnetic ( fm ) phase approaching another broken phase which is called ferromagnetic directional ( fmd ) phase , with the mass of the gauge field vanishing at the fm - fmd transition .
this was tried out in a u(1 ) smit - swift model and so far the results show that in the pure gauge sector , qed is recovered in the continuum limit @xcite and in the _ reduced _ model limit free chiral fermions in the appropriate chiral representation are obtained @xcite .
when one gauge transforms a gauge non - invariant theory , one picks up the longitudinal gauge degrees of freedom ( radially frozen scalars ) explicitly in the action .
the reduced model is then obtained by making the lattice gauge field unity for all links , _
i.e. _ , by switching off the transverse gauge coupling .
the reduced model is obviously a yukawa model .
the job of gauge fixing in the @xmath3 when translated into the reduced model is to find a continuous phase transition where these unwanted scalars are decoupled leaving only free fermions in the appropriate chiral representation .
we want to apply the gauge - fixing proposal to other previous proposals of a @xmath3 which supposedly failed due to lack of gauge invariance . for this purpose
we have chosen the waveguide formulation of the domain wall fermion where , without gauge fixing , mirror chiral modes appeared at the waveguide boundary in addition to the chiral modes at the domain wall or anti - domain wall to spoil the chiral nature of the theory @xcite .
our investigation of the reduced model in this case reveals that both at yukawa coupling @xmath1 @xcite and @xmath2 @xcite the scalars are fully decoupled , there are no mirror modes and the spectrum is that of free domain wall fermions .
for kaplan s free domain wall fermions @xcite on a @xmath4-dimensional lattice of size @xmath5 where @xmath6 is the 5th dimension , with periodic boundary conditions in the 5th or @xmath7-direction and the domain wall mass @xmath8 taken as @xmath9 the model possesses a lefthanded ( lh ) chiral mode bound to the domain wall at @xmath10 and a righthanded ( rh ) chiral mode bound to the anti - domain wall at @xmath11 . for @xmath12 ,
these modes have exponentially small overlap .
a 4-dimensional gauge field which is same for all @xmath7-slices can be coupled to fermions only for a restricted number of @xmath7-slices around the anti - domain wall @xcite with a view to coupling only to the rh mode at the anti - domain wall .
the gauge field is thus confined within a _ waveguide _ , @xmath13 with @xmath14 , @xmath15 . for convenience , the boundaries at ( @xmath16 ) and ( @xmath17 )
are denoted waveguide boundary @xmath18 and @xmath19 respectively .
the symmetries of the model remain exactly the same as in @xcite .
obviously , the hopping terms from @xmath20 to @xmath21 and that from @xmath22 to @xmath23 would break the local gauge invariance of the action .
this is taken care of by gauge transforming the action and thereby picking up the pure gauge _
dof _ or a radially frozen scalar field @xmath24 ( stckelberg field ) at the @xmath25 boundary , leading to the gauge - invariant action ( with @xmath26 , where @xmath27 gauge group ) : @xmath28 where we have taken the lattice constant @xmath29 and have suppressed all other indices than @xmath7 .
the projector @xmath30 is @xmath31 and @xmath32 is the yukawa coupling at the @xmath25 boundaries .
the @xmath33 and @xmath34 are the gauge covariant dirac operator and the wilson term ( with wilson @xmath35 ) respectively .
@xmath36 and @xmath37 are the free versions of @xmath33 and @xmath34 respectively .
the gauge - fixed pure gauge action for u(1 ) , where the ghosts are free and decoupled , is : @xmath38 where , @xmath39 is the usual wilson plaquette action ; the gauge fixing term @xmath40 and the gauge field mass counter term @xmath41 are given by ( for a discussion of relevant counterterms see @xcite ) , @xmath42 where @xmath43 is the covariant lattice laplacian and @xmath44 with @xmath45 and @xmath46 .
@xmath40 has a unique absolute minimum at @xmath47 , validating weak coupling perturbation theory ( wcpt ) around @xmath48 or @xmath49 and in the naive continuum limit it reduces to @xmath50 . obviously , the action @xmath51 is not gauge invariant . by giving it a gauge transformation the resulting action @xmath52 is gauge - invariant with @xmath53 and @xmath54 , @xmath55 . by restricting to the trivial orbit , we arrive at the so - called * reduced model * action @xmath56 where @xmath57 is obtained quite easily from eq.([wgact ] ) and @xmath58 \label{redb}\end{aligned}\ ] ] now is a higher - derivative scalar field theory action .
@xmath59 in ( [ redb ] ) is same as in ( [ bxsq ] ) with @xmath60
at @xmath1 , we carry out a wcpt in the coupling @xmath61 for the fermion propagators to 1-loop . in order to develop perturbation theory , in reduced model
, we expand , @xmath62 where @xmath63 , leading to free propagator for the compact scalar @xmath64 @xcite , @xmath65 where , @xmath66 . free fermion propagators at @xmath1 are obtained in momentum space for 4-spacetime dimensions while staying in the coordinate space for the 5th dimension following @xcite ( results in @xcite can not be directly used because of difference in finer details of the action ) . the appropriate free action to start from is , @xmath67 \tilde{\psi_p^t}\ ] ] where , @xmath68 , @xmath69 , @xmath70 , @xmath71 , @xmath72 and @xmath73 .
the free fermion propagator can formally be written as , @xmath74^{-1 } } \nonumber \\ & & = ( -i\overline{p\!\!\!/ } + m^\dagger ) p_l g^l+ ( -i\overline{p\!\!\!/ } + m ) p_r g^r \label{fprop}\end{aligned}\ ] ] explicit solution of @xmath75 are obtained from @xmath76 g^l_{s , t } - b(s+1 ) g^l_{s+1,t } } \nonumber \\ & & - b(s ) g^l_{s-1,t } = \delta_{s , t } \label{green}\end{aligned}\ ] ] and a similar equation for @xmath77 . in ( [ green ] ) , @xmath78 .
we show only the calculations for obtaining @xmath79 and henceforth drop the superscript @xmath80 . setting @xmath81 in the region @xmath82
where @xmath83 and @xmath84 in the region @xmath85 where @xmath86 , @xmath87 the solutions of these equations are expressed as sum of homogeneous and inhomogeneous solutions : @xmath88 } { 2a_{\pm}\sinh(\alpha_{\pm}(p))\sinh(\alpha_{\pm}(p)l/2)}. \label{lmsol}\end{aligned}\ ] ] where , @xmath89 .
the third term in the above is the inhomogeneous solution . in order to get the complete solution we need to determine the unknown functions @xmath90 and @xmath91 in ( [ lmsol ] ) , which are obtained by considering boundary conditions from eqs . ( [ lmslice ] ) at @xmath92 , @xmath93 with @xmath94 at @xmath95 . using the boundary conditions ( [ dwbc ] ) we arrive at an equation of the form , @xmath96 where
, @xmath97 is a 4-component vector , @xmath98 is another 4-component vector and @xmath99 is a @xmath100 matrix .
the solution to the equations ( [ mateq ] ) is very complicated in general , particularly for finite @xmath6 , however @xmath101 can be obtained by solving the above equations numerically for different @xmath102 values .
this way we can easily construct the free chiral propagators at any given @xmath7-slice , including @xmath103 .
the solutions for @xmath104 and the resulting propagators can be obtained in a similar way .
another issue of interest at the free fermion level is the spread of the wavefunctions of the two chiral zero mode solutions along the discrete @xmath7-direction and their possible overlap .
a finite overlap would mean an induced dirac mass .
the extra dimension can be interpreted as a flavor space with one lh chiral fermion , one rh chiral fermion and @xmath105 heavy fermions on each sector of @xmath106 and @xmath107 .
we consider flavor diagonalization of @xmath108 ( @xmath109 is not hermitian ) : @xmath110 ^ 2 \left ( \phi^{(0)}_j \right)_s \\ \left ( m_0 m^\dagger_0 \right)_{st } \left ( \phi^{(0)}_j \right)_t & = & \left [ \lambda_j^{(0 ) } \right]^2 \left ( \phi^{(0)}_j \right)_s , \label{eveq}\end{aligned}\ ] ] where the index @xmath111 for the eigenvalues and the eigenvectors is a flavor index . we have carried out explicit solutions for the chiral modes at the domain and anti - domain walls and for the heavy modes @xcite
. we do not present these results explicitly here except to point out that the overlap of the opposite chiral modes at the domain and anti - domain walls is exponentially damped anywhere in the @xmath7 direction .
next we calculate the fermion propagators to 1-loop .
half - circle _ diagram which is diagonal in flavor space contributes to @xmath112 and @xmath113 propagator self - energies and a flavor off - diagonal _ tadpole _ diagram produces the self - energy for the @xmath114 and @xmath115 parts .
however , the self - energies are nonzero only at the waveguide boundaries @xmath18 and @xmath19 .
besides , there is also a contribution to the fermion self - energy for the @xmath114 and @xmath115 parts coming from a flavor off - diagonal half - circle diagram ( we shall call it a _ global - loop _ diagram ) where the scalar field goes around the flavor space connecting fermions at the waveguide boundaries @xmath18 and @xmath19 . by expanding @xmath62 and retaining up to @xmath116 in the interaction term we find the vertices necessary to calculate the fermion self - energy to 1-loop .
( 290,85)(-10,65 ) ( 75,85)(125,85 ) ( 125,85)(175,85 ) ( 175,85)(225,85 ) ( 150,85)(35,180,0)3 ( 80,92 ) [ ] ( 92,76)[]@xmath117 , @xmath21 ( 125,92 ) [ ] ( 175,92 ) [ ] ( 150,76)[]@xmath118 , @xmath20 ( 220,92 ) [ ] ( 208,76)[]@xmath117 , @xmath21 ( 150,110)[]@xmath119 the @xmath112 propagator on the @xmath120-th slice at the waveguide boundary @xmath18 receives a nonzero self - energy contribution from the half - circle diagram , @xmath121_{s_0,s_0}\end{aligned}\ ] ] @xmath122 @xmath123_{s_0,s_0}{\cal g}(k ) \ ; \delta_{s , s_0 + 1}\delta_{t , s_0 + 1 } \label{fselfd}\ ] ] where @xmath124 and the expression in the square bracket in ( [ fselfl ] ) is the free @xmath113 propagator @xmath125_{s_0,s_0}$ ] on the @xmath20-slice .
eq.([fselfl ] ) assumes infinite 4 space - time volume while in eq.([fselfd ] ) a finite space - time volume @xmath126 is considered .
to avoid the infra - red problem in the scalar propagator , we use anti - periodic boundary condition in one of the space - time directions in evaluating ( [ fselfd ] ) . in a similar way , 1-loop corrected @xmath113 or @xmath112 propagators are obtained at all the @xmath7-slices of the waveguide boundaries @xmath18 and @xmath19 , _ i.e. _ , at the slices @xmath20 , @xmath21 , @xmath22 and @xmath23 .
( 300,85)(0,65 ) ( 90,85)(150,85 ) ( 150,85)(210,85 ) ( 150,105)(20,180,0)3 ( 150,105)(20,180,360)3 ( 95,92 ) [ ] ( 112,76)[]@xmath117 , @xmath21 ( 205,92 ) [ ] ( 198,76)[]@xmath117 , @xmath20 ( 150,115)[]@xmath119 for the @xmath114 propagator connecting @xmath20 and @xmath21 at the waveguide boundary @xmath18 , the self - energy contribution from the tadpole diagram is given by , @xmath127 where @xmath128 is the tadpole loop integral .
similarly the self - energy contribution to the @xmath114 propagator at the waveguide boundary @xmath19 connecting @xmath22 and @xmath23 comes from a tadpole diagram and is given by , @xmath129 the global - loop diagram originates from the fact that the @xmath130 field that couples the fermions at the waveguide boundary @xmath18 is the same @xmath130 field coupling the fermions at the waveguide boundary @xmath19 .
( 300,85)(0,65 ) ( 62,85)(125,85 ) ( 125,85)(195,85 ) ( 195,85)(258,85 ) ( 160,85)(47,180,0)3 ( 67,92 ) [ ] ( 82,76)[]@xmath117 , @xmath21 ( 125,92 ) [ ] ( 125,76)[]@xmath20 ( 195,92 ) [ ] ( 156,76)[]@xmath118 ( 193,76)[]@xmath23 ( 253,92 ) [ ] ( 247,76)[]@xmath117 , @xmath22 ( 160,121)[]@xmath119 self - energy contribution from the global - loop diagram for the @xmath114 propagator is : @xmath131_{s_1 + 1,s_0}\end{aligned}\ ] ] @xmath132 @xmath133 where @xmath124 and @xmath134 is the loop integral in eq.([global ] ) .
the mass parameter @xmath135 gets modified to @xmath136 at 1-loop as : @xmath137 \nonumber \\ & + & \left[-(\sigma^{ii}_{lr}(0))_{st}\right ] + \left[-(\sigma^{gl}_{lr}(0))_{st}\right ] \nonumber \\ & \equiv & ( m_0)_{st}p_l + b^2 ( { \bf \sigma}_{lr})_{st } p_l\,.\end{aligned}\ ] ] @xmath138 identically .
@xmath139 gets modified accordingly .
diagonalization of the fermion mass matrix at 1-loop @xcite shows that , i ) the zero modes are perturbatively stable , and ii ) the overlap of the opposite chiral modes at the domain and the anti - domain walls are still exponentially damped .
this clearly rules out the necessity of a fermion mass counter term .
in the quenched approximation , we have first numerically confirmed the phase diagram in @xcite of the reduced model in ( @xmath140 ) plane .
the phase diagram shown schematically in fig.1 has the interesting feature that for large enough @xmath141 , there is a continuous phase transition between the broken phases fm and fmd .
fmd phase is characterized by loss of rotational invariance and the continuum limit is to be taken from the fm side of the transition . in the full theory with gauge fields ,
the gauge symmetry reappears at this transition and the gauge boson mass vanishes , but the longitudinal gauge _ dof _ remain decoupled . in fig.1 pm
is the symmetric phase and am is the broken anti - ferromagnetic phase .
the numerical details involved in reconstruction of the phase diagram and the fermionic measurements that follow will be available in @xcite . for calculating the fermion propagators , as in @xcite
we have chosen the point @xmath142 , @xmath143 ( gray blob in fig.1 ) .
although this point is far away from @xmath49 , around which we did our perturbation theory in the previous section , the important issue here is to choose a point near the fm - fmd transition and away from the fm - pm transition .
the results below show that for the fermion propagators there is excellent agreement between numerical results obtained at @xmath142 , @xmath143 and perturbation theory .
numerically on @xmath144 and @xmath145 lattices with @xmath146 and @xmath147 we look for chiral modes at the domain wall ( @xmath10 ) , the anti - domain wall ( @xmath148 ) , and at the waveguide boundaries ( @xmath149 and @xmath150 ) .
error bars in all the figures are smaller than the symbols .
( @xmath146 ; a.p.b.c . in @xmath151 , @xmath152).,width=321,height=396 ]
figs.2 and 3 show the @xmath113 propagator @xmath153 and the @xmath112 propagator @xmath154 at the domain and anti - domain wall as a function of a component of momentum @xmath155 for both @xmath156 ( physical mode ) and @xmath157 ( first doubler mode ) at @xmath1 . from the figures ,
it is clear that the doubler does not exist , only the physical @xmath113 ( @xmath112 ) propagator seems to have a pole at @xmath158 at the anti - domain ( domain ) wall . in all the figures , ns , pt and
ff respectively indicate data from numerical simulation , from perturbation theory and from free fermion propagator by direct inversion of the free fermion matrix . for figs.2 and 3 ,
pt also mean zeroth order perturbation theory , _
i.e. _ , numerical solution of propagator following eq.([mateq ] ) .
we have pt results also for @xmath145 lattice but have chosen not to show them because they fall right on top of the numerical data . instead pt results shown for @xmath159 lattice for which the @xmath155 points are distinct .
the dotted line in all figures refer to the propagator from pt using a @xmath160 lattice . _
the curves stay the same irrespective of methods or lattice size_. based on the above , we can conclude that there are _ only free rh fermions _ at the anti - domain wall , and at the domain wall there are _ only free lh fermions_. ( @xmath146 ; a.p.b.c . in @xmath151 , @xmath152).,width=321,height=396 ]
figs.4 and 5 show no evidence of a chiral mode at the waveguide boundaries @xmath161 and 6 and excellent agreement with 1-loop perturbation theory . here
too doublers do not exist .
actually the agreement with the ff method ( direct inversion of the free domain wall fermion matrix on a given finite lattice ) is also excellent , because the 1-loop corrections are almost insignificant . for clarity , in figs.4 and 5
we have not shown the @xmath112 propagator on @xmath161 and the @xmath113 propagator at @xmath162 , but conclusions are the same .
-propagator at waveguide boundary @xmath161 ( @xmath146 , a.p.b.c . in @xmath151 , @xmath152).,width=321,height=396 ]
-propagator at waveguide boundary @xmath162 ( @xmath146 , a.p.b.c . in @xmath151 , @xmath152).,width=321,height=396 ]
similar investigation at the other waveguide boundary @xmath163 also does not show any chiral modes .
previous investigations of the domain wall waveguide model without gauge fixing @xcite have shown that the waveguide boundaries are the most likely places to have the unwanted mirror modes .
this is why we have mostly concentrated in showing that there are no mirror chiral modes at these boundaries , although we have looked for chiral modes everywhere along the flavor dimension .
in fact , we do not see any evidence of a chiral mode anywhere other than at the domain wall and the anti - domain wall .
at @xmath164 , the domain wall and the anti - domain wall are detached from each other .
fermion current considerations and numerical simulations clearly show in this case that mirror chiral modes form at the waveguide boundaries . on the other hand , at @xmath1 (
as presented above ) and @xmath165 @xcite , the mirror chiral modes are certainly absent at the waveguide boundaries .
actually in this case , the only chiral modes are at the domain wall and at the anti - domain wall and the spectrum is that of a free domain wall fermion . the interesting question is what happens at @xmath32 smaller than unity , especially at positive values near zero . to investigate this question
, we looked for chiral modes at @xmath166 at places other than the domain wall and the anti - domain walls , especially at the waveguide boundaries . on a @xmath167 lattice ,
the chiral propagators at the waveguide boundaries showed an increasing trend as the fourth component of momentum @xmath155 ( with @xmath168 ) was decreased to the minimum value possible for this lattice size , something that could signify a pole at zero momentum . to resolve this we took lattice sizes which were bigger in the 4-direction to accommodate lower @xmath155-values .
we found that for each @xmath169 there is a big enough lattice size for which the chiral propagators ultimately start showing a descending trend .
obviously the smaller the yukawa coupling , the bigger the @xmath151 extension was required .
moreover , all the chiral propagators at these small yukawa couplings matched exactly with the corresponding free case .
our results at the waveguide boundaries @xmath170 are summarized in the figs.6 and 7 for the @xmath113 and the @xmath112 propagators respectively .
poles for these chiral propagators are clearly not seen .
using the gauge fixing approach and tuning only a finite number of counterterms ( in this case , just the @xmath171-term ) , we end up in the reduced model with free domain wall fermion theory consisting only of lh and rh chiral modes respectively at the domain and the anti - domain wall . with the u(1 )
transverse gauge _ dof _ back on the waveguide , only the rh fermions on the anti - domain wall will be gauged ( according to our construction ) .
the gauge degrees of freedom are completely decoupled .
we reached our conclusions for the case of @xmath1 by performing a perturbation theory for the fermion propagators around @xmath49 and comparing them to the numerical results at @xmath143 .
the comparison is near perfect due to the robust properties of domain wall fermions ( this does not happen nearly as nicely for the wilson fermions ) and shows that as long as @xmath141 is big enough to avoid the fm - pm transition , it is as good as infinity , _
i.e. _ , the whole fm - fmd transition line is in the same universality class as the perturbation point @xmath49 .
9 y. shamir , phys .
d57 ( 1998 ) 132 ; m.f.l .
golterman , y. shamir , phys .
b399 ( 1997 ) 148 h. neuberger , phys .
b183 ( 1987 ) 337 w. bock et .
al . , hep - lat/9911005 w. bock , m.f.l .
golterman , y. shamir , phys . rev .
lett 80 ( 1998 ) 3444 m.f.l .
golterman , k. jansen , d.n .
petcher and j.c .
vink , phys .
d49 ( 1994 ) 1606 s. basak and asit k. de , hep - lat/9911029 s. basak and asit k. de , in preparation d.b .
kaplan , phys .
b288 ( 1992 ) 342 w. bock et .
al . , phys .
d58 ( 1998 ) 34501 s. aoki , h. hirose , phys . rev .
d54 ( 1996 ) 3471 w. bock et .
al . , phys .
d58 ( 1998 ) 54506 s. basak , asit k. de , nucl .
b ( proc . suppl . )
83 - 84 ( 2000 ) 615 | we investigate a u(1 ) lattice chiral gauge theory ( @xmath0 ) with domain wall fermions and gauge fixing . in the _ reduced _ model limit ,
our perturbative and numerical investigations at yukawa coupling @xmath1 show that there are no extra mirror chiral modes .
the longitudinal gauge degrees of freedom have no effect on the free domain wall fermion spectrum consisting of opposite chiral modes at the domain wall and the anti - domain wall which have an exponentially damped overlap .
our numerical investigation at small yukawa couplings ( @xmath2 ) also leads to similar conclusions as above . |
presolar silicon carbide ( sic ) is the abundant and well studied dust species among presolar grains extracted from meteorites .
based on the isotopic compositins of c , n , si and al , presolar sic grains are classified into six categories named `` mainstream '' , a , b , x , y and z grains ( e.g. , see @xcite for a review ) . among them , the mainstream grains populating about 90 % of presolar sic grains have been considered to be originated in carbon rich asymptotic giant branch ( c rich agb ) stars from the isotopic compositions of c , n , and si @xcite and the s
process signatures in the noble gases as well as in the heavy elements @xcite .
@xcite have reported that sic grains extracted from the murchison meteorite have a log normal distribution with the peak at @xmath8 0.4 @xmath12 in diameter , and 70 % of them by number are in the grains whose diameter ranges from 0.3 to 0.7 @xmath12 @xcite . although it was reported that sic grains are more fine grained in other meteorites than in murchison ( see @xcite ) , a significant fraction of presolar sic grains are larger than 0.3 @xmath12 in diameter ( see @xcite ) .
thus , submicron
sized sic grains are expected to form around c rich agb stars .
the presence of sic grains in c rich agb stars was confirmed from the emission feature around 11.3 @xmath12 @xcite , prior to the discovery of presolar sic grains @xcite .
also , the absorption feature attributed to sic grains was observed in several extreme carbon stars @xcite .
the radiative transfer models fitting the observed spectral energy distributions ( seds ) including the spectral feature attributed to sic grains have estimated the mass ratio of sic to carbon grains to be in the range of 0.010.3 @xcite .
the variation of the mass ratio of sic to carbon derived from the radiative transfer models arises from the grain models used , and the value could be also influenced by the optical constants of sic used in the radiative transfer models ( see the discussion in section 6 ) .
although the isotopic signatures of presolar sic grains and the observed spectral feature around 11.3 @xmath12 support the formation of sic grains in circumstellar envelopes of c rich agb stars , the formation process as well as the formation conditions remain unknown and debatable : thermodynamic equilibrium calculations have been applied to investigate the formation condition of dust grains in c rich agb stars ( e.g. @xcite ) , and the calculations have claimed that high gas pressure ( 0.3300 dyne @xmath13 ) and low c / o ratio ( 1.0 @xmath14 c / o @xmath15 1.1 ) is necessary for reproducing the condensation sequence of presolar tic , graphite , and sic grains in this order inferred from the presence of graphitic spherules containing a tic core @xcite . on the other hand , @xcite suggested the formation of sic grains earlier than carbon grains , considering the inverse greenhouse effect that the temperature of sic grains which is transparent in the optical to near infrared ( nir ) region @xcite is lower than the temperature of carbon grains in circumstellar envelopes of c rich agb stars .
in addition , the interferometric observation of irc+10216 at 11 @xmath12 has suggested the presence of dust species more transparent than carbon grains in the inner circumstellar envelope close to the photosphere @xcite .
although thermodynamic equilibrium calculations can tell us the physical conditions in which a condensate exists stably , the calculations as well as the astronomical observations provide directly no information on the number and size of grains that are essential to clarify the physical conditions prevailing in the formation sites of grains and the physical and chemical processing suffering in the interstellar space . also , it should be pointed out here that the c / o ratio in c rich agb stars is not restricted to be less than 1.1 , and tends to increase with decreasing the effective temperature @xmath6 ; the low c / o ratio observed in stars with @xmath6 @xmath15 2500 k is considered to be caused by the neglect of formation of dust in the model atmosphere @xcite . so far dust formation around c rich agb stars has been investigated in the framework of pulsation enhanced dust driven wind models since the study by @xcite , and it has been believed that the hydrodynamical models including only the formation of carbon grains well reproduce the observed dynamical behavior of c rich agb stars whose mass loss rates exceed 10@xmath16 @xmath2 yr@xmath17 ( e.g. , @xcite ) .
however , the formation of sic grains has not yet pursued in the hydrodynamical models .
on the other hand , formation of a variety of dust species including sic grains has been studied in the steady state dust driven wind models around agb stars in order to investigate the evolution of dust in the interstellar space ( see @xcite ) . however , in the models , the number of seed nuclei is assumed in the formation process .
although the models may predict the mass of dust species supplied from agb stars , no information is available for the size distribution . taking into account the nucleation process and including the
so called inverse greenhouse effect , @xcite have investigated the formation of sic grains in an outflowing gas with constant velocity around c rich agb stars and shown that sic grains condense prior to carbon grains for the mass loss rate smaller than @xmath8 1.5 @xmath18 10@xmath19 @xmath2 yr@xmath17 ; otherwise composite grains consisting of carbon and sic form
. the calculated radius of @xmath8 0.01 @xmath12 is too small to reproduce the size of presolar sic grains and the proposed grains consisting of a sic core and a carbon mantle have not been discovered except for one in presolar grains @xcite .
thus the investigation of formation of sic grains in a more realistic hydrodynamical model remains an important subject to be explored which will reveal the formation process and condition around c rich agb stars .
the knowledge of not only the amount but also the size distribution of dust grains is crucial for investigating the role of dust in the universe as shown by @xcite . in particular , the size distribution of sic grains formed around c rich agb stars is vital to get insight into the evolution and processing during the journey from the formation sites to the incorporation into meteorites from the analysis of presolar sic grains . in this paper , we aim at exploring the formation process of sic grains and investigating the amount as well as the size distribution in the framework of pulsation enhanced dust driven wind from c rich agb stars .
we develop the hydrodynamical model including formation of carbon and sic grains as well as a non grey radiative transfer .
the formulation by @xcite is applied for the nucleation and growth processes of carbon grains .
we assert that sic grains nucleate and grow homogeneously starting from sic molecule because all presolar sic grains whose internal structure has thus far been analyzed do not contain any seed nuclei at their centers @xcite .
furthermore there is no evidence in presolar graphite grains for a sic seed nuclei except for one grain @xcite .
the two cases are considered for the formation process of sic grains ; one is the lte case where the vibration temperature @xmath0 of sic cluster is the same as the gas temperature as usual .
another is the non lte case taking into account the inverse greenhouse effect on the formation , in which @xmath0 is assumed to be the same as the temperature of small sic grains .
we will show that the mass ratio of sic to carbon grains and the radius of sic grains are at most 10@xmath16 and 0.1 @xmath12 , respectively , in the lte case , while the mass ratio ranges from 0.098 to 0.23 and the typical radius ranges from 0.2 to 0.6 @xmath12 for the sticking probability @xmath10 of 0.11.0 in the non lte case .
this paper is organized as follows ; the chemical reactions and gas species relevant to the formation of sic grains are examined and the formation process of sic grains is formulated in section 2 , and the numerical scheme for calculation of grain formation is presented in section 3 , including the method for tracing the size distribution . section 4 describes the hydrodynamical model and the modeling procedure for pulsation enhanced dust
driven wind model including dust formation and radiative transfer .
the results of calculations are presented in section 5 and are discussed in section 6 .
summary is presented in section 7 .
in the hydrodynamical models of pulsation enhanced dust driven wind from c rich agb stars so far developed since the work of @xcite , the formation of carbon grains has been processed according to the recipe developed by @xcite ; the abundances of gas species responsible for the nucleation and growth of carbon grains are assumed to be in chemical equilibrium in the gas whose elemental abundance is solar except for carbon . in the recipe based on @xcite , the formation of clusters with number of constituent molecules ( size ) up to @xmath20=1000 is evaluated from the steady state nucleation rate , and the growth and evaporation of grains with size @xmath21 is treated as macroscopic grains .
the formation of carbon grains are considered to proceed through the two body reactions of c@xmath22h@xmath22 and c@xmath22h molecules with carbon clusters , starting from atomic carbon or diatomic carbon molecule . in this section
, we shall explore the formation process of sic grains around c rich agb stars , following the formulation for the formation of carbon grains @xcite .
first we examine the gas species responsible for the nucleation and the growth processes of sic grains from chemical equilibrium calculations . in the same way as the formation of carbon grains
, we consider that the nucleation process proceeds through the two body reactions of cluster with gas species , starting from the molecules whose stoichiometric composition is the same as sic grains . on the other hand , we allow three or more body reactions on the surface for the growth process . then , the basic equations describing the nucleation and growth processes of sic grains are presented .
figure 1 shows the temperature dependence of relative abundances of si bearing molecules ( left ) and c bearing molecules ( right ) in chemical equilibrium in the gas with the density @xmath23 10@xmath24 g @xmath25 and c / o = 1.4 , which is calculated by the method of the minimization of the gibbs free energies .
the elemental abundances except for carbon are taken from @xcite , and the thermodynamic data are taken from @xcite except for c@xmath22h@xmath22 @xcite , sic@xmath26 , sic@xmath27 , sic@xmath28 , si@xmath29c@xmath29 , si@xmath29c@xmath26 , si@xmath29c@xmath27 , si@xmath26c , si@xmath26c@xmath29 , si@xmath26c@xmath26 , si@xmath27c , si@xmath27c@xmath29 , and si@xmath28c @xcite .
we can see that , as the starting molecules for the formation of sic grains , sic is most abundant in the region whose gas temperature is higher than 1350 k , while si@xmath29c@xmath29 molecule is abundant in the lower temperature region . among si bearing molecules that are considered to be candidate reactants in the growth process of sic grains , si , sic@xmath29 , and sis molecules are abundant in the overall range of gas temperature presented in figure 1 , and si@xmath29c is also abundant in the low temperature region of @xmath30 1100 k. as shown in the right panel , hydrocarbon molecules such as c@xmath29h@xmath29 or c@xmath29h
are expected to be reactants participating in the growth process on the surface of sic grains unless these molecules are heavily depleted by the formation of carbon grains .
including the starting molecules sic and si@xmath29c@xmath29 for the nucleation process , the candidate reactants for the two body reaction in the nucleation process are confined to twelve si bearing molecules : sic , sic@xmath29 , sic@xmath26 , sic@xmath27 , si@xmath29c , si@xmath29c@xmath29 , si@xmath29c@xmath26 , si@xmath26c , si@xmath26c@xmath29 , si@xmath27c , si@xmath27c@xmath29 , and si@xmath28c .
we shall exclude the gas species with which the reaction for formation of sic clusters is endothermic .
table 1 presents the reaction enthalpies @xmath31 of the twelve candidates for the formation of si@xmath29c@xmath29 and si@xmath26c@xmath26 at 1000 and 1500 k. the molecules sic@xmath29 ,
sic@xmath26 , sic@xmath27 , si@xmath29c , si@xmath29c@xmath26 , si@xmath26c , si@xmath27c , and si@xmath28c are excluded from the candidates since the reactions are endothermic .
the reaction enthalpies for the formation of sic clusters with @xmath32 for which no thermodynamic data is available are replaced with those for the formation of solid sic , and the values at 1000 and 1500 k are listed in table 2 where the thermodynamic data of @xmath33sic sic ( 79.4 % by number ) , @xmath34sic ( 2.7 % ) , and the intergrowths of these two forms ( 17.1 % ) . note that the difference of thermodynamic data between @xmath34sic and @xmath33sic does not affect the results of calculations presented in this paper . ]
is used in the calculations .
the molecules sic@xmath29 and si@xmath28c are excluded because of the endothermic reactions . among the remaining candidates , as the reactants in the nucleation process of sic grains , we adopt sic and si@xmath29c@xmath29 since these molecules are much more abundant than the others in the temperature region of interest to us as shown in figure 1 ( left ) .
we discard the nucleation process starting from the two body reaction between si@xmath29c@xmath29 molecules since si@xmath29c@xmath29 is relatively stable and without the thermodynamic data of si@xmath27c@xmath27 it is difficult to judge whether the reaction for the formation of si@xmath27c@xmath27 proceeds or not .
thus we consider the following two reactions in the nucleation process of sic grains : @xmath35 where @xmath36 , and note that we take sic as the starting molecule for the nucleation process of sic grains .
we shall discuss the effect of the nucleation process starting from the two body reaction of si@xmath29c@xmath29 molecules in section 6 . in the nucleation process , clusters grow through the attachment of reactant and the growth rate depends on the number density and kinetic temperature of the reactants . on the other hand , clusters decay through the detachment of reactant molecules , and the decay rate depends on the vibration states of the cluster @xcite . as pointed out by @xcite , in the dust formation region around c rich agb stars , the vibration temperature of sic cluster @xmath0 is not always the same as the kinetic temperature of gas @xmath37 due to the low gas density and strong radiation field .
the principle of detailed balance being applied , the decay rate can be related with the growth rate by considering a hypothetical situation in which the cluster is in a local thermal equilibrium with the ambient gas whose temperature is @xmath0 .
then , under the assumption that @xmath0 is independent of the size of cluster , the steady state nucleation rate @xmath38 for the reaction with ( sic)@xmath39 is given by @xmath40^{-1}\end{aligned}\ ] ] where @xmath41=1 ( 2 ) denotes the reaction ( 1 ) ( the reaction ( 2 ) ) and the quantities related with the reactant sic ( si@xmath29c@xmath29 ) molecule , @xmath42 is the hypothetical radius of a monomer in the bulk phase , @xmath43 the number density of starting molecule sic .
the number density , sticking probability , mass , and partial pressure of the @xmath41th reactant is represented by @xmath44 , @xmath45 , @xmath46 , and @xmath47 , respectively .
the boltzmann constant is @xmath48 , @xmath49 with the floor function @xmath50 , and the surface tension of the cluster of size @xmath51 @xmath52 .
the supersaturation ratio @xmath53 is given by @xmath54 where @xmath55 is the standard pressure . in the above equation
, @xmath56 is the gibbs free energy for the formation of bulk sic from the @xmath57th reactant in the standard pressure , and the temperature dependence of @xmath58 is evaluated from the least squares fitting given by @xmath59 whose form is based on a polynomial expression of heat capacity at the constant pressure @xcite .
the coefficients @xmath60 , @xmath61 , @xmath62 , @xmath63 , @xmath64 and @xmath65 are given in table 3 .
in the calculations , we adopt @xmath42=1.71@xmath1810@xmath9 cm which is evaluated from the bulk density of solid sic ( @xmath66=3.16 g @xmath25 ) taken from the crc hand book @xcite .
the values of the sticking probabilities of sic and si@xmath29c@xmath29 molecules on the sic cluster are unknown and are assumed to be unity .
the surface tension @xmath52 of sic cluster with size @xmath67=2 or 3 is directly evaluated from the following definition of surface tension of @xmath51mer @xmath68 where @xmath69 and @xmath70 are the gibbs free energies of @xmath51mer in gas phase and the monomer in bulk phase in the standard state , respectively . on the other hand
, we adopt the bulk value of @xmath71=840 erg @xmath13 measured by @xcite as the surface tension of sic clusters with size @xmath72 .
we allow three or more body reactions for the growth of sic grains in addition to the two body reactions for the nucleation .
as shown in figure 1 , the abundant si and c bearing molecules expected to be the reactants for the grain growth are si , sic@xmath29 , si@xmath29c , c , c@xmath29h@xmath29 , and c@xmath29h , except for co and sis which are stable against the chemical reactions .
then we select the following exothermic reactions ; @xmath73 where @xmath74=1000 ; the reaction enthalpies at 1000 k are -1265 , -1739 , -1572 , -1206 , and -1240 kj mol@xmath17 for the chemical reactions from ( 9 ) to ( 13 ) in this order . except for the two body reactions
, it is assumed that each of the chemical reactions proceeds through the attachment of a key species which is defined as the species of the least collision frequencies among the reactants and is considered to control the kinetics of growth process @xcite .
then , the time evolution of the size of dust grains with radius @xmath75 is given by @xmath76 where the size independent net time scale of grain growth @xmath77 @xcite is the sum of the net time scales for the reactions from ( 7 ) to ( 13 ) and is written as follows : @xmath78 in the above equation , @xmath79 ( @xmath57=1,2 ) is the net time scale of grain growth by the two body reaction with ( sic)@xmath39 molecule and given by @xmath80 and @xmath81 ( n=9 to 13 ) is the net time scale of grain growth for each of the chemical reactions from ( 9 ) to ( 13 ) and is given in the same form of equation ( 16 ) by replacing @xmath82 , @xmath83 and @xmath84 with those for the key species . in equation
( 16 ) , @xmath85 represents the vapor pressure for the key species in the reaction ; for example , in the reactions ( 9 ) and ( 10 ) , the vapor pressure of si atom is assumed to be the same as that for the reaction : @xmath86 . in the reaction ( 11 ) , the vapor pressure of si@xmath29c molecule is assumed to be the same as that for the reaction : @xmath87 . in the calculations
, the sticking probability @xmath88 is assumed to be unity for all reactions .
the number densities of the gas species responsible for the formation process of carbon and sic grains in the pulsation enhanced dust driven wind from c
rich agb stars are evaluated from the chemical equilibrium calculations for given gas density and temperature coupled with the hydrodynamical calculation .
we include 4 atoms ( h , c , si , s ) and 14 molecules ( h@xmath29 , c@xmath29 , c@xmath26 , c@xmath29h , c@xmath29h@xmath29 , sih , sh , h@xmath29s , sic , sic@xmath29 , si@xmath29c , si@xmath29c@xmath29 , cs , sis ) in the chemical equilibrium calculations , and the abundances of elements except for carbon are taken from the table of @xcite .
we employ the formulation presented in section 2 for the formation process of sic grains , while the formulation by @xcite is applied for the formation of carbon grains .
as mentioned before , the sticking probability @xmath10 for all reactions at nucleation and growth of sic grains is assumed to be unity in the calculations , whose effect on the results of calculations is discussed in section 6 .
the number density and size of grains as well as the fraction of condensible molecules locked into grains are evaluated by taking the @xmath89th moment of the size distribution of grains per h
element @xmath90 defined as @xmath91 where @xmath92 is the number density of grains with the size @xmath93 at time @xmath94 and @xmath95=@xmath96 + 2@xmath97 .
normally , the moments are calculated by reducing to the simultaneous differential equations @xcite .
the time variation of gas ( grain ) temperature is not always monotonic and rather complicated in the pulsation
enhanced dust driven winds . in the case that the formed grains are being lost by the destruction due to evaporation at time @xmath94 , the extinction term is included in the differential equations by using the number density of grains with the minimum size @xmath98 which has to be derived by tracing the size distribution .
therefore , solving the differential equations must be compensated with the evaluation of the size distribution of grains with a limited number of size bins in the hydrodynamical simulations @xcite . thus , in this paper , we calculate the moments by employing the finite and adaptive size grids rather than solving a set of the simultaneous differential equations .
each of the size bins is comoved in size space @xcite .
the @xmath99th size bin is characterized by the number of grains per h
element @xmath100 in the size interval with lower limit @xmath101 and upper limit @xmath102 .
the lower and upper limits are calculated by integrating equation ( 14 ) . when the net time scale @xmath77 is positive ,
the number of grains per h element in the lowest size bin @xmath103 where @xmath104 is the total number of size bins is given by integrating the equation @xmath105 when the net time scale @xmath77 is negative , with the destruction rate of grains @xmath106 estimated by @xmath107 where @xmath108 , @xmath103 is derived by integrating the equation @xmath109 according to @xcite , the discrete moment @xmath110 is evaluated by @xmath111 for @xmath112=0 to 3 .
the 0th moment @xmath113 and the 3rd moment @xmath114 give the number of grains per h element and the number of condensible molecules incorporated into grains per h element , respectively .
the volume equivalent radius of grains @xmath115 at a time is defined by @xmath116 .
the basic equations for the time dependent spherically symmetric hydrodynamical model are written in standard finite difference form and solved by explicit integration along the lines of difference scheme given by @xcite .
this modeling method has been adopted by berlin group ( e.g. , @xcite ) . actually , this method is time consuming since the time steps are restricted by the courant friedrichs levy ( cfl ) stability condition ( see @xcite ) .
however , since the advection term drops out from the equation of motion in lagrangian coordinate , we can easily trace the size distribution of grains which is one of main subjects of this study .
the continuity equation representing the conservation of mass is replaced with the definition of gas velocity @xmath117 given by @xmath118 where @xmath119 is the radial position of the lagrangian mass element at a time @xmath94 . under the assumption of position coupling between gas and grains , the equation of motion
is described as @xmath120 where @xmath121 is the thermal gas pressure , @xmath66 is the gas density , @xmath122 is the gravitational constant , and @xmath123 is the mass inside a radius @xmath119 . in the above equation , @xmath124 is the ratio of radiation pressure force to gravity and expressed as @xmath125 where @xmath126 is the sum of the flux weighted mean of mass radiation pressure coefficient for gas @xmath127 and that for dust @xmath128 , @xmath3 the luminosity , and @xmath129 the speed of light .
the mass inside a radius @xmath119 in the above equations is replaced with the stellar mass @xmath1 because the mass contained in the circumstellar envelope is small compared to the stellar mass during the calculations presented in this paper . in the calculations ,
we adopt the constant gas opacity of @xmath130=2@xmath131 @xmath132 g@xmath17 @xcite . following the previous works for hydrodynamic models of dust driven
wind solved with frequency dependent radiative transfer calculation @xcite , we evaluate the dust opacity @xmath133 at a wavelength @xmath134 in the rayleigh approximation .
the optical constants of carbon and sic grains used in the calculation are taken from @xcite and @xcite , respectively .
the radiation field in the circumstellar envelope is derived from the frequency dependent radiative transfer calculation using the variable eddington factor method under the assumption of radiative equilibrium .
then , the gas temperature @xmath135 is given by @xmath136 where the subscript `` abs '' refers to absorption , @xmath137 is the mean intensity and @xmath138 the planck function .
the temperature of @xmath99th grain species @xmath139 is given by @xmath140 according to the hydrodynamical models developed so far for dust driven wind from c rich agb stars @xcite , the vibration temperature of carbon cluster is set to be equal to the gas temperature as usual . on the other hand , in order to investigate the influence of the inverse greenhouse effect suggested by @xcite for the formation of sic grains , we consider the two cases for the vibration temperature of sic cluster : one is the case that the vibration temperature of sic cluster is the same as the gas temperature and that is referred to as the lte case hereafter .
another is the non lte case where the vibration temperature is assumed to be the same as the temperature of small sic grains @xcite . assuming that the gas species mainly consisting of h , h@xmath29 , and he accommodate with grains completely upon colliding , and placing a small sic grain hypothetically at a position in the gas flow , we derive the vibration temperature of sic cluster @xmath0 by balancing the heating with the cooling through the interaction with radiation and gas as follows ; @xmath141d\lambda = \displaystyle\sum_{j}n_{{\rm gas},j } \left(\frac{k_{\rm b}t_{\rm v}}{2\pi m_{j}}\right)^{\frac{1}{2 } } ( c_{{\rm v},j}+\frac{1}{2}k_{\rm b } ) \left(t_{\rm v}-t_{\rm gas}\sqrt{\frac{t_{\rm gas}}{t_{\rm v}}}\right)\end{aligned}\ ] ] where @xmath142 is the absorption efficiency factor of sic grain with radius @xmath143 whose value is set to be @xmath144 @xmath12 in the calculation , and @xmath145 , @xmath146 , and @xmath147 are the number density , the mass , and the specific heat at constant volume of the @xmath89th gas species , respectively .
the models of pulsation enhanced dust driven wind are specified by the six parameters ; the stellar mass @xmath1 , the stellar luminosity @xmath3 , the effective temperature @xmath148 , the c / o ratio , the period of the pulsation @xmath7 , and the velocity amplitude of the pulsation @xmath149 . as the first step of the hydrodynamical simulation , a dust free static model atmosphere is constructed as follows ; the initial stellar radius @xmath150 is given by the stefan boltzmann law @xmath151 where @xmath71 is the stefan boltzmann constant .
given a guess value for gas density at @xmath150 , the radial profiles of density and temperature are calculated by solving the equation @xmath152 coupled with the radiative transfer calculation with sufficiently fine grids .
the procedure is iterated until the geometrically diluted optical depth at @xmath150 converges to 2/3 .
the dynamical model calculations including the formation of grains are carried out by placing the piston at @xmath153 whose position is two times pressure scale height around @xmath150 below from the photosphere of static model atmosphere and is defined as the inner boundary of dynamical model at @xmath94=0 .
then the radius @xmath154 of inner boundary is varied sinusoidally with time as @xmath155 and the velocity @xmath156 at the inner boundary is given by @xmath157 in the simulation , no mass flow is assumed through the inner boundary .
the radiative flux is assumed to be constant over time and be equal to @xmath158 at the photosphere of dynamical model @xmath159 where the diluted optical depth of gas is 2/3 , and thus the stellar luminosity @xmath160 varies with time .
the piston velocity amplitude @xmath149 is increased slowly from 1 cm s@xmath17 to the specified value to prevent the first outwardly moving subsonic wave from growing into a enormous transient shock , following @xcite .
after the onset of dust formation , the dust driven wind takes places .
as the gas expands , the spacing between radial grids in the acceleration region becomes wider and the rezoning procedure necessary for resolving the density structures of the circumstellar envelope is introduced by adopting the method proposed in fleischer et al.(1992 ) . after the outermost zone passes through the outer boundary of dynamical model which is placed at 25 @xmath150 in the simulation ,
the outermost lagrangian grid is eliminated .
then , at the outer boundary , we trace the time variations of mass loss rate , gas outflowing velocity , dust to gas mass ratio , and condensation efficiency @xmath161 ( @xmath162 ) which is defined as the fraction of carbon ( silicon ) atoms in c bearing molecules except for co ( si bearing molecules ) locked into carbon ( sic ) grains . note that , in the case that all si atoms in si bearing molecules are locked into sic grains , the maximum value of @xmath161 @xmath8 0.9 in the model with c / o=1.4 .
we adopt the following model parameters in the calculation ; @xmath6=2600 k , @xmath3=10@xmath4 @xmath5 , @xmath1=1.0 @xmath2 , c / o=1.4 , @xmath7=650 days , and @xmath163=2.0 km s@xmath17 , which is a typical set of the values in the previous studies ( e.g. , @xcite ) and is employed as the reference values in the models of c rich agb stars that reasonably reproduce the observed dynamical behaviors @xcite .
the number of radial grids used in the hydrodynamical calculations depends on the density structure and is in the range between 800 and 1200 throughout the calculations .
the wavelength region covered in the radiative transfer calculation ranges from 0.35 to 120 @xmath11 m and the number of wavelength grids is 100 .
the simulations have been performed for 180 cycles ( stellar pulsation periods ) . in this paper , we neglect the radiation pressure force acting on sic grains because of the transparency in the optical to infrared region ( up to @xmath8 9 @xmath12 ) @xcite . the formation of sic grains in the pulsation enhanced dust driven wind is closely related with the formation of carbon dust shell stemming from the rapid formation of carbon grains in the high density gas induced by the stellar pulsation and the resulting acceleration of dense region caused by the radiation pressure force acting on the carbon grains .
the carbon dust shell is characterized by the inversions of density distributions of gas and carbon grains and plays a crucial role in the time evolution of gas temperature and density through the backwarming caused by the thermal emission from carbon grains .
first , we outline the formation of carbon dust shells in this model , and then we show how sic grains form in the lte and non lte cases .
the carbon dust shell appears in the present model quasi periodically once a stellar pulsation period and is classified into two types by the peak value of @xmath161 in the shell .
here we shall overview the formation of each type of the carbon dust shells and its dynamics , referring figure 2 which displays the radial structure of the quantities related to gas dynamics and formation of carbon and sic grains in the lte case throughout the period from @xmath94=166.7 to 168.8 @xmath7 .
the first type of carbon dust shell ( hereafter referred to as the 1st type cds ) is defined as the high density region with the peak value of @xmath164 0.9 , and is located around 4 @xmath150 at @xmath94=167.9 @xmath7 ( see the bottom row in figure 2 ) .
the formation of this 1st type cds actually starts at @xmath165 166.7 @xmath7 and the behavior of formation process is almost the same as that of the 1st type cds whose formation starts at @xmath165 168.8 @xmath7 ; the nucleation of carbon grains takes place in the dust
free region ranging from 1.8 to 3.0 @xmath150 whose gas temperature decreases to 1300 k ( see the radial variation of the nucleation rate of carbon grains at @xmath94=168.8 @xmath7 and the time variation of the distribution of the number of carbon grains in this region from @xmath94=168.5 to 168.8 @xmath7 ) .
then , the backwarming by thermal emission from carbon grains subsequently formed in the outer part of this region not only raises up the gas temperature but also evaporates the carbon grains in the inner part ( see the 3rd rows at @xmath94=167.0 @xmath7 and @xmath94=167.3 @xmath7 ) . however , in the outer part , the nucleation and growth of carbon grains still proceed and the ratio of radiation pressure force to gravity @xmath34 around 2.5 @xmath150 exceeds unity at @xmath94=167.0 @xmath7 with increasing the volume equivalent radius @xmath166 and the condensation efficiency @xmath161 of carbon grains up to @xmath8 0.1 @xmath11 m and @xmath8 0.3 ( see the 3rd and bottom rows ) , as a result this part is accelerated outwards . on the other hand , the gas density in the region outer than the accelerated region is insuffcient to produce abundant enough carbon grains to accelerate this region efficiently despite the ongoing nucleation .
thus , the radial compression due to the velocity difference causes the outer part of the accelerated region to accelerate more through the increase of nucleation and growth rates of carbon grains , and results in the shell structure ; note that the shell structure is not associated with the pulsation shock but induced by the formation of carbon grains .
the shell structure becomes prominent with decreasing the gas density in the region inner than the accelerated region due to the radial expansion ( see the 2nd rows from @xmath94=167.3 to 167.6 @xmath7 ) .
the formation of 1st type cds completes at @xmath94 @xmath8 167.9 @xmath7 .
the formation of the second type of carbon dust shell ( hereafter referred to as the 2nd type cds ) defined as the high density region with the peak value of 0.3 @xmath167 0.7 onsets around 2.2 @xmath150 at @xmath94=168.2 @xmath7 , being triggered by the merging of the two consecutive shocks induced by the stellar pulsations , which can be seen from the time variations of gas velocity and @xmath34 in this region from @xmath94=167.9 to 168.2 @xmath7 ; the carbon grains in front of the precedent shock , which nucleated at @xmath94 @xmath8 166.7 @xmath7 and survived from the evaporation process associated with the evolution of 1st type cds starting at @xmath94 @xmath8 167.0 @xmath7 , grow up from @xmath166=0.05 to 0.1 @xmath12 in the region compressed further by the merging with the following shock ( see the bottom rows from @xmath94=167.6 to 168.2 @xmath7 ) .
after then , the nucleation and growth of carbon grains in the compressed outer part of this region develops the shell structure with increasing @xmath34 , while the rapid decrease of gas density caused by the outward velocity already exceeding 10 km s@xmath17 suppresses the growth rate of carbon grains and the increase of @xmath161 .
the evolution of this carbon dust shell ends up as the 2nd type cds with the peak value of @xmath161 @xmath8 0.7 at @xmath94 @xmath8 169.0 @xmath7 .
the density profile as well as the peak value of @xmath161 of 2nd type cds varies from cycle to cycle ( e.g. , see the 2nd type cds located around 4.5 @xmath150 at @xmath94=167.0 @xmath7 ) .
the alternating formation of the 1st and 2nd type cdss is sometimes disturbed , in which case the 2nd type cds fails to grow up without increasing the peak value of @xmath161 up to 0.3 .
the appearance and periodicity of carbon dust shell depend on the stellar parameters in a complicated manner as investigated by @xcite , @xcite , and @xcite .
for example , in our model with c / o=1.8 , only the 1st type cds forms every stellar pulsation period .
the values of mass loss rate , gas terminal velocity , and dust to gas mass ratio averaged over the last sixty pulsation periods at the outer boundary are 1.24@xmath1810@xmath19 @xmath2 yr@xmath17 , 20.4 km s@xmath17 , and 1.64 @xmath1810@xmath168 , respectively , in the hydrodynamical model with the lte case for the formation of sic grains .
as can be seen from figure 2 , the nucleation of sic grains onsets when the gas cools down to 1100 k after the formation of carbon grains , and is activated in the temperature range of 8001000 k. the corresponding formation region is located around 3.5 @xmath8 5.5 @xmath150 unless the gas temperature is raised up higher than 1100 k by the backwarming from the precedent 1st or 2nd type cds . in the formation region ,
si@xmath29c@xmath29 molecule is more abundant than sic molecule and the nucleation proceeds through the reaction ( 2 ) rather than the reaction ( 1 ) .
in particular , the formation of sic grains is most efficient in the region around the outer boundary of the 1st type cds because of the lower temperature and higher gas density , and the nucleation rate per h element reaches the maximum value of @xmath8 10@xmath169 s@xmath17 at @xmath94=167.9 @xmath7 .
however the subsequent decrease of the gas density caused by the accelerated radial expansion depresses the nucleation and growth processes of sic grains ; the number of sic grains per h element is limited to smaller than 10@xmath170 ( see the bottom row at @xmath94=168.8 @xmath7 ) , and the volume equivalent radius of sic grains @xmath171 is typically 36@xmath1810@xmath172 @xmath12 . in the other regions ,
the number and size of sic grains are much smaller .
thus the depletion of si bearing molecules does not take place substantially in lte case ( see the radial variation of @xmath162 in the bottom row throughout the time ) , and almost all of sic grains form around the outer boundary of carbon dust shell . figure 3 represents the the radial distribution of gas and grains in the region of @xmath173 25 @xmath150 at @xmath94=166.8 @xmath7 .
the radial distribution of @xmath161 ( the 2nd row ) clearly shows that the different type of carbon dust shell forms by turns as addressed in section 5.1 .
the carbon dust shells broaden with moving outwards from the formation sites because the outer part with more abundant carbon grains is accelerated more efficiently than the inner part , and the broadening in the outer circumstellar envelope of @xmath174 10 @xmath150 is caused by the pressure gradient .
in addition , the slower moving 2nd type cds seems to be caught up by the faster moving 1st type cds around 21 @xmath150 .
note that the shell ( shock ) structure seen in the radial distribution of gas density precedes the corresponding carbon dust shell in the outer circumstellar envelope ( see the top row ) .
the distributions of @xmath175 ( the top row ) and @xmath162 ( the 2nd row ) demonstrate that sic grains reside in thin shell around the outer boundary of carbon dust shell , reflecting the formation site .
the value of @xmath162 ( @xmath175/@xmath176 ) and the mass ratio of sic to carbon grains sic / c are at most 10@xmath16 ( 10@xmath177 ) and 10@xmath16 , respectively ( see the lower three rows ) .
the value of @xmath161 ( @xmath178/@xmath176 ) averaged over the last sixty pulsation periods at the outer boundary is 0.707 ( 1.64 @xmath18 10@xmath168 ) , while the averaged value of @xmath162 ( @xmath175/@xmath176 ) is in the order of magnitude of 10@xmath9 ( 10@xmath179 ) .
the resulting averaged value of sic / c is 4.93 @xmath18 10@xmath9 which is much smaller than the value inferred from the radiative transfer models ( 0.010.3 ) .
figure 4 shows the size distribution of sic grains by mass averaged over the last sixty pulsation periods at the outer boundary .
the distribution is characterized by the single peaked profile with the peak around @xmath75=6.0@xmath1810@xmath172 @xmath11 m , and the radius of sic grains is limited to be less than 0.1 @xmath11 m .
the modal size is almost a factor of four smaller than the size of presolar sic grains extracted from the murchison meteorite @xcite .
the lte case with lower c / o ratio resulting in lower gas outflow velocity makes the formation of sic grains with radius of 0.20.3 @xmath11 m possible as long as the gas outflow exhibits a layered structure with the cdss . on the other hand , regardless of c / o ratio , the amount of sic grains formed in the lte case is too small to reproduce amount of sic grains inferred from astronomical observations .
figure 5 displays the radial structure of the quantities related to gas dynamics and formation of grains in the non lte case throughout the period from @xmath94=166.7 to 168.8 @xmath7 .
the vibration temperature of sic cluster defined in section 4.2 and depicted in the top row ( solid red ) is evaluated by placing the small sic grain with @xmath143=10@xmath168 @xmath11 m hypothetically in the dynamical calculations .
the difference between the gas temperature @xmath135 and the vibration temperature @xmath0 is negligible in the high density region of @xmath66 @xmath180 10@xmath181 g @xmath25 where the collision with gas dominates the energy balance and controls @xmath0 . as the gas density decreases and the collision with gas gets less effective than the interaction with the radiation in the energy balance , @xmath0 deviates from and turns to be lower than @xmath135 because small sic grain is almost transparent from the visible to infrared region ( up to @xmath8 9 @xmath12 ) @xcite .
also , it should be noted that @xmath0 is not affected so much by the backwarming due to the thermal emission from the outer carbon dust shell compared with the gas temperature .
thus , the temperature difference @xmath182 is magnified in the less dense region inner than the opaque carbon dust shell ( e.g. , see around 2.5 @xmath150 at @xmath94=167.3 @xmath7 ) .
as the gas density decreases , the supersaturation ratio of sic grains quickly exceeds unity with increasing @xmath183 as well as decreasing @xmath0 .
the nucleation process of sic grains can be activated in the region with @xmath66 @xmath14 10@xmath181 g @xmath25 as long as @xmath184 1200 k and @xmath185 200 k , even in the region with @xmath186 1400 k where the nucleation process of carbon grains is almost depressed . on the other hand , if @xmath0 is higher than 1200 k , the nucleation process of sic grains is not activated regardless of the gas temperature since the abundances of molecules responsible for formation of sic grains are not sufficient to make the supersaturation ratio large enough for the onset of nucleation .
the active nucleation and growth of sic grains takes place in the infalling gas around 1.82.2 @xmath150 at @xmath94=166.7 @xmath7 when the gas density drops down to a few times 10@xmath24 g @xmath25 , @xmath0 decreases down below 1200 k and @xmath183 increases by larger than @xmath8 200 k , almost concurrently with the onset of formation of the 1st type cds as explained in section 5.1 .
then , the backwarming from the evolving 1st type cds raises up @xmath0 and @xmath135 high enough so as to evaporate sic grains in the inner dense region with @xmath66 @xmath187 10@xmath24 g @xmath25 at @xmath94=167.0 @xmath7 .
however , in the outer less dense region , sic grains grow up to @xmath8 0.3 @xmath12 thanks to the increase of @xmath183 and the condensation efficiency @xmath162 reaches 0.1 . in the region around 2.5 @xmath150 , the nucleation of sic grains proceeds , and at @xmath94=167.3 @xmath7 the increase of @xmath162 is distinguishable while carbon grains evaporate due to the backwarming , which continues up to @xmath165167.5 @xmath7 ( see the 3rd and bottom rows ) .
on the other hand , in the region around 2 @xmath150 , the evaporation of sic grains turns to the growth at @xmath94=167.3 @xmath7 with weakening the backwarming as the 1st cds moves outwards , and then the increase of gas density up to 10@xmath24 g @xmath25 caused by the pulsation shock leads to the efficient nucleation and growth of sic grains , and @xmath162 increases up to @xmath8 0.7 quickly until @xmath94=167.6 @xmath7 .
then , the efficient growth of sic grains in the region compressed by the merging of consecutive shocks starting from @xmath94 @xmath8 167.9 @xmath7 results in the complete consumption of si bearing molecules ( @xmath162=1.0 ) around @xmath165168.1 @xmath7 , and @xmath171 in this region is typically larger than @xmath8 0.2 @xmath12 , in contrast to the lte case .
the region with the condensation efficiency @xmath162 @xmath8 1.0 is referred to as the si
depletion region hereafter .
concurrently , the 2nd type cds formed and developed in this region stretches the si
depletion region outwardly ( see the radial distribution of @xmath162 in the bottom row from @xmath94=168.2 to 168.8 @xmath7 ) , while the backwarming from the 2nd type cds makes @xmath0 higher than 1200 k and depresses the nucleation of sic grains in the region inner than the si
depletion region for a while ( see @xmath0 in the top row at @xmath94=168.5 @xmath7 ) .
it should be noted that the 2nd type cds moves outward more slowly than the 1st type cds as mentioned in section 5.2 .
thus , the si depletion region with slower outward velocity is finally caught up and pushed by the 1st type cds as can be seen from the time evolution of the si
depletion region spreading over from 3.5 to 5.5 @xmath150 at @xmath94=167.3 @xmath7 which formed two cycles before . as mentioned above
, the formation of si depletion region onsets in the infalling gas in front of the coming pulsation shock , and the region moves outward with the formation of 2nd type cds triggered by the merger with the next pulsation shock .
thus , the si depletion region forms once every two pulsation periods . also , even in the other regions , the resulting value of @xmath162 often exceeds 0.1 .
thus , in contrast to the very small value of at most 10@xmath16 in lte case , a large amount of sic grains can condense through the inverse greenhouse effect in the non lte case .
figure 6 presents the radial variations of gas and grains at @xmath94=176.3 @xmath7 .
it seems that sic grains are concentrated around the outer boundaries of 1st type cdss in the density distribution ( see the top row ) because the shock associated with the formation of carbon dust shell sweeps up the gas .
reflecting the variation of gas density , the radial distribution of @xmath175 does not always corresponds to the si
depletion region shown in the 2nd row .
it should be noted that the extended si depletion region does not develop without the 2nd type cds , which can be demonstrated from an example of the failure of alternating formation of both types of carbon dust shells located at 18 @xmath150 where si depletion region restricted in a narrow region resides just ahead of the 1st type cds formed about 1 cycle after from the formation of the si depletion region ( see the 2nd row ) .
apart from the 1st type cds located at 18 @xmath150 , the extended sic dust shell corresponding to the si depletion region between 1st and 2nd type cdss can be clearly recognized in the radial distribution of @xmath188 ( see the 3rd row where @xmath188=8.46@xmath1810@xmath189 corresponding to @xmath162 = 1.0 ) .
the radial distribution of @xmath162 shows that the si depletion region stretched over by the slower moving 2nd type cds narrows with moving outwardly , being compressed by the faster moving 1st type cds .
thus , in the outer circumstellar envelope , the radial width tends to be of the order of the stellar radius ( see @xmath162 around 24 @xmath150 where the 2nd type cds almost merges with the 1st type cds ) and the narrower si depletion region seems to reside around the outer boundary of the 1st type cds .
the radial distribution of the mass ratio sic / c has a sharp peak in the region just ahead of the outer boundary of 1st type cds , and decrease by 0.1 ( 10@xmath189 ) toward the outer boundary of precedent 2nd ( 1st ) type cds ( see the bottom row ) .
the peak value exceeds ten in the inner circumstellar envelope , and decreases with increasing @xmath119 because of the subsequent growth of carbon grains .
however the value of sic / c in the si depletion region still keeps larger than 0.2 with the peak value exceeding unity .
the resulting values of mass loss rate , terminal velocity , dust to gas mass ratio averaged over the last sixty stellar pulsation periods are 1.10@xmath1810@xmath19 @xmath2 yr@xmath17 , 19.5 km s@xmath17 , and 1.85@xmath1810@xmath168 , respectively , for the non lte case .
the reduction of terminal velocity compared to the lte case reflects the somewhat weakened acceleration resulting from the reduction of amount of carbon grains due to the incorporation of c bearing gas species into sic grains ; on average 4.40 % of carbon atoms in the condensible c bearing molecules is incorporated into the sic grains .
the averaged value of @xmath190 ( @xmath161 ) is 1.51@xmath1810@xmath168 ( 0.652 ) , while the value of @xmath188 ( @xmath162 ) is 3.41 @xmath1810@xmath189 ( 0.403 ) .
the resulting averaged value of sic / c is 0.226 and is close to the upper value derived from the radiative transfer models .
the size distribution of sic grains by mass averaged over the last sixty stellar pulsation periods at the outer boundary is depicted in figure 7 .
the size distribution shows the broad profile with the mean radius of 0.3 @xmath12 , in contrast to the sharp and single peaked size distribution in the lte case presented in figure 4 . in the size range from 0.03 to 4.0 @xmath12 ,
80 % of sic grains by mass populate in the radii between 0.2 and 1.0 @xmath12 : the distribution well covers the size range of presolar sic grains found in the murchison meteorite . thus the hydrodynamical model with the non lte case for the formation of sic grains that the vibration temperature of sic cluster is assumed to be the same as the temperature of small sic grains reasonably reproduces the amount of sic grains inferred from astronomical observations and the size of presolar sic grains extracted from the murchison meteorite .
in the calculations of hydrodynamical models presented in section 5 , we assume that sic is the starting molecule for the nucleation process of sic grains , excluding the process starting from the two body reaction of si@xmath29c@xmath29 molecules .
this may underestimate the amount of sic grains in the lte case since si@xmath29c@xmath29 molecule is more abundant in the lower temperature region than sic molecule as shown in figure 1 ( left ) . in the lte case where carbon grains condense in prior to sic grains ,
the nucleation process starting from the two body reaction of si@xmath29c@xmath29 molecules dominates the nucleation rate in the region around 68 @xmath150 whose gas temperature ranges from 600 to 800 k. however the gas density in the region of @xmath19110@xmath192 g @xmath25 is too low to increase the amount of sic grains ; even if this process is included in the calculation , the resulting values of both @xmath162 and sic / c are limited to 10@xmath189 although the nucleation rate is much larger than that starting from sic molecule .
in addition , this nucleation process produces the population of grains whose radii are much smaller than 6.0 @xmath18 10@xmath172 @xmath12 , and make the modal radius smaller than 3.0 @xmath18 10@xmath172 @xmath12 .
thus the nucleation process starting from two body reaction of si@xmath29c@xmath29 molecules does not change the conclusion that the hydrodynamical models with the lte case can not reproduce the amount of sic grains around c rich agb stars inferred from the astronomical observation and the size range derived from the analysis of presolar grains . in the non lte case , this nucleation process dominates the formation of sic grains in the region of @xmath1744 @xmath150 , and is activated particularly in the outer part of the carbon dust shell .
although the number of sic grains produced through this process reaches about 10 times that produced through the nucleation starting from sic molecule , the typical radius of sic grains formed in the region with low gas density is limited to 3@xmath1810@xmath1686@xmath1810@xmath193 .
thus , the inclusion of this nucleation process does not affect the size distribution of sic grains whose radius is larger than 0.1 @xmath12 .
also it is true for the amount of sic grains since the averaged @xmath162 increases by only a few percent .
thus we conclude that the exclusion of nucleation process starting from the two body reaction of si@xmath29c@xmath29 molecules does not affect the result of calculations in the non lte case , except for the size distribution in the range of @xmath194 6@xmath1810@xmath193 .
the sticking probability set to be unity for nucleation and growth processes of sic grains may overestimate the size and amount of sic grains formed in the non lte case .
figure 8 shows the time averaged size distributions by mass of sic grains for given values of @xmath10 . in the range of @xmath10=0.11.0 , the mass fraction of large sized grains with @xmath75 @xmath180 1.0
@xmath12 decreases significantly and the profile becomes narrow with decreasing @xmath10 , while the reduction of @xmath10 does not affect the mass fraction of small
sized grains with @xmath75 @xmath1950.1 @xmath12 .
the modal radii are 0.30 and 0.22 @xmath12 for @xmath10=0.5 and 0.1 , respectively .
although the sic grains with radii larger than 1.0 @xmath12 do not form for the case of @xmath10=0.1 , the hydrodynamical models in the non lte case reasonably reproduce the size range of presolar sic grains extracted from the murchison meteorite @xcite as long as @xmath196 0.1 .
the averaged value of sic / c decreases with decreasing @xmath10 ; the values are 0.199 and 9.77@xmath1810@xmath172 for @xmath10=0.5 and 0.1 , respectively , which is within the range of the value inferred from the radiative transfer models .
although the reduction of @xmath10 changes the radial distribution of sic grains substantially , we conclude that the hydrodynamical models in the non lte case cover the ranges of size derived from the analysis of presolar grains and amount evaluated from the radiative transfer models unless the sticking probability @xmath1970.1 .
the mass ratio of sic / c inferred from the radiative transfer models depends on the models used in the calculation ; the evaluated values for c rich agb stars with the emission feature of sic grains are in the range of 0.010.29 @xcite , 0.010.15 @xcite , and 0.10.3 @xcite . the smaller value by @xcite and @xcite is considered to be ascribed to the grain model ; the temperature of composite grain consisting of carbon and sic used in their model is higher than that of isolated sic grains used in the other models as is shown in @xcite , since carbon grains are more absorptive in the visible to nir region than sic grains .
therefore , in order to reproduce the observed seds , if the inner boundary of dust layer is placed close to the photosphere , the high temperature of composite grains causes the value of sic / c to reduce . also , the smaller value of 0.010.06 derived by @xcite for c rich agb stars showing sic emission feature with optically thick shells ( @xmath198 ) would arise from the optical constant of sic grains used in the models
; the radiative transfer models usually employ the optical constants derived from the sic powder synthesized in the laboratory @xcite .
the absorption coefficients in visible to infrared region ( up to @xmath8 9 @xmath12 ) is much larger than the crystalline sic @xcite , which results in the higher temperature of sic grains and the smaller sic / c .
in addition , the mid infrared observation of c rich agb stars by _ spitzer _ shows that the 30 @xmath12 emission feature attributed to mgs grains becomes prominent with decreasing the strength of sic emission feature in the stars with mass loss rate larger than 10@xmath16 @xmath2 yr@xmath17 @xcite .
@xcite demonstrates that the absorption efficiency around 11.3 @xmath12 of grains consisting of a sic core and a mgs mantle decreases with increasing the volume fraction of mgs mantle .
therefore , if it is true , the radiative transfer model could underestimate the value of sic / c without including sic grains coated by mgs mantle .
thus , the mass ratio of sic to carbon grains derived from the the radiative transfer models depends on the optical constants of sic as well as the model of grains used in the calculations , and the value of sic / c up to @xmath8 0.23 calculated in the non lte case for c rich agb stars is considered to be not always in conflict with the values derived from sed fittings .
the formation of sic grains around c rich agb stars is investigated for the first time in the framework of hydrodynamical model for the pulsation enhanced dust driven wind .
we formulate the nucleation and growth processes of sic grains , considering that sic grains nucleate and grow homogeneously starting from sic molecule . in the calculations ,
the two cases are considered for the nucleation process of sic grains ; the lte case in which the vibration temperature @xmath0 of sic cluster is equal to the gas temperature and the non lte case in which @xmath0 is assumed to be the same as the temperature of small sic grain whose radius is 10@xmath168 @xmath12 . on the other hand ,
the vibration temperature of carbon cluster is considered to be the same as the gas temperature as is usually assumed in the hydrodynamical models for c rich agb stars .
the results of calculations for the model parameters @xmath1=1.0 @xmath2 , @xmath3=10@xmath4 @xmath5 , @xmath6=2600 k , c / o ratio=1.4 , @xmath7=650 days , and @xmath149=2.0 km s@xmath17 are summarized as follows : in the lte case , sic grains form in the accelerated and low density outflowing gas after the formation of carbon grains .
the resulting time averaged number fraction of si atoms locked into sic grains ( condensation efficiency @xmath162 ) of at most 10@xmath9 is too small to reproduce the amount of sic grains inferred from the astronomical observations and the radius is limited to less than 0.1 @xmath12 . on the other hand , in the non lte case , the time averaged mass ratio of sic to carbon grains ranges from 0.098 to 0.23 for the sticking probability @xmath10=0.11.0 , which is not in conflict with the value of 0.010.3 inferred from the radiative transfer models .
the time averaged size distribution of sic grains by mass with the peak at radius of 0.20.3 @xmath12 well covers the size range of presolar sic grains extracted from the murchison meteorite .
thus we conclude that the so called inverse greenhouse effect plays a crucial role in the formation process of sic grains in the pulsation enhanced dust driven winds from c rich agb stars , apart from the validity of the underlying assumption made in the non lte case .
the non lte effect should be explored for the formation process of dust grains in astrophysical environments .
the hydrodynamical model provides the two types of carbon dust shells ; the 1st ( 2nd ) type cds with the peak value of @xmath1640.9 ( 0.3 @xmath167 0.7 ) .
the formation region and the resulting radial distribution of sic grains in the hydrodynamical model with non lte case is closely related with the formation and dynamics of carbon dust shell , especially through the density enhancement and the backwarming effect . in the inner circumstellar envelope , the sic dust shell can be discriminated from the 1st type cds through the extended si
depletion region caused by the 2nd cds formed in the outer part of active sic formation region .
however , in the outer circumstellar envelope of @xmath17420 @xmath150 , the sic dust shell is localized around the outer boundary of 1st type cds .
not only the formation region and the radial distribution but also the amount and the size distribution of sic grains could heavily depend on the c / o ratio as well as the other model parameters ; for instance , in the non lte case with @xmath1=2.0 @xmath2 , the mass ratio of sic to gas and the size of sic are almost the same as those with @xmath1=1.0 @xmath2 while the mass loss rate decreases by a factor 5 .
anyway more comprehensive study covering a wide range of model parameters is necessary for revealing the amount and the size distribution of sic grains formed around c rich agb stars . also , the difference between the radial distributions of carbon and sic grains could greatly influence the appearance of the spectral feature attributed to sic grains .
these aspects will be investigated in the forthcoming papers .
we are grateful to the anonymous referee for valuable comments that improved the manuscript .
we thank dr .
k. ohnaka for his critical reading of the first draft and useful comments .
this research has been partly supported by the grant in
aid for scientific research of the japan society for the promotion of science ( 18104003 , 20340038 ) .
allen , c. w. 1973 , astrophysical quantities , the athlone press , london allen , b. c. , & kingery , w. d. 1959 , trans . metal .
aime , 215 , 30 amari , s. , lewis , r. s. , & anders , e. 1994 , , 58 , 459 bergeat , j. , knapik , a. , & rutily , b. 2002 , , 390 , 987 bergeat , j. , & chevallier , l. 2005 , , 429 , 235 berman , r. g. , & brown , t. h. 1985 , contr . miner .
, 89 , 168 bernatowicz , t. j. , cowsik , r. , gibbons , p. c. , et al .
1996 , , 472 , 760 bernatowicz , t. j. , fraundorf , g. , ming , t. , et al .
1987 , , 330 , 728 blanco , a. , borghesi , a. , fonti , s. , & orofino , v. 1994 , , 283 , 561 blanco , a. , borghesi , a. , fonti , s. , & orofino , v. 1998 , , 330 , 505 borghesi , a. , bussoletti , e. , colangeli , l. , & de blasi , c. 1985 , , 153 , 1 bowen , g. h. 1988 , , 329 , 299 chase , m. w. , jr . ,
davies , c. a. , downey , j. r. , jr .
1985 , j. phys .
chem . ref .
data , 14 , suppl .
1 cherchneff , i. , barker , j.r . , &
tielens , a.g.g.m .
1992 , , 401 , 269 choyke , w. j. , & palik , e. d. 1985 , in handbook of optical constants of solids , palik e.d .
( ed . ) , academic press , boston , p587 croat t. k. , stadermann , f. j. , & bernatowicz , t. j. 2005 , , 631 , 976 croat t. k. , stadermann , f. j. , & bernatowicz , t. j. 2010 , , 139 , 2159 danchi , w. c. , bester , m. , degiacomi , c. g. , mccullough , p. r. , & townes , c. h. 1990 , 359 , l59 daulton , t. l. , bernatowicz , t. j. , lewis , r. s. , et al .
2003 , , 67 , 4743 deng , j. l. , su , k. h. , wang , x. , et al .
2008 , eur .
j. d , 49 , 21 draine , b. t. 1985 , , 57 , 587 dreyer , c. , hegmann , m. , & sedlmayr , e. 2009 , , 499 , 765 dreyer , c. , hegmann , m. , & sedlmayr , e. 2011 , , 525 , a135 ferrarotti , a. s. , & gail , h .-
2006 , , 447 , 553 fleischer , a. j. 1994 , phd thesis , technische universit@xmath199 t , berlin , frg fleischer , a. j. , gauger , a. , & sedlmayr , e. 1992 , , 266 , 321 fleischer , a. j. , gauger , a. , & sedlmayr , e. 1995 , , 297 , 543 gail , h .-
, & sedlmayr , e. 1988 , , 206 , 153 gauger , a. , gail , h .-
p . , & sedlmayr , e. 1990 , , 235 , 345 goebel , j. h. , bregman , j. d. , goorvitch , d. , et al .
1980 , , 235 , 104 groenewegen , m. a. t. 1995 , , 293 , 463 groenewegen , m. a. t. , whitelock , p. a. , smith , c. h. , & kerschbaum , f. 1998 , , 293 , 18 hackwell , j. a. 1972 , , 21 , 239 hasegawa , h. , & kozasa , t. 1988 , prog . theor .
suppl . , 96 , 107 helling , ch . , winters , j. m. , & sedlmayr , e. 2000 , , 358 , 651 hofmeister , a. m. , pitman , k. m. , goncharov , a. f. , & speck , a. k. 2009 , , 696 , 1502 hfner , s. , feuchtinger , m.u . , & dorfi , e.a .
1995 , , 297 , 815 hfner , s. , fleischer , a. j. , gauger , a. , et al .
1996 , , 314 , 204 hfner , s. gautschy - loidl , r. , aringer , b. , & j@xmath200rgensen , u. , g. 2003 , , 399 , 589 hoppe , p. , amari , s. , zinner , e. , ireland , t. & lewis , r. s. 1994 , 430 , 870 hoppe , p. , & ott , u. 1997 , in astrophysical implications of the laboratory study of presolar materials , bernatowicz , t. j. & zinner , e. ( ed . ) , aip , new york , p27 hoppe , p. , & zinner , e. 2000 , , 105 , 10371 jones , b. , merrill , k. m. , puetter , r. c. , & willner , s. p. 1978
, , 83 , 1437 kozasa , t. , dorschner , j. , henning , t. , & stognienko , r. 1996 , , 307 , 551 kozasa , t. , & hasegawa , h. 1987 , prog .
theor . phys . , 77 , 1402 krger , d. , woitke , p. , & sedlmayr , e. 1995 , , 113 , 593 leisenring , j. m. , kemper , f. , & sloan , g.c .
2008 , , 681 , 1557 lide , d. r. 2003 , crc handbook of chemistry and physics , 84th ed . , crc press , new york lodders , k. , & fegley , b. , jr 1995 , meteoritics , 30 , 661 lorenz - martins , s. , & lef@xmath201vre , j. 1993 , , 280 , 567 lorenz - martins , s. , & lef@xmath201vre , j. 1994 , , 291 , 831 mattsson , l. , wahlin , r. , & h@xmath202fner , s. 2009 , , 509 , a14 mccabe , e. , m. 1982 , , 200 , 71 nowotny , w. , lebzelter , t. , hron , j. , & hfner , s. 2005 , , 437 , 285 ott , u. in lecture notes in physics 815 , henning , t. ( ed . ) , springer , berlin , p277 p@xmath203gouri@xmath203 , b. 1988 , , 194 , 335 pitman , k. m. , hofmeister , a. m. , corman , a. b. , & speck , a. k. 2009 , , 483 , 661 richtmyer r. d. , & morton , k. w. 1967 , difference methods for initial value problems , 2nd ed . , john wiley & sons , new york rouleau , f. & martin , p. g. 1991 , , 377 , 526 russell , s. s. , ott , u. , alexander , c. m. od . , zinner , e. , arden , j. w. , & pillinger , c. t. 1997 , meteorite .
, 32 , 719 sharp , c. m. , & wasserburg , g. j. 1995 , , 59 , 1633 speck , a. k. , barlow , m. j. , & skinner , c. j. 1997 , , 288 , 431 speck , a. k. , corman , a. b. , wakeman , k. , wheeler , c. h. , & thompson , g. 2009 , , 691 , 1202 spitzer , w. g. , kleinman , d. , & walsh , d. 1959 , phys .
rev . , 113 , 127 stroud , r. m. & bernatowicz , t. j. 2005 , lunar planet .
, 36 , 2010 treffers , r. , & cohen , m. 1974 , , 188 , 545 winters , j. m. , dominik , c. , & sedlmayr , e. 1994 , , 288 , 255 winters , j. m. , fleischer , a. j. , le bertre , t. , & sedlmayr , e. 1997 , , 326 , 305 winters , j. m. , le bertre , t. , jeong , k. s. , helling , ch . , & sedlmayr , e. 2000 , , 361 , 641 woitke , p. 2006 , , 460 , l9 woitke , p. , & niccolini , e. 2005 , , 433 , 1101 yamasawa , d. , habe , a. , kozasa , t. , et al .
2011 , , 735 , 44 zhukovska , s. , gail , h .-
p . , & trieloff , m. 2008 , , 479 , 453 zhukovska , s. , & gail , h .- p .
2008 , , 486 , 229 zinner , e. , tang , m. , & anders , e. 1987 , , 330 , 730 & & + @xmath204 & -751.5 & -751.0 + @xmath205 & -518.8 & -514.5 + @xmath206 & 76.98 & 78.12 + @xmath207 & 309.6 & 314.7 + @xmath208 & -54.88 & -58.76 + @xmath209 & 177.8 & 177.8 + @xmath210 & -123.0 & -131.1 + @xmath211 & 109.7 & 105.4 + @xmath212 & -110.8 & -109.8 + @xmath213 & 121.9 & 126.7 + @xmath214 & 232.7 & 236.5 + @xmath215 & -199.5 & -202.6 + @xmath216 & 110.1 & 112.1 + @xmath217 & 33.14 & 33.97 + @xmath218 & -53.73 & -52.46 + @xmath219 & 178.9 & 184.1 + @xmath220 & -122.4 & -119.9 + @xmath221 & -122.6 & -124.9 + @xmath222 & 110.1 & 111.6 + @xmath223 & -195.5 & -199.5 + @xmath224 & -16.53 & -15.41 + @xmath225 & 37.19 & 37.04 + @xmath226 & 352.2 & 347.9 + @xmath227 & 584.8 & 584.5 + & & + @xmath228 & -793.8 & -788.5 + @xmath229 & 34.64 & 40.66 + @xmath230 & -97.22 & -96.22 + @xmath231 & -165.3 & -168.6 + @xmath232 & -153.1 & -147.3 + @xmath233 & -836.2 & -826.0 + @xmath234 & -42.34 & -37.47 + @xmath235 & -207.2 & -199.4 + @xmath236 & -241.9 & -240.0 + @xmath237 & -96.07 & -89.92 + @xmath238 & -286.6 & -284.1 + @xmath239 & -439.8 & -431.3 + @xmath240 & -164.9 & -162.4 + @xmath241 & -237.8 & -237.0 + @xmath242 & -333.9 & -326.9 + @xmath243 & 309.8 & 310.4 + ccccccc i & @xmath60 & @xmath61 & @xmath62 & @xmath63 & @xmath64 & @xmath65 + 1 & 3.11656@xmath1810@xmath244 & -1.12418@xmath1810@xmath245 & 1.71784@xmath1810@xmath246 & -1.01082@xmath1810@xmath247 & 1.01432@xmath1810@xmath248 & -8.78548@xmath1810@xmath249 + 2 & -2.96847@xmath1810@xmath250 & 2.79219@xmath1810@xmath246 & 5.55156@xmath1810@xmath4 & -8.91964@xmath1810@xmath247 & 1.15759@xmath1810@xmath251 & -1.28618@xmath1810@xmath252 + | we investigate the formation of silicon carbide ( sic ) grains in the framework of dust driven wind around pulsating carbon rich asymptotic giant branch ( c rich agb ) stars in order to reveal not only the amount but also the size distribution .
two cases are considered for the nucleation process ; one is the lte case where the vibration temperature of sic clusters @xmath0 is equal to the gas temperature as usual , and another is the non lte case in which @xmath0 is assumed to be the same as the temperature of small sic grains .
the results of hydrodynamical calculations for a model with stellar parameters of mass @xmath1=1.0 @xmath2 , luminosity @xmath3=10@xmath4 @xmath5 , effective temperature @xmath6=2600 k , c / o ratio=1.4 , and pulsation period @xmath7=650 days show the followings : in the lte case , sic grains condense in accelerated outflowing gas after the formation of carbon grains and the resulting averaged mass ratio of sic to carbon grains of @xmath8 10@xmath9 is too small to reproduce the value of 0.010.3 inferred from the radiative transfer models . on the other hand , in the non lte case
, the formation region of sic grains is inner than and/or almost identical to that of carbon grains due to the so called inverse greenhouse effect .
the mass ratio of sic to carbon grains averaged at the outer boundary ranges from 0.098 to 0.23 for the sticking probability @xmath10=0.11.0 .
the size distributions with the peak at @xmath8 0.20.3 @xmath11 m in radius cover the range of size derived from the analysis of presolar sic grains .
thus the difference between temperatures of small cluster and gas plays a crucial role in the formation process of sic grains around c rich agb stars , and this aspect should be explored for the formation process of dust grains in astrophysical environments . |
in climate dynamics research , analysis of time series data has a central position .
detection and quantification of dependence between measured or modeled variables is often of interest .
apart from the dependences among different physical quantities , in many contexts the task is given to assess the dependence among the measurements of the same physical variable ( i.e. surface air temperature , sat ) measured at many different geographical locations .
the motivation for such a procedure commonly stems from the need to reduce the dimensionality of high - dimensional original data , such as in the application of empirical orthogonal function analysis @xcite to uncover the basic modes of dynamics of climate system . on the other side ,
dependence quantification might be used to uncover the complex structure of the climate system using approaches such as graph theory @xcite .
other applications exist including those combining the above named , see e.g. @xcite .
there is a wide range of methods available for detection of dependence between variables .
the most widely known and used is pearson s correlation coefficient , a measure particularly sensitive to linear dependence .
while the pearson s correlation detects dependence reliably in the case of multivariate gaussian probability distributions , it may be suboptimal in the case of complex non - gaussian dependence patterns .
for particular dependence patterns ( or bivariate probability distributions ) , it may also fail to detect statistical dependence between variables of interest completely .
note that in the following we use the terms ` linear ' and ` gaussian ' dependence interchangeably to denote patterns of dependence corresponding to bivariate normal distribution .
while the latter term is more precise , the former is more commonly used in the general community together with the distinction between linear and nonlinear methods .
however , alternative measures exist that are able to better reflect potential non - linear dependences .
these include the spearman s ordinal correlation coefficient @xcite and kendal s tau @xcite , that are designed to be sensitive to any monotonous dependence pattern , without the restriction to linear relationships .
an ultimate alternative to the pearson s correlation coefficient then lies in the utilization of mutual information @xcite , an information - theory based measure that is in principle sensitive to any dependence between variables . for this generality ,
mutual information is widely used to quantify statistical dependence in complex systems , and has been also introduced to the analysis of climate time series @xcite .
as the climatic system is highly nonlinear , it seems well motivated to use nonlinear dependence measures for analysis of the measured time series , as suggested e.g. in @xcite .
this may in theory allow more sensitive detection and quantification of dependences , potentially uncovering new climatic phenomena .
on the other side , nonlinear measures such as mutual information may have downsides including more difficult implementation and interpretation , increased computational demands and numerical stability issues .
these considerations motivate the central question of the current report : does the non - linear component of the climate time series dependences sufficiently motivate the use of nonlinear dependence measures ?
it is important to note , that the answer to this question might be complex , and certainly would be domain specific . to deal with this complication , we outline here first a generally applicable framework , and then show the results obtained by analyzing in detail a specific dataset of particular interest .
this is the monthly sat data from the ncep / ncar reanalysis dataset @xcite , as well as concatenated era-40 @xcite and era - interim @xcite data .
note that this data has been analyzed in many recent studies , utilizing both linear and nonlinear methods , and so constitutes a well motivated timely and relevant example of application of this framework to guide the decision regarding the method choice .
data from the ncep / ncar reanalysis dataset @xcite have been used . in particular , we utilize the time series @xmath0 of the monthly mean sat from january 1948 to december 2007 ( @xmath1 time points ) , sampled at latitudes @xmath2 and longitude @xmath3 forming a regular grid with a step of @xmath4 .
the points located at the globe poles have been removed , giving a total of @xmath5 spatial sampling points . to explore the generalizability of the results ,
analogous analysis has been carried out for the era-40 dataset @xcite concatenated with the era - interim dataset @xcite ( further referred to together as the era data ) . to minimize the bias introduced by periodic changes in the solar input , the mean annual cycle
is removed from the data to produce so - called anomaly time series .
we discuss two dependence measures throughout the paper , pearson s correlation coefficient ( linear correlation ) and ( nonlinear ) mutual information . given two real random variables @xmath6 , the well - known pearson s correlation coefficient is defined as @xmath7}{e((x - e(x))^2)e((y - e(y))^2)}},\ ] ] where @xmath8 denotes the expected value operator .
the corresponding finite sample estimate is denoted by @xmath9 . for two discrete random variables @xmath6 with sets of values @xmath10 and @xmath11 ,
the mutual information is defined as @xmath12 where @xmath13 is the probability distribution function of @xmath14 , @xmath15 is the probability distribution function of @xmath14 and @xmath16 is the joint probability distribution function of @xmath14 and @xmath17 . for continuous variables ,
mutual information is defined by the respective integral . however , in practice the mutual information is estimated using discretization of the theoretically continuous variables .
when the discrete variables @xmath6 are obtained from continuous variables on a continuous probability space , then the mutual information depends on a partition @xmath18 chosen to discretize the space .
a common choice is a simple box - counting algorithm based on marginal equiquantization method @xcite , i.e. , a partition is generated adaptively in one dimension ( for each variable ) so that the marginal bins become equiprobable .
this means that there is approximately the same number of data points in each marginal bin . in this paper
we use a simple pragmatic choice of @xmath19 bins for each marginal variable @xcite .
mutual information is a non - negative quantity , with @xmath20 corresponding to independence of the variables @xmath14 and @xmath17 , and units depending on the base of the logarithm ( base 2 corresponds to bits , while natural logarithm with base @xmath21 corresponds to nats , used here ) .
as the estimation of mutual information from finite sample size is prone to sample size dependent bias , to allow quantitative comparison we carry out an approximate correction by recalibration procedure described elsewhere @xcite .
this procedure is in general based on comparison with samples of the same size and coming from populations with analytically established mutual information . to elucidate our nonlinearity assessment strategy ,
we first point out that for a bivariate gaussian distribution , the correlation @xmath22 of the variables @xmath23 uniquely defines the mutual information between them , which is given by @xmath24 .
however , for a general non - gaussian bivariate distribution , this equation may not hold .
two cases of bivariate non - gaussianity can be distinguished .
firstly , the ` simpler ' nonlinearity consists in non - linear rescaling of one or both of the variables .
such a rescaling does not affect the mutual information between the variables , however , the correlation may change substantially .
rescaling of this type can be suspected in data e.g. due to non - linear properties of the measurement scale , and may be considered as a bias in the correlation estimation .
a remedy commonly adopted is the use of spearman rank correlation coefficient .
alternative procedure lies in preprocessing the data by applying a monotonous transformation to each variable separately that would render it gaussian ( `` marginal normalization '' ) , and computing the correlation on the transformed data .
a second , more ` substantial ' type of non - gaussianity lies in that some bivariate distributions differ from bivariate gaussian not only in their marginal distributions , but also in the form of the interdependence , which can not be resolved by univariate rescaling .
this substantial non - gaussianity is the key motivation for the use on nonlinear dependence measures , as the dependence pattern can not be recovered by only considering ranks or other rescaled version of the variables .
recently , a quantification method for such deviation from gaussian dependence has been proposed @xcite , building on the fact that for univariately gaussian random variables @xmath23 , the correlation gives a lower bound on the mutual information by @xmath25 with the minimum obtained for bivariate gaussian distribution . in particular , one can define the neglected ( ` extra - normal ' or ` non - gaussian ' ) information @xmath26 our investigation of the impact of nonlinear contributions to the climate dependence network is carried out in several steps with increasing level of detail . as a first step , for each pair of local time series @xmath27 , the correlation @xmath28 and mutual information @xmath29 of each pair of local time series
is computed , and their overall relation shown in a scatter plot for visual inspection of systematic deviation from the relation valid under gaussianity .
minor noisy deviations are of course expected due to estimation of the quantities from finite size samples .
the step is repeated for univariately normalized variables to isolate the effect of simple rescaling ; the normalized variables are further used in the subsequent analysis .
notably , the deviation of the relation among the two estimated quantities @xmath28 and @xmath29 from the theoretical prediction under gaussianity ( @xmath30 ) could be still attributed to the nonlinear properties of the bi - variate dependences , but also to different estimator properties of correlation and mutual information ( including residual mutual information estimator bias uncorrected by the procedure mentioned above ) . to isolate the genuine nonlinearity from apparent differences due to different estimator properties , and to allow more robust quantitative comparison ,
the mutual information estimates from data @xmath14 are further compared to mutual information estimates obtained for the respective pairs of _ linear surrogate data _ @xmath31 .
the surrogate data conserve the linear structure ( covariance and autocovariance ) and hence also the correlations of the original data , but remove non - gaussianity ( non - linearity ) from the multivariate distribution .
thus , in the large sample size limit , the mutual information between pairs of variables in the surrogate data should be @xmath32 technically , linear surrogate data are conveniently constructed as multivariate fourier transform ( ft ) surrogates @xcite ; i.e. obtained by computing the fourier transform of the series , keeping unchanged the magnitudes of the fourier coefficients ( the amplitude spectrum ) , but adding the same random number to the phases of coefficients of the same frequency bin ; the inverse ft into the time domain is then performed .
thus , instead of comparing the estimators of two different quantities @xmath28 and @xmath29 ( or using the rescaled version of the first one , i.e. @xmath33 , as an estimate of linear information based on correlation estimator @xmath9 ) , we can compare the values @xmath29 and @xmath34 obtained using the same estimator on both the original dataset @xmath14 and its linearized ( surrogate ) version @xmath35 .
as an added value , generation of the surrogates allows direct statistical testing of the gaussianity of the studied process , as they provide random samples with predefined covariance structure . for this second purpose , a set of @xmath36 such surrogate datasets is generated . for each pair of locations @xmath37 , one can test the hypothesis that the time series @xmath38 come from a bivariate linear stochastic process with gaussian dependence among the variables by comparing the obtained mutual information estimate @xmath29 with the empirical distribution of gaussian mutual information estimate @xmath39 .
the one - sided hypothesis is rejected at significance level @xmath40 if the data mutual information is higher than at least @xmath41 ( out of @xmath42 ) of the surrogate mutual information values .
apart from testing statistical significance of the deviations from linearity , we also describe the strength and localization of the effect .
this can be conveniently vizualized by estimating for each location the average mutual information between the location @xmath43 and all other locations : for original data by @xmath44 and for a surrogate dataset by @xmath45
. geographical rendering of the difference @xmath46 and the relative difference @xmath47 allows effective visual inspection of the most substantial nonlinearity localization . such automatic localization of substantially nonlinear dependencies can be followed by inspection of the corresponding temporal patterns and expert assessment of their relevance for the studied phenomena .
based on this investigation , the use of nonlinear measures for ( a subset of ) the data may be recommended to exploit the nonlinear information in the data , or alternatively further postprocessing may be proposed to clean the data from spurious sources of apparent nonlinearity , as shown below .
in particular , the intermediate results of the investigation motivated the introduction of two additional data preprocessing steps adopted in later sections of the paper .
the first is the removal of seasonality in variance ( variance normalization ) , which removes the differences in local temperature variability in different parts of year . to achieve that , standard deviation of temperature anomalies is computed for each month separately and
the anomaly data from given month are divided by this standard deviation .
the second additional preprocessing step was the removal of slow trends from the data . for simplicity ,
only a linear trend was considered here .
the general relation between the mutual information and linear correlation for pairs of sat anomalies is visualized in figure [ fig : figuremivsc ] .
the mutual information is in general very strongly related to the correlation .
the relation is even clearer after removing the estimator differences by representing the linear dependence in terms of mutual information in the surrogates , see figure [ fig : figuremivsmi ] .
the univariate gaussianization does not change the correlation coefficients substantially ( see figure [ fig : figurecvscmarg ] ) and does not affect much the relation between data and linear surrogates ( see figure [ fig : figuremivsmimarg ] ) .
the deviation from purely gaussian structure of the data has been tested separately for each pair of variables by comparison with a surrogate dataset distribution .
more than @xmath48% time series pairs showed significant at the @xmath49 level . to investigate the observed deviations from linear dependences in more detail
, we visualize the total and nonlinear contributions to dependence patterns in figure [ fig : lokalizace - skalovane - anom ] . at first inspection
one can see several well defined areas of relatively high nonlinear contribution to dependence patterns .
these are in particular an extensive ring around the anctarctica within the southern ocean , a few locations close to the north pole ( barentz sea , bering sea , baffin bay , greenland sea ) and areas in brazil and southwest asia . to understand the nonlinear dependence pattern in more detail , we select the locations with the highest relative nonlinear dependences and visualize both their linear and nonlinear dependence pattern with respect to all other locations , see figure [ fig:6fcpatterns ] . for most of these areas ,
the nonlinear dependences are not generally stronger , but rather include additional distant locations , in contrast with the mostly local character of linear dependence patterns .
this might suggest the existence of long - range interactions or `` teleconnections '' of highly or predominantly non - linear character , as discussed e.g. in @xcite . to elucidate the nature of these long - range connections we inspected the bivariate distributions and time series of the variables .
a representative example is shown in figure [ fig : figure_scatter ] .
the shape of the bivariate distribution together with close inspection of the time series suggests that the non - gaussianity might be related to seasonal variability in variance of the signal , which further differs between the two locations . in this particular case ,
the variability at the first location is the highest in december to february , when it is at its lowest at the second location and vice versa in july .
thus , the information shared by these time series would be explainable just by the seasonal differences of dynamics and ultimately just by variation in local solar influx . to test this hypotheses and control for this source of bias , we re - analyze the data after normalizing the seasonal variance , as described in section [ sec : mat ] .
after this additional preprocessing step , there was a marked decrease in detected pairs of locations with statistically significant nonlinear contribution to temperature dependence ( @xmath50 at the @xmath49 level ) , however , this is still more than expected by chance .
the localization of nodes with the strongest ( non)linear dependences is shown in figure [ fig : lokalizace - skalovane - varnorm ] . by comparison to figure [ fig : lokalizace - skalovane - anom ] we can see that the strongest contributors to apparent nonlinear dependences have been mitigated by this data cleaning step
. the maxima of the relative nonlinear contributions are now much lower and are located almost purely in the equatorial ocean regions .
as previously , we investigate the form of a nonlinear dependence pattern related to the strongest source of nonlinearity in thus preprocessed data .
this is shown in figure [ fig : figure_scatter_varnorm ] , showing typical example bivariate distributions and time series .
the source of the observed non - gaussianity of bivariate dependence can be commonly tracked down to an apparent non - stationarity of the time series .
in particular in the example shown in figure [ fig : figure_scatter_varnorm ] there is a strong ( almost linear ) trend , which might be of interest for other reasons , but can be considered as spurious with respect to detection of climate interactions on the time scale considered .
this motivates additional detrending of the time series followed by yet another replication of the analysis .
the results suggest that the trends are indeed responsible for a major part of the yet remaining apparent non - linearity .
in particular , the amount of detected pairs of locations with statistically significant nonlinear contribution to temperature dependence goes further down to @xmath51 at the @xmath49 level . similarly , the average nonlinear contribution to mutual information is further substantially reduced , see figure [ fig : lokalizace - skalovane - varnorm - detrend ] .
the analysis was replicated for the era dataset .
the general strength and distribution of nonlinear dependence within the era dataset is very similar to the ncep / ncar , see figures [ fig : figuremivsmi_era ] , [ fig : lokalizace - skalovane - anom_era ] , [ fig : lokalizace - skalovane - varnorm_era ] and [ fig : lokalizace - skalovane - varnorm - detrend_era ] .
the fraction of pairs of nodes with significant nongaussianity were at the @xmath49 level is @xmath52 , @xmath53 and @xmath54 percent in the original anomalies , variance normalized data and detrended variance normalized data respectively , suggesting just a slightly weaker contribution of the nonlinearity in the era dataset , and clearly not much more than expected by chance for linear data in the last case .
in the previous sections we have outlined an approach for a detailed multi - step analysis of the relevance of nonlinearity in the dependence structure of climate time series and the results for monthly sat reanalysis data .
the overall picture is in general that of negligible nonlinearities in the data ; the most substantial apparent ` nonlinearities ' are attributable to nonstationarity effects .
therefore , the tentative suggestion with respect to choice of dependence measure for this type of data would be to use a linear measure ( pearson correlation coefficient ) , potentially after removing data nonstationarity by preprocessing , as some of the nonstationarities may affect also pearson correlation estimates , albeit differently than the nonlinear mutual information .
an obvious question is that of generalizability of the specific findings .
explicitly , we have confirmed similar results in the era reanalysis sat data .
also in daily ncep / ncar data we have observed relatively negligible nonlinearity , after removing nonstationarities in variance as well as trends ( results not shown ) . while we have only showed results for the most commonly used grid with fixed angular resolution @xmath55 , based on inspection of the spatiotemporal structure of the data we conjecture that analogous results would be observed for other resolutions as well as area - corrected grids as used e.g. by @xcite .
the present results extend those of @xcite who tested for possible nonlinearity in the dynamics of the station ( prague - klementinum ) sat time series and found that the dependence between the sat time series @xmath56 and its lagged twin @xmath57 was well - explained by a linear stochastic process .
this result about a linear character of the temporal evolution of sat time series , as well as the results of this study about relations between the reanalysis sat time series from different grid - points can not be understood as arguments for a linear character of atmospheric dynamics .
these results rather characterize properties of measurement or reanalysis data at a particularly coarse level of resolution when the data reflecting a spatially and temporally averaged mixture of dynamical processes on a wide range of spatial and temporal scales are considered . for instance , a closer look on the dynamics on specific temporal scales in temperature and other meteorological data has led to identification of oscillatory phenomena with nonlinear behavior , exhibiting phase synchronization @xcite . also for other variables with vastly different dynamics
we can expect substantial nonlinearity in bivariate dependencies , especially if the measurement / model has sufficient spatial and temporal resolution .
conceptually , similar analysis to that presented is warranted to be carried out before the decision for use of linear or nonlinear dependence measure for each substantially different dataset .
the work on preparation of a semi - automated tool for such analysis is undergoing . despite the relative sparsity of the substantially nonlinear dependence patterns , even after the two additional preprocessing steps we have observed more than the expected proportion of significantly nonlinear dependences ( @xmath58 instead of expected @xmath59 ) . given the number of tests carried out ( several million of location - pairs ) , this small deviation likely consists a globally significant deviation from multivariate gaussianity , although the intricate interdependence of the pair tests themselves makes the estimation of the exact p - value for a global linearity hypothesis technically very difficult .
from practical point of view , it may be argued that the statistical determination of the above - random presence of apparent nonlinearity should not play a key role in method choice .
firstly , the more detailed quantitative analysis has already shown that even where present , the effect is relatively weak . secondly , even though for some region pairs there may be a statistical indication of nonlinear dependence , it is realistic to suspect ( given the results for the raw data , see figures [ fig : figure_scatter ] and [ fig : figure_scatter_varnorm ] ) that this apparent nonlinearity may be due to as yet not discussed type of nonstationarity , and visual inspection of the data is required .
indeed , several further spurious nonlinearities have been detected due to various uncorrected problems within the reanalysis data ( some of which corresponded to known problems as described in @xcite ) .
this leads to the consideration of the strongest nonlinear pattern observed in the current data even after the two additional preprocessing steps on top of those carried out in @xcite .
as shown in figure [ fig : lokalizace - skalovane - varnorm - detrend ] , ocean areas particularly in the tropical pacific bear still a slightly elevated non - gaussian contribution to dependence patterns .
an example of such dependence pattern and related time series is shown in figure [ fig : nonlinearremainder ] .
note that the area of strongest residual non - gaussianity roughly corresponds to regions implicated in enso dynamics ; for visual comparison we plot the area and time series for the el nino 3.4 index ( the nino 3.4 region is bounded by @xmath60w-@xmath61w and @xmath62s-@xmath62n @xcite ) .
the nino3.4 sst index data were downloaded from from noaa / nws climate prediction center , ( http://www.cpc.ncep.noaa.gov , accessed november 15th , 2012 ) .
we hypothesize that the observed pattern reflects the nonlinear character of enso dynamics observed e.g. by @xcite .
quantification of dependences between variables is a common task within the study of climate or other complex systems .
the choice of dependence measures is commonly based on some theoretical assumptions about the underlying system . in case of climatic time
series , this may lead to the choice of nonlinear methods due to the nonlinear nature of the underlying physical processes .
however , for various reasons including spatiotemporal sampling or averaging , measured or modeled data might be in fact well captured by linear measures , with their nonlinear counterparts potentially reducing sensitivity or introducing bias . we have presented a multi - step approach that allows the detailed assessment of the nonlinear contribution to dependence patterns in a dataset , including not only statistical testing , but also quantification , localization and analysis of sources of this contribution .
this approach can provide rationale for the decision regarding the choice of suitable dependence measure for given type of data , as well as direct the analyst attention to hidden crucial properties of the dataset .
importantly , the presented approach is quite general .
it is transferable to other geoscientific datasets , as well as to other disciplines such as neuroscience , see e.g. @xcite for an earlier application of a related approach and @xcite for an example focused to an assessment of the nonlinearity effects on graph - theoretical network characteristics .
let us note that the construction and interpretation of graphs representing climate networks poses further challenges @xcite .
for monthly sat in the ncep / ncar dataset and similar data , the analysis suggests that the use of linear dependence methods is generally sufficient , potentially after the treatment of described nonstationarity sources .
this has shown that the quantitative study of amount of nonlinearity in the data provides as a side product valuable hints about potentially hidden specific properties of the data , that may otherwise go unnoticed and bias the results , were just a single method used without detailed analysis .
this study is supported by the czech science foundation , project no . p103/11/j068 .
dee dp , uppala sm , simmons aj , berrisford p , poli p , kobayashi s , andrae u , balmaseda ma , balsamo g , bauer p , bechtold p , beljaars acm , van de berg l , bidlot j , bormann n , delsol c , dragani r , fuentes m , geer aj , haimberger l , healy sb , hersbach h , holm ev , isaksen l , kallberg p , koehler m , matricardi m , mcnally ap , monge - sanz bm , morcrette jj , park bk , peubey c , de rosnay p , tavolato c , thepaut jn , vitart f ( 2011 ) the era - interim reanalysis : configuration and performance of the data assimilation system
. quarterly journal of the royal meteorological society 137(656 , part a):553597 donges jf , schultz hch , marwan n , zou y , kurths j ( 2011 ) investigating the topology of interacting networks theory and application to coupled climate subnetworks .
european physical journal b 84(4):635651 hartman d , hlinka j , palu m , mantini d , corbetta m ( 2011 ) the role of nonlinearity in computing graph - theoretical properties of resting - state functional magnetic resonance imaging brain networks .
chaos 21(1 ) kalnay e , kanamitsu m , kistler r , collins w , deaven d , gandin l , iredell m , saha s , white g , woollen j , zhu y , chelliah m , ebisuzaki w , higgins w , janowiak j , mo k , ropelewski c , wang j , leetmaa a , reynolds r , jenne r , joseph d ( 1996 ) the ncep / ncar 40-year reanalysis project . bulletin of the american meteorological society 77(3):437471 kistler r , kalnay e , collins w , saha s , white g , woollen j , chelliah m , ebisuzaki w , kanamitsu m , kousky v , van den dool h , jenne r , fiorino m ( 2001 ) the ncep - ncar 50-year reanalysis : monthly means cd - rom and documentation .
bulletin of the american meteorological society 82(2):247267 palus m , novotna d ( 2004 ) enhanced monte carlo singular system analysis and detection of period 7.8 years oscillatory modes in the monthly nao index and temperature records .
nonlinear processes in geophysics 11(5 - 6):721729 palus m , novotna d ( 2011 ) northern hemisphere patterns of phase coherence between solar / geomagnetic activity and ncep / ncar and era40 near - surface air temperature in period 7 - 8 years oscillatory modes .
nonlinear processes in geophysics 18(2):251260 uppala s , kallberg p , simmons a , andrae u , bechtold v , fiorino m , gibson j , haseler j , hernandez a , kelly g , li x , onogi k , saarinen s , sokka n , allan r , andersson e , arpe k , balmaseda m , beljaars a , van de berg l , bidlot j , bormann n , caires s , chevallier f , dethof a , dragosavac m , fisher m , fuentes m , hagemann s , holm e , hoskins b , isaksen l , janssen p , jenne r , mcnally a , mahfouf j , morcrette j , rayner n , saunders r , simon p , sterl a , trenberth k , untch a , vasiljevic d , viterbo p , woollen j ( 2005 ) the era-40 re - analysis
. quarterly journal of the royal meteorological society 131(612 , part b):29613012 | quantification of relations between measured variables of interest by statistical measures of dependence is a common step in analysis of climate data .
the term `` connectivity '' is used in the network context including the study of complex coupled dynamical systems .
the choice of dependence measure is key for the results of the subsequent analysis and interpretation .
the use of linear pearson s correlation coefficient is widespread and convenient . on the other side , as the climate is widely acknowledged to be a nonlinear system , nonlinear connectivity quantification methods , such as those based on information - theoretical concepts , are increasingly used for this purpose . in this paper
we outline an approach that enables well informed choice of connectivity method for a given type of data , improving the subsequent interpretation of the results .
the presented multi - step approach includes statistical testing , quantification of the specific non - linear contribution to the interaction information , localization of nodes with strongest nonlinear contribution and assessment of the role of specific temporal patterns , including signal nonstationarities . in detail
we study the consequences of the choice of a general nonlinear connectivity measure , namely mutual information , focusing on its relevance and potential alterations in the discovered dependence structure .
we document the method by applying it on monthly mean temperature data from the ncep / ncar reanalysis dataset as well as the era dataset .
we have been able to identify main sources of observed non - linearity in inter - node couplings .
detailed analysis suggested an important role of several sources of nonstationarity within the climate data .
the quantitative role of genuine nonlinear coupling at this scale has proven to be almost negligible , providing quantitative support for the use of linear methods for this type of data . |
observations of edge - on , star - forming galaxies indicate that many ( possibly most ) such systems display extra - planar emission ( ` halos ' ) . where sufficient observations exist ,
these halos are known to be both multi - phase , displaying all of the components of the in - disk interstellar medium ( ism ) , as well as highly structured , revealing vertical features over a variety of spatial scales ( e.g. * ? ? ?
mass flux estimates thus far @xcite imply that disk - halo outflow is responsible for transporting large quantities of gas and , as a result , has an important role to play in the evolution of galaxies , from metallicity gradients in disks to altered star formation rates ( sfrs ) .
spatially resolved observations of nearby edge - on galaxies provide us with important data for studying the details of such disk - halo outflows .
early observations of edge - on galaxies showed that radio continuum emission can extend several kpc above the plane @xcite , and this emission is dominated by non - thermal synchrotron radiation ( e.g. * ? ? ?
* ; * ? ? ?
* ) , indicative of magnetic fields and cosmic rays ( crs ) .
galactic winds may also be present , since crs can drive outflows , as first pointed out by @xcite ; such outflows or ` feedback ' are crucial to galaxy formation scenarios .
dust in galaxy halos was first observed in the form of vertical filaments in absorption against starlight @xcite and , later in emission from high resolution space - based infra - red ( ir ) observations ( e.g. * ? ? ?
* ) . extracting the physical conditions of dust , which is present over a range of sizes , compositions and excitation conditions , requires careful modelling of the spectral energy distribution ( sed ) and therefore observations at a variety of infrared wavelengths .
although such modelling is routinely carried out for galaxy disks , it has proved more difficult for the weaker halo emission [ but see @xcite ] . with fewer data points , however , it is still possible to constrain some dust properties ( e.g. * ? ? ?
* ; * ? ? ?
* ) , especially in the ` classical ' regime in which grains reach equilibrium temperatures . in this paper , we present the results of observations at 617 mhz as well as at two sub - mm wavelengths , @xmath10 @xmath11 m and @xmath12 @xmath11 m , of two edge - on galaxies
ngc 3044 and ngc 4157 in order to study both the in - disk emission and any halo emission which might be present .
since synchrotron emission is stronger at lower frequencies , halos should be more easily detected than at high radio frequencies for comparable signal - to - noise ( s / n ) .
a clear example can be found in @xcite . as for dust ,
the sub - mm wavelengths are known to be probes of cold dust and are less likely to be contaminated by contributions from hot or warm components and/or stochastically heated very small grains .
such data also provide the best estimate of total dust mass .
adopting these wavebands also allows us to probe the far infrared ( fir ) - radio continuum relation ( see next subsection ) in these galaxies . in the next two subsections
, we discuss the fir - radio continuum relation and introduce the two galaxies . in sect .
[ sec : observations ] , we provide details of the observations and data reduction .
[ sec : results ] presents the results , including high latitude emission , where present , temperature fitting , mass estimates , and sub - mm - radio correlations . sects .
[ sec : discussion ] and [ sec : conclusions ] present the discussion and conclusions , respectively .
the well - known relationship between thermal fir emission from dust and the predominantly non - thermal synchrotron emission in galaxies @xcite is believed to result from the dependence of both on massive star formation @xcite .
the same stars that provide ultraviolet ( uv ) photons to heat the dust are also destined to become supernovae , thus producing non - thermal radio emission . what is still a puzzle
, however , is how the relation can be so tight over many orders of magnitude among galaxies that have a range of properties , for example , differing ism masses , dust masses , metallicities , sfrs and possible differences as to how optically thick or optically thin galaxies are to uv photons and cr electrons .
similar arguments apply to the correlation which has also been found _ within _ galaxies . as a result , other possibilities have been proposed for the origin of the relation , for example coupling between the magnetic field strength and the gas density ; since it is well known that gas and dust density vary together , one would expect higher synchrotron emission wherever the dust emission is higher @xcite .
one observational difficulty has been the traditional choice of the 1.4 ghz radio frequency ( which includes a small but non - negligible thermal contribution ) and fir emission at @xmath13 and @xmath14 @xmath11 m ( which includes contributions from both warm and cold dust ) .
an early attempt at separating the various spectral contributions within m 31 @xcite suggested that a good relation still exists between non - thermal radio emission and _ cold _ dust , the former driven by supernovae and the latter heated by the general interstellar radiation field ( isrf ) rather than solely by hot young stars , i.e. the direct link with star - forming regions is less clear .
@xcite later found a slightly non - linear relation between the radio continuum and cold dust ( @xmath15 ) and suggest that there may be a non - linear dependence of radio emission on sfr .
see also @xcite and @xcite for further discussion .
a better approach is to observe the radio emission at a frequency that is strongly dominated by the non - thermal component , so that assumptions and corrections for the thermal contribution are not required .
this is especially important when searching for correlations within a given galaxy since the thermal contribution can vary strongly from place to place .
we have taken this approach by choosing the low radio frequency of 617 mhz .
similarly , by searching for correlations with @xmath12 @xmath11 m emission , the isrf - heated cold dust component is more likely to be isolated .
such an approach has been taken by @xcite who find a good correlation between these two wavebands within the galaxy , ngc 5775 .
an important refinement is to consider whether a smoothed version of the sub - mm emission improves upon the relation , should one be found .
for example , @xcite proposed that , since the mean free path of a uv photon to dust absorption is much less than the diffusion length of a typical cr electron , we would expect that the fir - radio continuum relation ( or , in our case , the 617 mhz - 850 @xmath11 m relation ) will be improved if the fir emission is smoothed spatially .
such an improvement has been observed by murphy et al .
( 2006a , 2006b , 2008 , 2009 ) .
we will examine this refinement for ngc 3044 and ngc 4157 as well .
optical images of ngc 3044 and ngc 4157 are shown in fig .
[ fig : colourfig ] and their basic parameters are listed in table [ tab : galaxy_parameters ] . aside from the fact that ngc 4157 resides in a richer environment ( see below ) , the two galaxies have similar global properties : to within a factor of about 1.5 , the galaxies are of similar size , sfr , total mass , and gas content .
one difference is in the colour of the emission as seen in fig .
[ fig : colourfig ] and indicated by the ( b - v)@xmath16 colour of table [ tab : galaxy_parameters ] .
ngc 3044 shows predominantly blue emission throughout its projected disk , whereas ngc 4157 shows blue emission in the outer regions .
there is a small difference in inclination between the two galaxies which may contribute to this colour difference though it is nt clear to what extent .
for example , the slightly lower inclination of ngc 4157 appears to allow the redder inner regions of the galaxy to be more readily revealed .
we describe these galaxies in detail , below .
ngc 3044 ( fig .
[ fig : colourfig ] left ) is an isolated galaxy with no known companions . although @xcite place it within the leo cloud , the other 3 members are widely displaced in the sky ( @xmath17 or @xmath18 mpc ) .
some minor warping can be seen in the optical image at large galactocentric radii , however , and asymmetries between the north - west and south - east major axis emission have been noted by @xcite and @xcite .
ngc 3044 has been considered both a ` starburst ' and ` normal ' galaxy ; @xcite discuss this issue , indicating that it may be considered a starburst galaxy , but one in which the starburst is extended over most of the disk , rather than concentrated near the nucleus .
the presence of a galactic halo in ngc 3044 is well known , having first been discovered in the radio continuum by @xcite and further observed by various authors ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
extra - planar diffuse ionized gas ( edig ) has been detected by @xcite , @xcite and @xcite and extends as far as 5 kpc from the plane @xcite .
halo structure , including a kpc - scale diameter h@xmath19 arc on the south side of the disk is also observed @xcite .
ionization of the edig can be explained by a combination of photo - ionization and shock ionization @xcite .
xmm - newton observations by @xcite have detected an extended soft x - ray halo which reaches its maximum height over the inner disk ( @xmath20 kpc ) .
the x - ray emission correlates with both the edig and radio continuum .
halo hi emission is also observed , including hi extensions and several loops that have the signature of expanding supershells @xcite .
there appears to be some spatial correlation between the hi and radio continuum features in the inner disk , but the hi is more extended radially and shows vertical extensions that do not have counterparts at other wavebands .
infrared spectroscopy of the halo has also been recently obtained by @xcite .
a supernova was observed in ngc 3044 in the year , 1983 ] .
ngc 4157 ( fig .
[ fig : colourfig ] right ) is a member of the ursa major cloud @xcite and belongs to the galaxy group , lgg-258 @xcite .
previous radio continuum observations have been made by @xcite , @xcite , @xcite , @xcite and @xcite .
hi data can be found in @xcite , in @xcite , in the westerbork hi survey of spiral and irregular galaxies ( whisp ) catalog ( see e.g. * ? ? ?
* ) , and in @xcite .
a companion galaxy , ugc 7176 , with a dynamical mass @xmath21 1% of that of ngc 4157 , is located 12 arcmin to the south @xcite .
co parameters can be found in @xcite and @xcite . the galaxy was observed with the infrared space observatory ( iso ) at @xmath22 m @xcite and a uv image from the galaxy evolution explorer ( galex ) satellite can be found in @xcite .
a radio continuum halo was discovered in ngc 4157 by irwin et al .
( 1999 ) but , to our knowledge , the only previous search for extraplanar features in this galaxy was made by @xcite who searched for extraplanar dust absorption , with ambiguous results .
three historical supernovae have been detected in ngc 4157 in the years , 1937 , 1955 and 2003 ( see footnote [ footnote : iau ] ) . [ cols="<,^,^",options="header " , ] as indicated in sect .
[ sec : fir - rad ] , the fir - radio continuum relation has been shown to improve when the fir emission is smoothed .
this process effectively ` forces ' the uv photon dust optical depth to more closely match the synchrotron diffusion length .
following @xcite , we apply a sequence of smoothing kernels to the same @xmath1 m maps used above with consecutively increasing ` smearing scale - lengths ' . a @xmath1m/617 mhz ratio map is then formed , normalized , and the residuals compared to see which , if any , improve upon the relations plotted in figs .
[ fig : correlations]a and c. we tried both exponential smearing kernels ( @xmath23 ) where @xmath24 is the distance from each point and @xmath25 is the exponential scale length ( both in the plane of the sky ) as well as gaussian smearing kernels ( @xmath26 ) , where @xmath27 is the standard deviation of the gaussian . in both cases , a sequence of scales from 1 to 25 arcsec in 1 arcsec steps was tried .
like @xcite , we find that the residuals first decrease and then increase again as kernel size increases , resulting in a clear scale for which the fit is best .
the best fit results are given in table [ tab : fir_rad ] and are plotted in figs .
[ fig : correlations]b and d. there is no doubt that the fits are improved after applying this smoothing and , moreover , that the correlation becomes non - linear ( @[email protected] for ngc 3044 and [email protected] for ngc 4157 ) . for these cases with smoothing kernels ,
we also provide uncertainties on the other fitted parameters , as described above .
note that the intercept for ngc 3044 ( table [ tab : fir_rad ] ) is consistent with zero .
this is not the case for ngc 4157 although , for both galaxies , the forward and reverse bisectors encompass the ( 0,0 ) point .
since the intercept is zero for ngc 3044 , we have also provided a logarithmic plot for this galaxy which is shown in fig .
[ fig : logplot ] .
the best fit linear slope in this space with the displayed sampling ( 1.9 @xmath30 0.1 ) agrees with the non - linear fit of table [ tab : fir_rad ] .
m 617 mhz correlation for ngc 3044 .
the data from fig .
[ fig : correlations]b have been replotted here without the error bars .
the solid curve is the bisector from that figure and the dashed curve is the best linear fit in logarithmic space whose slope is 1.9 @xmath30 0.1 . , width=292 ] we find that the gaussian smoothing kernel produces a better fit than an exponential kernel .
it has been suggested that an exponential kernel is more likely to describe a situation in which cosmic ray electrons escape from the disk whereas a gaussian kernel would describe a random walk @xcite . for an edge - on galaxy
, it is not straightforward to interpret the resulting best fit scale ; however , since we are looking through a long in - disk line of sight in our two galaxies , we would expect that the in - disk ( random walk ) component should dominate , consistent with our better gaussian fit . for more face - on galaxies
, @xcite found that exponentials provided better fits and @xcite , with earlier data , found that both gaussians and exponentials were equally acceptable .
the non - linear relationship between the 617 mhz and @xmath12 @xmath11 m emission explains the fact that the major axis slices for these two bands , although roughly correlated , do not follow each other particularly well in fig .
[ fig : majoraxis ] ( red and green curves ) .
however , when the functional relationship shown in fig .
[ fig : correlations]b is applied to the @xmath1 m emission and renormalized , the widths of the two distributions ( not shown ) match each other to well within a beam size .
( note that squaring the @xmath1 m emission will _ narrow _ its distribution , bringing it in alignment with the 617 mhz emission after normalization . )
these results will be discussed further in sect .
[ sec : discussion ] .
the presence of a multi - phase gaseous halo in ngc 3044 has been known for some time @xcite .
see @xcite , for example , for measurements of the hi and edig scale heights in this galaxy .
our new observations are the first at 617 mhz and @xmath1 m and reveal several new high latitude features in these two wavebands ( fig .
[ n3044resultsfig ] and sect .
[ sec : radcont_images ] ) .
these disturbances are better seen in several selected overlays shown in fig .
[ fig : n3044_overlays ] ; the greyscales show h@xmath19 emission , hi total intensity emission , and smoothed h@xmath19 emission in frames a , b , and c , respectively . at a location of about 1.5 arcmin from the nucleus along the north - west major axis
there is a gap in the h@xmath19 emission in the disk ( fig .
[ fig : n3044_overlays]a ) . at this position is an hi supershell , part of which is visible as the high latitude feature , f10 ( fig .
[ fig : n3044_overlays]b ) , so labelled by @xcite .
the narrow vertical dust feature observed at @xmath1 m , although at a low s / n , occurs adjacent to f10 and the double - pronged 617 mhz feature is also at this position .
apparently , there is disturbance in the disk at this location which is affecting all of the components displayed and that disturbance has excavated an ionized gas cavity in the disk .
this is not the first time that high latitude features have been associated with a gap or absence of h@xmath19 emission in the underlying disk ; see @xcite for another prominent example in ngc 5775 .
m contours as in fig .
[ n3044resultsfig]d , over a greyscale h@xmath19 image from @xcite .
the h@xmath19 ` gap ' in the major axis that is discussed in sect .
[ sec : n3044-discuss ] is labelled . *
( b ) * @xmath1 m contours as in fig .
[ n3044resultsfig]d , over a greyscale hi total intensity map from @xcite , where f10 denotes an hi extension identified as part of an expanding shell in the latter reference . * ( c ) * 617 mhz contours as in fig .
[ n3044resultsfig]b over a greyscale h@xmath19 image from @xcite smoothed to 5 arcsec resolution .
, title="fig:",width=307 ] -0.3truein m contours as in fig .
[ n3044resultsfig]d , over a greyscale h@xmath19 image from @xcite .
the h@xmath19 ` gap ' in the major axis that is discussed in sect .
[ sec : n3044-discuss ] is labelled . *
( b ) * @xmath1 m contours as in fig .
[ n3044resultsfig]d , over a greyscale hi total intensity map from @xcite , where f10 denotes an hi extension identified as part of an expanding shell in the latter reference . * ( c ) * 617 mhz contours as in fig .
[ n3044resultsfig]b over a greyscale h@xmath19 image from @xcite smoothed to 5 arcsec resolution .
, title="fig:",width=307 ] -0.3truein m contours as in fig .
[ n3044resultsfig]d , over a greyscale h@xmath19 image from @xcite . the h@xmath19
` gap ' in the major axis that is discussed in sect .
[ sec : n3044-discuss ] is labelled .
* ( b ) * @xmath1 m contours as in fig .
[ n3044resultsfig]d , over a greyscale hi total intensity map from @xcite , where f10 denotes an hi extension identified as part of an expanding shell in the latter reference . * ( c ) * 617 mhz contours as in fig .
[ n3044resultsfig]b over a greyscale h@xmath19 image from @xcite smoothed to 5 arcsec resolution . , title="fig:",width=307 ] on the whole , there is global correspondence between the radio continuum and h@xmath19 emission as seen in fig .
[ fig : n3044_overlays]c ) .
this correspondence , which has been pointed out by many authors , including @xcite for ngc 3044 itself , can be understood from the fact that both components are associated with massive star formation . what is more unusual , as pointed out in sect .
[ sec : radcont_images ] , are the extraplanar features which appear to originate at the two ends of the major axis . in particular , at the south - east end , the emission is truncated approximately where the h@xmath19 emission also narrows abruptly ( fig . [
fig : n3044_overlays]c ) .
we then see a series of disconnected emission features in _ both _ the gmrt 617 mhz and vla 20 cm images ( figs .
[ n3044resultsfig]b and [ n3044vlafig]b , respectively ) , including the feature at ra @xmath31 9@xmath32 53@xmath33 51@xmath34 , dec @xmath31 01@xmath35 35@xmath36 45@xmath37 which we have labelled the non - thermal cloud ( ntc ) in fig .
[ n3044resultsfig]b .
the feature centers at approximately 2 arcmin ( 13 kpc ) from the galaxy s major axis . in fig .
[ fig : n3044_wise ] we show the 617 mhz ( red contours ) and @xmath2 cm ( blue dashed contours ) emission of ngc 3044 superimposed on a greyscale image of the @xmath38 @xmath11 m emission taken from the wide - field infrared survey explorer ( wise ) all sky survey @xcite , the latter enhanced to show low intensities that are @xmath39 2@xmath40 in brightness .
the @xmath38 @xmath11 m emission is an indicator of the presence of warm dust ( of order 130 k for classical grains in thermal equilibrium ) .
this map shows that the halo of ngc 3044 is even more extensive than previously known and shows considerable substructure , including vertical extensions and arcs as well as ` disconnected ' features towards the north .
it is beyond the scope of this paper to discuss the wise results in detail ; however , when considered together with the two independent radio continuum maps ( gmrt and vla ) , the results give credence to the reality of the northwards extensions that begin on the south - east end of the major axis .
note that , at low s / n , sensitivity to spatial scales ( which depends on an array s uv coverage ) is as important as s / n in determining what emission features will be detected .
consequently , the gmrt and vla emission , while overlapping , do not align perfectly and should not be used to determine spectral indices at these low emission levels . put together ,
though , the truncation of the radio continuum major axis where the northwards extension begins , the abrupt narrowing of the h@xmath19 emission at this position , and emission extending towards the north that is visible in gmrt 617 mhz , the vla @xmath2 cm map , and the wise @xmath38 @xmath11 m map , suggest that a disruption has occurred at the south - east end of the major axis in this galaxy and expelled material to the north of the plane .
b , and the @xmath2 cm emission ( blue dashed contours ) from fig .
[ n4157vlafig]b , are shown overlaid on a greyscale of the the wise @xmath38 @xmath11 m emission .
the wise image has been cut off at the 2@xmath40 level and is shown so as to enhance the faint broad - scale emission .
, width=313 ] bearing in mind the low s / n of the ntc , if we nevertheless take it at face value , we can estimate some physical parameters . excluding the smaller features closer to the major axis , the ntc , as shown in fig .
[ n3044resultsfig]b , has a flux density of 3 mjy , or a spectral power of @xmath41 w / hz . by comparison , the flux density of the galactic supernova remnant , cas a , is @xmath42 jy for epoch 2005.5 @xcite , corresponding to @xmath43 w / hz at its distance of 3.4 kpc @xcite .
thus , the radio power of the ntc alone corresponds to 36 equivalent supernova remnants .
if we include features closer to the disk , the value approximately doubles .
the environment of the ntc differs substantially from that of cas a ; nevertheless this comparison suggests that a massive star - forming region in the disk could indeed have been responsible for the observed emission from the point of view of energetics .
regarding timescales , from the mean magnetic field strength ( @xmath44 @xmath11 g ) found throughout the disk and the observed halo of ngc 3044 @xcite , the mean cosmic ray electron lifetime is @xmath45 yr ( see sect .
[ sec : correlation - discuss ] ) .
however , the magnetic field was found to vary from 3.5 to 8.9 @xmath11 g and it is likely that crs far from the disk would be associated with magnetic fields that are at the low end of this range .
consequently , a lifetime of @xmath46 yr ( @xmath47 ) may be more appropriate . for a distance of 13 kpc , then , the outflow velocity would be 91 km s@xmath48 which is much less than the 300 km s@xmath48 outflow velocity observed in ngc 253 by @xcite , for comparison .
the above calculations are order of magnitude only and are meant to illustrate the feasibility of non - thermal emission far from the disk of ngc 3044 having originated from activity within the disk .
we require more information on the magnetic field distribution and strength to make more definitive statements .
as noted earlier , the 617 mhz emission in ngc 3044 shows curvature towards the north on both ends of the major axis of ngc 3044 which sometimes suggests that ram pressure stripping could be occurring as the galaxy passes through an intergalactic medium ( igm ) in a southerly direction .
however , there is no such evidence in the hi emission and , in fact , hi extensions are observed on both the north and south sides of the major axis .
in addition , the isovelocity hi contours show a slight curvature which , if caused by motion through an igm , would imply a motion to the north - east rather than the south - west ( see fig . 4 of * ? ? ? * ) .
this suggests that the observed 617 mhz extensions at the ends of the major axis and the disconnected northern features may have been produced from internal activity which is asymmetrically placed with respect to gas distribution ( or magnetic field distribution ) in the galaxy .
@xcite also argue for an internal origin for the hi supershells and the energetics ( above ) are consistent with this .
with distributed star formation in the galaxy , it is reasonable to expect that outflows will more effectively eject material far from the plane at large galactocentric radii where the internal disk density is lower .
ngc 4157 is also known to exhibit high latitude emission @xcite but fewer observations of this galaxy have been carried out in comparison to ngc 3044 . as seen in figs .
[ n4157resultsfig]c and d , we do not have sufficient s / n for a clear detection of high latitude dust in ngc 4157 .
the radio continuum emission ( figs .
[ n4157resultsfig]a and b ) shows some extensions that are roughly at the locations of those seen in the vla 20 cm map ( fig .
[ n4157vlafig ] ) .
a new feature is seen in the 617 mhz map ( fig .
[ n4157resultsfig]b ) in the form of a radio continuum extension beginning at the far north - eastern end of the major axis and extending south , as described in sect .
[ sec : n4157_radio ] .
this southern extension can be better seen in fig .
[ fig : n4157_overlay ] where it is overlaid onto a greyscale image of the @xmath38 @xmath11 m wise emission .
this plot shows that warm dust exists in a halo that extends up to 1.3 arcmin ( 4.9 kpc ) from the plane .
the hi in this galaxy extends to roughly the same _ z _ distance @xcite ( although hi also extends much farther in the radial direction along the major axis ) .
the dust , therefore , may be a component of the high - latitude hi in the same way that it normally is within a galaxy disk .
for the purposes of our 617 mhz comparison , we note that the southern 617 mhz extension coincides with discrete dust emission in roughly the same direction which gives weight to our assertion that the extension is real .
indeed , some of the 617 mhz features on the north side of the disk also have dust counterparts .
although it is nt clear whether these features occupy the same volume , the evidence continues to mount that all constituents of the ism are represented in galaxy halos .
see @xcite for an earlier example .
b over a greyscale wise image at @xmath49 m .
the wise image ( spatial resolution of 12 arcsec ) has been cut off at 2@xmath40 and is shown so as to enhance the faint broad - scale emission .
, width=302 ]
as can be seen from fig .
[ spectra_fig ] , the @xmath1 m emission in ngc 4157 ( as modeled ) still includes a contribution from the warm dust component whereas the emission of ngc 3044 is strongly dominated by only cold dust at that wavelength .
in addition , the s / n of the ngc 3044 data is higher than that of ngc 4157 .
we therefore consider only ngc 3044 in our discussion of the 617 mhz - @xmath1 m relation which we take to represent a relationship between synchrotron emission and cold dust ( t@xmath50 k , table [ properties_table ] ) .
firstly , we note that the relationship between these two components is significantly improved when a smearing kernel is introduced to the sub - mm data ( fig .
[ fig : correlations ] ) , consistent with earlier results @xcite .
this kernel ` corrects ' for differences between the photon mean free path to dust absorption and the cr electron diffusion length .
we find , not only that the correlation improves , but also that it becomes clearly non - linear such that @xmath51 ; that is , the relationship is roughly consistent with a luminosity relation of @xmath52 . by comparison , @xcite found @xmath53 and @xcite found a roughly linear 617 mhz - @xmath1 m relation for ngc 5775 .
neither of these authors applied a smearing kernal and , as fig .
[ fig : correlations ] illustrates , the correlation becomes much flatter ( @xmath54 , table [ properties_table ] ) , when a smoothing kernel is ignored .
@xcite have shown that galaxies with higher sfrs per unit area have lower kernel scale lengths .
they interpret this result to imply that galaxies with high sfrs contain , on average , younger sf regions from which cr electrons have not had sufficient time to diffuse to large scales . when sf has not been so recent , by contrast , the difference between the radio continuum scale and the dust scale diminishes . as indicated in sect .
[ sec : correlations ] , it is not straightforward to interpret the meaning of the kernel size for an edge - on galaxy in which the line of sight that is being probed varies with position and may probe a range of sf regions of varying ages .
in addition , our best result corresponds to a gaussian kernel which can not be directly compared to the exponentials used by @xcite in more face - on systems .
nevertheless , we can consider some representative timescales to see whether our kernel size could reasonably represent a link between the cr electron and dust heating scales .
for example , based on spatially resolved images at @xmath2 cm and @xmath55 cm , the minimum energy magnetic field strength in ngc 3044 has a mean value of @xmath56 g , ranging from 3.5 to 8.9 @xmath11 g to within uncertainties of approximately a factor of 2 @xcite .
the lifetime for cosmic ray electrons against synchrotron and inverse compton losses , for an isotropic velocity distribution of cr electrons , is given by ( e.g. * ? ? ?
* ) , @xmath57 where @xmath58 is the magnetic field strength and @xmath59 and @xmath60 are the magnetic field and radiation energy densities , respectively .
taking @xmath61 and @xmath62 , then @xmath63 and we find @xmath64 yr ( ranging from to 3.3 to 14@xmath65 yr ) .
this implies that the 617 mhz emission is measuring a massive sf history of ngc 3044 that is integrated over @xmath7 yr ; it is not measuring the recent sf activity in the galaxy . in this time , how far can cr electrons diffuse ?
the diffusion timescale can be written @xcite , @xmath66 where @xmath67 represents the distance that a cr electron can diffuse from its point of origin via a random walk , assuming an energy - dependent diffusion coefficient . setting @xmath68 yields @xmath69 4.3 kpc .
this is a significant distance and sufficient to allow cr electrons to ` fill ' the galaxy disk , assuming that sf has been widely distributed over the past @xmath7 yr .
if the cold dust is heated by the isrf which is similarly widely distributed throughout the disk then , at least to order of magnitude , our smoothing kernel should represent roughly the difference between the cr electron diffusion length and the dust mean free path to isrf photons .
the latter is not known for ngc 3044 ( and likely varies ) , but in our own galaxy , a typical dust cross - section to absorption and scattering in the visible through uv part of the spectrum for big grains , @xmath70 , ranges from 0.2 to 1 @xmath71 @xmath72 per hydrogen atom @xcite . for
this range of dust cross - section and using the mean ism gas density in ngc 3044 of @xmath73 @xmath74 @xcite , the mean free path of isrf photons to dust is @xmath75 kpc .
our gaussian smoothing kernel of @xmath76 kpc ( table [ tab : fir_rad ] ) corresponds to a radius ( half - width at half maximum ) of @xmath771.0 kpc .
if we convolve this size scale with @xmath78 , the result is @xmath79 kpc which can now be compared to @xmath80 kpc estimated above .
although there are many uncertainties and assumptions in this comparison , the result indicates that the interpretation of the smoothing kernel as accounting for the difference between the cr electron diffusion scale and the photon mean free path to dust is reasonable .
our results suggest that there is a clear link between synchrotron emission , which measures the integrated massive sf history in the galaxy over a few 10@xmath81 yr , and the isrf - heated cold dust . in the latter case ,
the fraction of heating that is due to fuv photons from young massive stars is uncertain ( e.g. see the discussion in * ? ? ?
older stars contribute approximately an order of magnitude more photons than ob stars in the solar neighbourhood @xcite whereas the cross - section for large grains increases by only a factor of 5 from the infrared through the ultraviolet @xcite .
in addition , we are considering emission from regions ( of order 4.3 kpc , see above ) which are much larger than typical regions within which hot massive stars exist ; the heating source for the @xmath1 m dust emission over such size scales is then the cooler stellar population ( as has been found by other authors previously ) , rather than the hot , young massive stars that eventually give rise to the synchrotron emission via supernovae .
what then is the connection and the reason for the non - linear relation ?
we first note that non - linear relations have been observed before , especially when considering those of lower luminosity ( @xmath82 ) and/or galaxies that have been studied at low radio frequencies ( @xmath83 ghz ) . in such cases , the relation @xmath84 , where @xmath85 has been observed . such a trend has been interpreted as reflecting an increasing contribution from older stars in heating the dust , a non - linear relation between the sfr and radio continuum emission , or an underproduction of both radio and and infrared emission in low luminosity galaxies ( see e.g. * ? ? ?
* ) . to our knowledge , however , a slope as steep as @xmath86 has not previously been observed , likely because of the lack of smoothing as described above .
the closest comparisons are to the results of @xcite and @xcite for ngc 5775 and m 31 , respectively , who both find approximately linear relations between non - thermal radio emission and cold dust , in agreement with our result when a smoothing kernel is ignored . for both the 617 mhz and @xmath1 m emission
, we are observing optically thin emission over equivalent lines of sight .
then the relation compares the emissivities of the two components ( see also * ? ? ? * ) . for synchrotron emission ,
the emission coefficient can be expressed as , @xmath87 where @xmath88 represents number density of cr electrons , @xmath58 is the magnetic field strength ( we take @xmath89 ) , and @xmath90 is the power law slope of the cr electron spectrum .
there is substantial observational and theoretical evidence for an approximately constant alfvn speed in the ism leading to @xmath91 , where @xmath92 is the gas density ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and others ) .
in addition , we expect that @xmath93 , where @xmath94 is the massive sfr .
if we accept that the massive sfr is governed by a schmidt law such that @xmath95 , then the above equation becomes , @xmath96 for the dust emissivity , we have , @xmath97 which is essentially a variant of eqn . [ eqn : modified ] in which the optical depth is expressed explicitly as a function of the number density of dust grains ( @xmath98 ) .
if we take a constant gas / dust ratio , then @xmath99 . in addition , variations in gas density in the ism can be many orders of magnitude , whereas a reasonable variation in the cold dust temperature results in a variation in @xmath100 of factors of a few ( e.g. increasing the error bar on @xmath101 for ngc 3044 by a factor of two varies @xmath100 by a factor of approximately @xmath30 2 ) .
dust temperatures may vary a great deal in the ism , but since we are considering only the @xmath1 m emission which we have linked to the cold dust of ngc 3044 only , we will assume that gas density variations dominate and approximate the cold dust emission as , @xmath102 another way of expressing the above is that the isrf imposes an essentially constant heating effect on @xmath103 due to a presumed dominance of widespread and long - lived cooler stars to this heating .
combining eqns .
[ eqn : synch ] and [ eqn : cold_dust ] and restoring the flux nomenclature yields , @xmath104 taking the observational results for the schmidt law summarized by @xcite , i.e. @xmath105 , and using the mean spectral index between @xmath2 and @xmath106 cm measured by @xcite for ngc 3044 , @xmath107 , leading to @xmath108 , we find , @xmath109 which agrees with the measured @xmath110 for ngc 3044 , within errors .
see @xcite for an earlier example of this kind of approach .
it is important to note the variations that can occur in the above indices .
the measured spectral index of ngc 3044 , for example , ranges from -1.0 to 0.2 within the galaxy and may differ from these values at the lower frequency that has been used in these observations . as for the schmidt law , theoretical considerations result in values that can generally be found in the range , @xmath111 @xcite , though a free - fall value yields @xmath112 ( * ? ? ?
* his eqn .
observationally , variations in @xmath113 are also observed @xcite .
nevertheless , the above result shows that a simple application of the average spectral index and the most universally accepted schmidt law provides a reasonable explanation for the 617 mhz - 850 @xmath11 m correlation found in ngc 3044 . for this result , it is important that the non - thermal radio emission and cold dust emission are sufficiently isolated spectrally and that the dust scale length is adjusted through smoothing .
to summarize , if we assume a constant isrf heating of the cold dust , then the link between synchrotron emission and cold dust is via the gas density .
a higher gas density yields a higher sfr and a higher magnetic field , both of which affect the synchrotron emission , and a higher gas density yields a higher dust density leading to a higher cold dust emission .
we have observed the edge - on galaxies , ngc 3044 and ngc 4157 at 617 mhz using the gmrt and at @xmath10 and @xmath12 @xmath11 m using the jcmt .
these are the first results for these two galaxies at these wavelengths .
the main results are as follows .
@xmath114 for ngc 3044 , high latitude emission is observed at 617 mhz , consistent with previous radio continuum results and some evidence is present at low intensities for vertical extensions at @xmath12 @xmath11 m also .
@xmath114 at the far ends of the major axis of ngc 3044 , there appear to be disturbances which have resulted in 617 mhz emission extending away from the plane towards the north .
the most most obvious of these begin on the far south - east end of the major axis , including some disconnected features , one of which we have named the non - thermal cloud ( ntc ) .
order - of - magnitude calculations suggest that the ntc could have originated from activity related to star - formation in the galaxy s disk .
another radio continuum extension is seen on the nw end of the major axis at which an hi supershell is known .
such disturbances will be more effective at ejecting material away from the plane if they are at large galactocentric radii where the disk density is lower . @xmath114 some high latitude 617 mhz emission is observed in ngc 4157 but we do not have the sensitivity to unambiguously detect halo dust emission at either @xmath10 @xmath11 m or @xmath12 @xmath11 m for this galaxy .
a large radio continuum extension is seen towards the south beginning at the end of the north - eastern major axis in this galaxy and appears to correlate with high - latitude emission seen at @xmath38 @xmath11 m from the wise satellite .
@xmath114 we could not fit a single temperature model to the sub - mm spectrum for either ngc 3044 or ngc 4157 , but could fit both spectra with a two - temperature model . for ngc 3044 , we find t@xmath115 k and t@xmath50 k and for ngc 4157 , we find t@xmath116 k and t@xmath117 k. for the latter galaxy , we do not detect emission at the largest galactocentric radii .
dust masses are @xmath118 and @xmath119 for ngc 3044 and ngc 4157 , respectively .
there is more cold than warm dust in both galaxies .
@xmath114 we find a clear correlation between the 617 mhz and @xmath12 @xmath11 m emission for both ngc 3044 and ngc 4157 .
this correlation improves significantly if a smoothing kernel is applied to the @xmath12 @xmath11 m data to account for differences between the mean free path of a dust - heating photon and the diffusion length of cr electrons in the ism .
simple timescale and length scale arguments suggest that such a smoothing kernel can indeed account for these differences .
@xmath114 with an applied smoothing kernel , the 617 mhz and @xmath12 @xmath11 m correlation becomes strongly non - linear .
for the best data set and the one in which the @xmath12 @xmath11 m is strongly dominated by cold dust ( i.e. ngc 3044 ) , we find @xmath110 .
@xmath114 the non - linear relation between synchrotron emission and cold dust can be understood if the heating of the cold dust is the isrf in which cooler stars ( rather than hot young stars ) dominate , leading to variations in cold dust emission that are dominated by density rather than temperature variations .
synchrotron emission depends on the magnetic field strength and cr electron generation , both of which depend on gas density via @xmath120 and the schmidt law , respectively . with these assumptions , @xmath121 which agrees with the observed correlation .
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 .
we are grateful to dr .
rob swaters for providing whisp data and to dr .
siow - wang lee for providing hi and co(j=2 - 1 ) data .
thanks also to r. rand for providing the h@xmath19 image .
the co(j=1 - 0 ) data were originally provided by g. golla for comparative purposes .
we are especially grateful to loretta dunne for insightful comments . 99 ade , p. a. r. , et al .
2011 , a&a , 536 , 16 alton , p. b. , et al .
1998 , apj , 507 , l125 alton , p. b. , et al .
2000 , a&a , 356 , 795 ananthakrishnan , s. 2005 , proc .
of the int . cosmic ray conf . , 10 , 125 bell , e. f. 2003 , apj , 586 , 794 bendo , g. j. , et al . 2002 , aj , 123 , 3067 bicay , m. d. , & helou , g. 1990 , apj , 362 , 59 bigiel , f. et al . 2008 , aj , 136 , 2846 brar , r. s. , irwin , j. a. , & saikia , d. j. 2003 , mnras , 340 , 269 bregman , j. r. , & pildis , r. a. 1994 , apj , 420 , 570 chaboyer , b. , & henriksen , r. n. 1990 , a&a , 236 , 275 colbert , e. j. m. , et al . 1996 , apj , 467 , 551 collins , j. a. , rand , r. j. , duric , n. , & walterbos , r. a. m. 2000 , apj , 536 , 645 condon , j. j. 1987 , apjs , 65 , 485 condon , j. j. , helou , g. , sanders , d. b. , & soifer , b. t. 1990 , apjs , 73 , 359 cutri , r. m. , et al .
2012 , explanatory supplement to the wise all - sky data release products http://wise2.ipac.caltech.edu/docs/release/allsky/expsup/index.html de jong , t. , klein , u. , wielibinski , r. , & wunderlich , e. 1985 , a&a , 147 , 6 de vaucouleurs , g. , de vaucouleurs , a. , corwin , h. g. , jr . , et al .
1991 , third reference catalogue of bright galaxies ( rc3 ) ( springer verlag ) dsert , f .- x . , boulanger , f. , & puget , j. l. 1990 , a&a , 237 , 215 di francesco , j. , et al .
2008 , apjs , 175 , 277 dunne , l. , & eales , s. a. 2001 , mnras , 327 , 697 dunne , l. , et al .
2000 , mnras , 315 , 115 dupac , x. , et al .
2003 , mnras , 344 , 105 eales , s. a. , & edmunds , m. g. 1996 , mnras , 280 , 1167 fraternali , f. , cappi , m. , sancisi , r. , & oosterloo , t. 2002 , apj , 578 , 109 garcia , a. m. 1993 , a&as , 100 , 47 gil de paz , a. , et al . 2007 , apjs , 173 , 185 gioia , i. m. , & fabbiano , g. 1987 , apjs , 63 , 771 groves , b. a. , jungyeon , c. , dopita , m. , & lazarian , a. 2003 , pasa , 20 , 252 hammell , m. c. , & fesen , r. a. 2008 , apjs , 179 , 195 heesen , v. , beck , r. , krause , m. , & dettmar , r .- j .
2011 , a&a , 535 , 79 heesen , v. , beck , r. , krause , m. , & dettmar , r .- j .
2009 , a&a , 494 , 563 helou , g. , soifer , b. t. , & rowan - robinson , m. 1985 , apj , 298 , 7 helou , g. , & bicay , m. d. 1993 , apj , 415 , 93 henriksen , r. n. 1991 , apj , 377 , 500 hippelein , h. et al .
2003 , a&a , 407 , 137 hoernes , p. , berkhuijsen , e. m. , & xu , c. 1998 , a&a , 334 , 57 holland , w. s. , et al .
1999 , mnras , 303 , 659 howk , j. c. , & savage , b. d. 1999 , aj , 117 , 2077 hummel , e. , & van der hulst , j. m. 1989 , a&as , 81 , 51 hummel , e. , beck , r. , & dettmar , r .- j .
1991 , a&as , 87 , 309 ipavich , f. m. 1975 , apj , 196 , 107 irwin , j. a. , et al .
2012 , aj , submitted irwin , j. a. , english , j. , & sorathia , b. 1999 , aj , 117 , 2102 irwin , j. a. , & saikia , d. j. 1999 , mnras , 346 , 977 irwin , j. a. , saikia , d. j. , & english , j. 2000 , aj , 119 , 1592 irwin , j. a. , & madden , s. c. 2006 , a&a , 445 , 123 isobe , t. , feigelson , e. d. , akritas , m. g. , & babu , g. j. 1990 , apj , 364 , 104 james , a. , dunne , l. , eales , s. , & edmunds , m. g. 2002 , mnras , 335 , 753 jenness , t. , & lightfoot , j. f. 1999 , the surf programming interface , starlink system note 72 kennedy , h. j. 2009 , investigation of neutral hydrogen in three edge - on spiral galaxies , msc thesis , queen s university , canada , available at http://hdl.handle.net/1974/5134 kennicutt , r. c. , jr . 1998 , araa , 36 , 189 kennicutt , r. c. , jr .
2008 , pathways through an eclectic universe , asp conf .
series , vol .
j. h. knapen , t. j. mahoney , & a. vazdekis , san francisco , asp , p. 149
kennicutt , r. c. , jr . , &
evans ii , n. j. 2012 , araa , 50 , 531 komugi , s. , et al .
2008 , apjs , 178 , 225 larson , r. b. 1992 , in star formation in stellar systems , ed .
g. tenorio - tagle , m. prieto , & f. snchez ( cambridge : cambridge univ .
press ) , 125 lee , s .- w .
1998 , the interstellar medium and disk - halo interaction of the edge - on galaxies ngc 3044 and ngc 5775 , phd thesis , queen s university , canada lee , s .- w . , & irwin , j. a. 1997 , apj , 490 , 247 lee , s .- w . , et al .
2001 , a&a , 377 , 759 marengo , m. , et al .
1997 , a&a , 322 , 924 masi , s. , et al .
1995 , apj , 452 , 253 marsh , k. a. , & helou , g. 1998 , 493 , 121 miller , s. t. , & veilleux , s. 2003 , apjs , 148 , 383 murphy , e. j. , et al .
2006a , apj , 638 , 157 murphy , e. j. , et al . 2006b , apj , 651 , l111 ; erratum 2008 , apj , 672 , l85 murphy , e. j. , helou , g. , kenney , j. d. p. , armus , l , & braun , r. 2008 , apj , 678 , 828 murphy , e. j. , kenney , j. d. p. , helou , g. , chung , a. , & howell , j. h. 2009 , apj , 694 , 1435 niklas , s. , & beck , r. 1997 , a&a , 320 , 54 noordermeer , e. , van der hulst , j. m. , sancisi , r. , swaters , r. a. , & van albada , t. s. 2005 , a&a , 442 , 137 pierini , d. , popescu , c. c. , tuffs , r. j. , & vlk , h. j. 2003 , a&a , 409 , 907 rand , r. j. , wood , k. , benjamin , r. a. , & meidt , s. e. 2011 , apj , 728 , 163 rhee , m .- h . , & van albada , t. s. 1996 , a&as , 115 , 407 sanders , d. b. , & mirabel , i. f. 1996 , araa , 34 , 749 sanders , d. b. , et al .
2003 , aj , 126 , 1607 seaquist , e. , yao , l , dunne , l. , & cameron , h. 2004 , mnras , 349 , 1428 sofue , y. 1987 , pasj , 39 , 547 solomon , p. m. , & sage , l. j. 1988 , apj , 334 , 613 stevens , j. a. , & gear , w. k. 2000 , mnras , 312 , l5 swarup , g. 1991 , current science , 60 , 106 swarup , g. , et al .
1991 , current science , 60 , 95 tabatabaei , f. s. , beck , r. , & berkhuijsen , e. 2010a , highlights of astronomy , vol .
15 , p. 417 tabatabaei , f. s. , & berkhuijsen , e. 2010b , a&a , 517 , 77 thompson , t. a. , quataert , e. , & murray , n. 2009 , mnras , 397 , 1410 tielens , a. g. g. m. 2006 , the physics and chemistry of the interstellar medium , cambridge university press , cambridge troland , t. h. , & heiles , c. 1986 , apj , 301 , 339 tllmann , r. , & dettmar , r .- j .
2000 , a&a , 362 , 119 tllmann , r. , et al .
2006 , a&a , 448 , 43 tully , r. b. 1988 , nearby galaxies catalog , ( cambridge university press , cambridge ) tully , r. b. , et al . 1996 , aj , 112 , 2471 verheijen , m. a. w. , & sancisi , r. 2001 , a&a , 370 , 765 vinyakin , e. n. 2007 , arep , 51 , 87 vlahakis , c. , dunne , l. , & eales , s. 2005 , mnras , 364 , 1253 wang , q. d. , et al .
1995 , apj , 439 , 176 wang , q. , et al .
2001 , apj , 555 , l99 whaley , c. h. , irwin , j. a. , madden , s. c. , galliano , f. , & bendo , g. j. 2009 , mnras , 395 , 97 willmer , c. n. a. , et al .
2009 , aj , 138 , 146 wright , e. l. et al . 2010 , aj , 140 , 1868 wunderlich , e. , & klein , u. 1988 , a&a , 206 , 47 young , j. s. , et al .
1995 , apjs , 98 , 219 young , j. s. , et al .
1996 , aj , 112 , 1903 yun , m. s. , reddy , n. a. , & condon , j. j. 2001 , apj , 554 , 803 zhu , m. , gao .
y , seaquist , e. r. , & dunne , l. 2007 , aj , 134 , 118 | we present the first maps of ngc 3044 and ngc 4157 at @xmath0 m and @xmath1 m from the jcmt as well as the first maps at 617 mhz from the gmrt .
high latitude emission has been detected in both the radio continuum and sub - mm for ngc 3044 and in the radio continuum for ngc 4157 , including several new features .
for ngc 3044 , in addition , we find 617 mhz emission extending to the north of the major axis , beginning at the far ends of the major axis .
one of these low intensity features , more than 10 kpc from the major axis , has apparently associated emission at @xmath2 cm and may be a result of in - disk activity related to star formation . the dust spectrum at long wavelengths required fitting with a two - temperature model for both galaxies , implying the presence of cold dust ( t@xmath3 k for ngc 3044 and t@xmath4 k for ngc 4157 ) .
dust masses are @xmath5 and @xmath6 for ngc 3044 and ngc 4157 , respectively , and are dominated by the cold component .
there is a clear correlation between the 617 mhz and @xmath1 m emission in the two galaxies . in the case of ngc 3044 for which the @xmath1 m data are strongly dominated by cold dust
, this implies a relation between the non - thermal synchrotron emission and cold dust .
the 617 mhz component represents an integration of massive star formation over the past @xmath7 yr and the @xmath1 m emission represents heating from the diffuse interstellar radiation field ( isrf ) . the 617 mhz
@xmath1 m correlation improves when a smoothing kernel is applied to the @xmath1 m data to account for differences between the cr electron diffusion scale and the mean free path of an isrf photon to dust .
the best - fit relation is @xmath8 for ngc 3044 .
if variations in the cold dust emissivity are dominated by variations in dust density , and the synchrotron emission depends on magnetic field strength ( a function of gas density ) as well as cr electron generation ( a function of massive star formation rate and therefore density via the schmidt law ) then the expected correlation for ngc 3044 is @xmath9 , in agreement with the observed correlation .
[ firstpage ] galaxies : individual : ngc 3044 , ngc 4157 ; galaxies : ism ; infrared : galaxies ; radio continuum : galaxies |
the superconducting state of the underdoped cuprates shows anomalous properties when compared with the case of optimum or overdoped .
there has been a considerable recent effort to understand these in a model which includes the emergence of a pseudogap for doping @xmath6 below a quantum critical point ( qcp ) at @xmath7 .
the model of yang , rice , and zhang@xcite ( yrz ) on which this work is based , has the pseudogap forming about the antiferromagnetic brillouin zone ( afbz ) boundary with its own characteristic energy scale .
this model is different in many respects from other competing order proposals such as d - density waves@xcite and from a preformed pair model@xcite which involves a single energy scale . for a review of the successes of the yrz model in understanding the data ,
the reader is referred to schachinger and carbotte@xcite .
many more details can be found in references @xcite it is important to understand that this work goes beyond extensions of bcs theory to include effects such as strong coupling due to inelastic scattering@xcite and possible anisotropy@xcite beyond a simple @xmath0-wave superconducting gap . in a previous work@xcite
, we showed that the strong suppression of the specific heat jump at @xmath8 and corresponding reduction in condensation energy with increased underdoping can be understood as due to the emergence of a pseudogap . here
we consider the equally anomalous observation that the slope of the linear in temperature law as @xmath9 is a strongly increasing function of doping @xmath6 in the overdoped regime while it is nearly constant at underdoping.@xcite while the overdoped case is characteristic of a fermi liquid , the underdoped case requires a new ingredient for its understanding . in addition , we will also consider the recent data of wen et al.@xcite on bi@xmath2sr@xmath10la@xmath4cuo@xmath5 for which @xmath8 at optimum is only 30k and thus superconductivity can be quenched with a 9 tesla magnetic field providing access to the normal state as a function of doping in both fermi liquid ( overdoped ) and pseudogap ( underdoped ) state . in sec .
ii , we present the essential elements of the theory of yrz@xcite needed for the calculation of the specific heat . both pseudogap state alone and with the addition of superconductivity within a bcs formulation are considered . for @xmath6
greater than @xmath11 , the doping at which the qcp associated with the pseudogap formation is set , the fermi surface is the usual large open contour of band theory and of a fermi liquid . for @xmath12 , however , the fermi surface reconstructs into luttinger pockets which progressively shrink in size as the mott insulating state is more closely approached .
a consequence of this is that the density of states at the fermi level is reduced .
in addition , a gutzwiller factor enters the theory which accounts for the depletion of the coherent part of the electronic green s function due to increased correlations which eventually cause the transition to an insulating state . in sec .
iii , we present our results for the specific heat difference between superconducting and normal state as a function of temperature for values of doping ranging from 0.1 to 0.3 with @xmath13 the critical doping at optimum which is also the qcp in our model . the zero temperature limit of the specific heat difference is also considered more explicitly and compared with the density of states and the data of wen et al.@xcite . in sec .
iv , we show that while pseudogap formation strongly affects the overall temperature dependence of the specific heat , it does not change its slope at low @xmath1 .
this arises because this linear law only depends on the band structure near the dirac point on the heavily weighted part of the fermi surface in the nodal direction and this point is not changed by pseudogap formation in the model of ref .
@xcite . on the other hand
the size of the slope depends directly on the gutzwiller factor @xmath14 which gives the magnitude of the remaining coherent part of the green s function .
it also depends on the ratio of the superconducting gap to critical temperature which is expected to strongly increase with decreasing doping @xmath6 for @xmath12.@xcite these two effects combined lead to a rather constant value of the slope over a significant range of doping . for overdoping , fermi liquid theory is recovered and in this case the slope shows much greater variation with @xmath6 as in experiment .
comparison with data is presented . in sec .
v , we provide a summary and give our conclusions .
in the resonating valence bond spin liquid model@xcite , the self - energy due to the pseudogap is given by @xmath15 $ ] , where @xmath16 with @xmath17 the effective first neighbor hopping parameter and @xmath18 is a pseudogap which has @xmath0-wave symmetry in the brillouin zone .
it is given by @xmath19 here the amplitude @xmath20 is linear in doping @xmath6 as shown in fig .
[ fig1](a ) where the superconducting dome is also shown for easy orientation . in the above , @xmath21 is the momentum and @xmath22 is the lattice constant of the cuo@xmath2 plane . without superconductivity , the coherent part of the electronic green s function is given by @xmath23},\ ] ] where @xmath24 is the electron dispersion curve of band theory and @xmath14 is a gutzwiller factor equal to @xmath25 .
this latter quantity accounts for the effect of correlations which reduces the weight of the coherent part of @xmath26 and adds an incoherent background not considered in this work .
it also enters the renormalized band structure dispersion curve which includes up to third nearest neighbor hopping and narrows as the mott transition is approached with decreasing value of @xmath6 .
this narrowing is modeled by a second gutzwiller factor @xmath27 in addition to @xmath14 .
effectively , for a given @xmath21 , there are two electron branches @xmath28 with weights @xmath29 given by @xmath30 and @xmath31 ^ 2 + \delta^2_{\rm pg}(\boldsymbol{k})}}\biggr].\ ] ] in terms of @xmath28 , the fermi surface contours of zero excitation energy are given by @xmath32 and these are shown in fig .
[ fig1](b ) for three values of doping @xmath33 and 0.2 . in the first two ,
there is a hole pocket centered about the nodal direction @xmath34 .
this pocket is determined from the equation @xmath35 and @xmath36 is positive only for momenta falling within the area defined by the hole pocket .
the size of the pocket shrinks as @xmath6 decreases and we come closer to half filling and the mott insulating state . for the case
@xmath37 , close to the qcp at @xmath38 where pseudogap formation starts in our model , there is an additional electron pocket near the corner of the afbz .
this pocket is determined by the equation @xmath39 .
both electron and hole pockets have two sides , one weighted by @xmath29 of order one and the other , which takes on a close resemblance to the afbz , has only a small weight in comparison .
this small weight goes to zero in the limit of no pseudogap and the energy @xmath28 becomes the umklapp energy surface @xmath40 . on the other hand , in this same limit ,
the heavily weighted part traces out the large fermi surface of fermi liquid theory ( fig .
[ fig1](b ) , far right panel ) which has weight one everywhere .
these remarks make clear the evolution from large fermi surface into small hole pockets .
as the pockets shrink in size , the number of states which carry excitations of zero energy becomes small .
also , it should be kept in mind that because of the small weight on the backside of the hole pocket , we are effectively dealing with an arc when considering many properties . when superconductivity is included in a bcs formalism
, the electronic spectral density @xmath41 can be written in the form @xmath42 , \label{eq : a}\ ] ] with @xmath43 . here , the superconducting gap @xmath44 is assumed to have the same @xmath0-wave dependence in momentum space as in eq .
( [ eq : gapk ] ) for the pseudogap with amplitude @xmath45 replacing @xmath46 and @xmath45 is assumed to have the same doping dependence as the critical temperature @xmath47 dome .
this is shown in fig . [ fig1](a ) and for definiteness we will assume in all our numerical work that @xmath48 . when later we compare with experimental data
, we will remove this simplifying assumption .
the bogoliubov weights in eq .
( [ eq : a ] ) are @xmath49 the dispersion curves for @xmath50 and @xmath40 are taken from the work of ref .
@xcite and are unchanged here as are the other parameters , namely , @xmath51 \label{eq : delscx}\ ] ] and @xmath52 with @xmath53 an unrenormalized nearest neighbor hopping parameter characteristic of the cuo@xmath2 plane .
there are two equivalent ways for calculating the specific heat @xmath54 .
one is through the entropy @xmath55 , the other through the internal energy @xmath56 .
the entropy is @xmath57\ln[1-f(e^\alpha_{\boldsymbol{k}s})]\},\nonumber\\ & & \label{eq : s}\end{aligned}\ ] ] where @xmath58 is the fermi - dirac temperature distribution function .
alternatively , the internal energy can be expressed in terms of the single spin density of states @xmath59 as @xmath60 where the two is for spin degeneracy and the density of states is given by @xmath61 .
\label{eq : dos}\ ] ] in both eqs .
( [ eq : s ] ) and ( [ eq : dos ] ) , the sum is over the entire brillouin zone and @xmath62 .
( color online ) ( a ) phase diagram for the model discussed in the paper , showing the pseudogap @xmath63 and superconducting gap @xmath64 in units of @xmath53 as a function of doping @xmath6 .
( b ) illustration of the reconstruction of the fermi surface in the first quadrant of the brillouin zone for doping values of @xmath65 , 0.18 and 0.2 .
the red dashed line is the afbz boundary .
, scaledwidth=48.0% ]
( color online ) @xmath66 versus @xmath67 , where @xmath68 is in units of @xmath69 per volume .
curves are shown for ( a ) underdoped and ( b ) overdoped cases along with optimal doping for reference . ,
scaledwidth=48.0% ] in fig .
[ fig2 ] , we present our results for the difference between the superconducting and normal state specific heat @xmath70 , namely @xmath71 in units of @xmath69 per volume as a function of temperature @xmath1 .
several values of doping @xmath6 are considered as indicated in the figure .
the top frame covers the underdoped regime , while the bottom frame is for overdoped . as the superconducting dome given in eq .
( [ eq : delscx ] ) is symmetric in doping about optimum @xmath38 , the top and bottom frame curves come in pairs with the same value of critical temperature ( with the exception of not displaying @xmath72 for clarity ) .
comparison of top and bottom curves in a given pair shows that the formation of a pseudogap and associated fermi surface reconstruction [ fig .
[ fig1](b ) ] provides a drastic suppression of the jump at the critical temperature @xmath8 and also of the slope just below @xmath8 .
this is in qualitative agreement with experiment@xcite , as discussed already in ref .
these results are clearly not part of ordinary @xmath0-wave bcs theory where instead the jump is large and relatively independent of doping as seen in the fig .
[ fig2](b ) . for reference in assessing these curves
, we recall that for a constant density of states model with a superconducting @xmath0-wave gap defined on the fermi surface , the canonical value of the jump @xmath73 is 0.95 for a gap to critical ratio @xmath74 and is the same for all superconductors regardless of the size of @xmath8 . here
it deviates from this universal law because our superconducting gap is defined in the entire brillouin zone according to eq .
( [ eq : gapk ] ) with the pseudogap replaced by @xmath64 and our energy bands can be complicated even when the large fermi surface of fermi liquid theory is involved .
also , the magnitude of the jump itself is increased because we have used a gap to @xmath8 ratio of 6 rather than the weak coupling limit of 4.3 and this has resulted in a change of @xmath73 from 0.95 to @xmath75 .
thus , the large decrease in the jump @xmath76 seen in fig . [ fig2](a ) is a direct consequence of pseudogap formation and accompanying fermi surface reconstruction .
( color online ) ( a ) @xmath77 as @xmath9 normalized to the same quantity at optimal doping versus doping @xmath6 .
values extracted from the data of wen et al.@xcite are shown along with the dos at zero energy @xmath78 similarly normalized .
the inset shows @xmath59 in units of @xmath53 for several dopings as a function of @xmath79 .
@xmath59 and @xmath78 refer to the normal state with a pseudogap .
( b ) value of @xmath77 at @xmath8 from fig .
[ fig2 ] ( left axis ) and slope just below @xmath8 , @xmath80 ( right axis ) as a function of @xmath6 . , title="fig:",scaledwidth=48.0% ] ( color online ) ( a ) @xmath77 as @xmath9 normalized to the same quantity at optimal doping versus doping @xmath6 .
values extracted from the data of wen et al.@xcite are shown along with the dos at zero energy @xmath78 similarly normalized .
the inset shows @xmath59 in units of @xmath53 for several dopings as a function of @xmath79 .
@xmath59 and @xmath78 refer to the normal state with a pseudogap .
( b ) value of @xmath77 at @xmath8 from fig .
[ fig2 ] ( left axis ) and slope just below @xmath8 , @xmath80 ( right axis ) as a function of @xmath6 .
, title="fig:",scaledwidth=48.0% ] the size of the specific heat difference seen at zero temperature is also of interest and is very different in the underdoped than in the overdoped regime . while for near optimum , optimum and overdoped cases the value of @xmath66 as @xmath9 is around -1 in our units , for the underdoped cases , it has moved instead to a value of roughly @xmath81 . in this limit @xmath82
simply reduces to its normal state value @xmath83 and is a direct measure of the value of the density of states @xmath59 at the fermi energy , _
i.e. _ at @xmath84 .
[ fig3](a ) , we compare the results of our calculations for @xmath85 ( circles ) with the specific heat data of wen et al.@xcite on bi@xmath2sr@xmath3la@xmath4cuo@xmath5 ( squares ) . here
, opt refers to the value taken at optimal doping ( @xmath38 in our model ) . in carrying out this comparison ,
we have fit a parabolic form to the data of wen et al . for @xmath8 versus doping .
we found that the fit which best captured the data was when optimum was taken at their doping of @xmath86 and @xmath53 is set to 56 mev for @xmath87 .
similar , but less ideal fits would also work for optimum at @xmath88 . note that this material does not have a very high @xmath8 and consequently has a narrower dome compared with what is typically found in the literature .
it was chosen for experiment because reasonable magnetic fields can be used to suppress @xmath8 to zero and so access the normal state .
given our fit , we have shifted the value of @xmath89 in experiment by 0.035 to get correspondence with our theoretical work for which @xmath6 at optimum is kept at @xmath38 .
further discussion of this is presented in sec .
we have kept the pseudogap line to be as shown in fig .
[ fig1](a ) which coincides with optimum doping , but this could easily be changed .
returning to fig .
[ fig3](a ) , both theory and experiment show two distinct regimes : a somewhat flat or slowly rising region above @xmath90 with a sharp drop below this value . the behavior in the region above @xmath38 is sensitive to the choice of bandstructure parameters as shown in the original work of yrz ( ref.@xcite ) .
because the data is fairly flat in this region , we have opted to alter the bandstructure accordingly and have allowed the @xmath91 and @xmath92 parameters to continue to vary with @xmath6 rather than to become constant for @xmath93 as in ref .
@xcite . returning to the rapid drop around @xmath94
, we find that it is displaced to slightly lower values of @xmath6 in the data as compared to theory and we take this to mean that in bi@xmath2sr@xmath3la@xmath4cuo@xmath5 the qcp associated with the start of pseudogap formation occurs at a doping level close but slightly below optimum .
it is important to realize that in the present model , the point at which the value of the density of states at the fermi level @xmath78 begins to be strongly reduced by the growth of the pseudogap , is not exactly at the qcp but is instead displaced to smaller @xmath6 values .
the physics of this displacement is easily understood with reference to the inset in fig .
[ fig3](a ) .
first we show , in the main frame , that our numerical results for @xmath78 versus @xmath6 ( triangles ) are very close to the solid black circles as they must be .
these were obtained not from an extrapolation to zero temperature of a specific heat calculation as were the solid circles but instead from the full dos @xmath59 versus @xmath95 shown in the inset .
this good agreement is taken as providing a check on our numerical work .
more significantly it is important to realize that the fermi surface reconstruction from the large open fermi liquid surface of fig .
[ fig1 ] ( @xmath38 ) to a luttinger hole pocket does not immediately lead to change in the value of @xmath78 .
in fact , as can be seen clearly in the inset of fig .
[ fig3](a ) , at @xmath37 the dos at @xmath84 is hardly changed from its fermi liquid value ( see solid black curve for @xmath38 for comparison ) .
rather the effect of the pseudogap is to provide a depletion of states at negative @xmath95 . for @xmath96
however , the upper edge of this depletion region which is rather sharp , has moved across the fermi energy and @xmath78 has become significantly reduced as shown in the triangles of the main frame in fig . [ fig3](a ) .
this is not surprising .
the dos at @xmath84 depends on the luttinger contours of zero energy shown in fig .
[ fig1 ] . at @xmath37 , there are both hole and electron pockets and the number of states having zero energy is not very different from the number when the large open fermi surface applies .
it is only when we reach a single small hole pocket and no electron pockets that depletion of states at @xmath84 becomes significant . on the face of it
, one might even think that more zero energy states are involved when @xmath37 than when @xmath38 because the length of the fermi surface is larger when there are pockets .
but this is not so because , as we have already stated , parts of the luttinger contours carry very little weight . in fig .
[ fig3](b ) , we show results for the jump at @xmath8 of the specific heat @xmath76 ( circles ) and its slope out of @xmath8 ( squares ) as a function of doping . we note that for the jump @xmath76 , a notable drop occurs almost immediately below optimum @xmath38 , which can be used to identify the qcp . for the slope ,
the signature of the qcp is not as sharp , only a change in curvature arises at @xmath38 .
these results are to be contrasted with those in fig .
[ fig3](a ) . if one were to estimate the value of doping to be associated with the qcp from the behavior of @xmath66 at @xmath9 , it is necessary to account for a significant displacement downward towards smaller values of @xmath6 of the rapid drop in @xmath78 as compared to the value of @xmath6 at the qcp .
thus extracting a qcp from thermodynamics requires some care but it clearly can be done either from the jump at @xmath8 or the value of the difference @xmath66 as @xmath9 .
we now consider the low temperature behavior of the superconducting state specific heat @xmath97 which is emphasized in fig .
[ fig4](a)-(d ) for underdoped cases .
we base our discussion on the physics of the schematic shown in fig .
[ fig4](e ) .
the luttinger pockets are shown as heavy black lines for @xmath37 which has electron as well as hole areas .
also shown is a dirac cone centered on the highly weighted fermi surface contour in the nodal direction .
the cone illustrates the quasiparticle energies at low energy in the superconducting state as a function of @xmath98 and @xmath99 in the brillouin zone . only the upper right quadrant is depicted . at very low temperatures ,
the tip of this cone is the only region in momentum space where there is a finite thermal occupation of excited quasiparticles .
thus , the specific heat can depend only on characteristic parameters associated with the dirac point .
but this point is particularly simple .
it corresponds to @xmath35 in the normal pseudogap case and does not change with the onset of superconductivity . in the nodal direction , the pseudogap @xmath100 and therefore @xmath101 which is the condition for the underlying large fermi surface .
note that strictly speaking @xmath50 involves the chemical potential @xmath102 associated with the case when the pseudogap is present and is slightly different from the chemical potential of fermi liquid theory . in the former case
, it is determined from the luttinger sum rule while in the latter case of a large fermi surface one might determine @xmath102 from the dos filling up to @xmath103 . neglecting this small difference
, the band energy at the dirac point is unchanged from its fermi liquid value and the well known techniques@xcite for obtaining the @xmath104 limit of the dos @xmath59 apply unaltered .
the result is @xmath105 where @xmath106 and @xmath107 are the fermi and gap velocity , respectively .
note that eq .
( [ eq : nlow ] ) predicts that the only effect of the pseudogap formation on the dos around the dirac point is the appearance of the gutzwiller factor @xmath14 . while to a good approximation @xmath106 is unchanged , the gap velocity can be changed in magnitude if the ratio of zero temperature gap to critical temperature @xmath8 is affected by pseudogap formation as was found in the recent work of schachinger and carbotte@xcite .
this provides a second important change in eq .
( [ eq : nlow ] ) as compared with the more familiar fermi liquid case . inserting eq .
( [ eq : nlow ] ) into eq .
( [ eq : u ] ) for the internal energy @xmath56 , we obtain the simple analytic result for @xmath97 in the limit of low temperature : @xmath108 where @xmath109 is the boltzmann constant and @xmath110 is a number given by @xmath111 we have obtained , as in ordinary @xmath0-wave bcs theory , a linear in @xmath1 law with the same material factors appearing , _
i.e. _ @xmath106 and @xmath107 , but with an extra gutzwiller factor of @xmath14
. we also need to note that @xmath112 can be affected by the variation with doping @xmath6 of the gap ratio @xmath113 .
( color online ) @xmath70 versus @xmath67 shown in the absence of any gaps @xmath114 , with superconductivity only @xmath115 , pseudogap only @xmath116 and with both gaps present @xmath117 .
the red dashed line gives the theoretical expression for the slope from eq .
( [ eq : slope ] ) .
( a)-(d ) show a range of dopings for the underdoped case .
( e ) schematic of the superconducting dirac cone shown on the side of the @xmath37 fermi pocket with large quasiparticle weight .
, title="fig:",scaledwidth=48.0% ] ( color online ) @xmath70 versus @xmath67 shown in the absence of any gaps @xmath114 , with superconductivity only @xmath115 , pseudogap only @xmath116 and with both gaps present @xmath117 .
the red dashed line gives the theoretical expression for the slope from eq .
( [ eq : slope ] ) .
( a)-(d ) show a range of dopings for the underdoped case .
( e ) schematic of the superconducting dirac cone shown on the side of the @xmath37 fermi pocket with large quasiparticle weight .
, title="fig:",scaledwidth=48.0% ] in fig .
[ fig4 ] , we show our numerical results for @xmath97 at four values of @xmath6 namely 0.12 , 0.14 , 0.16 and 0.18 . in all cases ,
the solid line is for the case when no pseudogap is included while the dashed - dotted includes a pseudogap with its magnitude chosen to correspond to the value of doping @xmath6 chosen . otherwise , there are no other changes .
gutzwiller factors are included in these curves and the ratio @xmath118 is fixed at value 6 as in all other numerical work presented in this paper . while the inclusion of a pseudogap has drastic effects on @xmath97 , the slope as @xmath9 is completely unaffected and this slope agrees perfectly with the simplified analytic results of eq .
( [ eq : slope ] ) given as the red dashed line .
this constitutes an important prediction of yrz theory and will be verified later when we make comparison with experimental data . the heavy dotted lines in fig .
[ fig4 ] show the normal state specific heat @xmath119 in the fermi liquid and the light dotted , with a pseudogap included @xmath120 .
it is instructive to consider these in some detail and in particular to describe how they are related to the detailed variation of @xmath59 versus @xmath95 in the pure pseudogap state , _
i.e. _ with no superconductivity .
we begin with the case @xmath37 for which the pseudogap is small and the fermi contours include electron as well as hole pockets . in this case , the light dotted curve falls below the heavy dotted one but rises to meet it as @xmath9 .
this behavior can be traced to the @xmath95 variation seen in the inset of fig .
[ fig3](a ) which shows @xmath59 versus @xmath95 .
the solid black curve forms a reference and is the fermi liquid result for @xmath38 , zero pseudogap . comparing with the long - dashed green curve for @xmath37 , we note no visible change in the value of @xmath121 but
there is a significant depression of dos at negative energies .
as @xmath9 only @xmath78 is sampled in the specific heat and hence both cases , with and without a pseudogap , agree . at small but finite @xmath1
, however , the light - dotted curve falls below the heavy dotted line in fig .
[ fig4](d ) because the dip in its dos at @xmath122 starts to be sampled and this reduces the specific heat .
specific heat is , however , a rather broad spectroscopy for @xmath59 because the thermal factor in @xmath70 in eq .
( [ eq : u ] ) samples of order @xmath123 or so about @xmath84 .
the case @xmath124 is also noteworthy . in this instance
the dos about the fermi energy is nearly monotonic although depressed in value as compared to @xmath38 by about 50% and this correspondingly reduces the value of @xmath119 by the same amount with little other changes .
( color online ) the slope of the specific heat at low @xmath1 normalized to the value at optimal doping .
frame ( a ) shows the data extracted from wen et al.@xcite compared with the theory of eq .
( [ eq : slope ] ) for the @xmath8 dome given in the inset where we have fit a functional form for @xmath6 to their @xmath8 data ( see text for discussion ) .
the dashed curve includes the correction @xmath118 discussed in the text .
( b ) the extracted data of loram et al.@xcite is compared to the theoretical slope . shown also is the slope data from the sts dos of pushp et al.@xcite .
, title="fig:",scaledwidth=48.0% ] ( color online ) the slope of the specific heat at low @xmath1 normalized to the value at optimal doping .
frame ( a ) shows the data extracted from wen et al.@xcite compared with the theory of eq .
( [ eq : slope ] ) for the @xmath8 dome given in the inset where we have fit a functional form for @xmath6 to their @xmath8 data ( see text for discussion ) .
the dashed curve includes the correction @xmath118 discussed in the text .
( b ) the extracted data of loram et al.@xcite is compared to the theoretical slope . shown also is the slope data from the sts dos of pushp et al.@xcite .
, title="fig:",scaledwidth=48.0% ] we turn next to a comparison with data on the slope of the specific heat in the @xmath9 limit .
these are presented in fig .
[ fig5](a ) for the bi@xmath2sr@xmath3la@xmath4cuo@xmath5 data of wen et al.@xcite and in ( b ) for the data on bi-2212 and ( y , ca)-123 of loram et al.@xcite where we have read slopes from the published figures as best as was possible .
a more thorough analysis by the experimentalists might achieve better accuracy . in the inset of fig .
[ fig5](a ) , we show our results for the fit to the @xmath8 dome of the wen et al.@xcite data .
we find @xmath125t_0,\ ] ] which provides a good representation of the @xmath8 data as a function of doping @xmath89 .
we then take @xmath126 to place the data on our curves . in the main frame ,
the solid black circles are the data for the slope at @xmath9 normalized to its value at optimum doping .
the solid red line gives results of eq .
( [ eq : slope ] ) with the assumption of @xmath127 .
the agreement is very good and the sharp rise in slope in the overdoped region is captured by our model .
for the deeply underdoped case , our theoretical values are somewhat higher than experiment , but as we have already mentioned , the gap ratio may well vary with doping . solving a bcs - like pairing equation with pseudogap formation and corresponding
fermi surface reconstruction accounted for , schachinger and carbotte@xcite found that ( for @xmath128 ) approximately @xmath129.\ ] ] if this correction , scaled to 6 at optimum , is incorporated into the comparison with experiment , we get the red dashed curve which agrees better with the data at small @xmath6 .
note finally that to compare data with theory , we have shifted all @xmath6 values in fig . [ fig5](a ) by 0.035 because we wished to remain , as in the paper of yrz@xcite , with optimum doping at @xmath38 rather than the experimental value of about 0.165 .
this is also true for fig .
[ fig5](b ) where we compare with data on bi2212 ( triangles ) and ( y , ca)-123 ( squares ) .
the agreement with theory ( solid and dashed red curves ) is again good .
we have also included one further comparison with scanning tunneling spectroscopy ( sts ) results on bi2212 by pushp et al.@xcite ( circles ) .
sts gives the dos and not the specific heat , but this latter quantity follows directly from a knowledge of @xmath59 .
we can use the sts data in the limit of @xmath104 to determine @xmath59 as in eq .
( [ eq : nlow ] ) and so get @xmath97 for @xmath9 .
these are the solid circles which provide a confirmation of the specific heat results and also provide a significant cross check between these two important but very different probes of the microscopic structure of the superconducting state in the underdoped cuprates and indeed over the entire phase diagram .
the behavior of the specific heat of the underdoped cuprates differs profoundly from that observed on the overdoped side of their phase diagram . at optimum and overdoping , an ordinary bcs approach based on a fermi liquid normal state with constant dos
provides a first reasonable understanding on the assumption that the gap has @xmath0-wave symmetry .
this ensures a linear in @xmath1 low temperature law for the specific heat @xmath70 with @xmath130 .
it also gives a jump at @xmath8 normalized to its normal state value of 0.95 and this value can be increased if the gap to @xmath8 ratio is changed from 4.3 to a higher value . on the deeply underdoped side ,
however , there is new physics which can not be described even when strong coupling effects@xcite due to inelastic scattering@xcite are accounted for and/or effects of anisotropy@xcite . which go beyond a simple lowest harmonic @xmath0-wave picture for the superconducting gap . in the resonating valence bond spin liquid picture , a second energy scale , the pseudogap , emerges and grows in magnitude as @xmath6 is reduced towards the mott insulating state .
this pseudogap leads to a loss of metallicity .
it also radically reduces the size of the specific heat jump at @xmath8 as noted in experiment@xcite and also in theory@xcite .
however , as we show here the limit of @xmath9 of @xmath97 is not directly affected by the size of @xmath63 because this limiting value depends only on the band structure and superconducting gap right at the dirac point in the brillouin zone .
but this point is not importantly changed by pseudogap formation and fermi surface reconstruction so that the formula for the slope @xmath131 as @xmath9 remains unchanged in form but with two important modifications .
first , in the resonating valence bond spin liquid , there appears a gutzwiller factor @xmath14 which depends strongly on doping @xmath6 and represents the remaining weight in the coherent part of the electronic green s function as correlations become more important and consequently shift more spectral weight into an incoherent background at higher energies .
a second factor is that the ratio of @xmath132 can vary with doping . for overdoped and
optimally doped , in ordinary bcs @xmath0-wave it has a value of 4.3 , but schachinger and carbotte@xcite have found that it rises considerably in the underdoped region of the phase diagram .
these authors solve a bcs gap equation generalized to include pseudogap formation and attendant fermi surface reconstruction .
the results of such a theoretical study show that @xmath118 increases beyond a value of 7 before the lower end of the superconducting dome is reached .
these two effects , along with the linear in @xmath1 law which we have shown to still hold in yrz theory at underdoping , allow us to understand a previously anomalous feature of the data namely the slope remains reasonably constant at underdoping while it increases sharply in the overdoped region .
this last observation is consistent with a slope which varies inversely as the gap and this gap decreases toward zero as we approach the upper end of the dome .
comparison between theory and data on bisrlacuo , bi2212 , and ( y , ca)-123 show good agreement .
a further comparison was made with sts data which provides information on the average density of quasiparticle states .
it was noticed by pushp et al.@xcite that the slope of this quantity , in the @xmath133 limit , while increasing with reduced value of @xmath8 at overdoping , saturates and perhaps even decreases slightly with decreasing @xmath6 in the highly underdoped regime .
but the low @xmath95 dependence of the dos determines the low temperature behavior of the specific heat .
for the specific case of bi2212 , we found good agreement between sts and specific heat data further confirming our work and providing a strong test of the consistency between experimental data obtained by these two very different techniques and their consistency with the resonating valence bond spin liquid .
we have benefitted from discussions with james leblanc .
this work has been supported by the natural sciences and engineering research council of canada ( nserc ) and by the canadian institute for advanced research ( cifar ) . | the specific heat of the superconducting cuprates is calculated over the entire phase diagram .
a @xmath0-wave bcs approach based on the large fermi surface of fermi liquid and band structure theory provides a good description of the overdoped region . at underdoping it is essential to include the emergence of a second energy scale , the pseudogap and its associated gutzwiller factor , which accounts for a reduction in the coherent piece of the electronic green s function due to increased correlations as the mott insulating state is approached .
in agreement with experiment , we find that the slope of the linear in @xmath1 dependence of the low temperature specific heat rapidly increases above optimum doping while it is nearly constant below optimum .
our theoretical calculations also agree with recent data on bi@xmath2sr@xmath3la@xmath4cuo@xmath5 for which the normal state is accessed through the application of a large magnetic field .
a quantum critical point is located at a doping slightly below optimum . |
microwave induced magnetoresistance oscillation in high - mobility two - dimensional ( 2d ) electron systems has been a subject of intensive experimental @xcite and theoretical @xcite studies over the past decades . though a consensus on the period and phase of the oscillation has long been reached , the effect of the microwave polarization on the amplitude of the oscillation remains one of the most challenging and unsolved issues since the discovery of microwave - induced magnetoresistance oscillations .
early measurement on l - shaped hall bars indicated that the period and phase of radiation - induced magnetoresistance oscillations are insensitive to the relative orientation between the microwave polarization and the current.@xcite a later experiment@xcite carried out on specimens with a square geometry in a quasioptical setup reported a striking result that not only the frequency and phase but also the amplitude of radiation - induced resistance oscillations and the zero resistance regions are notably immune to the sense of circular and linear polarizations of microwaves .
this influential result raised a big challenge to the existing theoretical models , in which , though no detailed investigation was reported , some kind of polarization dependence was believed to exist,@xcite and thus expedited the emergence of different scenarios capable of showing polarization immunity of microwave magnetoresistance response.@xcite in a recent study , mani _ et al_.@xcite found a strong sensitivity in the amplitude of radiation - induced magnetoresistance oscillations to the relative orientation of the linear polarization with respect to the hall bar axis .
particularly , more detailed measurement@xcite by rotating , by an angle @xmath0 , the polarization direction of linearly polarized microwaves with respect to the long axis of the hall bar electron devices , showed , at relatively low microwave power , a strong sinusoidal variation in the diagonal resistance @xmath1 vs @xmath0 at the oscillatory extrema . and
, unexpectedly , the angle for the maximal oscillatory @xmath1 response under a given - power linear - polarized microwave , which is not at @xmath2 or @xmath3 , appears to depend upon the radiation frequency , the extremum in question , and the magnetic field orientation .
so far , there has been no theoretical explanation for these interesting observations .
there is an urgent need to analyze the detailed polarization dependence from a theoretical model for radiation - induced magnetoresistance oscillations .
we deal with an isotropic 2d system of short thermalization time , consisting of @xmath4 electrons in a unit area of the @xmath5-@xmath6 plane .
these electrons , scattered by random impurities and by phonons in the lattice , are subjected to a uniform magnetic field @xmath7 in the @xmath8 direction .
when an electromagnetic wave of angular frequency @xmath9 illuminates perpendicularly onto the 2d plane with the incident electric field @xmath10 at @xmath11 and a dc current flows within the plane , the electric field inside the 2d system involves a dc component , @xmath12 , and an ac component @xmath13 the steady - transport state of this electron system can be described by the drift velocity of the electron integrative ( the center of mass ) motion , consisting of a dc part , @xmath14 , and a stationary time - dependent part , @xmath15 in the 2d plane , together with an average temperature @xmath16 , characterizing the isotropic thermal distribution of electrons in the reference frame moving with the center of mass.@xcite they satisfy the following force and energy balance equations:@xcite @xmath17 @xmath18 here , @xmath19 is the time - averaged damping force against the electron drift motion due to impurity scattering , and @xmath20 is the time - averaged rate of the electron energy - absorption from the radiation field . in eqs.([exf0 ] ) and ( [ exsp ] ) , @xmath21 is the effective impurity potential , @xmath22 is the imaginary part of the electron density - correlation function at temperature @xmath16 in the presence of the magnetic field without the electric field , @xmath23 , and @xmath24 is the bessel function of order @xmath25 with argument @xmath26^{\frac{1}{2}}/\omega$ ] .
note that , although contributions of phonon scattering to @xmath27 and @xmath28 are neglected in comparison with those of impurity scattering at the considered low lattice temperature , it provides the main channel for electron energy dissipation to the lattice with a time - averaged energy - loss rate @xmath29 , having an expression as given in ref .. the ac components @xmath30 and @xmath31 of electron drift velocity should be determined selfconsistently in terms of the incident ac field @xmath32 by the electrodynamic equations connecting both sides of the 2d system , taking into account the scattering - related damping forces @xmath33 and @xmath34 .
@xcite however , for high - mobility systems at low temperatures , the effects of these scattering - related damping forces are much weaker in comparison to those of radiative decay @xcite and thus negligible , whence @xmath30 and @xmath31 are in fact directly given from eqs.([2deqv1 ] ) and ( [ 2deqv2 ] ) by the high - frequency electric field @xmath35 inside the 2d electron system . on the other hand , by solving the maxwell equations connecting both sides of the 2d electron gas which is carrying the sheet current density @xmath36 , @xmath35 is determined by the incident fields @xmath37 and @xmath38 based on the setup of the 2d system in the sample substrate.@xcite if the 2d electron gas locates within a thin layer under the surface plane at @xmath11 of a thick ( treated as semi - infinite ) semiconductor substrate having a refractive index , @xmath39 , the ac field @xmath35 driving the 2d electrons , which equals the sum of the incident and the reflected fields at @xmath11 and equals the transmitted field ( the field just passes through the 2d layer ) , can be expressed as @xcite @xmath40 here @xmath41 is the refractive index of the air and @xmath42 and @xmath43 are , respectively , the light speed and the dielectric constant in vacuum .
if the 2d electron gas is contained in a thin layer suspended in vacuum at the plane @xmath11 , then @xmath44 by combining eq.([thick ] ) or eq.([thin ] ) with balance eqs.([2deqv1 ] ) and ( [ 2deqv2 ] ) , the ac drift - velocity components @xmath30 and @xmath31 , and thus the argument @xmath45 , can be determined by the incident field components @xmath37 and @xmath38 . the imaginary part of 2d electron density correlation function in a magnetic field , @xmath46 , can be written in the landau representation as @xcite @xmath47\nonumber\\ & & \,\hspace{2cm}\times\,\,{\rm im}g_n(\varepsilon + { \it \omega})\,{\rm im}g_{n'}(\varepsilon ) , \label{pi_2ll}\end{aligned}\ ] ] where @xmath48 is the magnetic length , @xmath49^{-1}y^l{\rm e}^{-y}[l_n^l(y)]^2 $ ] with @xmath50 being the associate laguerre polynomial , @xmath51 + 1\}^{-1}$ ] is the fermi function at electron temperature @xmath16 , and @xmath52 is the density - of - states ( dos ) function of the broadened landau level @xmath25 . the landau - level broadening , which results from impurity , phonon , and electron - electron scatterings , is assumed to have a gaussian form [ @xmath53 is the center of the @xmath25th landau level , @xmath54 , @xmath55 , \label{gauss}\ ] ] with a @xmath56-dependent half width expressed as @xmath57 where @xmath58 , the single - particle lifetime or quantum scattering time in the zero magnetic field , is assumed to be related to the transport relaxation time @xmath59 or the zero - field linear mobility @xmath60 using an empirical parameter @xmath61 by @xcite @xmath62 for an isotropic system where the frictional force @xmath27 is in the opposite direction of the drift velocity @xmath14 and the magnitudes of both the frictional force and the energy - dissipation rate depend only on @xmath63 , we can write @xmath64 and @xmath65 . in the hall configuration with velocity @xmath14 in the @xmath5 direction @xmath66 or the current density @xmath67 , and @xmath68 , the longitudinal linear ( @xmath69 ) resistivity under the incident radiation of eq.([inci ] ) due to impurity scattering ,
can be written in the form @xmath70 where @xmath71 .
we consider an incident radiation of form ( [ inci ] ) linearly polarized along the direction having an angle @xmath0 with respect to the @xmath5-axis : @xmath72 and @xmath73 .
the 2d electron gas , having carrier sheet density @xmath74m@xmath75 and zero - temperature linear mobility @xmath76m@xmath77/vs from short - range impurity scattering in the absence of the magnetic field , is assumed to locate within a thin layer under the surface plane at @xmath11 of a thick gaas - based substrate with a refractive index of @xmath78 .
, electron temperature @xmath16 and energy absorption @xmath28 are plotted versus @xmath79 under the irradiation of linearly polarized ( lp ) microwaves of frequency @xmath80-ghz and incident power @xmath81w / m@xmath82 at several different polarization directions ( @xmath83 and @xmath84 ) for a system described in the text at temperature @xmath85k.,scaledwidth=38.0% ] induced by linearly polarized microwave of frequency @xmath80ghz and incident power @xmath81w / m@xmath82 at @xmath86 is shown for positive and reverse magnetic fields over the range @xmath87 ( a ) . the @xmath0 dependence of @xmath1 at extrema @xmath88 , @xmath89 , @xmath90 , and @xmath91 ( b ) and at extrema @xmath92 , @xmath93 , @xmath94 , and @xmath95 ( c).,scaledwidth=48.0% ] figure 1 shows the calculated energy absorption rate @xmath28 , the electron temperature @xmath16 and the longitudinal linear resistivity @xmath1 of this system irradiated by linearly polarized ( lp ) incident microwaves having frequency @xmath96ghz and amplitude @xmath97v / cm ( i.e. , incident power @xmath81w / m@xmath82 ) at different polarization directions ( @xmath83 , and @xmath84 ) , together with the dark resistivity , as functions of the normalized inverse magnetic field @xmath79 ( @xmath98 ) .
the lattice temperature is assumed to be @xmath85k and the landau level broadening parameter is taken to be @xmath99 .
we see that the electron energy absorption @xmath28 and thus the electron temperature @xmath16 , both showing a marked main peak at cyclotron resonance @xmath100 and secondary peaks around its harmonics @xmath101 , are essentially the same for all different polarization directions of radiation .
this can be understood directly from expression ( [ exsp ] ) of @xmath28 in the case of relatively low strength of incident microwave when the dominant contribution to it comes from the terms @xmath102 and @xmath103 .
writing explicitly the @xmath0-dependent expression of @xmath104 in the case of incident plane - polarized radiation having an angle @xmath0 with respect to the @xmath5-axis we can see that , after summing over all the directions of @xmath105 , the angle dependence disappears due to @xmath106 .
furthermore , the energy - loss rate @xmath29 has @xmath0-dependent behavior similar to that @xmath28 .
thus the electron temperature @xmath16 , determined by the energy - balance equation ( 7 ) involving only @xmath28 and @xmath29 , has similar @xmath0-dependent behavior .
the situation is different for the frictional force @xmath27 and the resistivity @xmath1 , because of additional @xmath105-direction dependent weighted factors showing up inside the @xmath107 summation in the expressions ( [ exf0 ] ) and ( [ rixx ] ) , leading to sensitive @xmath0-dependence of @xmath1 as seen from the numerical results shown in the upper part of fig.1 .
the linear - polarized microwave excited magnetoresistivity @xmath1 oscillates strongly , having the same oscillatory period and nearly the same phase and , at cyclotron resonance @xmath100 or its harmonics @xmath108 , taking equal values for all different polarization directions @xmath0 .
we can label , for all @xmath0 , the main peaks and valleys of the oscillatory resistivity respectively as @xmath109 , @xmath110 , and @xmath111 , @xmath112 , etc .
the amplitude of resistivity oscillation , however , varies sensitively with changing the polarization direction of the microwave .
the maximal or minimal amplitude of a peak or a valley shows up at different polarization angles @xmath0 for different peaks or valleys .
for instance , the maximal and minimal amplitudes show up , respectively , at @xmath113 and @xmath114 for @xmath109 , at @xmath115 and @xmath116 for @xmath111 , at @xmath117 and @xmath118 for @xmath110 , at @xmath119 and @xmath120 for @xmath112 , and show up at polarization angles closer to @xmath121 and @xmath122 for higher order peaks and valleys . the exact polarization angle for the maximal or minimal amplitude of an extremum appears to depend on the radiation frequency and the 2d electron setup in the substrate . in obtaining the above results the sample setup
is so assumed that 2d electrons locate within a thin layer under the surface plane at @xmath11 of a thick semiconductor substrate having a refractive index of @xmath78 , and the relevant damping in question , i.e. , the radiative decay , is fully determined by this sample setup from the electrodynamic equation ( [ thick ] ) .
except for this , no other damping parameter nor any mechanism capable of producing polarization rotation is introduced in the present model .
next we present the results obtained under magnetic field reversal .
figure 2(a ) plots the longitudinal resistivity @xmath1 vs @xmath123 over the range @xmath87 for the system irradiated by a linearly polarized microwave along the @xmath86 direction with frequency @xmath96ghz and incident power @xmath81w / m@xmath82 .
the extrema of interest here are labeled as @xmath88 , @xmath89 , @xmath90 , and @xmath91 for those in the domain @xmath124 , and @xmath92 , @xmath93 , @xmath94 , and @xmath95 for those in the domain @xmath125 .
the values of longitudinal resistivity @xmath1 at @xmath88 , @xmath89 , @xmath90 , and @xmath91 are plotted in fig.2(b ) , and those at @xmath92 , @xmath93 , @xmath94 , and @xmath95 qre plotted in fig.2(c ) , as functions of the microwave polarization direction @xmath0 .
the present treatment is for an isotropic system . by symmetry
we always have @xmath126 , @xmath127 , @xmath128 , etc .
if the maximal amplitude of an extremum , e.g. , @xmath88 , shows up at @xmath129 , the maximal amplitude of the corresponding extremum at the reverse magnetic field , @xmath92 , must be at a different polarization angle .
the effect of asymmetry in a real sample itself would produce further complexity .
we have examined the effect of changing the polarization angle @xmath0 of the incident linearly - polarized microwaves with respect to the dc current on radiation - induced magnetoresistance oscillations within the balance - equation formulation of the photon - assisted magnetotransport model .
the present investigation takes the incident microwave field , rather than the ac field inside the 2d electron system , as the input quantity , allowing a direct determination of the dominant damping mechanism in the high - mobility system , the radiative decay , from the experimental sample setup without introducing any artificial damping parameter , thus enabling a direct comparison with experimental measurement .
we find that the amplitude of radiation - induced magnetoresistance oscillation varies sensitively with changing @xmath0 . at an extremum the amplitude of oscillatory magnetoresistance @xmath1 exhibits a sinusoidal , up to a factor of 5 , magnitude variation with rotating the polarization angle @xmath0 .
the maximal amplitude shows up generally at a nonzero @xmath0 , which is dependent upon the extremum in question , the 2d electron setup , the radiation frequency and the magnetic field orientation .
i. a. dmitriev , a. d. mirlin , and d. g. polyakov , phys .
91 * , 226802 ( 2003 ) ; * 99 * , 206805 ( 2007 ) ; phys .
b * 70 * , 165305 ( 2004 ) ; * 75 * , 245320 ( 2007 ) ; i. a. dmitriev , m. g. vavilov , i. l. aleiner , a. d. mirlin , and d. g. polyakov , _ ibid .
_ * 71 * , 115316 ( 2005 ) .
j. dietel , l. i. glazman , f. w. j. hekking , and f. von oppen , phys .
b * 71 * , 045329 ( 2005 ) ; c. joas , j. dietel , and f. von oppen , _ ibid . _ * 72 * , 165323 ( 2005 ) ; j. dietel , _ ibid . _ * 73 * , 125350 ( 2006 ) . | we examine the effect of changing the linear polarization angle @xmath0 of incident microwaves with respect to the dc current on radiation - induced magnetoresistance oscillations in a two - dimensional ( 2d ) system within the balance - equation formulation of the photon - assisted magnetotransport model , considering the radiative decay as the sole damping mechanism . at an extremum the amplitude of oscillatory magnetoresistance @xmath1 exhibits a sinusoidal , up to a factor of 5 , magnitude variation with rotating the polarization angle @xmath0 .
the maximal amplitude shows up generally at a nonzero @xmath0 , which is dependent upon the extremum in question , the 2d electron setup , the radiation frequency and the magnetic field orientation .
these results provide a natural explanation for the experimental observations by mani _
et al . _
[ phys .
rev .
b * 84 * , 085308 ( 2011 ) ] , and ramanayaka _ et al . _
[ phys .
rev .
b * 85 * , 205315 ( 2012 ) ] . |
bilateral dyck paths are directed walks on @xmath2 starting at @xmath3 in the @xmath4-plane and ending on the line @xmath5 , which have steps in the @xmath6 ( up - step ) and @xmath7 ( down - step ) directions .
let @xmath8 be the set of all bilateral dyck paths .
note that in the literature these are also referred to as free dyck paths @xcite or grand - dyck paths @xcite .
dyck paths are paths in @xmath8 which have no vertices with negative @xmath9-coordinates .
let @xmath10 be the set of all dyck paths .
we define a negative dyck path to be a bilateral dyck path of nonzero length which has no vertices with positive @xmath9-coordinates .
given @xmath11 , we define the semilength @xmath12 to be half the number of its steps .
we say that an up - step is at height @xmath13 if it starts at a vertex @xmath14 and ends at a vertex @xmath15 ; it is at odd height if and only if @xmath13 is odd .
a down - step is at height @xmath13 if it starts at a vertex @xmath15 and ends at a vertex @xmath16 ; it is at odd height if and only if @xmath13 is odd .
therefore the first and last steps in any dyck path are at height @xmath17 , and an up - step and its matching down - step have the same height .
this definition is consistent with that in @xcite .
we define a peak as an up - step followed immediately by a down - step , and the height of a peak as the height of the steps which form it , or equivalently as the height of the vertex common to both steps of the peak .
we define a valley to be a down - step followed immediately by an up - step .
we define a contact as a down - step at height @xmath17 or an up - step at height @xmath18 .
we define a crossing to be either a down - step at height @xmath17 followed immediately by a down - step at height @xmath18 , or an up - step at height @xmath18 followed immediately by an up - step at height @xmath17 .
we define a prime dyck path to be a dyck path with exactly one contact .
let @xmath19 be the set of words with the symbol set @xmath20 . a bilateral dyck word of semilength @xmath21 is an element of @xmath19 such that the letters @xmath22 and @xmath23 each appear @xmath21 times . a dyck word of semilength @xmath21 is a bilateral dyck word such that no initial segment of the string has more @xmath23s than @xmath22s
there is an obvious bijection between paths and words , mapping up - steps to @xmath22s and down - steps to @xmath23s , and we will use words and paths interchangeably in this paper .
for example , we say that a dyck word is prime if it corresponds to a prime dyck path .
it is an established fact that many patterns in dyck paths are enumerated by narayana numbers @xmath24 for example , there are @xmath25 dyck paths of semilength @xmath21 with precisely @xmath1 peaks , or @xmath1 up - steps at odd height , or @xmath26 up - steps at even height , and many other statistics @xcite
. these are most easily proven by generating function techniques @xcite .
equinumeracy then follows from showing that different counting problems have the same generating function . in this note
we give a direct bijection between dyck paths of semilength @xmath21 with @xmath1 up - steps at odd height and dyck paths of semilength @xmath21 with @xmath1 peaks .
clearly such a bijection provides more structural insight than a proof using generating function techniques . in @xcite
a bijection between dyck paths of semilength @xmath21 with @xmath26 up - steps at even height and dyck paths of semilength @xmath21 with @xmath1 peaks is given , using the machinery of checkmark sequences . that bijection is given as a restriction of a bijection between bilateral dyck paths .
to make a connection of our result with the work in @xcite , we remark that any dyck path with @xmath1 up - steps at odd height can be mapped bijectively to a dyck path with @xmath26 up - steps at even height as follows .
if @xmath27 is a dyck path of nonzero length , the corresponding dyck word @xmath28 can be uniquely decomposed into a word of the form @xmath29 where @xmath30 and @xmath31 are dyck words .
exchanging @xmath30 and @xmath31 changes the parity of the steps of the sub - paths corresponding to @xmath30 and @xmath31 , and therefore the involution mapping @xmath29 to @xmath32 gives the desired bijection .
we will define this more carefully in section [ extension ] . by bijecting dyck paths of semilength @xmath21 with @xmath26 up - steps at even height to dyck paths with @xmath1 up - steps at odd height using the involution
just described , followed by application of the map @xmath33 given in the theorem below , we also have a bijection of dyck paths of semilength @xmath21 with @xmath26 up - steps at even height to dyck paths of semilength @xmath21 with @xmath1 peaks .
already from inspecting the five dyck paths of semilength @xmath34 one can see that this bijection is different from the one presented in @xcite .
we note that the extension of the bijection in @xcite to bilateral dyck paths needs a slightly modified definition of peaks .
more precisely , in @xcite an initial down - step or final up - step in a bilateral dyck path is also counted as a peak , and the bijection given is between bilateral dyck paths of semilength @xmath21 with @xmath26 up - steps at even height and bilateral dyck paths of semilength @xmath21 with @xmath1 ( modified ) peaks .
in contrast , our bijection extends to a bijection between bilateral dyck paths of semilength @xmath21 with @xmath1 up - steps at odd height and bilateral dyck paths of semilength @xmath21 with @xmath1 peaks .
note that the restriction of our bijection to dyck paths is essentially identical to a bijection given in @xcite .
there , the authors used two length - preserving bijections on dyck paths to show the equidistribution of statistics of certain strings occurring at odd height , at even height , and anywhere .
however , the problem discussed here , regarding the equidistribution of up - steps at odd height and peaks anywhere , was not addressed in @xcite .
if @xmath27 is a dyck path of nonzero length with @xmath35 down steps at height @xmath36 before the first contact , the corresponding dyck word @xmath28 can be uniquely decomposed into a word of the form @xmath37 where each @xmath38 is a dyck word . using this decomposition
, we define the map @xmath39 recursively by setting @xmath40 and @xmath41 , where @xmath42 denotes the empty word . as the decomposition expresses a dyck word in terms of strictly smaller dyck words , the recursion terminates .
therefore @xmath33 is well - defined on all dyck words .
[ theorem1 ] @xmath33 gives an explicit bijection from the set of dyck paths of semilength @xmath21 with @xmath43 contacts and @xmath1 up - steps at odd height to the set of dyck paths of semilength @xmath21 with @xmath43 contacts and @xmath1 peaks .
the decomposition implies that the parity of a step in the path corresponding to the sub - word @xmath38 is the same as its parity in the original path @xmath27 .
as the first step of each sub - path corresponding to nonempty @xmath38 is at odd height , each recursive iteration either maps the empty word to itself or maps exactly one up - step at odd height ( the first step of the path @xmath27 ) to a peak .
thus @xmath33 maps a path with @xmath1 up - steps at odd height to a path with @xmath1 peaks .
clearly , @xmath33 does not change the number of contacts .
we now show that @xmath33 is indeed a bijection by giving its inverse .
if @xmath27 is a dyck path with right - most peak at height @xmath44 with @xmath45 , the corresponding dyck word @xmath28 can be uniquely decomposed into a word of the form @xmath46 where each @xmath38 is a dyck word .
let @xmath47 be defined recursively as @xmath48 with @xmath41 , where @xmath42 denotes the empty word .
we note in passing that each recursive iteration of @xmath49 to a nonempty path maps exactly one peak to an up - step at odd height . to show that @xmath49 is the inverse of @xmath33 , we proceed by induction on @xmath21 , the semilength of word . for @xmath50
, @xmath51 now take a dyck word @xmath52 of semilength @xmath53 and assume that @xmath54 for all words @xmath55 of semilength less than @xmath21 .
@xmath52 can be decomposed as either @xmath56 or @xmath57 , where the @xmath38 are dyck words of semilength strictly less than @xmath21 . hence ,
by the inductive hypothesis , @xmath58 the number of dyck paths of semilength @xmath21 with @xmath43 contacts and @xmath1 up - steps at odd height is equal to the number of dyck paths of semilength @xmath21 with @xmath43 contacts and @xmath1 peaks .
mapping a dyck path @xmath52 ( top ) with @xmath59 steps , @xmath60 up - steps at odd height and @xmath17 contact to a dyck path @xmath61 ( bottom ) with @xmath59 steps , @xmath60 peaks and @xmath17 contact , and @xmath49 performing the inverse mapping .
intermediate stages after each recursive iteration of @xmath33 and @xmath49 are shown on the left and right , respectively .
black dots correspond to the occurrence of empty words in the decomposition of the corresponding words.,scaledwidth=90.0% ] an example of the bijection @xmath33 and its inverse @xmath49 is given in figure [ fig ] , explicitly showing the intermediate stages after each recursive iteration .
when writing the corresponding words below , we insert bracketing corresponding to the decompositions , e.g. we write @xmath62 note in particular that the double bracket @xmath63 signifies an empty word in the decomposition . for the sake of legibility
, we have avoided the occurrence of composite words in this example .
the mapping @xmath33 corresponding to the left arrows in figure [ fig ] is now performed as follows .
@xmath64 the first recursive iteration of @xmath33 gives @xmath65 and the second step applies the recursion to the sub - words @xmath30 and @xmath31 respectively . in the third step the recursion is applied to empty subwords , and hence the path no longer changes ; the corresponding figures look identical .
note in particular that while @xmath66 does not change the word @xmath67 , the position of the empty word @xmath63 , and correspondingly the position of the black dot in figure [ fig ] changes .
the inverse mapping @xmath49 corresponding to the right arrows in figure [ fig ] is performed similarly , with different intermediate stages .
we now give an extension of @xmath33 to bilateral dyck paths . before we do this
, we must define the following two maps .
we define the map @xmath69 as follows . if @xmath27 is a bilateral dyck path of semilength @xmath21 with corresponding bilateral dyck word @xmath70 , where @xmath71 , then ( i ) @xmath72 , ( ii ) @xmath73 and ( iii ) @xmath74 .
the map @xmath75 reflects bilateral dyck paths in the @xmath76axis .
in particular , it maps nonempty dyck paths to negative dyck paths and vice versa .
it maps steps at odd height to steps at even height , and vice versa .
it also maps peaks to valleys , and maps valleys to peaks .
we define @xmath77 as follows .
if @xmath27 is a dyck path of nonzero length , the corresponding dyck word @xmath28 can be uniquely decomposed into a word of the form @xmath29 where @xmath30 and @xmath31 are dyck words .
then @xmath78 the map @xmath79 is the one mentioned in section [ introduction ] , which illustrates the equinumeracy between paths with @xmath80 up - steps at odd height and @xmath1 up - steps at even height ; excepting the first up- and down - steps at height @xmath17 , @xmath79 maps steps at odd height to steps at even height , and vice versa .
it preserves the number of peaks or valleys , but does not preserve the number of contacts .
we note that both @xmath75 and @xmath79 are involutions .
we are now in a position to state the extended bijection .
let @xmath42 denote the empty word .
we define @xmath81 in the following way : * if @xmath27 is a dyck path with corresponding dyck word @xmath28 , then + @xmath82 * if @xmath27 is a negative dyck path with corresponding negative dyck word @xmath28 , then + @xmath83 * if @xmath27 is a bilateral dyck path with @xmath84 crossings then we can uniquely decompose the corresponding bilateral dyck word @xmath28 into @xmath85 , where each @xmath38 is alternately either a dyck word of nonzero length or a negative dyck word .
then + @xmath86 @xmath87 gives an explicit bijection from the set of bilateral dyck paths of semilength @xmath21 with @xmath1 up - steps at odd height to the set of bilateral dyck paths of semilength @xmath21 with @xmath1 peaks . by theorem [ theorem1 ] , ( a ) maps dyck paths with @xmath1 up - steps at odd height to dyck paths with @xmath1 peaks .
if @xmath27 is a negative dyck path with @xmath1 up - steps at odd height and corresponding negative dyck word @xmath28 , then @xmath88 corresponds to a dyck path with @xmath1 up - steps at even height .
applying @xmath79 to this new path gives a dyck path with @xmath80 up - steps at odd height .
then @xmath89 corresponds to a dyck path with @xmath80 peaks ( and so @xmath1 valleys ) .
finally , applying @xmath75 again will give a negative dyck path with @xmath1 peaks .
thus ( b ) maps negative dyck paths with @xmath1 up - steps at odd height to negative dyck paths with @xmath1 peaks .
application of ( c ) does not alter the parity of steps or the number of peaks .
thus @xmath87 maps bilateral dyck paths with @xmath1 up - steps at odd height to bilateral dyck paths with @xmath1 peaks .
we now show that @xmath87 is a bijection by giving its inverse .
let @xmath90 , be defined as follows .
* if @xmath27 is a dyck path with corresponding dyck word @xmath28 , then + @xmath91 * if @xmath27 is a negative dyck path with corresponding negative dyck word @xmath28 , then + @xmath92 * if @xmath27 is a bilateral dyck path with @xmath84 crossings then we can uniquely decompose the corresponding bilateral dyck word @xmath28 into @xmath85 , where each @xmath38 is alternately either a dyck word of nonzero length or a negative dyck word .
then + @xmath93 to show that @xmath94 is the inverse of @xmath87 , we proceed as follows .
let @xmath52 be a bilateral dyck word .
if the path associated to @xmath52 has @xmath84 crossings then we can uniquely decompose @xmath52 into @xmath85 , where each @xmath38 is alternately either a dyck word of nonzero length or a negative dyck word
. then @xmath95 if @xmath52 is a dyck word then @xmath96 and @xmath97 so @xmath98 using the fact that @xmath33 and @xmath49 are inverses of each other , together with the involutive properties of @xmath75 and @xmath79 , it is easy to show that for negative dyck words @xmath99 also holds : @xmath100 the number of bilateral dyck paths of semilength @xmath21 with @xmath1 up - steps at odd height is equal to the number of bilateral dyck paths of semilength @xmath21 with @xmath1 peaks .
the authors wish to thank judy - anne osborn for helpful comments on an early version of the manuscript , and the referees for their careful comments on our work . | it is known that both the number of dyck paths with @xmath0 steps and @xmath1 peaks , and the number of dyck paths with @xmath0 steps and @xmath1 steps at odd height follow the narayana distribution . in this paper
we present a bijection which explicitly illustrates this equinumeracy .
moreover , we extend this bijection to bilateral dyck paths .
the restriction to dyck paths preserves the number of contacts .
* keywords : * dyck path ; bilateral dyck path ; free dyck path ; grand - dyck path ; narayana numbers ; bijection |
following hertz@xcite , we describe the critical behavior of an antiferromagnetic metal entirely in terms of an effective ginzburg - landau - wilson theory of an order parameter field @xmath29 which represents the fluctuating ( staggered ) magnetization of the system . in the absence of a magnetic field ,
the quadratic part of the action takes the form@xcite ( assuming negligible spin - orbit coupling ) @xmath30 = \frac 1 \b \int { \frac{d^3 k}{(2\pi)^3 } } \sum_n \phi^*(r + k^2 + |\w_n| ) \phi,\ ] ] where @xmath31 measures the distance from the quantum critical point and momenta @xmath32 are given relative to the antiferromagnetic ordering wave vector @xmath23 .
the @xmath33 term arises from the ( landau- ) damping of the spin - fluctuations by gapless fermionic excitations in the vicinity of points on the fermi surface that are connected by @xmath23 ( assuming @xmath24 ) .
how will this effective action change in the presence of a magnetic field ?
first , @xmath34 will acquire a magnetic field dependence , for example , @xmath31 will grow for larger fields in systems where antiferromagnetism is suppressed by @xmath3 .
second , the magnetic field breaks the rotational invariance and components of @xmath16 parallel and perpendicular to @xmath35 will have different masses , @xmath36 and @xmath37 , respectively .
third , as argued above , the magnetization will precess around @xmath35 ; this is described by an extra term ( in coordinate and time space for convenience ) @xmath38 & = & \int \limits_0^\b \ !
d\tau \int \ !
d{\bf r } \ , { { \bf \boldsymbol b } } \cdot \i ( { { \bf \boldsymbol \phi } } \times \partial_t { { \bf \boldsymbol \phi } } ) \\ & = & \int\!\ !
\int b ( \i \phi_x \partial_t \phi_y - \i \phi_y \partial_t \phi_x ) = \int \!\
! \int b \tilde\phi_\perp^ * \partial_t \tilde\phi_\perp \nonumber\end{aligned}\ ] ] in the effective action , where @xmath39 is parallel to @xmath35 ( taken to point into the @xmath40-direction ) and we have introduced the complex field @xmath41 .
note that ( [ precession1 ] ) breaks time - reversal invariance .
therefore such a term is absent for @xmath42 .
above , we deduced the form of the effective action on phenomenological grounds but it can also be derived from a more explicit calculation starting from a hubbard - type model of electrons moving in the presence of a magnetic field , @xmath43 . here , the magnetic field enters only via a zeeman - term ; we do not take orbital effects into account , assuming that landau levels are broadened by disorder or thermal effects .
note that in the experimentally most relevant heavy fermion system , orbital effects are strongly suppressed compared to contributions from the zeeman term as the effective masses and magnetic susceptibilities are very large in those systems@xcite . for simplicity , we assume that the antiferromagnet is commensurate ( incommensurate antiferromagnets show the same qualitative behavior for all quantities discussed below ) and introduce a real order parameter vector @xmath44 as a hubbard - stratonovich field which decouples the spin - density part of the interaction .
following hertz , the electrons are now integrated out to obtain an effective action for the order parameter , generating a priori infinitely many interaction terms .
we truncate the effective action , retaining the leading frequency and momentum dependence of the gaussian part of the action as well as a constant @xmath45 interaction term , since all other terms are irrelevant in the renormalization group sense@xcite ( cubic terms are discussed in appendix [ cubic ] ) .
for the quadratic part one obtains @xmath46 where @xmath47 is the interaction in the spin - spin channel and @xmath48 the susceptibility in the presence of the finite field @xmath35 evaluated at @xmath49 . calculating these susceptibilities on the paramagnetic side of the transition
we obtain @xmath50 + s_4[{{\bf \boldsymbol \phi } } ] , \\\nonumber s_2[{{\bf \boldsymbol \phi } } ] & = \frac 1 \b \int { \frac{d^3 k}{(2\pi)^3 } } \sum\limits_n { { \bf \boldsymbol \phi}}_{\w_n,{{\bf \boldsymbol
k}}}^t \begin{pmatrix } r_\perp + |\w_n| \cos \th + k^2 & \w_n \sin \th & 0 \\ - \w_n \sin \th & r_\perp + |\w_n| \cos \th + k^2 & 0 \\ 0 & 0 & r_z + |\w_n| + k^2 \end{pmatrix } { { \bf \boldsymbol \phi}}_{\w_{-n},-{{\bf \boldsymbol k } } } \\ s_4[{{\bf \boldsymbol \phi } } ] & = \frac { g}{\b^4 } \int { \frac{d^3 k_1}{(2\pi)^3 } } \ldots { \frac{d^3 k_4}{(2\pi)^3 } } \sum\limits_{n_1 \ldots n_4 } \d({{\bf \boldsymbol k}}_1 + { { \bf \boldsymbol k}}_2 + { { \bf \boldsymbol k}}_3 + { { \bf \boldsymbol k}}_4)\d_{n_1 + n_2 + n_3 + n_4}({{\bf \boldsymbol \phi}}_{\w_{n_1},{{\bf \boldsymbol
k}}_1 } \cdot { { \bf \boldsymbol \phi}}_{\w_{n_2},{{\bf \boldsymbol k}}_2})({{\bf \boldsymbol \phi}}_{\w_{n_3},{{\bf
\boldsymbol k}}_3 } \cdot { { \bf \boldsymbol \phi}}_{\w_{n_4},{{\bf \boldsymbol k}}_4}).\end{aligned}\ ] ] here @xmath51 and @xmath52 is a matsubara frequency and @xmath32 is measured again from the ordering wave vector .
the coefficients of @xmath53 and @xmath54 have been made unity by an appropriate choice of the bare length scale @xmath55 and temperature / energy scale @xmath56 .
in general the prefactors of the @xmath57 and @xmath58 terms for @xmath59 and @xmath60 will be different ( even after rescaling ) , we suppress these prefactors to keep the notations simple as they will not lead to any qualitative changes in the results .
it is , however , essential to keep track of the dynamics of @xmath60 , i.e. of the ratio of precession and damping terms parametrized by an angle @xmath61 . for small @xmath61 the dynamics
is overdamped , while for @xmath62 precession dominates .
the value of @xmath61 depends on details of the band - structure and the size of the magnetic field with @xmath63 for small magnetic fields . as anticipated in ( [ precession1 ] ) ,
the @xmath64 and @xmath65direction are coupled for @xmath66 .
the gaussian part of the action can be diagonalized by introducing the complex field @xmath67 as above , and we obtain @xmath68 & = \
! \int \ ! { \frac{d^3 k}{(2\pi)^3 } } \ ! \frac 1\b \sum\limits_n { \phi_{\w_n,{{\bf \boldsymbol k}}}^\perp}^*(2 \chi_{{{\bf \boldsymbol k}}}(\i\w_n)^{-1 } ) \phi_{\w_n,{{\bf \boldsymbol k}}}^\perp \nonumber \\ & + \phi^z_{\w_n,{{\bf \boldsymbol
k}}}(k^2 + r_z + |\w_n| ) \phi^z_{-\w_n,-{{\bf \boldsymbol k } } } , \end{aligned}\ ] ] where @xmath69 is the propagator of @xmath70 .
as expected from the symmetry arguments given above , @xmath37 and @xmath36 turn out to be different with @xmath71 and @xmath72 for small @xmath3 .
as @xmath73 increases for increasing fields ( @xmath74 for small @xmath3 ) , an antiferromagnetic system sufficiently close to its quantum critical point can be tuned to the paramagnetic phase by applying a magnetic field ( assuming that no first order transition is induced ) . when discussing the behavior close to the quantum critical point , it is important to note that the magnetic field enters into the calculations both by the @xmath3-dependence of @xmath31 as well as through the @xmath3-dependent angle @xmath61 .
close to the quantum critical point tuned by a _ finite _ magnetic field @xmath75 ,
@xmath76 can be approximated by a constant ( as checked below ) while it is obviously essential to keep track of the leading @xmath3 dependence of the control parameter @xmath77 . at this point , it is worthwhile to take a closer look at @xmath78 $ ] in coordinate and time space for @xmath79 , i.e. if landau damping is absent as it is the case in an insulator like tlcucl@xmath13 @xcite or in a metal with @xmath80 ( see introduction ) .
the gaussian part of @xmath78 $ ] is minimized for a field @xmath70 that obeys the equation @xmath81 this has the form of a schrdinger equation for a particle in a constant potential given by @xmath82 . if one adds the interactions one obtains a non - linear schrdinger equation or gross - pitaevskii equation which describes the physics of weakly interacting bosons . in this interpretation
, @xmath31 takes over the role of the chemical potential .
the quantum critical point of a field tuned insulating antiferromagnet ( @xmath83 ) is therefore in the same universality class as the quantum phase transition of a dilute gas of bosons driven by a chemical potential .
the non - magnetic phase ( @xmath84 ) corresponds to a phase with _ negative _ chemical potential where no bosons are present in the @xmath85 limit , while the bose - einstein condensed phase corresponds to the magnetically ordered phase .
the physical properties of the effective action ( [ effact ] ) can be analyzed with the help of renormalization group equations .
as a first step it is useful to perform a simple scaling analysis of @xmath86 $ ] .
when momenta , frequencies and fields are rescaled as @xmath87 , @xmath88 , where @xmath27 is the dynamical critical exponent , and @xmath89 , @xmath86 $ ] remains invariant under scaling provided that @xmath22 .
the masses @xmath90 and the dimensionless coupling constant @xmath91 have the scaling dimensions 2 and @xmath92 , respectively . in an antiferromagnetic metal , damping as well as precession are linear in frequency and the terms therefore behave in the same way under scaling . in the renormalization group terminology
this implies that the precession term is an `` exactly marginal '' perturbation with respect to hertz fix - point ( @xmath93 ) which can be expected to modify the behavior of the system at the quantum critical point .
the renormalization group equations for the parameters @xmath94 , @xmath31 , and @xmath95 with corrections to scaling can be obtained by closely following the procedure employed by millis@xcite : we introduce a uv - cutoff in the linked cluster expansion of the free energy and express changes of that cutoff in terms of changes of the parameters of the model .
the rg equations are as follows : @xmath96 where @xmath97 is the running temperature and the expressions for @xmath98 as well as details of the calculation can be found in appendix [ derivation ] .
since the scaling dimension for @xmath95 is negative for an antiferromagnetic system in 3 spatial dimensions , we only consider contributions up to and including first order in @xmath95 . to this order ,
the scaling law for @xmath95 remains unmodified , and @xmath61 remains unrenormalized .
the parameter @xmath61 obtains , however , finite corrections by higher order contributions . equations
( [ trg1 ] ) and ( [ urg1 ] ) are solved trivially . as @xmath99
, @xmath59 remains massive at the quantum critical point ( @xmath100 for @xmath101 ) . in the following
we will concentrate on the regime @xmath102 , where the influence of the parallel mode @xmath59 can be absorbed in a redefinition of the bare @xmath37 .
( [ rperprg1 ] ) can be solved for low temperatures in the limits @xmath103 and @xmath104 , corresponding to quantum critical and ( renormalized ) fermi liquid regime , respectively ( see fig . [ phasedia ] ) .
this provides us with an expression for the correlation length @xmath105 .
we refer to appendix [ derivation ] for details of the calculation . in the quantum critical regime @xmath106
is given by @xmath107 and in the fermi liquid regime it has the form @xmath108 for all @xmath109 one obtains the same qualitative behavior as in the case of vanishing external magnetic field @xcite . only in the fermi liquid regime for @xmath83 ,
the @xmath110 correction is suppressed as landau damping is absent in this limit and our model is characterized by an energy gap which leads to an exponential dependence @xmath111 of the correlation length .
in this section we calculate the specific heat @xmath112 , the temperature dependence of the magnetization , the magnetocaloric effect @xmath113 , and the susceptibility .
the free energy can be calculated directly from rg equations following again ref .
however , as the quartic coupling @xmath95 is irrelevant , the leading behavior in the paramagnetic phase can equivalently be extracted just from the gaussian free energy @xmath114 \times \nonumber \\ & \quad \arctan \left(\frac{2 ( r+k^2 ) \w \cos \th}{(r+k^2)^2 - \w^2}\right ) \label{free}\end{aligned}\ ] ] measured in units of @xmath115 , and we have set @xmath116 . in eq .
( [ free ] ) and in the results shown below we ignore contributions from the massive , non - critical mode @xmath59 characterized by a finite mass @xmath36 . to leading order , the corresponding ( analytic ) corrections to the free energy and its derivative are just additive and
can be obtained by replacing @xmath31 by @xmath36 , by setting @xmath117 and by dividing the result by a factor @xmath118 ( as there are two modes perpendicular to @xmath3 ) in all formulas for thermodynamic quantities given below .
we first consider the specific heat coefficient @xmath119 .
more precisely , we calculate @xmath120 : @xmath121,\ ] ] this differs from the physical specific heat by a @xmath122independent ( but cutoff - dependent ) constant @xmath123 . , where a non - critical contribution has been subtracted , @xmath124 .
the function is not completely universal but depends on the parameter @xmath61 .
a crossover from @xmath125 for @xmath126 to @xmath127 for @xmath128 can be observed , where the signs depend on the value of @xmath61 .
@xmath129 shows a maximum for @xmath130 as can be seen more clearly in fig .
[ gamma1 ] . [ m ] ] and different values of @xmath31 .
note that the total specific heat coefficient @xmath131 is always positive .
the maximum for @xmath84 is characteristic for systems with @xmath130 .
[ gamma1 ] ] the integrals can be evaluated exactly in certain limits . for @xmath132 and low
@xmath133 the specific heat shows thermally activated behavior @xmath134,\end{aligned}\ ] ] as can be expected from a system with a gapped spectrum . for @xmath135 and @xmath133 ,
i.e. in the quantum critical regime and fermi liquid regime , respectively , we obtain for @xmath136 @xmath137 for @xmath138 , this reproduces well - known results @xcite ( correcting some factors of @xmath118 ) , and as expected from scaling , exponents do not change in the presence of the precession term . however , not only the size of the prefactors but interestingly also their sign changes when the dynamics begins to be dominated by precession rather than damping . in the quantum - critical regime the @xmath0 correction is _ negative _ for @xmath139 and positive for @xmath140 .
also in the fermi liquid regime a sign change can be observed in the @xmath141 contribution at @xmath142 . in fig .
[ m ] we show the scaling function @xmath143 obtained from a numerical integration of ( [ tildegamma ] ) . due to the presence of an exactly marginal perturbation ,
the scaling function is _ not _ completely universal but depends on the parameter @xmath61 . in an intermediate regime , @xmath144 , @xmath145 ( and the universal scaling function @xmath146 )
shows a characteristic maximum as a function of temperature as can be read off from the asymptotical results ( [ hight ] ) and ( [ lowt ] ) .
this maximum can not be seen directly at the quantum critical point ( @xmath147 ) but for any finite @xmath84 as long as the critical corrections to the specific heat dominate the non - critical ones . as was argued in ref .
[ ] , the specific heat is not the most sensitive thermodynamic quantity close to a quantum critical point as it tracks only variations of the free energy with respect to temperature ( vertical axis in fig .
[ phasedia ] ) but _ not _ with respect to the control parameter @xmath3 ( horizontal axis ) .
it is therefore interesting to study also the magnetization @xmath148 , the susceptibility @xmath149 , and the @xmath94-derivative of @xmath150 , @xmath151 .
this mixed derivative has the advantage that opposite to specific heat coefficient and susceptibility it vanishes in the @xmath85 limit due to the second law of thermodynamics .
therefore it is not necessary to subtract any constant non - critical contributions when measuring @xmath152 .
very interesting are also ratios of the thermodynamic derivatives@xcite . for a field
tuned critical point , one interesting combination is @xmath153 which describes the magnetocaloric effect , i.e. the temperature change in the sample after an adiabatic change of the magnetic field . for pressure - tuned quantum phase transitions , where @xmath154 is replaced by @xmath155 , the quantities related to @xmath156 , @xmath157 and @xmath158 are the thermal expansion , the compressibility ( and therefore also to the sound velocity ) , and the grneisen parameter@xcite . scaling function @xmath159 .
while for the specific heat shown in fig.[gamma1 ] it was necessary to subtract a non - critical constant contribution such a background does not exist for @xmath160 . ]
@xmath160 as a function of temperature for @xmath161 and different values of @xmath31 . at the quantum critical point
we obtain @xmath162 while @xmath163 for @xmath126 . ]
as both control parameter @xmath31 and @xmath61 depend on @xmath3 one can expect two independent critical contributions to @xmath156 : @xmath164 in the limits @xmath165 we obtain @xmath166 as @xmath31 is a _ relevant _ perturbation at the quantum critical point while @xmath61 is only _ marginal _ , the contributions due to the @xmath3-dependence of @xmath61 are subleading and can therefore be neglected .
with @xmath167 , we therefore find @xmath168 neither @xmath169 nor @xmath170 changes sign as a function of @xmath61 , and indeed , @xmath171 is a monotonous function of @xmath94 for all values of @xmath61 ( see fig . [ ms ] ) . for @xmath79 and at low @xmath94
, the temperature derivative of the magnetization also shows thermally activated behavior @xmath172 \frac{\partial r}{\partial b}.\ ] ] finally , we evaluate the magnetocaloric effect @xmath173 which is given by : @xmath174 in the limits @xmath165 . due to the non - critical contribution @xmath175
, the result for @xmath85 is _ not _ fully universal@xcite .
since @xmath176 and @xmath94 have the same scaling exponents , one might expect that the susceptibility @xmath177 and the specific heat coefficient @xmath178 show very similar behavior .
this , however , turns out to be _ not _ correct .
the technical reason for this is that the susceptibility in the quantum critical regime is a _ singular _ function of the ( dangerously irrelevant ) spin - spin interaction @xmath95 .
practically , this implies that a measurement of the susceptibility is complementary to other thermodynamic measurements as it is highly sensitive to a quantity which can otherwise be determined only by neutron scattering measurements of the correlation length .
the susceptibility @xmath179 gets contributions both from the @xmath3-field dependence of @xmath180 and of @xmath31 .
we only consider the leading corrections due to @xmath31 ( see discussion above ) and evaluate the quantity @xmath181 with @xmath182.\end{aligned}\ ] ] most interesting is the quantum critical regime , where the leading correction to the susceptibility takes the form @xmath183 note that this expression formally _ diverges _ for @xmath184 .
this implies that we have to take into account the interaction effects discussed in section [ rgeq ] and we have to replace the control parameter @xmath31 by @xmath185 given by eq .
( [ rquantcrit ] ) .
one therefore finds @xmath186 but @xmath187 scaling is violated , i.e. the susceptibility is no longer of the form @xmath188 , as the dangerously irrelevant coupling @xmath95 determines @xmath189 in this regime .
it is interesting to trace back the origin of the @xmath190 contribution in ( [ chiquantcrit ] ) .
it arises from the @xmath191 mode of the gaussian theory ( [ s2 ] ) which leads to a contribution of the form @xmath192 to the susceptibility .
note that the static @xmath193 contribution does not depend on the dynamics , i.e. it does not depend on @xmath180 .
however , the correlation length @xmath194 given in eq .
( [ rquantcrit ] ) does depend smoothly on @xmath180 which leads to a slight @xmath180-dependence of @xmath189 . in the fermi liquid regime ,
the susceptibility is given by @xmath195 while for @xmath79 and low @xmath126 it shows thermally activated behavior @xmath196\right ) \left(\frac{\partial r}{\partial b}\right)^2.\ ] ] the rapid crossovers between the @xmath110 , @xmath94 and @xmath197 regimes are shown in fig .
[ suscsmallangleplot ] . , @xmath198 , and different values of @xmath31 .
curves for other values of @xmath180 look essentially identical and are not shown . note that the susceptibility is _ much more _ sensitive to small deviations from the quantum critical point than the specific heat coefficient or @xmath156 ( c.f . fig .
[ gamma1 ] , [ ms2 ] , where much larger values for @xmath31 have been used ) .
the @xmath197 cusp at the quantum critical point [ eq . [ chiquantcrit2 ] ] is rapidly washed out by tiny deviations from the critical magnetic field and replaced by the linear dependence of eq .
( [ chiquantcrit3 ] ) .
[ suscsmallangleplot ] ]
due to energy and momentum conservation , the scattering of electrons from spin - fluctuations is most efficient close to `` hot lines '' on the fermi surface , where @xmath199 , where @xmath23 is the ordering vector of the antiferromagnet . in order to determine how spin precession modifies the results we calculate the scattering rate as a function of @xmath61 .
we will neglect all orbital effects of the magnetic field and will not try to calculate the conductivity and the hall effect . for an extensive discussion of orbital effects and magnetotransport in nearly antiferromagnetic metals see ref .
[ ] which does , however , not consider effects of spin precession . in second order perturbation
theory the lifetime of the spin - up electron scattering from fluctuations of @xmath200 at @xmath201 is given by @xcite @xmath202,\ ] ] where @xmath203 is a coupling constant , @xmath204 and @xmath205 the energy and velocity of spin up / down electrons and @xmath157 is the spin fluctuation spectrum of eq .
( [ phipropagator ] ) , @xmath206 with @xmath207 .
we split the momentum integration in an integral over the fermi surface and an energy integration @xmath208 and integrate first over @xmath209 , then over @xmath210 to obtain @xmath211 + \sin \th \arctan \left[\frac{- e^+_{{{\bf \boldsymbol k } } } \cos \th } { \w_{{{\bf \boldsymbol k } } - { { \bf \boldsymbol k^\prime}}}+r + e^+_{{{\bf \boldsymbol k } } } \sin \th } \right ] \right ) \\ & \approx \frac{g_s^2 q_0 ^ 2}{v_f^- ( 2 \pi)^2 } e^+_{{{\bf \boldsymbol k } } } { \rm min } \left\{\frac{e^+_{{{\bf \boldsymbol k}}}}{2 \d_k^2 } \cos \th , \frac \pi 2 - \th\right\ } , \end{aligned}\ ] ] where @xmath212 and @xmath213 is the distance of @xmath214 from the fermi surface or , approximately , the distance of @xmath32 from hot lines on the fermi surface .
analogously we obtain for spin - down electrons @xmath215 where the indices @xmath216 and @xmath217 have been exchanged w.r.t .
( [ tauup ] ) .
the scattering rate is strongly dependent on the distance from the hot lines : at the quantum critical point and for @xmath218 the scattering rate is linear in the quasiparticle energy . far away from the hot lines and the quantum critical point the usual scattering rate @xmath219
is recovered @xcite .
the main result of this section is that the spin - precession term does not lead to a qualitative change in the scattering rate .
in this paper , we have discussed the field - induced quantum phase transition of a clean , three - dimensional antiferromagnetic metal , restricting our attention to the non - magnetic side of the phase diagram .
the main question was how the interplay of precession of the spins in the presence of a finite magnetic field and landau damping modifies the quantum critical behavior .
one main qualitative result of our analysis is that the critical behavior is _ not _ completely universal as it depends on a continuous variable @xmath61 which parametrizes the ratio of precession and damping terms in the effective action .
while critical exponents do not depend on @xmath61 , this parameter strongly changes the scaling functions and even the sign of leading corrections e.g. to the specific heat .
a requirement for the validity of our analysis is that the two modes @xmath220 perpendicular to the magnetic field are characterized by the same mass . in the presence of sizable spin - orbit couplings , this will only be the case , if the crystal has a sufficiently high symmetry and if furthermore the external magnetic field is applied along a symmetry direction of a crystal .
presently , we are not aware of any experiments which show for example the maximum in the @xmath94 dependence of the specific heat coefficient which we predict for @xmath221 . note that in systems like cecu@xmath4ag@xmath5 @xcite or cecu@xmath6au@xmath7 @xcite strong anisotropies prohibit the precession of the spin , i.e. @xmath222 . under
what conditions can large values of @xmath61 be expected ? obviously , large uniform magnetizations are required . in heavy fermion systems with kondo temperatures of the order of a few kelvin
, one can introduce strong magnetic polarizations with moderate external fields and it should therefore be possible to induce sizable values of @xmath61 . a different class of systems which might be of interest in this context
are ferrimagnetic materials .
if it is possible to suppress only the staggered component of the magnetization in such systems either by external fields , pressure or doping , the critical theory within the hertz approach will be characterized by a finite ( and again sizable ) @xmath61 . as long as the ordering vector @xmath23 of the antiferromagnet can connect
the spin - up and spin - down fermi surfaces ( @xmath223 or , more precisely , @xmath224 for a line of momenta @xmath32 ) , landau damping is present and @xmath225 .
in contrast , one finds @xmath83 in all systems where no such connection exists ( @xmath226 ) .
however , the transition from @xmath225 to @xmath83 is _ not _ expected to be smooth , as the interactions of the spin - fluctuations diverge and become relevant@xcite at the point where @xmath227 . according to our analysis ,
the susceptibility @xmath228 is a particularly interesting experimental quantity to study close to a field - driven quantum critical point .
first of all , it is expected to be much more sensitive to small deviations from criticality compared to other thermodynamic quantities ( see fig .
[ suscsmallangleplot ] ) .
second , it allows to measure the correlation length , a quantity which can not be extracted from other thermodynamic quantities , as for @xmath15 we obtain from ( [ chiquantcrit ] ) @xmath229 third , it strongly violates the @xmath230 scaling . this deviation from scaling for @xmath157
can be used to show that the relevant critical theory is above its upper critical dimension , a central question for the interpretation of quantum criticality in systems like cecu@xmath231au@xmath10 or ybrh@xmath11si@xmath11 @xcite .
all these three statements are actually independent of the value of @xmath180 : they apply equally for a dynamics which is overdamped , @xmath232 , or for a bec system like tlcucl@xmath13 with @xmath233 .
note that in pressure - tuned quantum critical points the compressibility @xmath234 ( and therefore the sound velocity@xcite ) plays the same role as @xmath157 for field tuned quantum phase transitions .
an overview of the qualitative @xmath94 dependence of the specific heat and susceptibility is shown in fig . [ overview ] .
field tuned quantum - phase transitions in metals allow to study quantum critical behavior with a tuning parameter which can easily be controlled and with a conjugate field
the uniform magnetization which can directly be measured .
they are therefore especially well suited to answer some of the central questions in the field of quantum critical metals , for example , whether or not such systems can be described in terms of simple spin - fluctuation theories as have been used in this paper .
we thank t. lorenz , b. luethi , s. paschen , a. schofield , n. shah , q. si , and m. vojta for useful discussions and the emmy - noether program and sfb 608 of the dfg for financial support .
in this appendix we briefly discuss whether cubic terms @xmath235 are present in the low - energy effective lagrangian . as @xmath16 carries the momentum @xmath23 , the presence of such terms
is forbidden by momentum conservation in most systems with the exception of magnetic structures ( e.g. bcc lattices ) where the sum of three ordering vectors adds up to 0 .
if such a system has ising symmetry , then a cubic term does exist and the magnetic field driven transition will be first order .
however , for @xmath236 symmetry perpendicular to the magnetic field ( the case mostly discussed in this paper ) , a rotationally invariant cubic term of the form @xmath237 does _ not _ exist . while terms like @xmath238 _ are
_ allowed by symmetry , they lead effectively only to a renormalization of the @xmath239 interaction as @xmath59 remains massive .
we therefore neglect such terms .
following millis treatment@xcite , we perform the renormalization group analysis on the free energy after having converted all matsubara sums to integrals .
although we restrict our calculations to systems in three spatial dimensions and with a dynamical critical exponent of @xmath22 , we nonetheless keep the variables @xmath240 and @xmath27 in the calculation in order to make the origin of certain factors more transparent .
the free energy can be obtained via a linked cluster expansion in the coupling constant @xmath95 .
the scaling dimension of @xmath95 is @xmath241 , which is negative for an antiferromagnetic system in 3 spatial dimensions .
to first order in @xmath95 , only the diagram in fig .
[ freeenergydiag ] contributes to the free energy . .
the different contractions of the internal indices of the fields involved have been made explicit on the right hand side , where the dashed line represents the quartic interaction @xmath95.[freeenergydiag ] ] up to first order in @xmath95 , the free energy @xmath242 is therefore given by @xmath243,\ ] ] where @xmath244 is the gaussian free energy measured in units of @xmath115 with the cutoffs @xmath245 and @xmath246 , @xmath247 is given by @xmath248 and @xmath249 , @xmath250 are defined analogously . as a next step
, we separate out of the momentum and frequency integrals in the expressions on the right hand side of ( [ freeenexp1 ] ) the regions given by @xmath251 and @xmath252 . using that @xmath253 the change in @xmath242 upon such a variation of the cutoff can be expressed as a change of @xmath37 and @xmath36 , and this leads to the equations @xmath254 for the running masses @xmath255 , @xmath256 , where @xmath257 and @xmath258 are given by @xmath259 in the following we assume that the system is close to the quantum critical point at temperatures much smaller than @xmath36 . in this case , @xmath260 can be set to zero , and the renormalization group flow of @xmath37 is determined by @xmath261 only .
there are two contributions to @xmath257 , one from the renormalization due to the separated momentum shell , where momentum is set on shell @xmath262 , and one from the renormalization due to the frequency shell with @xmath263 . for subsequent calculations we note that @xmath264 \times \nonumber \\ & \hspace{-3.5cm}\frac{2 \w ( ( \lambda^2 + r_\perp)^2 + \w^2)\cos\th}{((\lambda^2+r_\perp)^2 + \w^2)^2 - 4 ( \lambda^2+r_\perp)^2 \w^2 \sin^2 \th } + \o ( e^{-\gamma/\t } ) , \end{aligned}\ ] ] in other words the contribution of the frequency shell renormalizes zero temperature properties only and is exponentially suppressed at finite temperatures . in order to obtain an expression for the correlation length , we first substitute @xmath265 to eliminate the naive scaling and then formally integrate equation ( [ rperprg2 ] ) : @xmath266 we then perform an expansion in temperature @xmath267 where three terms contribute . the first term @xmath268 is the running mass at zero temperature , @xmath269 the integrand can now be expanded in @xmath270 which leads to the following expression @xmath271 i.e. this defines the parameter which characterizes the distance from the critical point .
the other two terms in ( [ texpansion ] ) are of first order in temperature : one contribution is due to an explicit dependence of @xmath257 on the running temperature , @xmath272 and @xmath273 originates from the temperature dependence of the running mass , @xmath274 this term is of the order of @xmath275 and will be neglected from now on .
@xmath276 \frac{2 \w ( ( \lambda^2 + r_\perp(e^x ) e^{2x})^2 + \w^2)\cos\th}{((\lambda^2+r_\perp(e^x ) e^{2x})^2 + \w^2)^2 - 4 ( \lambda^2+r_\perp(e^x ) e^{2x})^2 \w^2 \sin^2 \th } \\
& = & \label{uglyxi } r_\perp + 16 \lambda^{d+z-2 } k_d t^{2/z } \int_{\ln(\frac{t^{1/z}}{\lambda})}^\infty d x \ , u(\e^x \lambda t^{-1/z } )
\e^{(z-2 ) x } \int_0^\infty \frac{d v}{\pi } ( \coth v - 1 ) \\ & & \frac{4 \lambda^z \e^{z x } v \big((\lambda^2 + r_\perp(e^x \lambda t^{1/z } ) e^{2 x } \lambda^2 t^{-2/z})^2 + 4 \lambda ^{2 z } \e^{2 z x } v^2 \big ) \cos \th}{\big((\lambda^2 + r_\perp(e^x \lambda t^{1/z } ) e^{2 x } \lambda^2 t^{-2/z})^2 + 4 \lambda ^{2 z } \e^{2 z x } v^2 \big)^2 - 16(\lambda^2 + r_\perp(e^x \lambda t^{1/z } ) e^{2 x } \lambda^2 t^{-2/z})^2 v^2 \lambda^{2 x } \e^{2 z x}\sin^2 \th } , \nonumber\end{aligned}\ ] ] where the transformations @xmath277 and @xmath278 have been introduced , and @xmath279 = @xmath280 . expression ( [ uglyxi ] ) for the correlation length can now be evaluated in the quantum critical and fermi liquid regime . in the quantum critical regime
we can neglect the dependence of the integrand of ( [ uglyxi ] ) on @xmath281 and extend the lower limit of the @xmath282-integral to @xmath283 . using the following integral @xmath284 we obtain @xmath285 in the fermi liquid regime and for low temperatures
, we can replace the running mass @xmath281 in ( [ uglyxi ] ) by the control parameter @xmath37 .
it is convenient at this point to introduce yet another variable transformation of the form @xmath286 .
to lowest order we can then neglect the term @xmath287 in the integrand .
furthermore , we can extend the lower limit of the @xmath282-integral to @xmath288 , thereby inducing an error of order @xmath289 , and obtain @xmath290 in the fermi liquid regime .
j. custers , p. gegenwart , h. wilhelm , k. neumaier , y. tokiwa , o. trovarelli , c. geibel , f. steglich , c. ppin , p. coleman , nature * 424 * , 524 ( 2003 ) ; p. gegenwart , j. custers , c. geibel , k. neumaier , t. tayama , k. tenya , o. trovarelli , and f. steglich , phys .
rev . lett . * 89 * , 056402 ( 2002 ) .
j. s. kim , n. o. moreno , j. l. sarrao , j. d. thompson , and g. r. stewart , phys .
b 69 , 024402 ( 2004 ) .
s. a. grigera , r. s. perry , a. j. schofield , m. chiao , s. r. julian , g. g. lonzarich , s. i. ikeda , y. maeno , a. j. millis , a. p.
mackenzie , science , * 294 * , 329 ( 2001 ) ; a. j. millis , a. j. schofield , g. g. lonzarich , s. a. grigera , phys . rev .
lett . * 88 * , 217204 ( 2002 ) ; r. s. perry , k. kitagawa , s. a. grigera , r. a. borzi , a. p. mackenzie , k. ishida , and y. maeno , phys .
* 92 * , 166602 ( 2004 ) .
h. v. lhneysen , t. pietrus , g. portisch , h.g .
schlager , a. schrder , m. sieck , and t. trappmann , phys .
lett . * 72 * , 3262 ( 1994 ) ; a. rosch , a. schrder , o. stockert , and h. v. lhneysen , phys .
lett . * 79 * , 159 ( 1997 ) ; o. stockert , h. v. lhneysen , a. rosch , n. pyka , and m. loewenhaupt , _ ibid . _ * 80 * , 5627 ( 1998 ) ; a. schrder , g. aeppli , r. coldea , m. adams , o. stockert , h. v. lhneysen , e. bucher , r. ramazasavili , and p. coleman , nature * 407 * , 351 ( 2000 ) . | a magnetic field applied to a three - dimensional antiferromagnetic metal can destroy the long - range order and thereby induce a quantum critical point .
such field - induced quantum critical behavior is the focus of many recent experiments .
we investigate theoretically the quantum critical behavior of clean antiferromagnetic metals subject to a static , spatially uniform external magnetic field
. the external field does not only suppress ( or induce in some systems ) antiferromagnetism but also influences the dynamics of the order parameter by inducing spin precession .
this leads to an exactly _ marginal _ correction to spin - fluctuation theory .
we investigate how the interplay of precession and damping determines the specific heat , magnetization , magnetocaloric effect , susceptibility and scattering rates .
we point out that the precession can change the sign of the leading @xmath0 correction to the specific heat coefficient @xmath1 and can induce a characteristic maximum in @xmath1 for certain parameters .
we argue that the susceptibility @xmath2 is the thermodynamic quantity which shows the most significant change upon approaching the quantum critical point and which gives experimental access to the ( dangerously irrelevant ) spin - spin interactions .
the study of quantum phase transitions is currently a very active field of research in theoretical as well as experimental condensed matter physics . particularly in a large number of metals mostly heavy fermion or transition metal compounds
the critical fluctuations associated with a quantum phase transition induce anomalous behavior in thermodynamic and transport quantities like diverging specific heat coefficients or a linear resistivity quite distinct from the behavior of a conventional fermi liquid .
experimentally there are three main methods to tune a system towards a quantum critical point : doping , pressure , and magnetic field .
doping has the disadvantage that it induces disorder and it can not be easily adjusted within a single sample .
these problems are absent if pressure is used as the control parameter of the quantum phase transitions .
however , the presence of a pressure cell makes many experiments difficult . for this reason
, many recent experiments @xcite investigate field - tuned quantum critical behavior , where an external magnetic field is used to control the distance from the quantum critical point .
generally it is expected that the presence of a magnetic field changes the universality class of the transition as in its presence time reversal invariance is broken . in this paper
, we will therefore analyze theoretically the quantum critical behavior of a clean itinerant antiferromagnet in three dimensions subject to a static , spatially uniform external magnetic field @xmath3 .
such a situation has been investigated in a number of experiments@xcite .
for example in cecu@xmath4ag@xmath5 @xcite and cecu@xmath6au@xmath7 @xcite magnetic order can be suppressed by moderate magnetic fields . in these systems the quantum critical behavior induced by a magnetic field @xmath3 appears to be qualitatively different compared to the critical properties for vanishing field ( controlled by pressure or doping ) . in the presence of a field
these systems seem to follow@xcite the predictions from spin - fluctuation theory @xcite for three - dimensional nearly antiferromagnetic metals , while this is not the case for @xmath8 @xcite .
similarly experiments@xcite in field tuned ybcu@xmath9al@xmath10 appear to be consistent with spin - fluctuation theory , which is not found to be the case in ybrh@xmath11si@xmath11 where magnetic order is suppressed by tiny magnetic fields @xcite .
recently , in cecoin@xmath12 @xcite the superconducting order was suppressed by a magnetic field
it is at the moment a controversial question whether the observed anomalous behavior is related to a superconducting quantum critical point or whether magnetism plays a role in this system .
another interesting class of systems are _ insulators _ like tlcucl@xmath13 @xcite , srcu@xmath11(bo@xmath13)@xmath11 @xcite or bacusi@xmath11o@xmath14 @xcite where antiferromagnetic order has been _ induced _ by the application of a magnetic field @xmath3 .
these transitions @xcite can be interpreted as a bose - einstein condensation ( see below ) of spin-1 excitations .
the energy of the `` spin - up '' component of such triplets is lowered by @xmath3 until it condenses at a critical field , @xmath15 , thereby inducing antiferromagnetic order perpendicular to the magnetic field .
in contrast to classical transitions , the dynamics , i.e. the temporal quantum fluctuations , of the order parameter determines the nature and universality class of a quantum phase transition .
for example , at the critical point of an insulating antiferromagnet , the dynamics of the order parameter @xmath16 can be described @xcite as in a klein gordon equation @xmath17 .
in such a system , typical frequencies @xmath18 scale linearly with the momentum , @xmath19 , where @xmath20 is the dynamical critical exponent .
in contrast , in a metal the excitation of particle - hole pairs leads to a landau damping @xcite of the antiferromagnetic order parameter , @xmath21 and therefore @xmath22 .
here we assumed that the ordering vector @xmath23 is sufficiently small , @xmath24 , such that low - energy particle - hole pairs with momentum @xmath23 exist .
a magnetic field will have two main effects : first it will suppress ( or in some cases@xcite also induce ) magnetic order .
more interesting is the second effect : it induces a precession of the magnetic moments @xmath25 perpendicular to the magnetic field @xmath26 and therefore modifies the dynamics of the order parameter .
the linear time - derivative also translates to a dynamical exponent @xmath22 , and therefore the question arises how the precession competes with damping in a metal , which is characterized by the same @xmath27 . for insulating systems the physics of the precession term
has been widely discussed @xcite .
the corresponding quantum critical behavior as a function of magnetic field of such an insulating magnet in an external field is actually well known : it is expected to be in the same universality class as the quantum phase transition of a low - density interacting bose einstein condensate as a function of chemical potential .
the linear time derivative @xmath28 of the schrdinger equation can in this case be identified with the precession term ( [ precession ] ) , see below . in the following ,
we study the interplay of ohmic damping and spin precession terms in the case of a nearly antiferromagnetic metal .
first we present the model for the order parameter field and a short derivation of the effective action .
then we list the renormalization group equations for the parameters of the model and use them to derive the behavior of the correlation length . in the following sections we calculate the specific heat , thermal expansion , magnetocaloric effect , and susceptibility .
we show for example that sufficiently large magnetic fields can induce sign changes in the critical contribution to the specific heat and that the susceptibility is particularly suited to probe the vicinity of the quantum critical point .
finally , we investigate the influence of the @xmath3-field on the scattering rate of the electrons . |
the observed accelerated expansion of the universe@xcite is currently one of the central issues of observational and theoretical cosmology .
it somehow came as a surprise to discover that the universe is not progressively slowing down its pace , but actually rocketing it up .
the usual explanatory paradigm for this striking observation assumes the existence of the so - called _ dark energy _
( de) a mysterious cosmic component with negative pressure , which , in contrast to dark matter ( dm ) , does not cluster ( at least in a comparable way ) and pervades all corners of the universe .
an obvious candidate for the de is a strictly constant cosmological term , @xmath0 .
nevertheless , this option precludes the possibility of a time evolution of the de ( which might be appealing so as to face the so - called coincidence problem , namely the fact that the current matter density @xmath8 is so close to the de density @xmath9 , or , equivalently , the fact that the acceleration epoch started so recently ) .
this drawback could be cured by admitting that the cc can be a dynamical quantity : @xmath6 , an option which is perfectly admissible by the cosmological principle .
actually , the mere possibility of a variable cc approach was explored long ago on purely phenomenological grounds cf.@xcite
although many of the models are little motivated . at present , we know that the idea of a slowly running cc is in general compatible with the experimental data@xcite and there are reasonable theoretical models grounded on fundamental physical principles that can support these ideas , meaning that there is no strict need to invoke ad hoc scalar fields to generate a time dependence for the de see@xcite and references therein , see also @xcite for recent alternative approaches reinforcing this point of view . still , the variable @xmath6 models can be dealt with in terms of a self - conserved de fluid in the manner of a scalar field ( as we can not know a priori the very nature of the de , and hence we may be naturally tempted to keep on using this simple scalar field approach ) . however , although this description in terms of scalar fields ( which may be called the `` de picture '' of the original @xmath6 model@xcite ) is perfectly possible , the corresponding effective equation of state ( eos ) for the de , @xmath10 , will be a non - trivial function of time or of the redshift , @xmath11 , and in general it can not be reduced to the standard forms proposed in the literature . in other words , the effect of having a variable cosmological term @xmath6 can not be described by the simple parameterizations usually employed for @xmath12 , characterized by two parameters @xmath13 that become constrained by the observations@xcite .
the effective eos of a variable vacuum model is in general more complicated than that .
this is shown in detail , with specific examples , in reference @xcite .
in particular , the vacuum models that we are going to consider can not be described with these simple parameterizations .
therefore , the @xmath6 models must be studied on their own and constitute an independent class of de models .
it means that we should be more open minded on the form that the effective eos can adopt , and we should not limit ourselves to some admittedly popular ( albeit perhaps too simple ) forms utilized too often in the literature . remarkably , the variable @xmath6 approach can also be advantageously employed for solving ( or highly palliating ) the big or `` old '' cc problem@xcite , viz .
the absolutely formidable task of trying to explain the value itself of the current vacuum energy density @xmath14 gev@xmath15 on the face of its enormous input value @xmath16 ( with @xmath17 gev ) left over in the very early times , presumably just after the inflation phase induced by a grand unified theory ( gut ) .
this option has been recently investigated in@xcite , and we will also review it here .
it is based on the idea of `` relaxation '' of the effective vacuum energy density @xmath18 .
relaxation offers indeed an entirely different perspective on the old cc problem ; in particular , it provides the extremely appealing possibility to solve the cc problem _ dynamically _ , hence _ _ without fine tuning__@xcite .
moreover , the current value of the expansion rate @xmath19 becomes determined by an approximate relation of the form @xmath20 .
interestingly enough , this relation tells us that @xmath21 is small just because the initial vacuum energy @xmath22 was very large ! in the next sections , i summarize the general notion and properties of the models with time - dependent vacuum energy , and consider specific scenarios possessing very interesting features concerning the possible solution , or at least significant alleviation , of some of the most fundamental cosmological constant problems of modern times .
suppose that we are given a model where the vacuum energy density @xmath2 is time evolving through its dependence on other cosmological function(s ) of the cosmic time , e.g. @xmath23 .
we take this starting `` picture '' as the original ( physical ) representation of the de , and we call it `` the cc picture '' . in general , other fundamental `` constants '' can also be varying , in particular the gravitational coupling @xmath24 .
therefore , in the cc picture we will have in general two functions , at least , of the cosmic time or the cosmological redshift , namely @xmath25 of course other fundamental parameters could also be variable .
for example , the fine structure constant has long been speculated as being potentially variable with the redshift .
however , here we focus purely on the potential variability of the most genuine fundamental gravitational parameters of einstein s equations , to wit : @xmath0 and @xmath24 . for a cosmological model of the kind ( [ varibleccg ] ) , the general bianchi identity of the einstein tensor leads to @xmath26=0 $ ] . in the flrw metric , and modeling the universe as a perfect fluid
, it implies @xmath27 + 3\,g\,h_{\rm cc}\,({\rho_m}+{p_m})=0\,.\ ] ] this mixed " conservation law connects the variation of @xmath28 and @xmath24 with that of the matter density @xmath8 . as a result , the matter fluid in the cc picture is generally non - conserved and its evolution may be non - canonical , as we will see later on in an specific example . the general expression for the hubble function in the cc picture , assuming a spatially flat flrw universe , is @xmath29\,,\ ] ] where @xmath30 , with @xmath31 for the matter and radiation epochs respectively . here
@xmath32 and @xmath33 are certain functions of the redshift .
they must satisfy @xmath34 and @xmath35 , in accordance with the current cosmic sum rule @xmath36 for a flat universe . on the other hand , the above cosmological dynamics in the cc picture can also be described in terms of an alternative fluid with covariantly self - conserved densities @xmath37 , therefore satisfying @xmath38 and @xmath39 , where @xmath40 will be , in general , a function of the redshift : @xmath41 .
of course @xmath42 and @xmath43 in general .
the hubble rate in the new picture takes on the form : @xmath44\,,\ ] ] where @xmath45 we have defined @xmath46 , with @xmath47 the current value of the critical density . notice that the parameter @xmath48 need not coincide with @xmath49 , as they are determined from two different parameterizations of the de data .
similarly , we have defined @xmath50 and its current value @xmath51 .
notice also that @xmath52 ( @xmath53 being the planck mass ) is strictly constant , whereas @xmath24 in eq.([hlambda ] ) is in general a variable function of the redshift : @xmath54 .
the cosmological picture characterized by the _ self - conserved _ pair of energy densities @xmath37 at fixed @xmath55 will be called the `` de picture''@xcite .
since the two pictures are assumed to be equivalent descriptions of the same cosmological evolution , their matching requires that the expansion history of the universe is the same in both pictures , i.e. that their hubble functions are numerically equal : @xmath56 recall that we take the original picture ( [ varibleccg ] ) as the physical one , whereas the de picture plays the role of an effective description of the former in terms of the self - conserved de component @xmath57 at fixed @xmath55 .
therefore , the function @xmath58 defined above can be considered as the `` effective eos '' ( in the `` de picture '' ) of the original model ( [ varibleccg ] ) .
a straightforward calculation , using the matching condition ( [ matching ] ) and the general bianchi identity ( [ bianchigeneral ] ) , allows to determine the effective eos function @xmath58 in terms of the fundamental parameters of the original theory ( [ varibleccg ] ) , as follows@xcite : @xmath59 where we have defined @xmath60 , and @xmath61\,.\ ] ] in particular , for a model of the kind ( [ varibleccg ] ) with fixed @xmath62 in the matter epoch ( @xmath63 ) , the effective eos simply reads @xmath64 .
the latter would shrink to just @xmath65 only in case of a truly cosmological constant @xmath66 because then @xmath57 would coincide with it . with the help of the matching condition ( [ matching ] ) of the two pictures , and the bianchi identity
, one can also show , with some effort , that @xmath67 this is a remarkable relation , if we recall the conditions @xmath68 fulfilled by the functions @xmath69 and @xmath70 in eq.([hlambda ] ) .
why is remarkable ?
as @xmath49 and @xmath48 must be very close , and @xmath71 is presumably monotonous , eq.([dzeta ] ) shows that a value @xmath72 _ always _ exists for which the _ l.h.s .
_ of eq.([dzeta ] ) vanishes and , therefore , the effective eos @xmath58 crosses the phantom divide at that point : @xmath73 . whether the crossing is in the recent past or in the immediate future
, it will depend on the details of the original model ( [ varibleccg ] ) .
notice , however , that if @xmath74 , then the crossing will be exactly at @xmath75 , i.e. now .
this shows that a generic model ( [ varibleccg ] ) with time varying vacuum energy , when described in the de picture ( i.e. as if it were a cosmological self - conserved de fluid ) generally leads to a crossing of the phantom divide , thus providing a possible natural explanation for the observational data which still admit a tilt in the phantom domain@xcite .
in contradistinction to purely phenomenological models of time dependent @xmath0@xcite , we adopt here the fundamental point of view that the @xmath0 term is a running parameter in qft in curved space - time@xcite .
as suggested in @xcite , we should have a corresponding renormalization group ( rg ) equation of the general form @xmath76 in which @xmath77 is an arbitrary scale associated to the rg running .
of course the full effective action is perfectly scale independent ( i.e. rg invariant)@xcite , and the running of the vacuum energy should ultimately reflect the dependence of the leading quantum effects with respect to some physical cosmological quantity @xmath3 associated with @xmath77 , hence @xmath78 . despite it being a matter of debate , it has been argued by different methods@xcite that the physical scale @xmath79 pointed by the @xmath77-dependence is the hubble rate @xmath19 .
since @xmath80 evolves with the cosmic time , the cosmological term inherits a time - dependence through its primary scale evolution with @xmath19 .
the coefficients @xmath81 in eq.([rgeg1a ] ) are obtained after summing over the loop contributions of fields of different masses @xmath82 and spins @xmath83 .
the general behavior is @xmath84@xcite .
therefore , for @xmath85 , the series above is ( for @xmath86 ) an expansion involving powers of the small quantities @xmath87 and @xmath88 .
given , however , that @xmath89 , we see that the heaviest fields furnish the dominant contribution .
this trait ( `` soft - decoupling '' ) represents a generalization of the decoupling theorem in qft . in addition , being @xmath90 , the condition @xmath91 is amply met for all known particles , and the series on the _
r.h.s _ of eq.([rgeg1a ] ) converges extremely fast at present .
also important is the fact that only even powers of @xmath87 are consistent with general covariance , see@xcite .
finally , the @xmath92 contribution is absent because it corresponds to terms @xmath93 that give an extremely fast evolution .
actually , from the rg point of view they are already excluded because , as noted above , @xmath91 for all known masses . in practice only the first term @xmath94 is needed , with @xmath82 of the order of the highest mass available .
as the dominant masses @xmath82 will be of order of a gut mass scale @xmath95 near the planck mass @xmath53 , it is convenient to introduce the ratio @xmath96 , in which @xmath97 depending on whether bosons or fermions dominate in their loop contributions to ( [ rgeg1a ] ) . if the effective mass @xmath95 of the heavy particles is just @xmath53 , the parameter @xmath98 takes the value @xmath99 , with @xmath100 . in general
we expect that @xmath98 will take values of this order or below . under the very good approximation @xmath94 , and with @xmath87 , eq.([rgeg1a ] ) abridges to @xmath101 whose solution is @xmath102 with @xmath103 [ cols="^,^ " , ] solving the model in the original cc picture@xcite leads to a non - standard evolution law for the matter density as a function of the redshift : @xmath104 , along with the following evolution of the vacuum energy density : @xmath105\,.\ ] ] for @xmath106 we recover the @xmath0cdm results , of course .
but for @xmath107 we have a model in which the vacuum energy decays into matter ( or vice versa ) according to the generalized conservation law ( [ bianchigeneral ] ) , with @xmath108 in this case . with this rg model of the cosmic evolution at hand ,
we may now use it to apply the procedure of section [ ccdepicture ] for obtaining the effective properties of the model in the `` de picture '' ( in which matter and dark energy are separately conserved ) . obviously , for this particular model , @xmath109 .
the corresponding effective eos can not be described with standard parameterizations@xcite . indeed , from equation ( [ we2 ] ) using ( [ if ] ) and ( [ crg ] )
we find : @xmath110-(1-\nu)\,[\tilde{\omega}_m^0\,(1+z)^3 - 1]}\,.\ ] ] here @xmath111 is the difference of the cosmological mass parameters in the two pictures ( corresponding to two different fits of the same data ) .
for @xmath112 we may expand the previous result to first order in @xmath98 .
assuming for simplicity that @xmath113 , we find @xmath114 this result reflects the essential qualitative features of the analysis presented in the previous section , where in this case @xmath115 .
indeed , for @xmath116 eq.([expwq ] ) shows that we get an ( effective ) phantom - like behavior ( @xmath117 ) , whereas for @xmath118 we have an ( effective ) quintessence behavior ( @xmath119 ) . in other words ,
this variable cc model can give rise to two types of very different qualitative behaviors by changing the sign of a single parameter .
in fig.[plot1 ] we show also more general cases where @xmath120 , corresponding to @xmath121 .
in contrast to the previous situation , here the variable cc model may exhibit phantom behavior for @xmath118 ( if @xmath122 ) , and manifests itself through the existence of a transition point @xmath123 in our recent past marked explicitly in the figure as predicted by the general formalism of section [ ccdepicture ] .
let us also examine another interesting situation . from eq.([bianchigeneral ] )
, we see that if @xmath7 is variable and we assume that matter is conserved ( thus no transfer of energy with the time dependent vacuum @xmath124 ) , we have @xmath125 . from eq.([if ] ) it is then not difficult to show that in eq.([if ] ) satisfies the differential equation @xmath126 .
next one notes that the previous equation is also solved by the alternative expression below : @xmath127 ] @xmath128\,.\ ] ] where @xmath123 is the crossing point of the phantom divide , which satisfies @xmath129 .
it follows that if @xmath24 is asymptotically free that is to say , if @xmath130 decreases with redshift we should observe quintessence behavior ( @xmath131 ) for @xmath132 ; whereas if @xmath24 is `` ir free '' ( i.e. it increases with @xmath133 , hence decreases with the expansion ) , we should then have phantom behavior ( @xmath134 ) in that interval .
an interesting example of model of this kind appears if we take once more the evolution law ( [ cch ] ) for the vacuum energy , but assume now that matter is conserved@xcite .
the basic cosmological equations can be formulated in the following compact way : @xmath135\ \ \ ( { \rm friedmann\ equation})\nonumber\\ & & ( { \omega_m}+{\omega_{\lambda}})dg+g\,d{\omega_{\lambda}}=0 \ \ \ \ \ \ \ \ \ ( { \rm bianchi\ identity})\label{bianchi2}\\ & & { \omega_{\lambda}}(z)={\omega_{\lambda}^0}+\nu\left[e^2(z)-1\right]\ \ \ \ \ \ ( { \rm running\ { \rho_{\lambda } } } ) \nonumber\ , , \label{rglaw3}\end{aligned}\ ] ] where @xmath136 and @xmath137 .
it is now straightforward to solve them for the gravitational coupling , which becomes a logarithmically running quantity of the hubble rate : @xmath138 we see that @xmath24 is asymptotically free ( resp .
ir free ) for @xmath116 ( resp .
@xmath118 ) . an alternative formulation of this running law for @xmath24 and @xmath139 , motivated by the quantum physics of inflation , was presented in @xcite .
the previous considerations show that we have plenty of theoretical and phenomenological motivations to support the possibility of observing running cosmological parameters , specially through the study of time - evolving vacuum cosmologies .
ultimately this issue must of course be backed up by observations , and in this sense it is very important to try to find all possible phenomenological indications suggesting such a vacuum dynamical picture .
it is also remarkable that one could even find signs of the running @xmath0 in the astrophysical context ; for example , through the study of the implications that a variable @xmath0 distributed ( homogeneously or inhomogeneously ) in a large domain could have in the process of local structure formation , see@xcite .
time - evolving vacuum models inspired by qft principles could do even more : they could provide a possible solution to the cosmic coincidence problem , and ultimately to the `` old cosmological constant problem '' , namely the cc fine - tuning problem the toughest cosmological conundrum of all times@xcite .
possible avenues leading to a significant step in this direction are summarized in the next two sections .
the @xmath0xcdm model@xcite makes a further step in our systematic approach to a cosmology with variable cosmological parameters .
it is based on assuming that we have a composite de energy density @xmath140 . in the canonical case where
@xmath108 , it can be taken as covariantly conserved : @xmath141 and in these conditions matter is self - conserved too of course : @xmath142 . here @xmath143 is the energy density of the running @xmath0 .
moreover , @xmath144 is the energy density of a new dynamical entity @xmath145 that we call the `` cosmon '' .
its dynamics is completely determined by the above local conservation law ( [ fosc ] ) once the evolution for @xmath146 is given .
as emphasized in@xcite , this implies that in general it will be an effective entity , not a fundamental one ( i.e. @xmath145 is _ not _ supposed to be a scalar field @xmath147 ) because the dynamics imposed by ( [ fosc ] ) will in general make it incompatible with the scalar potential provided for @xmath147 .
the cosmon @xmath145 , instead , embodies the behavior of terms in the effective action of qft in curved space - time , and , far from being an elementary scalar field , it represents a convenient way to parametrize the effect of those terms in the cosmological evolution .
the eos of the cosmon is indicated by @xmath148 ( assumed , for the moment , to be constant for simplicity ) .
it should not be confused with the effective eos of the total de , which reads @xmath149 next we will assume that @xmath150 is the same function of @xmath19 as in the running @xmath0cdm model presented in section [ rgmodel1 ] cf .
eqs.([cch ] ) and ( [ c0c1 ] ) .
notice that this does not mean that its redshift evolution is still given by eq.([crg ] ) , as here the overall conservation law is no longer ( [ bianchigeneral ] ) but ( [ fosc ] ) .
after solving explicitly the details of this new cosmological model which we refrain from reporting here@xcite the normalized energy densities @xmath151 ( @xmath152 ) with respect to the present critical density @xmath153 become determined as a function of the redshift ( cf .
fig.[fig4]a ) .
we see that the cosmon density @xmath154 can be negative , although this is not too surprising for an effective quantity that does not represent a fundamental field . on the other hand
we observe that the total de density @xmath155 behaves properly , and its current value is the measured de value , i.e. we have @xmath156 with @xmath157 ( flat space ) .
furthermore , the eos function @xmath158 can display a rich variety of behaviors while being in agreement with the most recent data@xcite .
this is shown in fig.[fig4]b , where we see that , depending on the value of @xmath98 , the eos can be quintessence - like , or mimic ( in some cases almost exactly ) a pure cc term ( @xmath159 ) , and it can even exhibit a mild transition from the phantom to the quintessence regime . from fig.[fig4]b
we recognize that this model is a spectacular example of cosmology , in which despite of the fact that the vacuum energy @xmath150 is a running quantity , the effective eos of the composite de can approach the standard @xmath0cdm model in an arbitrary way , and may display only small `` quintessence - like '' or `` phantom - like '' departures from it , thus mimicking a scalar field even in situations where such field could not behave canonically ( @xmath160 ) .
[ ratiordrm ] the truly nice feature of a model of this kind , however , is that it can cure or highly alleviate the `` cosmic coincidence problem '' . recall that this problem is related to the behavior of the ratio @xmath161 between the dark energy density and the matter energy density . in the standard @xmath0cdm model , that ratio increases arbitrarily with the cosmic evolution because @xmath57 is just the strictly constant vacuum energy density @xmath162 , whereas @xmath163 with the expansion , so there is no apparent reason why we should find ourselves precisely in an epoch where @xmath164 . or , to put in other terms : there is no obvious reason why the epoch where the universe started to accelerate is so recent .
this is an unexplained coincidence puzzle in the standard model . not so in the @xmath0xcdm model , in which , thanks to the presence of the cosmon @xmath145
, the function @xmath165 presents a maximum at the resdhift @xcite : @xmath166^{\frac{1}{3\,(1+{\omega_x}-\epsilon)}}-1\,,\ ] ] where we have defined @xmath167 .
one can show that this point can not be in our past@xcite , so it will generally be in the future ( hence @xmath168 ) . beyond that maximum
, there is a turning point @xmath169 where the universe stops and reverses its evolution . at the stopping point , @xmath170
this is possible because we can have @xmath171 , or @xmath172 .
the following observation is now in order : notice , from eq.([fosc ] ) , that for @xmath106 ( which implies no running of @xmath124 ) the cosmon energy density is conserved ; so , if in addition @xmath173 ( vanishing current cosmon density ) , we must then have @xmath174 at all times .
therefore , in the limit @xmath175 we exactly recover the standard @xmath0cdm model . and
it is precisely in this limit that the formula ( [ zs ] ) for the redshift location of the maximum of the ratio @xmath176 ceases to make sense , as it is evident by simple inspection of that formula . in contrast ,
when we make allowance for the parameter space of the @xmath0xcdm model , the maximum does exist and moreover the ratio @xmath176 stays bounded
typically @xmath177 for the entire history of the universe .
this is clearly seen in fig.3 , for various values of the parameters .
so , in fact , the @xmath0xcdm model can provide a nice solution to the cosmic coincidence problem without necessarily departing from the @xmath0cdm model in a significant way .
furthermore , the detailed analysis of the cosmic perturbations in the @xmath0xcdm model indicate excellent compatibility with the present observational data on structure formation@xcite .
the `` relaxed universe '' is a special realization of the @xmath0xcdm model which points to a more ambitious aim , to wit : solving the `` old cc problem''@xcite , i.e. understanding how the universe could start up its thermal history with a huge value of the initial vacuum energy @xmath16 ( e.g. triggered by a gut with , say @xmath17 gev , presumably responsible for inflation ) and then eventually ended up with its current tiny value @xmath178 gev@xmath179 !
the history of this profound theoretical conundrum , which keeps in check the fundamental physics principles since more than forty years ago , traces back to zeldovich s realization in the late sixties@xcite , that the vacuum fluctuations of the quantum fields should induce a very large value of the cc .
since then the problem stays with us , unbeaten and virtually invincible .
the `` traditional '' fine - tuning solution @xcite ( i.e. the sheer adjustment by hand of the initial big cc value @xmath22 by a carefully chosen finite `` counterterm '' ) is simply and fully inadmissible , especially at the quantum level see @xcite ( particularly appendix b of this reference ) for a detailed account .
a much more promising approach , instead , should be to find out a mechanism that neutralizes _ dynamically _ the presence of the initial vacuum energy , @xmath22 , whatever it be its seed value in the early universe .
we will call the cosmological model related to this relaxation mechanism , the `` relaxed universe''@xcite . as we would like that the expansion of the universe be responsible for the triggering of the relaxation dynamics ,
it is reasonable to assume that such mechanism should be closely related to the expansion itself , or more precisely , to the interaction governing the expansion , i.e. _ gravity_. a note of caution , however : although we aim at a modified form of gravity , we can not just content ourselves with a `` late time effect '' ( as usually done in the literature on modified gravity @xcite ) .
we rather must have a _
persistent _ effect , fully active throughout the _ entire history _ of the universe .
indeed , in order to solve the old cc problem we need that some automatic mechanism sets to work at a time prior to the onset of the radiation epoch at which time it must cancel most of the value of the original vacuum energy present in the early universe _ and _ , most important , we also need that this cancelation is maintained for the rest of the universe lifetime , leaving a small value today and maybe also in the future .
it is not easy to device such a mechanism , but some ideas on it are briefly explained here
see @xcite , and specially @xcite , for a detailed account ; and @xcite for alternative forms .
the effective action of the `` relaxed universe '' is proposed in the following way : @xmath180\,,\label{eq : cc - relax - action}\ ] ] in which we have an ( arbitrarily large ) initial vacuum energy @xmath22 , and we have included a functional ( the `` cosmon functional '' ) @xmath181 of the ricci scalar @xmath182 and the gauss - bonnet invariant @xmath183 .
the structure of this functional is convenient in order to avoid the presence of ghost degrees of freedom and also to dodge the ostrogradski instability , i.e. the appearance of vacuum states of negative energy@xcite .
the corresponding ( generalized ) einstein equations read @xmath184\,,\ ] ] where @xmath185 is the ordinary energy - momentum tensor of the matter fields , and @xmath186 is the new tensor in the field equations associated to the presence of the cosmon functional @xmath187 in the action .
we assume that matter is covariantly conserved , and therefore @xmath186 must also be conserved : @xmath188 .
clearly , our functional constitutes an extension of the einstein - hilbert s action with cc , in which we have added the @xmath189 part .
the next point is to make an appropriate choice for the functional @xmath187 such that it can have a bearing on the cosmological constant problem .
we make the following ansatz : @xmath190 where @xmath191 is a polynomial of the ricci scalar and the gau - bonnet invariant , and @xmath192 are dimensionful parameters to be determined later .
this model is in fact of @xmath0xcdm type , in the sense that the cosmon @xmath145 here is related to the energy density @xmath193 generated by @xmath187 in the effective action , and so @xmath194 plays the role of @xmath195 in the @xmath0xcdm model .
explicitly , one finds @xmath196\,.\label{eq : e00}\end{aligned}\ ] ] in our case , @xmath197 has the form @xmath198 as defined in ( [ eq : ffp ] ) , and @xmath199 are the partial derivatives of @xmath200 with respect to @xmath201 .
the parallelism with the @xmath0xcdm model becomes even more transparent if we notice that the mixed de energy density defined in ( [ fosc ] ) takes here the form @xmath202 , and is covariantly conserved .
indeed , since @xmath22 is constant , we have @xmath203 , where @xmath204 is the effective eos of the cosmon , i.e. the analog of @xmath148 in section [ lxcdmmodel2 ] , except that here @xmath205 is a non - trivial function of the cosmological evolution which is determined by the structure of the cosmon functional ( [ eq :
ffp ] ) .
this is indeed a good example of the situation we have mentioned in section [ lxcdmmodel2 ] , in which we emphasized that the cosmon @xmath145 is _ not _ a fundamental field , but a non - trivial field - theoretical object generated from terms of the effective action of qft in curved space - time . evaluating it in the flrw metric we obtain , for its denominator ( [ eq : defb ] ) : @xmath206^{3}\,,\label{eq : b}\ ] ] where @xmath207 is the deceleration parameter .
the previous formula shows why the cosmon functional ( [ eq : ffp ] ) can provide a dynamical mechanism for counterbalancing the value of @xmath22 , whatever it be its size .
the relaxation of the effective vacuum energy @xmath202 in the radiation epoch originates from the @xmath208 factor in the denominator ( [ eq : b ] ) of the cosmon functional , as this term grows as @xmath209 at large @xmath19 and dominates @xmath210 in the earlier epochs .
the large @xmath22 left over in the immediate post - inflationary period drives the deceleration parameter @xmath211 to larger values until @xmath212 , which corresponds to radiation - like expansion .
in other words , the very existence of the radiation period is triggered automatically by the presence of this term , which can be thought of as a countermeasure launched by the universe against the presence of the large `` residual '' vacuum energy @xmath22 in the pre - radiation era . as a result , @xmath213 is driven _ dynamically _ to very large values , opposite in sign to @xmath22 , and produce a small effective @xmath214 in the radiation epoch , for an appropriate sign of the parameter @xmath215 .
similarly , the @xmath216 factor in the denominator ( [ eq : b ] ) will be responsible for the relaxation in the matter era ( @xmath217 ) , at lower values of @xmath19 .
this is possible because the term with the @xmath216 factor is weighted by a power of @xmath19 smaller than the term carrying the @xmath208 factor .
the transition between the two regimes ( i.e. from radiation to matter ) is what fixes the parameter @xmath218 , where @xmath219 is the hubble rate just at the transition time between the two epochs . a numerical example illustrating all these features
can be appreciated in fig .
[ relaxhistory ] .
the relaxed universe is seen to follow very closely the past history of the standard @xmath0cdm model ( see especially the upper panel in that figure ) , and its effective eos loiters around the @xmath0cdm value @xmath220 near our time ( see the second panel of that figure ) . in the far past , @xmath221 turns out to track the values @xmath222 and @xmath223 in the matter and radiation epochs , respectively .
finally , when the universe leaves the matter epoch @xmath224 , there is a kind of impasse , as there is no longer a special value for @xmath211 to pick dynamically in order to maintain the counterbalance of the large @xmath22 by the cosmon density @xmath194 .
what to do now ? under these circumstances , the last resort of the relaxation mechanism is to enforce ( dynamically ) a very low value of @xmath19 .
since in the current universe we must have @xmath225 , with @xmath226 , it follows that the value of @xmath19 that solves this equation is approximately given by @xmath227 for an appropriate choice of @xmath215 , we may bring @xmath228 close enough to the current value @xmath21 . as @xmath215 has dimension @xmath229 of mass ,
we can rewrite it as @xmath230 .
the relaxation condition requires @xmath231 where @xmath232 .
thus , @xmath233 .
taking the standard gut scale @xmath17 gev , we obtain @xmath234 in the ballpark of a light neutrino mass : @xmath235 incidentally , this value for @xmath236 is the geometric mean of the two most extreme mass scales available in our universe below the planck mass .
it is rewarding to see that we do not obtain an extremely small mass scale of order @xmath237 ev , as in the case of quintessence models . from the above considerations , we see that the @xmath197-functional s main mission is to _ dynamically _ counterbalance the big initial @xmath22 at _ all epochs _ , not just now ; and in the last stage of this dynamical process the hubble rate is driven to the very small value @xmath21 that we observe today . at the end of the day
, what we get is not just a late time effect of the cosmological evolution ( as in the traditional modified gravity models ) , but rather a truly perennial large scale influence on all periods of the cosmic history .
one hope is that a relaxation mechanism of this sort should ultimate provide a dynamical solution to the tough fine tuning problem underlying the `` old cc problem''@xcite .
for example , in the standard model of electroweak interactions such fine - tuning involves @xmath238 decimal places , at least , and it must be retuned order by order in perturbation theory until reaching loop diagrams of order 20th !
@xcite for details . in the relaxed universe , in contrast
, the tuning to all orders is automatic and performed dynamically by the universe s expansion .
this is why such field - theoretical object associated to the functional @xmath197 is called the `` cosmon '' , as its primary mission is to compensate for the huge initial cc throughout the entire cosmological evolution .
clearly , it has the same aim as the cosmon scalar field first introduced in @xcite , except that here the cosmon has nothing to do with scalar fields , and moreover is _ not _ subdued by weiberg s `` no - go '' theorem
see@xcite for a more detailed account .
while in the present version the cosmon is not a fundamental quintessence field@xcite , it can perfectly mimic quintessence - like behavior ( cf .
[ relaxhistory ] ) .
the new cosmon , however , is a more complex object : it is a functional whose ultimate origin must be provided by a fundamental theory ( say string theory or any other theory aiming at a final explanation of quantum gravity ) . in the meanwhile ,
what we have shown here is a practical working example , actually a class of such functionals which are able to truly adjust the vacuum energy dynamically , whatever it be its initial value , and without performing fine - tuning at any stage of the cosmological evolution .
as this was the main aim of the original cosmon model@xcite , it is naturally suggested that the general class of these functionals is called the `` new cosmon '' . for more details , see @xcite .
let us also mention that the relaxed universe based on the cosmon functional ( [ eq : ffp ] ) and extensions thereof ( cf .
ref.@xcite ) , provides also a possible natural solution to the cosmic coincidence problem ( mentioned in section[lxcdmmodel2 ] ) .
this additional bonus is not too surprising if we recall that the relaxed universe is a particular implementation of the class of @xmath0xcdm models of the cosmic evolution@xcite .
we have already mentioned that , in the relaxed universe , there is a tracking behavior of matter / radiation and de during the @xmath239 ( radiation ) and @xmath240 ( matter epochs ) . however , when the matter epoch is left behind and the relaxation mechanism is enforced ( dynamically ) to pick a very small value of @xmath19 ( as we have seen above ) in order to insure the condition @xmath241 to hold , this crucial event decoupled once and forever the tracking behaviors of matter / radiation and de , and since then the de dominates the universe s expansion . such breakdown of the tracking property at the end of the matter epoch , on the other hand , pushed the subdominant de density curve upwards until crossing the decaying dm one at some point after the matter epoch and hence near our time .
this feature , which can be very well appreciated in fig .
[ relaxhistory ] ( cf .
third panel ) , encodes the very origin and the possible exciting explanation ( within the relaxed universe ) of the `` coincidence @xmath242 '' of the matter and de densities in our late time neighborhood . not only so : we can also understand now the smallness of both @xmath243 and @xmath21 in front of the huge initial vacuum energy density @xmath22 .
indeed , from ( [ hstar ] ) we predict @xmath244 gev , for @xmath234 not far from the @xmath245mev scale , which is the mass scale itself of the current cc : @xmath246 ev .
we have thus arrived at a most natural scenario ( `` the relaxed universe '' ) in which , at variance with the quintessence formulation of the de , there is no need to severely fine - tune any parameter of the model , and moreover it is not required the existence of any extremely tiny mass scale .
we have shown that models with variable vacuum energy @xmath28 ( and may be also with variable @xmath24 ) generally lead to a non - trivial effective eos when viewed from the point of view of a self - conserved de fluid at fixed @xmath24 . as a result
, these models can effectively appear as quintessence , and even as phantom energy , without need of invoking the existence of fundamental quintessence and phantom fields . combining a time - evolving @xmath28 with a variable `` cosmon '' energy density @xmath195 ( in general _
not _ representing any scalar field ) , we obtain the class of @xmath0xcdm models , which are able to deal with the cosmic coincidence problem .
finally , one can provide a version of the @xmath0xcdm model in which the new cosmon stands for a self - adjusting functional , namely one which is able to counteract the effects of an arbitrarily large cc in the early universe , leaving a small and completely innocuous net value of the cc at any stage of the cosmological evolution , in particular the tiny value @xmath14 gev@xmath15 that we observe at present .
this particular cosmon functional is thus able to reduce the value of the cc _ without fine - tuning _ , and may be eligible as a prototype model ( `` the relaxed universe '' ) for eventually solving the old cc problem .
i have been partially supported by diue / cur generalitat de catalunya under project 2009sgr502 and by mec and feder under project fpa2007 - 66665 , and also by the consolider - ingenio 2010 program cpan csd2007 - 00042 .
it is my pleasure to thank s. basilakos , f. bauer , j.c .
fabris , j. grande , a. pelinson , m. plionis , i.l .
shapiro and h. tefani for a nice and stimulating collaboration in some of the topics discussed here .
i would also like to thank r. woodard , b.f.l .
ward , m. maggiore and n. bili for discussions .
last but not least , i am also very grateful to l. perivolaropoulos and all the organizers of neb 14 , for the invitation to this excellent hellenic conference on recent developments in gravity .
10 r. knop et al . , astrophys .
j. * 598 * ( 2003 ) 102 ; + a. riess _ et al .
_ astrophys .
j. * 607 * ( 2004 ) 665 .
o. bertolami , nuovo cimento , * 93b * , 36 , ( 1986 ) ; + m. ozer m. and o. taha , nucl .
phys . , * b287 * , 776 , ( 1987 ) ; + o. k. freese k. , et al . , nucl .
phys . , * 287 * , 797 , ( 1987 ) ; + j. c. carvalho , j. a. s. lima and i. waga , phys .
rev . * d46 * , 2404 , ( 1992 ) .
s. basilakos , m. plionis and s. sol , phys . rev . * d 80 * , ( 2009 ) 3511 , arxiv:0907.4555 [ astro-ph.co ] ; + j. grande , j. sol , j.c .
fabris , i.l .
shapiro , class.quant .
* 27 * ( 2010 ) 105004 , arxiv:1001.0259
[ astro-ph.co ] . i. l. shapiro , j. sol , phys . lett . *
b682 * ( 2009 ) 105 , arxiv:0910.4925 [ hep - th ] ; + see also the reviews : i. l. shapiro , j. sol , j. phys . * a40*(2007 ) 6583 , + and specially arxiv:0808.0315 for a detailed exposition of the qft of the cosmological term , and references therein .
m. chevallier and d. polarski , int .
j. od mod .
d10 * ( 2001 ) 1939 ; + e.v .
linder , phys .
* 90 * ( 2003 ) 091301 ; + h. k. jassal , j.s .
bagla , t. padmanabhan , phys . rev . *
d72 * 103503 , ( 2005 ) ; + mon . not .
letters * 356 * , l11-l16 , ( 2005 ) .
s.weinberg , rev.mod.phys .
, * 61 * ( 1989 ) 1 ; + p.j.peebles , b.ratra , rev.mod.phys .
, * 75 * ( 2003 ) 559 ; + t.padmanabhan , phys.rept .
* 380 * ( 2003 ) 235 ; + v.sahni , a.starobinsky , int.j.mod.phys . * a9 * ( 2000 ) 373 .
f. bauer , j. sol , h. tefani , phys .
b688 * ( 2010 ) 269 , arxiv:0912.0677 [ hep - th ] ; + phys
b678 * ( 2009 ) 427 , arxiv:0902.2215 [ hep - th ] ; + f. bauer , j.phys .
* 259 * ( 2010 ) 012083 , arxiv:1010.1106 [ hep - th ] .
+ for a comprehensive exposition , see @xcite .
j. grande , a. pelinson , and j. sol , phys . rev .
* d79 * , ( 2009 ) 043006 , arxiv:0809.3462 [ astro - ph ] ; + j. grande , r. opher , a. pelinson , and j. sol , jcap * 0712 * ( 2007 ) 007 , arxiv:0709.2130 [ gr - qc ] ; + j. grande , a. pelinson , and j. sol , _ dark energy perturbations and a possible solution to the coincidence problem _ , proc . of dark 2009 ,
arxiv:0904.3293 .
tsamis , r.p .
woodard , phys .
* d81 * ( 2010 ) 103509 , arxiv:1001.4929 [ gr - qc ] ; + phys . rev .
* d80 * ( 2009)083512 , arxiv:0904.2368 [ gr - qc ] ; + v. branchina , d. zappala , gen . rel .
grav * 42 * ( 2010 ) 141 , arxiv:0705.2299 [ hep - ph ] ; + mod .
a25 * ( 2010 ) 2305 arxiv:1005.3657 ; + j. beltran jimnez , a. l. maroto , jcap * 0903 * ( 2009 ) 016 , arxiv:0811.0566 [ astro - ph ] ; + phys.lett .
* b686 * ( 2010 ) 175 , arxiv:0903.4672 [ astro-ph.co ] .
s. capozziello , s. carloni and a. troisi , recent res .
astrophys .
* 1 * ( 2003 ) 625 [ arxiv : astro - ph/0303041 ]
. s. m. carroll , v. duvvuri , m. trodden and m. s. turner , phys . rev .
d * 70 * ( 2004 ) 043528 , astro - ph/0306438 .
s. nojiri and s.d .
odintsov , econf * c0602061 * ( 2006 ) 06 [ int .
j. geom .
* 4 * ( 2007 ) 115 ] .
f. bauer , class .
grav . , * 27 * ( 2010 ) 055001 [ arxiv:0909.2237 [ gr - qc ] ] ; + arxiv:1007.2546 [ gr - qc ] . c. wetterich , nucl
* b302 * ( 1988 ) 668 ; + p.j.e .
peebles and b. ratra , astrophys . j. * 325 * ( 1988 ) l17 ; + b. ratra and p.j.e .
peebles , phys . rev .
* d37 * ( 1988 ) 3406 ; + p.g .
ferreira and m. joyce , phys . rev . *
d58 * ( 1998 ) 023503 ; + r.r .
caldwell , r. dave and p.j .
steinhardt , phys .
* 80 * ( 1998 ) 1582 ; + p.j .
steinhardt , l.m .
wang and i. zlatev , phys .
* d59 * ( 1999 ) 123504 ; + v. sahni and l.m .
wang , phys . rev .
* d62 * ( 2000 ) 103517 . | the idea that the cosmological term @xmath0 should be a time dependent quantity in cosmology is a most natural one .
it is difficult to conceive an expanding universe with a strictly constant vacuum energy density , @xmath1 , namely one that has remained immutable since the origin of time . a smoothly evolving vacuum energy density @xmath2 that inherits its time - dependence from cosmological functions @xmath3 , such as the hubble rate @xmath4 or the scale factor @xmath5 , is not only a qualitatively more plausible and intuitive idea , but is also suggested by fundamental physics , in particular by quantum field theory ( qft ) in curved space - time .
to implement this notion , is not strictly necessary to resort to ad hoc scalar fields , as usually done in the literature ( e.g. in quintessence formulations and the like ) .
a `` running '' @xmath0 term can be expected on very similar grounds as one expects ( and observes ) the running of couplings and masses with a physical energy scale in qft .
furthermore , the experimental evidence that the equation of state ( eos ) of the dark energy ( de ) could be evolving with time / redshift ( including the possibility that it might currently behave phantom - like ) suggests that a time - variable @xmath6 term ( possibly accompanied by a variable newton s gravitational coupling too , @xmath7 ) could account in a natural way for all these features . remarkably enough , a class of these models ( the `` new cosmon '' ) could even be the clue for solving the old cosmological constant problem , including the coincidence problem . |
* bistable transition of p5ga hairpin under constant tension :* to study the dynamics of biopolymers under tension , we used the self - organized polymer ( sop ) model , a new class of versatile coarse - grained model that have been proven powerful in describing diverse problems in biomolecular dynamics at single molecule level , which includes proteins , rna , and molecular motors @xcite . among the molecules
studied using sop model , the 22-nucleotide p5ga rna hairpin is the simplest one , whose dynamics was studied under various mechanical conditions such as force clamp and force ramp ( see methods ) , providing microscopic insights into rna hairpins in single molecule force experiments @xcite . here , we performed brownian dynamics simulation of the rna hairpin under varying tensions ( see methods ) . as shown previously @xcite ,
the time trajectories that describe the dynamics of molecular extension ( @xmath2 ) show that the rna hairpin undergoes thermally - induced hopping transitions when the molecule is held around the transition mid - force @xmath3 .
the force - dependent free energy profile @xmath4 with @xmath5 k , calculated from the distribution of molecular extension @xmath6 , becomes bistable at @xmath7 ( @xmath8 ) by forming the native basin of attraction ( nba ) and unfolded basin of attraction ( uba ) at @xmath9 nm and @xmath10 nm , respectively ( fig .
[ fig:1]a ) .
the tension greater ( or smaller ) than @xmath3 , @xmath11 ( or @xmath12 ) tilts the free energy profile towards the uba ( or nba )
. the force - dependent probability of being in the uba @xmath13 is calculated using @xmath14 where @xmath15 , @xmath16 is the extension corresponding to the barrier top ( transition state ) , @xmath17 and @xmath18 ( see fig .
[ fig:1]a ) are the minimum and maximum extensions , respectively .
@xmath19 , fitted to a two state model , @xmath20^{-1}$ ] @xcite , gives @xmath21 pn and @xmath22 nm ( fig .
[ fig:1]b ) . from the time trajectories
, the @xmath23-dependent mean folding ( @xmath24 ) and unfolding time ( @xmath25 ) are also calculated and found to approximately " obey the bell - like relation @xcite ( fig . [ fig:1]c ) .
it is of note that the deviation from linearity on a log folding time scale is due to moving transition state resulting from the underlying folding landscape that deforms under mechanical force @xcite . + * oscillation - induced coherent transition of the p5ga hairpin :* according to theories , the nonlinear and bistable characteristics of the folding landscape of rna hairpin guarantee the manifestation of sr and ra if an oscillatory driving is externally applied to the molecule just like the sr discussed in the context of brownian particle on a bistable potential .
however , given that there are many degrees of freedom whose dynamics experiences continuous dissipation over time , it is not obvious whether a small oscillatory force added to the rna can indeed be sensed along the backbone of biopolymer and renders the folding transition synchronous to the oscillatory force . for proteins and rnas where entropy play dominant role in shaping the underlying free energy landscape , the thermal noise strength ( temperature ) , kept at a physiological condition , is not a good variable to vary .
instead , it would be more appropriate to vary the frequency of oscillatory force .
to this end , we performed the many - body simulations using sop model by applying a small oscillatory force @xmath26 with @xmath27 pn to the 5-end of p5ga hairpin at @xmath28 pn and monitor the change in folding - unfolding dynamics , for various driving frequencies @xmath29 .
our many - body simulations using the sop model of rna hairpin show that although the probability of being in the nba at @xmath28 pn is only @xmath30 % so that the folding is improbable , folding transitions occur more frequently and in coherence to a small oscillatory force at the period of @xmath31 ms ( fig .
[ fig:2]c ) . here , it is noteworthy that @xmath32 for our rna hairpin at @xmath28 pn is similar to the sum of the two transition times in the absence of oscillatory driving , @xmath33 ms ( see fig .
[ fig:1]c at @xmath28 pn ) , a theoretically well - known sr condition for symmetric bistable potential @xcite . +
* mapping p5ga dynamics onto one - dimensional free energy profile :* to explore the details of sr and ra in the dynamics of rna hairpin more extensively , we simplified our many - body simulation using sop model into one - dimensional ( 1d ) simulation by showing that the dynamics of molecular extension @xmath34 from many - body dynamics can effectively be projected onto the dynamics of a quasi - particle under the 1d free energy profile @xmath35 . by using @xmath23-dependent free energy profile @xmath36 where the @xmath37 is directly interpolated from the energy profile at @xmath38 pn obtained by the sop simulation ( fig .
[ fig:1 ] ) , we modeled an associated langevin dynamics of the quasi - particle with a position @xmath34 as follows : @xmath39 . here , @xmath40 is the effective friction coefficient and @xmath41 is the random force that satisfies the fluctuation - dissipation theorem @xmath42 with @xmath43 . folding and unfolding times , determined from time trajectories @xmath34 on @xmath35 , @xmath44 and @xmath45 are in excellent agreements with those from the sop model ( @xmath24 and @xmath25 ) ( fig . [
fig:1]c ) .
in addition , by imposing reflecting boundary condition at @xmath46 nm and @xmath47 nm , which are the minimum and maximum molecular extension , we analytically calculated the mean first passage ( kramers ) times @xmath48 for the folding and @xmath49 for the unfolding .
we set the effective diffusion constant @xmath50 @xmath51 to match @xmath52 ( @xmath53 ) ( solid curves in fig .
[ fig:1]c ) to @xmath24 ( @xmath25 ) at @xmath54 from the simulation results using sop model ( circles in fig .
[ fig:1]c ) . the excellent agreement among differently calculated mean transition times @xmath55 justifies mapping the many - body dynamics of p5ga hairpin under varying @xmath23 onto the dynamics on 1d free energy profile described with molecular extension , @xmath2 . in fact , previous studies showed that @xmath2 is a good reaction coordinate on which both thermodynamics and kinetics of p5ga hairpin are faithfully described as long as @xmath23 does not greatly differ from @xmath3 @xcite .
this enables us to explore sr and ra of the rna folding - unfolding dynamics in a simpler way without compromising the goal of our study . to take into account the effect of external oscillatory force
, we simply added a term @xmath56 ; thus , the stochastic equation describing the dynamics of @xmath34 in the presence of external oscillatory signal reads as follows : @xmath57 + * resonant activation in the folding and unfolding transitions of rna hairpins :* to investigate the effect of oscillatory driving on transition kinetics , we calculated the average folding time @xmath58 as a function of @xmath59 for time trajectories generated using eq.[eq : sv ] at @xmath28 pn for @xmath27 , 2.0 pn or unfolding time @xmath60 at @xmath61 pn for @xmath62 , 1.4 pn ( fig .
[ fig:3 ] ) .
we found that the external oscillatory signal accelerates both folding and unfolding dynamics and maximizes the transition rates @xmath63defined by @xmath64 at @xmath65 ms by about 60 % and 30 % for both folding and unfolding , compared with those in the absence of the oscillatory driving .
the enhancement of folding ( unfolding ) rate could still be called ra , following the theoretical studies on symmetric bistable potential @xcite .
the presence of optimal condition of @xmath66 that minimizes the transition time indicates that there should be a matching condition for two frequencies between conformational fluctuation in folding landscape , which is induced by thermal noise , and the oscillation of input signal . for @xmath67 ,
net driving force provided by the fast oscillating force is essentially zero ; @xmath68 .
in contrast , when @xmath69 , the molecule experiences external tension that slowly varies between @xmath70 and @xmath71 during the transition time @xmath72 depending on the time interval @xmath73 . as a result , for @xmath74
ms , @xmath58 at @xmath28 pn and @xmath27 pn varies between 2 ms and 30 ms ( fig .
[ fig:1]c ) , and the corresponding mean folding time becomes @xmath75 ms as shown in fig .
[ fig:3]a , which is greater than the @xmath76 ms at @xmath66 @xcite .
the extent of ra varies with the amplitude of oscillatory force ( @xmath77 ) ( compare circles and squares in fig . [ fig:3 ] ) .
it is conceivable that @xmath78 where @xmath79 nm at @xmath28 pn ( fig .
[ fig:1 ] ) provides extra work to rectify the random fluctuation of molecule in uba .
therefore , the ratio of @xmath78 to free energy barrier @xmath80 from the uba to nba , i.e. , @xmath81 should determine the enhancement of folding rate at the ra condition . + * stochastic resonance in the refolding dynamics of rna :* folding time distributions @xmath82 ( fig .
[ fig:4]b ) calculated from time trajectories ( fig . [ fig:4]a ) are useful measure for characterizing sr . at fast driving @xmath83 ms ( fig .
[ fig:4]b-1 ) , @xmath82 decays exponentially with a decay time 10.5 ms , which is consistent with the folding time in the absence of driving , @xmath84 ( see fig . [
fig:1]c ) .
but at @xmath85 ms near the resonant condition , @xmath82 displays multiple peaks whose periodicity corresponds to the driving period of input signal , which is approximately equal to @xmath84 ( fig . [ fig:4]b-2 ) .
the multiple peaks are superimposed on the above - mentioned exponentially decaying contribution ( fig .
[ fig:4]b-1 ) @xcite . at a slow driving @xmath86 ms ( fig .
[ fig:4]b-3 ) , the @xmath82 has a broad and small peak around 25 ms and shows only a weak signature of coherence . the degree of synchronization is quantified by calculating the @xmath87 value corresponding to the area under the first peak of @xmath82 , which is defined as @xmath88 where the integration is done over a time interval between @xmath89 and @xmath90 that defines the first peak in the @xmath82 @xcite . for example , for @xmath83
ms , the @xmath82 exponentially decreases with no peak , thus @xmath91 .
in contrast , @xmath82s with @xmath85 ms and 40.96 ms ( fig .
[ fig:4]b-2 and [ fig:4]b-3 ) have the first peaks around 7 ms and 25 ms , respectively , leading to non - zero @xmath92 .
@xmath93 calculated for the data @xmath82 of our 1d simulation for p5ga hairpin ( filled circles in fig .
[ fig:4]c ) is non - monotonic with @xmath59 and is maximized around @xmath94 ms .
notably , @xmath95 ) rise sharply for @xmath96 ms ( fig .
[ fig:4]c ) and the @xmath32 that maximizes @xmath87 differs from the @xmath97 that minimizes @xmath58 .
the @xmath92 for the unfolding with @xmath61 pn and @xmath27 pn , by calculating unfolding time distribution @xmath98 , are also depicted with empty circles in fig .
[ fig:4]d .
we find that the unfolding is also synchronized with an optimal driving period around @xmath99 ms ( see fig .
[ fig:1]c ) .
thus , rna folding ( unfolding ) becomes maximally coherent at @xmath100 . due to asymmetric shape of potential leading to @xmath101 ,
the sr condition for folding becomes @xmath102 , which is worth comparing with the conventional sr condition @xmath103 for symmetric bistable systems .
there are many alternative measures to demonstrate the manifestation of sr .
one measure , directly amenable to experiment , is the power spectrum @xmath104 where @xmath105 denotes ensemble and long time averages for the trajectory @xmath34 @xcite .
the @xmath106 shows a narrow sharp peak with the frequency @xmath107 ms@xmath108 ( see the inset of fig .
[ fig:4]b-2 ) .
in addition , spectral behavior of dynamic susceptibility defined @xmath109 , which also becomes optimal at @xmath110 ms@xmath108 , can be used to quantify the response of system to an external driving force . +
* biopolymers can filter the resonance frequency from the spectrum of an external signal :* due to the multitude of cellular processes nonequilibrium noise in the cell is decomposed into various frequencies . to check
if our model can discriminate the resonant frequency using sr , we performed the 1d simulation for rna hopping transitions under @xmath28 pn by adding oscillatory forces consisting of different angular frequencies @xmath111 where @xmath112 ms@xmath108 , @xmath113 ms@xmath108 and @xmath114 ms@xmath108 ; here , the optimal folding period for the sr , @xmath115 ms , is included in the @xmath116 .
the @xmath82 for uniform amplitudes of the additive forces with @xmath117 pn ( fig .
[ fig:5 ] ) displays an oscillatory pattern with multiple peaks , suggesting that the folding is synchronous with the oscillatory force with the period @xmath32 . even for the case of nonuniform amplitudes , in which @xmath118 and @xmath119 have greater values than @xmath120 , the signature of synchronization to the oscillatory driving with @xmath32 is retained ( fig . [ fig:5 ] )
furthermore , even when we choose @xmath77 randomly from the values distributed between 0 and 2 pn at every time setp , we still found the @xmath82 with multiple peaks , which suggests that the p5ga system is capable of filtering the signal with frequency of @xmath121 out of the nonequilibrium noise . + * experimental demonstration of sr in biomolecular dynamics :* although the results discussed above are based on the numerical simulation that applies mechanical force directly acting to the ends of rna hairpin without mediating polymer handle as in actual optical tweezers experiment , our findings on sr and ra conditions for folding transition should not change qualitatively .
as quantitatively discussed by hyeon _
_ @xcite as well as recognized by others @xcite , the presence of handle polymers simply slows down the transition kinetics by pinning the motion of the two end points of biopolymers connected to the handles .
furthermore , in all practical situations , the conformational transition time of biopolymers is much slower than the tension propagation time @xcite .
therefore , even in the presence of handles sr is still realized at the optimal frequency , corresponding to the sum of the folding and unfolding transition times that are measured in the presence of handles .
many - body simulation using sop model of rna hairpin in the presence of 20 nm handles attached to the two ends of hairpin confirms the realization of sr in the presence of handles ( see _ si text _ and fig.s1 ) . for the purpose of experimental demonstration of sr using biopolymers
, it is recommended that the experiment is conducted at the transition mid - force ( @xmath3 ) , at which the folding and unfolding transitions are the most frequently observed @xcite ( fig.s2 ) . in this case , sr condition is expected to be formed around twice the mean hopping transition time ( @xmath122 ) .
originally , the sr refers to the phenomenological observation that a system under an optimal thermal noise can amplify a weak external signal if the system is nonlinear .
this paper restates the sr and ra condition by focusing more on the transition dynamics of biopolymers rather than the amplification of signal . furthermore , unlike the many studies that consider the sr using noise strength , i.e. , temperature , as a variable , the present study focuses on the bona - fide sr " with respect to the driving frequency . for biologically relevant environment ,
whose temperature range should be narrowly defined , it is more appropriate to consider the effect of driving frequency on sr . folding and unfolding transitions of biopolymers enhanced under and synchronized to an external periodic signals , which is clearly demonstrated in our simulations , suggest that as long as the sr condition is met , conformational sampling of biopolymers in vivo occurs more frequently than in vitro , in response to the spectrum of athermal noise in the cell .
true manifestation of sr and ra in the context of biopolymer dynamics in vivo should hinge on existence of the actual periodic frequency that is in resonance to the transition dynamics of biopolymer .
it would be of great interest to investigate the power spectra of cellular noise and how biological dynamics , for instance , conformational transitions of riboswitch , are modulated in the presence of athermal signals from other dynamic components such as motion of ribosome translocation and associated gene regulation @xcite .
* model :* we used the sop model to simulate the dynamics of p5ga rna hairpin under tension . in sop model ,
the effective hamiltonian is defined as follows @xcite : @xmath123\nonumber\\ & + \sum_{i=1}^{n-3}\sum_{j = i+3}^{n}\epsilon_{h}\left[\left(\frac{r_{i , j}^{o}}{r_{i , j}}\right)^{12}-2\left(\frac{r_{i , j}^{o}}{r_{i , j}}\right)^{6}\right]\delta_{i , j}\nonumber\\ & + \sum_{i=1}^{n-3}\sum_{j = i+3}^{n}\epsilon_{l}\left(\frac{\sigma}{r_{i , j}}\right)^{6}(1-\delta_{i , j})+\sum_{i=1}^{n-2}\epsilon_{l}\left(\frac{\sigma^{\ast}}{r_{i , i+2}}\right)^{6 } , \label{eq : sop}\end{aligned}\ ] ] where we modeled each nucleotide as an interaction center ; thus @xmath124 .
the first term restrains the bond length between the neighboring beads by using the finite extensible nonlinear elastic potential with @xmath125 kcal/(mol@xmath126@xmath127 ) , and @xmath128 nm . here , @xmath129 is the distance between neighboring interaction centers @xmath130 and @xmath131 , and @xmath132 is the distance in the native structure .
lennard - jones potential was employed to stabilize the base pair present in the native structure in the second term with @xmath133 kcal / mol , but make the non - native base pair repulsive in the third term , so that the model can easily reach the native topology without being trapped into non - native conformations . if @xmath130 and @xmath134 are in contact in the native state @xmath135 ; otherwise @xmath136 .
@xmath137 is the distance between @xmath130 and @xmath134 interaction centers that satisfy @xmath138 and @xmath135 . for
non - native pairs , the strength and range of interaction were set by @xmath139 kcal / mol and @xmath140 nm . the last term involving the repulsion with @xmath141 nm maintains a certain angle between @xmath130 and @xmath142 interaction centers . +
* simulation :* we performed brownian dynamics simulation by numerically integrating the discretized equation of motion @xmath143 where @xmath144 and the random force @xmath145 satisfies @xmath146 and it is assumed that the probability distribution of the noise in the random force is gaussian , i.e. , @xmath147\propto\exp\left(-\int d\tau\frac{[\xi_{\alpha , i}(\tau)]^2}{4\zeta k_bt}\right)$ ] and @xmath148 with @xmath149 .
we probed the folding - unfolding dynamics of rna hairpin by applying tension @xmath23 to the 3-end of the hairpin while fixing the position of 5-end at temperature @xmath5 k. also , we chose the friction coefficient @xmath150 with @xmath151 ps corresponding to the value that a nucleotide whose size @xmath152 nm feels in water environment whose viscosity is @xmath153 cp @xcite .
we applied the tension @xmath23 between @xmath154 pn ( fig . [ fig:1 ] ) to the molecule and added small sinusoidal forces of @xmath26 with varying @xmath155 values ( fig .
[ fig:2 ] ) .
we set the simulation time step @xmath156sec and recorded data at every @xmath157 time steps ( @xmath158sec ) .
after equilibrating the rna in native state for @xmath159 @xmath160 , we applied tension @xmath23 or @xmath161 and recorded the @xmath34 of each trajectory for @xmath162 ms .
this work was supported by brain - korea 21 program and national core research center for systems bio - dynamics to w.s . and national research foundation of korea grant ( 2010 - 0000602 ) to c.h .
we thank korea institute for advanced study for providing computing resources .
43 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , , , , , , _ _ ( , ) , ed .
, , , * * , ( ) . , , , , , , , * * , ( ) , issn . , , , , * * , ( ) . , , , , , , , , * * , ( ) . , , , * * , ( ) . , , , , * * , ( ) . , * * , ( ) . , , , , * * , ( ) . , , , , * * , ( ) . , , , , , * * , ( ) . , , , , , , * * , ( ) . , , , , , , * * , ( ) . , , , , , * * , ( ) . , , , , * * , ( ) . , , , , , * * , ( ) . , *
* , ( ) . ,
* * , ( ) . , , , * * , ( ) . , ,
, , * * , ( ) . , , , * * , ( ) . , , , ,
, , * * , ( ) . , , , * * , ( ) .
, , , , * * , ( ) . , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) , http://stacks.iop.org/0305-4470/14/i=11/a=006 . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) .
, , , * * , ( ) . , * * , ( ) . , , , , * * , ( ) , http://stacks.iop.org/1478-3975/7/i=3/a=034001 . , , , , * * , ( ) . , , , , , , , * * , ( )
. resonant activation ( ra ) in folding and unfolding transitions of p5ga .
( a ) mean folding time @xmath163 of p5ga under @xmath28 pn as a function of @xmath59 with @xmath27 pn ( filled squares ) and @xmath164 pn ( filled circles ) .
the transition times are minimized at @xmath165 ms .
( b ) mean unfolding time @xmath166 of p5ga under @xmath61 pn as a function of @xmath59 with @xmath62 pn ( empty squares ) and @xmath27 pn ( empty circles ) .
the transition times are minimized at @xmath165 ms.,width=604 ] sr in the folding and unfolding transitions of p5ga under time dependent tension @xmath167 .
( a ) time trajectories and ( b ) folding time distributions @xmath82 with @xmath28 pn and @xmath27 pn at @xmath83 ms , @xmath168 ms , and @xmath169 ms are shown . the synchronization ( sr condition )
occurs around @xmath170 ms , as shown in ( a-2 ) .
the inset in ( b-2 ) depicts the power spectrum @xmath106 , showing the sharpest peak at @xmath171 ms@xmath108 .
( c ) mean folding time @xmath163 and measure of coherence @xmath87 as a function of @xmath59 .
( d ) mean unfolding time @xmath166 and measure of coherence @xmath92 as a function of @xmath59 with @xmath61
pn and @xmath27 pn . note that the sr and ra conditions are made at different @xmath59s in ( c ) and ( d ) .
, width=453 ] filtering of sr condition .
shown are the @xmath82 of p5ga under @xmath172 at varying combinations of @xmath173 with @xmath112 ms@xmath108 , @xmath113 ms@xmath108 , @xmath114 ms@xmath108 and @xmath28 pn . for filled symbols , we fix @xmath174 pn , corresponding to the amplitude for the optimal frequency @xmath175 for sr , and vary the other amplitudes to be @xmath176 pn ( filled squares ) , 1.7 pn ( filled circles ) and 2.0 pn ( filled triangles ) , respectively .
the starred symbols represent the cases where @xmath118 , @xmath177 and @xmath119 are taken randomly between 0 and 2 pn every time step to make a nonequilibrium noise @xmath178 .
notably , the folding transition filters the optimal driving period despite other drivings with larger or transient amplitudes.,width=604 ]
* stochastic resonance in the presence of handles :* to study whether the sr is still realized when the weak signal is indirectly exerted to the system through handle polymers , just like in the real experimental condition in laser optical tweezers ( lots ) , we simulate folding dynamics of an rna hairpin by attaching stiff handles to the two of the p5ga rna hairpin modeled with sop representation ( eq.[2 ] in the main text ) .
each handle is modeled by using the following energy potential @xcite : @xmath179 the first term denotes the chain connectivity , with an equilibrium distance @xmath180 nm and spring constant @xmath181 .
the second term is the bending penalty potential with bending modulus @xmath182 pn@xmath183 nm , which gives rise to the persistence length @xmath184 nm .
we take @xmath185 for the length of handle , which amounts to the contour length of @xmath186 nm .
[ fig : s1 ] shows time trajectories of the extension of the entire system , which includes the extension of two handles , at @xmath28 pn and @xmath27 pn , for three different driving periods .
the molecular extension @xmath34 ( left ) and the total extension @xmath187 ( right ) are depicted in the red and black axes , respectively . as shown in the trajectories ,
the phase of @xmath187 is compatible with that of @xmath34 . even in the presence of handles ,
the sr is realized at an optimal period @xmath188 ms ( fig .
[ fig : s1]b ) where the folding transitions are synchronized to the frequency of external driving .
the other trajectories under relatively fast ( fig .
[ fig : s1]a ) and slow ( fig .
[ fig : s1]c ) driving show random transitions regardless of the drivings .
as it is discussed in the main text and in the ref.@xcite , the presence of the handles pins the overall molecular motion and slows down the transition kinetics of the molecule sandwiched between the handles .
we have estimated that the transition dynamics with handles with @xmath189 nm and @xmath186 nm tend to be almost about six - fold slower @xcite than the kramers times ( @xmath190 ms for folding and @xmath191 ms for unfolding ) without handles which we have considered in the text , and found sr ( fig .
[ fig : s1]b ) at an optimal driving period @xmath192 ms which is approximately sum of the kramers times .
+ * stochatstic resonance at the the transition mid - force :* to demonstrate the sr using lots experiment , it would be most straightforward to perform the experiment around the external force value near the transition mid - force , at which the transition from one basin to another is most frequent . to show sr under the transition mid - force ( @xmath38 pn ) , we have performed the sop simulations for hopping ( folding and unfolding ) dynamics of a p5ga under @xmath193 . fig .
[ fig : s2 ] shows time trajectories of the p5ga extension @xmath34 for three different driving periods , at @xmath27 pn .
sr is clearly shown in the fig .
[ fig : s2]b at an optimal driving period @xmath194 ms . in this case
, the optimal driving period is found to be around twice the mean hopping transition timee ( @xmath195 ms , see also fig.1c where @xmath1962 ms ) , in consistent with the _ time - scale matching condition _ as has extensively been discussed in the sr community @xcite . the other trajectories , under relatively faster ( fig .
[ fig : s2]a ) and slower ( fig .
[ fig : s2]c ) drivings which deviate from the optimal frequency , show random transitions . | in addition to thermal noise , which is essential to promote conformational transitions in biopolymers , cellular environment is replete with a spectrum of athermal fluctuations that are produced from a plethora of active processes . to understand the effect of athermal noise on biological processes
, we studied how a small oscillatory force affects the thermally induced folding and unfolding transition of an rna hairpin , whose response to constant tension had been investigated extensively in both theory and experiments .
strikingly , our molecular simulations performed under overdamped condition show that even at a high ( low ) tension that renders the hairpin ( un)folding improbable , a weak external oscillatory force at a certain frequency can synchronously enhance the transition dynamics of rna hairpin and increase the mean transition rate .
furthermore , the rna dynamics can still discriminate a signal with resonance frequency even when the signal is mixed among other signals with nonresonant frequencies .
in fact , our computational demonstration of thermally induced resonance in rna hairpin dynamics is a direct realization of the phenomena called stochastic resonance ( sr ) and resonant activation ( ra ) .
our study , amenable to experimental tests using optical tweezers , is of great significance to the folding of biopolymers in vivo that are subject to the broad spectrum of cellular noises .
the cell is a highly crowded and dynamic place with millions of biomolecules and their interactions @xcite . only recently
the notion that the self - assembly and function of biomolecules in vivo could be very different from those in isolation has been highlighted in terms of macromolecular crowding and confinement @xcite .
however , the implication from dynamics in vivo should not necessarily be limited to the confinement or crowding effect . besides the thermal noise , which is essential for activated internal motions and folding dynamics of biomolecules , of particular note are the nonequilibrium athermal signals produced from a plethora of active processes and collective actions among cellular constituents .
for example , ribosome translocate along mrna strand less than every 0.1 s per codon @xcite ; molecular motor , kinesin , takes step along the microtubule approximately every @xmath0 ms @xcite ; collective beating of cilia occurs approximately every @xmath1 min @xcite .
how these nonequilibrium signals or noises affect fundamental processes in biology , particularly the folding dynamics of biopolymers @xcite , is the main concern of this paper . to address this issue
, we study the effect of a weak oscillatory force on the transition dynamics of stretched rna hairpins as a test case . in recent years a combination of experiments , theoretical arguments and simulations have been used to extensively study the conformational dynamics of biopolymers under mechanical control @xcite . among others ,
paramount precision of optical tweezers , which can both exert piconewton force to molecules and probe structural changes in nm scale , not only allowed us to quantify thermodynamic and kinetic properties of the molecule at single molecule level @xcite , but also provided unprecedented chances to manipulate the process of life @xcite . to emulate nonequilibrium athermal noise in vivo and
study its effect on the conformational dynamics of biopolymers , high precision force instruments seem to hold good promise ; yet , no experiment to date has explored this issue .
given that the dynamics of proteins and nucleic acids occur at overdamped fluctuating media , it may be surprising to learn that conformational transitions of biopolymers could become enhanced or synchronized to an externally exerted weak oscillatory signal .
however , according to the notion of stochastic resonance ( sr ) , sensitivity of a system to external signal can be improved under thermal fluctuation , so that the system can discriminate the weak information carrying signals @xcite .
theoretically , it has been shown for a single brownian particle in a bistable potential that synchronization of barrier crossing events to an externally driven oscillation occurs not only at an optimal noise strength but also at resonance frequency @xcite .
furthermore , at another oscillation frequency that differs from the frequency of synchronization , the mean first passage time over the barrier is minimized , which is called the resonant activation ( ra ) @xcite .
hence , it is possible to demonstrate the sr and ra in the conformational transition dynamics of biopolymers like rna hairpins , whose folding landscape is bistable near transition mid - force , by properly tuning the frequency of additive oscillatory force . since the pioneering work of benzi _ et al . _
@xcite , there have been efforts to extend the idea of sr to a vast range of physical and biological phenomena @xcite .
however , sr often mentioned in terms of macroscopic systems such as recurrence of the ice age cycle and sensory systems for higher order organisms , especially the latter of which is wired with complex biochemical networks , is always a subject of debate because the inherent complexity of a system consisting of a formidable number of variables often makes the origin of such resonance - like phenomena inconclusive .
although phenomenological modeling of system dynamics using sr is still possible , there seems to be little convincing proof that the biological systems actually use sr in the cellular level @xcite . in this aspect ,
recent theoretical demonstrations of sr and ra of a polymer chain in a confined space has more direct relevance to the biology @xcite .
here we show that sr and ra can indeed be manifested in self - organization process of biopolymers by using molecular simulation .
although careful investigation is further required to confirm whether the sr and ra are actively utilized in the cell for its biological functions , it should be straightforward to demonstrate sr in the folding and unfolding dynamics of biopolymers by using single molecule force experiments .
we provide a simple experimental protocol for the demonstration . |
measurements of primordial density perturbations are consistent with adiabatic initial conditions , for which the ratios of neutrino , photon , baryon , and dark - matter number densities are initially spatially constant .
the simplest inflationary models predict adiabatic fluctuations @xcite .
isocurvature perturbations , on the other hand , are fluctuations in the relative number densities of different species .
they are produced in topological - defect models for structure formation @xcite , multi - field inflationary models , curvaton models @xcite , in which two different fields drive inflation and generate curvature perturbations , and in simple inflationary models if the dark matter is composed of axions @xcite .
cmb temperature anisotropies limit the contribution of baryon isocurvature perturbations ( fluctuations in the baryon - to - photon ratio ) @xcite and cdm isocurvature perturbations ( fluctuations in the dark - matter to photon ratio ) @xcite to the total perturbation amplitude @xcite .
the recent _ planck _ cmb results limit the cdm isocurvature fraction to be @xmath4 of the total perturbation amplitude .
it is therefore surprising that perturbations in the baryon density can be almost arbitrarily large , as long they are compensated by dark - matter perturbations such that the total - nonrelativistic - matter density remains unchanged @xcite .
these compensated isocurvature perturbations ( cips ) obey @xmath5 where @xmath6 , @xmath7 , and @xmath8 are fractional energy density perturbations in the cold dark matter , baryons , and photons , respectively , while @xmath9 and @xmath10 are the homogeneous dark matter and baryon densities .
cips induce no curvature perturbation at early times , and they therefore leave the photon density and thus large - angle cmb fluctuations homogeneous at linear order . curvaton models for inflation may generate cips @xcite , with amplitudes approaching the regime detectable by the proposed epic mission @xcite , and other inflationary models @xcite could generate even larger cip amplitudes .
recent theoretical ideas @xcite connecting the baryon asymmetry and dark - matter density could also have implications for cips . in any case
, our principal motivation in studying cips is curiosity : can we determine empirically , rather than simply assume , that the primordial baryon fraction is homogeneous and traces the dark matter ?
cips induce baryon motion through baryon - pressure gradients , but these motions occur only at the baryon sound speed . the resulting anisotropies would be imprinted on the baryonic sound horizon , at @xmath11 @xcite .
existing measurements at low redshift constrain the cip perturbation amplitude to be @xmath12 @xcite , while more sensitive proposed measurements of 21-cm absorption during the cosmic dark ages are a way off in the future @xcite .
more recently , it has been shown that cips would modulate the cmb anisotropies produced by adiabatic perturbations , both by inducing anisotropies in the optical depth to reionization @xcite , and more dramatically , by changing the silk damping length of the cmb in regions of sky containing a cip @xcite .
this modulation would induce a specific pattern of higher order - correlations in the temperature and polarization anisotropies , analogous to those induced by variations of other cosmological parameters @xcite , and by weak gravitational lensing @xcite .
this signature can be exploited to construct estimators for the cip perturbation @xcite .
cips generated in curvaton models for the primordial density fluctuation @xcite , lower - energy inflationary models @xcite , and perhaps other scenarios , are within the range of detectability for a cosmic - variance - limited cmb polarization experiment .
our motivation for studying cips , however is here one of curiosity : if pre - existing limits are so permissive , it behooves us to actually check if baryons and dark matter trace one another in the early universe !
here , we use wmap 9-year temperature maps to search for cips . our cip estimator is based on the full non - gaussian trispectrum of the observed cmb multipole moments , and can be used to perform either a model - independent reconstruction of the cip power spectrum @xmath13 , or to measure the amplitude of a scale - invariant spectrum of cips .
we impose a @xmath0-confidence upper limit of @xmath1 to the amplitude of a scale - invariant spectrum of cips , as well as model - independent constraints of @xmath14 to the root - mean - squared ( rms ) amplitude of the cip power spectrum at angular scales in the range @xmath15 , where @xmath16 is the multipole index of the cip .
we show that secondary contractions of the trispectrum contribute negligibly to the estimator , at least for wmap experimental parameters .
we show that known sources of non - gaussianity , such as gravitational lensing and unresolved point sources , do not provide a significant bias for our estimates of the cip power spectrum .
the same methodology which we have used here could be applied to the _ planck _
data , and has the potential to significantly improve on the constraints above .
we begin in sec .
[ cip_rederive ] with a derivation of the temperature anisotropies induced by cips . in sec .
[ esec ] , we present our cip estimator , based on that of ref .
@xcite , but generalized to be run on a partial sky - map with realistic noise properties .
we account for bias and estimator normalization using analytic estimates and monte carlo simulations . in sec .
[ bam ] we present our results and compare with the forecasted wmap sensitivity , and we conclude in sec . [ conclusions ] . throughout this work we use a fixed , flat @xmath17cdm cosmology consistent with the wmap-9 @xcite power spectrum , given by @xmath18 , @xmath19 , @xmath20 , adiabatic spectral index @xmath21 , reionization optical depth @xmath22 , and adiabatic scalar power spectrum normalization @xmath23 .
in refs . @xcite and @xcite , it was shown that a cip would induce off - diagonal correlations between cmb anisotropy multipole moments .
we rederive these results using a clearer method here , applying the line - of - sight ( los ) formalism of ref . @xcite .
this approach is more readily generalized to cips with wavelength smaller than the thickness of the surface of last - scattering , and is useful in computing the cip bispectrum . before launching into the formalism , we review the physical origin of cip - induced cmb correlations . as noted in refs .
@xcite , cips have no initial potential perturbations ( like other isocurvature modes ) , but also have no initial radiation pressure gradients . in linear theory
, flows then begin at the baryon sound speed , but are only effective in transferring fluctuations to the photons on very small scales ( @xmath24 ) .
even when baryons do begin to evacuate initial density fluctuations ( which would yield a net potential perturbation as the cip evolved ) , these flows will be diffusion damped , slowing down the growth of these small - scale potential perturbations .
this intuition is confirmed by running the boltzmann code camb @xcite with a cip initial condition , and noting that the induced cmb temperature anisotropy is negligible .
the fluctuating baryon fraction in the presence of a cip would lead to an inhomogeneous redshift of reionization ( when the first sources turn on ) , leading to a fluctuating optical depth and a distinct patchy reionization signal in the cmb @xcite .
more dramatically , however , if cips are in fact primordial , they will lead to a spatial modulation of coefficients in the early - time ( tight - coupling era ) equations of motion for fluid perturbations , altering observed cmb anisotropies in a detectable way @xcite .
[ cip_rederive ] consider a compensated isocurvature perturbation ( cip ) along the line of sight @xmath25 , which gives local changes in the baryon and cdm energy densities given by @xmath26 , \nonumber \\ \omega_c & \rightarrow \omega_c - \omega_b \delta(\hat{n}).\end{aligned}\ ] ] a positive ( negative ) value of @xmath27 corresponds to a more ( less ) baryon - loaded plasma , which decreases ( increases ) the sound speed and thus decreases ( increases ) the physical acoustic horizon .
the multipole index @xmath28 of the first cmb acoustic peak thus increases ( decreases ) as @xmath27 increases ( decreases ) , as shown in fig .
[ intuit ] ( from ref .
@xcite ) , generated using expressions in ref .
the other acoustic peak locations behave similarly .
cmb temperature anisotropies are suppressed on angular scales @xmath29 due to diffusion damping . using the expressions in ref .
@xcite and the camb @xcite code , we evaluate @xmath30 and show the results in the top right panel of fig .
[ intuit ] .
we see that , as photons diffuse over smaller distances , as a result of higher local baryon density in the presence of a cip with positive @xmath27 , the transition to exponential damping of cmb anisotropies occurs at higher @xmath31 . in the bottom panel of fig .
[ intuit ] ( from ref .
@xcite ) , we show the visibility functions @xmath32 for @xmath33 different values of @xmath27 ; @xmath34 is the optical depth due to thomson scattering and @xmath35 is the redshift .
the peak of the visibility function is the redshift @xmath36 , at which most cmb photons last scatter . in the presence of a positive ( negative ) @xmath27 cip , decoupling occurs later ( earlier ) due to higher ( lower ) baryon density .
the effects described above all result in modifications to the cmb power spectrum .
one can therefore imagine constructing an estimator for spatial variations in @xmath27 by forming localized power spectrum estimates and fitting for @xmath27 relative to a fiducial model based on the full - sky power spectrum .
we will do this in sec .
[ esec ] , however first we must quantify the effect of cips on the local power spectrum , which we do using the los formalism in the following sections .
the calculation of cmb anisotropies is greatly simplified using the line - of - sight ( los ) approach , introduced in ref . @xcite . at spatial location @xmath37 , conformal time @xmath38 , and for photon momentum direction vector @xmath39 ( where @xmath40 ) , the photon temperature perturbation @xmath41 obeys the equation ( derived from the boltzmann equation ) @xmath42,\]]where @xmath43 is a vector
whose entries are the fluid density / velocity perturbations ( as well as higher - order moments of the distribution function , for neutrinos ) and metric fluctuations characterizing the system , and @xmath44 is a linear differential operator which maps @xmath43 to a source term for temperature perturbations .
defining the lhs of this equation as the operator @xmath45 ( for boltzmann ) , and taking a fourier transform , we obtain @xmath46=\mathcal{d}_{\vec{k}}(\hat{p},\eta)[\vec{u}_{\vec{k}}(\eta)],\label{evo_eq}\]]where @xmath47 is a matrix operator in fourier space ( as opposed to a differential operator ) . for the usual adiabatic mode , @xmath48 , where @xmath49 is a time evolution operator mapping the initial potential perturbation @xmath50 to the solution for the fluid variables at subsequent times .
the components of the matrix operator @xmath47 ( which maps the fluid and metric variables to the observed temperature perturbation ) and vector - valued function @xmath51 are laid out in detail in refs . @xcite and others , so we use the operator notation to keep things simple and general .
( [ evo_eq ] ) may be formally integrated to obtain @xmath52\phi_{\vec{k}}\label{befit}\ ] ] where the source function is @xmath53=\mathcal{d}_{\vec{k}}(\hat{p},\eta)\vec{f}_{\vec{k}}(\eta)$ ] , @xmath54 is the conformal time today and @xmath55 .
it turns out that the source function depends on @xmath56 and @xmath39 only through @xmath57 ( as a polynomial in @xmath57 ) and @xmath58 , and so eq . ( [ befit ] ) may be integrated by parts to obtain @xmath59\phi_{\vec{k}},\]]in terms of a different source function @xmath60 $ ] , whose terms are specified in refs .
going back to real space , following suit with an inverse spherical harmonic transform to derive multipole coefficients , and expanding the exponential using a fourier - bessel series , we obtain @xmath61j_{l}[k(\eta_{0}-\eta)]\end{aligned}\ ] ] where @xmath62 is a spherical bessel function of index @xmath31 . the familiar cmb angular power spectrum
may then be obtained : @xmath63where the @xmath33-dimensional potential fluctuation power spectrum @xmath64 is defined by @xmath65 we now generalize the los solution to compute the off - diagonal temperature correlations induced by a cip .
+ + the operators @xmath66 and @xmath49 depend on the cosmological parameters and thus also on the amplitude @xmath25 of a cip . in the presence of a cip
, the real - space evolution equation will read @xmath67=\frac{d\mathcal{d}(\hat{p},\eta)[\vec{u}(\vec{y},\eta)]}{d\delta}\delta(\vec{y}).\label{modulated}\ ] ] as discussed above , on the scales of interest , the cip amplitude is frozen in time . in principle , it still has dependence on the conformal time along a photon trajectory , if the mode wavelength is shorter than the integration interval .
for modes of large angular scale @xmath68 , however , this radial dependence may be neglected , and @xmath69 .
fourier transforming eq .
( [ modulated ] ) , performing a spherical harmonic expansion of @xmath70 , we see that the evolution equation for @xmath71 is of the same form as eq .
( [ evo_eq ] ) , but with a source term that is a linear superposition of source terms like this in eq .
( [ evo_eq ] ) .
since this is a linear system , solutions may be superimposed .
fourier - bessel expanding and performing an inverse spherical harmonic transform of the solution , we obtain the perturbation to the los solution induced by a cip : @xmath72 expressed in terms of the familiar wigner-3j symbols @xcite .
as discussed in the previous section , a compensated isocurvature perturbation induces a small fluctuation in the cmb temperature which is proportional to the primordial potential @xmath50 . if we consider a fixed realization of isocurvature perturbations @xmath73 , the effect of cips is to introduce `` statistical anisotropy '' into the cmb , which is manifest as off - diagonal elements in the covariance matrix of the cmb fluctuations .
these are given by @xmath74 here @xmath75 is the usual cip - free temperature power spectrum .
the quantity @xmath76 is a weight function associated with the cips and @xmath77 the derivative power spectra of eq .
( [ toffdia_b ] ) are evaluated using the camb code , using numerical methods described in ref .
@xcite , including a spatial modulation in the optical depth to reionization , @xmath34 , as well as the much larger effect from physics near recombination .
we see above that a fixed cip realization breaks the diagonality of the covariance matrix of the @xmath78 in a very specific way , yielding a unique statistical signature which can be used to reconstruct the @xmath73 realization .
a similar derivation can reproduce the off - diagonal polarization correlations obtained in ref . @xcite . of course
, in reality we do not have a fixed cip realization to observe .
if the universe as a whole has no preferred orientation then the cip perturbations @xmath73 are themselves random variables , with some statistically isotropic distribution .
if we assume that @xmath73 are independent of the primordial fluctuations and gaussian , then they are completely characterized by their power spectrum @xmath79 . in this more realistic picture
, the first distinctive statistical signature of cips appears in the cmb `` trispectrum '' , or connected 4-point function .
the connected part of the 4-point function is zero for purely gaussian fluctuations , but in the presence of cips it becomes non - zero .
following ref .
@xcite , the connected 4-point function must take the form @xmath80 where @xmath81 is known as the trispectrum , and we have used the glebsch - morgan coefficient ^l_1 m_1 l_2 m_2 l_l_3 m_3 l_4 m_4 m = ( ccc l_1 & l_2 & l + m_1 & m_2 & -m ) ( ccc l_3 & l_4 & l + m_3 & m_4 & m ) .
symmetry of the four multipoles further requires that the trispectrum may be encoded as @xmath82.\label{bigtri}\end{gathered}\ ] ] the first term @xmath83 is referred to as the primary contraction of the trispectrum , while the final two terms are known as secondary contractions . the primary contraction introduced by cips
is given simply by p^l_1 l_2_l_3 l_4 ( l ) = c_l^ w^_l_1 l_2 l w^_l_3 l_4 l. [ eqn : cip_primary_contraction ] observationally , we will see that the picture of statistical anisotropy introduced initially above is still a useful ansatz ; as the universe shows us only a single realization of both @xmath50 and @xmath73 , a fixed @xmath73 model is experimentally indistinguishable from one in which the @xmath73 are gaussian random variables @xcite .
we can construct an estimator for the cip power spectrum utilizing the connected trispectrum which it induces .
if we are interested in a single mode of this power spectrum @xmath79 , then following ref .
@xcite , for example , an edgeworth expansion of the cmb likelihood leads to the following optimal ( minimum - variance ) trispectrum estimator @xmath84 { \bar c}^{tt}_{l_3 m_3 , l_4 m_4 } \\ - \left [ 2 \bar{t}_{l_1 m_1 } \bar{t}_{l_3 m_3 } - { \bar c}^{tt}_{l_1 m_1 , l_3 m_3 } \right ] { \bar c}^{tt}_{l_2 m_2 , l_4 m_4 } \\ - \left [ 2 \bar{t}_{l_1 m_1 } \bar{t}_{l_4 m_4 } - { \bar c}^{tt}_{l_1 m_1 , l_4 m_4 } \right ] { \bar c}^{tt}_{l_2 m_2 , l_3 m_3 } \big ) , \label{eqn : triest}\end{gathered}\ ] ] where @xmath85 is a normalization , @xmath86 are a set of so called `` inverse - variance filtered '' multipoles determined from the data map ( we will discuss how these are obtained in more detail shortly in sec . [ sec : estimator : filtering ] ) , and @xmath87 is their covariance matrix . for any modern cmb experiment , which measures thousands of modes , the covariance matrices @xmath87
have trillions of elements and are impossible to work with directly .
it is , however , computationally tractable to evaluate the expression above ( which distills these covariance matrices down to a single number ) using monte carlo simulations .
we will rewrite eq . in a form which makes the details of this evaluation clearer , and also makes connection to the discussion of statistical anisotropy in the previous section .
we begin by introducing the `` quadratic estimator '' @xmath88 , which is a function of two inverse - variance filtered temperature multipoles as |_lm [ |t_lm^(1 ) , ( ccc l & l & l + m & m & m ) w^_l l l |t_lm^(1 ) |t_lm^(2 ) .
if we were considering the cip realization to be fixed , this quadratic estimator is precisely the quantity which is required for optimal estimation of @xmath73 , following the formalism of quadratic maximum likelihood ( qml ) estimators @xcite .
this estimator is related to the actual cip multipole moment @xmath73 by an overall normalization .
we will often find it useful to work with such un - normalized quantities , which we will denote with overbars .
the reason is that for these maximum - likelihood estimators , the normalization is formed from the inverse of the estimator fisher matrix .
for this reason , we have denoted the un - normalized estimator @xmath89 with an overbear , in analogy to the inverse variance filtered temperature multipoles .
this quantity is also the variance of the estimator , and so these un - normalized estimates are effectively inverse - noise weighted .
this makes the un - normalized estimators useful for estimating other parameters , such as the overall amplitude of a scale - invariant spectrum of cips , as we shall see below .
we note that @xmath73 may be evaluated rapidly ( with computational cost @xmath90 using fast spherical harmonic transforms ) in position space as the product of two filtered maps @xmath91 = \\
\int d\hat{n } y_{lm}^{*}(\hat{n } ) \bar{t}^{(1)}(\hat{n } ) s^{(2)}(\hat{n } ) + [ ( 1 ) \leftrightarrow ( 2 ) ] , \end{gathered}\ ] ] where the filtered maps themselves are given by @xmath92where @xmath93 or @xmath94 as appropriate . using this notation , we rewrite the @xmath79 estimator as @xmath95 , where |c_l^ = c_l^| | - d_l^| | . [
eqn : clpphat ] the naive _ un - normalized _ power spectrum estimate @xmath96 is given by @xmath97 - \bar{\delta}_{lm } [ \bar{t}_{lm}^{(g ) } , \bar{t}_{lm}^{(g ) } ] \right)^ * \\ \times \left ( \bar{\delta}_{lm } [ \bar{t}_{lm } , \bar{t}_{lm } ] - \bar{\delta}_{lm } [ \bar{t}_{lm}^{(f ) } , \bar{t}_{lm}^{(f ) } ] \right ) \bigg>_{g , f } , \label{eqn : cl } \end{gathered}\ ] ] and the `` disconnected noise bias '' estimate is given by @xmath98 \bar{\delta}_{lm } [ \bar{t}_{lm } , \bar{t}_{lm}^{(g ) } ] \\
+ 2 \bar{\delta}_{lm } [ \bar{t}_{lm}^{(g ) } , \bar{t}_{lm}^{(f ) } ] \bar{\delta}_{lm } [ \bar{t}_{lm}^{(g ) } , \bar{t}_{lm}^{(f ) } ] \bigg>_{g , f}. \label{eqn : dl}\end{gathered}\ ] ] for both spectra , the ensemble average is taken over two sets of statistically independent monte carlo simulations of @xmath99 , labelled @xmath100 and @xmath101 .
this expression could equivalently be written with just a single set of simulations , however evaluating it as is done here reduces the susceptibility to numerical noise in the evaluation procedure . rather than a single mode @xmath79
, we may be interested in the amplitude @xmath102 of a fiducial power spectrum @xmath103 , for which the corresponding optimal estimator is = _
l = l_min^l_max ( 2l+1 ) normalization , not to be confused with @xmath104 .
note the use of the un - normalized @xmath105 here rather than @xmath106 . in the case of full - sky coverage with homogeneous noise levels ,
the inverse - variance filter is given simply by @xmath107 where @xmath108 is the beam- and pixel - transfer function , @xmath109 is the observed map indexed by pixel @xmath110 and the filter function @xmath111 is given by f_l = . for the more realistic case of a beam - convolved sky map with inhomogeneous noise ,
the construction of @xmath86 is more involved . to obtain @xmath86 from a set of wmap sky maps we use an inverse - variance filter which properly accounts for sky - cuts and the inhomogeneity of the map and its noise levels by solving the equation @xmath112 where the matrix @xmath113 is given by _
lm , lm _ p , .
[ eqn : cinvmat ] here the @xmath110 denotes a map pixel , @xmath114 denotes a particular channel map ( usually a given frequency band ) , @xmath115 is the observed ( beam convolved ) sky map in pixel @xmath110 at frequency @xmath114 , the pointing matrix @xmath116 gives the value of the spherical harmonic at the center of pixel @xmath110 , convolved with the appropriate beam+pixel transfer function @xmath117 .
@xmath118 represents is the noise covariance matrix .
we use a diagonal noise covariance in pixel space with @xmath119 where @xmath120 is a map - dependent noise level and @xmath121 is the number of observations of pixel @xmath110 for map @xmath114 .
@xmath122 is a map which is zero for masked pixels and unity elsewhere .
effectively , it sets the noise level to infinity for masked pixels , ensuring that they are ignored in the rest of the analysis .
there are known noise correlations in the wmap maps @xcite on scales @xmath123 , though we do not incorporate them in our filtering .
these modes are signal - dominated in the temperature maps which we use , and so this neglect of large - scale noise correlations does not affect our analysis .
we evaluate the matrix inverse of eq .
using conjugate descent with the fast multigrid preconditioner of ref .
generally , we will find it useful to work with individual wmap frequency maps ( using only a single entry for @xmath114 in the equations above ) , however for our final results we will combine all off the useable bands . for analytical purposes , it is useful to have a diagonal approximation to the full - blown inverse variance filter . for this
we use the @xmath111 function at the top of this section , estimating the average noise power spectrum of the channel - combined , beam - deconvolved map as c_l^tt , noise ^-1 , where @xmath124 is the unmasked sky fraction .
we now analytically derive an approximate normalization @xmath125 of our estimator .
we will eventually use but correct for deviations from this normalization using monte carlo simulations .
the analytical treatment , however , is also useful as a tool to explicitly compute the estimator bias induced by other physical sources of non - gaussianity at the trispectrum level .
consider full - sky coverage , with homogeneous noise , in which case the filtered cmb covariance matrix @xmath87 is diagonal .
if we ensemble average over cmb and cip realizations in eq . , we find that ( by construction ) @xmath106 is directly proportional to the connected four point function of @xmath78 @xmath126 here @xmath111 are the filter functions of the previous section . for generality , so that we can estimate possible biases to our estimator from known non - gaussian sources such as unresolved point sources and gravitational lensing by large - scale - structure ,
let us consider the response of this estimator to a trispectrum with the primary form ^x p^l_1 l_2_l_3 l_4 ( l ) = c_l^xx w^x_l_1 l_2 l w^x_l_3 l_4 l. this covers both the cip trispectrum of eq .
, as well as the trispectra due to gravitational lensing of the cmb by large - scale - structure , and unresolved point sources which pollute the map .
the trispectrum for cmb lensing is given by @xcite ^ p^l_1 l_2_l_3 l_4 ( l ) = c_l^ w^_l_1 l_2 l w^_l_3 l_4 l , where @xmath127\nonumber\\+&\left\{l_{1}\leftrightarrow l_{2}\right\}.\end{aligned}\ ] ] the trispectrum associated with point source shot - noise is given by @xcite ^s^4 p^l_1 l_2_l_3 l_4 ( l ) = s^4 w^s^2_l_1 l_2 l w^s^2_l_3 l_4 l , where @xmath128 is the kurtosis of the point sources and the weight function is given by w^s^2_l_1l_2l=. we discuss effective amplitudes for @xmath129 using the wmap point source masks in appendix [ ps_sec ] .
both point sources and gravitational lensing represent potential sources of bias in the reconstruction of @xmath130 and the estimator @xmath131 .
it is important to either verify that these sources of bias are negligible , or to construct appropriately debiased estimators ( as done , for example , for estimators of patchy reionization optical depth @xmath34 , in ref .
@xcite , or for cmb lensing estimators in @xcite ) .
we propagate the trispectrum described by the primary contraction through to eq .
, obtaining the ensemble - averaged contribution of the physical effect @xmath132 ( which can denote cips , weak gravitational lensing , or point sources ) to the cip trispectrum estimator @xmath131 , .
|_x = p_l^x + s_l^x .[hatexp ] here @xmath133 and @xmath134 capture the contributions from the primary and secondary contractions of any trispectrum respectively .
the primary term is given by @xmath135 , where the response function @xmath136 is given by _
[ eqn : resp ] the secondary term @xmath137 is more complicated , involving wigner-@xmath138j symbols which are numerically intensive to calculate .
we estimate these contributions using flat - sky expressions to evaluate the secondary contractions , given by @xcite @xmath139 where @xmath140 and @xmath141 are fourier space angular multipole vectors , @xmath142 and the flat - sky weight function is given by w^(_1 , _ 2 ) = c_|_1|^t , dt + c_|_2|^t , dt .
the flat - sky weight functions for cmb lensing and point source shot noise are @xcite @xmath143 + ( \bl_1 \leftrightarrow \bl_2 ) , \nonumber \\
w^{s^2}(\bl_1,\bl_2)&=1.\end{aligned}\ ] ] a special case of the bias calculation is for @xmath144 , which yields the response of @xmath105 to cip fluctuations themselves and therefore the normalization of the estimator .
we will see that for cips , the primary term is dominant .
we can therefore use eq . as an approximate , analytical normalization for @xmath106 , with _
l^approx = ^-1 [ eqn : normlapprox ] the corresponding approximate normalization for the fiducial power spectrum amplitude estimator @xmath131 is given by ^approx = ^-1.[napprox ] we will ultimately correct this normalization for our @xmath131 estimates using monte carlo simulations which implicitly include the contribution from the secondary contractions above , as well as cut - sky effects beyond simple @xmath145 scaling . for a scale - invariant power spectrum
we find that the approximation of eq .
is accurate to better than @xmath146 . in fig .
[ fig : cldd_sources ] we show estimates of the contribution of primary and secondary contractions of cips for a scale - invariant power spectrum , as well as weak lensing of the cmb and point sources . using fig .
[ fig : cldd_sources ] , we see that the primary and secondary contractions of the lensing trispectrum yield negligible contributions to the cip estimator , compared with the estimator noise power spectrum .
we also see that the point source ( discussed at length in appendix [ ps_sec ] ) trispectrum contributes negligibly to the cip estimator .
we also see that secondary contractions of the cip trispectrum are negligible compared with primary contractions . as a function of scale , for wmap v - band noise and beam . the estimator noise power spectrum @xmath147 is shown as black solid . for comparison ,
we plot a scale invariant power spectrum @xmath148^{-1}$ ] with an amplitude comparable to our one standard - deviation constraint ( thick black dashed ) .
we also calculate the contribution from the secondary contractions for this cip power spectrum ( @xmath149 , black long - dashed ) . the expected contribution from the primary and secondary contractions of the cmb lensing trispectrum
are shown as blue dashed and blue dotted respectively .
in addition , we plot a pessimistic ( larger than the maximum in appendix [ ps_sec ] ) estimate of the v - band contamination from unresolved point sources , with @xmath150 ( red dash - dotted ) .
the primary and secondary contributions from the shot - noise trispectrum have the same shape , and so we only plot their sum .
an analogous analysis in the q and w - bands yields the same conclusion , and so we omit those curves here . ]
we analyze the `` foreground - reduced '' cmb temperature anisotropy maps , with associated beam and noise characterization , from the final , 9-year wmap data release @xcite .
we use the q , v , and w - band maps , with nominal central frequencies of 40 , 60 , and 90ghz respectively .
to remove known galactic and bright point source contamination we use the `` kq85 '' temperature analysis mask produced by the wmap team . to compute the mean - field and disconnected noise bias of our estimator [ eqs .
( [ eqn : dl ] ) and ( [ eqn : cl ] ) ] , we use cmb+noise simulations of the data to compute the monte carlo averages ( @xmath151 realizations per wmap band ) .
we simulate a cmb temperature realization @xmath152 with power spectrum @xmath153 given by a flat @xmath17cdm cosmology consistent with the wmap power spectrum @xcite , with parameters given in the introduction , computed using the camb @xcite code .
we apply a beam and pixel transfer function to these temperature multipoles , and then project them onto an @xmath154 healpix map with a harmonic transform . to each pixel in this map , we add a gaussian noise contribution with variance given by eq . .
these simple simulations are then filtered in the same way as the real data [ that is , using eq .
( [ tfil ] ) ] to produce the simulated quantities @xmath155 and @xmath156 of eqs . and .
to test the normalization of our cip estimator , we also require non - gaussian simulations with @xmath157 .
we form these non - gaussian cmb simulations with power spectrum @xmath153 and trispectrum given by eq . using three pieces @xmath158 .
@xmath159 is drawn from a gaussian distribution with power spectrum @xmath160 ( the choice of @xmath160 is somewhat arbitrary for our purposes , we will use @xmath161 .
then we add the term @xmath162 which ensures that our simulations have the desired trispectrum , with @xmath163 \\
\times \left [ \sum_{l'm ' } y_{l'm'}(\hatn ) a_{l'm ' } \frac{c_{l'}^{\rm t , dt}}{c_{l'}^{\rm aa } } \right].\end{gathered}\ ] ] here @xmath73 are gaussians drawn for a fiducial power spectrum @xmath164 .
we add a final gaussian term @xmath165 which sets the overall power spectrum of our non - gaussian simulations to be @xmath153 .
the @xmath165 are drawn from the power spectrum @xmath166 , with @xmath167 in fig .
[ fig : cldd_specs ] we plot estimated cip power spectra @xmath106 from separate analyses of q , v , and w - band maps , as well as an inverse - variance weighted combination of the @xmath33 maps ( denoted qvw ) .
we use the approximate analytical normalization of eq . .
this neglects the secondary cip trispectrum contractions , which we see in fig .
[ fig : cldd_sources ] are negligible , as well as cut - sky effects beyond simple @xmath145 scaling . for the question we are asking in this figure ( `` is the measured cip power non - zero with statistical significance ? '' )
the accuracy of the normalization is not important , as corrections to eq .
show up in both the signal and noise , and thus leave the significance of any signal or limit unaffected .
we see no evidence for anomalous @xmath79 power in the individual q , v , or w - band maps .
we may now obtain limits to specific models for the cip power spectrum , as well as model - independent constraints to the cip power spectrum at different multipole numbers @xmath16 .
we now place limits to the amplitude of a scale - invariant power spectrum of cip fluctuations , which in multipole space projected onto a two - dimensional sky is @xcite c_l^ = c_l^,(fid ) , [ eqn : scalinv ] for some amplitude @xmath102 . for @xmath168 ,
corresponding realizations of @xmath27 would have fluctuations with amplitude of approximately unity .
for @xmath169 , the scaling in eq .
( [ eqn : scalinv ] ) actually turns over to @xmath170 as the wavelength of the cip falls below the thickness of the last - scattering surface .
these modes have very low signal - to - noise ratio using present data and thus contribute negligibly to our estimator of @xmath102 .
it is thus adequate to use eq .
( [ eqn : scalinv ] ) .
the results of these fits are shown in fig .
[ fig : cldd_amp_hist ] , and are also consistent with zero .
note that a value of @xmath171 is non - physical , but allowed by our measurement as we have not imposed a prior on the positivity of @xmath102 . for our measurement ,
we use eq . , @xmath172 , and @xmath173 .
we use the normalization @xmath174 , where @xmath175 is given in eq . .
although this normalization is only approximate , it is adequate to answer the question of whether or not the measured value of @xmath102 deviates from zero with statistical significance .
later , when limits to @xmath102 are set , we account for corrections to this normalization .
our constraint on @xmath131 comes mainly from very low-@xmath16 modes of @xmath105 .
the noise variance of @xmath106 at @xmath176 goes as @xmath177 , which falls off more slowly than the scale - invariant spectrum , and so most of our sensitivity to @xmath178 comes from a small number of low-@xmath16 modes .
our estimator probability density function ( pdf ) is thus slightly non - gaussian ( as can be seen from the histograms of fig .
[ fig : cldd_amp_hist ] ) , for the same reason that the distribution of low-@xmath16 power - spectrum estimates is non - gaussian .
this also occurs for trispectrum - based estimators of the amplitude of local - type non - gaussianity @xmath179 @xcite .
we find best - fit values in q , v , and w - bands separately , as well as for the qvw combination , of @xmath180 , @xmath181 , @xmath182 , @xmath183 . in fig .
[ fig : cldd_amp_hist ] , these values are compared with the distribution of @xmath131 values for simulations in which the real value @xmath184 .
we see that our best - fit values are consistent with the null hypothesis at @xmath0-confidence .
to establish upper limits to @xmath102 , we need the function @xmath185 , the probability distribution of @xmath102 values , given the estimated value @xmath131 . assuming flat priors for the ` data ' ( @xmath131 ) and theoretical parameter ( @xmath102 ) , we have @xmath186 . from our monte carlo simulation of the @xmath187 case , we know that the width of the non - gaussian function @xmath188 depends noticeably on the real value of @xmath102 , and so it is important to properly determine @xmath188 . to this end
, we conduct a suite of monte carlo simulations , estimating @xmath189 for multiple realizations ( @xmath151 per band and per real @xmath102 value ) , at the series of true parameter values @xmath190 . the results for @xmath188
are shown in fig .
[ fig : cip_pdf ] . to compute an upper limit to @xmath102
, we use a model for @xmath185 [ and thus @xmath188 ] .
the model for @xmath188 is as follows : a @xmath192 distribution @xmath193 is fit to the simulated random variable @xmath194 for @xmath195 . at each @xmath16 and for any given @xmath102 value , this distribution has two free parameters , the number of degrees of freedom @xmath196 , and a scale parameter @xmath197 such that @xmath198 .
the estimator used is given by = _
l = l_min^l = l_max_l.a model pdf @xmath199 is thus obtained by convolution over @xmath193 .
this model is shown as the dashed curve in fig .
[ fig : cip_pdf ] and is clearly a reasonable fit to the data .
alternatively , we may just fit @xmath193 for the @xmath184 case and then adjust the scale parameter @xmath200\right\}$ ] .
an analogous convolution then yields the semi - analytic function @xmath201 , which is comparably accurate to @xmath199 in fitting the estimator pdf from the simulation , and readily interpolated to obtain quantities of interest .
our final upper limit is computed by starting with @xmath201 , obtaining @xmath185 using bayes s theorem with a flat prior on @xmath102 in the simulated domain , computing the cumulative probability function ( cdf ) associated with @xmath185 , and interpolating this function , incrementing @xmath102 upwards to define a @xmath0-confidence interval . like the measurement of @xmath189 , these simulations use @xmath202 , and thus calibrate the relationship between @xmath102 and @xmath189 , accounting for errors induced in the approximation @xmath203 .
the resulting upper limits to @xmath102 are thus correctly normalized .
given the observed multi - band value @xmath204 , we compute @xmath205 [ and thus @xmath206 , and find that @xmath207 at @xmath0 confidence .
measurements of the baryon fraction in galaxy clusters @xcite impose the limit @xmath208 to the amplitude of a scale - invariant spectrum of cips @xcite .
the larger value quoted in ref .
@xcite is obtained using a different definition of @xmath102 , which we have appropriately rescaled here .
we may also restate our result in terms of @xmath209 , the rms fluctuation in the baryon - dark - matter density ratio on galaxy cluster scales , using our definition of @xmath102 and eq .
( 1 ) of ref .
@xcite ( with the associated wave - number range @xmath210 ) to obtain @xmath211 we obtain @xmath212 , to be compared with the result directly obtained from the cluster baryon fraction , @xmath213 . as predicted in the forecasts of ref .
@xcite , the cip - sensitivity of wmap is comparable to measurements of the baryon fraction in galaxy clusters .
the physics of this cmb probe of cips , however , is completely different than that used in the cluster probe , and offers an important and truly primordial constraint on the amplitude of cips .
[ limsec ] aside from placing limits to a scale - invariant cip power spectra , we may also place limits to the cip amplitude at different angular multipole numbers @xmath16 , without reference to a fiducial model .
.upper limits ( @xmath0-confidence level ) @xmath214 to the angular power spectrum of cips as a function of multipole number @xmath16 , as recovered from an inverse - variance weighted sum of the @xmath33 wmap 9-year cmb maps , and the monte carlo simulations described in sec .
[ limsec ] . also shown is the corresponding angular scale in degrees of @xmath215 and the @xmath0-confidence upper limit to the rms cip amplitude on that angular scale , @xmath216 .
[ cols="^,^,^,^,<",options="header " , ] -confidence level ) to the rms cip amplitude as a function of multipole number @xmath16 , as recovered from an inverse - variance weighted sum of the @xmath33 wmap 9-year cmb maps , and the monte carlo simulations described in sec .
[ limsec].,width=336 ] the pdf for @xmath201 used to obtain the above upper limits is derived by fitting a @xmath192 distribution to the pdf for each @xmath217 , assuming no covariance between @xmath217 for different @xmath16 , convolving over different multipole moments , and applying bayes s theorem .
this _ ansatz _ fits the monte carlo histograms well , and so we may use the same simulation to impose limits to @xmath13 independently for each scale @xmath16 , without making any assumptions about the functional form of the cip power spectrum .
we thus model the histogram of @xmath217 values using the function @xmath218 , where @xmath219\right\}$ ] . here
@xmath220 is the best - fit value of the @xmath192 scale - parameter @xmath197 for the null hypothesis simulation . for all multipoles in the range ,
@xmath221 , this pdf fits the monte carlo results . applying bayes s theorem on an @xmath16-by-@xmath16 basis with the resulting pdf , we obtain @xmath222 . using the estimated values @xmath217 from the minimum - variance - weighted qvw combination , we obtain @xmath0-confidence upper limits to @xmath13 in the @xmath16-range of interest .
the results are shown in table i and fig . [ flimfig ] , where the corresponding angular scale and rms cip amplitude upper limit is shown .
we thus see that the relative mass fractions of baryons and dark matter can vary by no more than @xmath14 on angular scales of @xmath2 at the sls .
this limit is forced upon us by cmb data alone , with no reliance on galactic abundance measurements , galaxy physics , or knowledge about the nature of cosmic reionization .
compensated isocurvature perturbations ( cips ) break the usual assumption that the primordial cosmic baryon fraction is spatially homogeneous . here , building on past work searching for cips using variations in the cluster baryon fraction , we use wmap 9-year maps of the cmb temperature fluctuations to search for cips .
we apply a trispectrum - based estimator to show that the amplitude of a scale - invariant spectrum of cips must obey the constraint @xmath223 , combining all @xmath33 cmb - dominated wmap bands .
independent of the functional form of the cip power spectrum , the rms cip amplitude must be no greater than @xmath14 on scales of @xmath2 when thomson scattering freezes out at @xmath224 .
as we saw , the estimator for @xmath13 is an appropriately filtered @xmath225-pt function of the data , probing correlations that are absent under the null hypothesis .
gaussian cips uncorrelated with the underlying adiabatic fluctuation can then be thought of as inducing a non - gaussian signal in the cmb , with vanishing @xmath33-pt function and non - vanishing @xmath225-pt function .
it may be that cips are not an independent gaussian random field ( as assumed here ) , but rather , as in some curvaton models @xcite , correlated with the usual adiabatic fluctuations . in this case , schematically , the induced temperature fluctuation is @xmath226 , as before , but now , since @xmath227 for some correlation coefficient , @xmath228 .
the resulting cmb @xmath33-pt function is then @xmath229 .
curvaton - inspired cips will thus cause an additional non - zero non - gaussianity , a @xmath33-pt correlation or bispectrum .
we note that the cip estimator described here could also be applied to the recently released nominal mission data of the _ planck _ satellite , for which we forecast a sensitivity improvement factor of @xmath230 in @xmath209 or @xmath102 @xcite .
future ground - based cmb polarization data ( such as actpol @xcite and sptpol @xcite ) also have the potential to significantly improve the sensitivity of the cip probe described here .
it is claimed in ref .
@xcite that cips would lead to negligible changes in the total matter power spectrum , and would thus not be detectable using galaxy surveys .
it stands to reason , however , that star formation efficiency , cooling , and other processes are strongly dependent on the baryon density , and that the statistics of actual galaxies would be drastically altered in the presence of cips . using semi - analytic models of galaxy formation ,
this could be tested .
large fluctuations in the jeans / cooling masses in the presence of a cip could also change the dynamics of reionization , leading to an additional probe .
we will explore these possibilities in future work .
[ conclusions ] we acknowledge useful conversations with d. n. spergel , c. dvorkin , and k. m. smith .
dg was supported at the institute for advanced study by the national science foundation ( ast-0807044 ) and nasa ( nnx11af29 g ) .
mk was supported by the department of energy ( doe sc-0008108 ) and nasa ( nnx12ae86 g ) .
part of the research described in this paper was carried out at the jet propulsion laboratory , california institute of technology , under a contract with the national aeronautics and space administration .
this work was supported by a cita national fellowship at mcgill .
some of the results in this paper have been derived using healpix @xcite .
this work was begun during the 2011 winter conference `` inflationary theory and its confrontation with data in the planck era '' at the aspen center for physics ( nsf grant 1066293 ) .
the authors are very grateful for the hospitality of the aspen center .
points sources which are bright enough to be detected are removed using the wmap analysis mask used in this paper .
there is , however , a population of residual sources which are too faint to be directly detected individually the maps , but which contribute non - negligibly to the cmb temperature power spectra , as well as higher order statistics like the trispectrum which we probe here . when modelling these sources , it is useful to consider an effective limiting flux threshold @xmath231 , above which all sources are assumed to be detected , and below which none are . at wmap frequencies and sensitivity ,
the dominant residual contribution is from a small number of radio sources directly below the detection threshold .
the statistics of this population are dominanted by the 1-point ( or `` shot - noise '' ) component .
if the differential number density of sources with flux @xmath232 is denoted @xmath233 , the unresolved shot noise contribution to the reduced trispectrum is given by @xcite @xmath234 where @xmath235 for experimental channel frequency @xmath114 and @xmath236 is a conversion factor from jansky flux units to thermodynamic @xmath237 , given by @xmath238\nonumber \\&\simeq&\frac{\left(e^{x}-1\right)^{2}}{x^{4}e^{x } } \frac{\mu{\rm k}}{99.27~{\rm jy}~{\rm sr}^{-1}}. \end{aligned}\ ] ] the physical constants here are the planck constant @xmath239 , speed - of - light @xmath240 , boltzmann constant @xmath241 , and cmb mean temperature @xmath242 .
we use two models for @xmath233 , to be sure that our conclusions about the contribution of shot noise to the trispectrum are robust . the first ( model 1 " hereafter )
is given in ref .
@xcite , and is parameterized by @xmath243 , where @xmath244 , and @xmath245 . the second ( model 2 " hereafter ) , is obtained from tables of source counts in ref .
we fit a power law of the form used above to these tables , and obtain @xmath246 and @xmath247 for wmap q - band source counts . for the v - band source counts , we obtain @xmath248 and @xmath249 .
w - band source counts are not given in these tables , but we use the q and v - band fits ( cases w@xmath250 and w@xmath251 ) as a model for w - band source counts . flux cuts and conversion functions @xmath236 appropriate for the w band , however , are used for the w - band trispectrum estimates .
we now use eq .
( [ trispec_ps ] ) to obtain estimates for the unresolved point - source trispectra .
we use a flux cut of @xmath252 , as it was found in ref .
@xcite that this value reproduces observed wmap point - source bispectra . for the q - band ,
we obtain @xmath253 and @xmath254 . for v - band data with model @xmath255
, we obtain @xmath256 . for v - band data with model @xmath257
, we obtain @xmath258 . for w - band data with model 1 , we obtain @xmath259 . for model 2 , case w@xmath250
, we obtain @xmath260 . for model 2 , case w@xmath251
, we obtain @xmath261 .
we plot a pessimistic estimate of the point source contamination in fig .
[ fig : cldd_sources ] for the wmap v - band .
as we know from the preceding discussion , the w - band contribution is negligible in comparison .
the q - band has the most worrisome amplitude of possible point source bias , however it is still more than an order of magnitude below the scale - invariant power spectrum shape at the low ( @xmath262 ) multipoles and amplitudes ( @xmath263 ) which we probe most sensitively .
we do not see evidence for any departure from zero in our model - independent @xmath79 measurement , which would be characteristic of point source contamination . | primordial isocurvature fluctuations between photons and either neutrinos or non - relativistic species such as baryons or dark matter are known to be sub - dominant to adiabatic fluctuations .
perturbations in the relative densities of baryons and dark matter ( known as compensated isocurvature perturbations , or cips ) , however , are surprisingly poorly constrained .
cips leave no imprint in the cosmic microwave background ( cmb ) on observable scales , at least at linear order in their amplitude and zeroth order in the amplitude of adiabatic perturbations .
it is thus not yet empirically known if baryons trace dark matter at the surface of last scattering . if cips exist , they would spatially modulate the silk damping scale and acoustic horizon , causing distinct fluctuations in the cmb temperature / polarization power spectra across the sky : this effect is first order in both the cip and adiabatic mode amplitudes . here
, temperature data from the wilkinson microwave anisotropy probe ( wmap ) are used to conduct the first cmb - based observational search for cips , using off - diagonal correlations and the cmb trispectrum .
reconstruction noise from weak lensing and point sources is shown to be negligible for this data set .
no evidence for cips is observed , and a @xmath0-confidence upper limit of @xmath1 is imposed to the amplitude of a scale - invariant cip power spectrum .
this limit agrees with cip sensitivity forecasts for wmap , and is competitive with smaller scale constraints from measurements of the baryon fraction in galaxy clusters .
it is shown that the root - mean - squared cip amplitude on @xmath2 scales is smaller than @xmath3 ( depending on the scale ) at the @xmath0-confidence level .
temperature data from the _ planck _ satellite will provide an even more sensitive probe for the existence of cips , as will the upcoming actpol and sptpol experiments on smaller angular scales . |
the techniques based on the factorization method , which try to identify the class of hamiltonians whose spectral problem can be algebraically solved , have attracted for years people s attention . in these lecture notes
we will elaborate a brief survey of this subject ; since we shall show that the factorization method , supersymmetric quantum mechanics ( susy qm ) , and intertwining technique are equivalent procedures for generating new solvable potentials departing from an initial one , these names will be indistinctly used to characterize the technique we are interested in . for a collection of books and review articles concerning factorization method and related subjects the reader can seek @xcite-@xcite .
i would like to emphasize that these lecture notes on susy qm are a sequel of those delivered by myself in 2004 at the latin - american school of physics @xcite .
for people wanting to have a global view of the way our group addresses the subject , and to know more details of our achievements , i recommend to read both , this article and @xcite , because they complement to each other . from a historical viewpoint
, it is nowadays accepted that the factorization method was introduced for the first time by dirac in 1935 in order to derive algebraically the spectrum of the harmonic oscillator @xcite . in coordinates
such that @xmath0 the oscillator hamiltonian admits the following factorizations : @xmath1 from these equations it is straightforward to derive the intertwining relationships , @xmath2 which lead to the commutators ruling the standard heisenberg - weyl algebra : @xmath3 = - a , \quad [ h , a^+ ] = a^+ , \quad [ a , a^+ ] = 1.\end{aligned}\ ] ] these expressions and the fact that @xmath4 is positive definite allow to generate the eigenfunctions @xmath5 and corresponding eigenvalues @xmath6 of @xmath4 .
the second important advance in the subject was done by schrdinger in 1940 , who realized that the factorization method can be as well applied to the radial part of the coulomb problem @xcite . in this case
the explicit factorizations in appropriate units read : @xmath7 as in the previous case , these factorizations imply some intertwining relationships : @xmath8 from which it can be derived once again the radial eigenfunctions @xmath9 and eigenvalues @xmath10 for the hamiltonian @xmath11 . in a set of works starting from 1941 , infeld and collaborators put forward the method @xcite , culminating their research with a seminal paper in which it was classified a wide set of schrdinger - type hamiltonians ( indeed four families ) solvable through the factorization method @xcite . the main tool for performing this classification
were the factorizations @xmath12 , \quad a_m^+ = \frac{1}{\sqrt{2 } } \left [ - \frac{d}{dx } + \beta_m(x)\right],\end{aligned}\ ] ] which lead straightforwardly to the intertwining relationships @xmath13 with these ingredients , infeld and hull were able to derive elegantly the eigenfunctions and eigenvalues for the hierarchy of hamiltonians @xmath14 @xcite .
moreover , the very technique allowed them to determine the several forms of the potentials @xmath15 admitting the factorization treatment . in this way , after @xcite the idea that the factorization methods was exhausted started to dominate ; indeed , at some time it was thought that for a potential @xmath16 to be solvable through factorization , it just has to appear in the infeld - hull classification .
revealing himself against this idea , in 1984 mielnik performed a simple generalization of the factorization method @xcite .
he proposed to look for the most general first - order differential operators @xmath17 factorizing the harmonic oscillator hamiltonian in the way : @xmath18 , \quad b^+ = \frac{1}{\sqrt{2}}\left [ -\frac{d}{dx } + \beta(x ) \right].\end{aligned}\ ] ] mielnik was able to determine the most general form of @xmath19 .
moreover , it turns out that the product @xmath20 supplies a new hamiltonian @xmath21 hence , the following intertwining relationships are valid : @xmath22 which interrelate the eigenfunctions of the harmonic oscillator hamiltonian @xmath4 with those of @xmath23 and vice versa .
mielnik s work was a breakthrough in the development of the factorization method , since it opened new ways to explore exactly solvable potentials in quantum mechanics .
in particular , his generalization was quickly applied to the radial part of the coulomb problem @xcite .
the important resultant expressions for that system become : @xmath24 they allow to generate the eigenfunctions and eigenvalues of @xmath25 departing from those of @xmath11 .
notice that , in the same year the connections between the factorization method with the darboux transformation and witten s supersymmetric quantum mechanics were established by andrianov et al @xcite and nieto @xcite respectively . in 1985
sukumar @xcite pushed further mielnik s factorization by applying it to general one - dimensional potentials @xmath16 and arbitrary factorization energies @xmath26 in the way @xmath27 , \
\ a^+ = \frac{1}{\sqrt{2 } } \left [ - \frac{d}{dx } + \beta(x)\right].\end{aligned}\ ] ] these relationships imply that @xmath28 which allow to generate the eigenfunctions and eigenvalues of @xmath23 from those of @xmath4 . along the years
, several particular cases and applications of the generalized factorization of @xcite have been performed successfully @xcite-@xcite . up to this point , however , the factorization operators , becoming the intertwiners in the equivalent formalism , were differential operators of first order .
a natural generalization , posed in 1993 by andrianov and collaborators @xcite , requires that eq.([intertwining ] ) be valid but @xmath29 are substituted by differential operators of order greater than one . through this higher - order case ,
it is possible to surpass the restriction arising from the first - order method that we can only modify the ground state energy of the initial hamiltonian . in 1995
bagrov and samsonov proposed the same generalization , and they offered an alternative view to this issue @xcite . on the other hand
, our group arrived to the subject in 1997 @xcite-@xcite , although there were several previous works related somehow to this generalization @xcite-@xcite .
it is worth to mention explicitly some important contributions that members of our group have made to the factorization method and related subjects . *
generation of susy partner of the harmonic oscillator @xcite * construction of coherent states for susy partners of the harmonic oscillator @xcite and for general one - dimensional hamiltonians @xcite . *
determination of susy partners of the radial oscillator potential @xcite * characterization of systems ruled by polynomial heisenberg algebras and connection with susy partners of the oscillator @xcite * analysis of susy partners of the radial coulomb problem @xcite * implementation of susy transformations involving complex factorization ` energies ' @xcite-@xcite * study of the confluent algorithm for second - order susy transformations and applications @xcite * generation of susy partners of the pschl - teller potentials @xcite-@xcite * application of susy techniques to the lam and associated lam potentials @xcite-@xcite it is important to mention that some other groups have elaborated the same subject from different viewpoints , e.g. , @xmath30-fold supersymmetry by tanaka and collaborators @xcite-@xcite , hidden nonlinear supersymmetries by plyushchay et al @xcite-@xcite , etc .
in this section we are going to show that the factorization method , intertwining technique and supersymmetric quantum mechanics are equivalent procedures for generating new solvable potentials departing from an initial one . the most straightforward way to see
this equivalence is starting from the technique which involves just first - order differential intertwining operators .
suppose we have two schrdinger hamiltonians @xmath31 which are intertwined by first - order differential operators @xmath32 in the way @xmath33 , \quad a_1^+ = \frac{1}{\sqrt{2}}\left[-\frac{d}{dx } + \alpha_1(x)\right].\end{aligned}\ ] ] by employing that , at the operator level , it is valid @xmath34 where @xmath35 is a multiplicative operator when acting on a generic wavefunction @xmath36 , the calculation of each member of the first intertwining relationship in eq.([inter1gen ] ) leads to @xmath37 thus : @xmath38 by plugging the @xmath39 of eq.([gv1 ] ) in eq.([derricc ] ) and integrating the result we get : @xmath40 .\end{aligned}\ ] ] this nonlinear first - order differential equation for @xmath41 is the well - known _
riccati equation_. in terms of a _ seed schrdinger solution _
@xmath42 such that @xmath43 it turns out that : @xmath44 which is the initial stationary schrdinger equation .
these equations allow to establish the link between the intertwining technique and the _ factorization method _ , since it is valid : @xmath45 suppose now that @xmath46 is a given solvable hamiltonian , with known eigenfunctions @xmath47 and eigenvalues @xmath48 , namely , @xmath49 in addition , let us take a nodeless mathematical eigenfunction @xmath42 associated to @xmath50 in order to implement the intertwining procedure . the equations ( [ inter1gen ] ) and the resulting factorizations ( [ genfac ] ) indicate that , if @xmath51 , then @xmath52 is an orthonormal set of eigenfunctions of @xmath53 with eigenvalues @xmath54 .
this is a basis if there is not a square - integrable function @xmath55 which is orthogonal to the full set .
thus , let us look for @xmath55 such that @xmath56 by solving this first - order differential equation we get : @xmath57 since @xmath58 ( compare eq.([genfac ] ) ) , then the spectrum of @xmath53 depends of the normalizability of @xmath55 .
we can identify three different situations .
\(i ) let us choose @xmath59 and @xmath60 , which is nodeless in the domain of @xmath61 , @xmath62 .
thus , @xmath63 is completely determined , and the corresponding hamiltonian @xmath53 has eigenfunctions and eigenvalues given by @xmath64 notice that @xmath65 sp@xmath66 since @xmath67 is not square - integrable .
\(ii ) let now a nodeless seed solution @xmath42 for @xmath68 be chosen so that @xmath69 is non - singular .
note that @xmath42 diverges at both ends of the @xmath70-domain @xmath71 tends to zero at those ends .
hence , the full set of eigenfunctions and eigenvalues is @xmath72 \(iii ) let us employ a solution @xmath42 with a node at one of the ends of the @xmath70-domain for @xmath68 .
thus , the transformation induced by @xmath43 is non - singular , @xmath67 diverges at the end where @xmath42 tends to zero , and then the eigenfunctions and eigenvalues of @xmath53 become @xmath73 this implies that @xmath53 and @xmath46 are isospectral hamiltonians .
let us build up now the so - called supersymmetric quantum mechanics , by realizing the witten supersymmetry algebra with two generators @xcite @xmath74=0 , \quad \
{ q_i , q_j \ } = \delta_{ij } h_{\rm ss } , \quad i , j=1,2 , \label{susyalg}\end{aligned}\ ] ] in the way @xmath75 notice that there is a linear relationship between @xmath76 and @xmath77 : @xmath78 there are two - fold degenerated levels of @xmath76 [ @xmath79 with @xmath80 for the case ( i ) and @xmath81 with @xmath82 for cases ( ii ) and ( iii ) ] which have associated the two orthonormal eigenvectors @xmath83 the ground state energy of @xmath76 in cases ( i ) and ( ii ) is non - degenerated ( and equal to @xmath84 ) , with the corresponding eigenstate being given respectively by @xmath85 in this case it is said that the supersymmetry is unbroken . on the other hand , in case ( iii ) the ground state energy ( @xmath86 ) is doubly degenerated , and then the supersymmetry is spontaneously broken .
let us apply the previous technique to the trigonometric pschl - teller potentials @xcite : @xmath87 first of all , we require the general solution of the schrdinger equation ( [ seg ] ) with the @xmath88 given in ( [ ptp ] ) , which is given by @xmath89 \nonumber \\ & & \hskip-1.3 cm + b \sin^{1 - 2\lambda}(x ) \ , { } _ 2f_1\left[\frac{1 + \nu - \lambda}{2 } + \sqrt{\frac{\epsilon}{2}},\frac{1 + \nu - \lambda}{2 } - \sqrt{\frac{\epsilon}{2 } } ; \frac32 - \lambda;\sin^2(x)\right]\bigg\ } , \label{vptgs}\end{aligned}\ ] ] where @xmath90 .
the physical eigenfunctions @xmath91 satisfy @xmath92 . if we ask that @xmath93 , it turns out that @xmath94 . moreover , in order to avoid that the divergent behavior at @xmath95 of the hypergeometric function of the remaining term dominates over the null one induced by @xmath96 for @xmath97 it has to happen that @xmath98 the corresponding normalized eigenfunctions are given by @xmath99^{\frac12 } \sin^\lambda(x ) \cos^\nu(x ) \ { } _ 2f_1[-n , n + \mu ; \lambda + \frac12;\sin^2(x)].\end{aligned}\ ] ] it will be important later to know the number of zeros which @xmath42 has , according to the value taken by @xmath26 .
this information can be obtained by comparing the asymptotic behavior of @xmath42 at @xmath100 and @xmath101 ; by picking up @xmath102 and @xmath103 in eq.([vptgs ] ) with @xmath104 it turns out that : * if @xmath68 , @xmath42 has either @xmath84 or @xmath105 nodes for @xmath106 or @xmath107 respectively * in general , if @xmath108 , @xmath42 will have either @xmath109 or @xmath110 nodes for @xmath106 or @xmath107 respectively , @xmath111 concerning the spectral modifications which can be induced through the standard first - order supersymmetric quantum mechanics , it is obtained the following .
\(a ) deleting the ground state of @xmath46 . let us choose @xmath59 and @xmath112 the first - order susy partner of @xmath61 becomes ( see eq.([gv1 ] ) ) @xmath113 since @xmath114 diverges at @xmath115 @xmath116 .
note that @xmath39 can be obtained from @xmath61 by changing @xmath117 ; this property is known nowadays as _ shape invariance _
@xcite , although it was known since infeld - hull work @xcite .
\(b ) creating a new ground state .
let us take now @xmath68 and a nodeless @xmath42 ( eq.([vptgs ] ) with @xmath118 , @xmath103 , @xmath119 ) .
as @xmath120 when @xmath121 @xmath122 @xmath123 , i.e. , @xmath124 is an eigenfunction of @xmath53 with eigenvalue @xmath26 , and therefore sp(@xmath53)=@xmath125 . in order to deal with the singularities at @xmath126 induced by @xmath42 on @xmath39
, it is expressed @xmath127 where @xmath128 is a nodeless bounded function in @xmath129 $ ] .
hence @xmath130 '' , \quad \lambda , \nu > 2 . \label{v1ptc}\end{aligned}\ ] ] \(c ) isospectral potentials .
they arise from the previous case ( for @xmath131 ) in the limit when @xmath42 acquires a node at one of the ends of the domain so that @xmath124 leaves to be an eigenstate of @xmath53 .
we can take , for example , the solution given in eq.([vptgs ] ) with @xmath132 , @xmath133 @xmath122 @xmath134 .
we isolate the singularities induced on @xmath39 by expressing @xmath135 , where @xmath128 is a nodeless bounded function in @xmath129 $ ] .
hence @xmath136 '' , \quad \lambda>1 , \nu > 2 . \label{v1pti}\end{aligned}\ ] ] notice that now @xmath53 and @xmath46 are isospectral .
an illustration of the several first - order susy partner potentials @xmath137 of @xmath88 ( see eqs.([v1ptd]-[v1pti ] ) ) is given in figure 1 .
the second - order supersymmetric quantum mechanics can be implemented through the iteration of two first - order transformations . however , by means of this approach it is not revealed in full the rich set of spectral manipulations which is offered by the second - order technique .
this is avoided if we address the method directly , by assuming that the intertwining operator @xmath138 is of second - order in the derivatives , namely @xcite @xmath139 a calculation similar as for the first - order case leads to the following set of equations
@xcite : @xmath140 by decoupling this system , assuming that @xmath61 is a given solvable potential , we arrive at : @xmath141 notice that @xmath142 and @xmath143 become completely determined once it is found @xmath144 ; this task is performed by assuming the anstz @xmath145 where the functions @xmath146 and @xmath147 are to be determined . therefore @xmath148 , \quad \epsilon=(d+\xi)/2,\end{aligned}\ ] ] which is once again a riccati equation for @xmath146 . by assuming that @xmath149 it turns out that @xmath42 satisfies the stationary schrdinger equation @xmath150 since @xmath151 , we get in general two different values for @xmath26 , @xmath152 , @xmath153 .
we thus arrive at a natural classification of the second - order susy transformations , depending on the sign taken by the parameter @xmath154 @xcite .
for @xmath155 we have that @xmath156 , @xmath157 ; denoting by @xmath158 , @xmath159 the corresponding solutions of the riccati equation it turns out that ( compare eq.([anzats ] ) ) @xmath160 by making the difference of both equations it is obtained : @xmath161 notice that now the new potential @xmath142 has no extra singularities with respect to @xmath61 if @xmath162 is nodeless ( compare eq.([v2sususy ] ) ) .
the spectrum of @xmath163 depends on the normalizability of the two eigenfunctions @xmath164 of @xmath163 associated to @xmath165 which belong as well to the kernel of @xmath166 , @xmath167 their explicit expressions in terms of @xmath168 and @xmath169 become @xcite @xmath170 we have observed several possible spectral modifications induced on the pschl - teller potentials .
\(a ) deleting 2 neighbor levels . by taking as seeds two physical eigenfunctions of @xmath46 associated to a pair of neighbor eigenvalues , namely @xmath171 @xmath172
, @xmath173 it is straightforward to show that @xmath174 where @xmath175,{}_2f_1[-i+1,i-1 + \mu ; \lambda + \frac12;\sin^2(x)]\}}{\sin(x ) \cos(x)}\end{aligned}\ ] ] is a nodeless bounded funcion in @xmath129 $ ] .
hence , @xmath176 notice that the two mathematical eigenfunctions @xmath177 of @xmath163 associated to @xmath178 do not vanish at @xmath179 , @xmath180 thus , @xmath181 sp@xmath182 , i.e. , we have deleted the levels @xmath183 of @xmath46 in order to generate @xmath184 .
\(b ) creating two new levels . let us choose now @xmath185 , @xmath186 , with the corresponding seeds @xmath42 as given in eq.([vptgs ] ) with @xmath187 , @xmath188 , @xmath189 .
this choice ensures that @xmath190 and @xmath191 have @xmath110 and @xmath109 alternating nodes , which produces a nodeless wronskian @xcite .
the divergence of the @xmath42solutions for @xmath192 is isolated by expressing @xmath193 @xmath194 being bounded for @xmath195 $ ] , @xmath196 . since the second term of the taylor series expansion of @xmath194 is proportional to @xmath197 @xmath122 @xmath198 tends to zero as @xmath199 for @xmath100 and as @xmath200 for @xmath101 .
thus , @xmath201 where @xmath202 $ ] is nodeless bounded in @xmath129 $ ] .
hence @xmath203 since @xmath204 @xmath122 sp@xmath205 , i.e. , we have created two new levels @xmath206 between the originally neighbor ones @xmath207 for generating @xmath184 .
it is worth to mention some other spectral modifications which can be straightforwardly implemented @xcite : * isospectral transformations * creating a new level * moving an arbitrary physical level * deleting an original physical level an illustration of the potentials @xmath184 induced by some of the previously discussed spectral modifications is given in figure 2 .
for @xmath209 we have that @xmath210 .
it this case , once the solution @xmath146 to the riccati equation for @xmath26 is found , one has to solve the bernoulli equation resultant for @xmath144 , @xmath211dx = w_0 + \int_{x_0}^x [ u^{(0)}(y)]^2\,dy , \\ & & \hskip3 cm u^{(0)}\propto\exp[\int\beta(x)\,dx ] , \end{aligned}\ ] ] with @xmath212 . notice that now @xmath213 has to be nodeless , which is obtained for seed solutions @xmath42 such that @xmath214 ^ 2\,dy < \infty \quad { \rm or } \\ & & \lim\limits_{x\rightarrow x_l } u^{(0 ) } = 0 , \quad \int_{x_l}^{x_0 } [ u^{(0)}(y)]^2\,dy < \infty,\end{aligned}\ ] ] and adjusting as well @xmath215 , where @xmath216 $ ] is the @xmath70-domain of the involved problem , @xmath217 $ ] .
there is one eigenfunction , @xmath218 , of @xmath163 associated to @xmath26 belonging as well to the kernel of @xmath166 , which implies that @xmath219 depends of the normalizability of @xmath220 . for the pschl - teller potentials , several possibilities for modifying the initial spectrum have been observed .
\(a ) creating a new level .
let us choose @xmath221 and @xmath42 as given in eq.([vptgs ] ) with @xmath222 : @xmath223 where @xmath128 is bounded for @xmath195 $ ] , @xmath224 . by calculating the involved integral with @xmath225 it is obtained : @xmath226 notice that @xmath227 is nodeless in @xmath129 $ ] for @xmath228 and it has one node for @xmath229 . by choosing now the @xmath227 of eq.([wconfluente ] ) with @xmath228 , we have to isolate anyways a divergence of kind @xmath230 arising for @xmath101 through the factorization @xmath231 @xmath232 being a nodeless bounded function for @xmath195 $ ] . from eq.([v2sususy ] ) we finally get @xmath233 since @xmath234 , it turns out that sp@xmath235 .
it is important to remark that the procedure can be implemented for solutions associated to @xmath236 , i.e. , by means of the confluent second - order transformation one is able to modify the levels above the ground state energy of @xmath46 some additional possibilities of spectral manipulation are now just pointed out @xcite . * isospectral transformations *
deletion of an arbitrary level the potentials @xmath184 induced by some of the mentioned spectral modifications are illustrated in figure 3 . for @xmath237
it happens that @xmath238 , @xmath239 .
as we want that @xmath142 to be real , it has to be chosen @xmath240 .
thus , @xmath241 } = \frac{w'}{w } , \quad w = \frac{w(u^{(0)}_1,\bar u^{(0)}_1)}{2(\epsilon_1 - \bar \epsilon_1)},\end{aligned}\ ] ] where @xmath168 is a complex solution of the schrdinger equation associated to @xmath242 . as in the previous cases
, @xmath227 has to be nodeless in order to avoid the arising of extra singularities in @xmath142 .
for this to happens , it is sufficient that @xmath243 in both cases we get that @xmath244 .
we illustrate now the technique through the trigonometric pschl - teller potentials , for which we take @xmath245 , @xmath42 as in eq.([vptgs ] ) with @xmath222 @xmath122 @xmath134 , @xmath246 when @xmath101 .
it is appropriate to express now @xmath247 hence , it is obtained : @xmath248 the potentials @xmath61 and @xmath142 turn out to be isospectral .
an illustration of @xmath88 and the @xmath184 of eq.([v2ptcom ] ) is given in figure 4 . a general conclusion , which can be inferred of the previously discussed first and second - order susy qm but it is valid for an arbitrary order , is that the spectra of the initial and new hamiltonians differ little .
thus , one would expect that some properties which are somehow related to the corresponding spectral problems would be essentially the same .
we will see next that this is true , in particular , for the algebraic structure ruling the two involved hamiltonians .
given an initial hamiltonian @xmath46 , which is characterized by a certain algebraic structure , it is natural to pose the following questions : * what is the algebra ruling the potentials generated through susy qm ?
* what other properties of @xmath46 are inherited by its susy partner hamiltonians ? these questions will be analyzed here through a simple system which algebraic structure generalizes the standard heisenberg - weyl ( oscillator ) algebra .
thus , let us consider a one - dimensional schrdinger hamiltonian @xmath249 whose eigenvectors @xmath250 and eigenvalues @xmath48 satisfy ( for calculational convenience we switch now to the dirac bra and ket notation ) @xmath251 it is supposed that there is an analytic dependence between the eigenvalues @xmath48 and the index labeling them , namely , @xmath252 let us define now the _ intrinsic algebra _ of @xmath46 , by introducing a pair of annihilation and creation operators @xmath253 whose action onto the eigenvector of @xmath46 is given by @xmath254 } \
\sqrt{e(n ) - e_0 } , \quad \alpha \in { \mathbb r } .\end{aligned}\ ] ] with these definitions it is straightforward to show that : @xmath255 it is introduced the number operator @xmath256 through its action onto the eigenvectors of @xmath46 : @xmath257 = \pm a_0^\pm .\end{aligned}\ ] ] now , the intrinsic algebra of the systems is characterized by @xmath258 = e(n_0 + 1 ) - e(n_0 ) \equiv f(n_0 ) .\end{aligned}\ ] ] it is important to realize that for general @xmath259 the intrinsic algebra becomes nonlinear ( since @xmath260 is not necessarily a linear function of @xmath256 ) . however
, this intrinsic nonlinear algebra can be linearized by deforming the corresponding annihilation and creation operators in the way @xmath261 = \pm a_{0_{l}}^\pm , \quad a_{0_{l}}^+ a_{0_{l}}^- = n_0 , \quad a_{0_{l}}^-
a_{0_{l}}^+ = n_0 + 1 , \quad [ a_{0_{l}}^- , a_{0_{l}}^+ ] = 1 .\end{aligned}\ ] ] in particular , for the harmonic oscillator hamiltonian the intrinsic algebraic structure reduces to the standard heisenberg - weyl algebra : @xmath262 = \pm a_0^\pm , \quad [ a_0 ^ -,a_0^+ ] = 1 .\end{aligned}\ ] ] on the other hand , for the trigonometric pschl - teller potentials we recover the su(1,1 ) algebra : @xmath263 = \pm a_0^\pm , \quad [ a_0 ^ -,a_0^+ ] = 4 n_0 + 2\mu + 2 .\end{aligned}\ ] ] let us characterize now the algebraic structure of the susy partner hamiltonians of @xmath46 @xcite . in order to illustrate the procedure , we restrict ourselves just to a second - order susy transformation which creates two new levels at the positions @xmath264 for @xmath163 , namely , @xmath265 notice that the eigenstates of @xmath163 , @xmath266 , associated to @xmath267 are obtained from @xmath268 and vice versa through the action of the intertwining operators @xmath138 , @xmath166 in the way :
@xmath269 the eigenstates of @xmath163 constitute a complete orthonormal set , @xmath270 it is important to notice that the two new levels @xmath271 can be placed at positions essentially arbitrary ( either both below the initial ground state @xmath272 or in between two neighbor physical levels @xmath273 ) . this fact somehow breaks the symmetry defined by @xmath259 .
thus , it should be clear that we need to isolate , in a sense , the two new levels @xmath271 . with this aim
, it is defined now the number operator @xmath274 through the following action onto the eigenstates of @xmath163 : @xmath275 departing now from the operators @xmath253 of the intrinsic algebra of @xmath46 , let us construct those of the _ natural algebra _ of @xmath163 @xcite , which represent a generalization of the operators introduced previously for the susy partners of the harmonic oscillator @xcite , @xmath276 an schematic representation of these operators is given in figure 5 .
their action onto the eigenvectors of @xmath163 is given by @xmath277[e(n-1)-\epsilon_i ] \right\}^\frac12 r_{i}(n ) , \quad i=1,2 , \
n=0,1,\dots.\end{aligned}\ ] ] using these equations it is found that @xmath278 \
! \left[\bar r_{n}(\ ! n_2 \!\
1 ) r_{n}(\ ! n_2 \!\ ! + \!\ !
1 ) \ ! - \ !
\bar r_{n}(n_2 ) r_{n}(n_2)\right ] \!\ !
\sum\limits_{m=0}^{\infty } \!\ !
\vert\psi_{m}^{(2 ) } \rangle \ !
\langle\psi_{m}^{(2 ) } \vert .\end{aligned}\ ] ] let us build up now the annihilation and creation operators @xmath279 which generate the intrinsic algebra of @xmath163 , @xmath280 = f(n_2 ) \sum\limits_{m=0}^\infty \vert\psi_{m}^{(2 ) } \rangle\langle\psi_{m}^{(2)}\vert .\end{aligned}\ ] ] their action onto the eigenstates of @xmath163 reads @xmath281 notice that this algebra of @xmath163 is the same as the intrinsic algebra of @xmath46 when we work on the restriction to the subspace spanned by the eigenvectors @xmath282 associated to the isospectral part of the spectrum .
let us construct , finally , the annihilation and creation operators @xmath283 of the linear algebra of @xmath163 , @xmath284 = \pm a_{2_{l}}^\pm , \qquad [ a_{2_{l}}^- , a_{2_{l}}^+ ] = \sum\limits_{m=0}^\infty \vert \psi_m ^{(2)}\rangle\langle \psi_m^{(2 ) } \vert .\end{aligned}\ ] ] their action onto the eigenstates of @xmath163 is given by @xmath285 once again , this algebra of @xmath163 coincides with the corresponding one of @xmath46 on the subspace spanned by @xmath286 .
once identified the algebraic structures associated to @xmath46 and @xmath163 , it would be important to generate the corresponding coherent states .
it is well know that there are several definitions of coherent states @xcite ; here we will derive them as eigenstates with complex eigenvalues of the annihilation operators of the systems . since there are available several annihilation operators for @xmath46 and @xmath163 , we will call them with the same name as the corresponding associated algebra .
since for @xmath46 we identified two different algebraic structures , intrinsic and linear one , we will look for the two corresponding families of coherent states .
the intrinsic coherent states @xmath287 of @xmath46 satisfy : @xmath288 expanding @xmath287 in the basis of eigenstates of @xmath46 , @xmath289 and substituting in eq.([dcsi0 ] ) , it is obtained a recurrence relationship for @xmath290 , @xmath291 by choosing @xmath292 so that @xmath287 is normalized we finally get @xmath293 an important property that should be satisfied is the so - called completeness relationship , namely , @xmath294 where the positive defined measure @xmath295 can be expressed as @xmath296 working in polar coordinates and changing variables @xmath297 , it turns out that @xmath298 must satisfy @xmath299 this is a moment problem depending on @xmath300 , i.e. , on the spectral function @xmath259 @xcite .
if one is able to solve eq.([mpi0 ] ) for @xmath298 , the completeness relation becomes valid .
once eq.([incom ] ) is satisfied , an arbitrary quantum state can be expressed in terms of coherent states . in particular
, an intrinsic coherent state @xmath301 admits the decomposition , @xmath302 where the reproducing kernel @xmath303 is given by @xmath304 as @xmath305 , it turns out that the eigenvalue @xmath306 is non - degenerate .
if the system is initially in an intrinsic coherent state @xmath307 , then it will be at any time in an intrinsic coherent state , @xmath308 where @xmath309 is the evolution operator of the system .
now , let us look for the linear coherent states such that @xmath311 a procedure similar to the previous one leads straightforwardly to @xmath312 which , up to the phases , has the form of the standard coherent states .
now the completeness relationship is automatically satisfied , @xmath313 therefore , an arbitrary linear coherent state @xmath314 is expressed in terms of linear coherent states , @xmath315 where the reproducing kernel is given by @xmath316 notice that , once again , the only eigenstate of @xmath46 which is also a linear coherent state is the ground state , @xmath317 . moreover , since @xmath318=1 $ ] then the linear coherent states also result from the action of the displacement operator @xmath319 onto the ground state @xmath320 : @xmath321 finally , an initial linear coherent state evolves in time as a linear coherent state , namely , @xmath322 let us remember that we have analyzed three different algebraic structures for @xmath163 , namely , natural , intrinsic and linear ones ; thus three families of coherent states will be next constructed .
the natural coherent states @xmath323 of @xmath163 are defined by @xmath324 a procedure similar to the previously used leads to @xmath325^{-\frac12 } \sum\limits_{m = 0}^{\infty } \ !
e^{-i\alpha ( e_{m } - e_{0})}\frac{z^{m}}{\sqrt{\widetilde\rho_m}}\vert\psi_{m } ^{(2)}\rangle , \end{aligned}\ ] ] where @xmath326 notice that now the completeness relationship has to be modified to include the projector onto the subspace spanned by the eigenvectors of @xmath163 associated to the two isolated levels @xmath327 , namely , @xmath328 where @xmath329 it is clear that the moment problem characterized by eq.([mpn ] ) is more complicated than the one associated to eq.([mpi0 ] ) ( compare the moments given by eqs.([momentsi0 ] ) and ( [ momentsn2 ] ) ) . since @xmath330 and @xmath331 , then the eigenvalue @xmath306 of @xmath332 is 3-fold degenerate .
finally , as for the coherent states of @xmath46 , it turns out that the natural coherent states @xmath323 evolve coherently with @xmath333 , @xmath334 where @xmath335 is the evolution operator of the system characterized by @xmath163 .
the intrinsic coherent states @xmath336 of @xmath163 are eigenstates with complex eigenvalue @xmath337 of the intrinsic annihilation operator @xmath338 of @xmath163 .
a treatment similar to the previous one leads to @xmath339 the modified completeness relationship reads now @xmath340 where @xmath295 is the same as for @xmath46 . since @xmath341 and @xmath342 , the eigenvalue @xmath306 of @xmath343 turns out to be @xmath344-fold degenerate . as for the natural coherent states of @xmath163 , the intrinsic coherent states
@xmath345 evolve in time in a coherent way .
let us consider , finally , the linear coherent states @xmath346 of @xmath163 as eigenstates of @xmath347 with eigenvalue @xmath337 .
a standard procedure produces now @xmath348 the completeness relationship is expressed as @xmath349 once again , the eigenvalue @xmath306 of @xmath347 is @xmath344-fold degenerate .
moreover , the linear coherent states @xmath346 appear from the action of the displacement operator @xmath350 onto the extremal state @xmath351 : @xmath352 as in the previous cases , the linear coherent states of @xmath163 evolve in time coherently .
let us end up these two sections , about the algebraic structure of the susy partner hamiltonians and the corresponding coherent states , with the following conclusions .
* the intrinsic and linear algebraic structures of @xmath46 are inherited by @xmath163 on the subspace associated to the levels of the isospectral part of the spectrum * this reflects as well in the corresponding coherent states of @xmath163 , linear and intrinsic ones , whose expressions on the same subspace are equal to the corresponding ones of @xmath46 * we generalized successfully the construction of the natural algebra of @xmath163 when the initial potential @xmath88 is more general than the harmonic oscillator
the exactly solvable models are important since the relevant physical information is encoded in a few expressions .
moreover , they are ideal either to test the accuracy ( and convergence ) of numerical techniques , or to implement some analytic approximate methods , as perturbation theory .
unfortunately , the number of exactly solvable periodic potentials is small , so it is important to enlarge this class . as we saw in the previous sections , a simple method to do the job is supersymmetric quantum mechanics . before applying the technique to periodic potentials
, however , let us review first the way in which it is classified the energy axis in allowed and forbidden energy bands . for one - dimensional periodic potentials such that @xmath353 , the schrdinger equation @xmath354 can be conveniently expressed in matrix form : @xmath355 & 0 \end{array } \right).\end{aligned}\ ] ] therefore : @xmath356 where the @xmath357 _ transfer matrix _
@xmath358 is symplectic , which implies that its determinant is equal to @xmath105 .
the transfer matrix can be expressed in terms of two real solutions @xmath359 such that @xmath360 , @xmath361 , @xmath362 , @xmath363 , as follows @xmath364 the general behavior of @xmath365 , and the kind of spectrum of the hamiltonian @xmath46 , depends on the eigenvalues @xmath366 of the _ floquet matrix _ @xmath367 , which in turn are determined by the so - called _ discriminant _ @xmath368 $ ] : @xmath369 by picking up @xmath370 as one of the eigenvectors of the floquet matrix @xmath367 with eigenvalue @xmath146 , it turns out that @xmath371 moreover , for these vectors in general we have that @xmath372 the corresponding @xmath36 are known as the bloch functions , and they exist for any @xmath373 . depending on the value of the discriminant , the bloch functions acquire three different asymptotic behaviors : * if @xmath374 for a given @xmath375 , it turns out that @xmath376 where @xmath377 is called the _ quasimomentum_. equation ( [ blochasimp ] ) implies that the bloch solutions @xmath378 in this case are bounded for @xmath379 , and hence they can be equipped with a physical meaning .
therefore , @xmath375 belong to an allowed energy band . *
the equation @xmath380 defines the band edges , denoted @xmath381 for these energy values it turns out that @xmath382 , the two bloch functions tend just to one which becomes periodic or antiperiodic .
thus , these energy values belong as well to the spectrum of @xmath46 .
moreover , the bloch eigenfunctions @xmath383 @xmath384 associated to @xmath385 are real , they have the same number @xmath386 of nodes , and both are either periodic or antiperiodic * if @xmath387 for a given @xmath375 , it turns out that @xmath388 then , the bloch solutions @xmath389 are unbounded either for @xmath390 or for @xmath391 so that they can not be equipped with a physical interpretation .
therefore , @xmath375 belongs to a forbidden energy gap as an illustration , we have plotted in figure 6 the discriminant @xmath392 as a function of @xmath375 for @xmath393 .
a very special class of periodic systems are given by the lam potentials @xmath394 where @xmath395 is a jacobi elliptic function .
these are doubly periodic special functions defined by @xcite @xmath396 with @xmath397 .
the real and imaginary periods are respectively @xmath398 , @xmath399 , @xmath400 , where @xmath401 the special nature of the lam potentials is due to the following properties : * they have @xmath402 band edges , which define @xmath403 allowed energy bands ( @xmath404 of them finite and @xmath105 infinite ) and @xmath403 energy gaps ( @xmath404 finite and @xmath105 infinite ) * it is possible to determine the general solution of the schrdinger equation for any real value of the energy parameter * in particular , there exist explicit analytic expressions for the band edge eigenfunctions let us illustrate this behavior through the simplest example . in this case
the bloch functions for a real arbitrary energy value @xmath26 are given by @xmath406 where @xmath407'=\zeta(z)$ ] , @xmath408 and @xmath409 note that @xmath410 is the weierstrass @xmath411-function , which is related to the jacobi elliptic function sn through @xcite @xmath412 there are now three band edges , and the explicit expression for the corresponding eigenvalues and eigenfunctions are given by @xmath413 the energy axis is classified as follows : the energy interval @xmath414 represents the infinite energy gap ( forbidden energies ) , then it comes the finite energy band @xmath415 $ ] ( allowed energies ) , then it arises the finite energy gap @xmath416 , and finally it emerges the infinite energy band @xmath417 .
let us choose in the first place as seed one real nodeless bloch function @xmath42 with @xmath50 .
it is straightforward to check that @xmath418 is periodic , which implies that @xmath137 is also periodic ( see eq.([gv1 ] ) ) .
the susy transformation maps bounded ( unbounded ) eigenfunctions of @xmath46 into bounded ( unbounded ) ones of @xmath53 . since for @xmath419
the chosen bloch function @xmath42 diverges either for @xmath390 or @xmath420 ( @xmath421 ) , then @xmath422 will diverge at the opposite limits and therefore @xmath423 .
on the other hand , for @xmath424 , the corresponding bloch solution @xmath42 is nodeless periodic and therefore @xmath422 is also a nodeless periodic eigenfunction of @xmath53 with eigenvalue @xmath272 . in conclusion , for first - order susy transformations involving a nodeless bloch function with @xmath50 it turns out that sp(@xmath53 ) = sp(@xmath46 ) , i.e. , the susy mapping is isospectral . in particular , for the lam potential with
@xmath425 the first - order susy transformation which employs the lowest band edge eigenfunction @xmath426 leads to @xcite @xmath427 and it is said that @xmath137 is self - isospectral to @xmath88 .
this property was discovered , for the first time , by braden and mcfarlane @xcite , although the name is due to dunne and feinberg @xcite .
moreover , by using as seed the bloch eigenfunction @xmath428 with @xmath68 is is straightforward to show that @xcite @xmath429 where @xmath430 is related to @xmath26 through eq.([a - epsilon ] ) .
let us choose now a real nodeless linear combination @xmath42 of the two bloch functions for @xmath68 .
although now @xmath418 is not strictly periodic , it is asymptotically periodic when @xmath431 .
this property is transferred as well to the new potential @xmath137 which is asymptotically periodic for @xmath431 .
as in the previous case , the bounded ( unbounded ) eigenfunctions of @xmath46 are mapped into bounded ( unbounded ) eigenfunctions of @xmath53 . since @xmath432 , it turns out that @xmath422 is square - integrable in @xmath433 .
we conclude that , when appropriate linear combinations of two bloch functions of @xmath46 associated to @xmath68 are used the implement the first - order susy transformation , we create a bound state for @xmath53 , i.e. , sp(@xmath53)@xmath434sp(@xmath46 ) .
we apply now this kind of transformation to the lam potential with @xmath425 , using as seed a real nodeless linear combination of the two bloch solutions @xmath428 given in eq.([blochlamen=1 ] ) with @xmath68 @xcite . in this way it is obtained an asympotically periodic potential , as illustrated in figure 7 , where the corresponding bound state of @xmath53 is as well shown .
let us apply now a second - order susy transformation using two real bloch solutions @xmath435 associated to @xmath436 $ ] or @xmath437 , j=1,2,\dots$ ] so that @xmath162 is nodeless @xcite-@xcite . for this choice it turns out that @xmath438 as well as @xmath184 are periodic ( see eqs.([v2sususy],[rcw2 ] ) ) .
the susy transformation maps bounded ( unbounded ) eigenfunctions of @xmath46 into bounded ( unbounded ) ones of @xmath163 .
thus , @xmath46 and @xmath163 have the same band structure , namely , sp(@xmath163)=sp(@xmath46 ) . for the lam potential with @xmath425
, if we employ the two band edge eigenfunctions @xmath439 bounding the finite gap , we get once again @xcite @xmath440 moreover , by using as seeds two bloch functions @xmath441 such that @xmath442 it is straightforward to show that @xcite @xmath443 let us choose now as seeds two real linear combinations of bloch function for @xmath436 $ ] or @xmath437 , j=1,2,\dots$ ] so that @xmath162 is nodeless @xcite . once again , + @xmath438 and consequently @xmath184 are asymptotically periodic for @xmath444 , the second - order susy transformation maps bounded ( unbounded ) eigenfunctions of @xmath46 into bounded ( unbounded ) ones of @xmath163 . since now the two eigenfunctions of @xmath163 , @xmath445 , @xmath446 associated to @xmath264 , can be made square - integrable for @xmath447 , in turns out that sp(@xmath163)@xmath448sp(@xmath46 ) .
let us illustrate the procedure for the lam potential with @xmath425 , using as seeds two real linear combinations @xmath435 of @xmath449 and @xmath450 for @xmath451 .
as in the general case , it is obtained a potential @xmath184 which is asymptotically periodic for @xmath431 .
the corresponding hamiltonian has two bound states at the positions @xmath206 .
an illustration of these results is shown in figure 8 .
our conclusions concerning susy transformations applied to periodic potentials are the following .
* it was employed successfully the susy qm for generating new solvable potentials departing from a periodic initial one * for seed solutions chosen as bloch functions associated to the band edges or to factorization energies inside the spectral gaps , the new potential is periodic , isospectral to the initial one * for seed solutions which are real linear combinations of bloch functions , the new potential is asymptotically periodic , and the spectrum of the corresponding hamiltonian will include a finite number of bound states embedded into the gaps * the technique was clearly illustrated by means of the lam potential with @xmath425
* along this paper it has been shown that supersymmetric quantum mechanics is a simple but powerful tool for generating potentials with known spectra departing from a given initial solvable one .
indeed , susy qm can be used to implement the spectral manipulation in quantum mechanics * one of our aims when writing this paper is to present the development of the factorization method , in particular the further advances arising after mielnik s paper about generalized factorizations @xcite * supersymmetric quantum mechanics has interrelations with several interesting areas of mathematical physics such as solutions of non - linear differential equations @xcite , deformation of lie algebras and coherent states @xcite , among others . in my opinion , a further exploration of these interrelations is needed , which represents the future of susy qm .
we will pursue these issues in the next years with the hope of getting as many interesting results as we were obtaining previously .
i. arefeva , d.j .
fernndez , v. hussin , j. negro , l.m .
nieto , and b.f .
samsonov , eds . of _ special issue dedicated to the subject of the international conference on progress in supersymmetric quantum mechanics _ , j. phys . a : math
37 * , number 43 ( 2004 ) d.j .
fernndez , and a. ramos , in _ topics in mathematical physics , general relativity and cosmology in honor of jerzy plebaski _ , eds . h. garca - compen et al , world scientific , singapore , 167 - 175 ( 2006 ) | supersymmetric quantum mechanics ( susy qm ) is a powerful tool for generating new potentials with known spectra departing from an initial solvable one . in these lecture notes
we will present some general formulae concerning susy qm of first and second order for one - dimensional arbitrary systems , and we will illustrate the method through the trigonometric pschl - teller potentials . some intrinsically related subjects , as the algebraic structure inherited by the new hamiltonians and the corresponding coherent states will be analyzed
. the technique will be as well implemented for periodic potentials , for which the corresponding spectrum is composed of allowed bands separated by energy gaps . |
dilatometry is a powerful technique for exploring the thermodynamic properties of solids .
measurements of the volume expansion coefficient , @xmath6 = ( @xmath7ln@xmath8/@xmath9)@xmath10 ( @xmath8 is the sample volume ) , or , when single - crystalline material is available , the uniaxial expansion coefficients , @xmath11 = ( @xmath7ln@xmath12/@xmath9)@xmath10 ( @xmath12 denotes the sample length along the principal crystal axes ) provide important information on the electronic , magnetic and lattice properties of the material .
the expansivity is thermodynamically related to the specific heat via @xmath13 with @xmath14 = @xmath15(@xmath7ln@xmath8/@xmath16)@xmath17 the isothermal compressibility , @xmath18 the molar volume , and @xmath19 = @xmath20 ( @xmath21/@xmath7ln@xmath8)@xmath17 is the grneisen function measuring the volume dependence of the system s entropy @xmath22 . in cases where a single characteristic energy @xmath23 governs the properties of a material within a certain temperature
range the corresponding grneisen parameter is given by @xmath19 = @xmath15dln@xmath23/dln@xmath8.@xcite hence by employing the grneisen formalism the volume dependence of characteristic energies can be derived .
likewise , for a second - order phase transition at a critical temperature @xmath24 , the application of the well - known ehrenfest relation ( dln@xmath24/d@xmath25)@xmath26 = @xmath27 , enables the determination of the pressure dependence of @xmath24 in the limit of vanishingly small pressure from the discontinuities in @xmath6 and @xmath28 , see _
e.g. _ ref . for a review .
[ grueneisen ] implies that the expansivity provides a highly sensitive probe for systems where @xmath19 is large or even tends to diverge .
in fact large and strongly temperature - dependent @xmath19 values have been predicted @xcite and experimentally observed upon approaching a quantum phase transition@xcite or a finite - temperature critical end point.@xcite hence in these cases thermal expansion measurements offer unprecedented possibilities for characterizing the phase transition . as an example of a finite - temperature critical end point where large anomalies in the expansivity are expected,@xcite we mention the first - order mott transition in organic charge - transfer salts,@xcite accessible by tuning the systems via pressure and/or chemical substitution . here
dilatometric measurements under well - controlled hydrostatic - pressure conditions would be of particular importance for unraveling the mystery surrounding the mott criticality in these systems.@xcite another area of potential applications for dilatometry under he - gas pressure concerns pressure studies on materials , the properties of which show an extraordinarily strong sensitivity to nonhydrostaticity .
prominent examples include low - dimensional organic conductors @xcite with strongly anisotropic pressure coefficients @xcite and some members of the recently discovered feas - based superconductors , notably cafe@xmath29as@xmath29.@xcite for the latter material , nonhydrostatic pressure conditions were found to induce a low - temperature superconducting phase @xcite which is absent under hydrostatic pressure.@xcite likewise , for systems with strongly pressure - dependent magnetic exchange coupling constants , tantamount to large magnetic grneisen parameters , thermal expansion measurements under pressure would provide a most sensitive tool for identifying and exploring the dominant interactions .
+ among the various methods for measuring temperature - induced length changes of a material , capacitive dilatometry@xcite excels by its extraordinarily high sensitivity reaching values of @xmath30 = 10@xmath31 , exceeding the resolution attainable by using strain gauges ( @xmath32__l / l _ _ @xmath33 10@xmath34),@xcite x - ray diffraction ( @xmath32__a / a _ _ @xmath33 10@xmath35),@xcite and optical interferometers ( @xmath32__a / a _ _
10@xmath36 - 10@xmath37)@xcite by several orders of magnitude .
in this work we describe the design and realization of an apparatus for high - resolution measurements of the coefficient of thermal expansion of solids under variable helium - gas pressure for a wide range of temperatures 1.4k @xmath0 300k and hydrostatic pressures @xmath1 250mpa .
since for capacitive dilatometers the pressure medium necessarily penetrates the dilatometer cell , the operation of which relies on the free motion of the sample and the capacitor plates , the experimentally accessible temperature - pressure range is restricted to the liquid ( or gas ) phase of @xmath2he .
this implies limitations of the operation range to @xmath38 25.5k at @xmath25 = 250mpa and @xmath38 14.2k at @xmath25 = 100mpa , for example .
apart from this constraint , liquid @xmath2he provides an ideal pressure - transmitting medium as it guarantees ideal hydrostatic - pressure conditions . a first attempt for capacitive dilatometry under high gas pressure , operating under quasi - constant - pressure conditions ,
has been reported in ref .. the performance of the apparatus described here is demonstrated by dilatometric measurements both at ambient pressure and @xmath25 = 104mpa on the natural mineral azurite , cu@xmath4(co@xmath4)@xmath5(oh)@xmath5 , a quasi - one - dimensional spin s = 1/2 heisenberg antiferromagnet a good realization of a distorted diamond chain.@xcite the expansivity data demonstrate a significant effect of pressure , indicating a strong pressure dependence of the magnetic interactions in this system .
for the thermal expansion measurements under variable pressure conditions , a high - resolution dilatometer cell , made of high - purity ( 99.999% ) silver , has been used .
the design of the cell is similar to the one described in ref . except for the reduced overall dimensions of the cell , especially the diameter of the capacitor plates which are 17 mm and 14 mm for the lower and upper plates , respectively .
for such a capacitive dilatometer , and assuming an ideal plate capacitor , the measured length change of the sample , @xmath39 , in response to a temperature change , @xmath40 = @xmath41 , is given by@xcite @xmath42 where _
c@xmath43 _ = _ c(@xmath44 _ are the capacitance values read off at the temperatures @xmath45 , @xmath23@xmath46 is the free space permittivity , @xmath47 is the dielectric constant of the medium between the capacitor plates of radius _ r _ , and @xmath48 = @xmath49 ) .
the constant @xmath50 , accounting for inhomogeneities of the electric field at the border of the capacitor plates,@xcite can be evaluated from geometrical parameters of the dilatometer cell .
a three - terminal capacitance bridge ( andeen hagerling ah2550a ) is used for measuring the capacitance _ _ c__@xmath43 . with the present cell , length changes as small as 5@xmath5110@xmath52 @xmath53
can be resolved .
the somewhat reduced resolution of this cell , compared to the cell described in ref .
, is due to the smaller size of the capacitor plates .
the dilatometer cell is placed inside a pressure cell , shown in fig.[pressure - cell ] .
the cell , designed in cooperation with the institute of high pressure physics , polish academy of sciences , unipress equipment division ( abbreviated unipress from hereon ) , is made of cube .
it has an inner / outer diameter of 36/56 mm , an inner length of 80 mm , an inner volume of @xmath54 = 81.4@xmath55 and a weight of 2.9 kg .
the maximum pressure of operation is limited to 250mpa .
the large volume of the cell and an inner seal diameter of 35.5 mm is required to house not only the above - mentioned dilatometer cell of diameter 22 mm and volume @xmath56 20@xmath55 .
it is designed for placing an even bigger cell with a diameter 33 mm providing a higher resolution of about 10@xmath57 . for precise _ in - situ _ measurements of the pressure ,
the resistivity of a doped n - insb single crystal is measured and used as a pressure gauge , cf.section @xmath58 the crystal is placed on the inner side of the plug which is unscrewed for assembling the sample into the dilatometer cell .
the plug is sealed against the pressure - cell body with a brass seal coated with tin .
a 60 nm torque is needed in order to screw the retaining screw so that the seal is crushed properly between the plug and the pressure cell .
the plug also includes electric feedthroughs for connecting the pressure gauge and the capacitor .
the four current - type electric feedthroughs , in the form of a cube cone provided with a single insulated cable of 0.8 mm , are pressed into the corresponding conical holes in the plug .
a single measuring - type electric feedthrough is formed by a cube cone , surrounded by twelve enamel - insulated cu leads of 0.12 mm .
this cone is also pressed in its corresponding conical hole .
the sealing material for both cases is hygroscopic pyrophyllite .
the capillary connector is sealed against the cell body using the principle cone - to - cone , widely used in high - pressure techniques .
( color online ) .
sketch of the pressure cell ( made of cube ) with an allowed maximum pressure of operation of 250mpa : 1 capillary with a connector on the top of the pressure cell , 2 pressure - cell body , 3 dilatometer cell , 4 n - insb pressure sensor , 5 metal seal , 6 plug with electrical feedthroughs , 7 retaining screw . ]
drawing of the lower part of the home - made insert holding the pressure cell . ]
the pressure cell is mounted to a home - made cryostat insert , shown in fig.[insert ] . in designing the insert ,
special care has been taken to ensure smooth and well - controlled temperature sweeps of the pressure cell and the dilatometer located therein . to this end
, the pressure cell is placed inside a thick - walled ( 2 mm ) copper can ( @xmath59 66 mm ) .
the can , wrapped by a thin heater foil ( kapton(r ) , thermofoil , minco company ) to ensure a homogeneous heat input , is surrounded by a stainless steel vacuum can .
the lower tail of the vacuum can ( @xmath59 75 mm ) feeds into the bore of a superconducting solenoid providing a magnetic field of maximally 12/14 t at 4.2/2.2k at the sample side .
the copper and vacuum can are connected to pump lines ( concentric tubes ) which can be vented separately with @xmath2he exchange gas . inside the copper
can the amount of exchange gas ( typically 0.02mpa at room temperature ) is kept constant throughout the whole accessible temperature range .
this is sufficient to provide a good thermal contact between all parts within the can .
in contrast , the amount of exchange gas inside the stainless steel can is varied , depending on the temperature range .
it provides the thermal coupling of the pressure cell to the surrounding @xmath2he - bath . for a given temperature interval the amount of exchange gas is adjusted such that the power applied by the temperature controller ( lakeshore ls 340 ) to the heater foil stays below 3.6 watt .
capillary heaters , mounted to copper blocks , are installed around the lower part of the capillary close to the pressure cell , see fig.[insert ] .
the heaters consist of 3 @xmath60 resistors , connected in series , with a maximum power of 10watt . in order to monitor the temperature change after switching on the capillary heater , a thermometer ( cernox resistor type cx-1080-cu )
is placed on the upper copper block . for pressurizing the cell at lower temperatures ,
the temperature of this point of the capillary is kept about 10k above the temperature of the pressure cell , to avoid a blockage of the capillary due to the freezing of spurious oxygen impurities in the @xmath61he gas . for the thermometry a pair of cernox resistors ( cx-1080-cu and cx-1050-cu ) , covering an overall temperature range 1.4k @xmath62 300k ,
have been installed on top of the pressure cell ensuring excellent thermal anchoring to the pressure cell and the dilatometer . on demand , a second set of thermometers
can be placed in the field - compensated region of the magnet , 354 mm above the centre of the main field , where @xmath63 0.01 t at maximum field value .
these sensors enable thermometry to be performed without accounting for magnetoresistance effects of the sensors . with this setup , reproducible temperature sweeps with sweep rates of typically 1.5k / h
can be performed , both upon decreasing and increasing temperatures .
the deviations from a preset in - time linear variation are less than 0.1mk for @xmath64 20k and 0.2mk for 20k @xmath65 300k .
the pressure cell is connected by a cube capillary with an inner / outer diameter of 0.3/3 mm and a length of @xmath33 10 m to an external pressure reservoir hold available at room temperature .
the reservoir , required to ensure , to a good approximation , constant - pressure conditions during temperature sweeps of the pressure cell , consists of two components .
( 1 ) a standard gas bottle of volume @xmath66 = 5@xmath5110@xmath61@xmath55 , filled with high - purity ( 99.999% ) @xmath2he gas .
this bottle is attached to ( 2 ) a compressor system , specially designed in cooperation with unipress , to which the capillary is connected . for low pressures
@xmath1 18mpa , the compressor is inactive so that the gas bottle itself serves as a reservoir . in this configuration
, the large volume of the gas bottle , as compared to the pressure cell , of @xmath67 , ensures that the pressure inside the cell stays constant within 0.1mpa during temperature changes ( see , _ e.g. _ , curve ( a ) in fig.[pressure - stability ] for a temperature sweep at a pressure of @xmath25 = 9.5mpa ) . for higher pressures @xmath68 18mpa
, the compressor is activated and pressurizes the gas from the bottle to the desired value .
the compressor consists of a low - pressure ( @xmath1 70 mpa ) and high - pressure ( 70mpa @xmath69 250mpa ) stage of volume @xmath70 = 720@xmath71 and @xmath72 = 103@xmath71 , respectively .
in addition , the compressor is equipped with an active pressure control monitoring the pressure in the high-@xmath25 stage .
in fact , the micropump can introduce a pressure change @xmath73 10@xmath35 .
however , the stability of the active pressure - control system is limited by the resolution of the manometer which determines the minimum pressure change @xmath74 the system can compensate for . in the present configuration
@xmath74 = 0.5mpa . as safety precautions
, the gas - pressure cell and all pressure - carrying parts of the compressor are located behind safety shields in order to protect the operator .
the capillary is fixed to a steel safety rope .
the pressure connectors should not be tightened under pressure .
in principle , eq . [ dell1 ] can also be used for analysing data taken under helium - gas pressure . for these finite - pressure measurements , however ,
pressure - induced changes of the dielectric constant @xmath75@xmath76 of the pressure - transmitting medium @xmath2he have to be taken into account . for temperatures and pressures sufficiently off the critical point of helium at _ p@xmath77 _ = 0.226mpa and _ t@xmath77 _ = 5.2k,@xcite the variation of @xmath75@xmath76 with pressure can be estimated by using the clausius - mossotti equation @xmath78 where _
@xmath79 _ is avogadro s number , _
@xmath80@xmath81 _ is the mass density , _
m _ is the molecular weight , @xmath82 is the molecular polarizability , and _
n _ is the number density .
( color online ) .
the variation of the dielectric constant @xmath83 of @xmath2he with temperature for various pressures calculated on the basis of eq.[clausius - mossotti01 ] by treating @xmath2he as a van der waals gas.@xcite the calculations were performed by assuming a constant volume .
inset : the variation of @xmath83 with pressure at @xmath84 = 302.3k and fixed volume .
open circles are the literature values@xcite and closed triangles are the calculated values . ]
the clausius - mossotti equation works well for describing various classes of dielectric liquids and gases over wide ranges of pressure and temperatures , see , _ e.g. _ refs . .
its application to @xmath61he , treated as a van der waals gas with the parameters ( @xmath85 = 0.03457l@xmath86atm / mol@xmath86 and @xmath87 = 0.0237l / mol),@xcite for @xmath1 250mpa and 4k @xmath0 300k yields an increase of @xmath75@xmath88 towards lower temperatures , reaching a maximum of 1.065 at 4k as shown in fig.[dielectric01].@xcite in the inset of fig.[dielectric01 ] we compare our model calculations for @xmath75@xmath88 with experimental results , unfortunately available only in a very limited parameter range .
as the figure indicates , our model reproduces well the increase of @xmath75@xmath76 observed for an isothermal pressure sweep at @xmath89 = 302.3k.@xcite at lower temperatures @xmath90 100k , however , test measurements on a high - purity ( 99.999% ) copper reference material , where a vanishingly small pressure effect in the expansivity is expected , see section * iii.b . *
below , indicate that eq.[clausius - mossotti01 ] is insufficient to describe @xmath75@xmath88 with the required accuracy .
part of the inadequacy of the description applied is likely to be related to the growth of critical density fluctuations of @xmath61he to @xmath75@xmath88 upon approaching the critical point ( @xmath91 ) , which are not properly treated in the above ansatz . for practical reasons , we therefore decided to account for the effect of @xmath75@xmath92 by experimentally determining the cell effect " at varying constant - pressure conditions , which includes the effect of pressure on @xmath75@xmath76(_p , t _ ) .
( color online ) . measured length changes at 0.1 mpa and 104 mpa for a copper sample used as a reference . ] fig .
[ p - inducedcelleffect ] shows length changes as a function of temperature at 0.1 mpa and 104 mpa for a copper sample used as a reference with a starting temperature @xmath93 = 300 k. the departure of the curve at 104 mpa from the data at 0.1 mpa at lower temperatures reflect the influence of pressure on @xmath75@xmath76(_p , t _ ) .
for the experiments on azurite , performed at the same pressure values of 0.1 mpa and 104 mpa , these data sets have been used in eqs .
[ celleffect3 ] and [ alpha ] for determining @xmath94 and @xmath95 ( see fig .
[ azurite01 ] ) . + in order to extract the relative length change of the sample at a given pressure from the measured data , see eq.[celleffect3 ]
below , all possible background effects have to be considered .
this includes the thermal expansion of the dilatometer cell and all other effects resulting from the presence of the dielectric medium @xmath61he at finite pressure .
the latter includes changes of the dielectric constant as a function of pressure and temperature as well as pressure - induced changes of the geometric cell parameters .
all these background effects are summarized in a generalized cell effect " which can be determined by measuring a proper reference material .
the requirements for such a reference material are that ( i ) it should be a simple system with a well - known and reproducible thermal expansion , which ( ii ) is not affected significantly by applying pressure in the pressure range of interest . high - purity copper serves this purpose since its thermal expansivity @xmath96 is well documented,@xcite and copper has a relatively small isothermal compressibility @xmath14 at room temperature , corresponding to a bulk modulus @xmath97 = @xmath98 = 132gpa.@xcite the variation of @xmath97 with pressure for cu at room temperature was determined to @xmath99 = 5.48,@xcite yielding an increase of @xmath97 by 0.43% at 104mpa . given that the product of the volume expansion coefficient @xmath6 and @xmath97 is , to a good approximation , a pressure - independent quantity for a given substance , _
i.e. _ , @xmath100 = @xmath101 = @xmath102 , see ref . and
references cited therein , a correspondingly small effect on the thermal expansivity of copper at 104mpa can be expected . at lower temperatures an even smaller effect on @xmath97 , and with it on @xmath6 ,
is expected because of the hardening of the material .
+ the relative length change of the sample at a constant pressure @xmath103 , @xmath94 , is then given by @xcite : @xmath104\\ & + \bigg(\frac{\delta\emph{l}}{\emph{l}}\bigg)^{cu}_{lit}(\emph{t , p = 0 } ) , \end{split } \label{celleffect3}\ ] ] where @xmath105 are the measured relative length changes of the sample , which includes the cell effect , and @xmath106 are the measured relative length changes of the copper sample used as a reference .
+ the coefficient of thermal expansion @xmath95 = dln@xmath107 at constant pressure @xmath108 is then approximated by @xmath109}{(t_2-t_1 ) } \label{alpha}\ ] ] with @xmath110 = @xmath111 , @xmath93 is the starting temperature of the experiment and @xmath89 = @xmath112 .
+ as an example for the data analysis , we show in fig.[data - analysis ] measured length changes of the sample ( left scale ) and changes of the capacitance ( right scale ) in a measurement performed at @xmath25 = 104mpa for the quasi - one - dimensional spin-1/2 antiferromagnet azurite , cf.section * iv.b.*. a peculiarity of the data set are the step - like changes in @xmath113 as a function of temperature .
these discontinuous changes are a consequence of the active pressure control , cf . section * iv.a . * , which compensates for temperature - induced changes of the pressure once @xmath74 has reached a preset threshold of @xmath74 = 0.5mpa in the present case .
the red spheres represent the average of the data between two subsequent pressure changes , with respect to both temperature and @xmath113 ( or @xmath114 obtained from eq.[dell1 ] ) .
the set of @xmath115 data derived from eq.[celleffect3 ] are then used to determine @xmath116 according to eq.[alpha ] .
( color online ) .
variation of the measured length changes ( left scale ) and capacitance ( right scale ) upon changing the temperature in a measurement on azurite at @xmath25 = 104mpa .
step - like changes are due to the active pressure control .
red spheres represent the average of the data between two subsequent steps , with respect to both axes ( _ t , c _ ) . ]
for a precise determination of the pressure at the sample site an n - insb single crystal , placed inside the pressure cell , is used as a pressure gauge .
the resistance of the gauge is measured in a four - terminal configuration using a nanovoltmeter ( keithley 2182a ) and a constant - current source ( hp 3245a ) operating at a maximum allowed current of 50ma . for each data point ,
an average is taken over typically six readouts with alternating polarity of the current in order to increase the signal - to - noise ratio and for getting rid of the effects of thermoelectric voltages . a prerequisite for its use as
a pressure gauge is an accurate determination of the effect of temperature changes on the resistance at @xmath25 = 0 , @xmath117 , of the n - insb crystal . to this
end the gauge has been calibrated as a function of temperature by using calibrated cernox resistors ( cx-1080-cu and cx-1050-cu ) as a reference . at a given temperature @xmath89
the pressure @xmath118 , in units of gpa , is related to the resistance @xmath119 according to the formula@xcite : @xmath120\cdot\mathrm{ln}\bigg[\frac{r(p , t)}{r(0,t)}\bigg ] .
\label{empiricalformula}\ ] ] with this setup pressure changes within the pressure cell of @xmath32__p _ _
@xmath121 @xmath122 0.1mpa can be resolved . in fig.[pressure - stability ] we show the variation of pressure upon increasing the temperature for measurements performed at @xmath25 = 9.5mpa ( curve ( a ) ) , 21mpa ( curve ( b ) ) and 104mpa ( curve ( c ) ) . apart from a slight increase by about 1mpa , corresponding to @xmath73 1% , for the high - pressure experiment ( curve ( c ) ) upon increasing the temperature over a wide range from 12 to 88k , the data reveal an excellent stability against temperature variations .
( color online ) .
variation of pressure inside the pressure cell , as measured by the n - insb pressure gauge , for @xmath25 = 9.5mpa ( a ) using the gas bottle as a reservoir , and 21mpa ( b ) as well as 104mpa ( c ) using the high - pressure compressor with active pressure control . for all measurements a sweep rate of 1.5k / h was used . ]
the performance of the dilatometer was checked by carrying out thermal expansion measurements both at ambient pressure and @xmath25 = 104mpa helium - gas pressure . for the latter experiment ,
the gas bottle in combination with the compressor was used as reservoir .
the material under investigation was a single crystal of the natural mineral azurite , cu@xmath4(co@xmath4)@xmath5(oh)@xmath5 , a quasi - one - dimensional spin s = 1/2 heisenberg antiferromagnet , which is a good realization of a distorted diamond chain .
this material has attracted considerable interest recently due to the discovery of a plateau in the low - temperature magnetization at 1/3 of the saturation magnetization.@xcite the crystal studied here stems from the same large single crystal explored in ref ..
fig.[azurite01 ] shows the uniaxial thermal expansion coefficient along the [ 010]-direction , @xmath123 , of azurite for temperatures @xmath124 100k at ambient pressure ( red spheres ) and at a pressure _ p _
= 104mpa ( olive squares ) .
the ambient - pressure results , which coincide with published data,@xcite reveal three distinct anomalies , seen also in magnetic susceptibility data , which can be assigned to characteristic energy scales of azurite.@xcite the rounded peak anomaly around 20k marks the formation of spin singlet s@xmath125 = 0 dimers , involving two thirds of the material s
cu@xmath126 s = 1/2 spins .
the corresponding antiferromagnetic intra - dimer exchange coupling @xmath127 33k represents the dominant magnetic energy scale of the system.@xcite the minimum around 4k has been assigned to interacting monomers , involving the remaining other one third of the cu@xmath126 spins , which form an antiferromagnetic spin-1/2 heisenberg chain .
the corresponding antiferromagnetic monomer - monomer interaction has been estimated to @xmath128 4.6k.@xcite finally , the phase transition anomaly at _ t@xmath129 _
= 1.88k , shown in the inset of fig.[azurite01 ] on expanded scales , reflects the onset of long - range antiferromagnetic order as a consequence of weak three - dimensional magnetic interactions among the monomers .
+ ( color online ) .
uniaxial thermal expansion along the [ 010]-direction of azurite , cu@xmath4(co@xmath4)@xmath5(oh)@xmath5 , measured at ambient pressure and a pressure of _ p _
inset ( a ) shows a blow - up of the @xmath25 @xmath121 0.1 mpa data around the phase transition into long - range antiferromagnetic order .
inset ( b ) displays the 20k anomaly on expanded scales .
the solid lines correspond to least squares fits to the function which serve to better visualize the shift in the position of the maximum under pressure . ]
fig.[azurite01 ] clearly demonstrates that an increase of the pressure from ambient pressure to _ p _
= 104mpa has a strong effect on the low - temperature expansivity of azurite .
note that due to the solidification of @xmath2he , the measurements at _ p _
= 104mpa were limited to @xmath130 14.5k.@xcite the data at _ p _
= 104mpa show a surprisingly strong reduction in the absolute value of the thermal expansion coefficient compared to the ambient - pressure data .
this is accompanied by a suppression of the height of the pronounced maximum in @xmath123 around 20k .
since changes of the lattice expansivity under pressure of 104mpa are expected to be less than 1% , cf.the corresponding discussion for cu in section @xmath131 , these results indicate the presence of strong magnetic contributions to @xmath82 in the entire temperature range investigated .
these magnetic contributions are most strongly suppressed at intermediate temperatures around 50k when pressure of 104mpa is applied .
a thorough inspection of the data around the 20k maximum , cf.inset ( b ) of fig.[azurite01 ] , discloses a slight shift of the position of the maximum to lower temperatures . in order to quantify the shift ,
the data have been fitted by a polynomial of fourth order .
the solid lines in the inset of fig.[azurite01 ] , corresponding to least - squares fits of these polynomials , yield a shift of the position of the maximum by about @xmath151k at 104mpa .
a suppression of the corresponding characteristic energy @xmath132 under pressure is qualitatively consistent with the expectation from the grneisen formalism , yielding @xmath133 = @xmath15dln@xmath134/dln@xmath8 @xmath135 0 , since the volume expansion coefficient ( @xmath6 @xmath121 @xmath1513@xmath5110@xmath34 k@xmath136 ) around this anomaly has a negative sign@xcite while the specific heat is positive.@xcite on a quantitative level , the present results are in excellent agreement with the observations in magnetic measurements under pressure up to about 600mpa @xcite yielding @xmath137 @xmath151k/100mpa .
we have constructed and assembled a thermal expansion setup enabling high - resolution dilatometric measurements to be performed under helium ( @xmath2he ) gas pressure for temperatures 1.4k @xmath0 300k and pressures @xmath138 250mpa .
the system excels by a very good temperature control and a high pressure stability during temperature sweeps .
the performance of the system has been demonstrated by measurements on the quasi - one - dimensional spin-1/2 heisenberg antiferromagnet azurite , cu@xmath4(co@xmath4)@xmath5(oh)@xmath5 , where pronounced pressure effects could be observed at low temperatures .
+ m. c. r. gibson , k. c. rule , a. u. b. wolter , j .- u .
hoffmann , o. prokhnenko , d. a. tennant , s. gerischer , m. kraken , f. j. litterst , s. sllow , j. schreuer , h. luetkens , a. brhl , b. wolf , and m. lang , phys .
b * 81 * , 140406(r ) ( 2010 ) .
h. jeschke , i. opahle , h. kandpal , r. valenti , h. das , t. saha - dasgupta , o. janson , h. rosner , a. brhl , b. wolf , m. lang , j. richter , s. hu , x. wang , r. peters , t. pruschke , and a. honecker , phys . rev
. lett . * 106 * , 217201 ( 2011 ) . | we report on the realization of a capacitive dilatometer , designed for high - resolution measurements of length changes of a material for temperatures 1.4k @xmath0 300k and hydrostatic pressure @xmath1 250mpa .
helium ( @xmath2he ) is used as a pressure - transmitting medium , ensuring hydrostatic - pressure conditions .
special emphasis has been given to guarantee , to a good approximation , constant - pressure conditions during temperature sweeps .
the performance of the dilatometer is demonstrated by measurements of the coefficient of thermal expansion at pressures @xmath3 0.1mpa ( ambient pressure ) and 104mpa on a single crystal of azurite , cu@xmath4(co@xmath4)@xmath5(oh)@xmath5 , a quasi - one - dimensional spin s = 1/2 heisenberg antiferromagnet .
the results indicate a strong effect of pressure on the magnetic interactions in this system . |
coupled nonlinear oscillators in the form of arrays arise in many branches of science .
these coupled arrays are often used to model very many spatially extended dynamical systems explaining reaction - diffusion processes .
one can visualize such an array as an assembly of a number of subsystems coupled to their neighbours , or as coupled nonlinear networks ( cnns ) [ chua & yang , 1988 ; chua , 1995 ] .
theoretically one can consider a system of coupled ordinary differential equations ( odes ) to represent a macroscopic system , in which each of the odes corresponds to the evolution of the individual subsystem .
specific examples are the coupled array of anharmonic oscillators [ marn & aubry , 1996 ] , josephson junctions [ watanabe _ et al . _ ,
1995 ] , continuously stirred tank reactors ( cstr ) exhibiting travelling waves [ dolnik & marek , 1988 ; laplante & erneux , 1992 ] , propagation of nerve impulses ( action potential ) along the neuronal axon[scott , 1975 ] , the propagation of cardiac action potential in the cardiac tissues [ allesie _ et al .
_ , 1978 ] and so on . for the past few years
several investigations have been carried out to understand the spatio - temporal behaviours of these coupled nonlinear oscillators and systems .
the studies on these systems include the travelling wave phenomena , turing patterns , spatio - temporal chaos and synchronization [ feingold _ et al .
_ , 1988 ; muuzuri _ et al .
_ , 1993 ; prez - muuzuri _ et al .
_ , 1993 ; sobrino _ et al .
_ , 1993 ; prez - muuzuri _ et al .
_ , 1994 , prez - muuzuri _ et al .
_ , 1995 ; kocarev & parlitz , 1996 ] .
one of the interesting behaviours of these coupled oscillators is that they can exhibit a failure mechanism for travelling waves which can not be observed in the homogeneous continuous models [ keener , 1987 ] .
similar failure phenomena in the nerve impulse propagation have been observed in biological experiments [ balke _ et al .
_ , 1988 ; cole _ et al .
_ , 1988 ] .
recently various spatio - temporal patterns have been found in discrete realizations of the reaction - diffusion models by means of arrays of coupled nonlinear electronic circuits [ prez - muzuri _ et al .
_ , 1993 ] .
for example prez - muuzuri and coworkers studied the various spatio - temporal patterns in a model of discretely coupled chua s circuits .
this system is a _ three _ variable model and coupled to its neighbours by means of a linear resistor . in this paper
we consider similar arrays of coupled nonlinear electronic circuits consisting of murali - lakshmanan - chua ( mlc ) circuits as a basic unit , which is basically a _ two _ variable model .
the dynamics of the driven mlc circuit including bifurcations , chaos , controlling of chaos and synchronization has been studied by murali _
[ 1994a ; 1994b ; 1995 ; lakshmanan & murali , 1996 ] . of particular interest among coupled arrays is the study of _ diffusively coupled driven systems _ as they represent diverse topics like faraday instability [ lioubashevski _ et al .
_ , 1996 ] , granular hydrodynamics [ umbanhawar _ et al .
_ , 1996 ; kudrolli _ et al .
_ , 1997 ] , self organized criticality [ bak & chen , 1991 ] and so on .
identification of localized structures in these phenomena has been receiving considerable attention very recently [ fineberg , 1996 ] . under these circumstances ,
study of models such as the one considered in this paper can throw much light on the underlying phenomenon in diffusively coupled driven systems . in the following by considering one and two dimensional arrays of coupled mlc circuits , we have studied the various spatio - temporal patterns in the presence as well as in the absence of external force .
we present a brief description on the one and two dimensional arrays of mlc circuits in sec .
then , in sec .
[ sec3 ] , in the absence of periodic external force , we study the propagation phenomenon of travelling wave front and its failure below a critical diffusive coupling coefficient . we show that the propagation phenomenon occurs due to a loss of stability of the steady states via subcritical bifurcation coupled with the presence of necessary basin of attraction of the steady states for appropriate diffusive coupling coefficient .
we also study the effect of weak coupling on the propagation phenomenon which causes a blocking in the propagation due to the presence of a stationary front at the weakly coupled cell .
in addition we also study the onset of turing instability leading to the spontaneous formation of hexagonal patterns in the absence of any external force .
further as mentioned earlier it is of great physical interest to study the dynamics of the coupled oscillators when individual oscillators are driven by external forces . in sec .
[ sec4 ] , we study the spatio - temporal patterns in the presence of periodic external force and investigate the effect of it on the propagation phenomenon and turing patterns . depending upon the choice of control parameters , a transition from hexagons to regular rhombic structures , hexagons to rolls and then to breathing oscillations from hexagons are observed .
the presence of external force with sufficient strength removes the penta - hepta defect pair originally present in the spontaneously formed hexagonal patterns leading to the formation of regular rhombic structures .
the inclusion of external periodic force can also induce a transition from hexagons to rolls provided there are domains of small roll structures in the absence of force .
we further show that in the region of hopf - turing instability , the inclusion of external periodic force with sufficiently small amplitude induces a type of breathing oscillations though the system shows a regular hexagonal pattern in the absence of any external force .
finally , we also study the spatio - temporal chaotic dynamics of the one dimensional array of mlc circuits in sec .
[ sec5 ] when individual oscillators oscillate chaotically . in this case , the emergence of spatio - temporal patterns depends on the system size . for larger size , above a critical number of cells , we observe a controlled space - time regular pattern eventhough the single mlc circuit itself oscillates chaotically .
however , synchronization occurs for a smaller system size , below the threshold limit . finally , in sec .
[ sec6 ] we give our conclusions .
as mentioned in the introduction the circuit proposed by murali , lakshmanan and chua ( mlc ) is one of the simplest second order dissipative nonautonomous circuit having a single nonlinear element [ murali _ et al_. , 1994a ] .
brief details of its dynamics are given in the appendix a for convenience . here
we will consider one and two dimensional arrays of such mlc circuits , where the intercell couplings are effected by linear resistors .
[ fig1](a ) shows a schematic representation of an one dimensional chain of resistively coupled mlc circuits . extending the analysis of single mlc circuit as given in the appendix a
, the dynamics vs. @xmath0 for the parametric choice @xmath1 @xmath2 in eq .
( c ) graph showing @xmath3 vs. @xmath0 for the parametric choice @xmath4 @xmath5 in eq .
( 2 ) . ] of the one dimensional chain can be easily shown to be governed by the following system of equations , in terms of the rescaled variables ( see appendix a ) , [ eqn1all ] @xmath6 where @xmath7 is the diffusion coefficient , @xmath8 is the chain length and @xmath3 is a three segment piecewise linear function representing the current voltage characteristic of the chua s diode , @xmath9 in ( [ eqn2 ] ) @xmath10 , @xmath11 and @xmath12 are the three slopes .
depending on the choice of @xmath10 , @xmath11 and @xmath12 one can fix the characteristic curve of the chua s diode . here
@xmath13 corresponds to the dc offset .
of special importance for the present study are the two forms of @xmath3 shown in figs .
[ fig1](b ) and [ fig1](c ) .
[ fig1](b ) corresponds to the choice of the parameters @xmath14 while fig .
[ fig1](c ) corresponds to the parameters @xmath15 .
as noted in the appendix a , the significance of the characteristic curve in fig .
[ fig1](b ) is that it admits interesting bifurcations and chaos , while fig .
[ fig1](c ) corresponds to the possibility of bistable states with asymmetric nature which is a prerequisite for observing wave front propagation and its failure . in the following studies we will consider a chain of @xmath16 mlc circuits . as in the case of one dimensional array
one can also consider a two dimensional array with each cell in the array being coupled to four of its nearest neighbours with linear resistors .
the model equation can be now written in dimensionless form as [ eqn3all ] @xmath17 this two dimensional array has @xmath18 cells arranged in a square lattice . in our numerical study
we will again take @xmath16 . in the following sections we present some of the interesting dynamical features of the array of mlc circuits such as propagation of wave front and its failure , effect of weak coupling in the propagation , turing patterns , effect of external periodic forcing in the turing patterns and spatio - temporal chaotic dynamics .
we have used zero flux boundary conditions for the study of propagation phenomenon and turing patterns and periodic boundary conditions for the study of spatio - temporal chaos .
transport processes in living tissues , chemical and physiological systems have been found to be associated with special types of waves called travelling waves .
earlier , continuous models were created to describe the wave propagation phenomenon in these systems , but these failed to cover all the important aspects of travelling wave behaviour .
one of the most important of them is the so called travelling wave propagation failure , occurring at weak coupling between cells .
it has been proved by keener [ 1987 ] that propagation failure can not be observed in a continuous , one variable , homogeneous reaction - diffusion system . therefore to study this kind of phenomenon , one has to use discrete models .
recently travelling wave phenomenon and its failure have been studied in arrays of discretely coupled chua s circuits [ prez - muuzuri _ et al . _ ,
wave propagation and its failure have also been observed even in one variable models like discrete nagumo equation [ leenaerts , 1997 ] . in this section
we carry out a study of such wave propagation phenomenon in one and two dimensional arrays of murali - lakshmanan - chua s ( mlc ) circuit oscillators , without forcing , and investigate the mechanism by which such a failure occurs . in order to observe wave front propagation in the one dimensional array of mlc circuits , we numerically integrated the eqs .
( 1 ) using fourth order runge - kutta method . in this analysis
we fix the parameters at @xmath19 @xmath20 so that the system admits bistability and this choice leads to the asymmetry characteristic curve for the chua s diode as shown in fig . [ fig1](c ) .
the existence of bistable state in the asymmetric case is necessary to observe a wave front .
zero flux boundary conditions were used in the numerical computations , which in this context mean setting @xmath21 and @xmath22 at each integration step ; similar choice has been made for the variable @xmath23 also . to start with , we will study in this section the force - free case , @xmath24 , and extend our studies to the forcing case @xmath25 in the next section .
the choice of the values of the parameters guarantees the existence of two stable equilibrium points @xmath26 , @xmath27 and @xmath28 , @xmath29 for each cell . in the particular case ,
corresponding to the above parametric choice , each cell in the array has three equilibrium points at @xmath30 , @xmath31 and @xmath32 . out of these three equilibrium points , @xmath33 and @xmath34
are stable while @xmath35 is unstable . due to the asymmetry in the function @xmath3 for the present parametric choices ,
the basin of attraction of the point @xmath33 is much larger than that of @xmath34 and it is harder to steer a trajectory back into the basin @xmath34 once it is in the basin of @xmath33 .
now we choose an initial condition such that the first few cells in the array are excited to the @xmath36 state ( having a large basin of attraction compared to that of @xmath37 ) and the rest are set to @xmath37 state . in other words a wave front in the array
is obtained by means of the two stable steady states . on actual numerical integration of eqs.([eqn1all ] ) with @xmath16 and with the diffusion coefficient and ( b ) propagation failure for @xmath38 . ]
chosen at a higher value , @xmath39 , a motion of the wave front towards right ( see fig .
[ fig2](a ) ) is observed , that is a travelling wave front is found .
after about 80 time units the wave front reaches the @xmath40 cell so that all the cells are now settled at the more stable state @xmath41 as demonstrated in fig .
[ fig2](a ) .
when the value of @xmath7 is decreased in steps and the analysis is repeated , the phenomenon of travelling wavefronts continues to be present
. however , below a critical value of the diffusion coefficient @xmath42 a failure in the propagation has been observed , which in the present case turns out to be @xmath43 .
this means that the initiated wavefront is unable to move as time progresses and fig .
[ fig2](b ) shows the propagation failure for @xmath38 . to understand the failure mechanism , we have carried out a detailed case study of an array of five coupled mlc circuits , represented by the following set of equations
we have verified our conclusions for the case of 3 , 4 , 6 and 7 oscillators also .
the analysis can be in principle extended to arbitrary but finite number of oscillators . from eq .
( [ eqn1all ] ) , the present array can be represented by the following set of ten coupled first order odes : @xmath44 where @xmath3 is as given in eq .
( [ eqn2 ] ) and the parameters are fixed at @xmath45 , @xmath46 , @xmath47 , @xmath48 , @xmath49 and @xmath50 . by carrying out a numerical analysis of the system ( [ eqsta ] ) with zero flux boundary conditions , we identify the following main results .
[ [ casea ] ] we solve the above system of equations ( [ eqsta ] ) for the chosen initial condition
@xmath51 @xmath52 @xmath53 @xmath54 @xmath55 @xmath56 @xmath57 @xmath58 @xmath59 @xmath59 @xmath60 and @xmath61 , @xmath62 for different values of the diffusion coefficient @xmath7 in the region @xmath63 .
note that the values @xmath64 and @xmath65 correspond to the stable equilibrium points @xmath36 and @xmath37 , respectively , of the uncoupled mlc oscillators .
when @xmath7 is decreased downwards from @xmath66 we find that in the region @xmath67 propagation of wavefront occurs as in fig .
[ fig2](a ) ( of the @xmath16 cells system ) . at the critical value @xmath38
, the wavefront fails to propagate and asymptotically ends up in a stationary wavefront which is nearer to the initial condition ( as in fig .
[ fig2](b ) ) .
the wavefront failure phenomenon persists for all values of @xmath68 .
[ [ caseb ] ] on the other hand , if we choose the initial condition as @xmath51 @xmath52 @xmath53 @xmath54 @xmath55 @xmath56 @xmath69 @xmath70 @xmath71 @xmath72 @xmath73 and @xmath61 , @xmath74 then the critical value @xmath7 , where wavefront propagation failure occurs , turns out to be at a slightly higher value @xmath75 .
in order to understand the failure mechanism of wavefront propagation we consider all the possible steady states ( equilibrium states ) associated with eqs.([eqsta ] ) for nonzero values of @xmath7 .
it is easy to check that eq .
( [ eqsta ] ) possesses a maximum of @xmath76 steady states .
first we consider the three trivial steady states which can be easily obtained by considering @xmath77 and @xmath78 , namely , @xmath79
@xmath56 @xmath80 , @xmath81 @xmath56 @xmath82 and @xmath83 @xmath56 @xmath84 .
( here the suffix @xmath85 indicates that the steady states are the same as that of the @xmath86 case , namely @xmath36 , @xmath87 and @xmath37 ) . from a linear stability analysis (
details are given in appendix b ) we can easily conclude that the steady states @xmath79 and @xmath83 are stable while @xmath81 is unstable irrespective of the value of the diffusion coefficient @xmath7 .
then any initial condition in the neighbourhood of @xmath88 will evolve into @xmath88 asymptotically .
also one has to note that , due to asymmetry in the function @xmath3 , the basin of @xmath79 is much larger than that of @xmath83 . since we are interested here to understand the propagation failure mechanism , we concentrate on those steady states which possess stationary wavefront structures _ near _ to the two types of chosen initial conditions in the numerical analysis discussed in sec . [ secnum ] . out of the 243 available steady states ,
we select a subset of them , @xmath89 , which is nearer to the numerically chosen initial state .
these three steady states are defined in eqs.([steady ] ) of appendix b. they are identified such that the components @xmath90 and @xmath91 are in the right extreme segment of the characteristic curve , fig .
[ fig1](c ) , while @xmath92 and @xmath93 are in the left extreme segment . on the other hand @xmath94
is allowed to be in any one of the three segments .
the reason for such a choice is that we are looking for the formation of a wavefront which is closer to the chosen initial conditions in sec .
[ secnum ] .
the corresponding forms of @xmath95 as given in eq .
( [ eqn2 ] ) are then introduced in eqs.([eqsta ] ) to find the allowed steady states .
then @xmath96 corresponds to the case in which @xmath94 as well as @xmath90 and @xmath91 are in the right most segment while the rest are in the left most segment and @xmath97 and @xmath98 correspond to the cases in which @xmath94 is , respectively , in the middle and the left extreme segments of the characteristic curve while @xmath90 , @xmath91 and @xmath92 , @xmath93 are chosen as in @xmath96 .
these steady states may be explicitly obtained as a function of the diffusion coefficient @xmath7 , whose forms are given in appendix b. we now consider the linear stability of these steady states @xmath96 , @xmath97 and @xmath98 . the corresponding stability ( jacobian ) matrices are obtained from the linearized version of eq .
( [ eqsta ] ) , whose form is given in eq .
( [ jacob ] ) .
the eigenvalues are evaluated by numerical diagonalization of ( [ jacob ] ) as a function @xmath7 and the stability property analysed .
it has been found that out of these three steady states both @xmath96 and @xmath98 are stable for @xmath99 in which all of the eigenvalues of ( [ jacob ] ) are having negative real parts , while @xmath97 is unstable for all values of @xmath100 . however , the real part of atleast one of the eigenvalues associated with each of @xmath96 and @xmath98 changes its sign to positive value at @xmath101 and thereby both the stationary fronts , @xmath96 as well as @xmath98 , also lose their linear stability at this critical value .
thus for @xmath102 , all the three steady states @xmath96 , @xmath97 and @xmath98 are linearly unstable .
first let us consider the second of the initial conditions chosen in sec .
[ secnum ] . at and above the critical value @xmath101 ,
all the three steady states are unstable and so the system transits to the other nearby steady states .
these are also found to be unstable by a similar analysis , ultimately ending up in the only nearby available stable state which is @xmath79 @xmath56 @xmath103 @xmath104 @xmath104 @xmath104 @xmath105 , thereby explaining the propagation of the wavefront starting from the initial condition @xmath106
@xmath56 @xmath107 @xmath108 @xmath109 @xmath110 @xmath111 @xmath112 ( which is the second of the numerical results given in sec .
[ secnum ] ) .
however for @xmath113 , since @xmath96 is stable and is also quite closer to the initial state , the system settles down in this state itself and no propagation occurs , thereby explaining the phenomenon of propagation failure as a subcritical bifurcation when the diffusion coefficient @xmath7 is increased from @xmath114 upwards .
note that eventhough @xmath98 is also stable , it is far away from the initial state and so no transition to this state will occur .
a table of these steady states @xmath96 and @xmath98 as a function of @xmath7 ( table [ table2 ] ) is given confirming that @xmath96 is always nearer to the initial state compared to @xmath98 .
now let us consider the first of the initial conditions of our numerical analysis ( see sec .
[ secnum ] ) , where wavefront propagation occurs at @xmath38 itself and not at @xmath101 , eventhough the steady states @xmath96 and @xmath98 are still linearly stable in the region @xmath115 . in order to understand this aspect , we start analysing the nature of the basin of attraction associated with the steady state @xmath98 for different @xmath7 values numerically ( note that for the present set of initial conditions @xmath98 is nearer to it than @xmath96 ) . we find that the basin of @xmath98 shrinks as @xmath7 is increased from @xmath114 upwards and vanishes completely at @xmath101 .
we also find that the chosen initial condition , @xmath106 @xmath56 @xmath103 @xmath104
@xmath116 @xmath116 @xmath117 falls fully within the basin of @xmath98 as long as @xmath118 .
however due to the for various values of the diffusive coupling coefficient .
in the figure each line corresponds to an initial condition .
the figure is drawn by fixing the initial state of the first oscillator at @xmath119 , @xmath120 and for clarity decreasing the rest downwards in steps of 0.2 . ] shrinking nature of the basin as @xmath7 increases ( as given in figs .
( [ figbas ] ) ) , a part of the initial state falls outside the basin for @xmath67 ( figs . [ figbas](b)-(d ) ) . as a result
the chosen initial condition does not end up in the stationary front @xmath98 for @xmath121 indicating a global instability ( eventhough the state is linearly stable ) .
therefore propagation starts to occur as soon as @xmath67 , while it fails for @xmath122 and it finally ends in the stable state @xmath79 described in the previous case .
however , if an initial condition which happens to fall completely within the basin of attraction of @xmath123 , the propagation starts to occur only for values of @xmath102 .
in fact this is what happens for the second chosen initial condition ( @xmath106 @xmath56 @xmath124 @xmath109 @xmath110 @xmath111 @xmath112 , as explained above ( note that the basin structure shown in fig .
[ figbas ] does not cover this case ) . at various time units ( a )
@xmath125 , ( b ) @xmath126 , ( c ) @xmath127 and ( d ) @xmath128 for @xmath129 and @xmath130 and ( e ) the same plot for propagation failure case @xmath131 and @xmath130 at @xmath132 . ] in conclusion , the propagation failure of wave fronts in arrays of coupled nonlinear oscillators is essentially due to the existence of stationary fronts ( stable steady states ) near to the initial states for low diffusion coefficients , below a certain critical value , which lose stability via subcritical bifurcation .
this coupled with the existence of necessary basin of attraction ( for a global stability ) of the steady state gives rise to the specific critical value of @xmath133 above which propagation begins for a given initial condition . for studying the propagation phenomenon in two dimensional array of coupled mlc circuits , we have considered the case in which @xmath24 , @xmath134 and @xmath130 .
the propagation of wave front and its failure have been observed as in the case of one dimensional array .
a two dimensional array of @xmath135 coupled mlc circuits has been considered for the numerical simulation with initial conditions chosen such that the first few cells in the array are excited to the @xmath33 state and the other cells are set at the @xmath34 state . figs .
[ fig3](a)-[fig3](d ) show the propagation of the wave front in the two dimensional array for @xmath129 and @xmath130 at various time units . again , below a certain critical diffusion coefficient , @xmath131 , propagation failure occurs .
one such case is illustrated in fig .
[ fig3](e ) for @xmath131 and @xmath130 .
the phenomenon can again be explained through a linear stability analysis as in the case of the one dimensional array . in the above
, the investigation has been made by considering the system as an ideal one ( as far as the circuit parameters are concerned ) .
but from a practical point of view , there are defects in the coupling parameters which may result in a weak coupling at any of the cell in the array . in the following
we study the effect of such weak coupling on the propagation of wave front .
let us consider a weak coupling at the @xmath136 cell .
by this we mean that the @xmath136 cell in the array is coupled to its nearest neighbour @xmath137 and @xmath138 cells by resistors with slightly higher values than that of the others .
we have studied the effect of this defect on the propagation of wave fronts .
numerical simulations have been carried out by considering an one dimensional array with 100 cells , where the initial conditions are chosen as in the case of propagation phenomenon in regular one dimensional arrays ( see sec .
[ sec3a ] ) . from the numerical simulation results
, we find that there is an abrupt stop in the propagation when the wave front reaches the weakly coupled cell .
this happens when the coupling coefficient on either side of the @xmath136 cell in the array has a value even above the critical value @xmath139 for propagation failure discussed in sec .
[ sec3a ] .
cell in the one dimensional array ( eq .
( 3 ) ) . ( a )
three dimensional space - time plot and ( b ) density plot . ]
[ fig4 ] shows a blocking in the propagation of the wave front when the coupling coefficients on either side of the @xmath136 cell @xmath140 , which we call as @xmath141 , is set to @xmath142 with the rest of the coupling coefficients set to @xmath143 @xmath144 .
we observe that the actual blocking occurs when the wave front reaches the @xmath136 cell .
we can say this is a kind of failure in the propagation ( because the wavefront will never reach the last cell ( that is the @xmath40 cell in the array ) by means of blocking .
as far as the initial conditions are concerned , since the system is in the propagation region @xmath145 any choice of initial condition of wavefront structure will lead to uniform propagation upto the weakly coupled cell .
now our motivation is to find the property of the above uniform wavefront when it reaches the weakly coupled cell .
one can analyse the above phenomenon by means of a linear stability analysis of the steady states as in the case of propagation phenomenon in sec .
[ sec3b ] .
we consider again as an example the system of five coupled mlc circuits as in eqs.([eqsta ] ) in which the middle cell ( @xmath146 ) is now taken as an example to be coupled weakly to its neighbours , respectively @xmath147 and @xmath148 .
this set up can be represented by the following set of ten coupled odes : @xmath149 here @xmath141 corresponds to the coupling coefficient of the weakly coupled cell and other parameters are the same as in ( [ eqsta ] ) . as in sec .
[ sec3b ] , we calculated the steady states associated with ( [ eqweak ] ) and analysed the linear stability properties .
let us now consider the three steady states of the system ( [ eqweak ] ) , namely , @xmath89 which possess wavefront structures near the chosen initial conditions .
these steady states can be found as discussed in sec .
[ sec3b ] . by fixing @xmath150 ,
we analyse the stability of @xmath151 as a function of @xmath141 .
we find that , the steady states , @xmath123 are linearly stable for values of @xmath141 , @xmath152 and become unstable for @xmath153 via subcritical bifurcation .
however @xmath97 state continues to be unstable irrespective of the value of @xmath141 .
at this point , one may think of an array with large number of mlc circuits . since all the cells , except the @xmath136 cell , have diffusion coefficient @xmath150 which is in the propagation region , the appropriately chosen initial condition will evolve as a propagating front upto the weakly coupled cell .
when this wave front reaches the weakly coupled cell , the existence of the stationary front @xmath98 for the array due to the presence of the weakly coupled cell will block the propagating front and hence it will end up in the stationary front @xmath98 asymptotically for @xmath154 . for @xmath155 ,
the stationary state @xmath98 is no longer stable and hence the propagation occurs without any blocking .
thus the blocking of the wave front can also be attributed to the existence of a stable stationary front when weak coupling is present in the system .
another interesting dynamical phenomenon in the coupled arrays is the formation of turing patterns .
these patterns are observed in many reaction diffusion systems when a homogeneous steady state which is stable due to small spatial perturbations in the absence of diffusion becomes unstable in the presence of diffusion[turing , 1952 ] . to be specific , the turing patterns can be observed in a two variable reaction - diffusion system when one of the variables diffuses faster than the other and undergo turing bifurcation , that is , diffusion driven instability[turing , 1952 ; murray , 1989 ] . treating the coupled array of mlc circuits as a discrete version of a reaction - diffusion system , one can as well observe the turing patterns in this model also .
for this purpose , one has to study the linear stability of system ( 3 ) near the steady state . in continuous systems ,
the linear stability analysis is necessary to arrive at the conditions for diffusion driven instability .
a detailed derivation of the general conditions for the diffusion driven instability can be found in murray[1989 ] . for discrete cases one can follow the same derivation as in the case of continuous systems by considering solutions of the form @xmath156 [ murray 1989 ; muuzuri _ et al .
_ , 1995 ] . here
@xmath157 and @xmath158 are considered to be independent of the position @xmath159 @xmath160 . for eq .
( 3 ) , the criteria for the diffusion driven instability can be derived by finding the conditions for which the steady states in the absence of diffusive coupling are linearly stable and become unstable when the coupling is present .
one can easily derive from a straightforward calculation that the following conditions should be satisfied for the general reaction - diffusion system of the form given by eq .
( [ eqn3all ] ) : @xmath161 the critical wave number for the discrete system ( 3 ) can be obtained as @xmath162 combining eqs .
( [ cond:1])-([cond:2 ] ) , one obtains[muuzuri _
et al . _ ,
1995 ] the condition for the turing instability such that @xmath163 we apply these conditions to the coupled oscillator system of the present study . for this purpose ,
we fix the parameters for the two - dimensional model ( 3 ) as @xmath164 @xmath165 and verify that this choice satisfies the conditions ( [ cond:1 ] ) to ( [ cond:3 ] ) .
the numerical simulations are carried out using an array of size @xmath166 and random initial conditions near the steady states are chosen oscillators for the parameters @xmath167 , @xmath46 , @xmath168 , @xmath169 , @xmath170 , @xmath171 , @xmath172 , @xmath173 , @xmath174 and @xmath175 in eq .
( 3 ) at various time units ( a ) @xmath126 , ( b ) @xmath127 , ( c ) @xmath176 and ( d ) @xmath177 ( e ) the same figure ( d ) in the fourier space . ] for the @xmath0 and @xmath23 variables .
[ fig5](a)-[fig5](d ) show how the diffusion driven instability leads to stable hexagonal pattern ( fig .
[ fig5](d ) ) after passing through intermediate stages ( figs .
[ fig5](a)-[fig5](c ) ) .
[ fig5](e ) shows the two dimensional fourier spectrum of the hexagonal pattern ( fig . [ fig5](d ) ) .
further , the spontaneously formed patterns are fairly uniform hexagonal patterns having a penta - hepta defect pair .
these defects are inherent and very stable .
we note that such turing patterns have already been observed in an array of coupled chua s circuits[muuzuri _ et al _ , 1995 ]
. however , our aim here is to investigate the effect of external forcing on these patterns , which is taken up in the next section .
the effect of external fields on a variety of dynamical systems has been studied for a long time as driven systems are very common from a practical point of view .
for example , in a large number of dynamical systems including the duffing oscillator , van der pol oscillator and the presently studied mlc circuits , temporal forcing leads to a variety of complex dynamical phenomena such as bifurcations , quasiperiodicity , intermittency , chaos and so on[guckenheimer & holmes , 1983 ; hao bai - lin , 1984 ; lakshmanan & murali,1996 ] . also the studies on the effect of external fields in spatially extended systems
have been receiving considerably increasing interest in recent times [ pismen , 1987 ; walgraef , 1988 ; 1996 ] .
particularly , with the recent advances in identifying localized and oscillating structures and other spatio - temporal patterns in driven nonlinear dissipative systems such as granular media , driven ginzburg - landau equations and so on , it is of considerable interest to study the effects of forcing on array of coupled systems such as ( 3 ) .
motivated by the above , we investigate the effect of external forcing in the propagation of wave front and formation of turing patterns in the coupled mlc circuits in one and two dimensions .
in this subsection we study the effect of external forcing on the propagation of wave fronts . for this purpose
we consider an one dimensional array of coupled mlc circuits with initial and boundary conditions as discussed in sec .
[ sec3a ] .
now we perform the numerical integration by the inclusion of external periodic force of frequency @xmath173 in each cell of the array ( see eq .
( 1 ) ) . by varying the strength , @xmath178 , of the external force we study the behaviour of the propagating wave front in comparison with the force free case ( @xmath24 ) as discussed in sec .
[ sec3a ] .
we find that in the propagation region ( @xmath179 ) , the effect of forcing is just to introduce temporal oscillations and the propagation continues without and @xmath180 and ( b ) the partial propagation observed for the @xmath181 and @xmath180 . ]
any disturbance ( see fig .
[ fig6](a ) ) as in the case @xmath24 ( fig .
[ fig2](a ) ) .
of course this can be expected as the external force is periodic in time . however , interesting things happen in the propagation failure region discussed in sec .
[ sec3a ] . in this region , beyond a certain critical strength of the external forcing , the wavefront tries to move a little distance and then stops , leading to a partial propagation .
[ fig6](b ) shows such a partial propagation observed for @xmath180 and @xmath181 ( this may be compared with fig .
[ fig2](b ) ) .
the phenomenon can be explained by considering the propagation failure mechanism discussed in sec .
[ sec3b ] in which one may look for a spatially stationary and temporally oscillating wavefront . the initial wavefront tries to settle in the nearby stationary state .
however , the system will take a little time and space to settle due the effect of forcing combined with the transient behaviour of the system .
thus the inclusion of external forcing in the propagation failure region can induce the wavefront to achieve a partial propagation .
it is well known that the defects are inherent in very many natural pattern forming systems . in most of the pattern forming systems , the observed patterns are not ideal .
for example , the patterns are not of perfect rolls or hexagons or rhombs .
a commonly observed defect in such systems is the so called penta - hepta defect ( phd ) pair which is the bound state of two dislocations [ tsimring , 1996 ] .
experiments on spatially extended systems often show the occurrence of phd in spontaneously developed hexagonal patterns[pantaloni & cerisier , 1983 ] . in the present case also
, the existence of phd pair can be clearly seen from fig . [ fig5](d ) .
in such a situation , it is important to study the effect of external periodic force in the coupled arrays of mlc circuits .
now we include a periodic force with frequency @xmath182 and amplitude @xmath178 in each cell of the array and we numerically integrate eqs.([eqn3all ] ) using fourth order runge - kutta method with zero flux at boundaries . by fixing the frequency of the the external periodic force as @xmath173 and varying the amplitude ( @xmath178 ) we analyse the pattern which emerges spontaneously .
interestingly for @xmath183 , the defects ( phd pair ) which are present in the absence of external force ( fig . [ fig5](d ) ) , gets removed resulting in the and ( b ) the same figure in fourier space . ]
transition to a regular rhombic array .
[ fig7](a ) shows the gray scale plot of the pattern observed for @xmath183 and the corresponding plot in fourier domain ( fig .
[ fig7](b ) ) .
thus , from the above we infer that the inclusion of external periodic force can cause a transition from hexagonal pattern to rhombic structures .
we note that prez - muuzuri _
et al_. have observed , in arrays of coupled chua s circuits , similar transition from hexagons to rhombs by means of _ sidewall _ forcing [ prez - muuzuri _ et al .
_ , 1995 ] .
however in our case we apply forcing simultaneously to all the cells , in order to mimic situations such as application in vertically vibrated granular media[umbanhowar _ et al .
_ , 1996 ] .
in addition to the transition from hexagons to rhombs by the influence of external periodic force , there are also other possible effects due to it . to realize them
, we consider a different set of parametric @xmath184 @xmath185 @xmath10 , @xmath186 @xmath187 @xmath56 @xmath188 @xmath189 @xmath190 @xmath191 @xmath192 @xmath193 with @xmath174 and @xmath194 .
( a ) @xmath24 , ( b ) @xmath195 , ( c ) @xmath196 and ( d ) @xmath197 . ] choice @xmath198 @xmath56 @xmath188 @xmath189 @xmath191 @xmath199 @xmath193 with @xmath200 , @xmath201 and @xmath202 . for this choice
the system shows hexagonal patterns with defects including domains of small roll structures ( fig . [ fig8](a ) ) .
now when the external periodic force is included a transition in the pattern from hexagonal structure to rolls starts appearing . by fixing the frequency of the external periodic force again at @xmath173
, we observed the actual transition from hexagons to rolls as we increase the forcing amplitude ( @xmath178 ) .
[ fig8](b)-[fig8](d ) show the gray scale plots for @xmath195 , @xmath203 and @xmath204 , respectively .
obviously the transition is due to the existence of small roll structures in the pattern for @xmath24 which nucleates the formation of rolls in the presence of forcing . in the above
, we have shown that the inclusion of the external periodic force can make a transition from one stationary pattern to another stationary pattern like the transition from hexagons to rolls .
besides these , are there any time varying patterns ? as mentioned above , patterns such as localized and breathing oscillations have considerable and @xmath205 @xmath184 @xmath185 @xmath10 , @xmath186 @xmath187 @xmath56 @xmath188 @xmath189 @xmath190 @xmath191 @xmath192 @xmath193 with @xmath206 and @xmath194 for various time units starting from @xmath207 . ] physical interest . in this regard , we considered the parameters @xmath208 @xmath56 @xmath188 @xmath189 @xmath209 @xmath191 @xmath199 @xmath193 with @xmath129 and @xmath202 such that a regular hexagonal pattern is observed in the _ absence _ of external periodic force . from numerical simulations , we observed that a space - time periodic oscillatory pattern ( breathing motion ) sets in for a range of low values of @xmath178 .
[ fig9 ] shows the typical snapshots of the oscillating pattern at various instants for the specific choice @xmath210 .
we have integrated over 10000 time units and the figure corresponds to the region @xmath211 .
typically we find that the breathing pattern repeats itself approximately after a period @xmath212 in the range of our integration .
one may conclude that the emergence of such breathing oscillations is due to the competition between the turing and hopf modes in the presence of external periodic force .
next we move on to a study of the spatio - temporal chaotic dynamics of the array of coupled mlc circuits when individual cells are driven by external periodic force .
the motivation is that over a large domain of ( @xmath178 , @xmath182 ) values the individual mlc circuits typically exhibits various bifurcations and transition to chaotic motion ( see appendix a ) .
so one would like to know how the coupled array behaves collectively in such a situation , for fixed values of the parameters . for this purpose
, we set the parameters at @xmath213 @xmath214 . in our numerical simulations , we have mainly considered the one dimensional array specified by eq .
( 1 ) and assumed periodic boundary conditions .
the choice of periodic boundary conditions here makes the calculation of lyapunov exponents easier so as to understand the spatio - temporal chaotic dynamics of ( 1 ) in a better way .
numerical simulations were performed by considering random initial conditions using fourth order runge - kutta method for seven choices of @xmath178 values .
the coupling coefficient in eq .
( 1 ) was chosen as @xmath150 . out of these , the first three lead to period-@xmath215 , period-@xmath216 , period-@xmath217 oscillations , respectively and the remaining choices oscillators for the external periodic forcing strength ( a ) @xmath210 , ( b ) @xmath218 , ( c ) @xmath219 , ( d ) @xmath220 , ( e ) @xmath221 , ( f ) @xmath222 and ( g ) @xmath195 ] correspond to chaotic dynamics of the single mlc circuit .
[ fig10](a)-[fig10](g ) show the space - time plots for @xmath210 , @xmath218 , @xmath219 , @xmath220 , @xmath221 , @xmath222 and @xmath195 , respectively . from the figs .
[ fig10](a)-[fig10](c ) , it can be observed that for @xmath210 , @xmath223 and @xmath224 the mlc array also exhibits regular periodic behaviour with periods @xmath215 , @xmath216 and @xmath217 respectively in time alone as in the case of the single mlc circuit .
however , for @xmath220 and @xmath225(figs . [ fig10](d ) and [ fig10](e ) ) one obtains _
space - time periodic _ oscillations eventhough each of the individual uncoupled mlc circuits for the same parameters exhibits chaotic dynamics .
we may say that a kind of controlling of chaos occurs due to the coupling , though the coupling strength is large here . from the above analysis
it can be seen that the macroscopic system shows regular behaviour in spite of the fact that the microscopic subsystems oscillate chaotically .
finally for @xmath195 the coupled system shows spatio - temporal chaotic dynamics ( fig . [ fig10](g ) ) and this was confirmed by calculating the lyapunov exponents using the algorithm given by wolf _ et al_. [ wolf _ et al .
_ , 1985 ] .
for example , we calculated the lyapunov exponents for @xmath226 coupled oscillators ( for @xmath16 , it takes enormous computer time relatively which we could not afford at present ) and we find the highest three exponents have the values @xmath227 , @xmath228 , @xmath229 and the rest are negative ( see next subsection for further analysis ) .
we also note that for @xmath222 , the system is in a transitional state as seen in fig . [
fig10](f ) .
since the above study of spatio - temporal chaos involves a large number of coupled chaotic oscillators it is of great interest to analyse the size dependence of the dynamics of these systems . to start with , we consider the case of 10 coupled oscillators with periodic boundary conditions and numerically solve the system using fourth order runge - kutta method with the other parameters chosen as in sec .
[ sec5a ] .
the value of @xmath178 is chosen in the range @xmath230 .
we find that this set up shows different behaviour as compared to the 100 cell case .
actually the system gets synchronized to a chaotic orbit rather than showing periodic behaviour as in the case of 100 cells described above .
first we analyse the dynamics for @xmath220 by slowly increasing the system size from @xmath231 .
we noted that for the system size , @xmath232 it gets synchronized to a chaotic orbit as mentioned above .
we have also calculated the corresponding lyapunov spectrum and found that there is one positive maximal lyapunov exponent as long as @xmath232 .
for example , for @xmath233 , @xmath234 , with the rest cell for the synchronized state @xmath235 and ( b ) the periodic orbit in the @xmath236 cell for the controlled state @xmath237 ] of the exponents being negative . fig .
[ fig11](a ) shows the dynamics of the @xmath236 cell in the array and [ fig12](a ) depicts the space - time plot of the synchronized spatio - temporal chaos .
the system shows entirely different behaviour when we increase the system size to @xmath238 . as noted in the previous subsection , [ sec5a ] , there occurs kind of suppression of spatio - temporal chaos .
[ fig11](b ) shows the resultant periodic orbit in the @xmath236 cell of the array and fig .
[ fig12](b ) shows the space - time plot of the spatio - temporal periodic behaviour of the array .
the maximal lyapunov exponent is found to be negative in this case @xmath239 .
we observed the same phenomenon for @xmath221 also . and ( b ) @xmath238 . ]
next we consider the case of @xmath195 in which the coupled system in the previous subsection showed spatio - temporal chaos . from numerical simulations
, we again observed a synchronized motion for @xmath240 and the corresponding lyapunov spectrum shows one positive exponent only with all other exponents being negative . but for @xmath241 ,
the coupled system shows spatio - temporal chaos .
the lyapunov spectrum in this case ( for @xmath241 ) possesses multiple positive exponents .
for example , for @xmath238 , @xmath242 , @xmath243 , @xmath244 . from the above analysis
we see definite evidence that the spatio - temporal chaotic dynamics in coupled mlc circuits depends very much on the size of the system .
a similar analysis can be performed to investigate the nature of the spatio - temporal dynamics on the strength of the coupling coefficient @xmath7 .
preliminary analysis shows similar phenomenon at critical values of @xmath7 .
further work is in progress on these lines .
in this paper we have discussed various spatiotemporal behaviours of the coupled array of murali - lakshmanan - chua circuits in one and two dimensions . in the first part we have reported the propagation phenomenon of the travelling wave fronts and failure mechanism in the absence of external forcing .
we have shown that the propagation phenomenon is due to the loss of stability of the steady states via subcritical bifurcation coupled with the presence of necessary basin of attraction of the steady states .
the effect of weak coupling on the travelling wave phenomenon has also been studied . in the two dimensional array we studied the onset of turing instability leading to the spontaneous formation of hexagonal patterns .
introduction of external periodic force has been shown to lead to a partial propagation in the failure region of one dimensional array , while in the two dimensional array a transition from hexagons to rhombs take place .
we have also showed that a transition from hexagons to rolls can occur by the influence of external periodic force provided domains of small roll structures already present in the absence of external force .
space - time breathing oscillations as a result of the competition between hopf and turing modes in the presence of external periodic force have also been observed .
we have also studied the spatio - temporal chaotic dynamics and showed that the mlc array exhibits both periodic and chaotic dynamics as in the case of single oscillator . in the chaotic regime
the dynamics is also critically dependent on the system size . for a large array ,
the collective behaviour shows periodic oscillations both in space and time eventhough the individual subsystems oscillate chaotically as the coupling brings the macroscopic system into a regular motion . on the other hand , below a critical size of the system ,
it gets synchronized to a chaotic orbit .
the study of array of diffusively coupled driven nonlinear oscillators is of intrinsic interest as they represent many natural phenomena such as faraday instability , patterns in granular media and so on .
of particular interest will be to look for localized structures in these systems .
whether such excitations exists in the present arrays of oscillators is a question to be answered in the near future .
work is in progress along these lines .
p.m. acknowledges with gratitude the financial support of council of scientific and industrial research in the form of a senior research fellowship . the work of m.l . has been supported by the department of science and technology , government of india and national board for higher mathematics ( department of atomic energy ) in the form of research projects .
the murali - lakshmanan - chua circuit , fig . [ fig13](a ) , is the simplest second order dissipative nonautonomous circuit , consisting of chua s diode as the only nonlinear element[murali _ et al_. , 1994a ] .
this circuit contains a capacitor @xmath245 , an inductor @xmath246 , a linear resistor @xmath247 , an external periodic forcing @xmath248 and a chua s diode . by applying the kirchoff s laws to this circuit ,
the governing equations for the voltage @xmath249 across the capacitor @xmath250 and the current @xmath251 through the inductor @xmath252 are represented by the following set of two first order nonautonomous differential equations : the piecewise nature of the characteristic curve of chua s diode is obvious from eq .
( [ app : eq2])[murali _ et al_. , 1994a ] .
the slopes of left , middle and right segments of the characteristic curve are @xmath257 , @xmath258 and @xmath259 , respectively . @xmath260 and @xmath261 are the break points and @xmath262 corresponds to the dc offset in the chua s diode . rescaling eq .
( [ app : eq1 ] ) as @xmath263 , @xmath264 , @xmath265 , @xmath266 @xmath267 and @xmath268 and then redefining @xmath269 as @xmath270 the following set of normalized equations are obtained : @xmath271 with @xmath272 where @xmath273 , @xmath274 and @xmath275 .
obviously @xmath3 takes the form as in eq .
( [ eqn2 ] ) with @xmath276 , @xmath277 and @xmath278 .
the dynamics of eq .
( [ app : eq3 ] ) depends on the parameters @xmath279 , @xmath280 , @xmath10 , @xmath11 , @xmath12 , @xmath281 , @xmath182 and @xmath178 .
the rescaled parameters corresponding to the experimental observations reported in murali
_ et al_.[1994a ] correspond to @xmath282 , @xmath283 , @xmath284 , @xmath285 , @xmath286 , @xmath50 and @xmath173 . by varying @xmath178 one
can observe the familiar period doubling bifurcations leading to chaos and several periodic windows in the mlc circuit .
[ fig14](a ) shows the one parameter bifurcation diagram in the @xmath178-@xmath0 plane [ @xmath292 .
a summary of the bifurcations that occur in this case for different @xmath178 values is given in table [ table1 ] .
further it is of great interest to consider the parametric choice @xmath293 @xmath294 which corresponds to the function @xmath3 having the form as shown in fig . [
fig1](c ) .
this choice of parameters provides the possibility of bistability nature in the asymmetric case in the absence of periodic forcing . in this case one
can easily observe from numerical simulations that the mlc circuit admits only limit cycles for @xmath295 .
the bifurcation diagram in the @xmath178-@xmath0 plane is depicted in fig . [ fig14](b ) .
the eigenvalues of the above matrix ( [ jacob1 ] ) as a function of @xmath7 can be obtained by solving the corresponding characteristic equation .
we have verified by numerical evaluation that all the eigenvalues of ( [ jacob1 ] ) have negative real parts for @xmath296 . similarly all the eigenvalues corresponding to the steady state @xmath83 have negative real parts irrespective of the values of @xmath7 @xmath297 .
we have also verified that atleast one of the eigenvalues corresponding to the state @xmath81 has positive real part for @xmath296 .
thus @xmath88 are stable , while @xmath81 is unstable for all values of @xmath296 . similarly one can obtain the stability properties of the set of steady states @xmath298 , by substituting the appropriate expressions from ( [ eqnb2 ] ) into ( [ jacob ] )
. numerical diagonalization of ( [ jacob ] ) for these steady states show that @xmath97 is unstable for all values of @xmath7 , while @xmath96 and @xmath98 are stable for @xmath299 and unstable for @xmath102 for the chosen initial conditions .
the numerical values of @xmath123 in the stable region are given in table [ table2 ] as a function of @xmath7 .
41 allesie , m. a. , bonke , f. i. m. & scopman , t. y. g. [ 1973 ] circus movement in rabit atrial muscle as a mechanism in tachycardia , " _ circ .
* 33 * , 54 - 62 .
bak , p. , & chen , k. [ 1991 ] self - organized criticality " _ sci .
_ * 264 * , 46 .
balke , c. w. , lesh , m. d. , spear , j. f. , kadish , a. , levine , j. h. & moore , e. n. [ 1988 ] effects of cellular uncoupling on conduction in anisotropic canine ventricular myocardium , " _ circ .
res . _ * 63 * , 879 - 892 .
chua , l. o. and yang , l. [ 1988 ] cellular neural networks : theory , " _ ieee trans .
* 35 * , 1257 - 1272 .
chua , l. o. ( ed . ) [ 1995 ] special issue on nonlinear waves , patterns and spatio - temporal chaos in dynamic arrays , " _ ieee trans .
syst . _ * 42 * , 557 - 823 .
cole , w. c. , picolne , j. b. & sperelakis , n. [ 1988 ] gap junction uncoupling and discontinuous propagation in the heart , " _ biophys .
j. _ * 53 * , 809 - 818 .
dolnik , m. & marek , m. [ 1988 ] extincion of oscillations in forced and coupled reaction cells , " _ j. phys .
_ * 92 * , 2452 - 2455 .
feingold , m. , gonzlez , d. l. , piro , o. & viturro , h. [ 1988 ]
phase locking , period doubling and chaotic phenomena in externally driven excitable systems , " _ phys .
a _ * 37 * , 4060 - 4063 .
fineberg , j [ 1996 ] physics of jumping sandbox , " _ nature _ * 382 * , 763 - 764 guckenheimer , j. and holmes , p. [ 1983
] _ nonlinear oscillations , dynamical systems and bifurcation of vector fields _ ( springer verlag , new york ) hao bai - lin [ 1984 ] _ chaos _ ( world scientific , singapore ) keener , j. p. [ 1987 ] propagation and its failure in coupled systems of excitable cells " _ siam j. appl .
math _ * 47 * , 556 kocarev , l. & parlitz , u. [ 1996 ] synchronizing spatiotemporal chaos in coupled nonlinear oscillators , " _ phys .
lett . _ * 77 * , 2206 - 2209 .
kudrolli , a. , wolpert , m & gollub , j.p . , [ 1997 ]
cluster formation due to collisions in granular material , " _ phys .
lett . _ * 78 * , 1383 - 1386 .
lakshmanan , m. & murali , k. [ 1995 ] experimental chaos from non - autonomous electronic circuits , " _ phil .
a _ * 353 * , 33 - 46 .
lakshmanan , m. & murali , k. [ 1996 ] _ chaos in nonlinear oscillators : controlling and synchronization _ ( world scientific , singapore ) .
laplante , j. p. & erneux , t. [ 1992 ] propagation failure and multiple steady states in an array of diffusion coupled flow reactors , " _ physica a _ * 188 * , 89 - 98 .
leenaerts , d. m . w. [ 1997 ] wave propagation and its failure in piecewise linear nagumo equation " proc .
lioubasheski , o. , arbell , h. & fineberg , j. [ 1996 ] dissipative solitary states in driven surface waves , " _ phys
lett . _ * 76 * , 3959 - 3962 .
marn , j. l. & aubry , s. [ 1996 ] breathers in nonlinear lattices : numerical calculation from the anticontinuous limit , " _ nonlinearity _ * 9 * , 1501 - 1528 muuzuri , a. p. , prez - muuzuri , v. , prez - villar , v. & chua , l. o. [ 1993 ] spiral wave in 2d array of nonlinear circuits , " _ ieee transactions circuits and systems _ * 40 * , 872 - 877 .
muuzuri , a. p. , prez - muuzuri , v. , gmez - gesteria , m. , chua , l. o. & prez - villar , v. [ 1995 ]
spatiotemporal structures in discretely - coupled arrays of nonlinear circuits : a review , " _ int .
j. bifurcation and chaos _ * 5 * , 17 - 50 .
murali , k. , lakshmanan , m. & chua , l. o. [ 1994a ] the simplest dissipative non - autonomous chaotic circuit , " _ ieee trans .
sys . _ * i41 * , 462 - 463 murali , k. , lakshmanan , m. & chua , l. o. [ 1994b ] bifurcation and chaos it the simplest dissipative non - autonomous chaotic circuit , " _ int .
j. bifurcation and chaos _ * 4 * , 1511 - 1524 murali , k. , lakshmanan , m. & chua , l. o. [ 1995 ] controlling and synchronization of chaos in the simplest dissipative nonautonomous circuit " _ int .
j. bifurcation and chaos _ * 5 * , 6498 - 6510 murray , j. d. [ 1989 ] _ mathematical biology _
( springer - verlag , newyork ) pantaloni , j. & cerisier , p. [ 1983
] structure defects in bnaed - marangoni convection , " _ cellular structures and instabilities , leccture notes in physics _
* 210 * , eds .
j. e. & zaleski , s. ( springer - verlag , berlin ) prez - muuzuri , v. , gmez - gesteria , m. , prez - villar , v. & chua , l. o. [ 1993 ] travelling wave propagation in one dimensional fluctuating medium , " _ int .
j. bifurcation and chaos _ * 3 * , 211 - 215 .
prez - muuzuri , v. , alonso , m. , chua , l. o. & prez - villar , v. [ 1994 ] resonance patterns in 1d arrays of coupled nonlinear excitable systems , " _ int .
j. bifurcation and chaos _ * 4 * , 1631 - 1638 .
prez - muuzuri v. , gmez - gesteria , m. , muuzuri , a. p. , chua , l. o. & prez - villar [ 1995 ] sidewall forcing of hexagonal turing patterns : rhombic patterns , " _ physica d _ * 82 * , 195 - 204 perraud , j. j. , de wit , a. , dulos , e. , de kepper , p. , dewel , g. & borkmans , p. [ 1993
] one - dimensional spirals : a novel asynchronous chemical wave source " , _ phys .
* 71 * , 1272 - 1275 pismen , l. [ 1987 ] bifurcations of quasi - periodic and nonstationary pattern under external forcing " , _ phys .
lett . _ * 59 * , 2740 scott , a. c. [ 1975 ] the electrophysics of nerve fiber , " _ rev .
_ * 47 * , 487 - 533 .
sobrino , t. , alonso , m. , prez - muuzuri , v. & prez - villar , v. [ 1993 ] reaction - diffusion process in a one - dimensional array of active nonlinear circuits , " _ eur .
j. phys . _
* 14 * , 74 - 79 .
tsimring , l. s. [ 1996 ] dynamics of penta - hepta defects in hexagonal patterns " , _ physica _ * d89 * , 368 - 380 .
turing , a. m. [ 1952 ] the chemical basis of morphogenesis , " _ phil .
. london _ * b237 * , 37 - 72 .
umbanhowar , p. b. , melo , f. & swinney , h. l. [ 1996 ] localized excitations in a vertically vibrated granular layers " _ nature ( london ) _ * 382 * , 793 - 796 walgraef , d. [ 1988 ] external forcing of spatio - temporal pattern " _ europhys .
lett . _ * 7 * , 485 walgraef , d. [ 1996 ] _ spatio - temporal pattern formation _
( springer , new york ) pp .
127 - 167 watanabe , s. , strogatz , s. h. , van der zant , h. s. j. & orlando , t. p. [ 1995
] whirling modes and parametric instablilities in the discrete sine - gordon equation : experimental tests in josephson rings , " _ phys .
* 74 * , 379 - 382 .
wolf , a. , swift , j. b , swinney , h. l. , and vastano , j. a. [ 1985 ] determining lyapunov exponents from a time series " _ physica _ * 16d * 285 - 317 . | the circuit recently proposed by murali , lakshmanan and chua ( mlc ) is one of the simplest non - autonomous nonlinear electronic circuits which shows a variety of dynamical phenomena including various bifurcations , chaos and so on . in this paper
we study the spatio - temporal dynamics in one and two dimensional arrays of coupled mlc circuits both in the absence as well as in the presence of external periodic force . in the absence of any external force ,
the propagation phenomena of travelling wave front and its failure have been observed from numerical simulations .
we have shown that the propagation of travelling wave front is due to the loss of stability of the steady states ( stationary front ) via subcritical bifurcation coupled with the presence of necessary basin of attraction of the steady states .
we also study the effect of weak coupling on the propagation phenomenon in one dimensional array which results in the blocking of wave front due to the existence of a stationary front .
further we have observed the spontaneous formation of hexagonal patterns ( with penta - hepta defects ) due to turing instability in the two dimensional array .
we show that a transition from hexagonal to rhombic structures occur by the influence of external periodic force .
we also show the transition from hexagons to rolls and hexagons to breathing ( space time periodic oscillations ) motion in the presence of external periodic force .
we further analyse the spatio - temporal chaotic dynamics of the coupled mlc circuits ( in one dimension ) under the influence of external periodic forcing . here
we note that the dynamics is critically dependent on the system size . above a threshold size
, a suppression of spatio - temporal chaos occurs , leading to a space - time regular ( periodic ) pattern eventhough the single mlc circuit itself shows a chaotic behaviour . below this critical size , however , a synchronization of spatiotemporal chaos is observed . |
preconditioning is a technique developed originally for the iterative solution of linear systems that aims at the acceleration of convergence of the iterations . in its simplest form ,
the system @xmath0 is multiplied by a matrix @xmath1 such that the spectral condition number of @xmath2 , the ratio of the largest to the smallest singular value thereof , is considerably smaller than that of @xmath3 , which generally leads to faster convergence . 0
we define a preconditioner of a matrix @xmath3 as a matrix @xmath1 such that their product @xmath2 has a smaller condition number than @xmath3 . instead of solving a linear system @xmath4 , one may solve a preconditioned system @xmath5 .
the small condition number of @xmath2 is expected to lead to fast convergence of iterations .
iterative methods for solving linear systems normally do not require @xmath3 and @xmath1 to be explicitly formed as matrices : it is sufficient that matrix - vector multiplications are implemented and performed via user - defined procedures .
the same is true for iterative methods that compute eigenvalues and eigenvectors of a very large matrix , as , e.g.,in @xcite , calculating one eigenvector of a 100-billion size matrix , or in @xcite .
a classical application area for preconditioned solvers is the discretized boundary value problems for elliptic partial differential operators ; see , e.g. , @xcite . with multigrid preconditioning , preconditioned solvers may achieve linear complexity on problems from this area ; see , e.g. , @xcite and references there for symmetric eigenvalue problems .
dyakonov seminal work , summarized in @xcite , proposes `` spectrally equivalent '' preconditioning for elliptic operator eigenvalue problems in order to guarantee convergence that does not deteriorate with the increasing dimension of the discretized problem .
owing to this , for large enough problems such preconditioners outperform direct solvers , which factorize the original sparse matrix @xmath3 .
inevitable matrix fill - ins , especially prominent in discretized differential problems in more than two spatial dimensions , destroy the matrix sparsity , resulting in computer memory overuse and non - optimal performance .
preconditioning has also long since been a key technique in _ ab initio _ calculations in material sciences ; see , e.g. , @xcite and references therein . in the last decade , preconditioning for graphs is attracting growing attention as a tool for achieving an optimal complexity for large data mining problems , e.g. , for graph bisection and image segmentation using graph laplacian and fiedler vectors since @xcite ; for recent work see , e.g. , @xcite .
preconditioned iterative methods for the original linear system @xmath6 are in many cases mathematically equivalent to standard iterative methods applied for the preconditioned system @xmath7 .
for example , the classical richardson iteration step applied to the preconditioned system becomes @xmath8 where @xmath9 is a suitably chosen scalar .
turning now to eigenvalue problems , let us consider the computation of an eigenvector of a real symmetric positive definite matrix @xmath3 corresponding to its smallest eigenvalue .
borrowing an argument from @xcite , suppose that the targeted eigenvalue @xmath10 , or a sufficiently good approximation thereof , is known .
then the corresponding eigenvector can be computed by solving a homogeneous linear system @xmath11 , or , equivalently , the system @xmath12 , where @xmath13 is an identity .
the richardson iteration step now becomes @xmath14 theoretically , the best preconditioners for @xmath6 and @xmath15 are , correspondingly , @xmath16 and @xmath17 , where @xmath18 denotes a pseudo - inverse , making both richardson iteration schemes , and , converge in a single step with @xmath19 . under the standard assumption @xmath20 , both in and ,
convergence theory is straightforward , e.g. , in terms of the spectral radius @xmath21 of @xmath22 .
sharp explicit convergence bounds , not relying on generic constants , can be derived in the form of inequalities that allow one to determine whether the convergence deteriorates with the increasing problem size by analyzing every term in the bound .
for some classes of eigenvalue problems , the efficiency of choosing @xmath20 has been demonstrated , both numerically and theoretically , in @xcite .
this choice allows the easy adaptation of a vast variety of preconditioners already developed for linear systems to the eigensolvers . in practice
, the theoretical value @xmath10 in the richardson iteration above has to be replaced with its approximation .
a standard choice for @xmath10 is a rayleigh quotient function @xmath23 , leading to @xmath24 it is well known that the rayleigh quotient @xmath25 gives a high quality ( quadratic ) approximation of the eigenvalue @xmath10 , if the sequence @xmath26 converges to the corresponding eigenvector .
thus , asymptotically as @xmath27 where @xmath28 , methods and are equivalent , and so may be their asymptotic convergence rate bounds .
however , asymptotic convergence rate bounds naturally contain generic constants , which are independent of @xmath29 , but may depend on the problem size . due to the changing value @xmath25 , a non - asymptotic theoretical convergence analysis is much more difficult , compared to the case for linear systems , even for the simplest methods , such as the richardson iteration .
dyakonov pioneering work from the eighties , summarized in ( * ? ? ?
* chapter 9 ) , includes the first non - asymptotic convergence bounds for preconditioned eigensolvers proving their linear convergence with a rate , which can be bounded above independently of the problem size .
just a few of the known bounds are sharp .
one of them is proved for the simplest preconditioned eigensolver with a fixed step size in a series of papers by neymeyr over a decade ago ; see @xcite and references therein .
the original proof has been greatly simplified and shortened in @xcite by using a gradient flow integration approach .
in this paper we present a new self - contained proof of a sharp convergence rate bound from @xcite for the preconditioned eigensolver , theorem [ t.1 ] . following the geometrical approach of @xcite , we reformulate the problem of finding the convergence bound for as a constrained optimization problem for the rayleigh quotient .
the main novelty of the proof is that here we use inequality constraints , which brings to the scene the karush - kuhn - tucker ( kkt ) theory ; see , e.g. @xcite .
kkt conditions allow us to reduce our convergence analysis to the simplest scenario in two dimensions , which is the key step in the proof .
we have also found several simplifications in the two dimensional convergence analysis , compared to that of @xcite .
we believe that the new proof will greatly enhance the understanding of the convergence behavior of increasingly popular preconditioned eigensolvers , whose application area is quickly expanding : see , e.g. , @xcite .
we consider a real generalized eigenvalue problem @xmath30 with real symmetric positive definite matrices @xmath3 and @xmath31 . the objective is to approximate iteratively the smallest eigenvalue @xmath32 by minimizing the rayleigh quotient @xmath33 .
a direct formulation of the convergence analysis with respect to this form of the eigenvalue problem has some disadvantages . instead , the inverted form @xmath34 with @xmath35 results in more compact representation of the problem and the proof ( many inverses like @xmath36 and @xmath37 can be avoided ) , cf . @xcite . for this inverted form
the objective is to approximate the largest eigenvalue @xmath38 of @xmath34 by maximizing the rayleigh quotient @xmath39 .
we denote the eigenvalues by @xmath40 , which can have arbitrary multiplicity .
corresponding eigenspaces are denoted by @xmath41 .
the increase of @xmath42 can be achieved by correcting the current iterate @xmath43 along the preconditioned gradient of the rayleigh quotient , i.e. @xmath44 see @xcite and references therein . if @xmath45 , then @xmath46 and method turns into with @xmath19 , discussed in the introduction . in all our prior work on preconditioned eigensolvers for symmetric eigenvalue problems , including @xcite , we have always assumed that the preconditioner @xmath1 is a symmetric and positive definite matrix , typically satisfying conditions @xmath47 or equivalent , up to the scaling of @xmath1 .
recently , the authors of @xcite have noticed and demonstrated that @xmath1 does not have to be symmetric positive definite , and a less restrictive assumption @xmath48 can be used instead , where @xmath49 denotes the matrix largest singular value , and @xmath50 is the symmetric positive definite square root of @xmath3 .
it is verified in @xcite that and are equivalent if @xmath1 is symmetric and positive definite . in what follows
, we give a complete and concise proof of the following convergence rate bound , first proved in @xcite , [ t.1 ] if @xmath51 and @xmath1 satisfies , then for @xmath52 given by it holds that either @xmath53 or @xmath54 the first step , lemma [ l.1 ] , of the proof of theorem [ t.1 ] is the same as that in @xcite , where we characterize a set of possible next step iterates @xmath52 in varying the preconditioner @xmath1 constrained by assumption , aiming at eliminating the preconditioner @xmath1 from consideration .
the only difference is that in @xcite we start with changing an original coordinate basis to an @xmath3-orthogonal basis , which transforms @xmath3 into the identity @xmath13 , resulting in a one - line proof of lemma [ l.1 ] . here
, we choose to present a detailed proof of lemma [ l.1 ] , for a general @xmath3 , demonstrating that the transformation of @xmath3 into the identity @xmath13 , made after lemma [ l.1 ] , is well justified .
[ l.1 ] let us denote @xmath55 and @xmath56 and define a closed ball @xmath57 centered at @xmath58 .
let @xmath1 satisfy , then for @xmath52 given by it holds that @xmath59 left - multiplying by @xmath60 gives @xmath61 or , in our new notation , @xmath62 resulting @xmath63 . since @xmath64 by , we get @xmath65 the second step of the proof is traditional reducing the generalized symmetric eigenvalue problem @xmath34 to the standard eigenvalue problem for the symmetric positive definite matrix @xmath66 by making the change of variables as hinted by lemma [ l.1 ] .
we use the standard inner product in @xmath67 variables , i.e. @xmath68 and the corresponding vector norm @xmath69 , so , e.g.,@xmath70 we later use @xmath71 and @xmath72 -based scalar products and norms defined as follows , e.g. , @xmath73 for brevity we drop the subscript @xmath67 in the rest of the paper . in the following @xmath1
refers to @xmath74 , @xmath31 refers to @xmath66 , @xmath43 refers to @xmath75 , and so on , cf .
lemma [ l.1 ] .
furthermore , @xmath76 , and method is @xmath77 .
the new form of condition is @xmath78 .
this means that @xmath79 approximates the identity matrix with respect to the notation used in lemma [ l.1 ] .
the closed ball has the form @xmath80 with the radius @xmath81 centered at @xmath82 . since @xmath83 and @xmath84
, we can estimate @xmath85 by using a minimizer of @xmath86 in @xmath87 ( i.e. by considering the worst case ) .
we observe that , effectively , we set @xmath88 without loss of generality .
[ s3 ] the main idea of the geometrical approach of @xcite , which we also employ in this paper , is that the convergence rate of iterations is slowest , in terms of the rayleigh quotient , if @xmath43 is a linear combination of two eigenvectors , which makes the further convergence analysis trivial .
a new proof of this fact actually occupies a major part of our paper . in order to illustrate how such a dramatic reduction in dimension becomes possible , in this section we apply our technique to a simplified case @xmath90 corresponding to @xmath91 .
it is not difficult to see that under this assumption turns into one iteration @xmath92 of the power method , 0 our proof of theorem [ t.1 ] in the next section uses powerful tools , which are uncommon in numerical linear algebra and thus may catch an unprepared reader unguarded .
the role of this section is to serve as a gentle introduction to the main proof .
here we assume @xmath93 so that @xmath91 in order to simplify the analysis , then relation turns into the power method , @xmath92 , and bound holds with @xmath91 and thus @xmath94 .
let us make a historic note that exactly this result has apparently first appeared in @xcite .
the left - hand side of bound is monotone in @xmath95 .
one way to find out at which @xmath43 the behavior of is the worst is to minimize @xmath96 for all @xmath43 that satisfy @xmath97 for some fixed @xmath98 . slightly abusing the notation in the proof
, we keep denoting by @xmath43 both the initial approximation in and the vector in the minimization problem .
we notice that @xmath97 is equivalent to @xmath99 . therefore , at a stationary point we have , using lagrangian multipliers , that @xmath100 where @xmath101 is some constant .
this yields @xmath102 which can be rewritten as @xmath103 where @xmath104 .
since @xmath105 implies @xmath106 , we obtain @xmath107 which shows that @xmath108 .
thus , equation can be viewed as a polynomial equation @xmath109 , where @xmath110 is a third degree polynomial with positive first and last coefficients , specifically @xmath111 and @xmath112 , correspondingly .
inserting @xmath113 , where @xmath114 are the projections of @xmath43 onto the eigenspaces @xmath115 , leads to @xmath116 .
since the eigenspaces are orthogonal to each other , the products @xmath117 must be zero for each @xmath118 .
owing to the positiveness of the first and last coefficients , the polynomial @xmath119 must have a non - positive root , and thus at most two positive roots , i.e. @xmath120 can be zero for some two indexes @xmath121 and @xmath122 at most , allowing the only possibly nonzero @xmath123 and @xmath124 from all projections @xmath114 .
we conclude that @xmath43 is a linear combination of at most two normalized eigenvectors @xmath125 and @xmath126 , corresponding to distinct eigenvalues @xmath127 and @xmath128 of the matrix @xmath31 . we assume without loss of generality that @xmath129 , then @xmath130 similarly , since @xmath131 , we obtain @xmath132 let @xmath133 , then @xmath134 implies @xmath135 . by using monotonicity of the ratio of the quotients in @xmath127 and @xmath128 and the fact that the vector @xmath43 here corresponds to the worst - case scenario , i.e. minimizing @xmath95 over all @xmath43 with the fixed value @xmath97 , we obtain with @xmath91 .
since is an equality , we also prove that the upper bound in with @xmath91 is sharp , turning into an equality if the initial approximation in satisfies @xmath136 . in the next section ,
we apply the described dimensionality reduction technique to the general case .
we formulate the conditions that `` the worst case '' @xmath43 must satisfy , which yield the generalization of equation , and rewrite this equation as a cubic equation @xmath109 .
we show that the first and last coefficients of this equation are positive , which , as we have just seen , implies that @xmath43 is a linear combination of two eigenvectors .
a simple two - dimensional analysis completes the proof of theorem [ t.1 ] .
[ s4 ] next the proof of theorem [ t.1 ] is given : let us denote @xmath138 and define @xmath80 , a closed ball with the radius @xmath81 centered at @xmath82 .
on the one hand , it holds that @xmath139 for any vector @xmath140 since @xmath43 is not an eigenvector and @xmath141 . indeed , taking into account @xmath142 ,
we have @xmath143 so that @xmath144 . on the other hand , @xmath145 , since @xmath146 , see lemma [ l.1 ] .
this proves @xmath147 and , thus , the left inequality in , provided that @xmath148 in the previous proof with @xmath91 , the ball @xmath87 shrinks to a single point @xmath82 and the only choice of @xmath149 is possible .
the present case @xmath150 is significantly more difficult for the worst - case scenario analysis , involving a minimization problem with two variables , @xmath43 and @xmath151 . in our previous work ,
see @xcite and references therein , we first vary @xmath140 intending to minimize @xmath152 for a given @xmath43 , and then vary @xmath43 fixing @xmath97 .
the first minimization problem defines @xmath151 as an implicit function of @xmath43 , and then lagrangian multipliers are used , as in section [ s3 ] , to analyze the second minimization problem , in @xmath43 .
it turns out that the proof is much simpler if we vary both @xmath43 and @xmath151 at the same time and attack the required two - parameter minimization problem in @xmath43 and @xmath151 directly by using the kkt arguments as provided below .
[ t.reduc ] for @xmath153 and a fixed value @xmath154 that is not an eigenvalue of @xmath31 , let a pair of vectors @xmath155 denote a solution of the following constrained minimization problem : @xmath156 if @xmath157 is not an eigenvector of @xmath31 , then both @xmath157 and @xmath158 belong to a two - dimensional invariant subspace of @xmath31 corresponding to two distinct eigenvalues , and @xmath159 where @xmath160 denotes an angle between two vectors defined by @xmath161 .
we consider the equivalent problem @xmath162 we first notice that the assumption @xmath163 implies @xmath164 because of the first constraint and @xmath165 .
thus , @xmath152 is correctly defined .
next , let us temporarily consider a stricter constraint @xmath166 , instead of @xmath163 .
combined with the other constraints , this results in minimization of the smooth function @xmath167 on a compact set , so there exists a solution @xmath155 .
finally , let us remove the artificial constraint @xmath166 and notice that any nonzero multiple of @xmath155 is also a solution .
thus we can consider the karush - kuhn - tucker ( kkt ) conditions , e.g. , ( * ? ? ?
* theorem 9.1.1 ) , @xcite , in any neighborhood of @xmath155 , which does not include the origin .
next we show that the gradients of @xmath168 and @xmath169 are linearly independent . for the gradient of @xmath169
, it holds that @xmath170 with @xmath141 , since @xmath43 is not an eigenvector of @xmath31 , and it holds that @xmath171 , since @xmath169 does not depend on @xmath151 . assuming the linear dependence of the gradients of @xmath168 and @xmath169 implies that @xmath172 , so that @xmath173 and @xmath149 . by using @xmath149
, it holds that @xmath174 while @xmath170 , i.e. ( using again the assumed linear dependence ) the vector @xmath175 is an eigenvector of @xmath176 , and , hence @xmath43 is an eigenvector of @xmath31 , contradicting the lemma assumption .
therefore , the gradients of @xmath168 and @xmath169 are linearly independent .
all functions involved in our constrained minimization are smooth .
we conclude that the stationary point @xmath155 is regular , i.e. , the kkt conditions are valid .
the kkt stationarity condition states that there exist constants @xmath101 and @xmath177 such that @xmath178 at the critical point @xmath155 .
the independent variables @xmath179 no longer appear , so to simplify the notation , in the rest of the proof we drop the superscript @xmath180 and substitute @xmath179 for @xmath155 .
we separately write the partial derivatives with respect to @xmath43 , @xmath181 and with respect to @xmath151 , @xmath182 the kkt complementary slackness condition @xmath183 must be satisfied , implying @xmath184 if @xmath151 is an eigenvector then @xmath185 in condition , leading to @xmath149 , i.e. vector @xmath43 is also an eigenvector of @xmath31 , thus we are done .
now we consider a nontrivial case , where neither @xmath43 nor @xmath151 is an eigenvector .
condition then implies @xmath186 , so identity holds unconditionally , condition turns into @xmath187 and taking the inner product of with @xmath151 gives @xmath188 taking the inner products of both sides of with @xmath189 results in @xmath190 therein we use and . denoting @xmath191 , we rewrite as @xmath192 taking the inner products of both sides of with @xmath193 yields @xmath194 where the orthogonality @xmath195 has been used again .
therefore , we obtain @xmath196 , which implies @xmath197 . substituting @xmath175 and multiplying through by @xmath31 in results in @xmath198 multiplying through by @xmath199 and substituting @xmath200 , which follows from ,
we obtain @xmath201 , where @xmath202 is a third degree polynomial with @xmath203 and @xmath204 , which can not have more than two positive roots . 0
inserting the expansion @xmath205 with the projections @xmath114 of @xmath151 on the eigenspaces @xmath115 leads to @xmath206 for every @xmath207 , cf .
section 3 . since @xmath208 and @xmath209 , the polynomial @xmath119 must have a non - positive root , and thus at most two positive roots .
hence @xmath210 can be nonzero for at most two @xmath207 ( since @xmath211 ) and thus , @xmath151 is a linear combination of two normalized eigenvectors @xmath125 and @xmath126 corresponding to two distinct eigenvalues ( cf .
section 3 ) , i.e. @xmath212 . since @xmath213 , by so is @xmath43 .
furthermore , the orthogonality @xmath195 from shows that @xmath214 and @xmath215 .
this leads to @xmath216 , since the angles between vectors have the range @xmath217 $ ] ( due to @xmath218 ) .
similarly , @xmath219 together with @xmath220 implies @xmath221 .
then we have @xmath222 by using .
we now derive bound in a two - dimensional @xmath31-invariant subspace .
[ t.2d ] let @xmath157 and @xmath158 belong to a two - dimensional invariant subspace of @xmath31 corresponding to the eigenvalues @xmath223 and satisfy , where @xmath157 is not an eigenvector .
it holds that @xmath224 in this proof we drop the superscript @xmath180 upon @xmath43 and @xmath151 .
the vectors @xmath43 , @xmath151 and @xmath82 can be represented by the coefficient vectors @xmath225 with respect to an orthonormal basis @xmath226 , where @xmath125 and @xmath126 are eigenvectors associated with @xmath127 and @xmath128 .
evidently , it holds that @xmath227 , @xmath228 , @xmath229 by using the orthonormal basis .
therefore , @xmath230 , and similarly @xmath231 .
this allows us to rewrite in the form @xmath232 . using the geometric property of the cross products @xmath233 we have @xmath234 which yields @xmath235 further
, we use the equalities @xmath236 which can be derived in a similar way to section [ s.cc ] .
then @xmath237 , so that @xmath238 since @xmath239 , we have @xmath240 which proves by using @xmath241 . the proof of theorem [ t.1 ] is completed by deriving convergence bound from its two - dimensional version .
we restate the assumption @xmath242 . according to lemma [ t.reduc ] , there exists a minimizer @xmath151 with @xmath243 , which satisfies with @xmath244 , and the interval @xmath245 is a subset of @xmath246 . using monotonicity arguments , , and the same arguments as section [ s.cc ] , we obtain @xmath247 & = \frac{\mu_k-\mu(y)}{\mu(y)-\mu_l}\ ; \frac{\mu(x)-\mu_l}{\mu_k-\mu(x ) } \le\left(\gamma+(1-\gamma)\frac{\mu_l}{\mu_k}\right)^2 \le\left(\gamma+(1-\gamma)\frac{\mu_{i+1}}{\mu_i}\right)^2 . \end{split}\ ] ] this proves since @xmath248 .
we have presented a succinct proof of the standard sharp convergence rate bound for the simplest fixed step size preconditioned eigensolver . the key argument of
the new proof is the characterization of the case of poorest convergence as a constrained optimization problem for the rayleigh quotient . employing the karush - kuhn - tucker conditions and some elementary matrix algebra , we dramatically simplify the convergence analysis by reducing it to a subspace spanned by two eigenvectors .
we expect the analytical framework developed in this paper to be a valuable tool in the convergence analysis of a variety of preconditioned eigensolvers ; see , e.g. , the analysis of the preconditioned steepest descent method in @xcite .
* i*. an alternative estimate for the left - hand side of , which is sharp with respect to all variables , can be derived as follows : with @xmath249 and @xmath250 , a new representation of @xmath251 is given by @xmath252 this results in a quadratic equation for @xmath253 with the roots @xmath254 since @xmath255 , a strictly sharp bound for the estimate in is given by @xmath256 .
we note that in the limit case @xmath257 it holds that @xmath258 , and @xmath259 turns into @xmath260 .
this coincides with the known bound in . + * ii*. the bound in contains a convex combination of @xmath111 and @xmath261 .
interestingly , this bound can also be derived by using a convex function as follows : without loss of generality , we assume that @xmath43 has a positive @xmath125 coordinate . then @xmath262 is an acute angle . since @xmath211 , @xmath263 and @xmath264 are also acute angles .
the equality together with @xmath165 shows further @xmath265 .
since @xmath264 and @xmath266 are acute angles , the vectors @xmath125 and @xmath151 are located in a half - plane whose boundary line is orthogonal to @xmath82 .
a simple case differentiation shows that @xmath267 is either equal to @xmath268 or equal to @xmath269 .
further , we use the equalities @xmath270 which can be derived in a similar way to section [ s.cc ] .
the last equality proves @xmath271 , since the tangent is an increasing function for acute angles , and @xmath272 .
this leads to @xmath273 , since @xmath274 are all in the same quadrant . in summary
, it holds that @xmath275 i.e. , @xmath267 is a further acute angle . using these acute angles , we write equivalently as @xmath276 in order to prove , we use again , together with @xmath277 , @xmath278 and the first inequality in .
it holds that @xmath279=:f(\gamma).\ ] ] because of @xmath280,\ ] ] @xmath281 is a monotonically increasing function in @xmath282 $ ] .
the numerator of @xmath283 is also a monotonically increasing function and its denominator is monotonically decreasing in @xmath284 $ ] .
these two functions are positive so that @xmath283 is also a monotonically increasing function .
thus @xmath281 is a convex function in @xmath282 $ ] , and @xmath285 which proves and hence .
, nonsymmetric preconditioning for conjugate gradient and steepest descent methods , _ procedia computer science _ , 51 ( 2015 ) ,
276285 . . a preliminary version available at arxiv:1212.6680 [ cs.na ] , 2012
http://arxiv.org/abs/1212.6680 , modern preconditioned eigensolvers for spectral image segmentation and graph bisection , workshop on clustering large data sets third ieee international conference on data mining ( icdm 2003 ) , 2003 .
http://math.ucdenver.edu/~aknyazev/research/conf/icdm03.pdf knn2003 a. v. knyazev and k. neymeyr , a geometric theory for preconditioned inverse iteration .
iii : a short and sharp convergence estimate for generalized eigenvalue problems , _ linear algebra appl .
_ , 358 ( 2003 ) , pp .
95114 .
kn2003 a. v. knyazev and k. neymeyr , efficient solution of symmetric eigenvalue problems using multigrid preconditioners in the locally optimal block conjugate gradient method , _ electronic transactions on numerical analysis _ , 15 ( 2003 ) , pp .
/ vol.15.2003/pp38 - 55.dir / pp38 - 55.pdf , preconditioned eigensolvers for large - scale nonlinear hermitian eigenproblems with variational characterizations .
i. conjugate gradient methods , research report 14 - 08 - 26 , department of mathematics , temple university , august 2014 .
revised april 2015 . to appear in
_ mathematics of computation_. https://www.math.temple.edu/~szyld/reports/nlpcg.report.rev.pdf , preconditioned eigensolvers for large - scale nonlinear hermitian eigenproblems with variational characterizations .
interior eigenvalues , research report 15 - 04 - 10 , department of mathematics , temple university , april 2015 . to appear in _ siam journal on scientific computing_. http://arxiv.org/abs/1504.02811 , high - performance computing for exact numerical approaches to quantum many - body problems on the earth simulator , in proceedings of the 2006 acm / ieee conference on supercomputing ( sc 06 ) .
acm , new york , ny , usa , article 47 , 2006 . | preconditioned iterative methods for numerical solution of large matrix eigenvalue problems are increasingly gaining importance in various application areas , ranging from material sciences to data mining .
some of them , e.g. , those using multilevel preconditioning for elliptic differential operators or graph laplacian eigenvalue problems , exhibit almost optimal complexity in practice , i.e. , their computational costs to calculate a fixed number of eigenvalues and eigenvectors grow linearly with the matrix problem size
. theoretical justification of their optimality requires convergence rate bounds that do not deteriorate with the increase of the problem size .
such bounds were pioneered by e. dyakonov over three decades ago , but to date only a handful have been derived , mostly for symmetric eigenvalue problems .
just a few of known bounds are sharp .
one of them is proved in [ ] for the simplest preconditioned eigensolver with a fixed step size .
the original proof has been greatly simplified and shortened in [ ] by using a gradient flow integration approach . in the present work ,
we give an even more succinct proof , using novel ideas based on karush - kuhn - tucker theory and nonlinear programming .
symmetric ; preconditioner ; eigenvalue ; eigenvector ; rayleigh quotient ; gradient ; iterative method ; karush kuhn tucker theory .
65f15 65k10 65n25 _ dedicated to the memory of evgenii g. dyakonov , moscow , russia , 19352006 . _ |
the korteweg - de vries ( kdv ) equation @xmath9 is globally well - posed in @xmath10 for @xmath11 ( see kenig - ponce - vega @xcite ) .
it possesses soliton solutions @xmath12 , where @xmath13 and @xmath14 ( so that @xmath15 ) .
benjamin @xcite , bona @xcite , and bona - souganidis - strauss @xcite showed that these solitons are orbitally stable under perturbations of the initial data .
we consider here the behavior of these solitons under structural perturbations , i.e. hamiltonian perturbations of the equation itself .
dejak - sigal @xcite , motivated by a model of shallow water wave propagation over a slowly - varying bottom , have considered the perturbed kdv ( pkdv ) @xmath16 where @xmath17 and @xmath2 .
they proved that the effects of this potential are small on the dynamically relevant time frame .
we consider instead @xmath1 , a slowly - varying _ but not small _ potential , .
see the comments below in
[ s : earlierwork ] . ] which allows for considerably richer dynamics . to state our main theorem
, we need the following definition : [ d : time - scale ] given @xmath18 , @xmath19 , @xmath20 , and @xmath21 , let @xmath22 , @xmath23 solve the system of odes @xmath24 with initial data @xmath25 and @xmath26 . let @xmath27 be the maximal time such that on @xmath28 , we have @xmath29 .
( @xmath27 could be @xmath30 . )
let @xmath31 .
[ t : main ] given @xmath18 , @xmath19 , @xmath20 , and @xmath21 , let @xmath27 be the time defined in def .
[ d : time - scale ] .
let @xmath32 and @xmath33 .
then for @xmath34 , there exist trajectories @xmath35 and @xmath36 , and positive constants @xmath37 and @xmath38 , such that the following holds . taking
@xmath39 the solution of with potential @xmath1 and initial data @xmath40 , let @xmath41 .
then @xmath42}^\infty h_x^1 } \lesssim h^{1/2}e^{cht } \,,\ ] ] @xmath43}^2h_x^1 } \lesssim h^{1/2 } e^{cht } \,,\ ] ] and @xmath44 moreover , @xmath45 , @xmath35 is of size @xmath46 and @xmath36 is of size @xmath47 , and gives leading - order in @xmath48 estimates for @xmath35 and @xmath36 that is , despite the differences in magnitudes , the estimates for @xmath35 and @xmath36 provided by are equally strong .
the strength of the local estimate , in comparison to the global estimate on the error @xmath49 , is that it involves integration in time over a ( long ) interval of length @xmath46 .
the estimate is _ on par _ , although slightly weaker than , the pointwise - in - time estimate @xmath50}^\infty l_x^2 } \leq he^{cht}$ ] .
the two estimates , are consistent ( but not equivalent to ) @xmath49 being of amplitude @xmath48 but effectively supported over an interval of size @xmath46 , which is suggested by numerical simulations .
the trajectory estimates state that we can predict the center of the soliton to within accuracy @xmath47 and the amplitude to within accuracy @xmath51 .
( this discussion does not include the @xmath52 loss that occurs when passing to the natural ehrenfest time scale @xmath3 . ) to define the hamiltonian structure associated with , let @xmath53 with @xmath54 we regard the function space @xmath55 as a symplectic manifold with symplectic form @xmath56 densely defined on the tangent space @xmath57 . then is the hamilton flow @xmath58 associated with the hamiltonian @xmath59 let @xmath60 denote the two - dimensional submanifold of solitons @xmath61 by direct computation , we compute the restricted symplectic form @xmath62 ( thus @xmath63 is a _
symplectic _ submanifold of @xmath64 ) and restricted hamiltonian @xmath65 , where @xmath66 the heuristic adopted in @xcite , essentially equivalent ( see @xcite ) to the `` effective lagrangian '' or `` collective coordinate method '' commonly applied in the physics literature , is the following : the equations of motion for @xmath67 , @xmath68 are approximately the hamilton flow of @xmath69 with respect to @xmath70 .
these equations are @xmath71 by taylor expansion , these equations are approximately @xmath72 note that the equations are the rescaled versions of these equations with the @xmath73 and @xmath74 error terms dropped .
the first of the orthogonality conditions in can be rewritten as @xmath75 and thus interpreted as symplectic orthogonality with respect to the @xmath67-direction on @xmath63 .
the other symplectic orthogonality condition @xmath76 is not defined for general @xmath77 functions @xmath49 since @xmath78 , where @xmath79 .
thus , we drop this condition , although it must be replaced with some other condition that projects sufficiently far away from the kernel ( @xmath80 ) of the hessian of the lyapunov functional .
we select @xmath81 ( i.e. , the second equation in ) since it is a hypothesis in the martel - merle local virial identity ( lemma [ l : mm ] ) . for the numerics , we restrict to time - independent potentials @xmath82 and use the rescaled frame @xmath83 , @xmath84 , @xmath85 , and @xmath86
. then @xmath87 solves the equation @xmath88 with initial data @xmath89 .
note that to examine the solution @xmath90 on the time interval @xmath91 , we should examine @xmath92 on the time interval @xmath93 .
as an example , we put @xmath94 and take @xmath95 , @xmath96 and @xmath97 .
then the width of the soliton is approximately the same width as the potential ( when @xmath98 ) , but note that the size of the potential is not small .
the results of numerical simulations for @xmath99 are depicted in the fig .
there , plots are given depicting the rescaled solution @xmath100 for each of these values of @xmath48 . in fig .
[ f:2 ] , we draw a comparison to the odes . in each of the numerical simulations , we record the center of the soliton as @xmath101 and the soliton scale as @xmath102 that is
, we fit the solution @xmath92 to @xmath103 .
let @xmath104 so that @xmath105 . to convert into the @xmath106 frame of reference , we plot @xmath107 versus @xmath108 in the top plot of fig .
[ f:2 ] together with @xmath109 solving . in the bottom frame ,
we plot @xmath107 versus @xmath110 together with @xmath111 solving .
we opted to only plot @xmath112 since the curves for @xmath113 were all rather close , producing a crowded figure .
theorem [ t : main ] predicts @xmath51 convergence in both frames of fig .
[ f:2 ] . the numerical solution to the equation
was produced using a ` matlab ` code based on the fourier spectral / etdrk4 scheme as presented in kassam - trefethen @xcite .
the odes were solved numerically using ` ode45 ` in ` matlab ` .
( see text ) for @xmath114 , @xmath95 , @xmath96 , on the time interval @xmath115 .
the three frames are , respectively , @xmath116 , @xmath112 , and @xmath117 .
[ f:1 ] , title="fig : " ] ( see text ) for @xmath114 , @xmath95 , @xmath96 , on the time interval @xmath115 .
the three frames are , respectively , @xmath116 , @xmath112 , and @xmath117 .
[ f:1 ] , title="fig : " ] ( see text ) for @xmath114 , @xmath95 , @xmath96 , on the time interval @xmath115 .
the three frames are , respectively , @xmath116 , @xmath112 , and @xmath117 .
[ f:1 ] , title="fig : " ] , the position was recorded as @xmath101 and the scale was recorded as @xmath118 ; that is , the solution @xmath100 was fitted to @xmath103 .
the top plot is @xmath107 versus @xmath119 for @xmath112 ( in blue ) compared to the value of @xmath109 obtained by solving the ode system ( in green ) .
the bottom plot is @xmath107 versus @xmath120 for @xmath112 ( in blue ) , compared to the value of @xmath111 obtained from the ode system ( in green ) .
[ f:2 ] , title="fig : " ] , the position was recorded as @xmath101 and the scale was recorded as @xmath118 ; that is , the solution @xmath100 was fitted to @xmath103 .
the top plot is @xmath107 versus @xmath119 for @xmath112 ( in blue ) compared to the value of @xmath109 obtained by solving the ode system ( in green ) .
the bottom plot is @xmath107 versus @xmath120 for @xmath112 ( in blue ) , compared to the value of @xmath111 obtained from the ode system ( in green ) .
[ f:2 ] , title="fig : " ] theorem 2 in dejak - sigal @xcite states ( roughly ) that for potential @xmath121 , the error @xmath122 can be achieved on the time - scale @xmath123 , and the equations of motion satisfy @xmath124 to reach the nontrivial dynamical time frame , one thus needs to take @xmath125 in their result . with this selection for @xmath126 ,
the @xmath73 errors in the odes can be removed as in our result with the effect of at least preserving the error estimate for @xmath127 in @xmath77 at the @xmath4 , rather than @xmath48 level .
but then the conclusion of their analysis is that the ( small and slowly varying ) potential has no significant effect on the dynamics .
we emphasize that in our case , we allow for @xmath128 and thus can see dramatic effects on the motion of the soliton . the paper dejak - sigal @xcite is modeled upon earlier work by frhlich - gustafson - jonsson - sigal @xcite for the nls equation , which controlled the error via the lyapunov functional employed in the orbital stability theory of weinstein @xcite . in @xcite , we improved @xcite by using the symplectic restriction interpretation as a guide in the analysis and introducing a correction term to the lyapunov estimate .
a correction term is not as easily applied to the study of since the leading order inhomogeneity in the equation for @xmath49 generates a `` nonlocal '' solution . to properly address the nonlocality of @xmath49 , we use both the global @xmath77 estimate as in @xcite , but also introduce the new local estimate , which is proved using the local virial identity of martel - merle @xcite .
we remark that our method does not use the integrable structure of the kdv equation , and we expect that our result will carry over to the perturbed @xmath5 subcritical gkdv-@xmath6 equation @xmath129 in the case @xmath8 , i.e. the second symplectic orthogonality condition @xmath130 ( where now @xmath131 and @xmath132 ) is well - defined for general @xmath77 functions @xmath49 . in this case , we are able to achieve stronger results by following the method of @xcite , and even treat double solitons
see @xcite .
the concurrent work by muoz @xcite considers the equation ( specializing to the case @xmath133 in his paper to facilitate comparison ) @xmath134 where @xmath135 , with @xmath136 increasing monotonically from @xmath137 to @xmath138 , and @xmath139 is constant , effectively corresponding to a moving frame of reference .
the equation is similar to our but not directly related to it through any known transformation .
his main theorem gives the existence of a solution @xmath140 which asymptotically matches the soliton @xmath141 as @xmath142 and matches the soliton @xmath143 as @xmath144 with error at most @xmath4 in @xmath145 .
here , @xmath146 is precisely given in terms of the solution to an algebraic equation ( see ( 4.17 ) in his paper ) .
he presents this problem as more of an obstacle scattering problem with a careful analysis of `` incoming '' and `` outgoing '' waves and thus his priorities are different from ours .
however , information from the `` interaction phase '' of his analysis can be extracted from the main body of his paper and compared with the results of our paper . in the course of his analysis
, he obtains effective dynamics ( here @xmath147 ) for an _ approximate solution _
@xmath148 he then shows that the approximate solution is comparable to a true solution in @xmath77 with accuracy @xmath149 ( same as in our result ) but only at the expense of a spatial shift for which he has the comparatively weak control of size @xmath46 . in our analysis
, we are able to achieve control of size @xmath47 on the positional parameter @xmath35 . at the technical level
, we are gaining an advantage by using the local virial estimate in the interaction phase analysis while muoz carries out a more direct energy estimate .
muoz does apply the local virial estimate in his `` post - interaction '' analysis to achieve a convergence statement as @xmath144 with a remarkably precise scale estimate .
it is convenient to work in both direct ( e.g. @xmath150 ) and `` pulled - back '' coordinates ( e.g @xmath151 ) .
our convention is that successive letters are used to define functions related in this way .
specifically , * @xmath152 and @xmath13 .
* @xmath153 and @xmath154 .
* @xmath155 * @xmath156 and @xmath157 .
in [ s : spectral ] , we deduce some needed spectral properties of the operator @xmath158 which are required to give the lower bound in the lyapunov functional method ( cor .
[ c : spec - bd ] ) . in
[ s : orth ] , we give the standard argument , via the implicit function theorem , that the parameters @xmath67 and @xmath68 can be adjusted so as to arrange that @xmath49 satisfies the orthogonality conditions ( lemma [ l : ift ] ) . in [ s : decomposition ] , we decompose the forcing term in the linearized equation into symplectically orthogonal and symplectically parallel components . in [ s : parameters ] , the orthogonality conditions are applied to obtain the equations for the parameters ( lemma [ l : param ] ) .
these equations include error terms expressed in terms of the local - in - space norm @xmath159 . in
[ s : locvir ] , an estimate on @xmath160 is obtained by the martel - merle local virial identity ( lemma [ l : locvir ] ) . in [ s : energy ] , the estimates on @xmath161 are obtained by the lyapunov energy method ( lemma [ l : energy ] ) .
the three key estimates ( lemmas [ l : param ] , [ l : locvir ] , [ l : energy ] ) are combined to give the proof of theorem [ t : main ] in [ s : proof ] .
galina perelman shared with me a set of notes illustrating how to apply the martel - merle local virial identity to this problem .
the present paper is essentially an elaboration of this note , and hence i am very much indebted to her generous assistance .
i thank also maciej zworski for initially proposing the problem , providing the numerical codes , and for helpful discussions .
i am partially supported by a sloan fellowship and nsf grant dms-0901582 .
recall that @xmath162 . since @xmath163
, we see that we must consider the schrdinger operator with pschl - teller potential @xmath164 with @xmath165 .
the spectral resolution of operators of the type @xmath166 is deduced via hypergeometric functions in the appendix of guillop - zworski @xcite . from this analysis
, we obtain the spectrum of @xmath167 is @xmath168 .
the @xmath5 normalized eigenfunctions corresponding to the first two eigenvalues are @xmath169 denote by @xmath170 the corresponding eigenspaces and @xmath171 the corresponding projections ( that is , the @xmath5 orthogonal projections and not the symplectic orthogonal projections ) .
suppose that @xmath172 and @xmath173
. then @xmath174 since @xmath167 preserves parity , it suffices to separately prove : _ claim 1_. if @xmath127 is even , @xmath175 , and @xmath172 , then @xmath176 .
_ claim 2_. if @xmath127 is odd , @xmath175 , and @xmath173 , then @xmath177 .
we begin with the proof of claim 1 . since @xmath127 is even , @xmath178 .
resolve @xmath127 as @xmath179 resolve also @xmath180 we compute that @xmath181 from which it follows that @xmath182 we then have @xmath183 which using , , and @xmath184 , implies @xmath185 by the spectral theorem , @xmath186 next , we prove claim 2 . since @xmath127 is odd ,
resolve @xmath127 as @xmath188 resolve also @xmath189 we compute that @xmath190 from which it follows that @xmath191 we then have @xmath192 which , using , , and @xmath184 implies @xmath193 by the spectral theorem , @xmath194 suppose that @xmath195 then @xmath196 by integration by parts , @xmath197 from which we obtain @xmath198 adding to this estimate @xmath199 the estimate , we obtain the claim . of course the above properties of @xmath167 can be converted to properties of @xmath158 , where @xmath200 by scaling and translation .
in particular , we have [ c : spec - bd ] suppose that @xmath201 then @xmath202 where the implicit constant depends on @xmath68 .
we next show by a standard argument that the parameters @xmath203 can be tweaked to achieve the orthogonality conditions .
[ l : ift ] if @xmath204 , there exist constants @xmath205 , @xmath206 such that the following holds . if @xmath207 with @xmath208 , then there exist unique @xmath67 , @xmath68 such that @xmath209 and @xmath210 satisfies @xmath211 define a map @xmath212 by @xmath213 the derivative of @xmath214 with respect to @xmath203 at the point @xmath215 is @xmath216 which is nondegenerate . by the implicit function theorem , the equation @xmath217 can be solved for @xmath203 in terms of @xmath218 in a neighborhood of @xmath219 .
since we will model @xmath220 and @xmath218 solves , we compute that @xmath49 solves @xmath221 where @xmath222 decompose @xmath223 , where @xmath224 is symplectically parallel to @xmath63 and @xmath225 is symplectically orthogonal to @xmath63 .
explicitly , we have @xmath226 by taylor expansion we obtain @xmath227 , where @xmath228 by definition of @xmath225 , we have @xmath229 and @xmath230 , which must then hold at every order in @xmath48 ; in particular , they hold for @xmath231 .
note that by parity @xmath231 in addition satisfies @xmath232 , although this is not expected to hold for @xmath225 at all orders .
it follows that @xmath233
[ l : param ] suppose that we are given @xmath234 and @xmath21 .
( implicit constants below depend only on @xmath235 and @xmath236 . )
suppose that @xmath237 , @xmath49 solves and satisfies , and @xmath238 .
then @xmath239 we first work with the orthogonality condition @xmath240 to obtain . applying @xmath241 to this orthogonality condition
, we obtain @xmath242 substituting the equation for @xmath49 and the relation @xmath243 , we obtain @xmath244 & { \langle}\partial_x \mathcal{k}v , \eta { \rangle}- 4c^2 { \langle}\partial_x v , \eta { \rangle}+{\langle}\partial_x(bv ) , \eta{\rangle}- 3{\langle}\partial_x v^2 , \eta { \rangle } & & \leftarrow \text{i}+\text{ii}+\text{iii}+\text{iv}\\ & + { \langle}f_\| , \eta { \rangle}+ { \langle}f_\perp , \eta { \rangle}+ \dot a { \langle}v , \partial_a \eta { \rangle}+ \dot c { \langle}v , \partial_c \eta { \rangle } & & \leftarrow \text{v}+\text{vi}+\text{vii}+\text{viii } \end{aligned}\ ] ] we have @xmath245 and @xmath246 .
next , we calculate @xmath247 we easily obtain @xmath248 .
next , @xmath249 from which it follows that @xmath250 next , we have @xmath251 and @xmath252 .
finally , @xmath253 using that @xmath254 , we obtain .
to establish , we apply @xmath241 to @xmath81 to obtain @xmath255 { \rangle}\ ] ] substituting the equation for @xmath49 and the relation @xmath243 , we obtain @xmath244 & { \langle}\partial_x \mathcal{k}v , ( x - a)\eta { \rangle}- 4c^2 { \langle}\partial_x v , ( x - a)\eta { \rangle}+ { \langle}\partial_x(bv ) , ( x - a)\eta{\rangle } & & \leftarrow \text{i}+\text{ii}+\text{iii } \\ & - 3{\langle}\partial_x v^2 , ( x - a)\eta { \rangle}+ { \langle}f_\| , ( x - a)\eta { \rangle}+ { \langle}f_\perp , ( x - a)\eta { \rangle}&&\leftarrow \text{iv } + \text{v}+\text{vi } \\ & + \dot a { \langle}v , \partial_a [ ( x - a)\eta ] { \rangle}+ \dot c { \langle}v , ( x - a)\partial_c \eta { \rangle}&&\leftarrow \text{vii}+\text{viii } \end{aligned}\ ] ] note that we do _ not _ have @xmath245 .
we would have @xmath245 if we were working with the orthogonality condition @xmath256 , but as explained previously , this condition can not be imposed on @xmath49 via the method of lemma [ l : ift ] , and even if it could , would not give the coercivity in corollary [ c : spec - bd ] .
we therefore keep term i as is for now .
next , we note that @xmath257 next , @xmath258 . also , @xmath259 it happens that @xmath260 and hence @xmath261 .
finally , @xmath262 , to which we can append the estimate . collecting , we obtain .
next , we begin to implement the martel - merle @xcite virial identity .
let @xmath263 , @xmath264 , @xmath265 on @xmath266 such that @xmath267 on @xmath268 $ ] and @xmath269 on @xmath270 , and @xmath271 on @xmath266 .
let @xmath272 , and for @xmath273 ( to be chosen later ) set @xmath274 .
the following is the ( scaled - out to unity version of ) martel - merle s virial estimate .
[ l : mm ] there exists @xmath166 sufficiently large and @xmath275 sufficiently small such that if @xmath127 satisfies the orthogonality conditions @xmath276 then we have the estimate @xmath277 step 2 in apx .
b of @xcite is a localization argument that shows that it suffices to consider the case @xmath278 and @xmath279 .
some integration by parts manipulations and the fact that @xmath172 convert this case to the estimate @xmath280 where @xmath281 .
the positivity estimate appears as prop . 3 in @xcite and as prop . 6 in @xcite , and is proved in @xcite . by scaling lemma [ l : mm ] , we obtain the following version adapted to @xmath158 .
[ c : mm ] there exists @xmath166 sufficiently large and @xmath275 sufficiently small such that if @xmath49 satisfies the orthogonality conditions , then ( with @xmath282 ) @xmath283 [ l : locvir ] suppose that we are given @xmath234 and @xmath21 .
( implicit constants below depend only on @xmath235 and @xmath236 . )
suppose that @xmath284 , @xmath237 , @xmath49 solves and satisfies , and @xmath285 . then with @xmath282
, we have @xmath286 where @xmath287 and @xmath288 . integrating over @xmath289 $ ]
, we obtain with @xmath290 , @xmath291}^2h_x^1 } \lesssim \|v\|_{l_{[0,t]}^\infty h_x^1 } + t^{1/2}h\,.\ ] ] recalling that @xmath292 , @xmath293 & - \dot a \int \psi ' v^2 + 2\int \psi \ , v \
, \partial_x \mathcal{k}v - 8c^2 \int \psi v \partial_x v + 2\int \psi v \partial_x(bv ) \\ & - 6 \int \psi v \partial_x ( v^2 ) + 2\int \psi v f_0 \end{aligned}\ ] ] we reorganize the terms in the equation to @xmath294 { \hspace{0.3in}&\hspace{-0.3in}}\underbrace{-2\int \psi v \ , \partial_x \mathcal{k } v}_{\text{a } } \underbrace{-2\int \psi v f_0}_{\text{b } } \\ & = \underbrace{-\partial_t \int \psi v^2}_{\text{i } } \underbrace{- \dot a \int \psi ' v^2}_{\text{ii } } \underbrace{- 8c^2 \int \psi v \partial_x v}_{\text{iii } } \underbrace{+ 2\int \psi v \partial_x(bv)}_{\text{iv } } \underbrace{- 6 \int \psi v \partial_x(v^2)}_{\text{v } } \ , .
\end{aligned}\ ] ] note that we have written this equation symbolically in the form @xmath295 and we now consider these terms separately .
integration by parts yields @xmath296 hence @xmath297 from which it follows that @xmath298 from which it follows that @xmath299 we obtain @xmath300 by , @xmath301 placing estimates , , and into , we obtain , for some constant @xmath302 , the bound @xmath303 & - \partial_t \int \psi v^2 + \kappa ( |\dot a - 4c^2 + b(a)|+ \|v\|_{h_x^1 } ) \| e^{-\epsilon |x - a| } v\|_{l_x^2}^2 \\ & + \kappa ( h+|\dot c| ) \|e^{-\epsilon |x - a| } v \|_{l_x^2 } + \kappa h \|v\|_{l_x^2}^2 \end{aligned}\end{aligned}\ ] ] using corollary [ c : mm ] and the assumptions @xmath304 , @xmath237 , we obtain , for some constants @xmath305 , the bound @xmath306 note that implies @xmath307 . substituting this into yields .
recall that @xmath308 . let @xmath309 [ l : energy ]
suppose that we are given @xmath234 and @xmath21 .
( implicit constants below depend only on @xmath235 and @xmath236 . )
suppose @xmath49 solves and satisfies , and @xmath310
. then @xmath311 & |-\dot a + 4c^2 -b(a)| \|e^{-\epsilon |x - a|}v\|_{h_x^1}^2 + h\|v\|_{h_x^1}^2 \\ & + h\|e^{-\epsilon |x - a|}v\|_{h^1 } + \|e^{-\epsilon|x - a|}v\|_{h_x^1}^2 \|v\|_{h_x^1}^2 \end{aligned}\ ] ] we remark that by integrating over @xmath289 $ ] , @xmath312 , and applying corollary [ c : spec - bd ] , we obtain @xmath313}^\infty h_x^1 } \lesssim \begin{aligned}[t ] & \|v_0\|_{h_x^1 } + \| \dot a -4c^2 + b(a)\|_{l_{[0,t]}^\infty}^{1/2 } \| e^{-\epsilon |x - a|}v\|_{l_{[0,t]}^2 h_x^1 } \\ & + t^{1/4}h^{1/2 } \|e^{-\epsilon |x - a| } v\|_{l_{[0,t]}^2h_x^1}^{1/2 } + \|e^{-\epsilon|x - a|}v\|_{l_{[0,t]}^2h_x^1 } \|v\|_{l_{[0,t]}^\infty h_x^1}\ , .
\end{aligned}\ ] ] we compute @xmath314 into i , we substitute .
this gives @xmath315 we have @xmath316 , while @xmath317 .
for @xmath318 , numerous applications of integration by parts gives @xmath319 and hence @xmath320 note @xmath321 but since @xmath322 , @xmath323 , and @xmath324 , we have @xmath325 .
we estimate the second term to obtain @xmath326 combining , we obtain @xmath327 & ( 12c^2 -3b(a)){\langle}\partial_a\eta , v^2{\rangle}-3{\langle}\mathcal{k}v , \partial_x v^2 { \rangle}\\ & + o ( h\|v\|_{h^1}^2 + h\|e^{-\epsilon|x - a|}v\|_{h^1 } ) \ , .
\end{aligned}\ ] ] substituting into ii , we obtain : @xmath328 in ii , we keep only the first term and estimate the rest to obtain @xmath329 note @xmath330 collecting , we obtain @xmath331 & |-\dot a + 4c^2 -b(a)| \|e^{-\epsilon |x - a|}v\|_{h_x^1}^2 + ( h+|\dot c|)\|v\|_{h_x^1}^2 \\ & + h\|e^{-\epsilon |x - a|}v\|_{h^1 } \end{aligned}\ ] ] note that in the addition of terms i and ii , the terms @xmath332 canceled , and in the addition of i and iv , the two @xmath47 coefficients @xmath333 and @xmath334 were combined to give the smaller coefficient @xmath335 . finally , we note that implies @xmath336 . substituting this into yileds .
it will be shown later that theorem [ t : main ] follows from the following proposition .
[ p : main ] suppose we are given @xmath234 and @xmath21 .
( implicit constants below depend only on @xmath235 and @xmath236 ) .
suppose that we are further given @xmath337 , @xmath338 , @xmath339 , @xmath340 , and @xmath341 satisfying , such that @xmath342 let @xmath39 be the solution to with @xmath343 and initial data @xmath344 .
then there exist a time @xmath345 and trajectories @xmath35 and @xmath36 defined on @xmath346 $ ] such that @xmath347 , @xmath348 and the following holds , with @xmath349 : 1 .
[ i : orth ] on @xmath346 $ ] , the orthogonality conditions hold
[ i : time - scale ] either @xmath350 , @xmath351 , or @xmath352 .
[ i : en - bd ] @xmath353}^\infty h_x^1 } \lesssim \kappa h^{1/2 } \,,$ ] 4 .
[ i : local - est ] @xmath354}^2 h_x^1 } \lesssim \kappa h^{1/2}$ ] .
[ i : dota ] @xmath355 .
[ i : dotc ] @xmath356 .
recall our convention that implicit constants depend only on @xmath235 and @xmath236 . by lemma [ l : ift ] and the continuity of the flow @xmath39 in @xmath77
, there exists some @xmath357 on which @xmath35 , @xmath36 can be defined so that holds . now take @xmath358 to be the maximal time on which @xmath35 , @xmath36 can be defined so that holds .
let @xmath359 be first time @xmath360 such that @xmath350 , @xmath351 , @xmath361 , or @xmath362 ( whichever comes first ) . here
, @xmath363 is a constant that will be chosen suitably small at the end of the proof ( depending only upon implicit constants in the estimates , and hence only on @xmath235 and @xmath236 ) .
[ r : bs1 ] we will show that on @xmath346 $ ] , we have @xmath364 , and hence by lemma [ l : ift ] and the continuity of the @xmath39 flow , it must be the case that either @xmath350 , @xmath351 , or @xmath362 ( i.e. the case @xmath361 does not arise ) .
let @xmath107 , @xmath365 , be the maximal time such that @xmath366}^\infty h_x^1 } \leq \alpha \kappa h^{1/2 } \,,\ ] ] where @xmath367 is a suitably large constant related to the implicit constants in the estimates ( and thus dependent only upon @xmath235 and @xmath21 ) .
in fact @xmath368 is taken to be @xmath369 times the implicit constant in front of @xmath370 in the energy estimate .
[ r : bs2 ] we will show , assuming that holds , that @xmath371}^\infty h_x^1 } \leq \frac12 \alpha \kappa h^{1/2}$ ] and thus by continuity we must have @xmath372 . in the remainder of the proof ,
we work on the time interval @xmath289 $ ] , and we are able to assume that the orthogonality conditions hold , @xmath373 , and that holds .
we supress the @xmath367 dependence in the estimates in and below . by lemma [ l : param ] , , and , ( just using that @xmath374 ) it follows that @xmath375 is satisfied . using in ( recall @xmath376 ) ,
we obtain @xmath377}^2 h_x^1 } \lesssim \kappa h^{1/2 } \,.\ ] ] inserting , , and into the energy estimate ( recall @xmath378 ) , we obtain @xmath379}^\infty h_x^1 } \leq \frac{\alpha}{4 } \|v_0\|_{h_x^1 } + c_\alpha ( \kappa^{1/2}h^{1/4}+ \kappa h^{1/2 } + \omega^{1/4 } ) \kappa h^{1/2}\ ] ] provided @xmath380 and @xmath381 , we obtain ( recall @xmath382 ) , we conclude that @xmath371}^\infty h_x^1 } \leq \frac12 \alpha \kappa^2 h$ ] , completing the bootstrap , and demonstrating that @xmath383 . in particular , we have established items , , , in the proposition statement . it remains to prove and . by lemma
[ l : param ] , @xmath384}^2 h_x^1 } + \|e^{-\epsilon |x - a| } v\|_{l_{[0,t]}^2 h_x^1}^2 + th^2\\ \notag & \lesssim h t^{1/2 } \kappa h^{1/2 } + \kappa^2 h + th^2\\ \label{e : c ' '' } & \lesssim \kappa^2 h \,,\end{aligned}\ ] ] establishing item .
similarly by lemma [ l : param ] , we obtain item .
the above proposition can be iterated to obtain : [ c : main ] suppose we are given @xmath234 and @xmath21 .
( implicit constants and the constant @xmath385 below depend only on @xmath235 and @xmath236 ) .
suppose that we are further given @xmath337 , @xmath338 , @xmath386 , @xmath340 , and @xmath341 satisfying , such that @xmath387 let @xmath39 be the solution to with @xmath343 and initial data @xmath344 .
then there exist a time @xmath345 and trajectories @xmath35 and @xmath36 defined on @xmath346 $ ] such that @xmath347 , @xmath348 and the following holds , with @xmath349 : 1 . on @xmath346 $ ] , the orthogonality conditions hold
. 2 . either @xmath350 , @xmath351 , or @xmath388 .
@xmath353}^\infty h_x^1 } \lesssim \beta h^{1/2 } e^{cht } \,,$ ] 4 .
@xmath354}^2 h_x^1 } \lesssim \beta h^{1/2 } e^{cht}$ ] . 5 .
@xmath390 .
let @xmath391 be the constant that appears in item of prop [ p : main ] , and @xmath363 be such that @xmath392 in item of prop .
[ p : main ] .
let @xmath393 for @xmath394 , where @xmath395 is such that @xmath396 .
let @xmath397 denote the time interval @xmath398 $ ] .
apply prop .
[ p : main ] on @xmath397 with @xmath399 .
now we complete the proof of theorem [ t : main ] .
recall that we are given @xmath400 , @xmath21 , @xmath337 , and @xmath338 .
let @xmath22 , @xmath23 , and @xmath27 be given as in def .
[ d : time - scale ] .
let @xmath359 , @xmath35 , @xmath36 be as given in cor .
[ c : main ] .
let @xmath401 and @xmath402 .
then @xmath403 on @xmath404
. then @xmath405 by gronwall s inequality , @xmath406 combining , and applying gronwall s inequality again , we obtain @xmath407 substitution back into yields @xmath408 this completes the proof of theorem [ t : main ] .
in this section , we prove that is globally well - posed in @xmath77 . the local well - posedness ( prop .
[ p : local ] below ) is a consequence of the local smoothing and maximal function estimate of kenig - ponce - vega @xcite and the global well - posedness follows from the local well - posedness and the nearly conserved @xmath5 norm and hamiltonian ( prop .
[ p : global ] below ) . a similar argument
is given in apx .
a of @xcite with an additional smallness assumption on @xmath409 . this smallness assumption could be removed by scaling their result . however , for expository purposes we present a shorter proof here , which also imposes fewer hypotheses on @xmath409 .
[ p : local ] let @xmath415 be the space of functions on @xmath289\times \mathbb{r}$ ] defined by the norm @xmath416 } h_x^1}\ ] ] suppose that @xmath417}^\infty } + \|\partial_x b\|_{l_{t\in [ 0,1]}^\infty l_x^2}<\infty\ ] ] and @xmath418
. then there exists @xmath419 and a solution @xmath420 to with initial data @xmath421 on @xmath289 $ ] .
this solution is the unique solution belonging to the function class @xmath415 .
moreover , the data - to - solution map is lipschitz continuous .
let @xmath422 denote the linear flow ( no potential ) operator , a mapping from functions of @xmath423 to functions of @xmath424 , defined by @xmath425 let @xmath426 denote the duhamel operator , a mapping from functions of @xmath424 to functions of @xmath424 , defined by @xmath427 that is , if @xmath428 , then @xmath127 solves the homogeneous initial - value problem @xmath429 with @xmath430 .
if @xmath431 , then @xmath127 solves the inhomogeneous initial - value problem @xmath432 with @xmath433 .
kenig - ponce - vega @xcite establish the estimates @xmath434 with implicit constants independent of @xmath435 .
in fact , is just the unitarity of @xmath436 on @xmath437 , is ( 2.12 ) in cor . 2.9 in @xcite ,
is ( 3.7 ) in theorem 3.5(ii ) in @xcite , and is not explicitly contained in @xcite , but can be deduced from the above quoted estimates as follows . by the christ - kiselev lemma as stated and proved in lemma 3 of molinet - ribaud @xcite , it suffices to show that @xmath438 by first applying and then the dual to the local smoothing estimate @xmath439 ( lemma 2.1 in @xcite ) , we obtain @xmath440 as claimed .
let @xmath214 be the mapping @xmath441 we seek a fixed point @xmath442 in some ball in the space @xmath415 .
to control inhimogeneities , we need the following four estimates , which are consequences of hlder s inequality : @xmath443 we prove .
@xmath444 which is .
the other estimates , , , are proved similarly . by , , @xmath445 by , ,
@xmath446 applying @xmath447 to and estimating with , , @xmath448 combining , , , and bounding the right - hand sides using , , , , we obtain @xmath449 let @xmath450 , and consider @xmath451 and @xmath452 .
then implies that @xmath453 .
[ p : global ] suppose that @xmath455 and satisfies the following .
suppose that for every unit - sized time interval @xmath426 , we have @xmath456 ( the bound need not be uniform with respect to all time intervals ) .
also suppose that for all @xmath457 , @xmath458 let @xmath418 .
then the local @xmath77 solution to with initial data @xmath421 given by prop .
[ p : local ] extends to a global solution with @xmath459}^\infty l_x^\infty } + \int_0^t \|b_t(s)\|_{l_x^\infty } e^{\gamma(s ) } \ , ds\right ) \,,\ ] ] where @xmath460 is given by @xmath461}^\infty l_x^\infty } \ , ds \,.\ ] ] let @xmath462 ( the momentum ) and recall the definition of @xmath463 , the hamiltonian .
direct computation shows that @xmath464 then @xmath465 , and hence @xmath466 \leq 0 $ ] . from this
, we conclude that @xmath467 in addition , we have @xmath468 hence @xmath469 by the gagliardo - nirenberg inequality @xmath470 and the peter - paul inequality , we have @xmath471 .
hence @xmath472 when combined with the inequalities for @xmath473 and @xmath474 , this gives the conclusion .
j. holmer , g. perelman , and m. zworski , _ effective dynamics of double solitons for perturbed mkdv _ , to appear in comm . math .
phys . , arxiv.org preprint ` arxiv:0912.5122 [ math.ap ] ` .
the numerical illustrations of the results and matlab codes can be found at ` http://math.berkeley.edu/ zworski / hpzweb.html ` . | we study the dynamics of solitons as solutions to the perturbed kdv ( pkdv ) equation @xmath0 , where @xmath1 , @xmath2 is a slowly varying , but not small , potential .
we obtain an explicit description of the trajectory of the soliton parameters of scale and position on the dynamically relevant time scale @xmath3 , together with an estimate on the error of size @xmath4 .
in addition to the lyapunov analysis commonly applied to these problems , we use a local virial estimate due to martel - merle @xcite .
the results are supported by numerics .
the proof does not rely on the inverse scattering machinery and is expected to carry through for the @xmath5 subcritical gkdv-@xmath6 equation , @xmath7 .
the case of @xmath8 , the modified korteweg - de vries ( mkdv ) equation , is structurally simpler and more precise results can be obtained by the method of holmer - zworski @xcite . |
given a graph @xmath0 , a graph @xmath1 is called a _ ramsey graph of @xmath0 _ if there is a monochromatic copy of @xmath0 in every coloring of the edges of @xmath1 with two colors . if @xmath1 is a ramsey graph of @xmath0 , we write @xmath3 and say that @xmath1 arrows @xmath0 .
we denote by @xmath4 the set of all graphs that arrow @xmath0 , and call it the _ ramsey class of @xmath0_. so the ramsey number @xmath5 is the smallest integer @xmath6 such that @xmath7 , where @xmath8 denotes the complete graph on @xmath6 vertices .
two graphs @xmath1 , @xmath0 are called _
ramsey equivalent _ if they have the same ramsey class .
we write @xmath9 if @xmath1 is ramsey equivalent to @xmath0 , and write @xmath10 otherwise .
the study of ramsey classes was initiated by the fundamental work of neetil and rdl @xcite and burr , erds , and lovsz @xcite .
however , the notion of ramsey equivalence of graphs was raised only recently by szab _
et al . _
it was shown in @xcite , that there are two non - isomorphic graphs that are ramsey equivalent , for example @xmath11 , where @xmath12 and @xmath0 is a vertex disjoint union of @xmath13 and @xmath14 for @xmath15 . in this example
@xmath0 is a disconnected graph .
et al . _ formulated a question for connected graphs : [ que : main ] are there two non - isomorphic connected graphs @xmath1 and @xmath0 with @xmath9 ?
note that @xmath16 if and only if there exists a graph @xmath17 such that @xmath18 and @xmath19 or @xmath20 and @xmath21 .
in this case we call @xmath17 a graph , _ distinguishing _ @xmath1 and @xmath0 .
so in order to prove that @xmath16 , it is sufficient to explicitly construct a distinguishing graph .
another approach is to identify a graph parameter @xmath22 , such that @xmath23 implies that @xmath24 . in this case
, we say that @xmath22 is a _
ramsey distinguishing parameter_. the only structural graph parameters that we know to be ramsey distinguishing are the clique number @xmath25 and the odd girth @xmath26 , where @xmath25 is the largest number of vertices in a clique of the graph , and @xmath26 is the length of its shortest odd cycle .
specifically , it is shown in @xcite that if @xmath27 and @xmath28 then there are ramsey graphs @xmath29 such that @xmath30 and @xmath31 .
note that if question [ que : main ] has a negative answer , then any graph parameter is a ramsey distinguishing parameter for the class of connected graphs .
+ in this paper , we provide a supporting evidence for a no-answer to question [ que : main ] by the following theorems , focusing on another graph parameter , the chromatic number , @xmath32 .
[ obs::bipartite ] if @xmath1 and @xmath0 are graphs , @xmath33 and @xmath34 then @xmath24 .
indeed , a sufficiently large complete bipartite graph arrows any fixed bipartite graph @xcite .
but it does not contain , and thus does not arrow any non - bipartite graph . here
, we prove that for several large classes of connected graphs , the chromatic number is a ramsey distinguishing parameter .
a graph is called _ clique - splittable _ if its vertex set can be partitioned into two subsets , each inducing a subgraph of smaller clique number .
note that any graph @xmath1 with @xmath35 is clique - splittable .
in particular all cliques and all planar graphs containing a triangle are clique - splittable .
the triangle - free clique - splittable graphs are precisely the bipartite graphs .
[ thm::differentchrno ] if @xmath36 are graphs , @xmath1 is clique - splittable and @xmath37 , then @xmath24 .
[ cor ] if @xmath36 are graphs , @xmath38 and @xmath39 , then @xmath24 .
theorem [ thm::differentchrno ] distinguishes pairs of graphs of distinct chromatic number under some splittability condition .
the following theorem requires stronger assumptions but also applies to graphs of the same chromatic number .
if @xmath0 is a subgraph ( proper subgraph ) of @xmath1 , we write @xmath40 ( @xmath41 ) .
[ thm::samechrno ] let a connected graph @xmath1 satisfy the following two properties : 1 .
there is an independent set @xmath42 such that @xmath43 .
2 . there is a proper @xmath44-vertex - coloring of @xmath1 in which some two color classes induce a subgraph of a matching .
let @xmath0 be a connected graph , not isomorphic to @xmath1 , such that either @xmath41 or @xmath45
. then @xmath24 . in theorems [ thm::differentchrno ] and [ thm::samechrno ]
we distinguish pairs of graphs under certain properties .
call a graph @xmath1 _ ramsey isolated _ if @xmath24 for any connected graph @xmath0 not isomorphic to @xmath1 .
note that question [ que : main ] asks whether every connected graph is ramsey isolated or not .
we apply the previous results to identify large families of ramsey isolated graphs .
the @xmath46-wheel is the graph on @xmath47 vertices obtained from a cycle of length @xmath46 by adding a vertex adjacent to all vertices of the cycle .
[ thm::pathstar ] 1 . if @xmath1 is connected , @xmath48 and there is a proper @xmath44-vertex - coloring of @xmath1 in which some two color classes induce a subgraph of a matching in @xmath1 , then @xmath1 is ramsey isolated .
any path and any star are ramsey isolated .
each connected graph on at most @xmath2 vertices is ramsey isolated .
[ smalldistinguishinggraphs ] if @xmath49 distinguishes @xmath1 and @xmath0 then @xmath49 has at least @xmath50 vertices .
the distinguishing graphs used in the proof of theorem [ thm::pathstar ] are rather large , except for stars . however , for most pairs @xmath36 of distinct connected graphs on at most @xmath2 vertices there is a distinguishing graph on @xmath50 vertices . a tree @xmath51 on @xmath46 vertices is called _ balanced _ if deleting some edge splits @xmath51 into components of order at most @xmath52 each .
the extremal function @xmath53 is the largest number of edges in an @xmath6-vertex graph with no copy of @xmath0 .
the erds - ss - conjecture states that @xmath54 for any tree @xmath51 on @xmath46 vertices .
we remark that recently , ajtai , komls , simonovits and szemerdi announced a proof of the conjecture for large @xmath46 @xcite .
we state here a much weaker conjecture : [ tree - conjecture ] there is a positive @xmath55 and an integer @xmath56 such that @xmath57 for any tree on @xmath46 vertices and @xmath58 .
[ thm::trees ] if conjecture [ tree - conjecture ] is true then any two trees of different order are not ramsey equivalent . if @xmath59 is a balanced tree on @xmath46 vertices and @xmath60 is any tree on @xmath61 vertices , then @xmath62 .
the next theorem makes use of multicolor ramsey numbers . here
@xmath63 is the minimum integer @xmath6 such that any coloring of the edges of @xmath8 in red , blue , and green has a red @xmath64 , a blue @xmath65 , or a green @xmath66 .
we write @xmath67 if @xmath1 and @xmath0
are not ramsey equivalent in @xmath46 colors , i.e. , there is a graph @xmath17 such that any @xmath46-coloring of its edges contains a monochromatic @xmath1 ( we write @xmath68 ) and there is such a coloring avoiding monochromatic @xmath0 , or vice versa .
[ thm::morecolors ] if @xmath1 and @xmath0 are graphs then @xmath24 if one of the following conditions holds : * there is a graph @xmath49 such that @xmath69 .
* @xmath70 and there is @xmath71 with @xmath67 .
the paper is structured as follows . in section [ sec::overview ]
we provide a more detailed summary of known results on ramsey equivalence , as well as some observations . in section [ sec::lemmas ]
we give known and our preliminary results that will be used in proving the main theorems .
section [ proofs ] contains the proofs of the main results , the appendix provides lemmas used to prove remark [ smalldistinguishinggraphs ] and other routine lemmas . finally , section [ conclusions ] contains conclusions and open questions .
we refer the reader to @xcite for all standard notations in graph theory and to figure [ fig::5vert ] for notations used for small graphs .
we omit floors and ceilings as long as the meaning is clear from context .
we also assume that the graphs under consideration have at least one edge .
as mentioned in the introduction , two connected non - isomorphic graphs @xmath1 and @xmath0 are not ramsey equivalent if one of them is a clique @xcite , or if @xmath72 @xcite or when @xmath73 @xcite .
note that it is an open question whether @xmath74 for all graphs @xmath1 @xcite , where @xmath75 is the length of a shortest cycle in @xmath1 .
there are several other results about ramsey classes that are useful in checking whether some two given graphs are ramsey equivalent or not .
note that two graphs are ramsey equivalent if and only if they have the same set of minimal ramsey graphs .
a graph @xmath17 is _ minimal ramsey _ for @xmath0 if @xmath18 , but @xmath76 for any proper subgraph @xmath77 of @xmath17 .
we need to define a few graphs to state the known results : @xmath78 , @xmath79 is the path , cycle on @xmath6 vertices , respectively , @xmath80 is a graph on @xmath81 vertices such that one vertex has degree @xmath82 and the other vertices induce @xmath13 , @xmath83 is the complete bipartite graph with parts of sizes @xmath84 and @xmath85 , respectively .
it was shown in @xcite that if @xmath0 does not contain @xmath83 , @xmath86 or @xmath87 , then there is @xmath88 such that @xmath17 does not contain @xmath83 . in general , for an integral graph parameter @xmath22 , we define @xmath89 if @xmath90
then clearly @xmath24 .
when @xmath91 , the minimum degree , a number of results have been obtained : @xmath92 , @xcite , @xmath93 , @xcite , @xmath94 , for @xmath95 , @xcite , @xmath96 , @xcite , @xmath97 if @xmath0 is a tree , @xcite , or when @xmath0 is @xmath98 plus a pending edge , @xcite , @xmath99 for even @xmath100 , @xcite .
burr , erds , lovsz @xcite conjectured that for every integer @xmath32 there is @xmath0 with @xmath101 and @xmath102 .
this conjecture has been proven by zhu @xcite .
_ proved that @xmath103 @xcite . in
@xcite the authors asked whether @xmath104 is bounded by a function depending on maximum degree @xmath105 only .
they prove that if @xmath0 is a tree , then @xmath106 . we observe that for a non - bipartite @xmath0 , @xmath107 , see lemma [ lem::maxdegree ] . in
@xcite a threshold @xmath108 is given , such that almost all graphs with density larger than @xmath108 are ramsey for @xmath0 and almost all with smaller density are not .
the introduction of @xcite covers several results on graphs with infinitely many minimal ramsey graphs .
it remains unclear whether a connected graph could or could not be ramsey equivalent to its subgraph .
note that an edge - transitive graph @xmath0 is not ramsey equivalent to any of its subgraphs because coloring the edges of @xmath109 for a ramsey minimal graph @xmath17 of @xmath0 in two colors , gives a monochromatic copy of @xmath110 for some edge @xmath111 .
since @xmath110 is isomorphic to @xmath112 for any two edges , @xmath110 contains any proper subgraph @xmath113 of @xmath0 .
we see that @xmath114 , but @xmath115 .
the following lemma is an easy generalization of the focusing lemma in @xcite .
[ lem::focusing ] let @xmath116 be a bipartite graph with a @xmath117-edge - coloring .
then there is a subset @xmath118 , @xmath119 , such that for each @xmath120 all edges from @xmath84 to @xmath121 are of the same color .
[ lem::ramseygraphsamecliqueno ] for any graph @xmath1 there is a graph @xmath122 with @xmath123 .
we write @xmath124 and @xmath125 for the vertex set , respectively the edge set , of a graph or hypergraph @xmath126 .
a hypergraph is @xmath46-uniform if every hyperedge has size @xmath46 .
the girth of a hypergraph is the smallest number @xmath71 of distinct vertices @xmath127 and distinct hyperedges @xmath128 such that for each @xmath129 , @xmath130 , @xmath131 ( indices taken modulo @xmath46 ) . the independence number of a hypergraph is the size of a largest set of vertices which does not contain a hyperedge completely .
the chromatic number of a hypergraph is the smallest number of colors in a proper vertex coloring , i.e. , a coloring without monochromatic hyperedges .
[ lem::hypergraphgirthindependence ] for any integers @xmath132 and any @xmath133 there is an integer @xmath6 and a @xmath46-uniform hypergraph on @xmath6 vertices with girth at least @xmath134 and independence number less than @xmath135 . from this lemma one easily derives the following well - known result .
[ lem::hypergraphgirthchromatic ] for any integers @xmath136 there is a @xmath46-uniform hypergraph with girth at least @xmath134 and chromatic number at least @xmath32 .
let @xmath137 and let @xmath126 denote a @xmath46-uniform hypergraph with girth at least @xmath134 and independence number at most @xmath138 , which exists by lemma [ lem::hypergraphgirthindependence ] . in any @xmath32-coloring of @xmath124 there is a set of at least @xmath139 vertices of the same color .
hence this color class induces an edge of @xmath126 . thus the coloring is not proper and @xmath140 . for graphs
@xmath49 , @xmath1 and for @xmath133 we write @xmath141 if for any set @xmath142 with @xmath143 , we have @xmath144\to g$ ] .
$ ] denotes the subgraph of @xmath49 induced by the vertices in @xmath145 .
[ lem::epsramsey ] for any @xmath133 and any graph @xmath0 , there is a graph @xmath49 with @xmath146 and @xmath147 .
let @xmath148 be a graph such that @xmath149 and @xmath150 .
such a graph exists by lemma [ lem::ramseygraphsamecliqueno ] .
further let @xmath126 denote a @xmath151-uniform hypergraph of girth at least @xmath152 and no independent set of size @xmath138 , which exists by lemma [ lem::hypergraphgirthindependence ] .
construct a graph @xmath49 by placing a copy of @xmath148 on the vertices of each hyperedge of @xmath126 .
then @xmath49 is a graph on @xmath153 vertices with @xmath154 .
each vertex set of size at least @xmath138 induces a hyperedge in @xmath126 and thus a copy of @xmath148 in @xmath49 which arrows @xmath0 .
we have the following corollary , since any graph which arrows @xmath0 contains @xmath0 .
[ lem::sparsegraphrichcopies ] for any @xmath133 and any graph @xmath0 , there is a graph @xmath49 with @xmath155 and each set of @xmath156 vertices in @xmath49 containing a copy of @xmath0 .
[ lem::maxdegree ] if a graph @xmath0 is not bipartite then @xmath107 .
the lower bound is tight .
let @xmath157 and suppose @xmath49 is a graph with @xmath158 .
it is sufficient to prove that @xmath159 .
consider a partition @xmath160 of @xmath161 with the maximum number of edges between @xmath162 and @xmath163 .
if there is a vertex @xmath164 with at least @xmath165 neighbors in @xmath162 , then @xmath166 has at most @xmath167 neighbors in @xmath163 .
thus the partition @xmath168 has at least one more edge between the parts than the original partition , a contradiction .
hence both @xmath169 $ ] and @xmath170 $ ] have maximum degree at most @xmath167 .
color all edges between @xmath162 and @xmath163 red and all other edges blue .
then the red subgraph is bipartite and the blue subgraph has maximum degree at most @xmath167 .
thus @xmath159 .
the lower bound is tight since @xmath171 , @xmath172 , where @xmath0 is the graph of maximum degree @xmath165 obtained from @xmath173 by adding an edge between two leaves .
[ lem::extremalno ] let @xmath1 and @xmath0 be graphs .
+ if @xmath174 or if @xmath0 is connected and @xmath175 , then @xmath24 .
in particular , if @xmath1 is a forest and @xmath0 contains a cycle , then @xmath16 .
assume first that @xmath176 .
let @xmath49 be a graph on @xmath6 vertices with @xmath177 edges without a copy of @xmath0
. in any @xmath117-coloring of the edges of @xmath49 one of the color classes contains at least @xmath178 edges , and thus a copy of @xmath1 .
hence @xmath179 , but @xmath159 .
assume now that @xmath0 is connected and @xmath180 .
let @xmath49 be a graph on @xmath6 vertices and @xmath53 edges not containing @xmath0 .
let @xmath181 be the cartesian product of @xmath49 with itself , that is , @xmath182 and @xmath183 if and only if @xmath184 and @xmath185 or @xmath186 and @xmath187 .
then @xmath188 has @xmath189 vertices and @xmath190 edges . in any @xmath117-edge - coloring of @xmath188
there is a color class with at least @xmath191 edges .
this color class contains a copy of @xmath1 , thus @xmath192 . on the other hand ,
we can color the edges of @xmath193 without creating monochromatic copies of @xmath0 by coloring an edge @xmath194 red if @xmath195 and blue otherwise .
note that each color class is a vertex disjoint union of @xmath6 copies of @xmath49 and thus does not contain @xmath0 , as @xmath0 is connected .
thus @xmath196 .
for the second part of the statement let @xmath1 be any forest and @xmath0 be any graph with a cycle @xmath197 .
we have @xmath198 and @xmath199 @xcite .
hence for sufficiently large @xmath6 we have @xmath200 and thus @xmath10 by the first part of the lemma .
let @xmath201 and @xmath202 denote the graphs obtained from @xmath203 by adding two , respectively three , pendent edges at some vertex , let @xmath204 denote the graph obtained from @xmath205 by adding a pendent edge at a vertex of degree @xmath206 , see figure [ fig : k23special ] .
, @xmath207 , @xmath208 , @xmath202 and @xmath204 . ]
the next technical lemmas we prove in the appendix . [ lem : mindegg4 ] @xmath209 .
[ lem : nomonokab ] in any @xmath117-edge - coloring of @xmath210 without monochromatic @xmath204 the vertex of degree @xmath117 is incident to exactly one red and one blue edge .
[ lem::mindegg5 ] @xmath211 .
[ lem : w4isolated ] if @xmath0 is connected and has at least @xmath212 vertices , then @xmath213 .
[ lem : k23isolated ] if @xmath0 is connected and has at least @xmath212 vertices , then @xmath214 .
if @xmath33 then @xmath1 is not ramsey equivalent to any graph @xmath0 of higher chromatic number by observation [ obs::bipartite ] .
so , assume that @xmath215 .
let @xmath216 , @xmath217 , @xmath218 .
we assume @xmath219 , otherwise @xmath24 by lemma [ lem::ramseygraphsamecliqueno ] .
let @xmath220 , @xmath221 , such that @xmath222 $ ] and @xmath223 $ ] each have clique number less than @xmath25 .
let @xmath224 be a vertex disjoint union of @xmath64 and @xmath65 , in particular @xmath225 .
we shall construct a graph @xmath17 such that @xmath226 and @xmath21 .
the building blocks of @xmath17 are a hypergraph @xmath126 and graphs @xmath49 and @xmath148 such that : * @xmath126 is a @xmath206-chromatic , @xmath46-uniform hypergraph of girth at least @xmath227 .
it exists by lemma [ lem::hypergraphgirthchromatic ] .
* @xmath49 is a graph such that @xmath228 and every set of at least @xmath229 vertices in @xmath49 contains a copy of @xmath224 , where @xmath230 .
such a graph exists by lemma [ lem::sparsegraphrichcopies ] .
* @xmath148 is a graph such that @xmath231 and @xmath232 for @xmath233 .
such a graph exists by lemma [ lem::epsramsey ] .
note that @xmath153 depends on @xmath234 and @xmath44 ; @xmath235 in turn depends on @xmath153 , @xmath236 , and @xmath1 , so @xmath235 depends only on @xmath0 and @xmath1 .
so , @xmath55 and @xmath237 are constants depending on @xmath0 and @xmath1 .
construct a graph @xmath17 by replacing the vertices @xmath238 of @xmath126 with pairwise vertex disjoint copies of @xmath148 on vertex sets @xmath239 and placing a complete bipartite graph between two copies of @xmath148 if and only if the corresponding vertices belong to the same hyperedge of @xmath126 , see figure [ fig : clique - splittable_1 ] .
-uniform hypergraph @xmath240 , @xmath241 .
right : the graph @xmath17 . ] to show that @xmath21 color each edge with both endpoints in some @xmath242 red , @xmath243 , and all other edges blue .
the red subgraph is a vertex disjoint union of copies of @xmath148 , its clique number is strictly less than @xmath25 , so it does not contain @xmath0 , whose clique number is @xmath25 .
the blue subgraph is a union of complete @xmath46-partite graphs induced by @xmath242 , @xmath243 . to see that the blue subgraph does not contain a copy of @xmath0 , consider any copy of @xmath0 in @xmath17 and consider sets @xmath244 intersecting the vertex set of this copy .
since @xmath126 has girth at least @xmath227 , @xmath245 do not form a cycle in @xmath126 , thus the blue graph induced by @xmath244 is @xmath46-partite .
however , @xmath218 , so the blue subgraph does not contain a copy of @xmath0 .
+ next we shall show that @xmath20 . consider a @xmath117-edge - coloring of @xmath17 .
recall that @xmath246 .
we write @xmath247 if there is a hyperedge in @xmath126 containing both @xmath248 and @xmath249 .
[ claim1 ] for any @xmath250 , @xmath251 , any @xmath129 , @xmath252 , @xmath242 contains a subset @xmath253 that is the vertex set of a monochromatic copy of @xmath49 and such that for any @xmath254 and any @xmath255 with @xmath247 , @xmath256 , all edges from @xmath166 to @xmath257 , are of the same color .
( left ) and claim [ claim2 ] ( right ) .
thin edges are red and thick edges blue . ]
we prove claim [ claim1 ] by induction on @xmath250 using the focusing lemma ( lemma [ lem::focusing ] ) .
when @xmath258 , we see that @xmath259 $ ] is isomorphic to @xmath148 and @xmath232 .
so in particular @xmath260 and there is a monochromatic copy of @xmath49 on some vertex set @xmath261 .
assume that @xmath262 form vertex sets of monochromatic copies of @xmath49 satisfying the conditions of claim [ claim1 ] .
apply the focusing lemma to the bipartite graph with parts @xmath263 and @xmath264 .
it gives a subset @xmath265 such that for any @xmath266 , all edges between @xmath166 and @xmath267 , if any , are of the same color and such that @xmath268 . thus @xmath269 $ ] contains a monochromatic copy of @xmath49 , because @xmath270 $ ] is isomorphic to @xmath148 and @xmath232 .
call the vertex set of this copy @xmath271 .
[ claim2 ] for any @xmath250 , @xmath272 , and @xmath129 , @xmath273 , each @xmath242 contains a subset @xmath274 that is the vertex set of a monochromatic copy of @xmath224 and such that for each @xmath255 with @xmath247 , @xmath275 , @xmath276 , @xmath277 are partite sets of a monochromatic complete bipartite graph .
we prove claim [ claim2 ] by induction on @xmath278 using the pigeonhole principle .
when @xmath279 , we see that @xmath280 forms the vertex set of a monochromatic @xmath49 , that in turn contains a monochromatic @xmath224 .
denote the vertex set of this @xmath224 as @xmath281 .
assume that @xmath282 form vertex sets of monochromatic copies of @xmath224 satisfying the conditions of claim [ claim2 ] .
consider @xmath283 and recall from claim [ claim1 ] that each vertex in @xmath283 sends only red or only blue edges to each @xmath276 with @xmath284 , @xmath285 . if @xmath286 then at least half of the vertices in @xmath283 send monochromatic stars of the same color to @xmath287 .
if @xmath288 then at least half of those send monochromatic stars of the same color to @xmath289 , and so on .
so at least @xmath290 vertices of @xmath283 send monochromatic stars of the same color to each @xmath276 with @xmath284 for @xmath285 .
we denote the set of these vertices by @xmath291 .
since @xmath292 $ ] forms the vertex set of a monochromatic @xmath49 , and @xmath293 , the definition of @xmath49 implies that @xmath294 $ ] contains a monochromatic copy of @xmath224 .
we denote the vertex set of this copy by @xmath295 .
red copies of @xmath224 corresponding to the @xmath46 vertices of a hyperedge of @xmath240 . here
the complete bipartite graph between any two of these @xmath46 copies is also monochromatic . ]
applying claim [ claim2 ] with @xmath258 , we see that each vertex @xmath248 of @xmath126 corresponds to a monochromatic copy of @xmath224 with vertex set @xmath276 , such that all edges between any two such copies from a common hyperedge have the same color . assigning the color of this @xmath224 to @xmath248
gives a @xmath117-coloring of @xmath124 .
since @xmath296 , there is a monochromatic hyperedge , without loss of generality with red vertices @xmath297 .
thus in @xmath17 there are @xmath46 red copies of @xmath224 on vertex sets @xmath298 , such that @xmath299 are partite sets of monochromatic complete bipartite graphs , for all @xmath300 , @xmath301 , see figure [ fig : clique - splittable_2 ] .
if at least one such bipartite graph is red , then there is a red copy of @xmath1 obtained by taking a red @xmath302 from one part and a red @xmath303 from the other part .
so we can assume that all such bipartite graphs are blue , forming a complete @xmath46-partite graph with each part of size @xmath304 .
since @xmath305 , there is a blue copy of @xmath1 .
thus @xmath226 .
since @xmath21 , we have that @xmath24 .
this concludes the proof of theorem [ thm::differentchrno ] .
let @xmath1 and @xmath0 be two graphs such that @xmath39 and @xmath38 . consider an arbitrary proper @xmath44-vertex - coloring of @xmath1 .
let @xmath162 denote the union of @xmath306 color classes and @xmath307 . since @xmath308 , every maximum clique contains a vertex from both sets @xmath242 , @xmath309 .
thus , @xmath1 is clique splittable .
so , if @xmath310 then @xmath24 by theorem [ thm::differentchrno ] .
if @xmath311 , then @xmath312 ( where we assume @xmath313 by lemma [ lem::ramseygraphsamecliqueno ] ) .
thus , @xmath0 is clique - splittable with the same arguments as above .
hence @xmath24 by theorem [ thm::differentchrno ] .
our construction is similar to the one from lemma 3.9 in @xcite .
consider a connected graph @xmath0 .
we may assume @xmath219 by lemma [ lem::ramseygraphsamecliqueno ] .
note that @xmath1 is clique - splittable since @xmath314 and @xmath315 .
further note that if @xmath1 is bipartite , the conditions of the theorem imply that @xmath1 is a union of a matching and a set of independent vertices .
however , @xmath1 is assumed to be connected , and thus it must be a single edge . since a single edge is ramsey isolated , we can assume that @xmath215 .
+ in the first part of the proof , we assume that @xmath316 and @xmath317 .
let @xmath318 , @xmath319 and let @xmath250 denote the size of a matching induced by two color classes of some proper @xmath46-vertex - coloring of @xmath1 .
note that @xmath320 since there is at least one edge between any two color classes .
further let @xmath321 , @xmath322 and @xmath323 be a vertex disjoint union of @xmath324 and @xmath145 independent vertices , i.e. , @xmath323 is the graph obtained from @xmath1 by deleting all edges incident to @xmath145 .
then @xmath325 . let @xmath326 be a vertex disjoint union of @xmath327 copies of @xmath323 and @xmath328 be a vertex disjoint union of @xmath329 copies of @xmath1 .
let @xmath330 .
let @xmath49 be a graph with @xmath331 and @xmath332 , which exists by lemma [ lem::epsramsey ] .
we construct a graph @xmath17 by taking the vertex disjoint union of a copy of @xmath328 and @xmath333 copies of @xmath49 denoted by @xmath334 and placing a complete bipartite graph between @xmath335 and @xmath336 , @xmath337 and between @xmath335 and @xmath328 , @xmath338 , see figure [ fig : matching - coloring_1 ] .
+ consisting of @xmath333 copies @xmath334 of @xmath49 and one copy of @xmath339 and all possible edges between distinct copies .
we have @xmath340 , @xmath326 consists of @xmath329 copies of @xmath323 and @xmath339 consists of @xmath329 copies of @xmath1 . ]
we shall show that @xmath20 , but @xmath21 .
color all edges within each @xmath335 and within @xmath328 red and all other edges blue . since @xmath341 , @xmath342 , and thus @xmath343 , @xmath338 . since @xmath316 and @xmath0
is connected , we have that @xmath344 . thus there is no red copy of @xmath0 . on the other hand ,
the blue subgraph is a complete @xmath345-partite graph , but @xmath346 .
thus there is no blue copy of @xmath0 .
it remains to show that @xmath226 .
consider a @xmath117-edge - coloring of @xmath17 .
assume for the sake of contradiction that there is no monochromatic copy of @xmath1 .
we prove the following claim , similar to claim [ claim-1 ] in the proof of theorem [ thm::differentchrno ] , by induction on @xmath347 ( up to renaming colors ) , see figure [ fig : matching - coloring_2 ] for an illustration . for each @xmath347 , @xmath348 , and each @xmath129 , @xmath349 , there is a red copy @xmath350 of @xmath326 in @xmath335 .
moreover for each @xmath129 , @xmath351 , each vertex @xmath166 in @xmath350 and each @xmath255 , @xmath352 , all edges between @xmath166 and @xmath353 are of the same color . there is a set @xmath162 of @xmath354 vertices in @xmath355 such that for each vertex in @xmath328 all edges to @xmath162 are of the same color by the focusing lemma ( lemma [ lem::focusing ] ) . since @xmath331 there is a monochromatic copy @xmath356 of @xmath326 in @xmath357 $ ] .
assume without loss of generality that @xmath356 is red .
this proves the claim for @xmath358 , and for @xmath241 .
suppose @xmath359 and there are red subgraphs @xmath360 , satisfying the conditions of the claim .
we apply the focusing lemma to the complete bipartite graph with one part @xmath361 and the other part @xmath362 .
there is a set @xmath363 of size @xmath364 , such that for each vertex @xmath166 in @xmath365 all edges from @xmath166 to @xmath366 are of the same color .
since @xmath331 there is a monochromatic copy @xmath367 of @xmath326 in @xmath368 $ ] . it remains to prove that @xmath367 is red .
assume @xmath367 is blue .
consider the vertices of @xmath369 .
all of them send monochromatic stars to @xmath367 . at most @xmath370 of these stars
are blue , as otherwise these stars together with a blue subgraph of @xmath367 isomorphic to @xmath323 form a blue copy of @xmath1 .
since the number of vertex disjoint copies of @xmath323 in @xmath369 is @xmath371 , there is a red copy @xmath372 of @xmath323 in @xmath356 whose vertices send only red stars to @xmath367 .
taking @xmath372 and @xmath22 vertices from @xmath367 gives a red copy of @xmath1 , a contradiction .
so we may assume that @xmath367 is red , which completes the proof of the claim .
+ consider the red @xmath350 , @xmath373 , given by the claim for @xmath374 .
we say that a vertex in @xmath375 , @xmath376 is bad for @xmath353 if it sends a red star to @xmath353 , for some @xmath377 . since for each @xmath353 there are at most @xmath370 bad vertices , there are at most @xmath378 bad vertices overall . since @xmath328 has @xmath379 vertex disjoint copies of @xmath1 , there are at least @xmath380 copies @xmath381 of @xmath1 in @xmath328 without bad vertices .
since each @xmath350 , @xmath382 , has @xmath379 disjoint copies of @xmath323 , there is at least one copy @xmath383 of @xmath323 in @xmath350 without bad vertices , @xmath384 .
note that all @xmath383s are red , @xmath338 , all edges between them are blue , and all edges between a @xmath383 and @xmath385 are blue , @xmath338 , @xmath386 , see figure [ fig : matching - coloring_3 ] . in each of @xmath334 and @xmath250 copies @xmath387 of @xmath1 in @xmath339 where all edges between distinct copies are blue .
if each @xmath388 , @xmath389 , has a blue edge we find a blue copy of @xmath1 in here . ] by assumption each @xmath385 , @xmath390 , has a blue edge , since otherwise there is a red copy of @xmath1 .
but then we can find a blue copy of @xmath1 by identifying these blue edges with the matching of size @xmath250 induced by the union of two color classes of @xmath1 , picking the other vertices of these two color classes from @xmath391 and the vertices of the other @xmath333 color classes of @xmath1 from @xmath383 , @xmath338 . since @xmath392 , there is sufficient number of vertices for each color class .
altogether we have a contradiction to our assumption that there are no monochromatic copies of @xmath1 .
hence @xmath226 .
this concludes the proof in case when @xmath316 and @xmath317 . + now , in the second part of the proof , we assume that @xmath40
. then @xmath393 .
since we assume that @xmath219 , we have @xmath394 .
thus , @xmath0 is clique - splittable .
assume first that @xmath311 .
then we have @xmath24 by theorem [ thm::differentchrno ] , applied with roles of @xmath1 and @xmath0 switched .
the last case to consider is when @xmath395 ( and @xmath40 ) .
now any proper @xmath44-vertex - coloring of @xmath1 with two color classes inducing a subgraph of a matching gives such a coloring of @xmath0 , too .
thus , the first part of the proof applied with roles of @xmath1 and @xmath0 switched shows that @xmath24 .
assume that @xmath48 and in some proper @xmath44-vertex - coloring of @xmath1 two color classes induce a subgraph of a matching .
then @xmath1 satisfies the requirements of theorem [ thm::samechrno ] . if @xmath396 then @xmath397 by lemma [ lem::ramseygraphsamecliqueno ] .
so , we can assume that @xmath313 .
if @xmath398 or @xmath317 , then @xmath24 by theorem [ thm::samechrno ] .
if @xmath316 and @xmath399 , then @xmath400
. thus @xmath401 , a contradiction . to see that a star @xmath402 is not ramsey equivalent to any other graph ,
observe that @xmath403 is a minimal ramsey graph for @xmath145 , but @xmath403 is minimal ramsey for neither any connected subgraph of @xmath145 nor any connected graph that is not a subgraph of @xmath145 .
it remains to show that a path is not ramsey equivalent to any other connected graph .
let @xmath404 , a path on @xmath250 vertices , and @xmath0 be a connected graph not isomorphic to @xmath1 . if @xmath0 is a path of different length , then @xmath24 since @xmath405 @xcite and hence @xmath406 .
so assume @xmath0 is not a path .
if @xmath0 is not a tree , then by lemma [ lem::extremalno ] we have @xmath24 . otherwise , @xmath0 is a tree and @xmath407 .
then @xmath408 @xcite , while an easy argument due to alon _ _ et al . _ _
@xcite shows that @xmath409 .
indeed , for any @xmath152-regular graph @xmath49 with girth at least @xmath410 we have @xmath411 as follows .
considering any @xmath117-edge - coloring of @xmath49 , we see that since @xmath49 has average degree @xmath152 at least one color class has average degree at least @xmath117 , i.e. , contains a cycle . since @xmath412 , this monochromatic cycle has length at least @xmath410 , and thus contains @xmath413 .
vertices with their ramsey numbers @xmath414 . ]
figure [ fig::5vert ] shows all non - trivial connected graphs on at most @xmath2 vertices .
let @xmath415 .
observe that any connected graph on at most @xmath2 vertices which is not in @xmath145 satisfies the conditions of theorem [ thm::pathstar].1 or [ thm::pathstar].2 and thus is ramsey isolated ; figure [ fig::5vert ] also indicates proper colorings for the graphs which satisfy the conditions of theorem [ thm::pathstar].1 .
it remains to prove that each graph in @xmath145 is ramsey isolated .
we consider the graphs in @xmath145 grouped according to their ramsey number .
let @xmath416 , @xmath417 , @xmath418 , and @xmath419 . here
@xmath420 contains the graphs from @xmath145 of ramsey number @xmath212 , @xmath421 the graph of ramsey number @xmath422 , @xmath423 those of ramsey number @xmath424 , and @xmath425 the graph of ramsey number @xmath426 , see @xcite .
first of all we consider @xmath427 . consider a connected graph @xmath0 which is not isomorphic to @xmath1 . if @xmath428 , then @xmath429 and hence @xmath397 . indeed ,
if @xmath0 is a star then coloring the edges of a @xmath430 in @xmath431 red and all other edges blue does not yield a monochromatic @xmath0 . if @xmath0 is not a star , then color a copy of @xmath432 in @xmath431 red and all other edges blue .
then the red edges form a star and the blue connected subgraph contains only @xmath2 vertices , so the coloring has no monochromatic @xmath0 .
so assume that @xmath433 .
if @xmath434 we have @xmath24 . indeed either @xmath435 and @xmath436 , or @xmath437 and @xmath0 is ramsey isolated by theorem [ thm::pathstar].1 or [ thm::pathstar].2 .
so it remains to distinguish the graphs in @xmath420 from each other .
we have @xmath438 and @xmath439 due to the coloring given in figure [ fig::h5x ] .
we claim that @xmath440 . indeed , consider a @xmath117-edge - coloring of @xmath441 and a vertex @xmath442 of degree @xmath2 . without loss of generality
@xmath442 is incident to @xmath206 red edges @xmath443 , @xmath444 and @xmath445 . then there is a red @xmath446 or all edges between @xmath447 and @xmath448 are blue .
but in the latter case the vertices in @xmath441 other than @xmath442 give a blue @xmath446 .
in particular @xmath440 and thus @xmath449 and @xmath450 . finally
@xmath451 @xcite and @xmath452 @xcite ( for the latter see a remark in the conclusion of @xcite ) .
thus @xmath453 and hence @xmath1 is ramsey isolated .
\(1 ) without monochromatic @xmath454 ( 1 ) , a coloring of @xmath441 without monochromatic @xmath203 ( 2 ) and a coloring of @xmath441 without monochromatic @xmath455 ( 3).,title="fig:",height=68 ] \(2 ) without monochromatic @xmath454 ( 1 ) , a coloring of @xmath441 without monochromatic @xmath203 ( 2 ) and a coloring of @xmath441 without monochromatic @xmath455 ( 3).,title="fig:",height=68 ] \(3 ) without monochromatic @xmath454 ( 1 ) , a coloring of @xmath441 without monochromatic @xmath203 ( 2 ) and a coloring of @xmath441 without monochromatic @xmath455 ( 3).,title="fig:",height=68 ] next consider @xmath456 , i.e. , @xmath457 and a connected graph @xmath0 which is not isomorphic to @xmath1 . if @xmath433 we have @xmath24 , because @xmath458 and hence @xmath436 or @xmath437 ( and @xmath0 is ramsey isolated by theorem [ thm::pathstar].1 or [ thm::pathstar].2 ) . if @xmath0 is bipartite then @xmath24 by observation [ obs::bipartite ] . if @xmath428 and @xmath0 is not bipartite , then @xmath459 .
indeed color the edges of @xmath460 with two vertex disjoint red copies of @xmath461 and all other edges blue .
then each connected component of the red subgraph has @xmath2 vertices and the blue subgraph is bipartite .
in particular there is no monochromatic copy of @xmath0 .
we conclude that @xmath24 , so @xmath1 is ramsey isolated .
next consider @xmath462 and a connected graph @xmath0 which is not isomorphic to @xmath1 . since @xmath205 is bipartite but
@xmath463 is not , the two graphs in @xmath423 are not ramsey equivalent by observation [ obs::bipartite ] . if @xmath433 then @xmath24 , because either @xmath464 , or @xmath436 , or @xmath437 .
so assume @xmath428 .
then @xmath465 by lemma [ lem : k23isolated ] . if @xmath0 is bipartite then @xmath466 by observation [ obs::bipartite ] . if @xmath0 is not bipartite then @xmath466 , since @xmath467 as argued above ( when considering @xmath421 ) .
altogether @xmath1 is ramsey isolated .
finally consider @xmath468 , i.e. @xmath469 , and a connected graph @xmath0 which is not isomorphic to @xmath1 . if @xmath433 we have @xmath24 , because @xmath470 and hence @xmath436 or @xmath437 . if @xmath428 then @xmath471 by lemma [ lem : w4isolated ] .
hence @xmath1 is ramsey isolated .
next we show that all but @xmath472 pairs from the @xmath473 pairs of distinct connected graphs on at most @xmath2 vertices are distinguished by a small graph . for @xmath474
such pairs of such graphs @xmath475 we give a distinguishing graph on @xmath50 vertices , which is clearly best - possible . among others , we will use graphs @xmath17 and @xmath77 given in figure [ fig::graphgamma ] and figure [ fig::gammaprimeg3 ] respectively .
the graph @xmath77 is obtained from @xmath476 by adding two independent vertices of degree @xmath2 such that these two vertices have exactly @xmath152 common neighbors . without monochromatic @xmath66 . ] without monochromatic @xmath66 . ]
first of all note that two graphs @xmath1 , @xmath0 of different ramsey number are distinguished by @xmath8 where @xmath477 which is the smallest possible order of a distinguishing graph .
this result distinguishes already lots of graphs using small graphs .
it remains to distinguish pairs of connected graphs on at most @xmath2 vertices of the same ramsey number .
hence we need to consider the following sets of graphs corresponding to ramsey number @xmath212 , @xmath422 , @xmath424 and @xmath426 respectively ; * ramsey number 6 : * @xmath478 .
we have @xmath479 ( pigeonhole principle ) but @xmath480 ( @xmath432 does not contain these ) , @xmath481 ( lemma [ lem::h52p4e ] ) but @xmath482 ( figure [ fig::h5x ] ) , @xmath483 ( lemma [ lem::h53p5 ] ) but @xmath484 ( figure [ fig::h5x ] ) , @xmath485 ( lemma [ lem::k55c4 ] ) but @xmath486 ( since @xmath455 not bipartite ) .
it remains open to distinguish @xmath203 and @xmath487 by some small graph . * ramsey number 9 : * @xmath488 .
we have @xmath489 ( lemma [ lem::gammag1 ] , [ lem::gammag3 ] ) but @xmath490 ( lemma [ lem::maxdegree ] ) , @xmath491 ( lemma [ lem::gammaprimeg1 ] ) but @xmath492 ( figure [ fig::gammaprimeg3 ] ) , @xmath493 ( lemma [ lem::h85g123 ] ) but @xmath494 ( figures [ fig::h85 ] , [ fig::h86 ] ) .
we conjecture @xmath495 ( motivated by lemma [ lem::h86g4 ] ) but @xmath496 ( figure [ fig::h86 ] ) .
it remains open to distinguish @xmath497 from @xmath498 and @xmath499 by small graphs . without monochromatic @xmath499 . ] without monochromatic @xmath499 . ] * ramsey number 10 : * @xmath500 .
we have @xmath501 ( lemma [ lem::k1212 ] ) but the other graphs are not bipartite , @xmath502 ( lemma [ lem::h96h12 ] ) but @xmath503 ( figures [ fig::h96h3 ] , [ fig::h98h4 ] ) . in this case
it remains to distinguish each pair within the sets @xmath504 and @xmath505 with a small graph . * ramsey number 18 : * @xmath506 .
we have not found any small distinguishing graph for the pairs in this case . without monochromatic @xmath463
is obtained by identifying vertices of the same label . ] without monochromatic @xmath463 is obtained by identifying vertices of the same label . ]
assume first that conjecture [ tree - conjecture ] is true .
let @xmath59 and @xmath507 be trees on @xmath46 and @xmath508 vertices respectively , @xmath509 .
note that @xmath510 .
indeed , just take @xmath511 vertex disjoint copies of @xmath512
. then @xmath513 for sufficiently large @xmath6 . thus @xmath514 and
lemma [ lem::extremalno ] implies that @xmath62 .
now , we shall prove the second statement of theorem [ thm::trees ] without assuming the validity of conjecture [ tree - conjecture ] .
let @xmath59 be a balanced tree on @xmath46 vertices and @xmath60 be any tree on @xmath515 vertices .
let @xmath1 be a @xmath46-regular graph of girth at least @xmath46 , which is known to exist @xcite .
we construct a bipartite @xmath46-regular graph @xmath516 of girth at least @xmath46 from @xmath1 by taking for each @xmath166 in @xmath1 two vertices @xmath517 in @xmath516 and for every edge @xmath518 in @xmath1 the edges @xmath519 and @xmath520 in @xmath516 .
finally , let @xmath521 be the line graph of @xmath516 .
we shall show that @xmath522 and @xmath523 . as @xmath516 is bipartite , @xmath521 is a union of two graphs @xmath355 , @xmath524 , each is a vertex disjoint union of copies of @xmath525 , where each clique in @xmath335 corresponds to a set of edges incident to a vertex in the @xmath129th partite set of @xmath516 , @xmath309 .
note that a clique in @xmath355 intersects a clique in @xmath524 by at most one vertex and that each vertex in @xmath49 belongs to two cliques , one from @xmath355 and one from @xmath524 .
coloring @xmath355 red and @xmath524 blue gives no monochromatic @xmath507 since each monochromatic connected component has @xmath526 vertices .
thus @xmath527 .
next , we show that @xmath528 .
let @xmath529 be an edge of @xmath59 such that the components of @xmath530 rooted at @xmath166 and @xmath531 have order at most @xmath532 .
consider any edge - coloring of @xmath49 with colors red and blue .
note that @xmath533 and @xmath534 . hence there are at least @xmath535 red edges or at least @xmath535 blue edges .
( note that conjecture [ tree - conjecture ] , if true , would imply that there is a red or blue copy of @xmath59 , independent of the girth of @xmath516 and whether @xmath59 is balanced . )
assume without loss of generality that there are at least @xmath535 red edges .
consider the red subgraph @xmath536 of @xmath49 and a subgraph @xmath1 of @xmath536 of highest average degree .
it follows that @xmath537 and @xmath538 , and so @xmath539 . if @xmath540 , then @xmath1 is @xmath345-regular and we can embed @xmath59 into @xmath1 greedily .
so without loss of generality we have @xmath541 .
let @xmath542 be a vertex of maximum degree in @xmath1 , i.e. , @xmath543 .
it follows that @xmath542 has incident red edges in both corresponding maximum cliques @xmath544 in @xmath49 .
without loss of generality @xmath542 has at least @xmath545 incident red edges in @xmath546 .
we embed @xmath166 onto @xmath542 , @xmath531 onto a neighbor of @xmath542 in @xmath536 in @xmath547 and all neighbors of @xmath166 different from @xmath531 onto neighbors of @xmath542 in @xmath536 in @xmath546 .
now we can greedily embed the subtrees @xmath548 of @xmath549 with their roots at the designated vertices in @xmath536 .
say @xmath550 is the subtree rooted at @xmath531 . as @xmath551 and @xmath516 has girth greater than @xmath46 , the embeddings of @xmath550 and @xmath552 are in disjoint sets of cliques .
it follows that @xmath528 .
we prove the first part of the theorem .
let @xmath553 . consider a @xmath206-edge - coloring @xmath554 of @xmath555 without red or blue @xmath0 and without green @xmath49 , which exists as @xmath556 .
let @xmath17 denote the graph obtained from @xmath555 by removing all green edges under @xmath554 .
thus @xmath557 due to the coloring @xmath554 restricted to @xmath17 . but @xmath226 , since any @xmath117-edge - coloring of @xmath17 without monochromatic @xmath1 can be extended by the green edges of @xmath554 to an edge - coloring of @xmath555 without red or blue @xmath1 and without green @xmath49 . + we prove the second statement by induction on @xmath46 with @xmath558 being obvious .
let @xmath17 be a graph such that @xmath68 , but @xmath559 , @xmath560 .
let @xmath554 be a @xmath46-edge - coloring of @xmath17 with no monochromatic @xmath0 .
let a graph @xmath77 be obtained from @xmath17 by deleting the edges of color @xmath561 .
we have that @xmath562 since @xmath554 restricted to @xmath77 is a @xmath345-coloring with no monochromatic @xmath0 .
we claim that @xmath563 , which , if true , gives @xmath564 and by induction @xmath24 , as desired .
let us assume for the sake of contradiction that @xmath565 , i.e. , there is a @xmath345-edge - coloring @xmath566 of @xmath77 without monochromatic @xmath1 .
we see that there is a copy of @xmath1 in color @xmath561 of @xmath554 , otherwise the coloring @xmath567 of @xmath17 that is the same as @xmath566 on @xmath77 and that colors all other edges with color @xmath561 has no monochromatic @xmath1 , a contradiction to the fact that @xmath68 . repeating the argument above to all colors in @xmath554 , we see that each of them contains @xmath1 .
more generally , we see that any edge - coloring of @xmath17 with @xmath46 colors avoiding monochromatic @xmath0 must have a monochromatic @xmath1 in each color .
however , since @xmath70 , @xmath77 has no monochromatic @xmath0 under @xmath566 , and hence the coloring @xmath567 of @xmath17 has no monochromatic @xmath0 .
thus @xmath567 must have monochromatic @xmath1 in each color , however there is no monochromatic @xmath1 in any of the colors @xmath568 , a contradiction .
this paper addresses ramsey equivalence of graphs and gives a negative answer to the question of fox _
et al . _
@xcite : `` are there two connected non - isomorphic graphs that are ramsey equivalent ? '' for wide families of graphs determined by so - called `` clique splitting '' properties and chromatic number . in particular , we find an infinite family of graphs that are not ramsey equivalent to any other connected graphs .
this extends the only such known family consisting of all cliques , paths , and stars . replacing @xmath25 with any other `` nice '' ramsey parameter , @xmath22 , generalizes theorems [ thm::differentchrno ] and [ thm::samechrno ] . here
, we say that a parameter @xmath22 is a `` nice '' ramsey parameter , if for any graph @xmath0 and any ramsey graph @xmath17 for @xmath0 , we have @xmath569 and for all @xmath133 equality is attained for at least one @xmath17 with @xmath570 .
so , both @xmath25 and @xmath571 ( the negative of the odd girth ) are nice ramsey parameters .
there are many questions that remain open in this area .
even the following weaker question is very far from being understood : `` are there other graph parameters that distinguish graphs in a ramsey sense ? '' , i.e. , is there a parameter @xmath22 such that @xmath23 implies that @xmath572 ? here , we showed that the chromatic number , @xmath32 , is very likely to be such a distinguishing parameter by proving this implication for graphs satisfying some additional properties .
interestingly enough , it is not clear , but most likely not true that @xmath39 implies that @xmath573 .
indeed , @xmath574 , but the positive answer to the burr - erds - lovsz - conjecture shows that there is a @xmath2-chromatic graph @xmath1 with @xmath575 , so @xmath576 but @xmath577 .
we believe that there are infinitely many pairs of graphs @xmath578 of different chromatic number and the same value for @xmath579 . in this paper
, we addressed the relation between other types of ramsey numbers and ramsey equivalence and got results in terms of multicolor ramsey numbers .
the following questions are open .
if @xmath580 for some graph @xmath49 , does this imply @xmath24 ?
for any two non - isomorphic graphs @xmath36 , is there an integer @xmath46 such that @xmath581 ? here @xmath582 is the smallest integer @xmath6 such that any coloring of edges of @xmath8 with @xmath46 colors contains a monochromatic copy of @xmath1 . for example
, we see that @xmath583 , and @xmath584 , for odd @xmath585 .
another question is whether the fact that @xmath581 for some @xmath46 implies that @xmath24 .
we answered the last question in positive only when @xmath1 is a subgraph of @xmath0 .
cliques play a special role in ramsey theory and got a particular attention in ramsey equivalence .
still , it is not clear for what graphs is a clique a minimal ramsey graph .
specifically , if the size ramsey number @xmath586 is less than @xmath587 , does it imply that @xmath588 is not a minimal ramsey graph for @xmath0 ?
a positive answer to the following question would immediately give a negative answer to the question of fox _
et al . _ : `` is there a graph in the ramsey class of any connected graph @xmath1 , that does not belong to the ramsey class of any other connected graph , except for subgraphs of @xmath1 ? ''
it is also not clear how small could be a distinguishing graph for two not ramsey equivalent graphs .
is there a function @xmath589 such that for any two not ramsey equivalent graphs @xmath1 , @xmath0 , the smallest order of their distinguishing graph is at most @xmath590 ? finally , we show that two trees of different order are not ramsey equivalent provided that the erds - ss - conjecture is true or if one of the trees is balanced
. we do not know whether there are two ramsey equivalent non - isomorphic trees on the same number of vertices
. 99 m. ajtai , j. komls , m. simonovits , and e. szemerdi .
( manuscript ) .
m. ajtai , j. komls , m. simonovits , and e. szemerdi .
( manuscript ) .
m. ajtai , j. komls , m. simonovits , and e. szemerdi .
( manuscript ) .
n. alon , g. ding , b. oporowski , and d. vertigan . . , 87:231243 , 2003 . l. w. beineke and a. j. schwenk .
in _ proceedings of the fifth british combinatorial conference ( univ .
aberdeen , aberdeen , 1975 ) _ , pages 1722 .
congressus numerantium , no .
utilitas math . ,
winnipeg , man . , 1976 .
b. bollobs , j. donadelli , y. kohayakawa , and r. h. schelp . .
, 7:27 37 , 00 2001 .
s. a. burr .
diagonal ramsey numbers for small graphs . , 7(1):5769 , 1983 . s. a. burr , p. erds , and l. lovsz .
, 1(1):167190 , 1976 . v. chvtal and f. harary .
, 32:389394 , 1972 .
p. erds and a. hajnal .
, 17:6199 , 1966 . j. fox , a. grinshpun , a. liebenau , y. person , and t. szab .
, 109:120133 , 2014 . j. fox and k. lin . ,
54(2):167177 , 2007 . l. gerencsr and a. gyrfs . , 10:167170 , 1967 .
a. grinshpun , r. raina , and r. sengupta .
, june 2014 . f. harary . in _ graph theory and applications ( proc .
conf . , western michigan univ . ,
kalamazoo , mich .
, 1972 ; dedicated to the memory of j. w. t. youngs ) _ , pages 125138 .
lecture notes in math .
, vol . 303 .
springer , berlin , 1972 .
g. r. t. hendry .
diagonal ramsey numbers for graphs with seven edges .
, 32:1134 , 1987 . g. r. t. hendry .
, 13(2):245248 , 1989 .
p. horn , k. g. milans , and v. rdl .
, 21(2):paper 2.5 , 6 , 2014 .
g. krolyi and v. rosta . , 263(1 - 2):8798 , 2001 .
combinatorics and computer science ( palaiseau , 1997 ) .
f. lazebnik , v. a. ustimenko , and a. j. woldar . , 32(1):7379 , 1995 . j. neetil . in _
erds centennial _ , volume 25 of _ bolyai soc .
math . stud .
_ , pages 383407 .
jnos bolyai math .
budapest , 2013 .
j. neetil and v. rdl . , 20(3):243249 , 1976 . j. neetil and v. rdl .
, 20(3):565582 , 1979 . j. neetil and v. rdl . , 1(2):199202 , 1981 . j. neetil and v. rdl .
101(1 - 3):223229 , 1992 . special volume to mark the centennial of julius petersen s `` die theorie der regulren graphs '' , part ii .
v. rdl and a. ruciski . , 8(4):917942 , 1995 . h. sachs .
regular graphs with given girth and restricted circuits .
, 38:423429 , 1963 . t. szab , p. zumstein , and s. zrcher .
, 64(2):150164 , 2010 .
d. b. west .
prentice hall , 2000 .
, 32(7):11681175 , 2011 .
[ lem : mindegg4 ] @xmath209 .
let @xmath17 denote the graph obtained from @xmath591 by adding @xmath422 new vertices and a matching between these and the vertices in @xmath591 .
we have @xmath592 , due to any @xmath152-factorization , and we shall show @xmath593 .
then each minimal ramsey graph of @xmath202 contained in @xmath17 contains at least one of the vertices of degree @xmath561 and thus @xmath209 .
consider the copy @xmath594 of @xmath591 in @xmath17 and let @xmath554 denote a @xmath117-edge - coloring of @xmath594 with no monochromatic copy of @xmath202 .
we will show that there is red and a blue copy of @xmath201 in @xmath594 with the same vertex @xmath542 of degree @xmath152 , see claim [ g4_claim2 ] .
then there is a monochromatic copy of @xmath202 in @xmath17 no matter which color is assigned to the edge pendent at @xmath542 . thus @xmath593 .
[ g4_claim1 ] there is no vertex in @xmath594 with @xmath2 incident edges of the same color under @xmath554
. for the sake of contradiction assume @xmath442 in @xmath594 has @xmath2 incident red edges .
let @xmath595 denote the @xmath2 neighbors of @xmath442 incident to these edges and @xmath596 denote the vertices in @xmath594 not incident to these edges
. then there is at most one red edge between @xmath595 and each vertex in @xmath447 , as otherwise there is a red @xmath202 in @xmath594 .
so there are two distinct vertices @xmath166 , @xmath531 in @xmath595 such that there are only blue edges between @xmath597 and @xmath447 .
since each vertex in @xmath447 is incident to @xmath152 blue edges to @xmath595 , each of the vertices in @xmath447 is the degree @xmath152 vertex in a blue copy of @xmath201 with another vertex from @xmath447 and four vertices from @xmath595 .
thus there are only red edges between @xmath442 and @xmath447 and only red edges within @xmath447 , as otherwise there is a blue @xmath202 .
but then @xmath598 forms a red @xmath203 with @xmath206 red edges pendent at @xmath442 ( those to @xmath595 ) , a monochromatic @xmath202 , a contradiction .
this proves claim [ g4_claim1 ] .
[ g4_claim2 ] there is a vertex in @xmath594 which is the vertex of degree @xmath152 in a red and a blue copy of @xmath201 under @xmath554 . by claim [ g4_claim1 ]
the red and the blue subgraph of @xmath594 under @xmath554 are @xmath152-regular .
consider a vertex @xmath442 in @xmath594 and let @xmath599 and @xmath600 denote the sets of neighbors in @xmath594 adjacent to @xmath442 via red respectively blue edges .
if there are vertices @xmath601 and @xmath602 with two red edges between @xmath166 and @xmath599 and two blue edges between @xmath531 and @xmath600 , then @xmath442 is the degree @xmath152 vertex in a red and in a blue copy of @xmath201 and we are done .
so we assume that there is at most one blue edge between @xmath600 and each vertex in @xmath599 . since @xmath603 and the blue subgraph is @xmath152-regular , each vertex in @xmath599 sends at most @xmath206 blue edges to the other vertices in @xmath599 and at least one blue edge to @xmath600 .
hence there is exactly one blue edge between @xmath600 and each vertex in @xmath599 and @xmath599 forms a blue @xmath604 .
if the blue edges between @xmath599 and @xmath600 form a matching , then @xmath600 induces a blue @xmath203 and two independent red edges .
then each vertex in @xmath600 is the vertex of degree @xmath152 in a blue and in a red copy of @xmath201 and we are done . if the blue edges between @xmath599 and @xmath600 do not form a matching then there is vertex @xmath601 with exactly two blue and two red edges between @xmath166 and @xmath599 ( three or four blue edges is not possible ) . hence @xmath166 is contained in a blue @xmath203 with three vertices from @xmath599 and there are two blue edges pendent at @xmath166 , one to @xmath442 and one within @xmath600 .
moreover @xmath166 is contained in a red @xmath203 with @xmath442 and two vertices from @xmath599 and there are two red edges pendent at @xmath166 within @xmath600 .
this proves claim [ g4_claim2 ] .
let @xmath204 denote the graph obtained from @xmath205 by adding a pendent edge at a vertex of degree @xmath206 , see figure [ fig : k23special ] .
[ lem : nomonokab ] in any @xmath117-edge - coloring of @xmath210 without monochromatic @xmath204 the vertex of degree @xmath117 is incident to exactly one red and one blue edge . let @xmath542 be the vertex of degree @xmath117 in @xmath210 , let @xmath516 denote the vertices of degree @xmath206 and @xmath605 .
then @xmath606 , @xmath607 and there is a complete bipartite graph between @xmath608 and @xmath516 .
consider a @xmath117-edge - coloring of @xmath210 without monochromatic @xmath204 .
we consider the edges between @xmath608 and @xmath516 first .
since @xmath606 , each vertex in @xmath516 is incident to at least @xmath117 red or @xmath117 blue edges by pigeonhole principle .
let @xmath609 denote the set of vertices in @xmath516 incident to at least @xmath117 red edges and @xmath610 those incident to @xmath117 blue edges . without loss of generality assume @xmath611 , thus @xmath612
. for @xmath613 let @xmath614 denote the vertices in @xmath608 adjacent to @xmath166 via red edges .
then @xmath615 .
assume there is a set @xmath616 of size @xmath117 and distinct vertices @xmath617 with @xmath618 , @xmath619 .
then all edges between @xmath620 and @xmath621 are blue , as otherwise there is a red @xmath204 .
but these edges form a blue @xmath622 which contains @xmath204 , a contradiction . hence for each set @xmath616 of size @xmath117
there are most two vertices @xmath166 in @xmath623 with @xmath624 . since @xmath606 and @xmath612 , there are exactly @xmath117 vertices @xmath613 with @xmath625 for each such @xmath620 and thus @xmath626 .
hence @xmath627 and the same arguments applied to @xmath628 and the blue edges show that for each @xmath629 there are exactly @xmath117 vertices in @xmath628 adjacent to @xmath620 with only blue edges too .
now consider the edges incident to @xmath542 .
if there are @xmath117 red edges , then together with the neighbors of @xmath542 and some @xmath206 vertices from @xmath623 there is a red @xmath204 , a contradiction .
the same argument holds for the blue edges and hence there is exactly one red and one blue edge incident to @xmath542 .
[ lem::mindegg5 ] @xmath211 .
consider the graph @xmath17 obtained from a complete bipartite graph on partite sets @xmath630 and @xmath631 by adding @xmath206 new vertices and a matching between these and the vertices in @xmath632 .
see figure [ fig : k23mindegree ] for an illustration .
we construct a graph @xmath49 as follows .
let @xmath197 denote the complete bipartite graph on parts @xmath608 and @xmath516 with @xmath606 and @xmath607 .
for each @xmath129 , @xmath633 , take a copy @xmath634 of @xmath197 and identify @xmath635 with two vertices in the smaller part of @xmath634 .
an illustration is given in figure [ fig : k23mindegree ] . which is ramsey for @xmath204 .
] from now on @xmath636 and @xmath531 refer to the identified vertices .
we prove @xmath637 next .
consider a @xmath117-edge - coloring of @xmath49 .
then each @xmath638 together with each @xmath634 , @xmath619 , induces a copy of @xmath210 .
hence there is a monochromatic copy of @xmath204 by lemma [ lem : nomonokab ] , if there are @xmath636 and @xmath639 such that both edges between @xmath639 and @xmath635 are of the same color , @xmath640 .
so assume that the edges between @xmath639 and @xmath635 are of different colors , @xmath641 .
by pigeonhole principle we may assume that there are @xmath117 vertices in @xmath632 , say @xmath642 , @xmath643 , such that the edge between @xmath639 and @xmath531 is red , @xmath644 .
then the edges @xmath645 are blue for @xmath644 .
thus there is a blue @xmath205 between @xmath646 and @xmath647 .
hence the edges @xmath648 are red , or otherwise there is a blue @xmath204 .
thus there is a red @xmath205 between @xmath646 and @xmath649 , see figure [ fig : k23mindegree_colored ] .
altogether there is a monochromatic copy of @xmath204 no matter which color is assigned to the edge pendent at @xmath642 . thus @xmath637 . in a @xmath117-edge - coloring of @xmath49 with the same degree @xmath206 vertices . ]
let @xmath148 denote the graph obtained from @xmath49 by removing the vertices of degree @xmath561 .
it remains to show that @xmath650 .
then every minimal ramsey graph of @xmath204 in @xmath49 contains at least one of the degree @xmath561 vertices and hence @xmath211 .
consider the following coloring of @xmath148 .
color all edges between @xmath646 and @xmath651 and all edges between @xmath652 and @xmath653 red .
color all other edges between @xmath632 and @xmath654 blue .
finally color the edges of each @xmath634 , @xmath619 , without monochromatic copy of @xmath205 .
such a coloring exists since @xmath655 @xcite .
next we show that there is no monochromatic @xmath204 under this coloring . each copy of @xmath205 in @xmath148 which does not contain any vertex of @xmath632
is contained in ( exactly ) one of the @xmath634 and hence is not monochromatic .
moreover each copy of @xmath205 in @xmath148 which contains a vertex from @xmath632 and a vertex from @xmath656 for some @xmath129 , contains @xmath531 and @xmath636 and hence a red and a blue edge .
thus the only monochromatic copies of @xmath205 in @xmath148 are contained in @xmath17 .
the part on @xmath117 vertices of all monochromatic copies of @xmath205 in @xmath17 is contained in @xmath632 .
hence there is no monochromatic copy of @xmath204 because each @xmath657 is incident to exactly @xmath206 red and @xmath206 blue edges .
the wheel @xmath658 is the graph on @xmath2 vertices obtained from a cycle @xmath203 of length @xmath152 by adding a new vertex adjacent to all vertices of the cycle .
[ lem : w4isolated ] if @xmath0 is connected and has at least @xmath212 vertices , then @xmath213 .
we assume @xmath659 due to lemma [ lem::ramseygraphsamecliqueno ] and @xmath660 due to observation [ obs::bipartite ] .
let @xmath661 and let @xmath49 be a graph with @xmath662 and @xmath663 , which exists by lemma [ lem::epsramsey ] .
we construct a graph @xmath17 by taking the vertex disjoint union of @xmath49 and a copy @xmath594 of @xmath461 and placing a complete bipartite graph between @xmath49 and @xmath594 .
we shall show that @xmath664 , but @xmath21 .
color all edges within @xmath49 and within @xmath594 red and all other edges blue . since @xmath665 , @xmath342 . since @xmath666 , @xmath667 .
since @xmath0 is connected there is no red copy of @xmath0 .
the blue subgraph is a complete bipartite graph and @xmath668 .
thus there is no blue copy of @xmath0 .
it remains to show that @xmath669 .
consider a @xmath117-edge - coloring of @xmath17 . by the focusing lemma ( lemma [ lem::focusing ] )
there is a set @xmath670 of @xmath671 vertices in @xmath49 such that between @xmath670 and each vertex in @xmath594 all edges are of the same color . since @xmath662 there is a monochromatic copy @xmath197 of @xmath203 in @xmath672 $ ] .
assume without loss of generality that @xmath197 is blue .
if there is a vertex in @xmath594 which sends a blue star to @xmath197 then there is a blue @xmath658 and we are done .
so assume all vertices in @xmath594 send red stars to @xmath197 .
if there is no blue copy of @xmath658 in @xmath594 , then there are two adjacent red edges @xmath673 and @xmath589 in @xmath594 ( since the complement of @xmath658 in @xmath461 is a maximum matching ) .
then @xmath673 , @xmath589 and any two vertices from @xmath197 form a red copy of @xmath658 , with the vertex of degree @xmath152 being the common vertex of @xmath673 and @xmath589 , see figure [ fig::w4gamma ] .
hence @xmath669 . in some @xmath117-edge - coloring of @xmath17 . ]
[ lem : k23isolated ] if @xmath0 is connected and has at least @xmath212 vertices , then @xmath214 .
we assume that @xmath0 is bipartite and contains a cycle by observation [ obs::bipartite ] and lemma [ lem::extremalno ] , respectively . since @xmath674 @xcite we assume @xmath675 , since otherwise @xmath214 .
hence one of the partite sets of @xmath0 contains at most @xmath117 vertices , since otherwise coloring the edges of @xmath676 with a red copy of @xmath677 and an edge disjoint blue copy of @xmath678 does not yield a monochromatic copy of @xmath0 .
so @xmath0 is @xmath679 with some edges pendent at the part of size @xmath117 for some @xmath680 with @xmath681 .
suppose @xmath0 has at least @xmath682 vertices .
we claim @xmath683 .
indeed , color the edges of a @xmath684 in @xmath460 red and all other edges blue .
then the blue edges form vertex disjoint copies of @xmath461 and do not contain @xmath0 since @xmath0 is connected .
the red subgraph does not contain @xmath0 because one of the bipartition classes of @xmath0 has at least @xmath212 vertices ( as the other has only @xmath117 ) .
hence @xmath214 because @xmath685 @xcite .
. then @xmath0 is isomorphic to one of the graphs @xmath687 given in figure [ fig : k23special ] or a supergraph of @xmath204 .
note that @xmath204 contains @xmath205 .
we have @xmath688 , @xmath689 by @xcite and @xmath690 , @xmath691 by @xcite .
thus @xmath214 if @xmath0 is isomorphic to one of the graphs @xmath687 , @xmath619 , because @xmath685 .
moreover @xmath214 if @xmath0 is isomorphic to @xmath202 or @xmath204 because @xmath692 by lemma [ lem : mindegg4 ] respectively lemma [ lem::mindegg5 ] , but @xmath693 .
so assume @xmath0 is a supergraph of @xmath204 .
let @xmath49 denote a minimal ramsey graph of @xmath204 with @xmath694 and obtain a graph @xmath148 by removing a vertex of degree @xmath561 from @xmath49 .
then @xmath695 because @xmath696 and @xmath697 ( since @xmath49 is not minimal ramsey for @xmath205 ) .
but @xmath650 , thus @xmath698 , and hence @xmath214 .
[ lem::h52p4e ] @xmath481 .
let @xmath442 denote the vertex of degree @xmath117 in @xmath699 and let let @xmath700 denote the edge incident to both neighbors of @xmath442 .
let @xmath542 , @xmath701 , @xmath702 denote the other vertices .
assume there is a @xmath117-edge - coloring of @xmath699 without monochromatic @xmath446 .
without loss of generality assume the edge @xmath518 is blue . if ( * case 1 * ) @xmath673 and two more edges @xmath703 , @xmath704 incident to @xmath166 are red , then either there is a red @xmath446 containing these edges or @xmath705 , @xmath706 , @xmath707 and @xmath708 are blue .
but then these form a blue copy of @xmath446 .
if ( * case 2 * ) @xmath673 and at most one other edge incident to @xmath166 is red , then assume @xmath703 , @xmath704 are blue . then the edges @xmath709 , @xmath710 , @xmath705 , @xmath706 and @xmath708 must be red , but these contain a copy of @xmath446 .
if ( * case 3 * ) @xmath673 and @xmath703 are blue , then @xmath710 , @xmath711 , @xmath712 and @xmath713 must be red .
but then the edges @xmath704 , @xmath714 and @xmath708 must be blue , but contain a copy of @xmath446 .
if ( * case 4 * ) @xmath673 is blue but all the edges @xmath703 , @xmath704 , @xmath714 are red , then @xmath709 , @xmath710 , @xmath711 are blue . but together with @xmath518 and @xmath529 this is a blue @xmath446 .
assume there is @xmath117-edge - coloring of @xmath716 without monochromatic @xmath454 .
let @xmath442 denote the vertex of degree @xmath206 in @xmath716 , let @xmath542 , @xmath701 , @xmath702 denote its neighbors and @xmath166 , @xmath531 the remaining vertices .
there are two edges of the same color incident to @xmath442 , assume @xmath443 , @xmath444 are red .
since there is no monochromatic @xmath205 ( it contains @xmath454 ) there is at least one edge from @xmath717 to @xmath718 in red .
if ( * case 1 * ) there are @xmath719 such that the edges @xmath720 , @xmath721 are red , then @xmath722 . then all edges from @xmath717 to @xmath723 are blue .
but then the edge from @xmath724 to @xmath723 can be neither red nor blue .
if ( * case 2 * ) , without loss of generality , @xmath542 has only blue edges to @xmath718 , then @xmath725 is red for some @xmath726 and all edges from @xmath347 to @xmath727 are blue .
this yields a blue @xmath203 on @xmath728 . since all edges which are incident to this @xmath203 ( but not contained ) are red we can find a red @xmath454 .
consider a @xmath117-edge - coloring of the edges of @xmath684 and vertex @xmath166 .
let @xmath670 denote the partite set of @xmath684 containing @xmath166 .
then @xmath166 is incident to three edges @xmath703 , @xmath704 , @xmath714 of the same color , say red .
from each of the four vertices in @xmath730 at most one edge to @xmath447 is red , otherwise there is a red @xmath731 .
but then there are two vertices in @xmath447 and two vertices in @xmath730 forming a blue @xmath731 by pigeonhole principle .
consider a @xmath117-edge - coloring of @xmath431 on vertices @xmath732 without monochromatic @xmath64 .
we may assume that @xmath733 forms a red @xmath455 since @xmath734 .
clearly there are no two independent red edges between @xmath594 and @xmath735 .
if there is a vertex in @xmath594 incident to at least two red edges to @xmath736 , then all edges between the two other vertices in @xmath594 and @xmath736 are blue .
thus these blue edges form a blue @xmath205 , and hence no edge in @xmath736 is blue .
but if all edges in @xmath736 are red , then there is a red @xmath455 with two vertices in @xmath736 and one vertex in @xmath594 and a pending red edge in @xmath594 and a pending in @xmath736 .
so we may assume that at most one vertex in @xmath736 is adjacent to @xmath594 in red , say @xmath702 .
then @xmath717 and @xmath594 form a blue @xmath205 .
this shows that @xmath712 is red .
we consider the cases how many edges between @xmath737 and @xmath702 are red .
if ( * case 1 * ) @xmath712 is the only red edge , then @xmath717 and @xmath738 induce a blue @xmath739 and any additional blue edge within this @xmath739 yields a blue @xmath64 .
hence the red edges form a @xmath604 plus disjoint @xmath740 , which corresponds to the rightmost coloring of figure [ fig::k6g1 ] . if ( * case 2 * ) there are at least two red edges , then @xmath702 is not part of any red @xmath455 on @xmath738 , since this @xmath455 would have a red pending edge in @xmath594 and another one in @xmath736 .
hence there is at most one red edge from @xmath702 to @xmath594 .
thus there is a blue copy of @xmath741 between @xmath594 and @xmath736 .
this shows that no edge in @xmath736 is blue and the coloring corresponds to the left or the middle coloring in figure [ fig::k6g1 ] .
assume @xmath554 is a @xmath117-coloring of the edges of @xmath17 , labeled like in figure [ fig::graphgamma ] , without monochromatic @xmath64 .
let @xmath594 denote the copy of @xmath431 in @xmath17 .
due to lemma [ lem::k6g1 ] , @xmath554 restricted to @xmath594 is isomorphic to one of three colorings of @xmath431 given in figure [ fig::k6g1 ] .
we will distinguish cases based on the coloring of @xmath594 under @xmath554 .
( * case 1 * : ) the red subgraph of @xmath594 under @xmath554 consists of two disjoint @xmath455 and the blue edges in @xmath594 form a copy of @xmath743 .
if one of these blue edges is contained in a blue @xmath455 with a vertex from @xmath447 , then there is blue @xmath64 . due to the construction of @xmath17 we can find three vertex disjoint copies of @xmath455 each with exactly one vertex from each of the red @xmath455 in @xmath594 and exactly one vertex from @xmath447 . since there is a red edge from @xmath594 to @xmath447 in each of these
, one of the red @xmath455 in @xmath594 has two independent pending red edges .
this gives a red @xmath64 , a contradiction .
( * case 2 * : ) the red subgraph of @xmath594 under @xmath554 consists of two disjoint @xmath455 connected by a single edge @xmath673 .
then all edges from @xmath594 to @xmath447 are blue if not adjacent to @xmath673 .
then there are two vertices in @xmath594 , not incident to @xmath673 , each having two blue edges to @xmath447 but only one common neighbor in @xmath447 .
since they are connected by an blue edge in @xmath594 , this gives a blue @xmath64 , a contradiction .
( * case 3 * : ) the red subgraph of @xmath431 consists of @xmath604 and a single edge @xmath673 .
then all edges from this @xmath604 to @xmath447 are blue .
the blue edges in @xmath594 form @xmath739 . again ,
if one of these blue edges in @xmath594 forms a blue triangle with a vertex from @xmath447 , then there is a blue @xmath64 .
thus all edges from @xmath673 to @xmath447 are red .
if @xmath744 , then @xmath673 together with @xmath447 forms a red copy of @xmath64 .
so assume @xmath745 .
if the edge @xmath712 is blue , then @xmath746 gives a blue @xmath64 .
if it is red , then @xmath747 gives a red @xmath64 , a contradiction .
if all edges from @xmath594 to @xmath736 are blue , then no edge among @xmath736 is blue .
thus the coloring corresponds to the left one in figure [ fig::k6g3 ] .
so assume the edge @xmath445 is red .
if @xmath738 forms a red @xmath604 , then all edges from this @xmath604 to @xmath717 are blue .
no matter which color is assigned to @xmath712 , the coloring has no monochromatic @xmath66 and corresponds to the middle or right color in figure [ fig::k6g3 ] . if ( * case 1 * ) @xmath709 is blue , then @xmath749 is a blue @xmath455 with pending blue edge @xmath748 .
thus @xmath444 , @xmath704 are red .
but then @xmath710 needs to be blue ( otherwise @xmath750 is red @xmath455 with pending red path @xmath751 ) and there is a blue @xmath604 .
thus the coloring corresponds to the middle or right coloring of figure [ fig::k6g3 ] with switched colors , as argued above . if ( * case 2 * ) @xmath709 is red , then @xmath712 is blue and @xmath703 , @xmath714 are blue
. then @xmath704 needs to be red , since @xmath752 is a blue @xmath455 with pending blue path @xmath753 otherwise .
then @xmath710 is blue , since otherwise @xmath750 is a red @xmath455 with pending red path @xmath751 otherwise .
but now @xmath754 is a blue @xmath455 with pending blue path @xmath755 , a contradiction .
assume @xmath554 is a @xmath117-coloring of the edges of @xmath17 , labeled like in figure [ fig::graphgamma ] , without monochromatic @xmath66 .
let @xmath594 denote the copy of @xmath431 in @xmath17 .
due to lemma [ lem::k6g3 ] , @xmath554 restricted to @xmath594 is isomorphic to one of three colorings of @xmath431 given in figure [ fig::k6g3 ] .
we will distinguish cases based on the coloring of @xmath594 under @xmath554 .
( * case 1 * : ) the red subgraph of @xmath431 consists of two disjoint @xmath455 . then the blue edges in @xmath594 form a copy of @xmath743 .
if one of these blue edges is contained in a blue @xmath455 with a vertex from @xmath447 , then there is blue @xmath66 . on the other hand
no vertex in @xmath735 sends a red edge to each of the red @xmath455s in @xmath594 .
since there are @xmath152 edges from each vertex in @xmath447 to @xmath594 , each is incident to at least one red edge and one blue edge .
if one of the edges induced by @xmath447 is red , then there is a red @xmath66 with a red @xmath455 from @xmath594 and an edge between them .
so @xmath447 induces a blue @xmath455 which forms a blue @xmath66 with an edge to @xmath594 and another contained in @xmath594 .
( * case 2 * : ) the red subgraph of @xmath431 consists of a red @xmath604 only .
then no edge incident to this @xmath604 is red .
let @xmath84 , @xmath85 denote the vertices in @xmath594 not contained in the red @xmath604 .
then every blue edge between @xmath447 and the red @xmath604 is part of a blue @xmath66 together with @xmath84 , @xmath85 , and another vertex in @xmath594 .
( * case 3 * : ) the red subgraph of @xmath431 consists of a red @xmath604 and a disjoint red edge @xmath673 .
again all edges incident to the red @xmath604 are blue and no blue edge in @xmath594 is contained in a blue @xmath455 with a vertex from @xmath736 .
thus all edges from @xmath673 to @xmath736 are red .
assume @xmath744 , say it is @xmath757 .
if @xmath712 is blue then @xmath758 forms a blue @xmath66 .
if @xmath712 is red then @xmath759 gives a red @xmath66 .
so assume @xmath745 .
if the edge @xmath712 is blue , then @xmath758 forms a blue @xmath66 .
if it is red , then @xmath747 gives a red @xmath66 . on the other hand , consider an arbitrary @xmath117-edge - coloring of @xmath761 without a monochromatic copy of @xmath762 for a fixed @xmath760 .
there is a monochromatic copy @xmath594 of @xmath763 , say in red ( i.e. a red copy of @xmath455 with a pending edge ) , since @xmath764 .
suppose there is no monochromatic @xmath64 .
then none of the two vertices of degree @xmath117 in @xmath594 is incident to another red edge in @xmath761 .
thus there is a blue copy of @xmath739 .
then the part with four vertices in this @xmath739 contains no further blue edge and induces a red @xmath604 .
but then no edge incident to this @xmath604 is red , and there is a blue copy of @xmath765 . since no other edge might be blue then , there are two disjoint red @xmath604 .
suppose there is no monochromatic @xmath65 .
the vertex of degree @xmath206 in @xmath594 has no other incident red edge .
so it is the center of a blue @xmath766 .
the degree @xmath561 vertices in this copy of @xmath766 do not induce a blue edge , so they induce a red @xmath604 .
but then no edge incident to this @xmath604 is red and there is a blue @xmath765 between @xmath594 and the other vertices . as argued above the red edges form two disjoint @xmath604s and the blue edges form @xmath765 .
suppose there is no monochromatic @xmath66 .
let @xmath736 denote the set of vertices not in @xmath594 .
then any edge connecting the vertex @xmath166 of degree @xmath561 in @xmath594 to a vertex in @xmath736 is blue .
assume there is a red edge from @xmath594 to a vertex @xmath767 .
then any edge connecting @xmath442 to a vertex in @xmath768 is blue .
then each edge @xmath673 within @xmath768 or from @xmath768 to @xmath769 is red , since otherwise there is a blue @xmath66 spanned by @xmath442 , @xmath166 and @xmath673 .
but then there is red @xmath66 , a contradiction .
assume there is a @xmath117-edge - coloring of @xmath771 without monochromatic @xmath762 for some @xmath760 .
we may assume that within the copy of @xmath761 the red edges form two disjoint @xmath604 with a blue @xmath765 in - between by lemma [ lem::k8g123 ] .
let @xmath166 denote the vertex of degree @xmath2 .
then @xmath166 has only blue incident edges since every neighbor of @xmath166 is part of a red @xmath604 .
but @xmath166 has a neighbor in each of the red @xmath604 s .
thus @xmath166 together with these two vertices forms a blue @xmath455 which is contained in a blue copy of @xmath762 for all @xmath760 , a contradiction .
assume there is a @xmath117-edge - coloring of @xmath77 without monochromatic @xmath64 . due to lemma [ lem::k7g12 ]
we may assume that the copy of @xmath476 in @xmath77 is colored such that the blue edges induce a copy of @xmath772 and red consists of two disjoint copies of @xmath604 and @xmath455 .
let @xmath594 denote the red @xmath455 and @xmath442 , @xmath166 the two vertices of degree @xmath2 in @xmath77 .
then each edge from @xmath773 to the red @xmath604 is blue and there are at least two such edges incident to each of @xmath442 , @xmath166 .
thus each edge from @xmath442 or @xmath166 to @xmath594 is red , since there is a blue @xmath64 otherwise .
due to construction of @xmath77 there are two independent edges from @xmath594 to @xmath773 and thus a red @xmath64 , a contradiction . [ lem::k8g4 ]
a @xmath117-edge - coloring of @xmath761 does not have monochromatic @xmath498 , if and only if one of the color classes induces two disjoint copies of @xmath604 with at most one edge of same color in - between ( i.e. the other color spans @xmath765 or @xmath774 ) .
assume there is a @xmath117-edge - coloring of @xmath775 without monochromatic @xmath498 .
we assume that the coloring of @xmath761 in @xmath775 has two disjoint red @xmath604 connected by at most one red edge according to conjecture [ lem::k8g4 ] .
let @xmath166 denote the vertex of degree @xmath212 .
it is incident to at most one red edge to to each of the red @xmath604 .
thus there is a blue @xmath455 with @xmath166 and one vertex from each red @xmath604 . but this forms a blue @xmath498 together with some of the other blue edges , a contradiction .
consider a @xmath117-coloring of the edges of @xmath776 and vertex @xmath442 .
let @xmath670 denote the partite set of @xmath776 containing @xmath442 .
then @xmath442 is incident to five edges @xmath518 , @xmath708 , @xmath443 , @xmath444 , @xmath445 of the same color , say red . from each of the @xmath472 vertices in @xmath777 at most two of the edges to @xmath778
are red , otherwise there is a red @xmath205 .
this means that there are three blue edges between each of the vertices in @xmath777 and @xmath778 .
there are @xmath424 sets of size @xmath206 in @xmath778 and @xmath472 vertices in @xmath730 .
hence there are two vertices in @xmath730 and three vertices in @xmath778 forming a blue @xmath205 by pigeonhole principle .
first of all observe that @xmath780 does not contain a copy of @xmath781 since every edge is contained in exactly one copy of @xmath455 .
moreover the complement of @xmath780 ( as a subgraph of @xmath591 ) is isomorphic to @xmath780 .
hence the edges of @xmath591 can be @xmath117-colored without monochromatic @xmath781 using two edge disjoint copies of @xmath780 .
on the other hand consider a @xmath117-edge - coloring @xmath554 of @xmath591 without monochromatic @xmath779 .
we will assign labels @xmath782 , @xmath640 , to the vertices of @xmath591 such that this labeling corresponds to an arrangement of the vertices in a @xmath783 grid where the red subgraph spans all rows and columns and all other edges are blue .
there is a monochromatic copy of @xmath499 under @xmath554 , say in red , since @xmath784 , see figure [ fig::5vert ] .
let @xmath785 , @xmath786 , @xmath787 , @xmath788 , @xmath789 denote the vertices of this @xmath499 such that @xmath790 is the vertex of degree @xmath152 and the edges of this red @xmath499 span the first row and first column in the grid , see figure [ fig::k9h32 ] . observe that no edge spanned by @xmath594 is red except for the edges in the red @xmath499 . indeed
if another edge is red , then there is a red @xmath781 in @xmath594 .
let @xmath736 denote the vertices not in @xmath594 .
we will use the following claim .
[ claim-1 ] if @xmath197 is a red copy of @xmath455 and @xmath518 is a vertex disjoint blue edge , then there is a vertex @xmath542 in @xmath197 such that @xmath791 and @xmath792 are blue , and there are two independent red and two independent blue edges between @xmath793 and @xmath518
. see figure [ fig::monok3edge ] for an illustration .
indeed , there is at most one red edge between each vertex in @xmath773 and @xmath197 and for at most one vertex in @xmath197 both edges to @xmath518 are blue . hence for exactly one vertex in @xmath197 both edges to @xmath518 are blue and there are exactly two further independent blue edges between @xmath197 and @xmath518 .
this proves claim [ claim-1 ] . by assumption the four vertices in @xmath736
do not induce a monochromatic @xmath781 and hence there are at least two blue edges @xmath673 , @xmath589 . let @xmath546 , @xmath547 denote the red copies of @xmath455 in @xmath594
. we will apply claim [ claim-1 ] to each of the pairs @xmath794 , @xmath795 , @xmath796 , @xmath797 .
there is a vertex @xmath798 in @xmath634 , @xmath309 , such that both edges between @xmath798 and a blue edge in @xmath736 are blue by claim [ claim-1 ] .
then @xmath799 , since otherwise there is blue copy of @xmath781 .
hence the blue edges in @xmath736 are independent , since two adjacent blue edges together with @xmath790 form a blue @xmath781 .
thus @xmath673 and @xmath589 are the only blue edges in @xmath736 .
see figure [ fig::k9h32 ] for the partial labeling .
furthermore there are two independent red edges and two independent blue edges from each of the edges @xmath673 and @xmath589 to each @xmath800 , @xmath309 , by claim [ claim-1 ] .
it remains to find labels for the vertices in @xmath673 and @xmath589 .
indeed , assume there are two such vertices @xmath803 in @xmath736 for some pair @xmath442,@xmath166 .
then the edge @xmath804 is red by claim [ claim-1 ] and there is a red @xmath781 .
thus there is at most one such vertex .
assume there is no such vertex in @xmath736 for some pair . then there is a red @xmath781 , since there are two independent red edges between each of @xmath673 and @xmath589 and each @xmath634 , @xmath309 , by claim [ claim-1 ] , a contradiction
this proves claim [ claim-2 ] .
let @xmath805 denote the vertex which is adjacent to @xmath786 and @xmath788 in red which exists by claim [ claim-2 ] . without loss of generality
assume @xmath805 is incident to @xmath673 .
let @xmath806 denote the other vertex incident to @xmath673 .
due to claim [ claim-1 ] applied to @xmath673 and @xmath546 and @xmath547 , the edges @xmath807 and @xmath808 are red and the edges @xmath809 , @xmath810 , @xmath811 and @xmath812 are blue . with the same arguments we choose @xmath813 accordingly .
this shows that the red color class is isomorphic to @xmath780 .
either there is a monochromatic copy of @xmath781 in the copy of @xmath591 in @xmath816 or we may assume by lemma [ lem::k9h32 ] that the @xmath591 in @xmath816 is an edge disjoint union of a red @xmath818 and a blue @xmath818 .
each edge in @xmath591 belongs to a unique monochromatic triangle . if @xmath166 sends two blue edges to vertices @xmath819 , where @xmath708 is blue , then the blue triangle containing @xmath708 in @xmath591 together with @xmath166 form a blue @xmath781 .
thus , we may assume that neighborhood of @xmath166 via blue edges forms a red clique , and , similarly , its neighborhood via red edges forms a red clique .
since degree of @xmath166 is @xmath212 , and the largest monochromatic clique in @xmath591 is a triangle , these cliques must be triangles .
however , there are no two disjoint red and blue triangles in @xmath591 , so we arrive at a contradiction .
thus , there is a monochromatic @xmath781 .
assume that the monochromatic copy @xmath594 of @xmath781 is red .
first , we assume that @xmath594 does not contain @xmath166 .
let @xmath736 denote the set of vertices from @xmath591 that are not in @xmath594 .
if there is a red edge between a vertex of degree @xmath206 of @xmath594 and @xmath736 , we have a monochromatic @xmath820 .
if there is a red edge between a vertex of degree @xmath117 of @xmath594 and @xmath736 , we have a monochromatic @xmath821 .
if all edges between degree @xmath206 vertices of @xmath594 and @xmath736 are blue and there are two adjacent blue edges in @xmath736 , then there is a blue copy of @xmath820 .
if all edges between degree @xmath206 vertices of @xmath594 and @xmath736 are blue and there are no two adjacent blue edges in @xmath736 , then @xmath736 forms a red @xmath461 minus a matching , and thus contains a copy of @xmath820 . if all edges between degree @xmath117 vertices of @xmath594 and @xmath736 are blue , then there is a blue copy of @xmath821 or there is no blue edge induced by @xmath736 , in the latter case @xmath822 induces a red @xmath461 that contains a red copy of @xmath821 .
now , assume that any monochromatic @xmath781 contains @xmath166 , i.e. , there is no monochromatic @xmath781 in a copy of @xmath591 of @xmath816 .
hence the coloring of @xmath591 is like it is described in lemma [ lem::k9h32 ] .
then , it is easy to see that @xmath594 and an appropriate edge of @xmath591 form a monochromatic copy of @xmath820 and , similarly , a monochromatic copy of @xmath821 . | given a graph @xmath0 , a graph @xmath1 is called a _ ramsey graph of @xmath0 _ if there is a monochromatic copy of @xmath0 in every coloring of the edges of @xmath1 with two colors .
two graphs @xmath1 , @xmath0 are called _
ramsey equivalent _ if they have the same set of ramsey graphs .
fox _
et al . _
[ _ j .
combin .
theory ser .
b _ * 109 * ( 2014 ) , 120133 ] asked whether there are two non - isomorphic connected graphs that are ramsey equivalent .
they proved that a clique is not ramsey equivalent to any other connected graph .
results of neetil _ et al . _ showed that any two graphs with different clique number [ _ combinatorica _ * 1*(2 ) ( 1981 ) , 199202 ] or different odd girth [ _ comment .
math .
univ .
carolin .
_ * 20*(3 ) ( 1979 ) , 565582 ] are not ramsey equivalent .
these are the only structural graph parameters we know that `` distinguish '' two graphs in the above sense .
this paper provides further supportive evidence for a negative answer to the question of fox _
et al . _ by claiming that for wide classes of graphs , chromatic number is a distinguishing parameter .
in addition , it is shown here that all stars and paths and all connected graphs on at most @xmath2 vertices are not ramsey equivalent to any other connected graph .
moreover two connected graphs are not ramsey equivalent if they belong to a special class of trees or to classes of graphs with clique - reduction properties .
* keywords : * ramsey , ramsey equivalence , chromatic number , ramsey classes |
the spectral energy distribution ( sed ) of the gravitational power released as electromagnetic radiation when matter accretes onto a compact object is far from universal .
different accretion modes are possible , and often the same initial conditions at the outer boundaries admit more than one solution for the accretion flow configuration at the inner boundary , with often very different radiative properties .
the main goal of accretion flows theory is to distinguish and understand all the possible different modes of accretion , and classify the observed sources in terms of such modes . in the case of
non - quiescent galactic ( stellar mass ) black holes candidates ( gbhc ) , an x - ray based classification in terms of their _ spectral states _ has emerged in recent years .
broadly speaking , the different states are distinguished by their broadband luminosity , and by the relative contribution to the luminosity of a ` soft ' , quasi - thermal and a ` hard ' , power - law like spectral components ( see e.g. the recent review by done 2001 and reference therein ) . at luminosities close to the eddington
one , these sources are often in the _ very high state _ , where both of the two components contribute substantially to the sed . at slightly lower luminosities
the quasi - thermal one dominates , and the power - law is usually steeper and extended to the gamma - ray band ( _ high / soft state _ ) . at even lower luminosities the spectra
are completely dominated by a hard power - law component , with the quasi - thermal one extremely weak or even absent : these are the so - called _ low / hard states_. sometimes , at luminosities intermediate between those of the soft and the hard states , an _ intermediate state _ is observed , with properties similar to those of the very high state .
it is often the case that the same source , either persistent or transient , undergoes a transition between spectral states , and therefore accretion modes . from the theoretical point of view
, the quasi - thermal component is usually associated with a geometrically thin , optically thick accretion disc .
its dynamical and geometrical properties could in principle be deduced solely by a careful comparison between the observed spectral properties of the quasi - thermal hump and a detailed model for the accretion disc . on the other hand ,
the observed hard x - ray power - laws represent a universal signature of a specific physical process ( most likely inverse compton scattering of soft photons on a population of hot electrons ; sunyaev & titarchuk 1980 ) more than of specific accretion dynamics .
this is the reason why , if there is little doubt that the standard accretion disc model ( shakura & sunyaev 1973 ) captures the basic physical properties of black holes in their high / soft state , the accretion mode responsible for the low / hard state is still matter of debate .
recently , multi - wavelength observations of gbhc in the low / hard state have been adding important pieces of information , which are helping to shed more light on the subject .
crucial in this respect is the discovery of the high - galactic latitude transient source xte j1118 + 480 .
this source was persistently in the low / hard state during its 2000 outburst , and being very little affected by galactic extinction , has provided the most accurate and complete sed to date of a galactic black hole in this state @xcite .
it possesses all the typical characteristics that distinguish such state @xcite , namely : a hard x - ray power - law with photon index @xmath0 and a high energy cut - off , indicative of thermal comptonization ; a dim and cold thermal disc component peaking in the euv band @xcite ; excess red and infrared emission compared the one expected from standard disc models ; a compact ( unresolved ) flat ( or inverted ) radio core , probably extending to the nir or optical regime @xcite .
the situation for supermassive black holes accreting in the centre of active galactic nuclei bears some resemblance with the gbhc one . beside the
best studied case of the black hole at the centre of our galaxy ( melia & falcke 2001 , and references therein ) , recent multi - wavelength campaigns on nearby low - luminosity agns ( llagn ) ( ho 1999 , nagar et al . 2000 , falcke et al .
2000 , ho et al . 2001 ) have revealed nuclear seds markedly different from the canonical broadband continuum spectrum of high luminosity agn ( as compiled , e. g. , by elvis et al .
the most striking difference is the lack of uv excess , the so - called ` big blue bump ' associated with the emission from an optically thick , geometrically thin accretion disc @xcite .
in addition , vla @xcite and vlba @xcite radio observations of llagn have revealed flat or inverted compact radio cores , with an incidence higher than in the case of normal seyfert galaxies , and , for the sources observed at higher spatial resolution with vlba , very high brightness temperatures ( @xmath1k ) . in light of the above observational similarities between the two classes
, we would like to discuss here a possible answer to the following question : what is the common accretion mode for low - luminosity black holes ?
since their rediscovery in recent years ( narayan & yi 1994 ) , radiatively inefficient , optically thin , advection dominated accretion flows ( adaf ; ichimaru 1977 , rees et al .
1982 , abramowicz et al . 1995
; see narayan , mahadevan & quataert 1999 for a recent review ) have been regarded as natural solutions .
they can explain the absence of the quasi - thermal hump , the low x - ray luminosities and the observed x - ray spectral slopes . when tested against the best data for gbhc in their low / hard state , though , as in the case of xte j1118 + 480 @xcite or cygnus x-1 @xcite , adaf models alone can not work . a transition between an inner adaf and an outer shakura sunyaev disc
is needed , with @xmath2 , as can also be inferred from studies of x - ray reflection components ( gilfanov , churazov & revnivtsev 2000 ; done 2001 , and reference therein ) .
detailed spectral fits to the broadband sed of the nearby llagns m81 and ngc 4579 ( quataert et al .
1999 ) lead to similar conclusions .
the physics of such a transition , which may be taking place through a gradual evaporation of the disc as proposed by meyer & meyer - hofmeister ( 1994 ) or via turbulent diffusive heat transport in the radial direction @xcite , is not included in the adaf models , so that @xmath3 is usually treated as a free parameter in the spectral fit procedure .
furthermore , the low energy part of the sed is also difficult to reconcile with the standard adaf picture , that in general predicts highly inverted spectra , instead of the observed flat ones .
in fact , morphology and spectral indices of the compact radio cores require additional outflows from the central inner part in order to be consistent with the prediction of an adiabatic flow @xcite .
finally , we must note that , both from the theoretical point of view @xcite and from numerical simulation @xcite , it is clear that radiatively inefficient flows are subject to strong convective instabilities ( and might therefore be better called cdaf ; stone , pringle & begelman 1999 , quataert & gruzinov 2000a ) , but are not in general able to generate strong outflows ( although , see misra & taam 2001 ) , unless the viscosity is very high .
flat or slightly inverted radio spectra from very compact sources are generally associated with jet models @xcite .
this has prompted the alternative view that spectra of llagn are essentially jet - dominated @xcite . as a matter of fact ,
in low - luminosity nearby radio galaxies observed with hst @xcite compact optical cores seem very common .
chiaberge et al . ( 1999 ) , analysing hst images of a complete sample of 33 fr i radio galaxies , demonstrated that the luminosity of compact optical cores in these sources correlates linearly with that of the radio cores , suggesting a common non - thermal synchrotron emission origin for both .
in particular , detailed spectral jet models have been demonstrated to successfully reproduce the sed of sgr a@xmath4 ( falcke & markoff 2000 ) .
interestingly , this might also be the case for the x - ray transient xte j1118 + 480 , @xcite .
the presence of a relativistic jet in the black hole candidate cygnus x-1 in the low / hard state has been recently confirmed by milliarcsecond resolution vlba observations @xcite , while fender ( 2001 ) , on the basis of the observational evidence of strong radio compact jet correlating with hard x - ray emission in gbhc in the low / hard state , concluded that , in these sources , the jet power is at least comparable with the accretion one . here
we explore further this latter hypothesis from the point of view of the standard accretion disc theory .
we shall propose a possible new model for low - luminosity black holes , in which a standard geometrically thin , optically thick disc at low accretion rates dissipates a large fraction of its gravitational energy in a magnetic corona .
we shall outline the reasons why energetically dominant coronae are the ideal site for launching powerful jets / outflows , and discuss the expected spectral properties of such accretion flows .
the important conclusion we reach is that if the jet / outflow is radiatively inefficient , then so is the source overall .
it is well known that micro - physical viscosity is in general too inefficient to drive accretion at rates comparable to the observed ones , and enhanced / turbulent viscosity is therefore a theoretical necessity .
as astrophysical discs are bound to be seeded with magnetic fields , magneto - rotational instability ( mri ) @xcite is the best candidate to date to produce the self - sustained turbulence that is able to transport angular momentum in rotationally supported discs .
any weak , sub - thermal magnetic field , coupled with outwardly decreasing differential rotation ( two conditions easily met by any accretion flow endowed with angular momentum ) generate the turbulence which in turn is able to regenerate the field @xcite .
this instability , which has a rapid growth rate of the order of the orbital frequency @xmath5 , results in a greatly enhanced effective viscosity that is able to transport angular momentum outward ( balbus & hawley 1998 ) .
what is the fate of the magnetic field which is amplified inside the disc ?
two competing mechanisms should operate in an accretion disc to saturate the field : dissipation and buoyancy . on one hand , part of the field _
must _ be dissipated , if the turbulence has to be sustained and mri relied upon to transport angular momentum .
in fact it has been shown @xcite that , in a steady state disc , if angular momentum transport is due solely to magneto - rotational instability , magnetic fields must be sub - thermal .
this in turn implies that buoyancy can not be a too efficient mechanism to get rid of the disc field : to cross one disc scaleheight in less than a dynamical time @xmath6 a velocity of the order of the sound speed is required , which would correspond to equipartition magnetic field .
such strong fields would suppress the mri @xcite .
on the other hand , we can expect the magnetic field configuration in the disc to be highly intermittent @xcite , with the field concentrated in small regions ( flux tubes and ropes ) with relatively low filling factor ( see e.g. blackman 1996 and the simulations presented in machida , hayashi & matsumoto 2000 ) .
these structures can more easily rise buoyantly and emerge in the low density part of the flow above the accretion disc to form a magnetic atmosphere , the corona , where the field is finally dissipated .
it is not clear at present what is the relative importance of the two mechanisms of field saturation inside a standard accretion disc , if they either coexist or instead alternate in a cyclic fashion ( as , for example , in the sequence : low disc field @xmath7 mri @xmath7 stronger field @xmath7 buoyancy @xmath7 field expulsion @xmath7 suppression of the mri @xmath7 low disc field @xmath7 mri ; see also the discussion in tout & pringle , 1992 ) .
numerical simulations of vertically stratified discs @xcite seem to suggest that , although dissipation is the main saturation mechanism for the instability , buoyancy is indeed capable of transporting vertically a substantial fraction of magnetic energy to power a strong corona ( see in particular miller & stone 2000 , where a larger vertical extent is explored ) , without quenching the magneto - rotational instability in the underlying disc . in this section
we investigate how changes in the internal disc structure , associated with variation in the external accretion rate , may influence the accretion disc corona system . to this end
we only consider standard optically thick and geometrically thin accretion discs sandwiched by a patchy magnetic corona @xcite .
we will assume that buoyant vertical transport of magnetic flux tubes is a relatively efficient way of disposing of the disc field , and discuss the self - consistency of such assumption at the end of the section .
the nature of the saturation mechanism determines the level up to which magnetic fields are amplified inside the disc .
many authors have argued that magnetic field amplification is likely to be limited to values such that the magnetic pressure is at most equal to the local _ gas _ pressure , even in its inner , radiation pressure - dominated parts @xcite . in a recent paper blaes & socrates ( 2001 )
have analysed local dynamical instabilities in magnetized , radiation pressure - supported ( and geometrically thin ) accretion discs , and found that , due to radiative diffusion , the growth rate of the mri is reduced in such systems to @xmath8 , where @xmath9 is the gas sound speed and @xmath10 is the alfvn speed , well below the usual rapid growth rate ( @xmath11 ) in gas pressure dominated discs .
we can therefore give an order of magnitude estimate of the amplitude of the field in the radiation pressure dominated part of a standard accretion disc .
if we assume that the buoyant vertical speed of a magnetic flux tube is approximately given by the alfvn speed , we get for the buoyant timescale @xmath12 , @xmath13 is the isothermal sound speed .
the magnetic field , with initial ( subthermal ) amplitude @xmath14 , rapidly grows to @xmath15 .
therefore we have @xmath16 .
neglecting the logarithmic dependence on the initial field , this in turn suggests that the field will saturate at an amplitude such that @xmath17 where @xmath18 and @xmath19 are the total ( gas plus radiation ) and the gas pressure at the disc midplane , respectively , and @xmath20 is a constant of the order of unity .
such a scaling was first suggested by taam & lin ( 1984 ) , and its stability properties have been studied by lin & shields ( 1986 ) and , for the case of slim discs at high accretion rates , by szuszkiewicz ( 1990 ) .
if magnetic turbulence is ultimately responsible for transporting angular momentum , we can still retain the usual modified @xmath21 viscosity prescription in which the stress tensor is proportional to the gas pressure @xmath22 @xcite , but now with @xmath23 .
as in many other previous treatments of the two - phase thermal models for accretion flows onto compact objects ( haardt & maraschi 1991 ; svensson & zdziarski 1994 ) , we parameterize the coronal dominance by means of the fraction @xmath24 of gravitational power associated with the angular momentum transport that is stored in the form of magnetic structures ( loops , flux tubes ) , transported vertically through the disc in a dissipationless fashion , and finally dissipated in the corona . thus , if the total released power of the accretion disc corona system is given by @xmath25 , we define the corona luminosity as @xmath26 .
the fraction @xmath24 of power dissipated in the corona can be expressed as the ratio of the magnetic poynting flux in the vertical direction to the locally dissipated energy flux in the disc : @xmath27 where @xmath28 is the vertical poynting flux and @xmath29 is the dissipation rate of gravitational energy per unit surface area , and we have defined the velocity @xmath30 with which the magnetic flux is transported in the vertical direction . from the standard equations of geometrically thin accretion discs with modified viscosity law we have @xmath31 where @xmath32 is the keplerian velocity .
the main source of uncertainties lies in the complex process of vertical transport of magnetic flux tubes in the body of the accretion disc , and therefore on the value of the velocity @xmath30 ( see coroniti 1981 ; stella & rosner 1984 ; vishniac 1995a , b ; for a more thorough discussion ) . here
we assume that the rise speed of the magnetic tubes is proportional to their internal alfvn speed , and consequently scales as @xmath33 , where @xmath34 is the standard plasma parameter , and @xmath35 is related to the efficiency of buoyant transport of magnetic structure in the vertical direction inside the disc , which is of the order of unity for extremely evacuated tubes .
equation ( [ f1 ] ) then becomes @xmath36 the fraction @xmath24 can then be calculated for any given accretion rate and at any given radius .
if the mass of the central black hole is @xmath37 and the radial distance from the central source is expressed in units of schwarzschild radii @xmath38 , from ( [ eq_f2 ] ) and the equations for the ratio @xmath39 with modified viscosity law , we obtain the following implicit equation for @xmath40 as a function of @xmath41 in the radiation and gas pressure dominated zones of the disc , respectively ( cfr .
stella & rosner 1984 , appendix a ) of the radiative diffusion equation has been set equal to 1 for the radiation pressure dominated solution and to 8/3 in the gas pressure one in order for the two solutions to match continuously at the boundary . also , for the ease of computation , we have approximated @xmath42 and @xmath43 in the expression for @xmath44 . ]
@xmath45^{8/5}\ ] ] @xmath46^{4/5},\ ] ] where the function @xmath47 comes from the no - torque inner boundary condition .
the numerical factor @xmath48 contains the unknown quantities in our treatment .
as we shall see , the value of @xmath49 is related to the maximal fraction of power released in the corona .
it is evident that high viscosities and vertical rise speeds ( that are associated with the presence of very low-@xmath50 , strongly evacuated , magnetic filaments in the disc ) are needed in order for the corona to be powerful .
the fraction of power dissipated in the corona is higher when gas pressure dominates over radiation pressure inside the disc .
this happens over a larger and larger portion of the disc as the accretion rate decreases .
the radial boundary between the two solution is found when the fraction of power released in the corona is @xmath51 , and can be calculated by solving the equation @xmath52 the left hand side has a minimum at @xmath53 , therefore , for accretion rates smaller than a critical value , @xmath54 , the disc is gas pressure dominated even in its inner part . equations ( [ eq_f_rad ] ) and ( [ eq_f_gas ] ) can be solved numerically to calculate the disc structure at every radius and for every accretion rate . a global value of @xmath55
can then be obtained by integrating over all the disc area : @xmath56 in figure [ f_mdot_coeff ] we plot such global value of @xmath55 as a function of the total accretion rate for two different values of @xmath49 ( 0.99 and 0.1 ) for typical stellar mass and supermassive black holes . as already pointed out , as the radiation pressure dominated part of the disc shrinks , the value of @xmath55 increases as a larger portion of the disc is gas pressure dominated . for accretion rates smaller than the critical one , @xmath57 , when the disc is completely gas pressure dominated , the @xmath55@xmath58 relation flattens , and the fraction of the power released in the corona increases only slightly as the disc becomes denser and the radiation pressure further decreases .
the maximum coronal dominance is attained for the lowest values of the accretion rate , and is roughly equal to @xmath59 . if @xmath60 eqs .
( [ eq_f_rad ] ) and ( [ eq_f_gas ] ) do not admit a solution with @xmath61 over a wide range of radii .
this might correspond to a situation in which the magnetic field is so strong and/or the vertical transport of magnetic flux tubes is so efficient that the standard accretion disc corona configuration in no longer stable and the accretion disc , which can not maintain its keplerian profile , is effectively fragmented by the magnetic field into a number of disconnected rings ( heyvaerts & priest 1989 ) . also , very low-@xmath50 discs are qualitatively different in terms of angular momentum transport processes .
in fact , in such cases the magneto - rotational instability is suppressed and magnetic stresses at the disc
corona interface may be the primary angular momentum vehicle ( blandford & payne 1982 ) and the standard disc structure we have assumed may not be appropriate .
we therefore did not investigate such a case further , restricting ourselves to the stable case in which @xmath62 in the disc , and @xmath63 .
the coronal strength depends weakly on the central source mass , as implicitly shown in eq .
( [ eq_f_rad ] ) and ( [ eq_f_gas ] ) . in fig .
[ mass ] we show explicitly the values of the accretion rate for which @xmath64 ( @xmath65 ) .
ghisellini & celotti ( 2001b ) , by studying the dividing line between fr i and fr ii radio - galaxies ( which correspond approximately to low and high radio power galaxies , respectively ) in the radio host galaxy luminosity plane conclude that such dividing line correspond to an approximately universal ( independent on the black hole mass ) transition value of @xmath66 , consistent with the values we found for the accretion rates at which the corona starts to be the dominant repository of gravitational energy . to summarize , in accretion discs where angular momentum is transported either by magnetic turbulence , or stresses ,
magnetic coronae should be dominant at low accretion rates , and their strength depends upon the nature of magnetic dissipation inside the disc .
furthermore , magnetic coronae are stronger in stellar mass black holes than in agn ( see also fig .
[ mass ] below ) .
what is the fate of the energy deposited in the corona by the magnetic structures that emerge from the disc when @xmath24 is large ? to indicate the global value of the fraction of gravitational power released in the corona @xmath55 , neglecting any radial dependence . ]
if the current paradigm for high energy emission from black holes is correct , involving inverse compton scattering of soft photons on a population of thermal electrons at temperatures @xmath67k , then part of this energy must be used to heat the corona , most likely via magnetic reconnection @xcite and then efficiently radiated away . on the other hand , because compton cooling is so efficient in the strong radiation field of an accretion disc , the energy densities of the thermal electrons and of the hard x - ray radiation are likely to be only a small fraction of the total energy stored in the corona , whose dominant contribution should come from stored magnetic field ( merloni & fabian 2001a ) and possibly hot ions , if the corona is two - temperature ( di matteo , blackman & fabian 1997 ) .
in such a situation , the condition can be favourable for the occurrence of strong coronal outflows .
the nature of the outflow depends crucially on the magnetic field strength and topology and on the physical properties of the particle heating mechanisms . in merloni & fabian ( 2001a )
it has been shown how the values of coronal temperatures and optical depths inferred from the observations of low luminosity black holes require that the main reservoirs of energy in the corona are either a structured magnetic field or the protons in a two - temperature plasma . the extent to which the protons may act as repository of most of the dissipated energy crucially depends on how efficient the heating of electrons in the corona is .
great theoretical effort has been devoted recently to the study of particle energization in accretion flows , mainly for its relevance for the adaf scenario .
quataert ( 1998 ) and quataert & gruzinov ( 1999 ) have shown that in optically thin , advection dominated accretion flows , as long as the mhd turbulence is dissipated on small scales via incompressible modes ( alfvn waves ) , protons will be mainly heated only in high @xmath50 plasmas ( @xmath68 ) , while electrons are more likely to be heated in low @xmath50 plasmas . on the other hand blackman ( 1999 ) argued that if modes of dissipation involving compression dominate , then , even for @xmath69 , electron heating is negligible .
furthermore , we stress that the field configuration in the corona ought to be more complicated than the uniform field considered by quataert ( 1998 ) and quataert & gruzinov ( 1999 ) , with the field concentrated in loops and tubes , similarly to those observed in the sun .
magnetic reconnection may play an important role in this case ( see e.g. bisnovatyi - kogan & lovelace 1997 ; di matteo 1998 ) .
unfortunately , the physics of particle energization in magnetic reconnection is still poorly understood , and no clear conclusions can be safely drawn at present .
however , the conclusion that a powerful corona is the ideal launching platform for powerful jets remains unscathed : as a matter of fact , magnetically dominated coronae ( @xmath70 ) may power mhd driven outflows which can carry away a substantial fraction of the gravitational power ( see section [ sec_mhd ] ) , even if the ions are subvirial . on the other hand , for @xmath62 , ions could be the main repository of the power dissipated in the corona ; they can be heated to supervirial temperature and launch powerful thermally - driven outflows ( see section [ sec_th ] ) . for the sake of simplicity ,
we keep the treatment of these two cases separate , bearing in mind that in a realistic situation they may both contribute to the determination of the total outflowing power .
we demonstrate that the power carried away by the outflow / jet may be a dominant fraction of the total accretion power in both cases . to be fairly general
, we consider that the total coronal power generated by the accretion disc , @xmath71 , can be either dissipated locally to heat the corona , and ultimately radiated away as hard x - ray radiation with a luminosity @xmath72 , or used to launch a jet with power @xmath73 . in the following ,
we investigate the dependence of the parameter @xmath74 on the disc parameters .
the coronal magnetic field intensity depends on the dissipation rate and on the field geometry .
here we assume , on the basis of the strong spectral and variability arguments @xcite , that the corona is structured and the magnetic field concentrated in a number of small active regions ( where dissipation occurs ) .
the size , height and distribution of such regions determines the details of the spectral and temporal properties of the high energy emission in these sources ( merloni & fabian 2001b ) . here
it will suffice to notice that a relatively small number ( @xmath75 , see merloni & fabian 2001b ) of spherical active regions of size and height of the order of a few schwarzschild radii ( @xmath76 ) is what is needed to reproduce the average properties of the hard x - rays emission of gbhc in their hard state and of seyfert 1s and low luminosity agn , namely a spectral index @xmath77 , high energy cut - off around 100 kev and strong variability .
we therefore assume that the magnetic energy density in the corona is @xmath78 where @xmath79 is the dissipation velocity and depends on the uncertain nature of the reconnection process @xcite , and can be assumed to be of the order @xmath80 . here
we discuss the most general mechanisms by which a coronal magnetic field can power strong outflows ( either in the form of poynting flux or in the form of a magnetically driven material wind ) .
models and simulations of jet production @xcite show that it is the _ poloidal _ component of the magnetic field which mainly drives the production of powerful jets , and the output power can be expressed as ( livio , ogilvie & pringle 1999 ) @xmath81 here @xmath82 is the size of the region where most of the coronal power is dissipated and @xmath5 is the typical angular velocity of the magnetic field lines .
maximal rotation of the central black hole may be important in this context more for its deeper potential well and the larger rotational velocities that the accreting gas can attain ( see also meier 2001 ) than for the extraction of rotational energy from the black hole itself
. the magnetic field that emerges from the disc into the corona must be mainly azimuthal , as simulations also suggest @xcite . if the disc is not threaded by any external large scale poloidal field , as it is likely to be in the case of central black holes , the strength of the poloidal field component depends on the typical scaleheight of a coronal magnetic flux tube and on the capability of reconnection events to create larger and larger coherent structures .
tout & pringle ( 1996 ) have proposed numerical and analytic models in which stochastic reconnection creates magnetic loops on all scales , with a power - law distribution . here , as the jet - launching region is identified with the corona itself , we consider the constraints put on the poloidal field by the coronal geometry rather than the disc one . a magnetic flux tube emerging from the disc should have a cross section of the order of the disc thickness @xcite .
nonetheless , the strong differential rotation at the tube footpoints causes a twisting of the tube magnetic field and increase its tension , counterbalancing its sideways expansion ( see parker ( 1979 ) , 9.1 , 9.6 ) . as both theoretical arguments and numerical simulations have shown @xcite , the net effect of this energy injection in the magnetic coronal loops is therefore a rapid loop expansion , followed by reconnection ( at a height @xmath83 much larger than the disc thickness ) and opening of the field lines .
then , if @xmath83 is the typical coronal flux tube scaleheight ( height of a reconnection site ) , we have @xmath84 from eqs .
( [ eq_b_lh ] ) , ( [ eq_lj ] ) and ( [ eq_bp_b ] ) we obtain @xmath85 which in turn gives , for the fraction of coronal power that goes into the mhd jet / outflow @xmath86 therefore , from eq .
( [ eq_eta ] ) , we can conclude that the jet / outflow power is stronger if : * the dissipation speed is low . in this case magnetic field
does not dissipate effectively via reconnection and can reach high intensities , favouring mhd outflow launching ; * the coronal scaleheight is large with respect to the distance from the central source .
this would help in increasing the relative strength of the poloidal component of the magnetic field , that is the one ultimately responsible for the powering of the jet ; * the rotational velocity of the field lines is high .
this is related to the gravitational potential of the central black hole ( maximally rotating kerr holes can increase the angular velocity of field lines relative to a distant observer due to the strong frame dragging effect , see e.g meier 2001 ) , and to the typical radius of maximal coronal dissipation . as an example , for @xmath87 , @xmath88 , @xmath89 and @xmath90 , we obtain @xmath91 : the mhd jet can carry away a substantial fraction of the coronal ( and accretion ) power .
the fraction of jet power @xmath92 strongly depends on the typical height of the coronal active regions .
interestingly , for a given active region size , @xmath93 , the greater its distance from the disc , @xmath83 , the more important will be synchrotron radiation as a source of soft photons for comptonization ( di matteo , celotti & fabian 1999 ; merloni & fabian 2001b ) as opposed to external radiation from the disc .
this immediately suggests that sources with observed variable high frequency synchrotron radiation are strong candidates for harbouring powerful mhd driven jets . as already discussed at the beginning of section 3 , a crucial uncertainty about the energetics of powerful coronae is whether dissipation of magnetic energy preferentially heats the protons or the electrons .
it seems very likely that the fraction of energy dissipated into the electrons is negligible at least for optically thin plasmas where magnetic pressure is smaller than gas pressure ( quataert & gruzinov 1999 ) . in this case , as discussed in di matteo , blackman & fabian ( 1997 ) , a two - temperature equilibrium ( with @xmath94 ) is easily established .
let us therefore consider the case of a two - temperature corona ( see also janiuk & czerny 2000 ; raska & czerny 2000 ) and assume that all the dissipated magnetic energy is dumped into the protons @xmath95 the protons exchange energy with the electrons only via coulomb collision , so that , in a steady state , @xmath96 where @xmath97 is the rate of energy transfer from ions to electrons due to coulomb collisions between populations with maxwellian distributions , calculated using the rutherford scattering cross section ( stepney & guilbert 1983 ) : @xmath98 where @xmath99 are the electrons and ions temperatures , respectively , @xmath100 ( @xmath101 ) are dimensionless temperatures , @xmath102 is the coulomb logarithm , @xmath49s are the modified bessel functions and @xmath103.\ ] ] in order to calculate the electron temperature , we equate the electron heating rate with the total cooling rate @xmath104 , which includes inverse compton scattering of both disc photons and self - absorbed synchrotron emission produced in the active region itself ( see e.g. di matteo , celotti & fabian 1997 , wardziski & zdziarski 2000 ) . once the heating rate is specified via eqs .
( [ eq_b_lh ] ) and ( [ eq_mag_heat ] ) , and the geometry of the corona is fixed , we can calculate electron and ion temperatures if we know the coronal optical depth .
observationally , both black hole candidates in their low / hard state and agn have similar values of @xmath105 , lying in a narrow range around @xmath106 @xcite . as an illustrative example
, we made here the assumption that , for each value of the central source mass , @xmath107 , the coronal optical depth is proportional to the accretion disc surface temperature .
in particular , for gas pressure dominated solution appropriate for low accretion rates , we have chosen @xmath108 . therefore @xmath105 is smaller for smaller accretion rates ( @xmath109 for @xmath110 ) and we have @xmath111 for @xmath112 , depending on the black hole mass .
the dependence of @xmath105 on the accretion rate for central black holes of masses @xmath113 and @xmath114 is shown in fig .
[ tau_alpha_th ] .
it is important to stress that the specific dependence of @xmath105 on the disc parameters we have assumed here has been chosen only for illustrative purposes : the results we present on the strength of thermally driven outflows would be qualitatively the same for any physical situation in which colder and denser accretion discs were sandwiched by lower optical depth coronae ( as would be the case , for example , if the coronal optical depth depended on the accretion disc ionization parameter @xmath115 ) .
in fact , the only requirement for the jet / outflow to dominate the energy budget at increasingly lower accretion rates is that the coronal optical depth decreases with decreasing accretion rate , regardless of the specific profile .
if instead @xmath105 was approximately independent of @xmath58 , the highest proton temperature would be reached for @xmath57 , where the absolute coronal power , @xmath71 is maximum .
once a value for the coronal optical depth is fixed , for each value of the accretion rate we can then calculate magnetic field intensity , electron and proton temperatures and the overall emitted spectrum by solving the equation for the cooling and heating rate , adopting an analytic approximation for the emissivities described in merloni & fabian ( 2001b ; appendix a ) . in figure [ tau_alpha_th ]
we also show , as functions of the accretion rate , the calculated values of the electron temperature @xmath116 and spectral index @xmath21 of the hard x - rays produced by inverse compton scattering on the hot coronal electrons .
we stress that the values of the electron temperature depends also on the details of the coronal geometry , namely , on the size , height and number of magnetic regions reconnecting in the corona at any time @xcite .
this is an inherently time dependent problem ( see e.g. poutanen & fabian 1999 , merloni & fabian 2001b ) and here we have chosen average values for the coronal geometric parameters ( @xmath117 ) that ensure that the value of the spectral index is close to the observed one , and we are not interested in detailed spectral modeling of the x - ray emission . if the coronal density is low enough , and if the power dissipated in the corona high ( and both conditions are more likely in low accretion rate discs , as we have shown in section [ sec_f_mdot ] ) , the protons can be heated up to temperatures higher than the virial one @xmath118 ( here we may take as @xmath119 the radius at which the coronal dissipation is higher ) . in this case
the corona is likely to drive powerful jets / outflows @xcite .
a fraction @xmath74 of the coronal power is advected away in the outflow in the form of bulk kinetic energy and the heating rate of the protons is therefore reduced . in such cases
we may estimate a value for the parameter @xmath74 simply by looking for the minimal value of @xmath74 for which the protons are at the virial temperature .
the resulting values for such @xmath120 as a function of the accretion rate for @xmath121 and two different black hole masses are shown in figure [ eta_mdot ] .
the outflow / jet power ( kinetic energy plus poynting flux ) decreases sharply as the accretion rate increases and is negligible as soon as the temperature up to which the protons are heated in the corona drops below the virial one .
regardless of the nature of the launching mechanism , a robust conclusion from the analysis of the previous sections is that when the fraction of gravitational power released into a structured magnetic corona approaches unity , strong outflows ought to be produced .
here we describe the basic spectral properties of the disc corona emission . as a reference , we will consider a sequence of models taken from the two temperature solutions illustrated in fig .
[ f_mdot_coeff ] and [ eta_mdot ] , but we emphasize that any of such spectra would be observed if the jet was mhd driven , as far as it had the same value of @xmath74 . for small and intermediate values of @xmath24 ( therefore for @xmath122 or @xmath123 for @xmath113 and @xmath114 , respectively ) the seed soft photons for comptonization come from the quasi - thermal accretion disc emission ( either intrinsic or reprocessed ) .
then , the coronal cooling decreases when the accretion rate decreases , and the x - ray spectrum gets harder , as can be seen if fig . [ tau_alpha_th ] .
the hardest spectra are obtained for @xmath124 and @xmath125 . at this point inverse compton scattering on synchrotron photons produced in
the corona dominates the cooling .
finally , as the ions become supervirial , more and more power is carried away in the form of a powerful jet / outflow and the coronal heating is reduced more rapidly than the cooling due to synchrotron emission and the spectrum gets slightly softer again .
this is illustrated in fig .
[ spectra_10 ] and [ spectra_108 ] which show a sequence of spectral energy distributions for the stellar mass and supermassive black hole cases , respectively .
we do not attempt to model in detail the emission from the jet / outflow .
given the high brightness temperatures measured in the cores of some llagn , we can expect self - synchrotron compton emission to be important at low @xmath58 .
also , the high energy spectrum would be modified by even a small fraction of non - thermal electrons in the corona and/or in the outflow ( see e. g. poutanen & coppi 1998 ; wardziski & zdziarski 2001 ) .
at the low accretion rates we are considering , both our corona dominated solution for a standard thin disc and adaf / cdaf solution are admitted .
a ` strong adaf principle ' has been advocated ( narayan & yi 1995 ) to ensure that the real systems will tend to chose the adaf / cdaf branch whenever possible .
here we have explored the possibility that such a strong adaf principle does not hold , and studied the expected properties ( spectral and energetic ) of the alternative accretion mode . nonetheless , there are admittedly some similarities between the powerful outflow - dominated , radiatively - inefficient coronae that we found here at the lowest accretion rates and a classical adaf , and we might well be exploring a region of the accretion flow parameter space where the two accretion modes are bridged together ( in this case mainly by virtue of strong magnetic field generation , rather than evaporation ; meyer & meyer - hofmeister 1994 ) .
indeed , predictions from adaf models of a transition in spectral properties at @xmath126 @xcite are similar to those of our coronal outflow dominated sources , at least at high energies . also , from our study we expect strong coronal outflows at high values of the viscosity , similar to what is predicted by the theory of advective flows @xcite .
the main testable difference between the two models is the presence of a geometrically thick , cold , accretion disc in the inner part of the flow . in the magnetic corona scenario , even at the lowest accretion rates , when most of the accreting power is dissipated in the corona ,
the presence of the disc in the inner part of the flow would manifest itself by reprocessing coronal radiation and producing both a cold , quasi - thermal and a reflection components .
nonetheless , it is well known that if the coronal luminosity is so high that the inner part of the accretion disc is fully ionized , the reflection features are so weak that it is difficult to distinguish between a truncated disc and an ionized one @xcite .
furthermore , when the corona is so powerful that it produces relativistic outflows , the bulk motion of the coronal outflowing material reduces the amount of high energy radiation impinging on the disc , thus reducing the strength of the reprocessed components @xcite . when @xmath124 and @xmath127 , the radiative efficiency of a coronal outflow dominated flow depends on the radiative efficiency of the outflow itself
. however , unless @xmath128 , we can not achieve the extremely low efficiencies ( @xmath129 ) required to explain some nearby llagn @xcite .
such estimates , though , are based on the bondi accretion rates estimated at relatively large distances from the source .
it is plausible to conceive that the strong outflows we envisage , by interacting with the intergalactic medium , would stifle accretion through the inner part of the disc ( an effect analogous to that of the adios solution of blandford & begelman 1999 , in the advective case ) , reducing the value of the accretion rate through the inner disc that we are considering here and alleviating the need for too low a radiative efficiency ( see also discussions in melia and falcke , 2001 , with regards to the galactic center , and di matteo et al . , 2001 , for ngc 6166 ) .
our coronal - outflow - dominated solutions are both thermally and viscously stable , as in general are all standard shakura
sunyaev accretion disc solution in the gas pressure dominated regime .
rapid and dramatic variability in the observed high - energy flux is nevertheless expected , as x - rays are produced by coronal structures that are the eventual outcome of the turbulent magnetic field generation inside the disc .
the geometry of these structures ( open vs. closed field lines , for example ) plays a very important role and may be such that , at times , parts of the corona become temporarily radiatively efficient .
the hard x - ray flaring event recently detected by chandra from sgr a@xmath4 @xcite might be an example of such an occasional coronal re - brightening . at the very low accretion rates inferred for this source
, we predict the x - rays to be produced by inverse compton scattering on synchrotron photons ( see section 4 ) , as indeed is suggested from the observations @xcite .
we note however that the interpretation of the galactic centre as a coronal - outflow - dominated disc is not straightforward .
firstly , the extreme faintness of the source requires an accretion rate in the inner part of the flow extremely low compared to that expected from simulations of the multiple wind sources in the inner @xmath130 pc @xcite and values of @xmath74 very close to unity .
the former requirement of small @xmath58 ( also consistent with the polarization characteristics of sgr a@xmath4 at millimeter and submillimeter wavelength , see agol , 2000 , quataert & gruzinov , 2000b , and melia & falcke , 2001 , and references therein ) may be due to the feedback from the coronal outflow suppressing the accretion rate at larger radii .
however , a second problem faced by our model is the over - prediction of the ir emission that should be produced by the infalling wind impacting on the outermost part of the cold disc @xcite .
apart from the possibility that the above mentioned interaction between outflowing and inflowing gas modify substantially the kinematics of the accretion flow , it is possible that in fact the inflowing matter forms a keplerian disc only relatively close to the central black hole @xcite .
such a small disc would still be dominated by a strong outflowing corona as the one we have presented here , with enough power to produce a jet responsible for the radio emission , as in the models of falcke & markoff ( 2000 ) .
clearly , the challenge posed by the wealth of data from the galactic centre to any accretion model requires more detailed work than the one we have outlined here , which would be beyond the aim of this paper .
an important related issue concerns the fate of the cold thin disc when we extrapolate our solution to extremely low accretion rates .
as we have discussed already in section [ sec_f_mdot ] , very strong outflowing coronae might have relevant effects on the nature of the angular momentum transport in the flow , and therefore on the disc structure .
the constant improvement of numerical simulations in recent years , make us believe that the relevance of the solutions we describe here will be assessed in the near future . at present , numerical simulations of mri unstable accretion discs with sufficient extension in the vertical direction
seem able to generate only modestly powerful coronae ( @xmath131 20 per cent , miller & stone 2000 ) . the inclusion of radiation in the mhd codes @xcite , and a detailed study of the dependence of the solution on the accretion rate will prove crucial to settle the issue .
we have discussed separately two different outflow producing mechanisms ( mhd and thermal ) .
their relative importance crucially depends on the uncertain nature of the magnetic energy dissipation in the corona and of particle heating .
however , as we have discussed in section 3 , according to our current theoretical understanding of these issues , at least one of the two mechanisms must be at work in the physical conditions we envisage for a powerful corona .
clearly both may be at work at the same time , and influence each other .
for example , when the conditions for mhd jet launching are favourable , and a larger number of open field lines thread the inner part of the corona , then , depending on the inclination angle of the field lines , a larger mass can be channeled into the outflow , reducing the coronal density and therefore the coulomb cooling of the hot ions and enhancing the thermally driven jet power .
radiatively inefficient flows are also expected at high accretion rate , when the flow is geometrically and optically thick and the radiation produced is trapped by the flow and advected with it @xcite .
when the accretion rate is of the order of the eddington one , the disc thickness stays moderate and a vertically integrated approximation may be retained ( slim discs , abramowicz et al .
1988 ) . in the limit @xmath132 ,
though , the behaviour of the disc still remains an open issue .
the problem is inherently 2d @xcite , and the simultaneous roles of convection , advection and outflows have to be assessed in order to properly model the expected sed .
if a strong outflow was produced , through a magnetic corona analogous to the one we envisaged for low accretion rates , or through any other radiative or hydrodynamical mechanism , the total amount of outflowing kinetic power could exceed the eddington limit ( and be much larger than the radiated one ; meier 1982 ) .
we just remark here that observationally , black holes believed to accrete at super - eddington rates bear many similarities with their low - accretion rate counterparts .
this class of object may include gbhc in the very high state ( see done 2001 and references therein ) , whose x - ray spectra show a power - law component usually as strong as the quasi - thermal one ; the microquasar grs 1915 + 105 @xcite ; broad line radio galaxies ( blrg , see e. g. zdziarski 1999 and reference therein ) ; powerful blazars ( see in particular ghisellini & celotti 2001a , where it is shown how the relativistic jet in these sources may dominate the total output power ) ; and possibly , narrow line seyfert 1s ( nls1 , mineshige et al .
most likely , newly formed supermassive black holes at the epoch of galaxy formation and quasars in their early evolutionary stages might be fed at super - eddington rates , therefore a better understanding of accretion flows and jet / outflow production in these sources will shed light on the processes of galaxy formation and evolution @xcite .
we studied the energetics and power output of magnetic structured coronae generated by geometrically thin , optically thick accretion discs at low accretion rates .
we have assumed that angular momentum transport in the disc is due to magnetic turbulent stresses .
the magnetic field , amplified by mri , saturates because of the strong buoyancy of flux tubes .
then , we have shown that both magnetic energy density and effective viscous stresses inside the disc are proportional to the geometric mean of the total ( gas plus radiation ) and gas pressure . in this case , the relative strength of the corona increases as the accretion rate decreases .
we have then studied the energetics of the corona itself , and assessed the relevance of the various mechanisms that can generate powerful outflows from the coronal region .
we have shown that , depending on the actual coronal field geometry , mhd launching mechanisms can indeed produce strong outflows , whose power is of the order of the radiated one . moreover ,
if the corona is two - temperature , a thermally driven outflow can easily dominate the power output from the source at low accretion rates .
thus , if the jet / outflow do not radiate efficiently , the whole system can be very under - luminous compared to a standard radiatively efficient accretion disc .
the model we have presented might be relevant to the accretion mode of low - luminosity black holes , either gbhc in the _ low / hard _ state or llagn observed in the nuclei of nearby galaxies .
the manuscript has been improved due to the comments and criticisms of the referee , prof .
ramesh narayan , and to the many suggestions from annalisa celotti , whom we both thank .
this work was done in the research network
accretion onto black holes , compact stars and protostars " funded by the european commission under contract number erbfmrx - ct98 - 0195. am and acf thank the pparc and the royal society for support , respectively .
zdziarski , a. a. , 2000 , invited review for iau symp . 195 : _ highly energetic physical processes and mechanisms for emission from astrophysical plasmas _ , p. c. h. martens , s. tsuruta , & m. a. weber , eds . ,
153 - 170 . | the spectral energy distributions of galactic black holes in the low / hard state and of low - luminosity agn possess many common features , the most prominent being : compact , flat ( or inverted ) spectrum radio cores with high brightness temperatures ; excess red and infrared emission , often correlated with the radio flux ; an extremely weak ( or absent ) quasi - thermal hump and a hard x - ray power - law with high energy cut - off .
these sources are thought to be accreting at low rates and advection ( or convection ) dominated accretion flows are usually considered the best candidates to explain them . here
we present an alternative possibility , involving strong , unbound , magnetic coronae generated by geometrically thin , optically thick accretion discs at low accretion rates .
first we show that , if angular momentum transport in the disc is due to magnetic turbulent stresses , the magnetic energy density and effective viscous stresses inside the disc are proportional to the geometric mean of the total ( gas plus radiation ) and gas pressure . therefore the corona is less powerful in a radiation pressure dominated disc , and the relative fraction of the power liberated in the corona increases as the accretion rate decreases .
furthermore , we discuss reasons why energetically dominant coronae are ideal sites for launching powerful jets / outflows , both mhd and thermally driven . in analysing the spectral properties of such coronal outflow dominated accretion discs , we reach the important conclusion that if the jet / outflow is , as is likely , radiatively inefficient , then so is the source overall , even without advection of energy into the black hole being relevant for the dynamics of the accretion flow .
[ firstpage ] accretion , accretion discs black hole physics magnetic fields |
the fabrication of individual graphene flakes @xcite , followed by an observation of the integer quantum hall effect in them @xcite , triggered an explosion of interest in the electronic properties of graphene .
it was fuelled , in part , by the realization that the low - energy band structure of a graphene monolayer consists of two dirac cones centered at inequivalent corners of the brillouin zone , @xmath1 and @xmath2 , which are called valleys [ fig .
[ fig:1 ] ( b ) ] .
they support chiral quasiparticles with opposite chirality in each valley , and a linear dispersion reminiscent of the quantum electrodynamics of massless fermions @xcite . in the presence of time reversal symmetry ,
the energy spectrum is degenerate between the different valleys since the time reversal operation connects electronic states at @xmath1 to those at @xmath2 . in graphene ,
not only time reversal symmetry , but parity , i.e. , spatial inversion symmetry with respect to the center of a hexagon @xcite , is also able to transform electronic states between valleys . in the presence of a magnetic field , parity ensures degeneracy of the electronic spectra at different valleys . in this paper , we study spatial symmetry in general @xmath0-stacked ( bernal ) multilayer graphene composed of @xmath3 layers , to illustrate how the electronic spectra at the two valleys are related in a magnetic field . in multilayers with even @xmath3 , including bilayers @xcite , the lattice obeys spatial inversion symmetry @xmath4 similarly to monolayers , which swaps electronic states between valleys ensuring valley degeneracy in the absence of time - reversal symmetry .
the picture is different in multilayers with odd @xmath3 , starting from trilayer graphene ( @xmath5 ) , because their lattices do not satisfy spatial inversion symmetry @xcite , fig .
1(c ) . here
we consider an additional transformation which imposes an approximate valley degeneracy on the electronic spectra of odd-@xmath3 multilayers .
it arises because the electronic hamiltonian may be decomposed into separate monolayer- and bilayerlike parts @xcite , with each part satisfying an approximate inversion symmetry corresponding to that of the real lattice of monolayer or bilayer graphene , respectively .
we predict a peculiar landau level spectra in bilayer and trilayer graphenes in the presence of interlayer potential asymmetry , with unusual structures of broken valley degeneracy that are markedly different from each other .
we present here general symmetry arguments without referring to model - specific details , in order to provide information about the valley degeneracy of landau levels .
we consider bernal - stacked multilayer graphene with @xmath3-layers in an external uniform magnetic field @xmath6 with arbitrary direction , and an external potential @xmath7 ( other than the lattice potential ) where @xmath8 . without resorting to any approximations
, we can formally express the total hamiltonian of the system as @xmath9 $ ] , and we take @xmath10 and @xmath11-axes to be parallel to the graphene layers , and the @xmath12 axis in the perpendicular direction . _
even - n multilayers : _ in multilayers with an even number of layers , the lattice is symmetric with respect to spatial inversion symmetry @xmath13 [ @xmath14 as shown in fig .
[ fig:1 ] ( c ) , because the point group of the lattice @xmath15 @xcite ( @xmath16 ) can be regarded as a direct product of group @xmath17 ( @xmath18 ) with the inversion group @xmath19 ( @xmath20 ) . in the absence of @xmath6 , there is a symmetry with respect to time reversal @xmath21 ( @xmath22 ) . the influence of @xmath13 and @xmath21 on the hamiltonians of even-@xmath3 multilayers is summarized as @xmath23 p^{-1 } & = & h \left[\mathbf{b } , u(-\vec{r } ) \right ] \ , , \label{eq_even_p}\\ t:\,\,\,\,\qquad h^{\ast } \left[\mathbf{b } , u(\vec{r } ) \right ] & = & h \left[-\mathbf{b } , u(\vec{r } ) \right ] \ , .
\label{eq_even_t}\end{aligned}\ ] ] the magnetic field @xmath6 does not change sign under the operation @xmath13 because it is an axial vector . manes _
@xcite used the combined @xmath24 operation to show that the fermi points are stable with respect to the opening of a gap . in multilayer graphenes , electronic properties
are well described by a @xmath25 approximation in the vicinity of @xmath1 and @xmath2 points @xcite .
we can consider the @xmath25 hamiltonian ( and thus eigenstates ) at each valley separately , as long as the potential @xmath7 is smooth in the @xmath26-plane compared to the atomic scale , and the magnetic field is not too strong as to mix states around the two valleys .
when @xmath27 , the time - reversal symmetry in eq .
( [ eq_even_t ] ) ensures valley degeneracy of the electronic spectrum , because the operation @xmath21 swaps eigenstates at @xmath1 and those of @xmath2 through flipping the bloch factor as @xmath28 . here
@xmath29 are the wave numbers corresponding to @xmath30 points @xcite .
the operation of spatial inversion @xmath13 also exchanges @xmath1 and @xmath2 because the point @xmath31 is equivalent to @xmath32 in the brillouin zone .
then the symmetry of eq .
( [ eq_even_p ] ) suggests that the eigenstates of @xmath1 at @xmath33 $ ] are related to those of @xmath2 at @xmath34 $ ] . from this
we immediately conclude that the landau levels in even-@xmath3 multilayers are degenerate in valleys as long as @xmath35 , or the external potential has inversion symmetry with the same symmetry point as that of the lattice potential . _
odd - n multilayers : _ in odd-@xmath3 multilayers with @xmath36 , the point group @xmath37 ( @xmath38 ) @xcite can be regarded as a direct product of the group @xmath17 ( @xmath18 ) with the reflection group @xmath39 ( @xmath40 ) . with respect to the even-@xmath3 multilayers ,
spatial inversion is replaced by mirror reflection @xmath41 [ @xmath42 as shown in fig .
[ fig:1 ] ( c ) , which does not reverse the in - plane electronic momentum .
the influence of @xmath21 and @xmath41 on the hamiltonians of odd-@xmath3 multilayers is summarized as @xmath43 \sigma_h^{-1 } & = & h \left[\mathbf{b } , u(x , y ,- z ) \right ] , \label{eq_odd_p}\\ t:\,\,\,\,\qquad \qquad h^{\ast } \left[\mathbf{b } , u(\vec{r } ) \right ] & = & h \left[-\mathbf{b } , u(\vec{r } ) \right ] . \label{eq_odd_t}\end{aligned}\ ] ] time reversal symmetry , eq .
( [ eq_odd_t ] ) , again ensures valley degeneracy of the electronic spectrum in the absence of an external magnetic field .
the mirror reflection symmetry would seem to play the role of parity .
however , it does not transform between states at the two valleys , and is , therefore , unable to guarantee valley degeneracy in a magnetic field .
it merely ensures that the spectrum at _ each _ valley is identical when the potential @xmath7 is inverted with respect to @xmath44 . actually , we can show that landau levels in odd - layered multilayer graphenes are approximately valley - degenerate when @xmath45 , by employing the effective mass hamiltonian described in the following section
although we focus on aba graphene multilayers in this paper , we point out that abc ( rhombohedral ) graphene multilayers , which have a different layer stacking @xcite , have inversion symmetry irrespective of their layer number , and thus the valley degeneracy of landau levels is always guaranteed .
to investigate the landau level structure in detail we adopt the effective - mass description of graphite in the slonczewski - weiss - mcclure parameterization @xcite .
we consider ab - stacked @xmath3-layer multilayer graphene in an external uniform magnetic field @xmath6 perpendicular to the layer , and with an external electrostatic potential @xmath46 at the @xmath47-th layer , which is uniform in the in - plane direction . in a basis with atomic components @xmath48 , @xmath49 , @xmath50 , @xmath51 , @xmath52 , @xmath53 , @xmath54 , the multilayer hamiltonian at the @xmath30 valley @xcite is @xmath55 with @xmath56 here @xmath57 , @xmath58 is the valley index and @xmath59 with the vector potential @xmath60 , which gives the external magnetic field as @xmath61 .
parameter @xmath62 is the potential at the @xmath63-th layer , which is constant within each layer , and @xmath64 represents the energy difference between sites which have neighboring atoms right above or below them and those sites which do not , and thus it only exists for @xmath65 .
parameter @xmath66 is the band velocity of monolayer graphene , which is written as @xmath67 .
other velocities are defined as @xmath68 and @xmath69 .
we neglect parameters @xmath70 and @xmath71 , which describe hopping between next - nearest neighboring layers .
they actually break exact valley degeneracy as mentioned later . for even-@xmath3 multilayers with @xmath72 , the hamiltonian eq .
( [ eq_h ] ) satisfies @xmath73 with @xmath74 being a @xmath75 matrix , @xmath76 which is nothing but the inversion symmetry discussed in the previous section . in odd-@xmath3 multilayers with @xmath77 ,
the landau levels are degenerate in valleys due to a different symmetry .
we can show this by employing a unitary transformation which decomposes the hamiltonian of @xmath3-layer graphene , eq .
( [ eq_h ] ) , into subsystems equivalent to monolayer and bilayer graphenes .
we construct the basis as @xcite @xmath78 where @xmath79 , and @xmath80,\\ g_m(j ) = -\frac{2}{\sqrt{n+1 } } \cos \left(\frac{\pi}{2 } j\right ) \sin \left[\frac{m\pi}{2(n+1 ) } j \right],\end{aligned}\ ] ] where @xmath81 is the layer index .
the label @xmath82 is the subsystem index which ranges as @xmath83 the superscript such as ( a , odd ) indicates that the wavefunction has an amplitude only on @xmath84 with odd @xmath47 s . when the hamiltonian ( [ eq_h ] ) with @xmath72 is written in the basis eq .
( [ eq_basis ] ) , the matrix is block - diagonalized in each @xmath82 .
the case of @xmath85 is special in that @xmath86 is identically zero , so that only two bases @xmath87 survive in eq .
( [ eq_basis ] ) .
the submatrix is written for this two component basis as @xmath88 which is equivalent to the hamiltonian of monolayer graphene except for the diagonal terms containing @xmath64 . for @xmath89 , the submatrix for @xmath90
where @xmath91 is defined by @xmath92 eq . ( [ eq_hm ] ) is identical to the hamiltonian of bilayer graphene , except that interlayer - coupling parameters @xmath93 , @xmath94 and @xmath95 are multiplied by the factor @xmath96 .
since the decomposed hamiltonian matrices are analogous to those of monolayer or bilayer graphene , they obey the corresponding inversion symmetry .
the bilayer - type submatrix , eq .
( [ eq_hm ] ) , obeys @xmath97 which guarantees valley degeneracy of landau levels . in the original basis , this `` effective '' inversion process eq . ( [ eq_inv_bi ] ) exchanges the wave amplitudes on odd-@xmath47th layers and those on even-@xmath47th ones . in even-@xmath3 multilayers , the operation eq .
( [ eq_inv_bi ] ) for each eigenstate becomes equivalent to the original inversion symmetry eq .
( [ eq_inv_even ] ) except for a phase factor . for the monolayer - type submatrix eq .
( [ eq_h0 ] ) ( exists in odd-@xmath3 ) satisfies @xmath98 but it is only approximate since @xmath64 breaks this symmetry , unlike in the bilayer - type symmetry , eq .
( [ eq_inv_bi ] ) . the operation of eq .
( [ eq_inv_mono ] ) exchanges the wave amplitudes within odd-@xmath47 layers .
the extra parameters @xmath70 and @xmath71 neglected here generally mix the states between different @xmath82 s , and also appear in diagonal elements within each @xmath82 @xcite . in odd-@xmath3 multilayers , they lift the valley degeneracy by breaking the effective inversion symmetry eq .
( [ eq_inv_bi ] ) through additional matrix elements .
it should be noted that , in even-@xmath3 multilayers , valley degeneracy is never influenced by any extra lattice parameter , because it is protected by inversion symmetry inherent in the lattice .
here , we illustrate the valley degeneracy of the electronic spectra in even- and odd-@xmath3 multilayers through an analytical description of the landau level spectra of bilayer and trilayer graphene .
we use the magnetic length @xmath99 and an energy scale @xmath100 related to the inverse of the magnetic length .
the landau level spectrum , in a magnetic field of magnitude @xmath101 perpendicular to the graphene sheet , may be found using the relation @xmath102 for @xmath1 , and @xmath103 for @xmath2 @xcite , where @xmath104 and @xmath105 are raising and lowering operators , respectively , which operate on the landau - level wave function @xmath106 as @xmath107 and @xmath108 . _ bilayer graphene : _ the hamiltonian of bilayer graphene @xcite is given by the multilayer hamiltonian , eq .
( [ eq_h ] ) , with @xmath109 . here
, we consider a simple form of the hamiltonian that only contains non - zero parameters @xmath110 , describing nearest - neighbor intralayer hopping , @xmath93 , describing the dominant interlayer coupling , and the layer potential defined as @xmath111 . in the absence of interlayer asymmetry , @xmath112 ,
the spectrum consists of fourfold ( valley and spin ) degenerate landau levels @xcite , except for the level at zero energy which is eightfold degenerate .
the valley degeneracy is guaranteed by spatial inversion symmetry eq .
( [ eq_even_p ] ) .
finite @xmath113 breaks spatial inversion symmetry and it splits the valley degeneracy of the levels @xcite , as indicated in the numerically - calculated spectrum with a magnetic field of @xmath114 , plotted in fig .
[ fig:2 ] @xcite .
there we can see that the energy levels of two valleys are related as @xmath115 , due to the inversion symmetry arguments presented in sec .
[ sec_sym ] .
the valley splitting @xmath116 is thus an odd function in @xmath113 , and generally begins with a term linear in @xmath113 . in the limit @xmath117 ,
the hamiltonian is approximately described by an effective hamiltonian @xcite operating in the space of two - component wave functions @xmath48 , @xmath51 @xmath118 the energy levels are obtained by assuming the wave function @xmath119 , where @xmath120 is regarded as 0 .
they are given by @xmath121 ^ 2 } \nonumber \\ & & - \xi \frac{\delta \gamma_b^2}{\gamma_1 ^ 2 } \ , , \label{bi1 } \\ \varepsilon_{0 } & = & \xi \delta \left ( 1 - \frac{2 \gamma_b^2}{\gamma_1 ^ 2 } \right ) \ , , \label{bi2 } \\
\varepsilon_{-1 } & = & \xi \delta \ , . \label{bi3}\end{aligned}\ ] ] the splitting of the valley degeneracy in @xmath122 ( by terms containing @xmath123 ) is linear in the asymmetry potential @xmath113 , and also in the magnetic field @xmath101 .
_ trilayer graphene : _ the trilayer graphene @xcite hamiltonian is given by the multilayer hamiltonian , eq .
( [ eq_h ] ) , with @xmath5 .
we again retain the parameters @xmath110 and @xmath93 , and set the layer potential as @xmath124 , to focus our attention on interlayer asymmetry @xmath125 .
as described in section [ s : emh ] , we perform a unitary transformation to decompose the hamiltonian into a monolayerlike part , eq .
( [ eq_h0 ] ) , and a bilayerlike part , eq .
( [ eq_hm ] ) ( such a decomposed trilayer hamiltonian is written explicitly in ref .
@xcite ) . in the absence of interlayer asymmetry , @xmath126 ,
the spectrum consists of superimposed monolayerlike and bilayerlike spectra @xcite , with fourfold ( valley and spin ) degenerate landau levels , except for the level at zero energy which is twelvefold degenerate .
the valley degeneracy is guaranteed by the effective spatial inversion symmetry of the separate , monolayerlike and bilayerlike , parts of the decomposed hamiltonian .
finite @xmath127 breaks the effective spatial inversion symmetry and it splits the valley degeneracy of the levels as indicated in the numerically - calculated spectrum at @xmath114 , plotted in fig .
[ fig:3 ] @xcite .
unlike in bilayer , the energy spectrum at each valley is an even function of @xmath127 , due to the reflection symmetry argued in sec .
[ sec_sym ] .
the valley splitting is therefore quadratic in @xmath127 , except for zero energy where the different levels are degenerate at each single valley , and the energy level ( and thus splitting ) is allowed to be linear in @xmath127 . in the limit @xmath128 , it is possible to obtain a simplified description of four electronic bands near zero energy , by eliminating atomic components associated with bands split away from zero by energy @xmath129 , to obtain a four - component effective hamiltonian in basis @xmath130/\sqrt{2}$ ] , @xmath131/\sqrt{2}$ ] , @xmath132/\sqrt{2}$ ] , @xmath51 , @xmath133 the energy levels are obtained by assuming the wave function @xmath134 . for levels with index @xmath135 at valley @xmath136 , the hamiltonian eq .
( [ h4b ] ) yields four energy levels for each @xmath137 , given by @xmath138}{\sqrt{\delta_1 ^ 2 + ( n+1 )
\gamma_b^2 } } \ , .\end{aligned}\ ] ] for index @xmath139 , there are three energy levels , @xmath140 and , for @xmath141 , two energy levels , @xmath142 here , the superscript @xmath143 means that the level is related to an `` upper '' energy band which is split away from zero energy by @xmath127 ( _ i.e. _ the landau level energy approaches @xmath144 as @xmath145 ) , whereas superscript @xmath146 means that the level is related to a `` lower '' energy band near zero energy ( _ i.e. _ the landau level energy approaches zero as @xmath145 ) .
the landau level spectrum at the second valley , @xmath147 , differs as compared to that at @xmath136 . for levels with index @xmath135 , the hamiltonian eq .
( [ h4b ] ) again yields four energy levels for each @xmath137 given by @xmath148}{\sqrt{\delta_1 ^ 2 + n \gamma_b^2 } } \ , .\end{aligned}\ ] ] for index @xmath139 , there are two energy levels at @xmath147 , @xmath149 and , for @xmath141 , one energy level , @xmath150 in the limit @xmath126 , the levels are degenerate in pairs of @xmath151 , and @xmath152 , which form monolayerlike and bilayerlike spectra , respectively . in the high - field limit , @xmath153 , the valley splitting of those levels is proportional to @xmath154 but independent of @xmath101 , unlike bilayer graphene .
the other levels give degenerate zero - energy states at @xmath155 . for @xmath156 , one of the levels
@xmath157 corresponds to the zero - energy level of the monolayerlike spectrum , whereas @xmath158 and one of @xmath157 correspond to two zero - energy levels of the bilayer - like spectrum .
this yields an overall twelvefold degeneracy with spin degeneracy included ( as opposed to the eightfold degeneracy in bilayers and fourfold degeneracy in monolayers ) . in the presence of finite interlayer asymmetry , two of the otherwise - zero levels at each valley are hybridized and split away from zero [ eq.([zp2 ] ) at @xmath136 and eq.([zm1 ] ) at @xmath147 ] , whereas one level at each valley remains at zero [ eq.([zp1 ] ) at @xmath136 and eq.([zm2 ] ) at @xmath147 ] .
zero energy states are formed primarily by different atomic orbitals at different valleys , and , as in bilayers @xcite , the orbitals in trilayers are on different layers , so that those levels exhibit quite different dependences on the interlayer asymmetry between @xmath1 and @xmath2 . in realistic experimental situations ,
interlayer potential asymmetry can be produced by gate - induced electric fields .
there the potentials @xmath62 , which are taken as external parameters in the present work , should be determined self - consistently including the screening effect of graphene electrons @xcite . the self - consistent landau level structure and quantum hall effect as a function of gate voltage would be an important future study @xcite .
it has been pointed out @xcite that a number of level crossings will occur in bilayer graphene at finite asymmetry , fig .
[ fig:2 ] .
we note that the landau level spectrum of trilayer graphene , fig .
[ fig:3 ] , has a very rich pattern of level crossings at a range of different values of energy and asymmetry .
the presence of additional terms in the hamiltonian will tend to produce some anti - crossings in the spectrum .
the precise position and nature of level crossings in the spectra of multilayer graphenes will be the subject of future investigation .
the authors thank t. ando and m. mucha - kruczyski for discussions .
this project has been funded by epsrc first grant ep / e063519/1 , the royal society , and the daiwa anglo - japanese foundation , and by grants - in - aid for scientific research from the ministry of education , culture , sports , science and technology , japan .
note that charge conjugation symmetry @xcite of the low - energy effective hamiltonians of monolayer , bilayer , and multilayer graphene ensures electron - hole symmetry of the spectra that , depending on whether @xmath3 is odd or even , occurs between spectra at either the same or at opposite valleys , even in the presence of finite interlayer asymmetry . | we study spatial symmetry in general @xmath0-stacked multilayer graphene to illustrate how electronic spectra at the two valleys are related in a magnetic field .
we show that the lattice of multilayers with an even number of layers , as well as that of monolayer graphene , satisfy spatial inversion symmetry , which rigorously guarantees valley degeneracy in the absence of time - reversal symmetry .
a multilayer with an odd number of layers ( three or more ) lacks inversion symmetry , but there is another transformation imposing an approximate valley degeneracy , which arises because the low - energy hamiltonian consists of separate monolayerlike and bilayerlike parts .
we show that an external electrostatic potential generally breaks valley degeneracy in a magnetic field , in a markedly different manner in odd and even multilayers . |
tides play an important role in binary star systems and in star - planet systems . while numerous studies of tidal effects have been based on the so - called equilibrium tide theory , which parametrizes tidal dissipation by an effective tidal lag angle / time or tidal quality factor ( e.g. , darwin 1880 ; goldreich & soter 1966 ; alexander 1973 ; hut 1981 ) , the underlying physics of tides in fluid stars and planets involves dynamical excitations of waves and oscillations by the tidal force ( see ogilvie & lin 2007 and zahn 2008 for recent reviews ) .
tides in highly eccentric systems are particularly rich in their dynamical behavior , since wave modes with a wide range of frequencies can be excited and participate in the tidal interaction .
various aspects of dynamical tides in eccentric binaries have been studied by lai ( 1996,1997 ) , kumar & quataert ( 1998 ) , witte & savonije ( 1999,2001 ) , willems et al .
( 2003 ) , ivanov & papaloizou ( 2004 ) and papaloizou & ivanov ( 2010 ) .
recent observations of the binary star system hd 187091 ( koi-54 ) by the _ kepler _ satellite provide a unique opportunity for studying dynamical tides in eccentric binaries .
koi-54 consists of two a stars ( mass @xmath2 and radius @xmath3 ) in an eccentric ( @xmath4 ) orbit with period @xmath5 days ( welsh et al .
the binary is nearly face - on with orbital inclination @xmath6 .
in addition to periodic brightening events caused by tidal distortion and irradiation of the two stars during their close periastron passages , the power spectrum of the _ kepler _ light curve revealed 30 significant ( with a signal - to - noise ratio @xmath7 ) stellar pulsation modes . the observed mode periods range from 45 hours to 11 hours , corresponding to the mode frequency @xmath8 ranging from @xmath9 to @xmath10 ( where @xmath11 is the orbital frequency ) .
most interestingly , twenty - one of these mode frequencies are integer multiples of @xmath12 ( with the ratio @xmath13 differing from an integer by @xmath14 or less ) .
the two dominant modes have frequencies that are exactly 90 and 91 times @xmath12 , with the corresponding flux amplitudes of 297.7 @xmath15mag and 229.4 @xmath15mag , respectively . while dynamical tides in massive - star binaries have been studied before ( e.g. , zahn 1977 , goldreich & nicholson 1989 for circular binaries ; lai 1996,1997 , kumar & quataert 1998 and witte & savonije 1999,2011 for eccentric binaries ) , koi-54 represents the first example where tidally excited oscillations are directly observed and therefore serves as an explict demonstration of dynamical tides at work in the system . as discussed in welsh et al .
( 2011 ) , the observed oscillation modes are puzzling : over 20 of the observed modes are nearly exact integer multiples of the orbital frequency , yet several others are not .
it is not clear why the dominant modes are so prominent , e.g. , why modes with frequencies 90 , 91 , 44 , and 40 times @xmath16 are clearly visible , and yet modes with frequencies greater than @xmath17 and those less than @xmath18 appear to be absent . the goal of this paper is to explain some of the observational puzzles related to koi-54 and to develop the general theoretical framework for studying tidally - excited oscillations in eccentric binary systems . our paper is organized as follows . in section 2 , we derive the general equations for calculating the energies of tidally excited oscillation modes in an eccentric binary .
our theory improves upon previous ( and less rigorous ) works , and provides a clear relationship between the resonant mode energy and non - resonant mode energy . in section 3
we study the properties of non - radial g - modes relevant to the stars in the koi-54 system and calculate the non - resonant mode energies these serve as a benchmark for examining the effect of resonances . in section 4
we present our calculations of the flux variation due to tidally - forced oscillations .
we show that the observed flux variation in koi-54 can be largely explained when a high - frequency mode is locked into resonance ( with the mode frequency equal to @xmath19 ) . in section 5
we study the possibility of resonance locking .
we show that the combination of the secular tidal orbital / spin evolution and the intrinsic spindown of the star ( e.g. , due to stellar evolution ) may naturally lead to resonance locking of a particular mode .
our analysis demonstrates that ( in the koi-54 system ) a mode with frequency around @xmath19 can be resonantly locked , while modes with higher frequencies can not . in section 6
we discuss the origin of the observed modes in koi-54 with frequencies that are not an integer multiple of @xmath16 , including the evidence of nonlinear mode coupling .
we conclude in section 7 with a discussion of future prospects and remaining puzzles .
we consider the tidally - excited oscillations of the primary star of mass @xmath20 and radius @xmath21 by the companion of mass @xmath22 .
the gravitational potential produced by @xmath22 can be written as u(*r*_i , t)=-gm_lmw_lmr^ld^l+1 e^-im(t ) y_lm(,_i ) , where @xmath23 is the position vector ( in spherical coordinates ) relative to the center of star @xmath20 ( the azimuthal angle @xmath24 is measured in the rotating frame of the star , with the rotation rate @xmath25 and the spin axis aligned with the orbital angular momentum ) , @xmath26 is the binary separation and @xmath27 is the orbital true anomaly .
the dominant terms have @xmath28 and @xmath29 , @xmath1 , and for these terms @xmath30 and @xmath31 .
the linear response of star @xmath20 is specified by the lagrangian displacement @xmath32 , which satisfies the equation of motion ( in the rotating frame of the star ) + 2_s+=-u , [ eq : eqnmotion2 ] where @xmath33 is a self - adjoint operator ( a function of the pressure and gravity perturbations ) acting on @xmath34 ( see , e.g. , friedman & schutz 1978 )
. a free mode of frequency @xmath35 ( in the rotating frame ) with @xmath36 satisfies -^2_-2i_s_+_=0 , where @xmath37 denotes the mode index , which includes the radial mode number @xmath38 , the polar mode number @xmath39 ( which reduces to @xmath40 for spherical stars ) and the azimuthal mode number @xmath41 .
we carry out phase space mode expansion ( schenk et al .
2002 ) = _ c_(t ) , where the sum includes not only mode indices , but also both positive and negative @xmath35 .
note that the usual mode decomposition , @xmath42 ( with the sum including only mode indices ) , adopted in many previous studies ( e.g. , lai 1997 , kumar & quataert 1998 ; witte & savonije 1999 ) , are rigorously valid only for non - rotating stars . using the orthogonality relation @xmath43 ( for @xmath44 ) , where @xmath45 , we find ( lai & wu 2006 ) @xmath46 where @xmath47 is the mode ( amplitude ) damping rate , and @xmath48 and we have used the normalization @xmath49 . the quantity @xmath50 ( called the `` tidal overlap integral '' or `` tidal coupling coefficient '' ) directly relates to the tidally excited mode amplitude .
we shall focus on @xmath29 , @xmath1 and @xmath51 modes in the following ( although we will continue to use the notations @xmath40 and @xmath41 so that it would be easy to generalize to high - order tides ) .
the general solution equation ( [ eq : adot ] ) is c_(t)=e^-i_t-_t _
t_0^t e^i_t+_t - imdt , assuming @xmath52 , where _ = _
+ m_s is the mode frequency in the inertial frame .
let @xmath53 ( with @xmath54 ) be the times at apastron .
after the @xmath55th periastron passage , the mode amplitude becomes @xmath56,\end{aligned}\ ] ] with [ deltacalpha ] c_=_-p/2^p/2dt e^i_+_t - im . for @xmath57 , the steady - state mode energy in the inertial frame [ ealpha ] e_=2__|c_|^2 becomes ( lai 1997 ) ) was derived in lai ( 1997 ) in an approximate manner ( since mode decomposition was not done rigorously ) , and physical arguments were used to get rid of a fictitious term . ]
@xmath58 where the second equality assumes @xmath59 . here
@xmath60 is the energy transfer to the mode in the `` first '' periastron passage : e_=gm^2r(rd_p)^2(l+1 ) |q_k_lm(_)|^2 , [ eq : dealpha ] where @xmath61 is the periastron distance ( @xmath62 is the orbital semi - major axis ) and [ k ] k_lm(_)=w_lm2_-p/2^p/2dt ( rd)^l+1 e^i_t - im . note that in equation ( [ eq : dealpha ] ) , both @xmath50 and @xmath63 are dimensionless ( in units such that @xmath64 ) .
equation ( [ eq : ealpha ] ) shows that when @xmath65 is not close to @xmath66 ( where @xmath67 is an integer ) , the steady - state mode energy is approximately @xmath60 .
thus @xmath60 serves as a benchmark for the non - resonant mode energy .
equation ( [ eq : ealpha ] ) provides a simple relationship between the actual mode energy @xmath68 and the non - resonant mode energy @xmath60 .
we construct an @xmath69 stellar model using the mesa code ( paxton et al .
we assume solar metallicity and evolve the star until its radius reaches @xmath70 .
these parameters are close to star @xmath71 in koi-54 .
figure [ plot0 ] displays a propagation diagram for our stellar model .
the star has a small convective core inside radius @xmath72 .
we make sure that the stellar model has thermodynamically consistent pressure , density , sound speed and brunt - vsl frequency profiles . we have computed the @xmath29 adiabatic g - modes for this non - rotating stellar model , including @xmath35 , @xmath50 , and the mode mass @xmath73 . here
, the magnitude of the surface displacement is defined by @xmath74 where the @xmath75 and @xmath76 subscripts denote the radial and horizontal components of the displacement vector , respectively .
we use equation ( [ eq : dealpha ] ) to compute the non - resonant mode energy @xmath77 .
the corresponding surface displacement @xmath78 is then obtained from figure [ plot1 ] shows the energy @xmath60 and surface amplitude of the radial component of the displacement , @xmath79 , of tidally excited modes away from resonances for the koi-54 parameters .
the radial displacement is directly related to the flux variation due to the oscillation mode ( see section 4 ) .
the most energetic modes have frequencies @xmath80 , where @xmath81 is the orbital angular frequency , depending on the stellar spin rate and the value of @xmath41 .
low - order ( high frequency ) modes have larger values of @xmath50 but have smaller values of @xmath82 , so medium - order ( @xmath83 ) modes have the largest values of @xmath60 .
, thus @xmath0 modes with @xmath84 have the largest values of @xmath63 . for @xmath1 , modes with @xmath85 have the largest values of @xmath63 . ]
figure [ plot1 ] also shows the magnitude of the displacement @xmath86 for @xmath0 modes in the zero spin limit .
the total displacement is much larger than the radial displacement for low - frequency modes because these modes are characterized by large horizontal displacements and are concentrated near the surface .
consequently , these modes have lower mode mass and the maximum of @xmath86 shifts to lower frequencies compared to the maximum of @xmath60 .
the rotation rates of the koi-54 stars are unknown .
spectroscopic observations constrain @xmath87 km s@xmath88 , corresponding to @xmath89 km s@xmath88 if the spin inclination angle @xmath90 equals @xmath91 ( welsh et al .
this implies that the spin period @xmath92 days and that @xmath93 .
although the classical equilibrium tide theory ( e.g. , hut 1981 ) is not expected to be valid for our system ( lai 1997 ) , we can adopt the pseudosynchronous rotation frequency @xmath94 [ corresponding to @xmath95 as a fiducial value . to account for the effect of stellar rotation on the tidally excited modes , we adopt the perturbative approximation , valid when @xmath25 is less than @xmath96 ( the mode frequency in the zero - rotation limit ) .
the mode wave functions and @xmath50 are unchanged by the stellar rotation , while the mode frequencies are modified according to @xmath97 and @xmath98 , where @xmath99 is a constant ( e.g. unno et al .
1989 ) our calculation gives @xmath100 for all relevant modes .
note that in this approximation , @xmath101 .
more accurate results can be obtained using the method of lai ( 1997 ) .
stellar rotation increases the inertial - frame frequency @xmath102 of the @xmath103 modes , causing the dominant modes to have higher observed frequencies .
however , rotation also shifts the tidal response to higher order g - modes with smaller values of @xmath50 , so we expect rotation to lower the mode energies .
figure [ plot1 ] confirms that finite ( prograde ) stellar rotation indeed shifts the dominant mode energy and surface amplitude to higher - order g - modes , giving rise to smaller @xmath60 and @xmath79 .
the somewhat erratic features of @xmath60 and @xmath104 as seen in figure [ plot1 ] for high - order ( low frequency ) modes are due to mode trapping effects created by the thin sub - surface convection zone in our stellar model . for
these high - order modes , the tidal overlap integrals ( @xmath50 ) depend on the precise shape of the mode wave function , and they can be easily affected by the detailed properties of the stellar envelope .
care must be taken in order to obtain reliable tidal overlap integrals for these high - order modes ( see fuller & lai 2010 ) .
the damping rate of a mode , @xmath47 , can be estimated in the quasi - adiabatic limit via _ ^r_+_r_- k_r^2 ( _ r^2 + l(l+1 ) _
^2 ) r^2 dr , [ gamma ] where @xmath105 is the thermal diffusivity , @xmath106 is the local radial wave number , and @xmath107 and @xmath108 are the boundaries of the mode s propagation cavity . equation ( [ gamma ] ) can be derived in the wkb limit from the quasi - adiabatic work function of a mode ( unno et al .
1989 ) . ) to estimate the mode damping rate .
the bottom panel of figure [ plot1 ] was added after we saw the burkart et al . paper . ]
the bottom panel of figure [ plot1 ] shows the values of @xmath47 calculated via equation ( [ gamma ] ) for the modes of our stellar model .
lower frequency g - modes
have larger damping rates due to their larger wavenumbers and because they propagate closer to the surface of the star where the diffusivity is larger . equation ( [ gamma ] ) provides an estimate for mode damping rates via radiative diffusion in the quasi - adiabatic limit ; however , fully adiabatic oscillation equations must be solved for low - frequency modes for which @xmath47 is comparable to the mode frequency .
furthermore , modes may also damp via nonlinear processes ( see section [ nonlinear ] ) .
in section [ general theory ] , we considered the tidal response to be composed of the sum of the star s natural oscillation modes , each having its own steady - state energy @xmath68 .
this provides a simple relation between the resonant mode energy and non - resonant energy .
the observed magnitude variation of koi-54 has over 20 components with frequencies that are almost exact multiples of the orbital frequency , i.e. , they have @xmath109 with @xmath110 . in this section ,
we examine the tidal response as a sum of oscillations at exact multiples of the orbital frequency .
each component @xmath111 of the tidal potential can be decomposed as [ u2 ] u_lm = - y_lm(,_i ) _ n=-^ f_nm e^-int , where @xmath112 is defined by the expansion ( ad)^l+1 e^-im = _ n=-^ f_nm e^-int and is given by [ fn ] f_nm = ^_0 d. here , @xmath62 is the semi - major axis of the orbit , and @xmath113 is the eccentric anomaly .
note that @xmath112 is related to @xmath63 [ see equation ( [ k ] ) ] by @xmath114 . inserting equation ( [ u2 ] ) into equation ( [ eq : adot ] ) yields
@xmath115 whose non - homogeneous solution is [ csol ] c_(t ) = _ n=-^ .
the total tidal response is @xmath116 , where @xmath117 .
when the displacements are expressed in terms of @xmath118 ( the position vector in the inertial frame , with @xmath119 ) , we have @xmath120 + c.c.\end{aligned}\ ] ] where @xmath121 denotes the complex conjugate , and @xmath122 implies that the sum is restricted to modes with @xmath123 ( including both @xmath103 and @xmath124 modes ) ( by contrast , @xmath125 includes both @xmath123 and @xmath126 terms , as well as both @xmath103 and @xmath124 ) .
we have omitted the @xmath127 term for simplicity because it is not part of the dynamical response .
each @xmath128 ( oscillating at frequency @xmath129 ) is then a sum over the star s oscillation modes , with large contributions coming from nearly - resonant modes with @xmath130 .
we use the result of buta & smith ( 1979 ) ( see also robinson et al . 1982 ) to estimate the magnitude variation @xmath131 associated with each component of the tidal response .
the magnitude variation of the star has two primary components : a geometrical component , @xmath132 , due to distortions in the shape of the star , and a temperature component , @xmath133 , due to perturbations in the surface temperature of the star s photosphere . for an @xmath134 mode with surface radial displacement @xmath135 ,
the amplitudes of the bolometric magnitude variations are @xmath136 \frac{{\xi_r}_\alpha(r)}{r } \end{aligned}\ ] ] and [ deltam2 g ] ( ) _ g^(m=2 ) -0.42(_l + _ l ) ^2 i_s . here
, @xmath137 is the adiabatic index of gas at the surface of the star , @xmath138 is the dimensionless mode frequency , and @xmath139 and @xmath140 are bolometric limb darkening coefficients appropriate for a stars . for koi-54 , the value of the spin inclination angle @xmath90 ( the angle between the spin axis and the line of sight ) is unknown , but we may use the system s orbital inclination of @xmath141 as a first guess , although it is possible that the star s spin axis is inclined relative to the orbital angular momentum axis . similarly , for @xmath29 , @xmath1 modes , the amplitudes of the magnitude variations are @xmath142 \frac{{\xi_r}_\alpha(r)}{r}\end{aligned}\ ] ] and [ deltam0 g ] ( ) _ g^(m=0 ) -0.34 ( _ l + _ l ) ( 3 ^2 i_s -1 ) . in equations ( [ deltam2 t ] ) and ( [ deltam0 t ] ) ,
the factor of @xmath143 $ ] arises from the outer boundary condition formulated by baker & kippenhahn ( 1965 ) ( see also dziembowski 1971 ) , in which the radial derivative of the lagrangian pressure perturbation vanishes at the outer boundary ( in our case , the stellar photosphere ) , such that the mode energy vanishes at infinity in an isothermal atmosphere .
it is not clear how well this boundary condition applies for a real star , nor how it should be modified when non - adiabatic effects are taken into account . for the @xmath144-modes considered here , the factor @xmath145 , causing the temperature effect to overwhelm the geometrical effect for magnitude variations .
the magnitude variation for each frequency @xmath129 is a sum of the variations due to individual modes : @xmath146 + c.c.,\end{aligned}\ ] ] where @xmath147 is the constant in front of @xmath148 in equations ( [ deltam2t])-([deltam0 g ] ) , and @xmath149 . using the values of @xmath150 , @xmath50 , @xmath151 , and @xmath47 calculated in section [ general theory ]
, we compute each term @xmath152 in equation ( [ deltamn ] ) .
figure [ plot2 ] shows a plot of the magnitude variation as a function of @xmath67 , along with the observed magnitude variations in koi-54 as determined by welsh et al .
( 2011 ) . to make this plot ,
we have subtracted out the contribution from the equilibrium tide , because the equilibrium tide is responsible for the periodic brightening of koi-54 near periastron and was subtracted out by welsh et al .
( 2011 ) in order to obtain the magnitude variations due to tidally - induced stellar oscillations . the equilibrium tide can be computed by taking the limit @xmath153 , i.e. , by setting @xmath154 for each term inside the sum in equation ( [ deltamn ] ) .
we adopt a spin inclination angle @xmath155 .
figure [ plot2 ] includes the magnitude variations due to the temperature effect and those due to geometrical effects .
for the adopted spin inclination ( @xmath156 ) , the @xmath1 modes dominate the observed magnitude variations , although a nearly - resonant @xmath0 mode can be visible .
however , it is obvious from figure [ plot2 ] that our computed variations due to the temperature effect of @xmath1 modes are appreciably larger than those observed .
in fact , the magnitude variations due to geometrical effects reproduce the observed variations much more accurately .
the over - prediction of the magnitude variations from the temperature effect is mostly likely due to non - adiabatic effects in the stellar atmosphere , which renders equations ( [ deltam2 t ] ) and ( [ deltam0 t ] ) inaccurate for low - frequency modes .
indeed , buta & smith ( 1979 ) also found that for main sequence b stars , the predicted magnitude variations due to the temperature effect for low - frequency modes were much larger than what was observed , and they speculated that the mismatch was due to non - adiabatic effects in the outer layers of the star .
our results depicted in figure [ plot2 ] suggest that most of the observed magnitude variations ( with the exception the highest - frequency modes ) in koi-54 are due the geometrical surface distortions produced by @xmath1 modes that happen to be nearly resonant with a harmonic of the orbital frequency .
the actual magnitude variations due to the temperature effect ( when non - adiabatic effects are properly taken into account near the stellar photosphere ) should not be much larger than the geometric effect .
note that in figure [ plot2 ] we have plotted the absolute values of the magnitude variations for the temperature and geometrical effects , but that the temperature and geometrical effects have opposite sign and tend to cancel each other out . to produce
the highest - frequency modes ( @xmath19 and @xmath17 ) observed in kio-54 , in figure [ plot2 ] we have fine - tuned the spin of the star to @xmath157 such that the @xmath158 oscillation is nearly resonant with one of the star s oscillation modes . in section 5
we argue that the @xmath158 and @xmath159 oscillations are likely due to an @xmath0 mode in each star that is locked in resonance .
the fine - tuning produces a magnitude variation due to the temperature effect that is comparable to one of the observed oscillations , although the predicted geometrical magnitude variation is somewhat less than what is observed .
if the @xmath158 and @xmath159 oscillations are indeed due to resonant @xmath0 modes , then it is likely that the predicted magnitude variation due to the temperature effect ( based on adiabatic approximation ) is more accurate for these modes due to their higher frequencies .
this is not unreasonable since non - adiabatic effects in the stellar photosphere are expected to be less important for low - order ( high frequency ) modes , a point also emphasized by buta & smith ( 1979 ) . in any case , a more in - depth analysis of the non - adiabatic oscillation modes ( particularly their flux variations ) for the stars of the koi-54 system is needed .
in the previous sections we proposed that a resonance with @xmath160 creates the largest observed magnitude variations in koi-54 . here
we study the secular evolution of the stellar spin and binary orbit , and how resonances may naturally arise during the evolution .
several aspects of tidal resonance locking in massive - star binaries were previously explored by witte & savonije ( 1999 , 2001 ) .
dynamical tides lead to spin and orbital evolution , with the orbital energy and angular momentum evolving according to [ dedj ] e_=-_2_e _ , j_=-_2_e _ , where we have used the fact that the mode angular momentum ( in the inertial frame ) is related to the mode energy by @xmath161 .
the orbital frequency @xmath81 therefore evolves as = _ , where f_=^-1 , and @xmath162 specifies the orbital decay rate due to a single non - resonant mode
@xmath163 : @xmath164 using the koi-54 parameters and assuming @xmath165 , we have t_d^-1 8.910 ^ -10_()^2 .
resonance occurs when a mode has frequency @xmath166 for an integer value of @xmath67 .
we first consider how likely it is to observe a mode near resonance when @xmath102 is held constant ( i.e. , it does not change during the period of resonance - crossing ) . as can be seen from figure [ plot1 ] ,
the @xmath0 modes dominate tidal energy transfer in the koi-54 system , yet ( as evidenced from figure [ plot2 ] ) , most of the visible modes ( except @xmath167 ) are @xmath1 modes . consider a particular @xmath1 mode near resonance , with @xmath168 and @xmath169 . since the mode contributes little to the tidal energy transfer ,
the probability of being close to resonance ( @xmath170 ) is simply @xmath171 .
figure [ plot2 ] indicates that an @xmath1 mode with @xmath172 requires @xmath173 to be visible [ mode visibility scales as @xmath174 .
in each star there are about forty @xmath1 modes in this frequency range , thus we should expect about 8 of these modes from each star to be visible , in rough agreement with figure [ plot2 ] and the observations . now let us consider modes that significantly influence the tidal energy transfer ( these include @xmath0 modes such as the @xmath167 modes , but may also include @xmath1 modes very near resonance if they occur ) . for the koi-54 system ,
the modes that contribute significantly to the tidal energy transfer have @xmath175 , thus @xmath176 .
consider a particular mode near resonance , and suppose that the tidal energy transfer is dominated by the resonant mode ( @xmath163 ) .
the orbital decay rate during the resonance is given by @xmath177^{-1}$ ] .
thus the time that the system spends `` in resonance '' ( @xmath170 ) is @xmath178 \sim 8 \pi^2 t_{d \alpha } |\epsilon_0|^3 /(3n)$ ] .
by contrast , the time it takes the orbit to evolve between resonances ( for the same mode ) is @xmath179 , where @xmath180 is the orbital decay timescale due to all non - resonant modes ( @xmath180 may be a factor of @xmath181 smaller than @xmath182 ) .
the probability of seeing a mode very near resonance is thus p_res = |_0|^3 .
[ probability ] figure [ plot2 ] indicates that ( for in inclination of @xmath155 ) we require @xmath183 for an @xmath0 mode to be visible , for which the probability is @xmath184 .
thus , at first glance , the chances of observing even a single tidally - excited @xmath0 mode in the koi-54 system are slim .
the @xmath167 modes require @xmath183 , even if they are produced by @xmath1 modes .
it is therefore extremely unlikely to observe the system with such large amplitude modes , unless ( by some mechanism ) they are locked into resonance . in the sections that follow ,
we outline such a resonance locking mechanism and how it applies to the koi-54 system .
we now consider the possibility of resonance locking for the @xmath0 modes .
tidal angular momentum transfer to the star increases the stellar spin @xmath25 , thereby changing the mode frequency @xmath102 . there exists a particular resonance , @xmath185 , for which @xmath102 and @xmath81 change at the same rate , i.e. , = , so that the resonance can be maintained once it is reached . with @xmath186 and the mean stellar rotation rate @xmath187 ( where @xmath188 is the moment of inertia ) , we have @xmath189 , where @xmath190 . assuming that a single resonant mode dominates the tidal energy and angular momentum transfer , we find @xmath191 .
thus ( _ _ ) _ tide = ( ) ^2()_tide , [ eq : rates]where [ nc ] n_c = m()^1/2 ( with @xmath192 the reduced mass of the binary ) corresponds to the critical resonance - locking mode .
the subscript `` tide '' in equation ( [ eq : rates ] ) serves as a reminder that the changes of @xmath102 and @xmath81 are due to the tidal interaction . for the koi-54 parameters ( with @xmath193 and @xmath0 )
, we have [ nc2 ] n_c 131(b_0.84)^1/2 ( 0.05)^-1/2 , where @xmath194 .
if we assume that the star maintains rigid - body rotation during tidal spin up , then @xmath195 . numerical calculation for our stellar model gives @xmath196 ( with very weak dependence on modes ) and @xmath197 , giving @xmath198 ; a less evolved star would have @xmath199 , giving @xmath200 .
note that in the above we consider only tides in one star ( @xmath20 ) when tides in the other star are considered , @xmath201 would be larger and the effective value of @xmath202 would be reduced .
if identical resonances occur in the two stars , the effective value of @xmath202 would be reduced by a factor of @xmath203 to @xmath204 ( for @xmath205 ) .
see section 5.3.3 and equations ( [ nc_1])-([nc_2 ] ) below .
the above consideration assumes that the spin evolution of the star is entirely driven by tide . in reality
, the star can experience intrinsic spindown , either due to a magnetic wind or due to radius expansion associated with stellar evolution .
we may write @xmath206 , where the second term denotes the contribution due to the intrinsic spindown . the corresponding rate of change for @xmath102 is then = ( _ _ ) _ tide -|__|_sd .
[ eq : sigdot]on the other hand , the orbital evolution remains solely driven by tides : = ( ) _ tide = ( ) ^2 ( _ _ ) _ tide .
[ eq : omegadot]comparing equations ( [ eq : sigdot ] ) and ( [ eq : omegadot ] ) , we see that _ @xmath202 represents the upper boundary for resonance locking : _ for @xmath207 , resonance locking ( @xmath208 ) is not possible . for @xmath209 ,
resonance locking can be achieved : the value of @xmath210 depends on the closeness to the resonance , which in turn depends on the intrinsic spindown timescale of the star [ see equations ( [ eq : falpha ] ) and ( [ eq : epsiloneq ] ) below ] . the fact that @xmath202 naturally falls in the range close to @xmath158 and @xmath159 , the most prominent modes observed in koi-54 , is encouraging . in the next subsection
, we consider how the system may naturally evolve into a resonance - locking state with @xmath209 .
as noted above , even in the absence of tides , the star can experience spin - down , either due to magnetic wind or due to radius expansion associated with stellar evolution .
furthermore , the intrinsic mode frequencies will change as the internal structure of the star changes due to stellar evolution .
we now show that these effects can naturally lead to evolution of the system toward resonance locking .
the evolution equation for the stellar spin reads @xmath211 , where @xmath212 is the `` intrinsic '' stellar spin - down time scale associated with radius expansion and/or magnetic braking . using equation ( [ dedj ] ) , we have [ doms ]
_ s_s=_(ma^23i ) ( ^2_s _ ) f_t_d-1t_sd . from equation ( [ dedj ] ) , we find that the orbital frequency and eccentricity evolve according to [ dom ] = _ , [ de ] ee1-e^2=-13 _ . to leading order in @xmath25 ,
the mode frequency depends on @xmath25 via [ dsig ] _ = _ ^(0)+mb__s . for simplicity ,
we assume that the secular change of @xmath102 is only caused by the @xmath25-evolution ( e.g. , we neglect the variation of @xmath96 due to stellar evolution
this effect can be absorbed into the spindown effect on the mode ; see witte & savonije 1999 ) . to gain some insight into the evolutionary behavior of the system , we first consider the case where one of the modes is very close to resonance , i.e. , _ = ( n+ ) , [ epsilon ] while all the other modes are non - resonant .
we then write the orbital and spin frequency evolution equations as [ dom2 ] = f_t_d+1t_d , [ doms2 ] _ s_s = n_c^2mb_^2_s_f_t_d-1t_sd + 1t_su . here
@xmath213 and @xmath214 are the orbital decay rate and spin up rate due to all the non - resonant modes .
( see lai 1997 ) .
] they are approximately given by @xmath215 where @xmath216 and @xmath217 are the energy and angular momentum transferred from the orbit to the star in the `` first '' periastron passage ( see section [ general theory ] ) , and @xmath218 is the characteristic mode damping rate of the most important modes in the energy transfer . since @xmath219 ( where @xmath220 is the orbital frequency at periastron )
, we find ~a^23i^2_s_p .
note that @xmath213 can be a factor of a few ( @xmath181 ) larger than @xmath162 , and @xmath214 is a factor of a few larger than @xmath213 .
equations ( [ epsilon])-([doms2 ] ) can be combined to yield the evolution equation for @xmath221 : = -1t_d , [ eq : epdot]where = mb__sn(1t_sd -1t_su)+1t_d .
equation ( [ eq : epdot ] ) provides the key for understanding the condition of achieving mode locking : \(i ) for @xmath207 : the rhs of equation ( [ eq : epdot ] ) is always negative , and the system will pass through the resonance ( @xmath222 ) without locking .
physically , the reason is that for @xmath207 , the orbit decays faster ( during resonance ) than the mode frequency can catch up , so the system sweeps through the resonance .
\(ii ) for @xmath209 and @xmath223 : starting from a non - resonance initial condition ( @xmath224 , or @xmath225 ) , the system will evolve toward a resonance - locking state ( @xmath226 ) , at which f_=t_d_n t_d1 [ eq : falpha ] where @xmath227
. the `` equilibrium '' value of @xmath228 is given by _
[ eq : epsiloneq ] this resonance - locking state can be achieved when [ saturation ] t_d_n
/ t_d>(_p)^2 , as otherwise resonance `` saturation '' [ @xmath229 occurs , and the system will sweep through the resonance .
note that during the resonance - locking phase , the stellar spin increases as [ domsres ] _
s_s=1_n ( t_sd^-1-t_su^-1)+n_c^2n_n mb__s t_d^-1 , and the orbital frequency increases as [ domres ] = ( t_sd^-1 - t_su^-1 ) + t_d^-1 .
the above analysis assumes @xmath230 . in the absence of the intrinsic stellar spin - down ( i.e. , @xmath231 )
, we have t_d^-1~-n_c^2mn(20_p ) t_d^-1+t_d^-1 .
thus @xmath232 can be negative for @xmath233 .
in this case , one may expect mode - locking for @xmath207 .
nevertheless , @xmath234 is only moderately larger ( a factor of 10 or so ) than @xmath162 without intrinsic spin - down , so a close resonance with @xmath235 ( or @xmath169 ) is unlikely . above ( and in other sections of this paper ) , we considered the frequency of a mode in the inertial frame to first order in @xmath25 : @xmath236 . in this approximation ,
@xmath1 modes have constant frequency and thus can not lock into resonance .
however , lai ( 1997 ) find that to second order in @xmath25 , the frequency of @xmath1 modes can be approximated by [ sigm0 ] 1 + ( ) ^2 .
recall that @xmath150 is the mode frequency in the absence of rotation .
thus , to second order , the frequency of @xmath1 modes can change and resonance locking is possible .
let us examine the scenario in which an @xmath237 mode with @xmath238 is locked in resonance , as we suspect is the case for koi-54 . assuming this mode dominates the tidal interaction , the system evolves such that @xmath239 then there will be a value of @xmath240 for an @xmath1 mode such that [ dm0 ] n_m=0 = _ , m=0 . using equation ( [ dm2 ] ) for the lhs of equation ( [ dm0 ] ) , equation ( [ sigm0 ] ) for the rhs of equation ( [ dm0 ] ) , and the condition @xmath241 , we find [ nm0 ] n_m=0 = ^1/2 . using @xmath242 , @xmath237 , @xmath243 , and @xmath244 , we find that @xmath245 .
thus , an @xmath1 mode with @xmath246 may be able to stay close to resonance for an extended period of time .
koi-54 has two very visible modes at @xmath247 and @xmath248 .
we speculate that these modes may correspond to an @xmath1 mode in each star that is nearly locked in resonance due to the orbital evolution produced by a locked @xmath0 mode with @xmath249 .
however , because the value of @xmath240 depends on @xmath25 ( which is currently unknown for koi-54 ) , we will not consider @xmath1 mode locking in the remainder of this paper . in section 5.3.1 ,
we considered resonant locking of an @xmath0 mode in the primary star @xmath20 .
since the two stars in the koi-54 system are quite similar , resonant locking may be achieved in both stars simultaneously .
we now consider the situation in which an @xmath0 mode @xmath102 in star @xmath20 and an @xmath250 mode @xmath251 in star @xmath22 are both very close to orbital resonance , i.e. , _
= ( n+),_=(n+ ) , while all other modes ( in both stars ) are non - resonant . here
, all primed quantities refer to star @xmath22 , and unprimed quantities refer to star @xmath20 .
the orbital evolution equation then reads = f_t_d + f_t_d+1t_d , while the spin evolution is governed by equation ( [ domsres ] ) for star @xmath20 and a similar equation for @xmath22 .
we then find @xmath252 { f_\alpha\over t_{d\alpha}}-{1\over t_d},\\ & { \dot\epsilon'\over n'}=\left[\left({n_c'\over n'}\right)^2 - 1-{1\over x}\right ] { f_{\alpha'}\over t_{d\alpha'}}-{1\over t_d'},\end{aligned}\ ] ] where x = e_e _ , with @xmath253 and @xmath254 the energy dissipation rates ( including resonances ) due to the resonant modes in star @xmath20 and @xmath22 , respectively .
thus , analogous to section 5.3.1 , for @xmath255 and @xmath256 , the necessary conditions for resonant mode locking in both stars are n < n_c , eff , n<n_c , eff , with [ nc_1 ] n_c , eff = n_c = m^1/2 , [ nc_2 ] n_c , eff=n_c=m ^1/2 . obviously , if the two stars have identical resonances ( @xmath257 ) , then @xmath258 would be smaller than @xmath202 by a factor of @xmath203 .
if the energy dissipation rates in the two stars differ by at least a factor of a few ( e.g. , @xmath259 ) , then @xmath258 is only slightly modified from @xmath202 , while @xmath260 will be a factor of a few smaller than @xmath261 . for a given set of modes , the solution of the system of equations ( [ doms])-([de ] ) depends on the dimensionless parameters @xmath262 , @xmath202 and @xmath263 , as well as the initial conditions . in general , these parameters change as the system evolves .
-2.5 cm
-2.5 cm to illustrate the essential behavior of the secular evolution toward resonance locking , we first consider the simple case where @xmath264 and @xmath263 are assumed to be constant and identical for all modes considered . figure [ evolve ] depicts an example : we include two @xmath0 modes ( @xmath265 and @xmath266 ) , both have @xmath267 , @xmath268 and @xmath269 .
the parameter @xmath202 has an initial value of @xmath270 , but we allow @xmath202 to evolve via @xmath271 . at @xmath272 ,
the spin frequency is @xmath273 , and the two modes have @xmath274 and @xmath275 ( i.e. , both are initially `` off - resonance '' ) .
we see that the stellar spindown first causes @xmath265 to lock into resonance at @xmath276 .
the star then spins up , driven by the resonant tidal torque .
the second mode passes through the @xmath158 resonance , but it produces a negligible effect on the @xmath276 resonance . in the meantime , the orbit decays , reducing @xmath202 .
when @xmath202 becomes less than @xmath277 , the resonance locking can no longer be maintained , and @xmath265 breaks away from the @xmath278 resonance .
the stellar spin then decreases ( due to the @xmath279 term ) , which leads to @xmath266 to capture into the @xmath158 resonance . eventually , as @xmath202 decreases ( due the resonant tidal torque ) below @xmath280 , the second mode breaks away from the @xmath158 resonance .
this example corroborates our analytical result of section 5.3 , and demonstrates that resonance locking can indeed be achieved and maintained for an extended period during the evolution of the binary system .
having demonstrated how resonance locking can occur in a simplified system consisting of only two oscillation modes , we now investigate how resonance locking is likely to occur in the actual koi-54 system .
we solve the orbital evolution equations ( [ doms])-([dsig ] ) using the actual values of @xmath150 , @xmath50 , and @xmath47 found in section [ modes ] to calculate each @xmath182 .
we include eighteen modes in our equations : they are the @xmath29 , @xmath1 , @xmath281 modes for six values of @xmath150 such that the initial frequencies of the @xmath0 modes range from @xmath282 to @xmath283 .
we allow the value of @xmath63 to evolve as the values of @xmath102 , @xmath81 , and @xmath284 change with time .
figures [ evolution ] and [ evolution2 ] display the evolution of @xmath81 , @xmath25 , @xmath202 , and the values of @xmath102 and @xmath68 for a selected sample of modes over a time span of tens of millions years .
we use initial values of @xmath285 ( here @xmath286 is the observed orbital frequency of koi-54 ) , @xmath287 , @xmath288 , and a spindown time of @xmath289 years . in figure
[ evolution2 ] , we have doubled the orbital decay rate to account for an equal amount of energy dissipation in the companion star .
let us start by examining figure [ evolution ] .
the system quickly enters a resonance locking state with the @xmath290 mode , where the notation @xmath291 identifies the mode with the @xmath55th largest frequency in our simulation with azimuthal number @xmath41 .
the resonance locking lasts for over 100 million years , until the @xmath292 mode reaches resonance .
the @xmath292 mode then locks in resonance for over 100 million years . during resonance
locking , the orbital frequency and spin frequency increase rapidly .
the energies @xmath68 of the resonant locked modes can exceed @xmath293 their non - resonant values @xmath60 , corresponding to an increase in visibility of over 25 times the visibility during a non - resonant state .
when no mode is locked in resonance , the orbital evolution is relatively slow , with the dominant effect being the spindown of the star due to the @xmath294 term . for the initial conditions and orbital evolution depicted in figure [ evolution ] , only two of our 18 modes pass through resonant locking phases : two @xmath0 modes with initial frequencies of @xmath295 and @xmath296 .
the highest frequency @xmath0 mode included in our evolution does not lock in resonance because it has @xmath207 at all times .
the lowest frequency modes do not lock in resonance because they do not satisfy equation ( [ saturation ] ) , i.e. , they become saturated before they can begin to resonantly lock .
this occurs because the value of @xmath50 for these high - order modes is small , resulting in a small orbital decay rate @xmath297 for these modes .
which mode locks into resonance is dependent on the initial conditions .
however , as can be seen in figure [ evolution ] , the resonance with the @xmath292 mode ends the resonance with the @xmath290 mode because the @xmath292 mode has a larger value of @xmath298 and will thus dominate the orbital decay .
therefore , we can expect the system to evolve into a state in which the mode with the largest value of @xmath68 ( and with @xmath209 ) is locked in resonance . in figure [ evolution ] , the resonance locking with the @xmath292 mode is ended by a resonance with the @xmath299 mode .
the @xmath299 mode does not lock into resonance because an @xmath1 mode can not change the stellar spin frequency .
however , its resonance causes the value of @xmath300 to exceed the value of @xmath301 for the locked mode , so that the system sweeps through the resonance .
in other words , the non - locked resonant mode temporarily decreases the value of @xmath180 , decreasing the value of @xmath232 such that resonant saturation occurs for the locked mode , causing it to pass through resonance .
also note that even though the @xmath299 mode does not lock into resonance , it maintains a relatively large energy ( @xmath302 ) for a period of over 50 million years .
this is indicative of the pseudo - resonance locking for @xmath1 modes described in section [ m0 ] .
let us now examine figure [ evolution2 ] , in which the orbital decay rate has been doubled ( i.e. , we multiply @xmath303 by 2 ) to account for an equal tidal response in the companion star .
the results are significantly different : due to the increased orbital decay rate , the initial effective value of @xmath258 has dropped to @xmath304 ( see section 5.3.3 ) so that the @xmath292 mode can no longer lock into resonance .
also , the maximum energy @xmath305 of the modes is larger in this scenario ( because the modes must enter deeper into resonance to become locked ) , so the modes can create larger magnitude variations while being locked in resonance . however , the resonance locking events are shorter because the modes are nearly saturated during resonance locking , allowing resonances with @xmath1 modes or higher frequency @xmath0 modes to easily disrupt the resonance locking state . in the bottom
three panels of figure [ evolution2 ] , we have plotted the energy @xmath68 of the @xmath306 mode . in figure
[ evolution2 ] , three of the resonant locking events for the @xmath290 mode are ended due to a resonance with the @xmath306 mode .
we have examined the results carefully and found that all of the resonance locking events were ended due a resonance with an @xmath1 mode or an @xmath0 mode for which @xmath207 .
although the resonance locking events are brief at the end of the simulation , the locking event with the @xmath307 mode lasts for about 20 million years at the beginning of the simulation when the system s parameters most closely resemble those of koi-54 .
we thus conclude it is likely to observe a system such as koi-54 in a resonance locking state .
we note the secular evolution analysis presented above does not take into account the fluctuation of mode amplitudes due to the changing strength of the tidal potential in an elliptical orbit .
therefore , in addition to the results displayed above , we have also performed calculations of the full dynamic evolution of the stellar oscillations and binary orbit , including the back - reaction of the modes on the orbit .
we find that resonance locking can indeed be maintained for extended periods of time , and that non - secular effects have little impact on the results discussed above .
as discussed in section 4 , in the linear theory , tidally - forced stellar oscillations give rise to flux variations at integer multiples of the orbital frequency .
many of these oscillations have been observed in the koi-54 system .
however , welsh et al . ( 2011 ) also reported the detection of nine modes that do not have frequencies that are integer multiples of the orbital period , and their observation requires an explanation .
one possibility is that one ( or both ) of the stars in koi-54 are @xmath308-scuti variable stars .
the masses , ages , and metallicities of the koi-54 stars put them directly in the instability strip and so @xmath308-scuti - type pulsations are not unexpected .
however , as pointed out by welsh et al .
( 2011 ) , the complete absence of any modes detected with periods less than 11 hours is troublesome for the @xmath308-scuti interpretation , as most observed @xmath308-scuti pulsations have periods on the order of a few hours . one
would then have to explain why only long period ( @xmath309 hours ) @xmath308-scuti - type oscillations are visible in this system .
another possibility is that some ( or all ) of the non - harmonic modes are excited via nonlinear couplings with resonant modes . in particular
, it is thought that three - mode resonant coupling ( parametric resonance ) plays an important role in limiting the saturation amplitudes of overstable g - modes in zz - ceti stars ( wu & goldreich 2001 ) and @xmath308-scuti stars ( dziembowski & krolikowska 1985 ) . in the koi-54 system ,
when a resonantly excited mode reaches sufficiently large amplitude , it will couple to two non - resonant , lower - frequency daughter modes ( see kumar & goodman 1996 ) , thereby explaining the observed non - harmonic oscillations .
parametric resonance can occur when @xmath310 , where @xmath311 is the frequency of the parent mode and @xmath35 and @xmath312 are the frequencies of the two daughter modes .
the additional requirement that @xmath313 , where @xmath41 is the azimuthal number of the mode , implies that @xmath314 .
examination of the koi-54 data given in welsh et al .
( 2011 ) reveals that @xmath315 ( 25.195 @xmath15hz = 6.207 @xmath15hz + 18.988 @xmath15hz ) exactly to the precision of the measurements , where @xmath316 is the frequency of the magnitude oscillation with the @xmath317th largest magnitude variation .
this provides strong evidence that the @xmath318 and @xmath319 oscillations are due to modes excited via parametric resonance and not via direct tidal forcing .
however , no other pair of observed non - harmonic magnitude oscillations have frequencies which add up to that of an observed oscillation .
some of these non - resonant oscillations may be due to modes excited via parametric resonance that have undetected sister modes .
this possibility is especially appealing if one considers the scenario in which one of the two daughter modes has @xmath1 ( and could thus be easily detected for small values of @xmath90 ) , while its sibling has @xmath0 ( and is thus very difficult to detect for small @xmath90 ) .
we therefore suggest that all non - harmonic flux oscillations are produced by nonlinear mode coupling . deeper observations
( with better photometry ) and analysis may reveal additional evidence for nonlinear mode couplings in the koi-54 system
we have shown that many properties of flux oscillations detected in the koi-54 binary system by _ kepler _ can be understood using the theory of dynamical tidal excitations of g - modes developed in this paper .
in particular , our analysis and calculation of the resonance mode locking process , which is driven by dynamical tides and intrinsic stellar spindown , provides a natural explanation for the fact that only those modes with frequencies ( @xmath102 ) less than about 90 - 100 times the orbital frequency ( @xmath81 ) are observed .
our result suggests that the koi-54 system is currently in a resonance - locking state in which one of the stars has a rotation rate such that it possesses a @xmath0 mode with frequency @xmath320 and the other star has a similar mode with @xmath321 these modes produce the largest flux variations detected in koi-54 .
our analysis shows that the system can evolve into and stay in such a resonance locking state for relatively long time periods , and it is reasonable to observe the system in such a state .
other less prominent flux variations at lower frequencies can be explained by the @xmath1 tidally forced oscillations , many of which may be enhanced by the resonant effect .
we have also found evidence of nonlinear three - mode coupling from the published _
kepler _ data and suggested that the nonlinear effect may explain the flux variabilities at non - harmonic frequencies . in our study ,
we have used approximate quasiadiabatic mode damping rates .
obviously , it would be useful to repeat our analysis using more realistic mode damping rates , calculated with the full non - adiabatic oscillation equations .
our calculations of the flux variations of tidally induced oscillations have also highlighted the importance of accurate treatment of non - adiabatic effects in the stellar photosphere .
we have only considered g - modes ( modified by stellar rotation ) in this paper .
our general theory allows for other rotation - driven modes , such as inertial / rossby modes . it would be interesting to study tidal excitations of these modes as well as the nonlinear 3-mode coupling effect in the future .
more detailed modelling of the observed oscillations may provide useful constraints on the parameters of the koi-54 system , particularly the stellar rotation rates and spin inclinations . on a more general level
, koi-54 may serve as a `` laboratory '' for calibrating theory of dynamical tides , which has wide applications in stellar and planetary astrophysics ( see references in section 1 ) . while we believe that our current theory provides the basis for understanding many aspects of the koi-54 observations , some puzzles remain . in our current interpretation
, the most prominent oscillations ( at @xmath322 and @xmath323 ) occur in the two different stars , each having a @xmath0 mode resonantly locked with the orbit . while one can certainly appeal to coincidence ,
it is intriguing that the strongest observed oscillations are at the two consecutive harmonics ( @xmath324 ) of the orbital frequency .
more importantly , as discussed in section 5.3.3 , when both stars are involved in resonance locking , the effective @xmath202 ( above which resonance locking can not happen ) is reduced .
for example , using our canonical stellar parameters ( @xmath205 ) , we find @xmath325 [ see equation ( [ nc2 ] ) ] , but the effective @xmath202 would be reduced to @xmath326 if the two stars have identical modes in resonance locking . in the case where the resonant energy dissipation rates in the two stars differ by a factor of more than a few , the effective @xmath202 of one star would remain close to 131 , while the effective @xmath202 for the other star would likely be reduced to a value below 90 . in this case
, it would be unlikely to find the binary system in the resonance - locking state involving both stars with @xmath327 close to 90 , because it is most likely for modes with @xmath67 just less than @xmath202 to be locked in resonance ( see section 5.4.2 . ) .
thus , we find it likely that the energy dissipation rates are nearly equal ( to within a factor of 2 ) in each star in the koi-54 system .
finally , we may use our results to speculate on the evolutionary history ( and future ) of the koi-54 system .
the star with @xmath328 and radius @xmath329 has an age about @xmath330 years , according to our stellar model generated by the mesa code . during the resonance locking phase ,
the orbital eccentricity decreases on a relatively short time scale ( of order @xmath331 years ) .
the current large eccentricity of koi-54 then suggests that resonance locking has not operated for a large fraction of the system s history .
when resonance locking is in effect , the orbital evolution time scale is set by primarily by the spindown time scale , @xmath294 , and the value of @xmath332 for the resonant mode [ see equation ( [ domres ] ) ] . in the future
, the stars will continue to expand into red giants and the spindown time scale will decrease .
this will cause the orbital evolution time scale to correspondingly decrease , leading to rapid orbital decay and circularization of the system . while our paper was under review , a paper by burkart et al .
( 2011 ) appeared on arxiv .
burkart et al .
carried out non - adiabatic mode calculations and also attribute most of the oscillations to @xmath1 modes , although they do not reach a definite conclusion on the source of the @xmath167 oscillations .
they did not consider resonance locking ( instead they consider the qualitatively different phenomenon of pseudosynchronous locks " ) , and appeared to attribute all resonances to random chance .
they also showed that 3-mode coupling ( see section 6 ) is possible .
this work has been supported in part by the nsf grant ast-1008245 . | recent observation of the tidally - excited stellar oscillations in the main - sequence binary koi-54 by the _ kepler _
satellite provides a unique opportunity for studying dynamical tides in eccentric binary systems .
we develop a general theory of tidal excitation of oscillation modes of rotating binary stars , and apply our theory to tidally excited gravity modes ( g - modes ) in koi-54 .
the strongest observed oscillations , which occur at 90 and 91 times the orbital frequency , are likely due to prograde @xmath0 modes ( relative to the stellar spin axis ) locked in resonance with the orbit .
the remaining flux oscillations with frequencies that are integer multiples of the orbital frequency are likely due to nearly resonant @xmath1 g - modes ; such axisymmetric modes generate larger flux variations compared to the @xmath0 modes , assuming that the spin inclination angle of the star is comparable to the orbital inclination angle .
we examine the process of resonance mode locking under the combined effects of dynamical tides on the stellar spin and orbit and the intrinsic stellar spindown .
we show that koi-54 can naturally evolve into a state in which at least one @xmath0 mode is locked in resonance with the orbital frequency .
our analysis provides an explanation for the fact that only oscillations with frequencies less than 90 - 100 times the orbital frequency are observed .
we have also found evidence from the published _ kepler _
result that three - mode nonlinear coupling occurs in the koi-54 system .
we suggest that such nonlinear mode coupling may explain the observed oscillations that are not harmonics of the orbital frequency .
[ firstpage ] binaries : close stars : oscillations stars : rotation stars : individual : hd187091 ( koi-54 ) |
single molecule magnets ( smm s ) have been under intense study recently , due to their potential uses in magnetic storage and quantum computing.@xcite the materials consist of insulating crystalline arrays of identical smm s 1 - 3 nm in size , each containing two or more magnetic ions .
since the magnetic ions in each smm are surrounded by non - magnetic ligands , the intermolecular magnetic interactions are usually negligible .
although the most commonly studied smm s are the high - spin mn@xmath20 and fe@xmath21,@xcite such smm s contain a variety of ferromagnetic ( fm ) and antiferromagnetic ( afm ) intramolecular interactions , rendering unique fits to a variety of experiments difficult.@xcite in addition , there have been many studies of afm fe@xmath22 ring compounds , where @xmath23 etc.@xcite in these studies , analyses of inelastic neutron diffraction data and the magnetic induction @xmath9 dependence of the low - temperature @xmath3 specific heat and magnetization steps were made , using the isotropic heisenberg near - neighbor exchange interaction , the zeeman interaction , and various near - neighbor spin anisotropy interactions.@xcite however , the rings were so complicated that analyses of the data using those simple models were inaccessible to present day computers.@xcite thus , those authors used either simulations or phenomenological fits to a first - order perturbation expansion with different spin anisotropy values for each global ring spin value.@xcite here we focus on the simpler cases of equal spin @xmath24 magnetic dimers , for which the full spin anisotropy effects can be evaluated analytically , investigated in detail numerically , and compared with experiment .
afm dimers with @xmath25,@xcite and various forms of fe@xmath17 with @xmath16 were studied recently.@xcite several fe@xmath17 dimers and effective @xmath18 dimers of the type [ mn@xmath19@xmath17,@xcite have magnetic interactions weak enough that their effects can be probed at @xmath26 1k with presently available @xmath9 .
a comparison of our results with magnetization @xmath27 versus @xmath9 step data on a fe@xmath17 dimer strongly suggests a substantial presence of local spin anisotropy.@xcite the paper is organized as follows . in section
ii , we present the model in the crystal representation and given exact formulas for the matrix elements .
the general thermodynamics are presented in section ii , along with the universal behavior of the @xmath28 and @xmath29 behavior associated with the energy level crossing . in section iv , we solve the model exactly for @xmath12 , giving analytic expressions for the magnetization and specific heat . in section
v , we discuss the exact solution for @xmath30 , present the equations from which the eigenvalues are readily obtained , and give numerical examples of the low-@xmath3 magnetization and specific heat curves . in section vi , we present numerical examples of the low-@xmath3 @xmath16 magnetization and specific heat curves . in section vii , we rotate to the induction representation , and give the eigenstates to first order in the anisotropy energies . these are used to obtain asymptotic expressions for the magnetization and specific heat for arbitrary @xmath0 that are highly accurate at sufficiently low @xmath3 and/or large @xmath14 . in addition ,
analytic formulas for the magnetic inductions at which the level crossings occur are provided , accurate to second order in the anisotropy energies .
finally , in section viii we discuss our results with regard to experiments on fe@xmath17 dimers .
we represent the @xmath24 dimer quantum states , @xmath31 in terms of the global ( total ) spin and magnetic quantum numbers @xmath6 and @xmath32 , where @xmath33 and @xmath34 satisfy @xmath35 and @xmath36 , where @xmath37 , @xmath38 , and we set @xmath39 .
we also have @xmath40 , where @xmath41 and @xmath42 for an arbitrary @xmath9 , we assume the hamiltonian has the form @xmath43 , where @xmath44 contains the heisenberg exchange and zeeman interactions , the gyromagnetic ratio @xmath45 , where @xmath46 and @xmath47 is the bohr magneton .
the global axial and azimuthal anisotropy terms @xmath48 and @xmath49 respectively , only involve components of @xmath50 , but have been the main anisotropy terms discussed in the smm literature,@xcite so we have included them for comparison .
such terms have been commonly studied as an effective hamiltonian for a singlet orbital ground state , in which the tensor global spin interaction with a fixed spin quantum number @xmath6 has the form @xmath51 , resulting in the principal axes @xmath52 and @xmath53.@xcite for a dimer , we take @xmath54 to be diagonal in the orientation pictured in fig
then @xmath55 , @xmath56 , and @xmath57 .
taking @xmath58 and @xmath59 still leaves @xmath60 unrestricted .
the single - ion axial and azimuthal anisotropy terms , @xmath61 respectively , arise from spin - orbit interactions of the local crystal field with the individual spins .
these terms have usually been neglected in the smm literature , but have been studied with regard to complexes containing a single magnetic ion , such as ni@xmath62,@xcite and with regard to clusters of larger numbers of identical magnetic ions.@xcite the local axially and azimuthally anisotropic exchange interactions @xmath63 satisfy @xmath64 so we need only include either either @xmath65 or @xmath66 and @xmath67 or @xmath68 , respectively.@xcite that is , if we stick to the hamiltonian @xmath69 , we may incorporate the effects of @xmath66 and @xmath68 by letting @xmath70 , @xmath71 , and @xmath72 , @xmath73 , respectively .
since @xmath65 and @xmath67 describe the axial and azimuthal anisotropy each single ion attains from its surrounding environment , they are the physically relevant local anisotropy interactions . for the case of fe@xmath17,@xcite a constituent of the high - spin smm fe@xmath21 and the afm fe@xmath22 rings,@xcite the exchange between the fe@xmath74 @xmath16 spins occurs via two oxygen ions , and these four ions essentially lie in the same ( @xmath75 ) plane.@xcite we set the @xmath53 axis parallel to the dimer axis , as pictured in fig .
1 . since the quantization axis is along the dimer axis , which is fixed in a crystal , we denote this representation as the crystal representation .
we generally expect each of the @xmath76 for @xmath77 to satisfy @xmath78 , but there are not generally any other restrictions upon the various magnitudes of the @xmath76 . since all dimers known to date have predominantly afm couplings ( @xmath79 ) , and also because their magnetizations and specific heats are particularly interesting , we shall only consider afm dimers .
in addition , since no studies on unequal - spin dimers have been reported to our knowledge , we shall only treat the equal - spin @xmath24 case .
we note that for equal spin dimers , the group symmetry of the dimer environment is @xmath80 , so that dzaloshinski - moriya interactions do not arise.@xcite hence , our hamiltonian @xmath69 is the most general quadratic anisotropic spin hamiltonian of an equal - spin dimer .
dimer , with two bridging o@xmath81 ions ( o ) .
ligands ( not pictured ) are attached to the fe@xmath74 ions ( fe ) .
the arrows signify spins.,scaledwidth=15.0% ] for @xmath82 , we have @xmath83 @xmath84 and @xmath68 contain the individual spin operators @xmath85 and @xmath86 for @xmath87 . after some clebsch - gordon algebra involving the wigner - eckart theorem , for arbitrary @xmath88 , @xmath89 where @xmath90 and @xmath91/(4s^2 - 1)}.\label{eta}\end{aligned}\ ] ] for @xmath92 , we require @xmath93 , for which @xmath94 .
we then find @xmath95\alpha_{s , s_1},\label{gss1m}\\ h_{s , s_1}^{m,\sigma'}&=&d_{s+(\sigma'+1)/2,s_1}^md_{s+(3\sigma'+1)/2,s_1}^m,\label{hss1msigma}\\ k_{s , s_1}^{x,\sigma'}&=&c_{s+(\sigma'+1)/2,s_1}^{-\sigma'x-(1+\sigma')/2}c_{s+(3\sigma'+1)/2,s_1}^{-\sigma'x-(3\sigma'+1)/2},\label{kss1xsigma}\\ l_{s , s_1}^x&=&2f_s^x\alpha_{s , s_1},\label{lss1}\\ \noalign{and } \alpha_{s , s_1}&=&\frac{3s(s+1)-4s_1(s_1 + 1)-3}{(2s-1)(2s+3)}.\label{alphass1}\end{aligned}\ ] ] we note that @xmath96 and that eq .
( [ alphass1 ] ) holds for @xmath97 .
equations ( [ sipm ] ) and ( [ siz ] ) allow for an exact solution to the most general hamiltonian of arbitrary order in the individual spin operators . in all previous treatments of more complicated spin systems with similar anisotropic interactions , it was only possible to obtain numerical solutions , and therefore the full anisotropy of the magnetization and specific heat was not calculated.@xcite the operations of @xmath98 , @xmath99 and @xmath100 satisfy the selection rules @xmath101 , @xmath102 . the local anisotropy interactions @xmath65 and @xmath67 allow transitions satisfying @xmath103 , @xmath104 , and @xmath103 , @xmath105 , respectively , so in the presence of either of these interactions , @xmath6 is no longer a good quantum number , unless @xmath12 .
in order to obtain the thermodynamic properties , we first calculate the canonical partition function , @xmath106 . since @xmath69 is not diagonal in the @xmath107 representation , we must construct the wave function from all possible spin states .
we then write @xmath108 where @xmath109 is constructed from the @xmath110 basis as @xmath111 where @xmath112 and @xmath113 is boltzmann s constant . to evaluate the trace , it is useful to diagonalize the @xmath114 matrix . to do so ,
we let @xmath115 , where @xmath116 is constructed from the new orthonormal basis @xmath117 , and @xmath118 is a unitary matrix of rank @xmath119 . choosing @xmath118 to diagonalize @xmath120
, we generally obtain @xmath121 and the partition function for a smm dimer , @xmath122 the specific heat @xmath123 is then easily found at all @xmath124 , @xmath125.\label{sh}\end{aligned}\ ] ] it is easily seen from eq .
( [ sh ] ) that at the induction @xmath126 corresponding to the @xmath6th level crossing in which @xmath127 , @xmath128 as @xmath11 .
the magnetization @xmath129 requires @xmath130 for each @xmath9 . as @xmath11 ,
at most two eigenstates are relevant . for most @xmath14 values ,
only one @xmath131 is important . but near the @xmath6th level - crossing induction @xmath15 at which @xmath127 , two eigenstates are relevant .
we then have @xmath132&{\rightarrow\atop{t\rightarrow0}}&0,\label{cvzeroes}\\ c_v\bigl[b_{s , s_1}^{\rm lc}(\theta,\phi)\pm \frac{2c}{\gamma\beta}\bigr]&{\rightarrow\atop{k_bt\ll|j|}}&c_v^{\rm peak}\label{peak}\\ c_v^{\rm peak}/k_b&= & \bigl(\frac{c}{\cosh c}\bigr)^2\nonumber\\ & \approx&0.439229,\label{peakheight}\\ \noalign{\rm and } m\bigl[b_{s , s_1}^{\rm lc}(\theta,\phi)\bigr]/\gamma&{\rightarrow\atop{t\rightarrow0 } } & s-\frac{1}{2},\label{msteps}\\ \frac{dm}{\gamma db}\bigr|_{b_{s , s_1}^{\rm lc}(\theta,\phi)}&=&\frac{\beta\gamma}{4},\label{slope}\end{aligned}\ ] ] where @xmath133 and @xmath134 is the solution to @xmath135 . the easiest way to obtain eq .
( [ msteps ] ) is to first rotate the crystal so the quantization axis is along @xmath9 , as discussed in appendix b. we note that @xmath136 depends upon @xmath137 , and the direction of @xmath9 when anisotropic interactions are present .
hence , the heights and midpoint slopes of the @xmath5 @xmath28 steps are universal , but the step positions and hence their plateaus are not .
correspondingly , the @xmath5 positions of the @xmath29 zeroes are non - universally spaced , but each zero is equally surrounded by two peaks of equal , universal height , the universal spacing between which is @xmath138 as @xmath11 . hence , the non - universal level - crossing inductions @xmath136 fully determine the low-@xmath3 thermodynamics of afm dimers . in the next three sections
, we consider the special cases of @xmath139 and 5/2 .
then , in section vii and appendix d , we present out general expression for @xmath136 accurate to second order in each of the @xmath76 .
we remark that a double peak in the low-@xmath3 @xmath29 curve has been seen experimentally in a much more complicated fe@xmath140 ring compound , and was attributed to level crossing.@xcite
plots of @xmath141 and @xmath142 versus @xmath143 for the isotropic spin 1/2 dimer were given previously.@xcite for @xmath12 with an arbitrary @xmath9 and @xmath76 for @xmath77 , the rank 4 hamiltonian matrix is block diagonal , since @xmath144 is a good quantum number .
the eigenvalues are given by @xmath145 where @xmath146\nonumber\\ & & -b^2\sin^2\theta[j_b - j_d\cos(2\phi)].\label{halfeigen}\end{aligned}\ ] ] for the special cases @xmath147 for @xmath148 , the @xmath149 satisfy @xmath150 where @xmath151 respectively , and where @xmath152 ( @xmath153 ) corresponds to the upper ( lower ) sign .
the magnetization for @xmath154 is given by @xmath155,\\ \delta_z&=&\exp(-\beta j_b)[\exp(-\beta j)+1],\end{aligned}\ ] ] respectively .
when the interactions are written in terms of the less physical @xmath156 , and @xmath157 , then @xmath158 and @xmath159 , so that for @xmath12 , @xmath160 and @xmath157 merely renormalize @xmath161 and @xmath162 .
neither of the single - ion spin anisotropy terms @xmath65 and @xmath67 affect the thermodynamics for @xmath12 .
we note that @xmath163 for each @xmath14 , as expected from @xmath164 . from eqs .
( [ epsilon1 ] ) , ( [ epsilon234 ] ) , ( [ lambdaz ] ) , and ( [ lambdaxy ] ) , the single level crossing induction ( @xmath165 ) for am @xmath12 dimer with @xmath154 occurs at @xmath166 provided that @xmath167 and @xmath168 , respectively . to distinguish the different effects of the global anisotropy interactions @xmath161 and @xmath162 that affect the magnetization of @xmath12 dimers , in figs . 2 and 3 ,
we have respectively plotted the low-@xmath3 @xmath169 versus @xmath143 with @xmath170 and @xmath171 for @xmath172 ( solid ) and @xmath173 ( dashed ) , along with the isotropic case @xmath174 ( dotted ) . from fig .
2 , @xmath175 and @xmath176 causes a greater shift to higher @xmath14 values at the magnetization step with @xmath177 than for @xmath173 , consistent with eq .
( [ bstephalf ] ) .
in addition , @xmath162 finite with @xmath178 has a very different effect upon the anisotropy of the magnetization step , as shown in fig .
although for @xmath179 , @xmath14 at the step is slightly reduced from its isotropic interaction value , for @xmath173 , the magnetization step occurs at a larger @xmath14 .
these results are consistent with eq .
( [ bstephalf ] ) .
the midpoint slopes are universal , in accordance with eq .
( [ .]slope ) . versus @xmath143 at @xmath180 for the afm spin 1/2 dimer with @xmath181 , @xmath176 , with @xmath177 ( solid ) , @xmath182 (
dashed ) , along with the isotropic case @xmath174 ( dotted).,scaledwidth=45.0% ] versus @xmath143 at @xmath180 for the afm spin 1/2 dimer with @xmath183 , @xmath178 , with @xmath177 ( solid ) , @xmath182 ( dashed ) , along with the isotropic case @xmath174 ( dotted).,scaledwidth=45.0% ] the specific heat of an @xmath12 dimer with @xmath147 is @xmath184 where the @xmath185 are given by eq .
( [ di ] ) , and the @xmath186 are given in appendix a. plots at low @xmath3 of @xmath141 versus @xmath143 for @xmath12 dimers with the corresponding global anisotropies @xmath181 and @xmath183 are shown in figs . 4 and 5 , respectively .
we note the universal curve shapes , but non - universal level - crossing positions , in quantitative agreement with eqs .
( [ cvzeroes])-([peakheight ] ) . in fig .
4 , the positions of the maxima and the central minimum in @xmath10 track that of the magnetization step in fig .
2 with the same parameters . with @xmath183 ,
the behaviors in @xmath10 and @xmath187 for @xmath182 are also very similar .
however , there is a slight difference in the behaviors for @xmath177 .
note that @xmath187 ( fig .
3 ) shows a slight reduction for @xmath177 in the induction required for the step , whereas @xmath10 ( fig .
5 ) shows a slight increase in the positions of the peaks .
this detail only appears when @xmath188 , for which the effective temperature is slightly higher than for @xmath176 . versus @xmath143 for the afm spin 1/2 dimer with @xmath189 at @xmath180 with the same curve notation as in fig .
2.,scaledwidth=45.0% ] versus @xmath143 for the afm spin 1/2 dimer with @xmath183 , @xmath178 at @xmath180 with the same curve notation as in fig .
for dimers with @xmath30 , the allowed @xmath6 values are @xmath190 .
the three @xmath165 states are decoupled from the six remaining @xmath191 states .
they satisfy a cubic equation given in appendix a. for @xmath154 with @xmath148 , this cubic equation simplifies to a linear and a quadratic equation , as for the @xmath165 eigenstates of @xmath12 dimers .
the remaining six eigenstates corresponding nominally to @xmath191 are in general all mixed .
the matrix leading to the hexatic equation from which the six eigenvalues can be obtained is given in appendix a. that is sufficient to evaluate the eigenvalues for @xmath30 using symbolic manipulation software . when combined with the three @xmath165 eigenvalues , one
can then use eqs .
( [ sh ] ) and ( [ mag ] ) to obtain the resulting exact magnetization and specific heat at an arbitrary @xmath9 .
the combined nine eigenvalues depend upon all four anisotropy parameters @xmath76 for @xmath77 . to the extent that the eigenvalues can be obtained from the solutions to either linear or quadratic equations ,
the expressions for the @xmath192 are simple , and are given in appendix a. in the global anisotropy case @xmath193 , the first level crossing induction @xmath194 with @xmath154 is identical to @xmath195 , the level crossing with @xmath12 given by eq .
( [ bstephalf ] ) . for comparison ,
we expand the first and second level crossing inductions to first order in each of the @xmath76 for @xmath77 for @xmath147 where @xmath148 , @xmath196 where the upper ( lower ) sign is for @xmath173 ( @xmath197 ) , respectively . from these simple first - order results , it is possible to understand the qualitatively different behavior obtained with local , single - ion , anisotropy from that obtained with global anisotropy .
in the isotropic case @xmath198 , the first and second level crossings occur at @xmath199 and @xmath200 , respectively . for each induction direction , the signs of the @xmath161 and @xmath162 contributions to @xmath201 and @xmath202 are the same , whereas the signs of the @xmath203 and @xmath204 contributions to @xmath201 and @xmath202 are the _
opposite_. in figs .
6 - 15 , we plot @xmath142 and @xmath141 versus @xmath205 for five low-@xmath3 cases of afm @xmath30 dimers , taking @xmath180 .
the @xmath142 curves all exhibit the universal features predicted by eqs .
( [ msteps ] ) and ( [ slope ] ) .
the corresponding @xmath141 curves also obey the universal features predicted in eqs .
( [ cvzeroes])-([peakheight ] ) . in these figures , only one of the five anisotropy interactions @xmath76 is non - vanishing , and we take @xmath206 , for @xmath207 and @xmath208 , respectively . unlike the case of @xmath12 dimers , for which @xmath157 merely renormalizes @xmath162 , for @xmath30
all of these interactions lead to distinct anisotropy effects in the low-@xmath3 magnetization and specific heat .
of course , since the case @xmath209 ( and the remaining @xmath210 ) can be evaluated by setting @xmath211 and @xmath212 for any @xmath0 , it is not really distinct from the other anisotropy interactions , but only represents that particular combination of global and local azimuthal anisotropy interactions . and @xmath213 of @xmath142 versus @xmath143 for the afm spin 1 dimer .
curves for @xmath179 ( solid ) , @xmath173 ( dashed ) , and the isotropic case ( @xmath178 , dotted ) are shown.,scaledwidth=45.0% ] and @xmath213 of @xmath141 versus @xmath143 for the afm spin 1 dimer .
curves for @xmath179 ( solid ) , @xmath173 ( dashed ) , and the isotropic case ( @xmath178 , dotted ) are shown.,scaledwidth=45.0% ] we first show the results for the global anisotropy interactions . in figs . 6 and 7 , we plot @xmath28 and @xmath29 for @xmath181 ( and all other @xmath210 ) at @xmath180 for the afm @xmath30 dimer .
the level - crossing induction increases monotonically with level - crossing number , with the largest induction required occurring for @xmath177 , as for @xmath12 , quantitatively consistent with eqs .
( [ b11z])-([b21xy ] ) . in figs . 8 and 9 ,
the corresponding results for @xmath183 ( and the other @xmath210 ) are shown .
again , the level - crossing induction increases monotonically with level - crossing number , and the largest induction required for each level crossing occurs for @xmath214 , quantitatively consistent with eqs .
( [ b11z])-([b21xy ] ) , as for @xmath12 .
we note that the solid curves for @xmath215 are nearly identical with the dotted curves for the isotropic case @xmath216 .
however , there is a very small distinction between these @xmath187 and @xmath10 curves .
this distinction is also present for @xmath12 , as shown in figs . 3 and 5 , and is a detail of the shapes of the @xmath187 and @xmath10 curves arising from a slightly higher effective temperature with @xmath188 than in the other cases studied . and
@xmath217 of @xmath218 versus @xmath143 for the afm spin 1 dimer .
curves for @xmath177 ( solid ) , @xmath173 ( dashed ) , and the isotropic case ( @xmath176 , dotted ) are shown.,scaledwidth=45.0% ] and @xmath217 of @xmath141 versus @xmath143 for the afm spin 1 dimer .
curves for @xmath179 ( solid ) , @xmath173 ( dashed ) , and the isotropic case ( @xmath178 , dotted ) are shown.,scaledwidth=45.0% ] next , in figs .
10 - 13 , we show the corresponding curves for @xmath28 and @xmath29 at low @xmath3 for the local anisotropies @xmath219 and @xmath220 , respectively , with the other @xmath210 .
in contrast to the global anisotropies , the increase in @xmath14 required with the level crossing is not monotonic in the level - crossing number , quantitatively consistent with eqs .
( [ b11z])-([b21xy ] ) .
in particular , we note that in each of these figures , the deviations in the level crossings @xmath221 and @xmath222 are opposite in sign .
a careful examination of the exact formula for @xmath10 reveals that it vanishes at precisely the level crossing inductions as @xmath11 .
similarly , the magnetization curve attains its midpoint values @xmath223 and @xmath224 between the two neighboring steps at the level crossing inductions as @xmath11 . and
@xmath225 of @xmath142 versus @xmath143 for the afm spin 1 dimer .
curves for @xmath179 ( solid ) , @xmath173 ( dashed ) , and the isotropic case ( @xmath226 , dotted ) are shown.,scaledwidth=45.0% ] and @xmath225 of @xmath141 versus @xmath143 for the afm spin 1 dimer .
curves for @xmath179 ( solid ) , @xmath173 ( dashed ) , and the isotropic case ( @xmath178 , dotted ) are shown.,scaledwidth=45.0% ] and @xmath227 of @xmath142 versus @xmath143 for the afm spin 1 dimer .
curves for @xmath179 ( solid ) , @xmath173 ( dashed ) , and the isotropic case ( @xmath228 , dotted ) are shown.,scaledwidth=45.0% ] and @xmath213 of @xmath141 versus @xmath143 for the afm spin 1 dimer .
curves for @xmath179 ( solid ) , @xmath173 ( dashed ) , and the isotropic case ( @xmath178 , dotted ) are shown.,scaledwidth=45.0% ] finally , in figs .
14 and 15 , we show the corresponding @xmath28 and @xmath29 curves for @xmath209 , corresponding to the parameter choices @xmath212 and @xmath211 . with this combination , the magnetization and specific heat curves shown in figs .
14 and 15 resemble those for @xmath183 shown in figs . 8 and 9 , except that the second magnetization steps and the second set of specific heat peaks have nearly the same behavior as do the first magnetization steps and specific heat double peaks . in short , the case @xmath30 , for which the exact expressions from which the eigenvalues can be readily obtained are sufficiently short as to be writable on paper , are sufficient to exhibit the very different behaviors obtained from the single - ion , local spin anisotropy interactions from those obtained from the global spin anisotropy interactions .
as @xmath0 increases beyond 1 , the situation becomes not only more complicated , but also more interesting , as shown in the following . and
@xmath229 of @xmath142 versus @xmath143 for the afm spin 1 dimer .
curves for @xmath179 ( solid ) , @xmath173 ( dashed ) , and the isotropic case ( @xmath230 , dotted ) are shown.,scaledwidth=45.0% ] and @xmath229 of @xmath141 versus @xmath143 for the afm spin 1 dimer . curves for @xmath179 ( solid ) , @xmath173 ( dashed ) , and the isotropic case ( @xmath178 , dotted ) are shown.,scaledwidth=45.0% ]
for @xmath16 , one of the cases of greatest experimental interest , when @xmath65 and @xmath67 are present , none of the allowed @xmath231 values is a true quantum number .
that is , @xmath84 and @xmath67 cause all of the states with nominally odd or even @xmath6 to mix with one another .
for @xmath154 for @xmath232 , this simplifies in the crystal representation as for @xmath30 , since only states with odd or even @xmath32 in the appropriately chosen representation can mix . by using symbolic manipulation software ,
it is possible to solve for the exact eigenvalues of the @xmath16 dimer .
however , because the analytic expressions for the eigenvalues are much more complicated than those for @xmath30 presented in appendix a , we shall not attempt to present them , but will instead focus upon their numerical evaluation for specific cases . to first order in the @xmath76 , the first three level crossings for @xmath154 with @xmath232 are @xmath233 of @xmath142 versus @xmath205 for the afm spin 5/2 dimer with @xmath213 .
@xmath179 ( solid ) , @xmath173 ( dashed ) , and the isotropic case ( @xmath178 , dotted ) are shown.,scaledwidth=45.0% ] and @xmath213 of @xmath141 versus @xmath143 for the afm spin 5/2 dimer .
curves for @xmath179 ( solid ) , @xmath173 ( dashed ) , and the isotropic case ( @xmath178 , dotted ) are shown.,scaledwidth=45.0% ] versus @xmath143 for the afm spin 5/2 dimer at @xmath180 with @xmath217 .
curves for @xmath177 ( solid ) , @xmath173 ( dashed ) , and the isotropic case ( @xmath178 , dotted ) are shown.,scaledwidth=45.0% ] and @xmath217 of @xmath141 versus @xmath143 at @xmath180 for the afm spin 5/2 dimer .
curves for @xmath177 ( solid ) , @xmath234 ( dashed ) , and the isotropic case ( @xmath178 , dotted ) are shown.,scaledwidth=45.0% ] in figs .
16 - 23 , we plot @xmath142 and @xmath141 versus @xmath205 for four low-@xmath3 cases of afm @xmath16 dimers , @xmath181 , @xmath183 , @xmath219 , and @xmath209 , respectively , and the other @xmath210 , taking @xmath180 . in fig .
24 , examples of @xmath235 at fixed @xmath14 and @xmath236 are shown .
figures 16 - 24 are sufficient to distinguish the more interesting local spin anisotropy effects in afm dimers with higher @xmath0 values from the non - existent or less interesting ones present with @xmath139 , respectively . in figs .
16 , 18 , 20 , and 22 , the solid and dashed curves represent the cases of @xmath177 and @xmath234 , and the dotted curve is the isotropic case , @xmath216 , as in figs
. 2 - 15 . because of the number of peaks in the specific heat curves , in figs .
17 , 19 , 21 , and 23 , we ol = nly showed @xmath141 versus @xmath143 for @xmath179 ( solid ) and @xmath173 ( dashed ) .
the isotropic curve with @xmath237 was published previously.@xcite that curve has minima at @xmath238 for @xmath239 , each of which is central to double peaks . @xmath84 and @xmath99 are invariant under @xmath240 , so curves for @xmath173 in figs .
16 , 17 , 20 , and 21 are identical to those for @xmath241 . since @xmath242 and @xmath68 are odd under @xmath240 , the @xmath173 curves in figs .
18 , 19 , 22 , and 23 correspond to @xmath241 with @xmath243 , @xmath244 , respectively .
we first examine the global spin anisotropy effects of @xmath99 and @xmath100 in figs .
these figures exhibit the same behavior shown for @xmath139 in the corresponding figs . 2 - 9 .
note that @xmath245 is largest for @xmath246 with @xmath181 and for @xmath247 with @xmath183 , and increases monotonically with @xmath6 , as for @xmath30 in both cases , nearly quantitatively consistent with eqs .
( [ b52step1])-([b52step3 ] ) .
by contrast , @xmath248 for @xmath246 with @xmath181 and for @xmath247 for @xmath183 are nearly indistinguishable from the isotropic case , also nearly quantitatively consistent with eqs .
( [ b52step1])-([b52step3 ] ) .
in contrast , the local field anisotropy interactions show very different and much more interesting behaviors . in figs .
20 - 21 , we present our results for the effects of the single - ion axial anisotropy interaction @xmath65 , eq .
( [ lz ] ) . as in fig . 10 for @xmath30 ,
@xmath249 changes sign with increasing @xmath6 .
@xmath250 , whereas for @xmath251 , @xmath252 and increases monotonically with @xmath6 . for @xmath246 ,
nearly the opposite situation occurs .
@xmath253 is positive for @xmath165 and decreases monotonically with increasing @xmath6 . in both cases,@xmath254
is a minimum for @xmath255 .
these local axial anisotropy effects , consistent with eqs .
( [ b52step1])-([b52step3 ] ) , are very different than the global anisotropy ones pictured in figs .
they are also much richer and interesting than the corresponding case for @xmath30 pictured in figs . 10 and 11 . in figs .
22 - 23 , we present our @xmath142 versus @xmath143 results for the case of the local azimuthally anisotropic exchange interaction , @xmath68 , eq .
( [ hc ] ) , evaluated for afm dimers at @xmath180 with @xmath209 and the remaining @xmath210 .
@xmath256 for this case is obtained from eqs .
( [ b52step1])-([b52step3 ] ) by setting @xmath257 , respectively , and @xmath258 .
@xmath259 is nearly indistinguishable from the isotropic case , as in figs .
12 and 13 for @xmath30 , except for some minor curve shape effects far from the step midpoints .
@xmath253 is always positive , as for @xmath30 shown in figs .
12 and 13 , but has a minimum at @xmath260 .
this is also in stark contrast to the monotonic global anisotropy behavior for @xmath16 seen in figs .
both of these behaviors are nearly quantitatively consistent with eqs .
( [ b52step1])-([b52step3 ] ) as modified to include @xmath157 . and @xmath225 of @xmath142 versus @xmath143 for the afm spin 5/2 dimer .
curves for @xmath179 ( solid ) , @xmath173 ( dashed ) , and the isotropic case ( @xmath226 , dotted ) are shown.,scaledwidth=45.0% ] and @xmath225 of @xmath141 versus @xmath143 at @xmath180 for the afm spin 5/2 dimer .
curves for @xmath177 ( solid ) , @xmath234 ( dashed ) , and the isotropic case ( @xmath178 , dotted ) are shown.,scaledwidth=45.0% ] of @xmath142 versus @xmath205 for the afm spin 5/2 dimer with @xmath229 .
@xmath179 ( solid ) , @xmath173 ( dashed ) , and the isotropic case ( @xmath178 , dotted ) are shown.,scaledwidth=45.0% ] and @xmath229 of @xmath141 versus @xmath143 at @xmath180 for the afm spin 5/2 dimer .
curves for @xmath177 ( solid ) , @xmath234 ( dashed ) , and the isotropic case ( @xmath178 , dotted ) are shown.,scaledwidth=45.0% ] in contrast , the local field anisotropy interactions show very different behaviors . in figs .
20 and 21 , our results for the effects of the single - ion axial anisotropy interaction @xmath84 , eq .
( [ lz ] ) , are shown .
as in figs .
10 and 11 , the induction anisotropy effects of @xmath65 change sign with increasing @xmath14 . in fig .
20 , for @xmath177 the first magnetization step appears at a lower value of @xmath261 than in the isotropic case , and for steps 3 - 5 , there is a monotonic increase in the extra field required for each step .
the opposite is true for @xmath173 , for which the first step appears at a larger @xmath261 value than for the isotropic case , and subsequent magnetization steps appear at monotonically decreasing values of @xmath261 .
a cross - over occurs at about the second step , for which the effects of this type of anisotropy are small .
the corresponding shifts in the positions of the double peaks in the specific heat are shown in fig .
these local axial anisotropy effects are very different than the global anisotropy ones pictured in figs .
16 - 19 . in figs . 22 and 23
, we present our @xmath27 and @xmath10 results for the case of the local azimuthally anisotropic exchange interaction , @xmath68 , eq .
( [ hc ] ) , evaluated for afm dimers at @xmath180 with @xmath209 , or equivalently with @xmath211 and @xmath212 ( and the remaining @xmath210 ) . for @xmath177 , there is almost no change from the isotropic case , as occurred with the global azimuthal anisotropy pictured in figs .
128 and 19 . for @xmath173 ,
the field required for each step is larger than in the isotropic interaction case , as in figs .
14 and 15 for @xmath30 , and for global azimuthal anisotropy shown in figs .
18 and 19 .
however , in this case , the extra induction required for each level crossing is non - monotonic in the crossing number , with the largest extra induction required for the first crossing , and the minimum extra induction required for the intermediate , third crossing .
this is in stark contrast to the monotonic global anisotropy behavior for @xmath16 seen in figs .
although not pictured explicitly , the behavior for @xmath220 with the other @xmath210 is rather like that of the @xmath219 curves pictured in figs .
20 and 21 with @xmath262 , differing in ways similar to those differences between the @xmath219 and @xmath220 curves pictured for @xmath30 in figs .
10 - 13 . at @xmath236 versus @xmath263 near the second step at @xmath264 and @xmath180 for each @xmath206 with @xmath265 .
curves are labelled with @xmath266 values.,scaledwidth=45.0% ] although not pictured explicitly , the behavior for @xmath220 with the other @xmath210 is rather like that of the @xmath219 curves pictured in figs .
20 and 21 with @xmath267 , differing in ways similar to those differences between the @xmath219 and @xmath220 curves pictured for @xmath30 in figs .
as indicated in eqs .
( [ b52step1])-([b52step3 ] ) , the @xmath259 are nearly independent of @xmath204 .
however , the @xmath268 with @xmath220 are nearly three times as large as for @xmath219 case , so that the minimum @xmath269 is also a minimum for @xmath255 .
more details are given in section vii and appendix d. in addition , the angular dependencies of @xmath27 are different for each of the four @xmath76 we presented for @xmath16 . in fig .
24 , we present the results for @xmath270 versus @xmath263 near the second level crossing at @xmath271 and @xmath180 for each of the four @xmath16 afm magnetization cases pictured in figs .
16 , 18 , 20 , and 22 .
note that @xmath272 .
the @xmath203 and @xmath161 curves , while rather similar at @xmath273 , are very different at @xmath274 .
hence , @xmath275 depends strongly upon the particular type of spin anisotropy .
since the diagonalization of the hamiltonian matrix is difficult for an arbitrary magnetic field @xmath276 direction and for an arbitrary combination of spin anisotropy interactions , and must be done for each value of @xmath0 separately , it is useful to consider a perturbative solution in the relative strengths @xmath277 of the the anisotropy interactions .
we nominally assume @xmath278 for @xmath77 . however , to compare with low-@xmath3 @xmath4 and @xmath13 experiments at various @xmath276 directions and magnitudes
, one can not take @xmath14 to be small . in order to incorporate an arbitrary @xmath9 accurately ,
we rotate the crystal axes @xmath279 to @xmath280 , so that @xmath281 . the rotation matrix and a brief discussion of its ramifications
are given in appendix b. in these rotated coordinates , the zeeman interaction @xmath282 is diagonal .
we therefore denote this representation as the induction representation .
the hamiltonian @xmath283 in this representation is given in appendix b. in the induction representation , we choose the quantum states to be @xmath284 . in the absence of the four anisotropy interactions @xmath76 , @xmath285 is diagonal , @xmath286 the operations of the remaining terms in @xmath283 on the eigenstates @xmath284 are given in appendix c. the first order correction to the energy in this representation is @xmath287 , which is found to be @xmath288-\frac{j_a}{2}[s(s+1)-1]\nonumber\\ & & + \frac{\tilde{j}_{b , a}^{s , s_1}}{2}[m^2+s(s+1)-1]\nonumber\\ & & + \frac{1}{2}[s(s+1)-3m^2]\nonumber\\ & & \times\bigl(\tilde{j}_{b , a}^{s , s_1}\cos^2\theta + \tilde{j}_{d , e}^{s , s_1}\sin^2\theta\cos(2\phi)\bigr ) , \label{esm1}\end{aligned}\ ] ] where @xmath289 and @xmath290 is given by eq .
( [ alphass1 ] ) .
since the @xmath8 dependence of @xmath291 arises from the term proportional to @xmath292 , it is tempting to think that the thermodynamics with @xmath293 and @xmath177 are equivalent to those with @xmath294 and @xmath173 .
however , as shown explicitly in the following , the @xmath8-independent parts of eq .
( [ esm1 ] ) strongly break this apparent equivalence , causing the @xmath136 for those two cases to differ .
this implies that @xmath161 and @xmath162 are inequivalent , as are @xmath203 and @xmath204 , even to first order in the anisotropy strengths .
to first order in the anisotropy interactions , @xmath6 and @xmath32 are still good quantum numbers , so the partition function @xmath295 where @xmath296 .
although it is difficult to perform the summation over the @xmath32 values analytically , it is nevertheless elementary to evaluate @xmath297 numerically for an arbitrary @xmath298 , and @xmath3 from the eigenstate energies .
the magnetization is obtained from @xmath299 similarly , the specific heat to first order in the anisotropy interactions is found from @xmath300.\label{cvapprox}\end{aligned}\ ] ] as a test of the accuracy of this first - order calculation , we have compared the first - order and exact @xmath28 obtained for the @xmath16 dimer with @xmath183 and @xmath177 in fig .
[ magtest ] . the corresponding comparison between the first - order and exact @xmath29 is shown in fig .
[ shtest ] .
we see that the curves evaluated using the first - order and the exact expressions for @xmath187 and @xmath10 with @xmath16 are indistinguishable at @xmath180 . the @xmath10 curves are noticeably different at @xmath301 for @xmath302 , and at @xmath303 , they are noticeably different for @xmath304
. corresponding noticeable differences in the @xmath187 curves at the same @xmath14 values appear at @xmath3 values roughly three times as high as in the @xmath10 curves . at very low @xmath3 , @xmath305 ,
the most important states in this perturbative scheme are the minima for each @xmath6 value , @xmath306 , which determine the first - order level crossings . as @xmath11
, we can ignore all of the @xmath307 states in eqs .
( [ zapprox])-([cvapprox ] ) .
versus @xmath143 obtained using the first - order asymptotic form ( dotted ) with the exact calculation ( solid ) , for the @xmath16 afm dimer with @xmath183 , @xmath308 , at @xmath309 , as indicated .
inset : expanded view of the region @xmath310.,scaledwidth=45.0% ] versus @xmath143 obtained using the first - order asymptotic form ( dotted ) with the exact calculation ( solid ) , for the @xmath16 afm dimer with @xmath183 , @xmath308 , at @xmath311 , as indicated.,scaledwidth=45.0% ] we can find an expression for the @xmath312 afm level crossing at the induction @xmath313 to first order in the anisotropy interactions for a general @xmath0 spin dimer by equating @xmath314 to @xmath315 , @xmath316\nonumber\\ & & + 3c_{s , s_1}[j_a\cos^2\theta+j_e\sin^2\theta\cos(2\phi)],\\ c_{s , s_1}&=&\frac{[3 + 3s-5s^2 - 4s^3 + 4s_1(s_1 + 1)]}{2(2s+1)(2s+3)}.\label{bstep}\end{aligned}\ ] ] this expression is consistent with those obtained for @xmath139 given by eqs .
( [ bstephalf ] ) and ( [ b11z])-([b21xy ] ) .
in addition , this expression is nearly quantitatively in agreement with the @xmath28 and @xmath29 behaviors pictured for @xmath16 in figs .
we note that @xmath317 contains the @xmath8-independent terms , @xmath318 , which distinguish @xmath161 from @xmath162 and @xmath203 from @xmath204 .
in particular , we note that the single - ion anisotropy interactions behave very differently with increasing step number than do the global anisotropy interactions , especially for large @xmath0 .
for the three cases we studied in detail , for @xmath12 , @xmath319 , so that the local anisotropy terms are irrelevant , for @xmath30 , @xmath320 and @xmath321 have different signs , and for @xmath16 as in fe@xmath17 dimers , the first three @xmath322 coefficients are @xmath323 , @xmath324 , and @xmath325 , respectively , the second being an order of magnitude smaller than the other two , and opposite in sign from the first . for @xmath18 dimers such as [ mn@xmath19@xmath17 ,
the first four @xmath326 are @xmath327 , @xmath328 , @xmath329 , and @xmath330 , which changes sign between @xmath255 and @xmath260 , where its magnitude is a minimum .
this is in sharp contrast to the global anisotropy interactions , for which the analogous coefficient @xmath331 increases monotonically from @xmath223 to @xmath332 as @xmath6 increases from 1 to 2 , independent of @xmath0 .
these differences should be possible to verify experimentally in careful low-@xmath3 experiments at high magnetic fields applied at various directions on single crystals of those @xmath16 fe@xmath17 and @xmath18 [ mn@xmath19@xmath17 dimers for which @xmath333 is sufficiently small . to aid in the analysis of experimental data
, we have extended this perturbative calculation to second order .
since we expect the @xmath334 in most circumstances , this extension ought to be sufficient to accurately analyze most experimentally important samples .
to second order in the anisotropy interactions , the eigenstate energies @xmath335 are given in appendix c. we note that the @xmath336 contains divergences at @xmath337 and @xmath338 , so that near to those values , one would need to modify the perturbation expansion to take proper account of the degeneracies .
hence , the expressions for @xmath336 can not be used in the asymptotic expressions for the thermodynamics , eqs .
( [ zapprox])-([cvapprox ] ) . however , as the @xmath6th afm level crossing occurs approximately at @xmath238 , which is far from any divergences , we can safely use this second order expansion to obtain an expression for the level crossings second order in the anisotropy interaction energies .
we find @xmath339 where @xmath340 is given by eq .
( [ bstep ] ) .
the full expression for @xmath341 is given in appendix d. from this expression , it is easy to see that for @xmath12 , @xmath342 , where @xmath343 and @xmath344 are given in appendix d. for @xmath179 , @xmath345 , and with @xmath346 , @xmath347 , in agreement with the expansion to second order in the @xmath76 of our exact formulas in eq .
( [ bstephalf ] ) .
however , the second order functions have more complicated @xmath8 dependencies than do the first order @xmath313 in eq .
( [ bstep ] ) .
we have also explicitly checked each formula in appendix d for @xmath30 and @xmath348 .
thus , eq . ( [ bstep2 ] ) is a highly accurate expression for the full @xmath8 dependencies of all @xmath349 level crossings of a single crystal dimer of single ion spin @xmath0 . by superposing this
non - universal level - crossing formula , eq .
( [ bstep2 ] ) , combined with the universal behavior presented in eqs .
( [ cvzereos])-([slope ] ) , it is easy to use our results in accurate fits to experimental data at low temperatures and high magnetic fields on dimers with arbitrary @xmath0 values .
in summary , we solved for the low - temperature magnetization and specific heat of equal spin @xmath0 antiferromagnetic dimer single molecule magnets , including the most general forms of anisotropic spin exchange interactions quadratic in the spin operators .
the magnetization and specific heat exhibit steps and zeroes , respectively , at the non - universal level - crossing induction values @xmath15 , but the magnetization steps and their midpoint slopes , plus the two peaks surrounding the specific heat zeroes all exhibit universal behavior at sufficiently low temperatures .
local ( or single - ion ) anisotropy interactions lead to low - temperature magnetization step plateaus that have a much richer variation with the magnetic induction @xmath9 than do those obtained from global anisotropy interactions , provided that @xmath2 .
we derived simple , accurate asymptotic analytic expressions for the low - temperature magnetization and specific heat for the most general quadratic anisotropic spin interactions at an arbitrary @xmath9 , and an accurate expression for @xmath15 , enabling fast and accurate fits to experimental data .
there were two low-@xmath3 @xmath28 studies of fe@xmath17 dimers.@xcite for @xmath350-oxalatotetrakis(acetylacetonato)fe@xmath17 , all five peaks in @xmath351 were measured in pulsed magnetic fields @xmath352 .
these evenly spaced peaks indicated little , if any , spin anisotropy effects.@xcite on the other hand , studies of the first 2 - 3 @xmath351 peaks in powdered samples of [ fe(salen)cl]@xmath17 , where salen is @xmath353-ethylenebis(salicylideneiminato ) , were much more interesting.@xcite these data showed a broad first peak at @xmath354 t that was only partially resolvable into two separate peaks , followed by a sharp second peak at @xmath355 t , consistent with local axial anisotropy of strength @xmath356 , as obtained from the derivatives of the curves shown in fig .
20 from eqs .
( [ b52step1])-([b52step3 ] ) , one could also have @xmath357 .
these values might perhaps be combined with a smaller @xmath358 , obtained from the derivatives of the curves pictured in fig .
22 . without a detailed single crystal study with
the magnetic field directed along a number of different crystal directions , it is impossible to determine the relative amounts of @xmath203 and @xmath204 that would best fit the data . however , the existing data on [ fe(salen)cl]@xmath17 appear to be inconsistent with a predominant global anisotropy interaction of either type , as obtained from the derivatives of the curves shown in figs . 16 and 18 .
only single crystal studies could determine if a small amount of such global anisotropy interactions were present in addition to one or more presumably larger local spin anisotropy interactions .
in comparing the two materials cited above , it appears that the interaction of a cl@xmath359 ion neighboring each fe@xmath360 ion leads to strong local ( or single - ion ) anisotropy effects . in order to verify this hypothesis and to elucidate the details of the interactions , further experiments using single crystals in different field orientations on this and related fe@xmath17 dimers with 1 - 3 similarly bonded cl@xmath359 ions
are urged.@xcite we also urge single crystal data on some of the @xmath361 [ mn@xmath19@xmath17 dimers,@xcite as well as on @xmath12 dimers lacking in predicted local spin anisotropy effects . @xcite to aid in the fits , we derived simple , useful formulas for the magnetization and specific heat at low temperature and sufficiently large magnetic induction , and accurate formulas for the level crossing inductions . with local anisotropy interactions ,
the total spin @xmath6 is not a good quantum number , potentially modifying our understanding of quantum tunneling processes . it might also be possible to fit a variety of experimental results using a smaller
, consistent set of model parameters.@xcite we emphasize that the study of smm dimers , for which the most general anisotropic quadratic exchange interactions can be solved exactly , may be our best hope for attaining a more fundamental understanding of the underlying physics of single molecule magnets .
we thank the max - planck - institut fr physik komplexer systeme , dresden , germany , the university of north dakota , grand forks , nd , usa , and talat s. rahman for their kind hospitality and support .
this work was supported by the netherlands foundation for the fundamental research of matter and by the nsf under contract ner-0304665 .
we first present the numerators of the exact expressions for the specific heat with @xmath12 and @xmath154 for @xmath148 .
we have @xmath362}\nonumber\\ & & + f_{x , y}\sinh(\beta f_{x , y})e^{-\beta j_{x , y}}\nonumber\\ & & \times[(j+j_{x , y})e^{-\beta j}+(j_{x , y}-2j_{y , x})e^{2\beta j_{y , x}}]\nonumber\\ & & + \frac{1}{2}e^{-\beta j_{x , y}}\cosh(\beta f_{x , y})\nonumber\\ & & \times\bigl([(j+j_{x , y})^2+f_{x , y}^2]e^{-\beta j}\nonumber\\ & & + [ ( 2j_{y , x}-j_{x , y})^2+f_{x , y}^2]e^{2\beta j_{y , x}}\bigr),\\ { \cal n}_z&=&f_z^2\cosh(2\beta f_z)+\frac{j^2}{4}e^{-\beta(j+2j_b)}\nonumber\\ & & + \frac{1}{2}\cosh(\beta f_z)\bigl(\delta_z(j_b^2+f_z^2)\nonumber\\ & & \qquad+e^{-\beta(j+j_b)}j(j+2j_b)\bigr)\nonumber\\ & & + f_z\sinh(\beta f_z)\bigl(j_b\delta_z+je^{-\beta(j+j_b)}\bigr).\end{aligned}\ ] ] in the remainder of this appendix , we provide the details of our exact results for @xmath30 . the cubic equation for the three @xmath165 eigenvalues
is given by @xmath363\nonumber\\ & & + ( j_a - j_b)[b_z^2+(j_d - j_e)^2]\nonumber\\ & & -(j_d - j_e)(b_x^2-b_y^2),\end{aligned}\ ] ] where the @xmath364 as in eq .
( [ bi ] ) . for @xmath365
, the cubic equation is easily solved to yield @xmath366 for @xmath346 , we have @xmath367,\end{aligned}\ ] ] where @xmath368 ( @xmath369 ) corresponds to the upper ( lower ) sign .
the rank ordering of these eigenvalues depends upon @xmath14 and the various anisotropy parameters , of course .
the six eigenvalues for the mixed @xmath191 states satisfy @xmath370 we then define @xmath371 and obtain the hermitian matrix @xmath372 for the six @xmath191 states , the eigenvalues of which are the @xmath373 . for brevity
, we let @xmath374 then , the matrix @xmath372 is given by @xmath375 after some rearrangement , the @xmath373 are found to satisfy @xmath376\nonumber\\ & & \qquad-21\tilde{j}_d(b_x^2-b_y^2)\bigr)\nonumber\\ & & + \tilde{j}_a^2(3\tilde{j}_b^2 + 9\tilde{j}_d^2 + 2b^2 + 3b_z^2)\nonumber\\ & & + 2\tilde{j}_b^2[4\tilde{j}_b^2 + 3\tilde{j}_e^2 + 11(b^2 - 3b_z^2)]\nonumber\\ & & + 2\tilde{j}_d^2[54(\tilde{j}_b^2+\tilde{j}_d^2)+9\tilde{j}_e^2 - 12(b^2 - 3b_z^2)]\nonumber\\ & & + 2\tilde{j}_e^2(4b^2 - 3b_z^2)+4b^4\nonumber\\ & & + 2(3\tilde{j}_b\tilde{j}_d-\sqrt{6}\tilde{j}_a\tilde{j}_e)(b_x^2-b_y^2),\\ c_1&=&-12\tilde{j}\bigl(b^4 + 3\tilde{j}_b^4 + 18\tilde{j}_b^2\tilde{j}_d^2 + 27\tilde{j}_d^4\nonumber\\ & & + 2(\tilde{j}_b^2 - 3\tilde{j}_d^2)(b^2 - 3b_z^2)\nonumber\\ & & + 12\tilde{j}_b\tilde{j}_d(b_x^2-b_y^2)\bigr)\nonumber\\ & & -2\tilde{j}_b\bigl(2(9\tilde{j}_d^2+b^2)(b^2 + 3b_z^2)\nonumber\\ & & -\tilde{j}_a^2(b^2 - 3b_z^2)+6\tilde{j}_b^2(b^2 - 5b_z^2)\nonumber\\ & & + ( 42\tilde{j}_b\tilde{j}_d+\sqrt{6}\tilde{j}_a\tilde{j}_e)(b_x^2-b_y^2)\nonumber\\ & & -2\tilde{j}_a^2(\tilde{j}_b^2 + 9\tilde{j}_d^2)+8\tilde{j}_b^2(2\tilde{j}_b^2 + 6\tilde{j}_d^2+\tilde{j}_e^2)\nonumber\\ & & + 6\sqrt{6}\tilde{j}_a\tilde{j}_b\tilde{j}_d\tilde{j}_e\bigr)\nonumber\\ & & -2\tilde{j}_d\bigl(6(b^2 - 3\tilde{j}_d^2+\tilde{j}_e^2)(b_x^2-b_y^2)\nonumber\\ & & -\sqrt{6}\tilde{j}_a\tilde{j}_e(b^2 - 3b_z^2 - 18\tilde{j}_d^2)\bigr),\\ c_0&=&-12\tilde{j}\bigl(\tilde{j}_bb^2(b^2 - 3b_z^2)\nonumber\\ & & -3\tilde{j}_d(b^2-\tilde{j}_b^2 - 3\tilde{j}_d^2)(b_x^2-b_y^2)\nonumber\\ & & + \tilde{j}_b^3(b^2 + 3b_z^2 - 2\tilde{j}_b^2 + 12\tilde{j}_d^2)\nonumber\\ & & -3\tilde{j}_b\tilde{j}_d^2(7b^2 - 3b_z^2 - 18\tilde{j}_d^2)\bigr)\nonumber\\ & & -\tilde{j}_a^2\bigl((b^2 - 3b_z^2 + 2\tilde{j}_b^2)^2 + 36\tilde{j}_d^2(b_z^2-\tilde{j}_b^2)\bigr)\nonumber\\ & & -4\tilde{j}_b^2\bigl(2b^2(b^2 - 3b_z^2+\tilde{j}_e^2)-2\tilde{j}_b^2(2\tilde{j}_b^2+\tilde{j}_e^2-b^2 - 3b_z^2)\nonumber\\ & & + 6\tilde{j}_d^2(-7b^2 + 3b_z^2 + 4\tilde{j}_b^2 + 18\tilde{j}_d^2 + 3\tilde{j}_e^2)\nonumber\\ & & + ( 6\tilde{j}_b\tilde{j}_d-\sqrt{6}\tilde{j}_a\tilde{j}_e)(b_x^2-b_y^2)\bigr)\nonumber\\ & & + 24\tilde{j}_b\tilde{j}_d(b_x^2-b_y^2)(b^2 - 3\tilde{j}_d^2 - 2\tilde{j}_e^2)\nonumber\\ & & + 2\sqrt{6}\tilde{j}_a\tilde{j}_e(b_x^2-b_y^2)(b^2 - 3b_z^2 - 6\tilde{j}_d^2)\nonumber\\ & & -6\tilde{j}_e^2(b_x^2-b_y^2)^2\nonumber\\ & & + 8\sqrt{6}\tilde{j}_a\tilde{j}_b\tilde{j}_d\tilde{j}_e(2b^2 - 3b_z^2+\tilde{j}_b^2 - 9\tilde{j}_d^2).\end{aligned}\ ] ] we recall that the six hamiltonian matrix eigenvalues @xmath131 are generally given by eq .
( [ epsilonnspin1 ] ) . for the special case @xmath177
, @xmath372 is block diagonal .
the resulting eigenvalues of the nominal @xmath377 notation are @xmath378 the remaining four states are obtained from @xmath379 similarly , for @xmath346 , @xmath380 is also block diagonal .
the resulting eigenvalues of the nominal @xmath381 notation are @xmath382 the remaining four states are obtained from @xmath383 \nonumber\\ & & + 2\tilde{j}_b^2[\tilde{j}_a^2 - 4\tilde{j}_b^2 - 6\tilde{j}\tilde{j}_b-12\tilde{j}_d^2 - 2\tilde{j}_e^2]\nonumber\\ & & \pm12\tilde{j}_d(2\tilde{j}_b+3\tilde{j})(\tilde{j}_b^2 + 3\tilde{j}_d^2)\nonumber\\ & & \mp6\tilde{j}_b\tilde{j}_d(\tilde{j}_a^2\pm6\tilde{j}\tilde{j}_d-2\tilde{j}_e^2)\nonumber\\ & & -4\sqrt{6}\tilde{j}_a\tilde{j}_d\tilde{j}_e(\tilde{j}_b\mp3\tilde{j}_d).\end{aligned}\ ] ] when only one of the @xmath384 , the eigenvalues for @xmath179 simplify considerably . for @xmath385 , we have @xmath386 for @xmath188 , we find @xmath387 for @xmath388 , the eigenvalues can also be found analytically , @xmath389 we note that the ground state energy in this case is @xmath390 which explicitly involves mixing of the @xmath391 and the @xmath392 states . for @xmath393 ,
three of the eigenvalues are @xmath394 and the remaining three satisfy the cubic equation , @xmath395 we note that the cubic equation must be solved to obtain the ground state energy , since the @xmath391 state mixes with the @xmath396 states . for @xmath346 , with @xmath397 ,
the eigenvalues are @xmath398 where @xmath399 is given by eq .
( [ tildejxy ] ) , and the other three eigenvalues satisfy the cubic equation , @xmath400 for @xmath388 , two of the eigenvalues satisfy @xmath401 and the other four eigenvalues satisfy the quartic equation , @xmath402 finally , for @xmath393 , two eigenvalues are @xmath403 and the other four eigenvalues satisfy the quartic equation , @xmath404x\nonumber\\ & & \pm4j_e(-3b^2j\pm2b^2j_e+jj_e^2 ) .
\end{aligned}\ ] ] at the first level crossing with @xmath179 , we have for @xmath388 , @xmath405^{1/2}\bigr).\label{jastep1}\end{aligned}\ ] ] at the second level crossing , simple formulas are only obtained for @xmath177 with one @xmath384 . for @xmath179 and @xmath388 , @xmath385 and @xmath188 ,
respectively , we have @xmath406^{1/2}\bigr)^{1/2}.\end{array}\right.\label{secondstep}\end{aligned}\ ] ]
the rotation from the crystal representation to the induction representation is obtained from @xmath407 leading to @xmath408.@xcite this operation is equivalent to a rotation by @xmath409 about the @xmath53 axis , a rotation by @xmath410 about the transformed @xmath411 axis , and then a rotation by @xmath412 about the transformed @xmath53 axis.@xcite in effect , in using the above rotation matrix , we made the arbitrary choice that the rotated @xmath53 axis lies in the @xmath413 plane . after the above rotation , it is still possible to rotate the crystal by an arbitrary angle @xmath414 about the @xmath415 axis , keeping @xmath416 .
hence , there are in effect an infinite number of equivalent rotations leading to @xmath408 .
the resulting hamiltonian matrix will then have off - diagonal elements that depend upon @xmath414 .
however , all such rotations necessarily lead to the identical , @xmath414-independent , set of eigenvalues of the resulting diagonalized hamiltonian matrix . we have explicitly checked that the above rotation gives the exact cubic expression , eq .
( [ halfeigen ] ) , for the @xmath12 eigenvalues , and also leads to the correct eigenstate energies second order in each of the @xmath76 for @xmath30 .
we also showed explicitly that @xmath414 does not enter the eigenstate energies second order in @xmath161 for arbitrary @xmath417 .
the global axial and azimuthal anisotropy interactions in the rotated frame are @xmath418,\nonumber\\\end{aligned}\ ] ] where @xmath419 is the anticommutator .
the operations of the rotated global anisotropy interactions upon these states may be written as @xmath420\sin^2\theta\bigr)| \varphi_s^m\rangle\nonumber\\ & & -\sin(2\theta)\sum_{\sigma=\pm1}(2m+\sigma)a_s^{\sigma m}|\varphi_s^{m+\sigma}\rangle\nonumber\\ & & + \sin^2\theta\sum_{\sigma=\pm1}f_s^{\sigma m}|\varphi_s^{m+2\sigma}\rangle\bigr ] , \label{hbprime}\end{aligned}\ ] ] and @xmath421|\varphi_s^m\rangle\nonumber\\ & & + 2\sin\theta\sum_{\sigma=\pm1}(2m+\sigma)a_s^{\sigma m}\nonumber\\ & & \times[\cos\theta\cos(2\phi)+i\sigma\sin(2\phi)]|\varphi_s^{m+\sigma}\rangle\nonumber\\ & & + \sum_{\sigma=\pm1}f_s^{\sigma m}\bigl[(1+\cos^2\theta)\cos(2\phi)\nonumber\\ & & + 2i\sigma\cos\theta\sin(2\phi)\bigr]|\varphi_s^{m+2\sigma}\rangle\bigr),\label{hdprime}\end{aligned}\ ] ] where @xmath422 is defined by eq .
( [ fsx ] ) .
we note that in this representation , both @xmath99 and @xmath100 preserve the global spin quantum number @xmath6 , but allow @xmath423 transitions . as a check on the algebra
, we verified that for @xmath12 , eq . ( [ halfeigen ] ) with @xmath424 is obtained from eqs .
( [ hbprime ] ) and ( [ hdprime ] ) . with regard to the local spin anisotropy terms in the rotated coordinate system ,
we write @xmath425 and @xmath426,\end{aligned}\ ] ] where @xmath427 and@xmath428 the operations of these interactions are given by @xmath429 where @xmath430(1-\alpha_{s , s_1}),\nonumber\\ & & \\
n_{s , s_1}^{m,\sigma'}&=&\sum_{\sigma=\pm1}c_{s+(\sigma'+1)/2,s_1}^{-\sigma\sigma'm-(\sigma'+1)/2 } c_{s+(3\sigma'+1)/2,s_1}^{\sigma\sigma'm+(\sigma'-1)/2},\label{nss1msigma}\\ p_{s , s_1}^{m,\sigma}&=&2a_s^{\sigma m}(2m+\sigma)\alpha_{s , s_1},\\ \noalign{\rm and } r_{s , s_1}^{x,\sigma'}&=&c_{s+(\sigma'+1)/2,s_1}^{-x\sigma'-(\sigma'+1)/2}d_{s+(3\sigma'+1)/2,s_1}^{m+\sigma}\nonumber\\ & & + c_{s+(3\sigma'+1)/2,s_1}^{-x\sigma'-(\sigma'+1)/2}d_{s+(\sigma'+1)/2,s_1}^m,\label{rss1xsigma}\end{aligned}\ ] ] where @xmath431 , @xmath432 , @xmath433 , @xmath434 , and @xmath435 are given by eqs . ( [ eta ] ) and ( [ gss1m])-([lss1 ] ) , respectively .
we note that for @xmath436 , @xmath437 .
in this appendix , we evaluate the corrections to the eigenstate energies second order in the four anisotropy interaction energies @xmath76 for @xmath77 .
the operations of the rotated hamilatonian @xmath283 upon the eigenstates @xmath284 may be written as @xmath438 where @xmath439,\\ { \cal v}_{s , s_1}^{m,\sigma}(\theta,\phi)&=&-\frac{1}{4}f_s^{\sigma m}\bigl[\tilde{j}_{b , s}^{s , s_1}\sin^2\theta\nonumber\\ & & + \tilde{j}_{d , e}^{s , s_1}(1+\cos^2\theta)\cos(2\phi)\nonumber\\ & & + 2i\sigma\tilde{j}_{d , e}^{s , s_1}\cos\theta\sin(2\phi)\bigr],\\ { \cal w}_{s , s_1}^{m,\sigma'}(\theta,\phi)&=&-\frac{1}{2}h_{s , s_1}^{m,\sigma'}[j_a\cos^2\theta+j_e\sin^2\theta\cos(2\phi)]\nonumber\\ & & + \frac{1}{8}n_{s , s_1}^{m,\sigma'}\sin^2\theta[j_a - j_e\cos(2\phi)],\\ { \cal x}_{s , s_1}^{m,\sigma,\sigma'}(\theta,\phi)&=&-\frac{\sigma\sigma'}{8}r_{s , s_1}^{\sigma m,\sigma'}\bigl(\sin(2\theta)[j_a - j_e\cos(2\phi)]\nonumber\\ & & \qquad-2i\sigma j_e\sin\theta\sin(2\phi)\bigr),\\ \noalign{\rm and } & & \nonumber\\ { \cal y}_{s , s_1}^{\sigma m,\sigma'}(\theta,\phi)&=&-\frac{1}{8}k_{s , s_1}^{\sigma m,\sigma'}\bigl[j_a\sin^2\theta\nonumber\\ & & \qquad+j_e\bigl((1+\cos^2\theta)\cos(2\phi)\nonumber\\ & & \qquad+2i\sigma\cos\theta\sin(2\phi)\bigr)\bigr],\end{aligned}\ ] ] where @xmath440 and @xmath441 are given by eqs .
( [ e0 ] ) and ( [ esm1 ] ) , respectively , @xmath433 and @xmath434 are given by eqs .
( [ hss1msigma ] ) and ( [ kss1xsigma ] ) , respectively , @xmath442 and @xmath443 are given by eqs .
( [ nss1msigma ] ) and ( [ rss1xsigma ] ) , respectively , and @xmath444 and @xmath445 are given by eqs .
( [ tildejba ] ) and ( [ tildejde ] ) , respectively . from eq .
( [ hprime ] ) , the second order eigenstate energies may be written as @xmath446}\nonumber\\ & & + \sum_{\sigma,\sigma'\pm1}\bigl(\frac{|{\cal x}_{s , s_1}^{m,\sigma,\sigma'}|^2}{j[2+(2s+1)\sigma']+\sigma\gamma b}\nonumber\\ & & + \frac{|{\cal y}_{s , s_1}^{m,\sigma,\sigma'}|^2}{j[2+(2s+1)\sigma']+2\sigma\gamma b}\bigr).\end{aligned}\ ] ] for simplicity , we rewrite this as @xmath447\nonumber\\ & & \times\bigl(\cos^2\theta[\tilde{j}_{b , a}^{s , s_1}-\cos(2\phi)\tilde{j}_{d , e}^{s , s_1}]^2\nonumber\\ & & \qquad+\sin^2(2\phi)(\tilde{j}_{d , e}^{s , s_1})^2\bigr),\label{esm2}\\ e_{s , s_1}^{m,(2){\cal v}}&=&-\frac{m}{8\gamma b}[2s(s+1)-2m^2 - 1]\nonumber\\ & & \times\bigl([\sin^2\theta\tilde{j}_{b , a}^{s , s_1}+(1+\cos^2\theta)\cos(2\phi)\tilde{j}_{d , e}^{s , s_1}]^2\nonumber\\ & & \qquad+4\cos^2\theta\sin^2(2\phi)(\tilde{j}_{d , e}^{s , s_1})^2\bigr),\\ e_{s , s_1}^{m,(2){\cal w}}&=&\frac{d_{s , s_1}^m}{16j}\bigl(j_a-3[j_a\cos^2\theta+j_e\sin^2\theta\cos(2\phi)]\bigr)^2,\nonumber\\ & & \label{esm2w}\\ e_{s , s_1}^{m,(2){\cal x}}&=&\frac{f_{s , s_1}^m(\gamma b / j)\sin^2\theta}{2j}\nonumber\\ & & \times\bigl([j_a - j_e\cos(2\phi)]^2\cos^2\theta\nonumber\\ & & \qquad+\sin^2(2\phi)j_e^2\bigr),\label{esm2x}\\ e_{s , s_1}^{m,(2){\cal y}}&=&\frac{g_{s , s_1}^m(\gamma b / j)}{64j}\bigl[\bigl(j_a\sin^2\theta\nonumber\\ & & \qquad+j_e(1+\cos^2\theta)\cos(2\phi)\bigr)^2\nonumber\\ & & \qquad+4\cos^2\theta\sin^2(2\phi)j_e^2\bigr],\label{esm2y}\end{aligned}\ ] ] where @xmath448\eta_{s , s_1}^2\eta_{s-1,s_1}^2}{(2s-1)}\nonumber\\ & & + \eta_{s+2,s_1}^2\eta_{s+1,s_1}^2\nonumber\\ & & \times \frac{[(s+1)^2-m^2][(s+2)^2-m^2]}{(2s+3)},\nonumber\\ f_{s , s_1}^m(x)&=&-\frac{\eta_{s , s_1}^2\eta_{s-1,s_1}^2(s^2-m^2)}{(2s-1)^2-x^2}\nonumber\\ & & \times\bigl((2s-1)[(s-1)(s-2)+m^2]\nonumber\\ & & \qquad - m(2s-3)x\bigr)\nonumber\\ & & + \frac{\eta_{s+2,s_1}^2\eta_{s+1,s_1}^2[(s+1)^2-m^2]}{(2s+3)^2-x^2}\nonumber\\ & & \times\bigl((2s+3)[(s+2)(s+3)+m^2]\nonumber\\ & & \qquad - m(2s+5)x\bigr),\\ g_{s , s_1}^m(x)&=&\frac{2\eta_{s , s_1}^2\eta_{s-1,s_1}^2}{(2s-1)^2 - 4x^2}\bigl((2s-1)[m^4\nonumber\\ & & + m^2(6s^2 - 18s+11)\nonumber\\ & & + s(s-1)(s-2)(s-3)]\nonumber\\ & & -4mx(2s-3)(s^2 - 3s+1+m^2)\bigr)\\ & & -\frac{2\eta_{s+2,s_1}^2\eta_{s+1,s_1}^2}{(2s+3)^2 - 4x^2}\bigl((2s+3)[m^4\nonumber\\ & & + m^2(6s^2 + 29s+34)\nonumber\\ & & + ( s+1)(s+2)(s+3)(s+4)]\nonumber\\ & & -4mx(2s+5)(s^2 + 5s+5+m^2)\bigr).\end{aligned}\ ] ] there is a remarkable amount of symmetry in the angular dependence of the eigenstate energies .
we note that @xmath449 and @xmath450 have the same forms , differing in the replacements of the interactions @xmath444 and @xmath445 with @xmath203 and @xmath204 , respectively , and with different overall constant functions .
the same comparison can also be made with @xmath451 and @xmath452 .
in addition , we note that there is a remarkable similarity in the @xmath8 dependence of @xmath453 with that of the local spin anisotropy part of @xmath454 in eq .
( [ bstep ] ) .
the contributions to the @xmath6th level crossing second order in the anisotropy interactions are calculated as indicated in eq .
( [ bstep2 ] ) , and are found to be @xmath455 @xmath456 ^ 2\cos^2\theta\nonumber\\ & & \qquad\qquad+j_d^2\sin^2(2\phi)\bigr),\\ f_2(\theta,\phi)&=&\frac{\sin^2\theta}{j}\bigl([j_b - j_d\cos(2\phi)][j_a - j_e\cos(2\phi)]\nonumber\\ & & \qquad\times\cos^2\theta + j_dj_e\sin^2(2\phi)]\bigr),\\ f_3(\theta,\phi)&=&\frac{\sin^2\theta}{j}\bigl([j_a-\cos(2\phi)j_e]^2\cos^2\theta\nonumber\\ & & \qquad\qquad+j_e^2\sin^2(2\phi)\bigr),\\ f_4(\theta,\phi)&=&\frac{1}{j}\bigl(j_b\sin^2\theta+j_d(1+\cos^2\theta)\cos(2\phi)\bigr)^2\nonumber\\ & & + 4\frac{j_d^2}{j}\cos^2\theta\sin^2(2\phi),\\ f_5(\theta,\phi)&=&\frac{1}{j}[j_b\sin^2\theta+j_d(1+\cos^2\theta)\cos(2\phi)]\nonumber\\ & & \times[j_a\sin^2\theta+j_e(1+\cos^2\theta)\cos(2\phi)]\nonumber\\ & & + 4\frac{j_dj_e}{j}\cos^2\theta\sin^2(2\phi),\\ f_6(\theta,\phi)&=&\frac{1}{j}\bigl(j_a\sin^2\theta+j_e(1+\cos^2\theta)\cos(2\phi)\bigr)^2\nonumber\\ & & + 4\frac{j_e^2}{j}\cos^2\theta\sin^2(2\phi),\\ f_7(\theta,\phi)&=&\frac{1}{j}\bigl(j_a-3[j_a\cos^2\theta+j_e\sin^2\theta\cos(2\phi)]\bigr)^2.\nonumber\\\end{aligned}\ ] ] the coefficients are given by @xmath457}{s(2s+1)(2s+3)},\\ a_{2,0}(s)&=&-3(9+s-23s^2 + 4s^3 + 12s^4),\nonumber\\ a_{2,1}(s)&=&-9 + 8s+4s^2,\\ a_3(s , s_1)&=&a_3^{\cal u}(s , s_1)+a_3^{\cal x}(s , s_1),\\ a_3^{\cal u}(s , s_1)&=&\frac{[3 - 3s-3s^2 + 4s_1(s_1 + 1)]^2}{2(2s+3)^2}\nonumber\\ & & -\frac{(s-1)[3 + 3s-3s^2 + 4s_1(s_1 + 1)]^2}{2s(2s+1)^2},\nonumber\\ & & \\ a_3^{\cal x}(s , s_1)&=&\frac{[s(s+2)-4s_1(s_1 + 1)]}{2(s+1)(s+3)(2s+1)^2(2s+3)^2}\nonumber\\ & & \times\frac{a_{3,0}^{\cal x}(s)+4s_1(s_1 + 1)a_{3,1}^{\cal x}(s)}{(2s+5)(3s+1)},\nonumber\\
a_{3,0}^{\cal x}(s)&=&(s+1)(s+3)(51 + 114s\nonumber\\ & & + 209s^2 + 302s^3 + 164s^4 + 24s^5),\\ a_{3,1}^{\cal x}(s)&=&129 + 318s+395s^2\nonumber\\ & & + 442s^3 + 300s^4 + 72s^5,\nonumber\\ a_{4}(s , s_1)&=&\frac{(4s-3)}{8s},\\ a_5(s , s_1)&=&-\frac{3[a_{5,0}(s)+4s_1(s_1 + 1)]}{4s(2s+1)(2s+3)},\\ a_{5,0}(s)&=&3 + 3s-5s^2 - 4s^3,\\ a_6(s , s_1)&=&a_6^{\cal v}(s , s_1)+a_6^{\cal y}(s , s_1),\\ a_6^{\cal v}(s , s_1)&=&\frac{[3 - 3s-3s^2 + 4s_1(s_1 + 1)]^2}{8(2s-1)(2s+3)^2}\nonumber\\ & & -\frac{(s-1)[3 + 3s-3s^2 + 4s_1(s_1 + 1)]^2}{8s(2s-3)(2s+1)^2},\nonumber\\ & & \\ a_6^{\cal y}(s , s_1)&=&\frac{-1}{96(2s-3)(2s-1)(2s+1)^2(2s+3)^2}\nonumber\\ & & \times\frac{\sum_{n=0}^2a_{6,n}^{\cal y}(s)[4s_1(s_1 + 1)]^n}{(2s+5)(4s+1)(4s+3)},\nonumber\\ a_{6,0}^{\cal y}(s)&=&s\bigl(-5184 + 20898s+123003s^2 \nonumber\\ & & + 81669s^3 - 304665s^4 - 521611 s^5\nonumber\\ & & -168482s^6 + 163132s^7 + 124888s^8\nonumber\\ & & + 12008s^9 + 256s^{10}+128s^{11}\bigr),\\ a_{6,1}^{\cal y}(s)&=&5832 + 6939s+855s^2 + 17694s^3\nonumber\\ & & + 22242s^4 - 1548s^5 - 3464s^6\nonumber\\ & & + 3728s^7 + 1920s^8 - 512s^9 - 256s^{10},\\ a_{6,2}^{\cal y}(s)&=&4536 + 5787s-3111s^2 + 4782s^3\nonumber\\ & & + 16972s^4 + 5144s^5 - 864s^6\nonumber\\ & & + 256s^7 + 128s^8,\\ a_7(s ,
s_1)&=&\frac{[s(s+2)-4s_1(s_1 + 1)]}{4(2s+1)^3(2s+3)^3(2s+5)}\nonumber\\ & & \times\bigl(4s_1(s_1 + 1)(-1 + 38s+24s^2 + 60s^3)\nonumber\\ & & + ( s+1)(3 + 67s+94s^2 + 44s^3 + 8s^4)\bigr).\nonumber\\\end{aligned}\ ] ] by expanding the solutions in the crystal representation to second order in the @xmath76 , we have explicitly checked these expressions for @xmath458 , and for @xmath30 , @xmath348 .
we note that for @xmath12 , only @xmath459 and @xmath460 are non - vanishing . | we present an exactly solvable model of equal spin @xmath0 dimer single molecule magnets .
the spins within each dimer interact via the heisenberg and the most general quadratic global and local ( single - ion ) anisotropic spin interactions , and with the magnetic induction @xmath1 . for antiferromagnetic couplings and @xmath2 , the low temperature @xmath3 magnetization @xmath4 exhibits @xmath5 steps of universal height and midpoint slope
, the @xmath6th step of which occurs at the non - universal level - crossing magnetic induction @xmath7 , where @xmath8 define the direction of @xmath9 . the specific heat @xmath10 exhibits zeroes as @xmath11 at these @xmath7 values , which are equally surrounded by universal peak pairs as @xmath11 .
the non - universal @xmath7 values lead to a rich variety of magnetization plateau behavior , the structure and anisotropy of which depend upon the various global and local anisotropic spin interaction energies .
we solve the model exactly for @xmath12 , 1 , and 5/2 , and present @xmath4 and @xmath13 curves at low @xmath3 for these cases .
for weakly anisotropic dimers , rather simple analytic formulas for @xmath4 and @xmath13 at arbitrary @xmath0 accurately fit the exact solutions at sufficiently low @xmath3 or large @xmath14 .
an expression for @xmath15 accurate to second order in the four independent anisotropy energies is derived .
our results are discussed with regard to existing experiments on @xmath16 fe@xmath17 dimers , suggesting further experiments on single crystals of these and some @xmath18 [ mn@xmath19@xmath17 dimers are warranted . |
the band structure calculations as well as the arpes measurements show that the fermi surface of iron - based superconductors consists of two hole ( h ) pockets centered around the @xmath61 point and two electron ( e ) pockets centered around the @xmath62 point of the _ folded _ brillouin zone ( bz ) @xcite .
we model the resulting band structure by using the following single - electron model hamiltonian @xmath63 where @xmath64 refer to the band indices , @xmath65 are the on - site single - electron energies , @xmath66 is the electronic dispersion that yields hole pockets centered around the @xmath67 point , and @xmath68 is the dispersion that results in the electron pockets around the @xmath69 point .
this model has been used previously to fit the available arpes data@xcite in optimally doped ba@xmath0k@xmath1fe@xmath70as@xmath70 .
it has been found recently@xcitethat the fermi surface and the corresponding electronic structure of fe@xmath36te@xmath37se@xmath38 is very similar to the 122 compounds except for the different level of nesting .
to account for this change we use the following parameters [ @xmath71 @xmath72 and @xmath73 for the @xmath74 and @xmath75 bands , respectively , and @xmath76 and @xmath77 for the @xmath78 and @xmath79 bands , correspondingly ( all values are in ev ) .
this parametrization is almost the same as the one for ba@xmath0k@xmath1fe@xmath70as@xmath70@xcite though the slightly different on - site energies and different parametrization of the @xmath75 band have been used to ensure the incommensurate nesting wave vectors between @xmath41 and @xmath42-bands .
next we consider the one - loop contribution to the spin susceptibility that includes the intraband and the interband contributions : @xmath80 \label{eq : bare_chi}\end{aligned}\ ] ] where @xmath81 , @xmath82 again refer to the different band indices . @xmath83 and @xmath84 are the normal and anomalous ( superconducting ) green functions , respectively . for
the four - band model considered here the effective interaction will consist of the intraband and interband repulsion denoted by @xmath85 and @xmath86 , respectively . within rpa
the spin response can be written in a matrix form : @xmath87^{-1 } \hat{\chi}_0({\bf q},{\rm i}\omega_m ) \label{eq : chi_rpa}\end{aligned}\ ] ] where @xmath88 is a unit matrix and @xmath89 is @xmath90 matrix formed by the interband and intraband bare susceptibilities determined by eq .
( [ eq : bare_chi ] ) .
the vertex is given by @xmath91 , \label{eq_coupling}\end{aligned}\ ] ] and we assume here @xmath92ev and @xmath93ev .
note that the value of @xmath85 was chosen in order to stay in the paramagnetic phase . | we report on inelastic neutron scattering measurements that find incommensurate itinerant like magnetic excitations in the normal state of superconducting fete@xmath0se@xmath1 ( @xmath2=14k ) at wave - vector @xmath3 with @xmath4=0.09(1 ) . in the superconducting state
only the lower energy part of the spectrum shows significant changes by the formation of a gap and a magnetic resonance that follows the dispersion of the normal state excitations .
we use a four band model to describe the fermi surface topology of iron - based superconductors with the extended @xmath5 symmetry and find that it qualitatively captures the salient features of these data .
magnetism is a key ingredient in the formation of cooper pairs in unconventional high temperature superconductors@xcite .
a consequence of magnetism and the symmetry of the superconducting order parameter is a magnetic resonance that has been detected through inelastic neutron scattering for a wide range of magnetic superconductors ranging from the heavy fermion systems@xcite , to cuprates@xcite , and more recently in the iron based superconductors@xcite . in each class of materials
the features of the resonance in @xmath6 are different and are directly related to the electronic degrees of freedom , providing vital clues to the role of magnetic fluctuations in each case .
we report inelastic neutron scattering measurements showing that the magnetic excitations in fete@xmath0se@xmath1 are incommensurate and itinerant - like , and that upon entering the superconducting state only the lower energy part of the spectrum shows significant changes by the formation of a gap and a magnetic resonance that follows the dispersion of the normal state excitations . using a four band model that describes the fermi surface topology of iron - based superconductors and the extended @xmath5 symmetry
, we can qualitatively describe the salient features of the data .
the good agreement between theory and experiment found here may provide clues to a better understanding of the magnetic resonance in high-@xmath2 cuprate superconductors . in cuprate superconductors a magnetic resonance , whose energy scales with the superconducting energy gap , is a saddle point of a broader spectrum of magnetic excitations @xcite .
the interpretation of the magnetic resonance and its relationship to superconductivity is complicated in the cuprates due to the occurrence of charge stripes and the pseudogap phase that lead to hotly debated models such as quantum excitations from charge stripes@xcite , or spin excitons from a particle - hole bound state of a @xmath7-wave superconductor@xcite .
a far clearer picture has emerged in the iron based superconductors partly due to their more itinerant nature .
it was realized early on that the nesting between hole and electron fermi surfaces related by the antiferromagnetic wavevector * q*@xmath8 in the undoped iron superconductors , might also be responsible for unconventional superconductivity of the so - called extended @xmath5-wave symmetry@xcite .
the remarkable feature of this superconducting order parameter is the different sign of the superconducting gap on bands that are separated by * q*@xmath9 can yield a magnetic resonance in the form of a spin exciton @xcite at the same wavevector .
this picture suggests that the continuum of magnetic excitations will be gapped in the superconducting state up to twice the energy of the superconducting gap , @xmath10 and that an additional feature , the magnetic resonance will occur at an energy @xmath11 .
the latter is a consequence of both the unconventional symmetry of the superconducting order parameter which enhances the continuum at @xmath10 , and the residual interaction between the quasiparticles that shifts the pole in the total susceptibility to lower energies .
while so far the magnetic resonance has been observed in iron based superconductors at commensurate wavevectors , the itinerant origin of the magnetic fluctuations suggests that in the doped superconducting compounds , the nesting could be incommensurate .
[ scale= 0.5 ] scans2 [ scale= 0.6 ] inset in order to understand the extended magnetic excitation spectrum in iron based superconductors we have chosen to examine the doped binary superconductor fete@xmath0se@xmath1 with a superconducting @xmath2=14k , using inelastic neutron scattering ( ins ) .
the availability of large and high - quality single crystals and the simplicity of its crystal structure ( tetragonal @xmath12 unit cell@xcite ) compared to other iron based superconductors make this system attractive for further examination . in this work
we describe the measurements in momentum space in units of q=(@xmath13,@xmath13 ) while we omit reference to the @xmath14-axis direction for clarity as q@xmath15=0 throughout . the portion of reciprocal space probed is shown in fig . [ highe](e ) , which was not explored in a previous study@xcite .
fete@xmath0se@xmath1 crystallize in a primitive tetragonal structure p@xmath16 , with @xmath17=3.80and @xmath14=6.02 .
high quality single crystal samples of this composition were prepared as described previously@xcite .
inelastic neutron scattering measurements were collected using the triple axis spectrometer in8 operated by the institut laue - langevin ( ill ) , with the sample mounted with the @xmath18axis vertical giving us access to spin excitations within the basal @xmath19-plane .
the neutron optics were set to focus on a virtual source of 30 mm .
measurements were made with the ( 002 ) reflection of a pyrolytic graphite monochromator and analyzer with an open / open / open / open configuration .
a graphite filter was placed after the sample position .
diaphragms were placed before and after the sample .
cooling of the sample was achieved by a standard orange cryostat .
data were collected around the ( 1/2,1/2,0 ) and ( 3/2,3/2,0 ) brillouin zones using a fixed final wave vector @xmath20 of either 2.66 or 4.1 @xmath21 . in the normal state we find steeply dispersive magnetic excitations that we have measured up to 80 mev , fig .
[ highe](a , b ) . at higher energies ,
two excitations are clearly resolved in the transverse scans . at lower energies ( 2 - 6 mev )
, the two peaks merge to a broad response , far wider than the resolution function ( @xmath22q@xmath23 0.06 r.l.u . measured on the ( 200 ) bragg peak ) .
these lower energy data at 20k are best modeled with two excitations as shown for example for the 6 mev data in fig .
[ highe](a ) . in fig .
[ lowe](a ) we show the lower portion of these transverse scans in the normal state as a color map where in addition we plot as black points the @xmath24-position of the excitations determined by fitting gaussian peaks to the data of fig .
[ highe ] .
the dispersion of these excitations is linear up to 80 mev as shown in fig.[res2](a ) , and the points for the lower energy excitations fall on the same line confirming our above analysis . extrapolating the centers of these excitations to zero energy , red lines in fig .
[ lowe](a ) , it is clear that they do not converge to the commensurate * q*@xmath25 wavevector but to the incommensurate positions of @xmath3 with @xmath4=0.09(1 ) .
we note that for the fete end - compound with lower excess fe , static antiferromagnetism appears at q=(1/2,0)@xcite .
longitudinal scans ( not shown ) between 2 and 80 mev through these spin excitations show that they are well defined isolated maxima and not parts of spin - wave cones as found in insulators , a conclusion concurred also in other recent ins measurements @xcite .
therefore , we argue that magnetic excitations in the normal state of the superconductor fete@xmath0se@xmath1 consist of a single - lode excitation continuum from each incommensurate @xmath26 .
this behavior in @xmath27 is reminiscent of fincher - burke excitations reported for many itinerant magnetic materials that are close to an antiferromagnetic instability@xcite and indicates that an itinerant , rather than a local moment picture is relevant for this superconductor .
[ scale= 0.35 ] maps cooling below @xmath2 we find substantial changes to the lower energy magnetic excitations , while our measurements indicate that the higher energy excitations remain largely unchanged when compared to the normal state measurements , see fig .
[ highe](b , d ) .
the most pronounced changes are in the lower energy part of the magnetic excitation spectrum where the opening of a spin gap and the development of the magnetic resonance is observed as shown in fig .
[ lowe](b ) . here
a spin gap opens below @xmath28=4 mev , while there is an enhancement of the spectral weight above the spin gap that peaks at @xmath28@xmath29=6 mev . at higher energies the spectral weight returns to approximately the same values as the data found in the normal state , see fig .
[ highe ] .
constant energy scans at the commensurate position @xmath30=(1/2,1/2 ) between these dispersions also reflect these changes as shown in reference .
although the normal state magnetic excitations extend from incommensurate points it is important to establish if the magnetic resonance itself is centered on the commensurate wavevector @xmath30=(1/2,1/2 ) or if it consists of two excitations originating from the two incommensurate wavevectors ( at @xmath26 ) found in the normal state . for a more detailed analysis of the q - scans across the resonance
we subtract the 20k from the 2k data in order to obtain only the magnetic scattering contributing to the resonance ( see below ) .
these data are shown in fig .
[ res2](b ) , while the centers of the resonant excitations are plotted in fig .
[ lowe](b ) . between @xmath28=7 and 12 mev , the data shows well separated resonant excitations that follow the normal state dispersions , while their intensity decreases with energy transfer giving only a minor superconducting enhancement at 12 mev . at the resonance ,
however , @xmath31 mev , a `` flat top '' excitation is evident that can be modeled with two incommensurate peaks at @xmath32 with @xmath4=0.05(1 ) , a value that departs slightly from the one found in the normal state .
the salient point of these measurements is that in the normal state the magnetic excitations show a behavior consistent with an itinerant antiferromagnet .
the onset of superconductivity affects only the lower energy magnetic excitations and spectral weight is shifted upwards to follow the normal - state incommensurate wave - vector dependence except at the spin resonance energy .
our ins results are qualitatively consistent with the extended s(@xmath33 ) model of the superconducting order parameter as we indeed observe the opening of a spin gap and the enhanced peak just above the gap as expected .
however it is less clear if the magnetic resonance and its spectral weight in @xmath27 is consistent with this model . to clarify this
we compute the magnetic susceptibility within the four - band model used previously to describe the fermi surface topology in iron - based superconductors@xcite and use the conventional random phase approximation ( rpa ) which describes the enhancement of the spin response of a metal in presence of the moderately strong ( interband and intraband ) repulsive interactions . in fig .
[ lowe](c , d ) we show the results for the total _ physical _ rpa susceptibility , @xmath34i@xmath35 , calculated for the fermi surface topology that is consistent with arpes results on fe@xmath36te@xmath37se@xmath38 @xcite as a function of the momentum along the transverse direction ( 1/2+@xmath4,1/2-@xmath4 ) and @xmath28 in the normal ( c ) and superconducting ( d ) states ( details of these calculations are presented in the supplementary information ) . in order to simulate the experimental resolution we have convoluted the results by the experimental gaussian with @xmath39r.l.u . and @xmath40mev and the original calculated figures
are shown in the appendix . due to the nesting condition ] of one of the hole - bands ( @xmath41 ) and the two electron - bands ( @xmath42 ) ( see fig . [ disp ] in the appendix ) we find that in the normal state the spin response shows antiferromagnetic excitations centered on an incommensurate transverse wavevector with @xmath4=0.03 .
these low - energy magnetic excitations in the metals , often called _ paramagnons _ , result from the proximity of the fermi surface to a magnetic instability and in our model we find that they disperse linearly from the incommensurate wavevectors . while the rpa theory indicates two peaks for each incommensurate wave vector ( counter propagating linearly dispersive modes ) ,
only a single peak is observed experimentally dispersing from each @xmath43 wave vector , suggesting stronger electronic damping than the current rpa treatment , as demonstrated previously by moriya et al.@xcite .
[ scale= 0.32 ] resonance2 in the superconducting state , the magnetic susceptibility changes due to opening of the superconducting gap and the corresponding change of the quasiparticle excitations . here
the imaginary part of the bare interband susceptibility ( i.e. without electron - electron interaction that yields rpa ) is gapped for small frequencies and at the antiferromagnetic wavevector @xmath44 , the value of this gap is determined as @xmath45mev @xcite where @xmath46 is the superconducting gap .
furthermore , above @xmath47 the imaginary part of the bare interband susceptibility shows the discontinuous jump implied by the bogolyubov coherence factor ( @xmath48 ) and the condition @xmath49 .
the latter is true for the @xmath50-wave symmetry of the superconducting order parameter .
correspondingly , the real part of the bare susceptibility is positive , diverges logarithmically at @xmath51 , and scales as @xmath52 at small frequencies@xcite . then , switching on the electron - electron interactions , we find within rpa that for any positive value of the interband interaction the total ( rpa ) spin susceptibility acquires a pole or _
spin exciton _ below @xmath47 with infinitely small damping .
therefore , other than the gapping of the spin spectrum , the most salient point of our analysis is that it predicts a spin exciton at @xmath53mev in the superconducting state that is located at the same incommensurate wavevectors as paramagnons in the normal state , a feature that is clearly found in our ins data .
we stress here that the spin exciton is a property of the superconducting state and requires opposite sign of the superconducting order at the corresponding wave vector , _
i.e. _ @xmath54 .
we remark that the actual spin gap in the neutron spectrum is smaller than @xmath55 which is again a consequence of the spin exciton formation in the superconducting state .
our itinerant rpa like description is obviously not sufficiently accurate to describe all the detailed features of the spin excitation spectrum .
for example , the value for the incommensurability @xmath4 is smaller than the experimentally observed and the dispersion of the paramagnons are steeper than observed as shown in fig [ lowe](c ) .
nevertheless , our treatment captures the key points of the data , these being the incommensurate paramagnon excitations , the emergence of the spin gap and the superposition of the exciton in the superconducting state onto the paramagnon scattering . at this point
it is tempting to compare the structure in @xmath27 of the magnetic resonance between the iron based superconductors and the cuprates .
we argue that there are two basic differences .
first , in the iron based compounds the superconducting gap is almost constant at each of the fermi surfaces , though the magnitude of the gap may differ on each of the pockets .
therefore , the resonance is not expected to show any additional features like a downward dispersion , as found in the cuprates .
secondly , in the iron based superconductors the main electronic interactions are local ( weakly momentum - dependent ) and smaller than the bandwidth .
the resonance in the superconducting state is an additional feature to the initial structure of the magnetic susceptibility in the normal state , determined by the nesting the fermi surface pockets .
once the nesting is incommensurate the resonance also shifts to the incommensurate wavevectors as we demonstrate in our measurements . to confirm this
we have checked that if the nesting is restored the resonance returns to be fully commensurate .
this is in contrast to the cuprates where the fermi surface is far from being nested and the maximum intensity at @xmath56 of the spin response is a consequence of the momentum - dependent strong superexchange interaction .
we note that there are intriguing similarities between this iron superconductor and the rare - earth- and actinide - based superconductors upd@xmath57al@xmath58@xcite , cecu@xmath57si@xmath57@xcite and cecoin@xmath59 @xcite . in all three compounds the magnetisation dynamics in the normal state
are modulated in @xmath27 . in the superconducting state ( i )
the low energy response is suppressed by the formation of a gap , ( ii ) a resonance is formed at energies related to the superconducting gap energy and ( iii ) the higher energy dynamic response remains unchanged . in this iron superconductor @xmath2 is much higher than the rare - earth- and actinide - based superconductors making this behavior far clearer to observe . to summarize , we find using inelastic neutron scattering that in the fete@xmath0se@xmath1 superconductor the magnetic resonance appears superimposed at the incommensurate wavevector as paramagnon excitations found in the normal state .
our simple rpa model of the magnetic excitations on the basis of the @xmath5 symmetry of the superconducting order parameters can qualitatively reproduce the main features of the neutron scattering data .
the agreement supports an itinerant character of the spin excitations in iron - based systems and a spin spectrum determined by the single poles of im@xmath60 with no extra contributions from localized moments .
the success of this model stands in sharp contrast to the uncertainties still faced to fully understand the magnetic resonance in the cuprates .
work at tulane was supported by the nsf under grant dmr-0645305 ( for materials ) and the doe under de - fg02 - 07er46358 ( for graduate students ) .
work at jhu was funded by the doe under de - fg02 - 08er46544 .
dna benefited from helpful discussions with jan zaanen and alan goldman .
mmk is grateful to p.j .
hirschfeld for useful discussions and acknowledges support from rfbr ( grant n 09 - 02 - 00127 ) and ofn ras program on strong electronic correlations |
due to the high centre of mass energy @xmath1 and presence of a lepton and quark in the initial state , searches for physics beyond the standard model ( sm ) in @xmath0 collisions at hera complement those made at lep and the tevatron .
indeed , much interest has focused on the recent results published by the zeus and h1 experiments @xcite in which an excess of events above standard model predictions was reported in the previously unexplored region of high @xmath2 .
no attempt will be made in this review , however , to discuss this excess within the context of new physics .
further discussion and the latest results in the high @xmath2 region can be found in the conference proceedings @xcite .
preliminary results are presented on the search for excited fermions and selectron and squark production within the framework of r - parity conserving supersymmetric models using a total integrated luminosity of 9.3 and 20 respectively .
results on the search for r - parity violating squarks are also discussed taking into account all possible r - parity violating and gauge decays and using an integrated luminosity of 2.83 .
the existence of excited leptons or quarks would provide compelling evidence for a new layer of fermion structure predicted within theories of compositeness . at hera single excited electrons , quarks and neutrinos can in principle be produced with masses up to the kinematic limit of 300 gev via the processes @xmath3 shown in figure [ fig : fstar_feyn ] .
the resonance is searched for via its electroweak decay to a light fermion , @xmath4 , and a vector boson , @xmath5 .
direct searches for excited leptons and quarks have also been carried out at lep and the tevatron . in @xmath6 collisions ,
excited leptons may be pair or singly produced up to @xmath7 . at 95% confidence level ( cl ) ,
any excited lepton @xmath8 ) with mass below approximately 80 gev has been ruled out @xcite .
this limit can be extended when the excited fermion is produced singly , however , becomes dependent on the coupling parameters at the @xmath9 vertex as shown in figure [ fig : fstar ] .
experiments at the tevatron have set stringent limits on the production of @xmath10 in @xmath11 collisions at @xmath12 , excluding @xmath10 produced via the strong coupling @xcite with masses between 80 and 760 gev , except for a window between 570 and 580 gev .
hera therefore extends the mass range over which @xmath13 and @xmath14 can be searched and offers unique sensitivity for excited quarks produced via electroweak mechanisms .
in describing excited fermion production and decay it has become conventional to use the phenomenological model of hagiwara et al @xcite , extended to include excited quark production by baur et al @xcite .
the effective lagrangian describing magnetic transitions from spin @xmath15 excited fermions @xmath16 to ordinary fermions @xmath4 has the form @xmath17 f + h.c.\ ] ] where @xmath18 corresponds to the compositeness scale and the constants @xmath4 , @xmath19 and @xmath20 describe the effective changes from the standard model coupling constants @xmath21 , @xmath22 and @xmath23 .
it is conventional to relate these unknown constants according to the chosen decay mode , thereby reducing the cross section dependence to a single parameter , @xmath24 . for excited electron and excited quark searches @xmath25 and @xmath26 is chosen , exploring excited quark production via electroweak mechanisms . the radiative decay of excited neutrinos ( @xmath27 ) is only allowed when @xmath28 and in this case @xmath29 and @xmath30 is chosen .
the zeus search for excited fermions in positron - proton collisions uses an integrated luminosity of 9.4 .
the @xmath16 decay modes which have been studied are shown in figure [ fig : fstar_feyn ] and the subsequent decay channels of the heavy boson are listed in table [ table : fstar ] .
a detailed description of the analysis is given in @xcite .
.[table : fstar ] the decay modes and subsequent final states considered in this search for excited fermions . [ cols="^,^,^,^",options="header " , ] figure [ fig : estarmass ] shows the mass of the reconstructed @xmath31 final state in events with two isolated , high @xmath32 electromagnetic clusters .
the data are compared to the predicted monte carlo background , comprising mainly qed compton and neutral current dis events .
the unshaded histogram shows the expected lineshape for an excited electron of mass 150 gev .
the cluster of four events at masses around 135 gev have features consistent with background expectations . in the absence of a positive signal for excited leptons or quarks in any of the eight decay modes considered , upper limits on the characteristic coupling @xmath24 as a function of the excited fermion mass
were determined at 95% confidence level .
figure [ fig : fstar ] shows the limits for @xmath13 , @xmath14 and @xmath10 production derived from each of the final states considered and from their combination using the assumptions about the couplings discussed previously . in each case
the most stringent limits are derived from the @xmath33 channel and dominate the combined limit from all three decay modes .
the limits on @xmath13 and @xmath14 production from the lep experiments are also shown and demonstrate that hera extends the limits for @xmath16 production well beyond the @xmath34 gev limits currently achieved at lep .
also shown on the @xmath35 plot is the limit from the previous search by zeus in @xmath36 collisions with a much lower integrated luminosity ( 0.55 ) .
this previous limit is more stringent for excited neutrino masses greater than 130 gev because the @xmath35 production cross section in @xmath37 collisions is heavily suppressed relative to @xmath36 collisions , out - weighing the increase in integrated luminosity taken with positrons .
high luminosity running with an @xmath38 beam planned for the 1998 run - period will substantially improve this limit .
exclusion limits on the excited fermion masses can be determined by further assuming that @xmath39 . for this case , at 95% cl , excited electrons are ruled out in the mass range between 30 and 200 gev using the combined limit from all three decay modes .
excited neutrinos with masses in the range 40 to 96 gev are excluded while excited quarks with electroweak - only couplings are excluded over the range 40 to 196 gev .
supersymmetry ( susy ) is presently considered to be a promising candidate for a theory beyond the standard model . while there is strong theoretical motivation for its existence ,
there is currently no experimental evidence to support its prediction that each sm particle has a supersymmetric partner differing by a half - unit of spin . within the minimal supersymmetric standard model ( mssm ) it is assumed that r - parity is conserved .
this implies that susy particles can only be produced in pairs and that the lightest susy particle ( lsp ) , taken here to be the lightest neutralino is stable and weakly interacting thereby escaping detection .
existing limits from the lep experiments at @xmath40 exclude selectrons with mass less than @xmath41 80 gev @xcite and spartners to the light quarks with masses less that 45 gev ( lep1 ) @xcite .
although the experiments at the tevatron set strong limits on squark masses , these are dependent on the gluino mass @xcite and can only be related to the hera results assuming additional relations motivated by grand unified theories ( gut ) . at hera
the dominant mssm process is the production of a selectron and squark via @xmath42 channel neutralino exchange @xmath43 as shown in figure [ fig : mssm_feyn ] .
although the and can decay into any lighter gaugino and their sm partners , the cleanest experimental signature involves both the @xmath44 and @xmath45 decays and is the one considered here .
the final state topology consists therefore of an electron which is acoplanar to the hadronic system and missing momentum .
limits from a earlier search using the h1 detector and an integrated luminosity of 6.4 have been published in @xcite .
the current search by zeus uses an integrated luminosity of 20 .
no signal was observed and preliminary exclusion limits on production were derived at @xmath46 confidence level .
the limits are interpreted as exclusion regions in mssm parameter space .
the quantities used are @xmath47 and @xmath48 , the mass parameters for the @xmath49 amd @xmath50 gauginos , the higgsino mass parameter @xmath51 , and @xmath52 , the ratio of the two higgs doublet vacuum expectation values .
figure [ fig : m0mq ] shows the excluded region in the plane defined by the mass of the lightest neutralino and half the sum of the selectron and squark masses for @xmath53 and different values of @xmath51 . for small values of @xmath51 ( @xmath54 ) and low neutralino masses ,
competing decays of the @xmath55 and @xmath56 to gauginos other than the lead to reduced limits for @xmath57 . in the region @xmath58 ,
the branching ratios @xmath59 and @xmath60 are close to one because the other gauginos are heavier than @xmath61 or @xmath62 . at large @xmath63
the mass difference between the , and becomes small leading to a reduction in cross section together with a drop in detection efficiency limiting the upper exclusion bound .
the excluded mass range extends to 72 gev for @xmath57 and 47 gev for @xmath63 , extending the previous limits set by h1 @xcite with lower luminosity .
figure [ fig : memq ] shows the excluded region in @xmath61 versus @xmath62 for different values of fixed mass at @xmath54 .
the limits are approximately functions of the sum @xmath64 due to the cross section and efficiency depending almost essentially on this sum . for @xmath65 squarks with masses between 55 gev and 89 gev
are excluded for @xmath66 while selectrons with mass between 55 gev and 92 gev are excluded for @xmath67 .
the selectron mass exclusion limits from lep2 @xcite are also indicated in figure [ fig : memq ] and show that the new hera limits exclude a further region at high selectron and low squark masses . in figure
[ fig : mum2 ] the exclusion limits on @xmath47 versus @xmath51 are shown for @xmath53 and @xmath68 , a good approximatation for the only susy gut relation used here @xmath69 . for @xmath70 , the is dominated by its photino component leading to couplings to the selectron and squark that are electromagnetic in strength and allow for a sizable cross section .
as @xmath71 the becomes higgsino - like and the couplings and cross section become very small .
the zeus limits are compared to those from chargino and neutralino searches at lep @xcite and extend considerably the limits on @xmath47 at negative @xmath51 .
the search for r - parity violating squarks possessing a yukawa coupling @xmath72 to lepton - quark pairs is particularly promising at hera .
such squarks can be singly produced as an @xmath73 channel resonance up to the kinematic limit of @xmath74 via positron - quark fusion .
the squarks subsequently decay via their yukawa coupling into fermions or via their gauge couplings into a quark and neutralino or a chargino as shown in figures [ fig : rparity_feyn](a ) , ( c ) and ( b ) , ( d ) respectively . the @xmath75 and @xmath76 are mixed states of the supersymmetric partners to the sm gauge bosons and neutral higgses and are in general unstable .
in contrast to the mssm , this also holds for the lsp which can decay into a quark , antiquark and lepton . in the case
where both the squark production and decay occur through a yukawa coupling indices correspond to to the generations of superfields @xmath77 and @xmath78 containing the left - handed lepton doublet and quark doublet , and the right - handed quark singlet respectively ] the final state signatures consist of a lepton and jet and are indistinguishable from standard neutral and charged current deep inelastic scattering on an event - by - event basis .
the strategy then involves searching for a resonance at high mass ( @xmath79 ) where @xmath80 is the bjorken scaling variable .
no attempt is made here to interpret the recently reported excess of events at high @xmath80 and @xmath2 observed at the hera experiments @xcite within the context of r - parity violating susy . using an integrated luminosity of 2.8 the squark mass
is reconstructed using the @xmath81 energy and polar angle in the selected neutral current sample and the hadronic final state in the charge current sample .
figures [ fig : h1_squark](a ) and ( b ) show the reconstructed mass for the neutral and charged current event samples respectively .
the shaded histogram indicates the background monte carlo while the dashed histogram compares the mass distribution for a mixture of background and susy mc events for @xmath82 and @xmath83 respectively .
no significant deviation from the number of expected events is observed when including all sources of systematic error .
squark decays via gauge couplings @xmath84 or @xmath85 and @xmath86 have also been considered ( figure [ fig : rparity_feyn ] ( b ) and ( d ) ) .
furthermore , both gauge and r - parity violating decays of the chargino have been taken into account together with the dependence of the decay on its gaugino - higgsino composition governed by the fundamental susy parameters .
in particular if the is dominated by its photino component it will tend to decay @xmath87 , giving the possibility of a `` wrong sign '' lepton ( with respect to the charge of the incoming lepton beam ) in the final state . ' '' ''
a & , & @xmath56 & & @xmath88 & & @xmath89 & high @xmath90 @xmath81 + & , , & @xmath91 & & @xmath92 & & @xmath93 & + + & , & @xmath91 & & @xmath94 & & @xmath95 & multiple jets + & & & & & & @xmath96 & + ' '' '' b & , & @xmath56 & & @xmath97 & & @xmath98 & high @xmath90 @xmath38 + & , & @xmath91 & & @xmath94 & & @xmath95 & + + & & & & & & @xmath99 & multiple jets + ' '' '' c & , & @xmath56 & & @xmath97 & & @xmath100 & + & , & @xmath91 & & @xmath94 & & @xmath101 & missing @xmath90 + & & & & & & @xmath102 & + + & , , & @xmath91 & & @xmath94 & & @xmath103 & multiple jets + & & @xmath91 & & @xmath94 & & @xmath101 & + & & & & & & @xmath104 & + ' '' '' d & & @xmath91 & & @xmath105 & & @xmath95 & high @xmath90 @xmath81 or @xmath106 + & & & & & & @xmath107 & missing @xmath90 + 1 jet + ' '' '' e & , & @xmath91 & & @xmath105 & & @xmath101 & high @xmath90 @xmath108 + @xmath81 or @xmath106 + & & & & & & @xmath109 & + missing @xmath90 + & & & & & & & + multiple jets + ' '' ''
f & , & @xmath91 & & @xmath105 & & @xmath101 & high @xmath90 @xmath81 or @xmath106 + & & & & & & @xmath110 & + missing @xmath90 + & & & & & & & + multiple jets + six distinguishable event topologies are listed in table [ table : susy_top ] and have been considered in this search , full details of which are given in @xcite .
no significant excess of events was observed in any channel , however , an interesting event with a final state @xmath51 and jet passed the selection cuts for topology d in table [ table : susy_top ] and is shown in figure [ fig : theevent ] .
detailed analysis of the event @xcite shows the @xmath106 has a @xmath111 and there is an overall missing @xmath112 , taking the @xmath90 of the @xmath106 into account .
the dominant source of background is expected from single @xmath113 production followed by the decay @xmath114 .
however the properties of this event are such that there is only a @xmath115 probability for such an interpretation and a negligible contribution from misidentified charged current dis .
a second similar candidate has been observed during 1997 data - taking and is shown in figure [ fig : the2ndevent ] , however more data will be required to establish the origin of these events . in the absence of a significant deviation from sm expectations , 95% confidence level limits limits on the yukawa coupling
are derived as a function of squark mass combining all contributing channels .
figure [ fig : h1squark_limits ] shows the limits assuming that the lightest neutralino is a pure photino and for different photino masses .
first generation squarks with masses up to 240 gev are excluded at 95% confidence level for coupling strengths @xmath116 , thereby extending the limits inferred from the tevatron di - lepton data @xcite . from the analysis of the case , limits on the couplings
can be deduced by folding in the proper parton densities .
the results given in table [ table : yukawa ] are for @xmath82 and @xmath117 for -dominant and -dominant .
no other direct limits for where @xmath118 or @xmath119 have been published . & & & + ' '' '' @xmath120 & @xmath121 & @xmath122 & 0.056 & 0.048 + & @xmath123 & @xmath124 & & + ' '' '' @xmath125 & @xmath121 & @xmath126 & 0.14 & 0.12 + & @xmath127 & @xmath124 & & + ' '' '' @xmath128 & @xmath129 & @xmath130 & 0.18 & 0.15 + & @xmath131 & @xmath132 & & + ' '' '' @xmath133 & @xmath134 & @xmath122 & 0.058 & 0.048 + & @xmath135 & @xmath136 & & + ' '' '' @xmath137 & @xmath134 & @xmath126 & 0.19 & 0.16 + & @xmath127 & @xmath136 & & + ' '' '' @xmath138 & @xmath134 & @xmath130 & 0.30 & 0.26 + & @xmath139 & @xmath136 & & + ' '' '' @xmath140 & @xmath141 & @xmath122 & 0.06 & 0.05 + & @xmath142 & @xmath143 & & + ' '' '' @xmath144 & @xmath141 & @xmath126 & 0.22 & 0.19 + & @xmath127 & @xmath143 & & + ' '' '' @xmath145 & @xmath146 & @xmath130 & 0.55 & 0.48 + & @xmath147 & @xmath143 & & +
the cases where @xmath148 or @xmath148 are of particular interest as they allow for the direct production of stop . a light
stop mass eigenstate , , much lighter than the top quark or other squarks could exist .
the search for such a stop mass eigenstate is extended towards lower masses by considering @xmath149 pair production via photon - gluon fusion as shown in figure [ fig : stop_feyn ] . here
the is assumed to be lighter than the top quark and the lightest chargino so that the decays into @xmath150 and @xmath151 are forbidden so that the decay @xmath152 will dominate .
the final state for such a signature will comprise two isolated and acollinear high @xmath32 electrons together with two high @xmath32 jets and no missing momentum .
details of the analysis are published in @xcite . as no signal
was observed using = 2.83 , exclusion limits were derived at 95% cl , combining those channels from the previous search in which single is produced .
figure [ fig : stop_limits ] shows the limits on the coupling @xmath153 as a function of @xmath154 .
the exclusion limits from the pair production analysis , indicated by the shaded area , are independent of the coupling and exclude masses between 9 and 24.4 gev .
the region above the curve in figure [ fig : stop_limits ] is excluded by the single stop search .
for coupling strengths @xmath155 stop masses below 138 gev are excluded . in the region where the stop mass limits extend beyond the top mass , the limits are only valid when @xmath156 .
the zeus and h1 experiments have found no evidence for excited fermions or supersymmetry using an integrated luminosity of up to 20 .
upper limits at 95% confidence level have been established on excited electron and neutrino production which are more stringent than the current limits at lep .
the limits on excited quark production via electroweak mechanisms complement those measurements made at the tevatron where their strong production is explored .
new limits on susy particle production at hera have been established within the context of the mssm and r - parity violating susy , including limits on stop production .
the future prospects for discovery at hera continue to improve as the luminosity delivered by the hera accelerator continues to increase . in 1997
each experiment expects to collect an integrated luminosity of @xmath157 of @xmath37 collision data , doubling the total data sample collected pre-1997 . in 1998 - 99
hera will operate with an @xmath38 beam and the goal is to deliver @xmath158 per experiment , thereby improving the sensitivity for the search for excited neutrinos , certain leptoquarks and r - parity violating squarks .
hera will undergo a major luminosity upgrade in the shutdown between 1999 - 2000 , beyond which each experiment expects to collect a total of 1000 . 99 zeus collaboration , j. breitweg et al .
, z. phys .
c74 ( 1997 ) 207 ; + h1 collaboration , c. adloff et al , z. phys .
c74 ( 1997 ) 191 .
transparencies shown by j. meyer ( h1 ) , d.c . williams ( zeus ) and d.k .
krakauer are available from + http://sbhep1.physics.sunysb.edu/@xmath159hadrons/program.html opal collaboration , k. ackerstaff et al .
b391 ( 1997 ) 197 ; + delphi collaboration , p. abreu et al .
b393 ( 1997 ) 245 ; + l3 collaboration , m. acciarri et al .
lett b401 ( 1997 ) 139 .
cdf collaboration , f. abe et al .
, phys . rev lett 74 ( 1995 ) 3538 ; + cdf collaboration , f. abe et al .
d55 ( 1997 ) 5263 .
k. hagiwara , s. komamiya and d. zeppenfeld , z. phys .
c29 ( 1985 ) 115 .
u. baur , m. spira and p.m. zerwas , phys .
rev d42 ( 1990 ) 815 .
zeus collaboration , j. breitweg et al .
, desy-97 - 112 , june 1997 .
opal collaboration , k. ackerstaff et al .
b396 ( 1997 ) 301 .
opal collaboration , m.z .
akrawy et al . , phys .
b240 ( 1990 ) 261 ; + delphi collaboration , p. abreu et al .
lett . b247 ( 1990 ) 148 ; + aleph collaboration ,
d. decamp et al .
216 ( 1992 ) 253 .
d0 collaboration , s. abachi et al .
, phys . rev lett 75 ( 1995 ) 618 ; + cdf collaboration f. abe et al , phys .
76 ( 1996 ) 2006 .
h1 collaboration , s. aid et al .
b380 ( 1996 ) 461 .
aleph collaboration , d. buskulic et al .
, z. phys .
c72 ( 1996 ) 549 .
h1 collaboration , s. aid et al . , z. phys .
c71 ( 1996 ) 211 .
h1 collaboration , t. ahmed et al .
, desy 94 - 248 ( december 1994 ) d.p .
roy , phys .
lett . b283 ( 1992 ) 270 . | recent results on the searches for new physics in @xmath0 collisions at hera using the zeus and h1 detectors are presented .
no evidence for excited fermions or supersymmetric particles within the context of the minimal supersymmetric standard model nor r - parity violation has been found using an integrated luminosity of up to 20 .
new limits have therefore been established .
+ _ talk given at the hadron collider physics conference , stony brook , june 1997 , representing the zeus and h1 collaborations _ |
work at the university of tennessee is supported by the materials science division , office of basic energy sciences at the u.s .
department of energy under grant de - fg02 - 01er45885 .
research at binghamton is supported by the division of materials research , national science foundation under grant 0313963 .
work at university of arizona is supported by materials science division , office of basic energy sciences at the u.s .
department of energy under grant de - fg02 - 06er46315 .
we thank h. zhao for constructing fig .
1(b ) .
j. cao , j. choi , j. l. musfeldt , s. lutta , and m. s. whittingham , chem .
* 16 * , 731 ( 2004 ) ; j. cao , j. choi , j.l .
musfeldt , s. lutta , and m.s .
whittingham , nanoscale materiales : from science to technology , ch .
21 , nova science publishers ( 2006 ) .
a. barbour , r. d. luttrell , j. choi , j. l. musfeldt , d. zipse , n. s. dalal , d. w. boukhvalov , v. v. dobrovitski , m. i. katsnelson , a. i. lichtenstein , b. n. harmon , and p. kgerler , phys .
b. * 74 * , 014411 ( 2006 ) . | we measured the optical properties of mixed valent vanadium oxide nanoscrolls and their metal exchanged derivatives in order to investigate the charge dynamics in these compounds .
in contrast to the prediction of a metallic state for the metal exchanged derivatives within a rigid band model , we find that the injected charges in mn@xmath0 exchanged vanadium oxide nanoscrolls are pinned .
a low - energy electronic excitation associated with the pinned carriers appears in the far infrared and persists at low temperature , suggesting that the nanoscrolls are weak metals in their bulk form , dominated by inhomogeneous charge disproportionation and madelung energy effects .
the discovery that low - dimensional inorganic solids can curve or fold into nanoscale objects provides an exciting opportunity to investigate bulk versus nanoscale chemistry using molecular - level strain as the tuning parameter .
@xcite these materials are targets of intensive research efforts , driven by the need to further miniaturize electronic devices and the potential to exploit unusual mechanical and optical properties . some of the beautiful , flexible , and functional nanomorphologies include tubes , wires , octahedra , particles , urchins , and spheres .
@xcite they offer molecular - level control of size , shape , mechanical response , and chemical composition as well as unusual confinement effects due to finite length scales .
( color online ) ( a ) representative x - ray powder diffraction scan of ( c@xmath1h@xmath2nh@xmath3)@xmath4-vo@xmath5 scrolls showing only 00l and weak hk0 reflections .
the inset displays a representative tem image with lattice fringing and a center cavity .
( b ) probable double - layer structure of the vo@xmath5 scrolls , analogous to the bav@xmath6o@xmath7 model compound .
@xcite ( c ) close - up view of the local structure around the v centers , showing the arrangement of octahedra and tetrahedra in the bav@xmath6o@xmath7 model compound .
@xcite , width=3 ] among the transition metal oxides , vanadates show particularly rich chemistry due to the tunable vanadium oxidation state and flexible coordination environment , which ranges from octahedral to square pyramidal to tetrahedral with increasing vanadium oxidation state . @xcite vanadium oxides form many layered and nanoscale compounds with open structural frameworks , making them prospective materials for ion intercalation , exchange and storage . @xcite the nanoscale vanadates of interest here , ( amine)@xmath4vo@xmath5 , are formally mixed - valent@xcite with @xmath8 @xmath9 2.4 and @xmath10 @xmath9 0.28 .
stoichiometric considerations do not , of course , distinguish between a true mixed - valent v@xmath11 state that may be metallic and a state with an inhomogenous charge distribution , consisting of v - ions with distinct small and large charges .
such charge disproportionation is common in oxides of ti and v with formal oxidation states of ti@xmath12 and v@xmath11 , respectively , and the convention is to assign integer charges ( ti@xmath13 and ti@xmath14 ; v@xmath14 and v@xmath15 ) to the ions with different oxidation states.@xcite we adopt the same convention here , and consider the possibility of charge disproportionation in the nanoscale vanadates .
the compounds we consider are actually scrolls , consisting of vanadate layers between which organic molecules are intercalated .
@xcite the size of the amine or diamine template determines scroll winding , providing an opportunity to tune the size of these materials .
the typical scrolls are @xmath915 - 100 nm in diameter , containing up to 30 vanadium oxide layers ( inset of fig .
[ fig_dvostruc](a ) ) .
the exact crystal structure is complicated and is not completely understood .
figure [ fig_dvostruc](a ) shows a representative x - ray powder diffraction pattern for the ( c@xmath1h@xmath2nh@xmath3)@xmath4-vo@xmath5 scrolls .
the ( 00l ) reflections from the vanadium oxide layers are clear , but the ( hk0 ) reflections from the atomic structure within each layer are weak .
the data suggest a tetragonal lattice of basal repeat size 6.144 , closely related to that of bav@xmath6o@xmath7 , which has a planar structure of basal size 6.160 .
@xcite the arrangement in bav@xmath6o@xmath7 is shown in fig .
[ fig_dvostruc](b ) .
@xcite it contains zig - zag chains of edge - sharing vo@xmath16 square pyramids .
these chains share some corners with each other to form a two dimensional sheet .
the sheets also share corners with vo@xmath17 tetrahedra , bringing the two vo@xmath5 layers together to form a characteristic double sheet .
figure [ fig_dvostruc](c ) displays the local structure around the v centers in the bav@xmath6o@xmath7 model compound .
each two - dimensional layer contains two octahedrally coordinated vanadium atoms , v(1 ) and v(2 ) , with one tetrahedrally coordinated v(3 ) occurs in between the layers .
scrolled vanadates exhibit very interesting physical properties including a large spin gap , diameter - dependent optical features , and potential battery and optical limiting applications .
@xcite recently , electron- and hole - doped vanadium oxide nanoscrolls were reported to be 300 k ferromagnets , raising fundamental questions about the underlying magnetic exchange mechanism .
@xcite it was further suggested that charge - injection via `` doping '' introduces free carriers into the pristine mott insulating scrolls , resulting in partially filled bands .
the `` doped '' nanoscrolls are predicted to exhibit drude - type metallic behavior within this picture .
@xcite direct measurement of the low - energy electronic structure is clearly important to test this prediction .
further , the rigid band model raises fundamental questions about the chemical nature of the `` doping '' or ion exchange process in the scrolled vanadates .
@xcite from the chemical perspective , metal intercalation does not appear as simple as `` putting in electrons '' ; both reduction of vanadium as well as ion exchange with the proton on the amine are possible consequences of the exchange process . to date , only mn@xmath0 has been shown to completely replace the organic template , even though some other ions such as na@xmath18 , li@xmath18 , zn@xmath0 , cu@xmath0 , and ca@xmath0 can partially replace the amine template .
@xcite in order to understand the fundamental charge dynamics and test the applicability of the rigid band model in vanadium oxide nanoscrolls , we investigated the variable temperature optical spectra of the pristine and mn@xmath0 exchanged materials .
in contrast to the rigid band model expectation for a metallic state , we find that charge is pinned in the metal substituted scrolls .
the low - energy electronic excitation associated with the pinned carriers appears in the far infrared and persists at low temperature , suggesting that the nanoscrolls are weak metals in their bulk form , dominated by inhomogeneous charge disproportionation and madelung energy effects .
we propose an alternate model for charge injection to account for our observations .
analysis of the vibrational properties shows that the 575 @xmath19 v - o - v equatorial stretching mode is very sensitive to metal substitution , indicating that ion exchange modifies both the local curvature and charge environment .
charge effects also shift excitations in the higher energy optical conductivity .
vanadate nanoscrolls were prepared by an initial sol - gel reaction followed by hydrothermal treatment , as described previously .
@xcite appropriate amines ( c@xmath20h@xmath21nh@xmath22 with @xmath23=4 - 18 ) were employed as the structure - directing agent .
ion exchange was preformed by stirring a mixture of vanadate nanoscrolls and mcl@xmath22 ( m = mn@xmath0 , zn@xmath0 , and na@xmath18 ) for 2 hours in ethanol / water mixture .
the spectroscopic experiments were carried out over a wide energy range ( 4 mev - 6.2 ev ; 30 - 50000 @xmath19 ) using a series of spectrometers .
@xcite variable temperature work was done between 4.2 and 300 k using an open - flow helium cryostat and temperature controller .
the optical constants were calculated by a kramers - kronig analysis of the measured reflectance : @xmath24 .
@xcite ( color online ) 300 k optical conductivity of pristine vo@xmath5 scrolls and the mn@xmath0 exchanged compound in the far - infrared regime .
mode assignments include : v - o axial stretching at 945 and 995 @xmath19 , weak v - o - v equitorial stretching at 790 and 866 @xmath19 , v - o - v equitorial stretching at 585 , 652 , and 727 @xmath19 , v - o - v axial stretching at 486 and 498 @xmath19 , v - o bending at between 179 and 381 @xmath19 , and screw - like motion of the scroll at 113 @xmath19.@xcite the dashed green line guides the eye , highlighting the additional localized carrier contribution in the substituted scrolls .
the inset shows a magnified view of the low - frequency response of pristine and mn@xmath0 exchanged scrolls .
the extrapolated dc conductivity is indicated by filled circles at @xmath25 = 0 in both panels.,width=3 ] figure [ fig_dvolowfre ] shows the optical conductivity of the pristine vo@xmath5 scrolls and the mn@xmath0 substituted compounds at room temperature .
the pristine scrolls display semiconducting behavior , with clearly resolved vibrational modes below 1000 @xmath19 and a low background conductivity .
these modes were previously assigned as v - o - v axial and equatorial stretching , v - o bending , and screw - like motion of the scrolls .
@xcite subtle spectral changes occur when the amine template is exchanged for mn@xmath0 , zn@xmath0 , or na@xmath18 . in the mn@xmath0 substituted compound ( fig .
[ fig_dvolowfre ] ) , the optical conductivity develops a broad electronic background in the far infrared region , with strong phonons riding top of this excitation .
the fano - like lineshapes of several phonons , especially those near 250 and 380 @xmath19 , indicate that the background is electronic in nature.@xcite this localized electronic excitation has a width of @xmath9400 - 500 @xmath19 . extrapolating @xmath26 to zero frequency , we find @xmath27 @xmath91 @xmath28 @xmath19 .
such a pinned carrier response is characteristic of a weak or `` bad '' metal rather than a traditional drude metal and has been observed in other complex oxides.@xcite we obtain similar results at low temperature , suggesting that the vanadium centers in both pristine and ion - exchanged scrolls are charge disproportionated , with v@xmath14 and v@xmath15 centers rather than the mixed valent state ( v@xmath11 ) that might be anticipated for a highly conducting material . based on the observation of a pinned electronic excitation in the far infrared , the ion exchange processes does add carriers to the scrolled vanadates .
the carriers are not , however , mobile in the ion substituted scrolls .
single scroll scanning tunneling microscope measurements are in progress to complement this spectroscopic work with local di / dv measurements
. previous individual tube measurements of li@xmath18 exchanged scrolls actually showed a decrease in conductivity with li@xmath18 hole doping.@xcite ( color online ) schematic representation of possible electronic structure changes resulting from the ion exchange process in the scrolled vanadates .
( a ) band structure of pristine vanadium oxide nanoscrolls . here , @xmath29 denotes the @xmath90.5 ev optical gap.@xcite ( b ) schematic view in which ion exchange adds carriers to a new defect - level band that forms near @xmath30 .
( c ) cartoon view in which the new charge defect band splits due to the madelung energy difference that exists in any charged system , electron - electron interactions , and/or chemical disorder effects . here , @xmath31 represents the pinned low - energy electronic excitation .
this structure will likely have a relatively narrow bandwidth because the excitation is between split defect levels .
( d ) alternate approach to the exchange process in which ion exchange adds carriers within the rigid band model.@xcite a metallic state might be anticipated to arise from the partially - filled band .
( e ) schematic view in which the v @xmath32 band splits due to chemical disorder effects . here
, @xmath31 represents a possible low energy excitation , although it might be anticipated to display a wider band width than in scheme ( c ) due to joint density of states considerations .
, width=3 ] we now discuss the evolution of the electronic structure of the nanoscrolls due to the metal exchange process within a heuristic correlated - electron model .
we use an `` effective '' band picture , even though a localized description may be more suitable for charge - disproportionated v - oxides , @xcite in order to maintain continuity with the existing literature .
@xcite the pristine scrolls , depicted in fig .
[ fig_dvomodel](a ) , are semiconductors with an optical gap of @xmath90.5 ev .
@xcite this gap derives from a superposition of both v on - site @xmath33 to @xmath33 excitations and v@xmath14 to v@xmath15 charge transfer excitations . @xcite
thus charge disproportionation exists already in the pristine material .
optical gaps with similar magnitude are observed in bulk vo@xmath22 , ladder - like @xmath34@xmath35-nav@xmath22o@xmath16 , tubular na@xmath22v@xmath3o@xmath36 , and certain molecular magnets.@xcite within the rigid band scenario , the metal exchange process merely involves charge injection into the v 3@xmath33 conduction band , which is now partially - filled ( see fig .
[ fig_dvomodel](d)).@xcite the spectral response of the mn@xmath0-substituted nanoscrolls in fig .
[ fig_dvolowfre ] , however , clearly precludes the metallic state expected from the bandfilling model . even if the partially - filled band of fig .
[ fig_dvomodel](d ) is split due to chemical disorder effects ( fig .
[ fig_dvomodel](e ) ) , the resulting low energy excitation would be fairly broad due to joint density of states considerations in which the narrow occupied level below @xmath30 is coupled to the wider empty conduction band arising from the v @xmath32 states .
we believe that the optical response indicates that the ion exchange process injects charges that are strongly localized due to disorder and madelung energy considerations .
any charge introduced into the system can in principle reduce a v@xmath15 to a v@xmath14 ion .
this extra electron remains pinned , however , as mn@xmath0 ions will `` bind '' preferentially ( though not entirely ) to the oxygen atoms bonded to a v@xmath14 ion than to a v@xmath15 ion , since proximity to a v@xmath15 ion would raise the madelung energy substantially . thus even in the `
doped' system there exists a barrier to electron hopping between neighboring v@xmath14 and v@xmath15 ions . within the effective band model
therefore the charging process leads to strongly localized deep defect levels as shown in fig .
[ fig_dvomodel](b ) .
the d.c .
conductivity may be dominated by hopping between the disordered v@xmath14 and v@xmath15 ions .
the nonbonding character of the localized levels gives them their narrow widths .
the local potentials of the two kinds of ions continue to be different due to electron - electron repulsions between neighboring v@xmath14-v@xmath14 ions and long - range madelung energy involving the mn@xmath0 ions .
we have depicted this in fig .
[ fig_dvomodel](c ) by an energy splitting between `` occupied '' ( predominantly v@xmath14 ) and `` unoccupied '' ( predominantly v@xmath15 ) defect levels . a dipole - allowed transition between these split defect levels would be consistent with a localized charge excitation rather than a traditional metallic response in the low energy region of the optical conductivity .
such a feature would be relatively narrow , consistent with the 400 - 500 @xmath19 linewidth observed in the spectrum of mn@xmath0 substituted scrolls .
the above localized picture is in better accord with our experimental results for the pristine and mn@xmath0 exchanged scrolls .
careful examination of vibrational mode patterns can also provide information on the ion exchange process , although in this case , we must account for both charge and structural modifications .
this analysis is possible because vibrational features are observed in both the pristine and ion exchanged materials . in the pristine vanadium oxide nanoscrolls , the mode at @xmath9575 @xmath19 is assigned as the v - o - v equatorial stretch .
@xcite this mode is very sensitive to curvature , widening as the size of the amine template is reduced , characteristic of a more strained lattice .
@xcite the v - o - v equatorial stretching mode ( fig .
[ fig_dvolowfre ] ) redshifts with ion substitution , moving from @xmath9575 @xmath19 in the pristine scrolls , to @xmath9570 @xmath19 in na@xmath18 and @xmath9560 @xmath19 in mn@xmath0 exchanged compounds .
similar shifts in peak position are also observed in other doped and intercalated vanadates .
@xcite from the structural point of view , the redshifted v - o - v equitorial streching mode in the scrolled vanadates is unexpected because x - ray results indicate that the interlayer distance decreases as the smaller metal ions replace the larger amines .
@xcite the v - o - v equatorial stretching mode might therefore be expected to widen in the substituted scrolls due to their increased curvature , but the center peak position ought to be relatively insensitive to size .
@xcite ion exchange also adds charge to the scrolls .
we attribute changes in the v - o - v equatorial stretching mode position ( fig .
[ fig_dvolowfre ] ) to charging effects which overcome the aforementioned local strain that tends to widen and slightly harden the v - o - v resonance .
this overall softening trend ( @xmath9575 @xmath19 in the pristine scrolls to @xmath9560 @xmath19 in the mn@xmath0 substituted compound ) confirms that the ion exchange process adds carriers to the scrolls .
the sharp , unscreened vibrational features in the ion substituted scrolls show , however , that the carriers are not mobile , consistent with the observation of a pinned charge excitation in the far infrared .
the metal exchange process also leads to some chemical disorder , as evidenced in the splitting of a low frequency bending motion ( @xmath9179 @xmath19 ) and the screw - like mode of the scroll at @xmath9112 @xmath19 that is analogous to the radial breathing mode in carbon nanotubes ( inset , fig .
[ fig_dvolowfre ] ) .
disorder in cu@xmath0 exchanged nanoscrolls has been reported recently as well .
@xcite ( color online ) expanded view of the 300 k optical conductivity of pristine vo@xmath5 scrolls and the mn@xmath0 exchanged compound .
the inset shows a close - up view of the @xmath95 ev excitation , which changes modestly with sheet distance in the unsubstituted scrolls.@xcite in the inset , the curves are offset for clarity .
, width=3 ] figure [ fig_dvoelectr ] displays the 300 k optical conductivity of the pristine vo@xmath5 scrolls and the mn@xmath0 exchanged compound over a broader energy range . with the exception of the aforementioned bound carrier excitation localized in the far infrared ( fig .
[ fig_dvolowfre ] ) , the spectrum of the ion exchanged scrolls is similar to that of the pristine compound , in line with the fact that these excitations derive from the vanadium oxide framework .
@xcite both materials display similar @xmath90.5 ev optical gaps and excitations centered near 1 ev that derive from a superposition of v @xmath33 to @xmath33 on - site excitations and v@xmath14 to v@xmath15 charge transfer excitations.@xcite the o @xmath37 to v @xmath33 charge transfer excitations at @xmath93.9 and @xmath95.0 ev redshift with ion substitution . that this is a charge ( rather than size ) effect is evident from a comparison of the data in the inset of fig .
[ fig_dvoelectr ] . here , the excitation centered at @xmath95.0 ev , which is probably polarized in the radial direction , shifts to higher energy as the size of the amine template decreases ( corresponding to a reduction in the interlayer distance).@xcite the 3.9 ev feature does not change with curvature .
@xcite the mn@xmath0 substituted scrolls display the opposite trend , indicating that charge effects also modify the charge transfer excitations .
first principles electronic calculations will offer microscopic insight on this problem . we measured the optical spectra of pristine and mn@xmath0 substituted vanadium oxide nanoscrolls in order to understand the charge dynamics in the pristine and metal exchanged materials and to test the applicability of the rigid band model .
in contrast to expectations from the rigid band model , the spectra display a pinned low - energy electronic excitation in the mn@xmath0 substituted nanoscrolls rather than a traditional metallic response .
we propose that the injected charge is localized due to disorder and madelung energy effects .
our model is consistent with inhomogeneous charge disproportionation ( v@xmath14 and v@xmath15 ) in both the pristine and the ion - exchanged materials , and explains the observed far infrared charge localization .
analysis of the vibrational properties shows that the 575 @xmath19 v - o - v equatorial stretching mode redshifts with ion substitution , indicating that ion exchange modifies both the local curvature and charge environment .
charge effects also redshift two high energy electronic excitations . |
a fundamental theorem of dirac @xcite states that any graph on @xmath1 vertices with minimum degree at least @xmath3 contains a hamilton cycle .
a natural question is whether this theorem can be extended to hypergraphs . for this
, we first need to extend the notions of minimum degree and of hamilton cycles to hypergraphs .
a _ @xmath0-uniform hypergraph _ or _ @xmath0-graph _ @xmath4 consists of a vertex set @xmath5 and a set of edges each consisting of @xmath0 vertices
. we will often identify @xmath4 with its edge set and write @xmath6 if @xmath7 is an edge of @xmath4 . given a @xmath0-graph @xmath4
, we say that a set of @xmath8 vertices @xmath9 has _
the _ degree of @xmath11 _ is @xmath12 .
the _ minimum degree _ of @xmath4 is the minimum size of such a neighbourhood , that is , @xmath13 .
we say that a @xmath0-graph @xmath14 is a _ cycle of order @xmath1 _ if its vertices can be given a cyclic ordering @xmath15 so that every consecutive pair @xmath16 lies in an edge of @xmath14 and every edge of @xmath14 consists of @xmath0 consecutive vertices .
a cycle of order @xmath1 is _ tight _ if every set of @xmath0 consecutive vertices forms an edge ; it is _ loose _ if every pair of adjacent edges intersects in a single vertex , with the possible exception of one pair of edges , which may intersect in more than one vertex .
this final condition allows us to consider loose cycles whose order is not a multiple of @xmath8 .
figure [ fig : cycles ] shows the structure of each of these cycle types .
a _ hamilton cycle _ in a @xmath0-graph @xmath4 is a sub-@xmath0-graph of @xmath4 which is a cycle containing every vertex of @xmath4 .
rdl , ruciski and szemerdi @xcite showed that for any @xmath17 there is an @xmath18 so that if @xmath19 then any @xmath0-graph @xmath4 on @xmath1 vertices with minimum degree @xmath20 contains a tight hamilton cycle ( this improved an earlier bound by katona and kierstead @xcite ) .
they gave a construction which shows that this result is best possible up to the error term @xmath21 .
in this paper , we prove the analogous result for loose hamilton cycles .
[ main ] for all @xmath22 and any @xmath17 there exists @xmath18 so that if @xmath19 then any @xmath0-graph @xmath4 on @xmath1 vertices with @xmath23 contains a loose hamilton cycle .
the case when @xmath24 was proved by khn and osthus @xcite .
we will use a similar method of proof for general @xmath0-graphs , but this will be greatly simplified by the use of the recent blow - up lemma of keevash @xcite .
proposition [ bestpos ] shows that theorem [ main ] is best possible up to the error term @xmath21 .
in fact , proposition [ bestpos ] actually tells us more than this , namely that up to the error term , this minimum degree condition is best possible to ensure the existence of any ( not necessarily loose ) hamilton cycle in @xmath4 .
this means that the minimum degree needed to find a hamilton cycle in a @xmath0-graph of order @xmath1 is @xmath25 .
whilst finalizing this paper we learnt that hn and schacht @xcite independently and simultaneously proved theorem [ main ] , using a different approach .
the result in @xcite also covers the notion of a @xmath0-uniform @xmath26-cycle for @xmath27 ( here one requires consecutive edges to intersect in precisely @xmath26 vertices ) .
more recently khn , mycroft and osthus @xcite further developed the method of hn and schacht to include all @xmath26 such that @xmath28 ( the remaining values of @xmath26 are covered by the results of rdl , ruciski and szemerdi @xcite ) .
there is also the notion of a _
berge - cycle _ , which consists of a sequence of vertices where each pair of consecutive vertices is contained in a common edge .
this is less restrictive than the cycles considered in this paper .
hamiltonian berge - cycles were studied in @xcite . [
fig : cycles ]
the next proposition shows that theorem [ main ] is best possible , up to the error term @xmath21 .
[ bestpos ] for all integers @xmath29 and @xmath30 , there exists a @xmath0-graph @xmath4 on @xmath1 vertices such that @xmath31 but @xmath4 does not contain a hamilton cycle .
* let @xmath32 and @xmath33 be disjoint sets of size @xmath34 and @xmath35 respectively .
let @xmath4 be the @xmath0-graph on the vertex set @xmath36 , with @xmath37 an edge if and only if @xmath38 , that is , if @xmath7 contains at least one vertex from @xmath32 .
then @xmath4 has minimum degree @xmath39 .
however , any cyclic ordering of the vertices of @xmath4 must contain @xmath40 consecutive vertices @xmath41 from @xmath33 , but then @xmath42 and @xmath43 can not be contained in a common edge consisting of @xmath0 consecutive vertices , and so @xmath4 can not contain a hamilton cycle . in our proof of theorem
[ main ] we construct the loose hamilton cycle by finding several paths and joining them into a spanning cycle . here
a @xmath0-graph @xmath44 is a _ path _ if its vertices can be given a linear ordering such that every edge of @xmath44 consists of @xmath0 consecutive vertices , and so that every pair of consecutive vertices of @xmath44 lie in an edge of @xmath44 .
similarly as for cycles , we say that a path @xmath44 is _ loose _ if edges of @xmath44 intersect in at most one vertex .
the ordering of the vertices of @xmath44 naturally gives an ordering of the edges of @xmath44 .
we say that any vertex of @xmath44 which lies in the initial edge of @xmath44 , but not the second edge of @xmath44 , is an _
initial vertex_. similarly , any vertex of @xmath44 which lies in the final edge of @xmath44 but not the penultimate edge is a _
final vertex_. also , we refer to vertices of @xmath44 which lie in more than one edge of @xmath44 as _ link vertices_. thus , for example , a loose path @xmath44 has @xmath8 initial vertices , @xmath8 final vertices , and one link vertex in each pair of consecutive edges . in section [ sec : regularity ]
, we shall introduce various ideas we will need in the proof of theorem [ main ] . in particular , we will state a version of the hypergraph regularity lemma due to rdl and schacht @xcite and theorem [ robust - universal ] due to keevash @xcite .
the latter provides a useful way of applying the hypergraph blow - up lemma . in section [ sec : prelims ] , we shall prove various auxiliary results , including a result on finding loose paths in complete @xmath0-partite @xmath0-graphs , and an approximate minimum degree condition to guarantee a near - perfect packing of @xmath4 with a particular @xmath0-graph @xmath45 .
finally , in section [ proof ] we shall prove theorem [ main ] as follows .
in section [ structure ] we use the hypergraph regularity lemma to split @xmath4 into @xmath0-partite @xmath0-graphs @xmath46 on disjoint vertex sets @xmath47
. these @xmath0-graphs @xmath46 will be suitable for embedding almost spanning loose paths , and all the vertices of @xmath4 not contained in any of the @xmath47 will be included in an ` exceptional ' loose path @xmath48 ( actually , if @xmath49 is not divisible by @xmath8 , then @xmath48 will contain two consecutive edges which intersect in more than one vertex ) . the requirement that @xmath46 contains an almost spanning loose path means that the vertex classes of the @xmath46 must have suitable size .
we achieve this by first defining a suitable ` reduced @xmath0-graph ' @xmath50 of @xmath4 .
then we cover almost all vertices of @xmath50 by copies of a suitable auxiliary @xmath0-graph @xmath51 .
for each copy of @xmath51 , the corresponding sub-@xmath0-graph of @xmath4 is then split into the same number of disjoint @xmath46 . in section [ linking ] we shall use the structure imposed on @xmath4 to find a hamilton cycle in @xmath4 by the following process . * the @xmath0-graphs @xmath46 are connected by means of a walk @xmath52 in the ` supplementary graph ' . this graph ( which we will define in section [ supp ] )
has vertices @xmath53 corresponding to the @xmath0-graphs @xmath46 . *
using lemma [ interpath ] , each edge @xmath54 of @xmath55 is used to create a short ` connecting ' loose path @xmath56 in @xmath4 joining two different @xmath46s . *
@xmath48 and the paths @xmath56 are extended to ` prepaths ' ( these can be thought of as a path minus an initial vertex and a final vertex ) @xmath57 and @xmath58 , where @xmath59 and all @xmath60 are sets of size @xmath61 .
these prepaths have the property that there are large sets @xmath62 and @xmath63 such that @xmath64 can be extended to a loose path by adding any vertex of @xmath62 as an initial vertex and any vertex of @xmath63 as a final vertex .
similarly there are large sets @xmath65 and @xmath66 so that @xmath67 can be extended to a path by adding any vertex of @xmath65 as an initial vertex and any vertex of @xmath66 as a final vertex . @xmath68 and @xmath63
both lie in the same @xmath46 ( for all @xmath69 ) . * for each @xmath46 and for all those pairs @xmath70 which lie in @xmath46 , we choose a loose path @xmath71 inside @xmath46 from @xmath63 to @xmath68 . for each @xmath72
, we will use the hypergraph blow - up lemma ( in the form of theorem [ robust - universal ] ) to ensure that together all those @xmath73 which lie in @xmath46 use all the remaining vertices of @xmath46 . *
the loose hamilton cycle is then the concatenation @xmath74 .
note that the number of vertices of a loose path is @xmath75 modulo @xmath8 .
so in order to apply theorem [ robust - universal ] to obtain spanning loose paths in a subgraph of @xmath46 , we need this subgraph to satisfy this condition .
so we choose our paths sequentially to satisfy the following congruences modulo @xmath8 .
* @xmath48 is chosen with @xmath76 .
* let @xmath77 be the subset of @xmath47 obtained by removing @xmath78 .
( all the @xmath47 will be disjoint from @xmath79 . )
let @xmath80 be the number of times that @xmath55 visits @xmath46 .
when choosing @xmath56 , for every @xmath47 it traverses ( except the final one ) we arrange to intersect @xmath77 in a set of size @xmath81 ( the size modulo @xmath8 of the intersection of @xmath56 with the final @xmath47 it traverses is then determined by the sizes of the other intersections ) .
the choice of @xmath48 in ( a ) ensures that after all @xmath56 have been picked , the remaining part @xmath82 of @xmath47 has size @xmath83 . *
each @xmath56 is extended to a prepath @xmath64 by adding @xmath84 and @xmath85 .
similarly , @xmath48 is extended into a prepath @xmath67 by adding @xmath86 and @xmath87 .
now the remaining part of @xmath47 has size @xmath88 . *
it remains to select @xmath80 paths @xmath73 within each @xmath47 : each uses @xmath89 vertices , so the divisibility conditions are satisfied .
we begin with some notation . by @xmath90
$ ] we denote the set of integers from 1 to @xmath91 . for a set @xmath92
, we use @xmath93 to denote the collection of subsets of @xmath92 of size @xmath0 , and similarly @xmath94 to denote the collection of non - empty subsets of @xmath92 of size at most @xmath0 .
we write @xmath95 to mean that @xmath96 .
we shall omit floors and ceilings throughout this paper whenever they do not affect the argument .
a _ hypergraph _ @xmath4 consists of a vertex set @xmath97 and an edge set , such that each edge @xmath7 of the hypergraph satisfies @xmath98 .
so a @xmath0-graph as defined in section [ intro ] is a hypergraph in which all the edges are of size @xmath0 .
we say that a hypergraph @xmath4 is a _
@xmath0-complex _ if every edge has size at most @xmath0 and @xmath4 forms a simplicial complex , that is , if @xmath99 and @xmath100 then @xmath101 . as for @xmath0-graphs
we identify a hypergraph @xmath4 with the set of its edges .
so @xmath102 is the number of edges in @xmath4 , and if @xmath103 and @xmath4 are hypergraphs then @xmath104 is formed by removing from @xmath103 any edge which also lies in @xmath4 .
if @xmath4 is a hypergraph with vertex set @xmath5 then for any @xmath105 the _ restriction @xmath106 $ ] of @xmath4 to @xmath107 _ is defined to have vertex set @xmath107 and all edges of @xmath4 which are contained in @xmath107 as edges . also , for any hypergraphs @xmath103 and @xmath4 we define @xmath108 to be the hypergraph @xmath109 $ ] .
we say that a hypergraph @xmath4 is _
@xmath91-partite _ if its vertex set @xmath110 is divided into @xmath91 pairwise - disjoint parts @xmath111 , in such a way that for any edge @xmath6 , @xmath112 for each @xmath72 .
we call the @xmath113 the _ vertex classes _ of @xmath4 and say that the partition @xmath111 of @xmath110 is _ equitable _ if all the @xmath113 have the same size .
we say that a set @xmath114 is _
@xmath91-partite _ if @xmath115 for each @xmath72 . so every edge of an @xmath91-partite hypergraph is @xmath91-partite . in the same way we may also speak of @xmath91-partite @xmath0-graphs and @xmath91-partite @xmath0-complexes . given a @xmath0-graph @xmath4
, we define a @xmath0-complex @xmath116 and @xmath117 and a @xmath118-complex @xmath119 and @xmath117 .
conversely , for a @xmath0-complex @xmath4 we define the @xmath0-graph @xmath120 to be the ` top level ' of @xmath4 , i.e. @xmath121 .
( here @xmath122 . )
given a @xmath0-graph @xmath103 and a set @xmath55 of vertices of @xmath103 , we denote by @xmath123 $ ] the sub-@xmath0-graph of @xmath103 obtained by removing all vertices and edges not contained in @xmath55 ( in this case , we say @xmath103 is _ restricted to @xmath55 _ ) .
for a @xmath0-graph @xmath103 and a sub-@xmath0-graph @xmath124 write @xmath108 for @xmath125 $ ] .
let @xmath111 be pairwise - disjoint sets of vertices , and let @xmath126 .
given @xmath127}{\leq k}$ ] , we write @xmath128 for the complete @xmath129-partite @xmath130-graph whose vertex classes are all the @xmath113 with @xmath131 .
the _ index _ of an @xmath91-partite subset @xmath132 of @xmath110 is @xmath133 : s \cap x_i \neq \emptyset\}$ ]
. furthermore , given any set @xmath134 , we write @xmath135 . similarly , given @xmath127}{\leq k}$ ] and
an @xmath91-partite @xmath0-graph or @xmath0-complex @xmath4 on the vertex set @xmath110 we write @xmath136 for the collection of edges in @xmath4 of index @xmath92 and let @xmath137 .
in particular , if @xmath4 is a @xmath0-complex then @xmath138 is the set of all those vertices in @xmath113 which lie in an edge of @xmath4 ( and thus form a ( singleton ) edge of @xmath4 ) . in general , we will often view @xmath136 as an @xmath91-partite @xmath130-graph with vertex set @xmath110 . also , given a @xmath0-complex @xmath4 we similarly write @xmath139 and @xmath140 .
we write @xmath141 for the @xmath130-graph whose edges are those @xmath91-partite sets @xmath142 of index @xmath92 for which all proper subsets of @xmath132 belong to @xmath4 .
( in other words , a set @xmath132 with index @xmath92 satisfies @xmath143 if and only if for all @xmath144 the edges of @xmath4 which have size @xmath145 and are subsets of @xmath132 form a complete @xmath145-graph on @xmath146 vertices . )
then the _ relative density of @xmath4 at index @xmath92 _ is @xmath147 .
the _ absolute density of @xmath136 _ is @xmath148 .
( note that @xmath149 . )
if @xmath4 is a @xmath0-partite @xmath0-complex we may simply write @xmath150 for @xmath151})$ ] . similarly , the _ density _ of a @xmath0-partite @xmath0-graph @xmath4 on @xmath152 is @xmath153}(x)|$ ] . finally , for any vertex @xmath154 of a hypergraph @xmath4
, we define the _ vertex degree _ @xmath155 of @xmath154 to be the number of edges of @xmath4 which contain @xmath154 .
note that this is not the same as the degree defined earlier , which was for sets of @xmath8 vertices .
the _ maximum vertex degree _ of @xmath4 is then the maximum of @xmath155 taken over all vertices @xmath156 .
the _ vertex neighbourhood _ @xmath157 of @xmath154 is the set of all vertices @xmath158 for which there is an edge of @xmath4 containing both @xmath159 and @xmath154 .
for a @xmath0-partite @xmath0-complex @xmath4 on the vertex set @xmath160 we also define the _ neighbourhood complex _
@xmath161 of a vertex @xmath162 for some @xmath72 to be the @xmath118-partite @xmath118-complex with vertex set @xmath163 and edge set @xmath164 .
in this subsection we shall define the concept of regular complexes ( which was first introduced in the @xmath0-uniform case by rdl and skokan @xcite ) in the form used by rdl and schacht @xcite .
this is a generalization of the standard concept of regularity in graphs , where we say that a bipartite graph @xmath165 on vertex classes @xmath166 and @xmath5 forms an @xmath167-regular pair if for any @xmath168 and @xmath105 with @xmath169 and @xmath170 we have @xmath171 ) = d(b)\pm { \epsilon}$ ] . in the same way
, we say that a @xmath0-complex @xmath103 is regular if the restriction of @xmath103 to any large subcomplex of lower rank has similar densities to @xmath103 .
more precisely , let @xmath103 be an @xmath91-partite @xmath0-complex on the vertex set @xmath126 .
for any @xmath127}{\leq k}$ ] , we say that @xmath172 is _ @xmath167-regular _ if for any @xmath173 with @xmath174 we have @xmath175 we say @xmath103 is _ @xmath167-regular _ if @xmath172 is @xmath167-regular for every @xmath127}{\leq k}$ ] .
note that if @xmath103 is a graph without isolated vertices , then the definition in the previous paragraph is equivalent to the 2-complex @xmath176 being @xmath167-regular . to illustrate the definition for @xmath24 ,
suppose that @xmath177 $ ] .
then for instance the top level of @xmath178}$ ] is the bipartite subgraph of @xmath103 induced by @xmath179 and @xmath180 and @xmath181 is the set of ( graph ) triangles in @xmath103 .
so roughly speaking , the regularity condition states that if we consider a subgraph of @xmath178 } \cup g_{\{1,3\ } } \cup g_{\{2,3\}}$ ] which spans a large number of triangles , then the proportion of these which also form an edge of @xmath172 is close to @xmath182 , i.e. close to the proportion of ( graph ) triangles in @xmath103 between @xmath179 , @xmath180 and @xmath183 which form an edge of @xmath103 . roughly speaking ,
the hypergraph regularity lemma states that an arbitrary @xmath0-graph can be split into pieces , each of which forms a regular @xmath0-complex .
the version of the regularity lemma we shall use also involves the notion of a ` partition complex ' , which is a certain partition of the edges of a complete @xmath0-complex .
as before , let @xmath126 be an @xmath91-partite vertex set .
a _ partition @xmath0-system @xmath44 on @xmath110 _ consists of a partition @xmath184 of the edges of @xmath128 for each @xmath127}{\leq k}$ ] .
we refer to the partition classes of @xmath184 as _ cells_. so every edge of @xmath128 is contained in precisely one cell of @xmath184 .
@xmath44 is a _ partition @xmath0-complex on @xmath110 _ if it also has the property that whenever @xmath185 lie in the same cell of @xmath184 , we have that @xmath186 and @xmath187 lie in the same cell of @xmath188 for any @xmath189 .
this property of @xmath190 forms an equivalence relation on the edges of @xmath128 , which we refer to as _ strong equivalence_. to illustrate this , again suppose that @xmath24 and @xmath177 $ ] . then if @xmath44 is a partition @xmath0-complex , @xmath191 , @xmath192 and @xmath193 together yield a vertex partition @xmath194 refining @xmath195 .
@xmath194 naturally induces a partition @xmath196 of the @xmath197 complete bipartite graphs induced by the pairs @xmath198 .
@xmath199 , @xmath200 and @xmath201 also yield a partition @xmath202 of these complete bipartite graphs .
the requirement of strong equivalence now implies that @xmath202 is a refinement of @xmath196 . at the next level
, @xmath202 naturally induces a partition @xmath203 of the set of triples induced by @xmath204 and @xmath183 .
as before , strong equivalence implies that the partition @xmath205 of these triples is a refinement of @xmath203 .
let @xmath44 be a partition @xmath0-complex on @xmath206 . for @xmath207 $ ] ,
the cells of @xmath208 are called _ clusters _
( so each cluster is a subset of some @xmath113 ) .
we say that @xmath44 is _ vertex - equitable _ if all clusters have the same size .
_ @xmath209-bounded _
if @xmath210 for every @xmath92 ( i.e. if @xmath128 is divided into at most @xmath209 cells by the partition @xmath184 ) .
also , for any @xmath91-partite set @xmath211 , we write @xmath212 for the set of all edges lying in the same cell of @xmath44 as @xmath213 , and write @xmath214 for the @xmath91-partite @xmath0-complex whose vertex set is @xmath110 and whose edge set is @xmath215 .
( since @xmath44 is a partition @xmath0-complex , @xmath216 is indeed a complex . )
the partition @xmath0-complex @xmath44 is _
@xmath167-regular _ if @xmath214 is @xmath167-regular for every @xmath91-partite @xmath217 .
given a partition @xmath118-complex @xmath44 on @xmath110 and @xmath127}{k}$ ] , we can define an equivalence relation on the edges of @xmath128 , namely that @xmath185 are equivalent if and only if @xmath186 and @xmath187 lie in the same cell of @xmath44 for any strict subset @xmath218 .
we refer to this as _ weak equivalence_. note that if the partition complex @xmath44 is @xmath209-bounded , then @xmath128 is divided into at most @xmath219 classes by weak equivalence .
if we let @xmath103 be an @xmath91-partite @xmath0-graph on @xmath110 , then we can use weak equivalence to refine the partition @xmath220 of @xmath128 ( i.e. two edges of @xmath172 are in the same cell if they are weakly equivalent and similarly for the edges not in @xmath172 ) .
together with @xmath44 , this yields a partition @xmath0-complex which we denote by @xmath221 $ ] .
if @xmath221 $ ] is @xmath167-regular then we say that @xmath103 is _
perfectly @xmath167-regular with respect to @xmath44_. note that if @xmath221 $ ] is @xmath167-regular then @xmath44 must be @xmath167-regular too .
finally , we say that @xmath91-partite @xmath0-graphs @xmath103 and @xmath4 on @xmath110 are _
@xmath222-close _ if @xmath223 for every @xmath127}{k}$ ] , that is , if there are few edges contained in @xmath103 but not in @xmath4 and vice versa .
we can now present the version of the regularity lemma we shall use to split our @xmath0-graph @xmath4 into regular @xmath0-complexes .
it actually states that there is some @xmath0-graph @xmath103 which is close to @xmath4 and which is regular with respect to some partition complex .
this will be sufficient for our purposes , as we shall avoid the use of any edges in @xmath104 , so every edge used will lie in both @xmath103 and @xmath4 .
there are various other forms of the regularity lemma for @xmath0-graphs which give information on @xmath4 itself ( the first of these were proved in @xcite ) but these do not have the hierarchy of densities necessary for the application of the blow - up lemma ( see @xcite for a fuller discussion of this point ) . the version below is due to rdl and schacht @xcite ( actually it is a very slight restatement of their result ) .
[ eq - partition ] suppose integers @xmath224 and reals @xmath225 satisfy @xmath226 and where @xmath227 divides @xmath1 .
suppose also that @xmath4 is an @xmath91-partite @xmath0-graph whose vertex classes @xmath111 form an equitable partition of its vertex set @xmath110 , where @xmath228 .
then there is an @xmath209-bounded @xmath167-regular vertex - equitable partition @xmath118-complex @xmath44 on @xmath110 and an @xmath91-partite @xmath0-graph @xmath103 on @xmath110 that is @xmath222-close to @xmath4 and perfectly @xmath167-regular with respect to @xmath44 . here
( and later on ) we write @xmath229 to mean that we can choose the constants @xmath230 from right to left .
more precisely , there are increasing functions @xmath231 such that , given @xmath232 , whenever we choose some @xmath233 , @xmath234 and @xmath235 , all calculations needed in the proof of the subsequent statement are valid .
hierarchies with more constants are defined similarly .
one important property of regular complexes is that they remain regular when restricted to a large subset of their vertex set .
for regular @xmath0-partite @xmath0-complexes this property is formalised by the following lemma , a special case of lemma 6.18 in @xcite .
[ restrict ] suppose @xmath236 , and that @xmath103 is an @xmath167-regular @xmath0-partite @xmath0-complex on the vertex set @xmath237 such that @xmath238 for each @xmath72 and @xmath239 .
let @xmath55 be a subset of @xmath110 such that @xmath240 for each @xmath72 .
then the restriction @xmath123 $ ] of @xmath103 to @xmath55 is @xmath241-regular , with @xmath242 ) > d(g)/2 $ ] and @xmath243}(g[w ] ) > d_{[k]}(g)/2 $ ] . apart from theorem [ eq - partition ] , the other main tool we shall use in the proof of theorem [ main ] is the recent hypergraph blow - up lemma of keevash .
this result involves not only a @xmath0-complex @xmath103 , but also a @xmath0-graph @xmath244 of ` marked ' edges on the same vertex set .
if the pair @xmath245 is ` super - regular ' , then this blow - up lemma can be applied to embed any spanning bounded - degree @xmath0-complex in @xmath246 , that is , within @xmath103 but avoiding any marked edges .
we will apply this with @xmath247 where @xmath103 is the @xmath0-graph given by theorem [ eq - partition ] .
super - regularity is a stronger notion than regularity .
a result in @xcite states that every @xmath167-regular @xmath0-complex can be made super - regular by deleting a few of its vertices .
unfortunately , the notion of hypergraph super - regularity is very technical , but the following definition from @xcite avoids many of these technicalities . let @xmath248 be a @xmath0-partite @xmath0-complex . roughly speaking
, we say that @xmath248 is robustly @xmath249-universal if the following holds : even after the deletion of many vertices of @xmath248 , the resulting complex @xmath250 has the property that one can find in @xmath250 a copy of any @xmath0-partite @xmath0-complex @xmath251 which has vertex degree at most @xmath249 and whose vertex classes are the same as those of @xmath250 . condition ( i ) puts a natural restriction on the number of vertices we are allowed to delete from the neighbourhood complex of a vertex of @xmath250 and condition ( iii ) states that for a few vertices @xmath159 of @xmath251 we can even prescribe a ` target set ' in @xmath252 into which @xmath159 will be embedded . * definition .
( robustly universal complexes ) * suppose that @xmath248 is a @xmath0-partite @xmath0-complex on @xmath253 with @xmath254 for each @xmath255 $ ] .
we say that @xmath248 is _
@xmath256-robustly @xmath249-universal _ if whenever * @xmath257 are sets with @xmath258 for all @xmath259 $ ] , such that writing @xmath260 } v_j$ ] and @xmath261 $ ] we have @xmath262 for any @xmath259 $ ] and @xmath263 , * @xmath251 is a @xmath0-partite @xmath0-complex of maximum vertex degree at most @xmath249 on some vertex set @xmath264 with @xmath265 for all @xmath259 $ ] , * @xmath266 satisfies @xmath267 for every @xmath259 $ ] , and sets @xmath268 satisfy @xmath269 for each @xmath270 , where for each @xmath159 we let @xmath271 be such that @xmath272 , then @xmath250 contains a copy of @xmath251 , in which for each @xmath259 $ ] the vertices of @xmath273 correspond to the vertices of @xmath274 , and @xmath159 corresponds to a vertex of @xmath275 for every @xmath270 .
so our use of the blow - up lemma will be hidden through this definition . of course
, we shall also need to obtain robustly universal complexes .
this is the purpose of the next theorem , which states that given a regular @xmath0-partite @xmath0-complex @xmath103 with sufficient density , and a @xmath0-partite @xmath0-graph @xmath244 on the same vertex set which is small relative to @xmath103 , we can delete a small number of vertices from their common vertex set so that @xmath246 is robustly universal .
it is a special case of theorem 6.32 in @xcite .
[ robust - universal ] suppose that @xmath276 , @xmath103 is a @xmath0-partite @xmath0-complex on @xmath277 with @xmath278 for every @xmath259 $ ] , @xmath103 is @xmath167-regular with @xmath243}(g ) \ge d$ ] and @xmath279 } ) \ge d_a$ ] , and @xmath280 with @xmath281
. then we can delete at most @xmath282 vertices from each @xmath274 to obtain @xmath283 , @xmath284 $ ] and @xmath285 $ ] such that * @xmath286 and @xmath287 for every @xmath288 , and * @xmath289 is @xmath256-robustly @xmath249-universal .
in this section we will collect the preliminary results we need to prove theorem [ main ] . in order to apply theorem [ robust - universal ]
, we need to know under what conditions we can find particular loose paths in complete @xmath0-partite @xmath0-graphs , which is the topic of the next subsection .
the problem of when we can find particular loose paths in a complete @xmath0-partite @xmath0-graph can be reformulated in terms of the question of which strings satisfying certain adjacency conditions can be produced from a fixed character set ; the following lemma is the result we will need .
[ strings ] let @xmath26 and @xmath290 be integers such that @xmath291 for all @xmath72 , and @xmath292 .
then for any @xmath293 $ ] there exists a string of length @xmath26 on alphabet @xmath294 such that the following properties hold : 1 .
no two consecutive characters are equal , 2 .
the first character is not @xmath295 and the final character is not @xmath296 , 3 .
the number of occurrences of character @xmath297 is @xmath298 .
* proof . *
note that the conditions on @xmath26 and the @xmath298 imply that @xmath299 .
we will construct the required string by starting with an ` empty string ' of @xmath26 blank positions , and for each @xmath72 inserting precisely @xmath298 copies of character @xmath297 .
this ensures that condition ( 3 ) will be satisfied .
we shall fill the empty positions in the following order : first the first position , then the third , and so on through the odd - numbered positions , until we reach either position @xmath26 or position @xmath300 ( dependent on whether @xmath26 is odd or even ) .
we then fill the second position , then the fourth , and so on until all positions are filled .
note that if we proceed by inserting all copies of one character , then all the copies of another character , and so forth , then condition ( 1 ) must be satisfied .
this is because to get two consecutive copies of @xmath297 , we must have inserted a copy of @xmath297 at some odd position @xmath301 , then @xmath302 , @xmath303 , and so on until reaching @xmath26 or @xmath300 , and then filled even positions @xmath304 . however
, this would imply that we had inserted at least @xmath305 copies of character @xmath297 , contradicting the fact that @xmath306 .
we therefore only need to determine an order to insert the different characters so as to satisfy ( 2 ) .
we first consider the case @xmath307 , say @xmath308 and @xmath309 . in this case
we insert @xmath310 first , @xmath311 last , and the remaining character blocks in any order in between .
clearly this prevents the first character from being @xmath311 and the last from being @xmath310 , and so ( 2 ) is satisfied .
now we may assume @xmath312 , say @xmath313 . then if @xmath26 is odd , we insert the characters in the following order : @xmath314
. then all the copies of @xmath311 must be in even positions ( since @xmath315 ) , and so ( 2 ) is satisfied .
alternatively , if @xmath26 is even , we insert first @xmath297 for some @xmath316 with @xmath317 , then @xmath311 , and then the remaining blocks of characters in any order .
( note that these include at least one character other than @xmath311 and @xmath297 since @xmath318 and @xmath319 imply that at least three @xmath145 have @xmath320 . )
so neither the first nor last character can be @xmath311 , and so ( 2 ) is again satisfied .
the next lemma is the result we were aiming for in this section , giving information about which loose paths can be found in complete @xmath0-partite @xmath0-graphs .
note that the maximum vertex degree of a loose path is two , and so this lemma will tell us when we can find a loose path in a robustly universal @xmath0-complex .
[ loosepath ] let @xmath103 be a complete @xmath0-partite @xmath0-graph on the vertex set @xmath321 .
let @xmath322 be integers with @xmath323 for each @xmath72 .
suppose that * @xmath324 is an integer , and * @xmath325 for all @xmath72 .
then for any @xmath293 $ ] , there exists a loose path in @xmath103 with an initial vertex in @xmath326 , a final vertex in @xmath327 , and containing @xmath328 vertices from @xmath329 for each @xmath207 $ ] .
* proof . * note first that @xmath1 is the number of edges such a path must contain .
let @xmath330 for each @xmath72 , so that @xmath331 .
by lemma [ strings ] we can find a string @xmath132 of length @xmath332 on the alphabet @xmath333 such that @xmath329 appears @xmath298 times , no two consecutive characters are identical , the first character is not @xmath326 and the final character is not @xmath327 .
let @xmath334 be the @xmath72th character of @xmath132 . to construct a loose path @xmath44 in @xmath103 , first choose any vertex from @xmath326 to be the initial vertex of @xmath44 , and any vertex from @xmath327 to be the final vertex of @xmath44 .
we also use @xmath132 to choose the link vertices of @xmath44 : choose the @xmath72th link vertex ( i.e. the vertex lying in the intersection of the @xmath72th and @xmath335th edges of @xmath44 ) to be any member of @xmath334 not yet chosen .
we have now assigned two vertices to each edge of @xmath44 .
finally , we complete @xmath44 by assigning to each edge one as yet unchosen vertex from each of the @xmath61 classes not yet represented in that edge .
this is possible since precisely @xmath298 link vertices are from the class @xmath329 and so the total number of vertices used from @xmath329 is @xmath336 .
since @xmath103 is complete we know that each edge of @xmath44 is an edge of @xmath103 , and so @xmath44 is a loose path satisfying all the conditions of the lemma .
a _ walk @xmath55 _ in a hypergraph @xmath4 consists of a sequence of edges @xmath337 of @xmath4 and a sequence @xmath338 of ( not necessarily distinct ) vertices of @xmath4 , satisfying @xmath339 for all @xmath340 $ ] , and also @xmath341 , @xmath342 and @xmath343 for all @xmath344 $ ] .
the _ length of @xmath55 _ is the number of its edges .
we say that @xmath345 is the _ initial vertex of @xmath55 _ , @xmath346 is the _ final vertex of @xmath55 _ , and that @xmath347 are the _ link vertices of @xmath55_. by a _ walk from @xmath348 to @xmath349 _ we mean a walk with initial vertex @xmath348 and final vertex @xmath349 .
note that the vertices of a hypergraph @xmath4 can be partitioned using the equivalence relation @xmath350 , where @xmath351 if and only if either @xmath352 or there exists a walk from @xmath348 to @xmath349 .
we call the equivalence classes of this relation _ components _ of @xmath4 .
we say that @xmath4 is _ connected _ if it has precisely one component .
observe that all vertices of an edge of @xmath4 must lie in the same component .
finally , note that if @xmath4 is a connected hypergraph of order @xmath1 , then for any two vertices @xmath353 of @xmath4 we can find in a walk from @xmath348 to @xmath349 of length at most @xmath1 in @xmath4 .
in this section we shall obtain , with high probability , a lower bound on the density of a subgraph of a @xmath0-partite @xmath0-graph chosen uniformly at random .
we will use azuma s inequality on the deviation of a martingale from its mean .
[ azuma ] suppose @xmath354 is a martingale , i.e. a sequence of random variables satisfying @xmath355 , and that @xmath356 for some constants @xmath357 and all @xmath358 $ ] .
then for any @xmath359 , @xmath360 [ randomsplit ] suppose @xmath361 , and that @xmath4 is a @xmath0-partite @xmath0-graph on the vertex set @xmath362 , where @xmath363 for each @xmath207 $ ] .
suppose also that @xmath4 has density @xmath364 and that for each @xmath72 we have @xmath365 .
if we choose a subset @xmath366 with @xmath367 uniformly at random and independently for each @xmath72 , and let @xmath368 , then the probability that @xmath369 $ ] has density @xmath370 ) > c/2 $ ] is at least @xmath371 .
moreover , the same holds if we choose @xmath372 by including each vertex of @xmath113 independently with probability @xmath373 .
* let @xmath374 . to prove the first assertion
, we obtain our subsets @xmath366 through the following two - stage random process , independently for each @xmath72 .
first we assign the vertices of each @xmath113 into sets @xmath375 and @xmath376 independently at random , with each vertex being assigned to @xmath375 with probability @xmath373 , and assigned to @xmath376 otherwise .
then , in the ( highly probable ) event that we have @xmath377 we shall select uniformly at random a set of vertices to transfer between @xmath375 and @xmath376 to obtain from @xmath375 the set @xmath372 with @xmath367 .
for each @xmath72 , no subset @xmath366 of size @xmath378 is more likely to result from this process than any other , so we have chosen each @xmath372 uniformly at random .
it remains to show that @xmath369 $ ] is likely to have high density .
we do this by noting that @xmath379 $ ] is likely to have high density ( where @xmath380 ) and that with high probability we will only need to transfer a small number of vertices to form @xmath381 , which can have only a limited effect on the density .
more precisely , let @xmath382 be an ordering of the vertices of @xmath110 , and for each @xmath358 $ ] let the random variable @xmath383 take the value 1 if @xmath384 , and 0 otherwise . recall that we write @xmath102 to denote the number of edges of a @xmath0-graph @xmath4 . for all @xmath385
we now define random variables @xmath386 by @xmath387| \mid y_1 , \dots , y_i)$ ] .
then the sequence @xmath388 is a martingale , @xmath389|$ ] , and as we formed each @xmath375 by assigning vertices of @xmath113 independently at random into @xmath375 and @xmath376 , we have @xmath390| ) \ge c \prod_{i=1}^k t_i$ ] . also ,
for any vertex @xmath297 , let @xmath391 be such that @xmath392 ( i.e. @xmath391 is the index of @xmath297 ) .
then @xmath393 for all @xmath358 $ ] .
thus we can apply lemma [ azuma ] to obtain @xmath394 therefore the event that @xmath395 ) > 3c/4 $ ] has probability at least @xmath396 . also , by a standard chernoff bound , for each @xmath207 $ ] the event that @xmath397 has probability at least @xmath396 .
thus with probability at least @xmath371 all of these events will happen .
now , if @xmath398 , we choose a set of @xmath399 vertices of @xmath375 uniformly at random and move these vertices from @xmath375 to @xmath376 .
similarly , if @xmath400 , then we choose a set of @xmath401 vertices of @xmath376 uniformly at random and move these vertices to @xmath375 . in either case , for any @xmath72 this action can decrease @xmath395)$ ] by at most @xmath402 .
thus if we let @xmath55 be the set obtained from @xmath403 in this way , we have @xmath370 ) > c/2 $ ] , proving the first part of the lemma .
the proof of the ` moreover part ' is the same except that we can omit the ` transfer ' step at the end of the proof .
let @xmath404 denote the @xmath0-graph whose vertex set @xmath405 is the union of @xmath40 disjoint sets @xmath406 of size @xmath8 and whose edges consist of all @xmath0-tuples of the form @xmath407 , with @xmath408 and @xmath409 ( see figure 2 ) .
[ fig : a_kdiag ] -graph @xmath410 ( only edges involving @xmath411 are shown),title="fig : " ] so @xmath412 .
an _ @xmath404-packing _ in a @xmath0-graph @xmath103 is a collection of pairwise vertex - disjoint copies of @xmath404 in @xmath103 .
[ ak - pack - lem ] suppose @xmath413 , and that @xmath103 is a @xmath0-graph on @xmath414 $ ] such that @xmath415 for all but at most @xmath416 sets @xmath417}{k-1}$ ]
. then @xmath103 has an @xmath404-packing which covers more than @xmath418 vertices of @xmath103 .
* let @xmath419 be an @xmath404-packing of @xmath103 of maximum size , so @xmath420 .
let @xmath110 be the set of uncovered vertices , and suppose that @xmath421 .
let @xmath422 .
our first aim is to choose disjoint sets @xmath423 in @xmath424 so that @xmath425 and @xmath426 for all @xmath427 $ ] .
note that @xmath428 implies that @xmath429 .
so we can greedily choose disjoint @xmath430 such that @xmath431 for all @xmath432 $ ] .
let @xmath433 : |n_g(s_i ) \cap x| \geq \theta m/2\}$ ] .
we claim that @xmath434 .
otherwise , consider the bipartite graph @xmath165 with vertex classes @xmath11 and @xmath110 , where we join @xmath435 to @xmath436 if @xmath437 is an edge of @xmath103 .
note that @xmath165 can not contain a complete bipartite graph with @xmath438 vertices in @xmath11 and @xmath8 vertices in @xmath110 , as this would correspond to a copy of @xmath404 contained in @xmath110 , which is impossible as @xmath419 is a maximum size @xmath404-packing .
however , by definition of @xmath11 we have @xmath439 for every @xmath435 , and double - counting pairs @xmath440 with @xmath435 and @xmath441 gives @xmath442 a contradiction .
this proves the claim , and by relabelling the @xmath334 we can assume that @xmath431 and @xmath426 for all @xmath427 $ ] .
now we show how to enlarge the @xmath404-packing @xmath443 . for @xmath427 $ ] let @xmath444 : |n_g(s_i ) \cap v(a_j)| \ge k \}.\ ] ] since @xmath445 for each @xmath446 $ ] we have @xmath447 and so @xmath448 .
we now double - count pairs @xmath449 with @xmath427 $ ] and @xmath450 .
the number of such pairs is @xmath451 so we can find some @xmath452}{k-1}$ ] and @xmath453 $ ] with @xmath454 such that @xmath455 for every @xmath456 . for each @xmath456 and each @xmath457
fix some @xmath0-set @xmath458 ( which is possible by definition of @xmath459 )
. then we can choose @xmath460 with @xmath461 so that @xmath462 for all @xmath463 and every @xmath457 .
for each @xmath457 we write @xmath464 for @xmath465 with @xmath466 .
we will now use the @xmath464 to find @xmath0 new copies of @xmath404 that only intersect @xmath8 of the copies in our packing .
we arbitrarily divide @xmath467 into @xmath0 sets @xmath468 of size @xmath438 and label @xmath469 for all @xmath457 .
the new copies @xmath470 of @xmath404 are obtained for each @xmath255 $ ] by identifying @xmath471 with @xmath472 and @xmath473 with @xmath474 . replacing the copies @xmath475 by @xmath470
we obtain a larger @xmath404-packing .
this contradiction completes the proof .
[ ak - pack ] lemma [ ak - pack - lem ] still holds if we insist that the sub-@xmath0-graph of @xmath103 induced by the vertices covered by the @xmath404-packing must be connected . * proof . *
apply lemma [ ak - pack - lem ] to obtain an @xmath404-packing @xmath476 in @xmath103 with @xmath477 , and let @xmath92 be the sub-@xmath0-graph of @xmath103 induced by @xmath478 . by hypothesis at most @xmath416 sets @xmath417}{k-1}$ ]
have fewer than @xmath479 neighbours in @xmath103 and
so at most @xmath416 sets @xmath480 have no neighbours in @xmath481 . by the definition of a component ,
no edges of @xmath92 contain vertices from different components of @xmath92 .
therefore the largest component @xmath14 of @xmath92 must contain at least @xmath418 vertices .
indeed , if not then there are at least @xmath482 sets @xmath480 which meet at least two components of @xmath92 and thus have no neighbours in @xmath92 , a contradiction ( we can obtain such a set @xmath11 by choosing @xmath61 vertices arbitrarily in @xmath481 and then choosing the final vertex in a different component of @xmath92 than the first vertex ) .
thus we may take the @xmath404-packing consisting of all those copies @xmath483 of @xmath404 with @xmath484 .
in our proof we will use constants that satisfy the hierarchy @xmath485 furthermore , for any of these constants @xmath486 , we use @xmath487 and assume that the above hierarchy also extends to the additional constants , e.g. @xmath488 .
let @xmath489 be the sub-@xmath0-graph obtained from @xmath4 by removing up to @xmath227 vertices so that @xmath490 is divisible by @xmath227 .
let @xmath491 be an equitable @xmath91-partition of the vertices of @xmath489 , and let @xmath492 consist of all those edges of @xmath489 that are @xmath91-partite sets in @xmath11 .
then @xmath492 is an @xmath91-partite @xmath0-graph with order divisible by @xmath227 , and so we may apply the regularity lemma ( theorem [ eq - partition ] ) , which yields an @xmath209-bounded @xmath167-regular vertex - equitable partition @xmath118-complex @xmath44 on @xmath11 and an @xmath91-partite @xmath0-graph @xmath103 on @xmath11 that is @xmath222-close to @xmath492 and perfectly @xmath167-regular with respect to @xmath44 .
let @xmath493 .
so any edge of @xmath494 is also an edge of @xmath4 .
let @xmath495 be the clusters of @xmath44 .
so @xmath496 and @xmath103 is @xmath497-partite with vertex classes @xmath498 .
note that @xmath499 since @xmath44 is @xmath209-bounded .
moreover , since @xmath44 is vertex - equitable , each @xmath329 has the same size .
so let @xmath500 . as is usual in regularity arguments
, we shall consider a reduced @xmath0-graph , whose vertices correspond to the clusters @xmath329 , and whose edges indicate that within the cells of @xmath44 corresponding to the edge we can find a subcomplex to which we can apply theorem [ robust - universal ] . for this we would like @xmath103 to have high density in these cells , and @xmath244 to have low density .
thus we define the _ reduced @xmath0-graph @xmath50 _ on @xmath414 $ ] as follows : a @xmath0-tuple @xmath132 of vertices of @xmath50 corresponds to the @xmath0-partite union @xmath501 of clusters .
the edges of @xmath50 are precisely those @xmath502}{k}$ ] for which @xmath503 $ ] has density at least @xmath504 ( i.e. @xmath505| > c '' |k_s(s')|$ ] ) and for which @xmath506 $ ] has density at most @xmath507 ( i.e. @xmath508| < \nu^{1/2}|k_s(s')|$ ] ) .
now , the edges in the reduced graph are useful in the following way
. given an edge @xmath509 , let @xmath501 again . using weak equivalence ( defined in section [ regcomplexes ] ) ,
the cells of @xmath44 induce a partition @xmath510 of the edges of @xmath511 .
recall that @xmath512 .
therefore at most @xmath513 edges of @xmath511 can lie in sets @xmath514 with @xmath515 .
furthermore , @xmath508|
< \nu^{1/2 } |k_s(s')|$ ] ( as @xmath509 ) and so at most @xmath516 edges of @xmath511 can lie in sets @xmath514 with @xmath517 .
together with the fact that @xmath505| > c '' |k_s(s')|$ ] this now implies that more than @xmath518 edges of @xmath503 $ ] lie in sets @xmath514 with @xmath519 and @xmath520 .
thus there must exist such a set @xmath514 that also satisfies @xmath521 .
fix such a choice of @xmath514 and denote it by @xmath522 .
let @xmath523 be the @xmath0-partite @xmath0-complex on the vertex set @xmath524 consisting of @xmath525 and the cells of @xmath44 that ` underlie ' @xmath522 , i.e. for any edge @xmath526 we have @xmath527 ( recall that @xmath528 was defined in section [ regcomplexes ] . )
we also define the @xmath0-partite @xmath0-graph @xmath529 on the vertex set @xmath524 .
then the following properties hold : * @xmath523 is @xmath167-regular . * @xmath523 has @xmath0-th level relative density @xmath243}(g^s ) \geq d'$ ] .
* @xmath523 has absolute density @xmath530 .
* @xmath531 satisfies @xmath532 .
* @xmath533 for any @xmath534
. indeed , ( a1 ) follows from ( [ eqgs ] ) since @xmath103 is perfectly @xmath167-regular with respect to @xmath44 . to see ( a2 ) , note that @xmath535})^*=c^s$ ] and so @xmath243}(g^s ) = |g^s_{[k]}|/|(g^s_{[k]})^*|=|g^s \cap c^s|/|c^s| > c''/2 $ ] by our choice of @xmath522 .
similarly , ( a3 ) follows from our choice of @xmath522 since @xmath536}|}{|k_s(s')| } = \frac{|g^s \cap c^s|}{|c^s| } \cdot \frac{|c^s|}{|k_s(s')| } > \frac{(c'')^2}{6a^k}>d'_a.\ ] ] ( a4 ) holds since @xmath537 and @xmath538 .
finally , ( a5 ) follows from ( [ eqgs ] ) and the fact that @xmath539 for all @xmath540 .
the next step in our proof is to use corollary [ ak - pack ] to find an @xmath404-packing in the reduced @xmath0-graph @xmath50 . for this
we shall need an approximate minimum degree condition for @xmath50 .
let @xmath541}{k-1 } : |n_r(i)| \le \left(\frac{1}{2(k-1 ) } + \phi \right)m \right\}.\ ] ] we shall show that @xmath250 is small , that is , that almost all @xmath118-tuples of vertices of @xmath50 have degree at least @xmath542 in @xmath50 . consider how many edges of @xmath4 do not belong to @xmath503 $ ] for some edge @xmath543 .
( recall that @xmath544 . )
there are three possible reasons why an edge @xmath545 does not belong to such a restriction : * @xmath7 is not an edge of @xmath103 .
this could be because @xmath7 lies in @xmath4 but not @xmath489 , in @xmath489 but not @xmath492 , or in @xmath492 but not @xmath103 .
there are at most @xmath546 edges of the first type , at most @xmath547 of the second type , and at most @xmath548 of the third type .
* @xmath549 contains vertices from @xmath550 such that the restriction of @xmath244 to @xmath544 satisfies @xmath508| \geq \nu^{1/2}|k_s[s']|$ ] , where @xmath551 .
( note that since @xmath103 and thus @xmath244 is @xmath497-partite , @xmath552 are all distinct . ) since @xmath103 and @xmath492 are @xmath222-close and thus @xmath553 there are at most @xmath554 edges of this type .
* @xmath549 contains vertices from @xmath550 such that the restriction of @xmath103 to @xmath555 has density less than @xmath504 .
there are at most @xmath556 edges of this type .
therefore there are fewer than @xmath557 edges of @xmath4 that do not belong to the restriction of @xmath103 to @xmath524 for some @xmath543 , and so we have @xmath558 since @xmath559 we deduce that @xmath560 , and so @xmath561 ( since @xmath562 ) .
this allows us to apply corollary [ ak - pack ] ( with @xmath563 ) to obtain an @xmath404-packing @xmath419 in @xmath50 with @xmath564 , such that the sub-@xmath0-graph of @xmath50 induced by @xmath565 is connected . for each @xmath566 $ ] ,
let the vertex set of @xmath483 be @xmath567 , with each @xmath568 of size @xmath8 , so that the edge set is @xmath569 , x \in u^i_0\}$ ] . given a sub-@xmath0-graph @xmath467 of @xmath50 and a cluster @xmath329 , we say that _
@xmath329 belongs to @xmath467 _ if @xmath570 .
let @xmath571 contain the at most @xmath227 vertices of @xmath4 we removed at the start of the proof , and also the vertices in all those clusters not belonging to some copy of @xmath404 in our packing ( there are at most @xmath572 of the latter ) .
we will incorporate these vertices into a path @xmath48 which will later form part of our loose hamilton cycle .
we also include in @xmath571 an arbitrary choice of @xmath573 vertices from each @xmath574 for which @xmath575 for some @xmath576 $ ] and some @xmath566 $ ] ( we do not modify any of the @xmath574 for which @xmath577 ) .
we add up to @xmath578 more vertices from @xmath579 ( say ) to @xmath571 so that @xmath580 .
we delete all these vertices from the clusters they belonged to and still write @xmath574 for the subcluster of a cluster @xmath574 obtained in this way .
this gives @xmath581 .
now , we shall construct a path @xmath48 in @xmath4 , which will contain all the vertices in @xmath582 and avoid all the clusters @xmath574 with @xmath577 .
let @xmath583 , i\in [ t]\}$ ] .
so we shall use only vertices from @xmath571 and @xmath584 in forming @xmath48 . recall that if @xmath49 is not a multiple of @xmath8 , then a loose hamilton cycle contains a single pair of edges which intersect in more than one vertex
: we shall make allowance for this here .
choose @xmath585 satisfying @xmath586 , @xmath587 and @xmath588 .
now choose distinct @xmath589 such that @xmath590 and @xmath591 ( we shall see in a moment that such @xmath592 exist ) .
these edges will be the first 2 edges of @xmath48 . to complete @xmath48 ,
let @xmath593 be any partition of the vertices of @xmath571 into sets of size @xmath61 .
we proceed greedily in forming @xmath48 : for each @xmath594 choose any @xmath595 such that @xmath596 ( where the @xmath297 are all chosen to be distinct ) .
let us now check that there will always be such a vertex available .
indeed , every set in @xmath597 has at least @xmath598 neighbours and we can choose any such neighbour which lies in @xmath584 and has not already been used .
but @xmath599 and at most @xmath600 vertices have been used before .
thus ( since @xmath601 ) for each choice of an @xmath297 we have at least @xmath602 vertices of @xmath584 to choose from .
moreover , these vertices must be contained in at least @xmath603 different @xmath574 such that @xmath604 ( @xmath605 ) .
thus we can avoid choosing a vertex from any single @xmath574 more than @xmath606 times .
the path @xmath48 thus formed has edges @xmath607 , @xmath608 and @xmath609 for all @xmath610 $ ] .
so all the vertices of @xmath571 are included in @xmath48 .
for each cluster @xmath574 , we still denote the subset of @xmath574 lying in @xmath611 by @xmath574 .
then each @xmath574 with @xmath577 for some @xmath72 still satisfies @xmath612 , and each @xmath574 with @xmath575 for some @xmath605 satisfies @xmath613 in addition latexmath:[\[\label{exceppathsize } the next step of the proof will be to split the copies @xmath443 of @xmath404 ( more precisely the clusters belonging to the @xmath483 ) into sub-@xmath0-complexes of @xmath103 that we shall later use to embed spanning loose paths . consider any @xmath483 . for convenient notation
we identify each @xmath568 in @xmath483 with @xmath616 $ ] ( but recall that they are disjoint sets ) .
for each @xmath617 $ ] we have @xmath612 , and so we can partition @xmath574 uniformly at random into @xmath438 pairwise disjoint subsets @xmath618 , each of size @xmath619 .
similarly , given @xmath620 $ ] with @xmath576 $ ] , ( [ vy ] ) and the fact that @xmath621 imply that we can partition @xmath622 uniformly at random into @xmath8 pairwise disjoint subsets @xmath623 and @xmath624 { \setminus}\{z\}}$ ] so that @xmath625 and @xmath626 for all @xmath627 { \setminus}\{z\}$ ] .
figure 3 shows how we do this in the case @xmath24 .
we arrange these pieces into @xmath628 collections of @xmath0 sets as follows : for each @xmath629 and each @xmath630 $ ] we have a collection consisting of @xmath631 , @xmath632 and @xmath633 .
( @xmath197 of these collections are illustrated in figure 3 . ) for convenient notation we relabel these collections as @xmath634 with @xmath635 , where for all @xmath636 $ ] we have @xmath637 and @xmath638 ( ( [ splitsizes2 ] ) follows from ( [ vy ] ) using the fact that all the @xmath639 have equal size . )
let @xmath640 } x_{i , j}$ ] , so each @xmath113 is a @xmath0-partite set , on which we shall now find a sub-@xmath0-complex @xmath641 of @xmath103 that is suitable for applying theorem [ robust - universal ] .
consider any copy @xmath642 in our @xmath404-packing .
note that for each of the @xmath628 collections @xmath643 obtained by splitting up the clusters belonging to @xmath642 there is an edge @xmath644 such that each @xmath645 lies in a cluster belonging to @xmath646 ( and these clusters are distinct for each of @xmath647 ) .
recall that @xmath648 denotes the union @xmath649 of all the clusters belonging to @xmath646 .
let @xmath641 denote the restriction of the @xmath0-partite @xmath0-complex @xmath650 ( which was defined in section [ regsection ] ) to @xmath113 , i.e. @xmath651 $ ] .
let @xmath652 $ ] .
we claim that we may choose the above collections @xmath634 such that @xmath653)\ge \frac{c''}{4 } \ \
\text{for all } i\in [ t'].\ ] ] indeed , since @xmath654 , @xmath655 $ ] has absolute density at least @xmath504 and @xmath656 $ ] has density at most @xmath507 .
since @xmath657 and @xmath658 this shows that @xmath659 $ ] has density at least @xmath660 .
lemma [ randomsplit ] now implies that each @xmath661 $ ] has density at least @xmath662 with probability @xmath663 , and so with non - zero probability this is true for all @xmath664 $ ] .
lemma [ restrict ] and properties ( a1)(a3 ) and ( a5 ) imply that @xmath641 is an @xmath241-regular @xmath0-partite @xmath0-complex on the vertex set @xmath113 , with absolute density @xmath665 , relative density @xmath243}(g_i ) \geq d$ ] , and @xmath666 for each @xmath145 .
moreover , using @xmath667 , property ( a4 ) and the fact that @xmath668 we see that @xmath669 so by theorem [ robust - universal ] we can delete at most @xmath670 vertices from each @xmath645 so that if we let @xmath671 consist of the undeleted vertices , and let @xmath672 , @xmath673 $ ] and @xmath674 $ ] , then @xmath675 is @xmath676-robustly @xmath677-universal , @xmath678 and @xmath679 for every @xmath680 . in particular , the latter two conditions together imply that @xmath681 for every @xmath682 .
let @xmath683 denote the set of vertices deleted from any @xmath645 , so @xmath684 .
by deleting up to @xmath578 more vertices if necessary , we may assume that @xmath685 is divisible by @xmath61 .
the latter will help us to extend @xmath48 into a path which contains all the vertices in @xmath683 .
when extending @xmath48 in order to incorporate @xmath683 , we shall have to remove some more vertices from some of the @xmath686 , and we wish to do this so that the remainder satisfies ( i ) in the definition of robust universality .
for this reason , we partition each @xmath686 into two parts @xmath687 and @xmath688 as follows ( where we write @xmath689 for @xmath690 } bx'_{i , j}$ ] ) : ( recall that for a @xmath118-complex @xmath695 , @xmath696 denotes the ` @xmath118th level ' of @xmath695 . ) to see that such a partition exists , consider a partition obtained by assigning each vertex to a part with probability @xmath697 independently of all other vertices .
( b2 ) is then satisfied with high probability by a standard chernoff bound .
now consider ( b1 ) .
the ` moreover ' part of lemma [ randomsplit ] implies that with high probability we have for all @xmath698 and for all @xmath680 that @xmath699)_= ) \ge d(g_i'(v)_=)/2 $ ] .
also , a standard chernoff bound implies that with high probability @xmath700 for all @xmath701 $ ] .
thus @xmath702)_= |=d((g_i'(v ) [ bx'_i])_=)\prod_{j'\neq j}|bx'_{i , j'}|\ge \frac{d(g_i'(v)_=)}{2}\prod_{j'\neq j}\frac{|x'_{i , j'}|}{3}\ge 2c|g_i'(v)_=|.\ ] ] now , we shall extend our path @xmath48 to include the vertices in @xmath683 , using only vertices from @xmath694 .
we proceed similarly to when constructing @xmath48 .
so we split @xmath683 into sets @xmath703 of size @xmath61 ( so @xmath704 ) .
letting @xmath345 be a final vertex of @xmath48 , for @xmath705 $ ] , we successively choose @xmath297 to be a neighbour of the @xmath118-tuple @xmath706 contained in some @xmath707 and not already included in @xmath48 , and extend @xmath48 by the edge @xmath708 , continuing to denote the extended path by @xmath48 .
recall that @xmath48 originally contained at most @xmath709 vertices . since @xmath684 , after each extension of @xmath48 we shall have @xmath710 .
so ( b2 ) implies that for each choice of @xmath297 we have at least @xmath711 suitable vertices and hence at least @xmath712 of the sets @xmath713 contain such a suitable vertex .
this shows that we can choose the @xmath297 in such a way that at most @xmath714 vertices are chosen from any single @xmath713 . for each @xmath664 $ ] let @xmath715 be the vertices remaining after the removal from @xmath716 of the at most @xmath714 vertices used in extending @xmath48 , let @xmath717 $ ] , and let @xmath718 $ ] . by ( [ desityhg ] ) there are at least @xmath719 vertices @xmath156 such that @xmath154 lies in some @xmath47 for which at least @xmath720|/(2|x^i|)$ ] edges of @xmath721 $ ] contain @xmath154 .
so we may add two further edges of @xmath4 to @xmath48 ( one at each end ) so that the new path @xmath48 has an initial vertex @xmath722 and a final vertex @xmath723 which each lie in at least @xmath720|/(2|x^i|)$ ] edges of their respective @xmath721 $ ] .
( we also delete the vertices of these additional two edges from their @xmath47 , @xmath724 and @xmath725 ) .
note that @xmath722 may be contained in some @xmath688 ( and the same is true of @xmath723 ) , but by ( b2 ) we may choose these two additional edges so that all other vertices used lie in some @xmath726 .
we claim that the above steps give us the following useful structure : a path @xmath48 which is ready to form part of a loose hamilton cycle , and disjoint @xmath0-partite vertex sets @xmath715 supporting @xmath0-complexes @xmath724 and @xmath0-graphs @xmath725 for each @xmath636 $ ] which satisfy the following properties .
* every vertex of @xmath4 lies in either the path @xmath48 or precisely one of the @xmath0-partite sets @xmath47 . * for each @xmath72 , @xmath724 is a @xmath0-partite sub-@xmath0-complex of @xmath103 on the vertex set @xmath47 .
@xmath725 is the @xmath727-partite @xmath0-graph @xmath728 , and @xmath729 .
clearly these statements remain true after the deletion of up to @xmath730 vertices of @xmath47 . * even after the deletion of up to @xmath730 vertices of @xmath47 , the following statement holds .
let @xmath251 be a @xmath0-partite @xmath0-complex on the vertex set @xmath264 , where @xmath731 for each @xmath145 , and let @xmath251 have maximum vertex degree at most @xmath677 .
let @xmath732 and suppose we have @xmath733 and sets @xmath734 with @xmath735 for each @xmath736 $ ] ( where @xmath737 is such that @xmath738 ) . then @xmath739 contains a copy of @xmath251 , in which for each @xmath145 the vertices of @xmath273 correspond to the vertices of @xmath740 , and each @xmath741 corresponds to a vertex in @xmath742 . * for each @xmath72 , @xmath743 $ ] has density at least @xmath744 , even after the deletion of up to @xmath730 vertices of @xmath47 . *
if we delete up to @xmath730 vertices from any @xmath47 , and let @xmath745 for each @xmath746 $ ] after these deletions , and let @xmath747 , then @xmath748 for all @xmath145 . * the initial vertex @xmath722 of @xmath48 lies in at least @xmath720|/(2|x^i|)$ ] edges of @xmath721 $ ] , where @xmath72 is such that @xmath749 .
the analogue holds for the final vertex @xmath723 of @xmath48 .
( when we talk of removing a vertex of @xmath47 we implicitly mean that @xmath724 , @xmath725 and @xmath46 are all restricted to the remaining vertices of @xmath47 . )
these properties hold for the following reasons .
( c1 ) holds as every vertex deleted from an @xmath113 has been added to @xmath48 , whilst ( c2 ) is clear as whenever we deleted vertices we simply restricted @xmath103 and @xmath244 to the remaining vertices . for ( c3 ) , recall that @xmath675 was @xmath676-robustly @xmath677-universal . moreover , for all @xmath664 $ ] and all @xmath259 $ ] we have @xmath750 , since we ensured that we only deleted @xmath751 vertices from any single @xmath752 ( and at most two from @xmath689 ) . furthermore by ( b1 ) we know that @xmath753)_= | \ge c|g'_i(v)_=|$ ] for any @xmath754 .
( also , even if we had arbitrarily deleted a further @xmath730 vertices from @xmath755 when obtaining @xmath47 , @xmath724 and @xmath725 , these bounds would still hold . )
so @xmath739 satisfies ( i ) in the definition of a robustly universal complex ( where @xmath756 plays the role of @xmath274 ) .
the sets @xmath742 satisfy ( iii ) in the definition and so we can find the required copy of @xmath251 ( even after the deletion of up to @xmath730 more vertices of @xmath47 ) .
( c4 ) follows from ( [ desityhg ] ) and the fact that @xmath47 was formed by deleting at most @xmath757 vertices from @xmath113 .
similarly , for ( c5 ) note that ( even after up to @xmath730 more deletions ) we have deleted at most @xmath758 vertices from each @xmath113 since we split the clusters to form the @xmath113 .
so by ( [ splitsizes1 ] ) , after these deletions we must have roughly speaking , our aim is to find a spanning loose path in each @xmath767 ( and thus in @xmath46 ) such that all these paths together with @xmath48 form a loose hamilton cycle in @xmath4 .
so we have to ensure that the complete @xmath0-partite @xmath0-graph on @xmath47 contains a spanning loose path ( for this , we will need @xmath768 ) and we need to join up all the loose paths we find in the @xmath46 .
the purpose of this section is to find the ` connecting loose paths ' which join up the @xmath47 in such a way that the divisibility problems are dealt with as well . to do this ,
we first define a supplementary hypergraph @xmath769 whose vertices correspond to the @xmath47 .
we will show that @xmath769 is connected and that ` along ' edges of @xmath769 we can find our loose paths in @xmath4 which join up all the @xmath47 .
the vertex set of the _ supplementary hypergraph @xmath769 _ is @xmath770 $ ] .
a subset @xmath771 $ ] of size at least 2 is an edge of @xmath769 if there exists an edge @xmath772 such that for all @xmath773 there are @xmath774 and @xmath775 $ ] with @xmath776 and @xmath777 .
( we fix one such edge @xmath778 for every @xmath779 . )
then every edge of @xmath769 has size at most @xmath0 .
we say that @xmath47 _ belongs _ to an edge @xmath779 if @xmath780 .
similarly , @xmath47 _ belongs to _ some subhypergraph @xmath781 if @xmath570 .
* proof . *
recall that we chose the copies @xmath782 of @xmath404 in such a way that the sub-@xmath0-graph @xmath92 of @xmath50 induced by @xmath783 is connected .
suppose that @xmath769 is not connected .
let @xmath784 be a component of @xmath769 and let @xmath785 .
let @xmath786 : x^i_s \subseteq v_j$ ] for some @xmath787\}$ ] .
so @xmath788 corresponds to the set of all those clusters which meet some @xmath47 belonging to @xmath784 .
define @xmath789 similarly .
then @xmath790 and thus @xmath92 contains some edge @xmath132 intersecting both @xmath788 and @xmath789 .
but then @xmath132 corresponds to an edge of @xmath769 intersecting both @xmath791 and @xmath792 , a contradiction .
the next lemma shows that within the @xmath47 belonging to an edge of @xmath769 , we can find a reasonably short loose path in @xmath4 and we may choose ( modulo @xmath8 ) how many vertices this path uses from each @xmath47 .
using the connectedness of @xmath769 , this will allow us to find the connecting loose paths which join up the @xmath47 whilst having control over the divisibility properties .
we shall also insist that the path in lemma [ interpath ] avoids a number of ` forbidden vertices ' , to enable us to ensure that our connecting loose paths are disjoint , and that the endvertices of these paths lie in many edges of the relevant @xmath46 .
[ interpath ] suppose that @xmath779 and that for every @xmath780 there is an integer @xmath378 such that @xmath793 and @xmath794 .
let @xmath795 be distinct .
moreover , suppose that @xmath796 is a set of at most @xmath797 ` forbidden ' vertices of @xmath4 .
then in the sub-@xmath0-graph of @xmath4 induced by @xmath798 we can find a loose path @xmath251 with the following properties .
* @xmath251 contains at most @xmath799 vertices .
* @xmath251 has an initial vertex @xmath159 in @xmath800 and a final vertex @xmath154 in @xmath801 .
* @xmath802 for each @xmath780 . *
@xmath251 contains no forbidden vertices , i.e. @xmath803 .
* @xmath159 lies in at least @xmath804 edges of @xmath805 , and @xmath154 lies in at least @xmath806 edges of @xmath807 .
* recall that in section [ regsection ] we assigned a @xmath0-partite @xmath0-complex @xmath523 to every edge @xmath509 such that ( a1)(a5 ) are satisfied . to simplify notation ,
we write @xmath132 for the edge @xmath772 corresponding to @xmath7 and suppose that @xmath808 $ ] . for
each @xmath809 $ ] choose @xmath774 and @xmath775 $ ] such that @xmath776 and such that @xmath810\}$ ] . to simplify notation we write @xmath811 for @xmath812 , @xmath813}y_j$ ] and
assume that @xmath814 and @xmath815 .
for each @xmath780 let @xmath816 be the set of all @xmath809 $ ] with @xmath817 .
so the sets @xmath816 are disjoint and their union is @xmath818 $ ] . pick some @xmath819 and let @xmath820 and @xmath821 for all @xmath822 .
our path @xmath251 will consist of @xmath823 vertices from each @xmath811 ( modulo @xmath8 ) and thus of @xmath378 vertices from each @xmath47 ( modulo @xmath8 ) .
since @xmath523 satisfies ( a1)(a3 ) and ( a5 ) , lemma [ restrict ] implies that the restriction @xmath824 $ ] is @xmath241-regular , with absolute density at least @xmath825 , relative density at index @xmath818 $ ] at least @xmath826 and @xmath827=y_j$ ] .
furthermore , ( a4 ) together with the fact that @xmath828)\ge d(g^s)/2 $ ] imply that @xmath829| < |m^s| < \frac{2\nu^{1/4}|g^s|}{c''}\le \theta |g^s[y]|.\ ] ] thus theorem [ robust - universal ] implies that we can delete @xmath830 vertices from each @xmath811 to obtain subsets @xmath831 such that @xmath832 { \setminus}m^s[y']$ ] is @xmath676-robustly @xmath677-universal , where @xmath833 } y'_j$ ] .
now , let @xmath834 .
then @xmath835 and so @xmath836 is an integer .
furthermore , @xmath837 , and so @xmath838 for each @xmath145 .
thus by lemma [ loosepath ] we can find a loose path in the complete @xmath0-partite @xmath0-graph on the vertex set @xmath839 , beginning in @xmath840 , finishing in @xmath841 and using @xmath842 vertices from each @xmath843 .
since @xmath832 { \setminus}m^s[y']$ ] is @xmath844-robustly @xmath677-universal , we can find such a loose path @xmath251 in @xmath832{\setminus}m$ ] and hence in @xmath845 .
( indeed , we can do this by finding the complex @xmath846 , which has maximum vertex degree at most @xmath677 .
note that we use the definition with @xmath847 in ( i ) ) .
note that @xmath251 contains at most @xmath848 vertices . to see that we can insist on the final condition of the lemma , recall that @xmath849 by ( c4 ) .
thus for all @xmath746 $ ] at least @xmath850 vertices of @xmath740 lie in at least @xmath851 edges of @xmath46 , and so we may restrict the initial and final vertices of @xmath251 to these sets of vertices ( minus the vertices of @xmath796 ) by ( iii ) in the definition of robust universality . as discussed before , our hamilton cycle in @xmath4 will consist of @xmath48 and paths in each @xmath46 as well as paths connecting the @xmath47 . however , we need to make sure that all these paths join up nicely , motivating the following definition .
suppose @xmath251 is a path in some @xmath0-graph @xmath852 with initial vertex @xmath853 and final vertex @xmath854 .
also , let @xmath855 be disjoint sets of size @xmath61 .
then @xmath856 is a _
prepath_. note that @xmath857 is not ( the vertex set of ) a @xmath0-graph , but that if we can find vertices @xmath858 such that @xmath859 , then adding @xmath348 and @xmath349 to @xmath857 gives another path .
we refer to all such vertices @xmath860 as _ possible initial vertices _ of @xmath857 and to all such vertices @xmath861 as _ possible final vertices_. if @xmath251 , @xmath862 and @xmath863 are disjoint loose paths , @xmath864 are as before , @xmath348 is also the final vertex of @xmath862 and @xmath349 is also the initial vertex of @xmath863 then @xmath865 and @xmath695 together with @xmath866 form a single loose path , illustrating how we shall join paths together .
we start by converting our exceptional path @xmath48 into a prepath .
recall that @xmath867 and that the initial vertex @xmath722 of @xmath48 and its final vertex @xmath723 satisfy ( c6 ) .
let @xmath868 $ ] and @xmath869 $ ] be such that @xmath870 .
pick any @xmath871 $ ] with @xmath872 .
( c4 ) and ( c6 ) together imply that there is a set @xmath873 for which @xmath874 contains at least @xmath875 vertices @xmath154 which form an edge of @xmath876 together with @xmath877 .
let @xmath878 be such a set of vertices .
similarly , letting @xmath879 $ ] , @xmath880 $ ] be such that @xmath881 , there is a set @xmath882 for which @xmath883 contains at least @xmath884 vertices @xmath154 which form an edge of @xmath885 together with @xmath886 .
let @xmath887 be such a set of vertices .
let @xmath888 be the prepath @xmath889 .
then @xmath890 is a set of possible initial vertices of @xmath67 and @xmath66 is a set of possible final vertices .
( we do not remove @xmath86 from @xmath891 and @xmath87 from @xmath892 at this stage . ) since by lemma [ suppgraphconnected ] the supplementary graph @xmath769 is connected , we can find a walk @xmath55 from @xmath893 to @xmath209 in @xmath769 such that every @xmath664=v(r^*)$ ] appears as an initial , link or final vertex in @xmath55 ( these vertices were defined in section [ walks ] ) and such that @xmath55 has length @xmath894 .
let @xmath895 be the edges of this walk , let @xmath896 , and let @xmath897 be the link vertices of the walk
. for each @xmath664 $ ] , let @xmath898 : r_j = i\}|$ ] , that is , the number of times @xmath72 appears as an initial , link or final vertex in @xmath55 .
so @xmath899 for every @xmath72 and @xmath900 .
our next aim is to apply lemma [ interpath ] to each edge @xmath54 in order to find a loose path @xmath56 in @xmath4 , which we will extend to a prepath @xmath64 with many possible initial vertices in @xmath901 and many possible final vertices in @xmath902 .
we shall do this for each @xmath895 in turn .
so suppose that @xmath903 $ ] and that for all @xmath904 we have defined loose paths @xmath56 in @xmath4 as well as sets @xmath905 extending @xmath56 to a prepath @xmath906 which satisfy the following properties : * @xmath56 lies in the sub-@xmath0-graph of @xmath4 induced by @xmath907 and contains at most @xmath799 vertices .
* the initial vertex @xmath908 of @xmath56 lies in @xmath901 and its final vertex @xmath909 lies in @xmath902 .
* @xmath910 and @xmath911 .
* there is a set @xmath912 of at least @xmath913 possible initial vertices for @xmath906 .
similarly , there is a set @xmath914 of at least @xmath915 possible final vertices for @xmath906 . *
all the prepaths @xmath916 are disjoint .
* for each @xmath664 $ ] and all @xmath917 let @xmath918 , where @xmath919 . for each @xmath920
$ ] set @xmath921 .
then for every @xmath922 with @xmath923 we have @xmath924 .
moreover @xmath925 .
let us now show how to find @xmath926 , @xmath927 and @xmath928 . apply lemma [ interpath ] with @xmath929 , @xmath930 , @xmath931 and with @xmath932 to find a loose path @xmath926 which satisfies ( d1 ) , ( d2 ) , ( d6 ) and is disjoint from @xmath916 .
moreover , the initial vertex @xmath295 of @xmath926 lies in at least @xmath933 edges of @xmath934 , and the final vertex @xmath935 of @xmath926 lies in at least @xmath936 edges of @xmath937 .
we can now use the latter property to choose sets @xmath927 and @xmath928 which extend @xmath926 to a prepath @xmath938 satisfying ( d3)(d5 ) .
the argument for this is similar to that for the extension of @xmath48 to @xmath888 .
altogether this shows that we can find prepaths @xmath939 satisfying ( d1)(d6 ) . for each @xmath636 $ ]
we let @xmath940 be the maximal integer such that @xmath941 .
thus @xmath942 by ( d1 ) .
but if @xmath943 then ( d5 ) and ( d6 ) together imply that @xmath944 and so @xmath945 . we claim that this also holds if @xmath946 . to see this , recall that since @xmath56 is loose , we have @xmath947 for each @xmath948 $ ] . hence @xmath949,\ i\neq r_{\ell+1}}|x^i(\ell)|\\ & \stackrel{(\ref{exceppathsize})}{\equiv } & -1-\ell+\sum_{i\in [ t ' ] \
i\neq r_{\ell+1 } } d_i \equiv -d_{r_{\ell+1 } } \mod k-1\end{aligned}\ ] ] as @xmath950 } d_i$ ] .
let @xmath951 .
since by ( d3 ) for each @xmath664 $ ] there are exactly @xmath952 vertices of @xmath47 which lie in @xmath953 but not in @xmath954 , this in turn implies that @xmath955 , @xmath956 , @xmath957 , @xmath958 and @xmath959 . in order to complete our prepaths
@xmath960 to a hamilton cycle we wish to choose @xmath80 disjoint loose paths @xmath961 within each @xmath962 $ ] which together contain all the vertices in @xmath963 and which ` connect ' successive prepaths @xmath64 .
we achieve this as follows .
let @xmath816 be the set of all @xmath964 $ ] with @xmath965 .
so @xmath816 is the set of positions at which @xmath72 occurs as an initial , final or link vertex in our walk @xmath55 and @xmath966 .
let @xmath967 be the elements of @xmath816 .
then we choose the @xmath968 ( @xmath969 $ ] ) in such a way that the initial vertex of @xmath968 lies in @xmath970 and its final vertex lies in @xmath971 , all the @xmath968 are disjoint and together they cover all the vertices in @xmath963 . to see that this can be done , first note that @xmath972 .
so using lemma [ loosepath ] together with ( c5 ) and ( [ yi ] ) it is easy to check that the complete @xmath0-partite @xmath0-graph on @xmath963 contains such paths ( e.g. first choose @xmath973 , each consisting of precisely 2 edges , and then apply ( c5 ) and lemma [ loosepath ] to find a loose path @xmath968 containing all the remaining vertices of @xmath963 ) . now
( c3 ) and ( d4 ) together imply that @xmath974{\setminus}m^i[y^i]$ ] contains the @xmath0-complexes induced by these paths ( i.e. it contains @xmath975 ) . but
this means that we can find the required paths @xmath961 in each @xmath962 $ ] . finally , for each @xmath969 $ ] write @xmath976 for @xmath968 and @xmath977 for its initial and @xmath978 for its final vertex ( where @xmath979 is as defined in the previous paragraph ) .
to obtain our hamilton cycle in @xmath4 we first traverse @xmath980 , then we use the edge @xmath981 in order to move to the initial vertex @xmath982 of @xmath983 .
( this is possible since @xmath984 . )
now we traverse @xmath983 and use the edge @xmath985 to get to @xmath311 .
( again , this is possible since @xmath986 . )
next we traverse @xmath987 and use the edge @xmath988 to move to @xmath989 .
we continue in this way until we have reached the initial vertex @xmath990 of @xmath980 again .
( so in the last step we traversed @xmath991 and used the edge @xmath992 . )
this completes the proof of theorem [ main ] . | we prove that any @xmath0-uniform hypergraph on @xmath1 vertices with minimum degree at least @xmath2 contains a loose hamilton cycle .
the proof strategy is similar to that used by khn and osthus for the 3-uniform case .
though some additional difficulties arise in the @xmath0-uniform case , our argument here is considerably simplified by applying the recent hypergraph blow - up lemma of keevash . |
consider the dynamics @xmath32 , where @xmath24 for any pure state @xmath137 , let @xmath138 , @xmath139 .
as shown in the main text , the optimal probe states are those states that minimize @xmath140 for all @xmath38 , note that @xmath141\\ = & f(|\varphi_{sa}\rangle \langle \varphi_{sa}| , u'\otimes i_a|\varphi_{sa}\rangle \langle \varphi_{sa}|u'^\dagger\otimes i_a)\\ = & |\langle \varphi_{sa}| u'\otimes i_a|\varphi_{sa}\rangle|\\ = & |tr(\rho_su')| , \endaligned\end{aligned}\ ] ] where @xmath142 and @xmath48 which can be written as @xmath143}$ ] with @xmath144 as a unit vector .
the eigenvalues of @xmath145}$ ] are @xmath146 , we can thus diagonalize @xmath89 as @xmath147 where @xmath148 is a unitary which depends on @xmath144 .
let @xmath149 , then @xmath150|\\ & = |\tilde{\rho}_{11}e^{ia}+\tilde{\rho}_{22}e^{-ia}|\\ & = \sqrt{\cos^2(a)+(\tilde{\rho}_{11}-\tilde{\rho}_{22})^2\sin^2(a)}\\ & \geq |\cos(a)| , \endaligned\end{aligned}\ ] ] the minimum is achieved when @xmath151 , i.e. , when @xmath152 , from which we obtain the optimal @xmath153(note that other choices of @xmath154 will not be invariant under the conjugation of @xmath148 , which then does not lead to any fixed probe state ) .
the optimal probe state @xmath137 is thus any maximally entangled state .
[ [ the - maximal - quantum - fisher - information - matrix - for - the - estimation - of - a - magnetic - field - under - free - evolution ] ] the maximal quantum fisher information matrix for the estimation of a magnetic field under free evolution ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ in this section we show how to calculate @xmath69 in the estimation of a magnetic field under free evolution . in the main text we showed that @xmath155\}. \endaligned\end{aligned}\ ] ] to calculate @xmath69 , let @xmath156}e^{-i ( b+db)t[\vec{n}(\theta+d\theta,\phi+d\phi)\cdot \vec{\sigma}]}\\ & = e^{ia(x , dx)[\vec{k}(x , dx)\cdot \vec{\sigma } ] } \endaligned\end{aligned}\ ] ] where @xmath157 is a unit vector , @xmath158 which can be easily obtained from @xmath159=\cos(bt)\cos[(b+db)t]+\vec{n}(\theta,\phi)\cdot\vec{n}(\theta+d\theta,\phi+d\phi ) \sin(bt)\sin[(b+db)t]$ ] .
since the eigenvalues of @xmath160 are @xmath161 we have @xmath162 , @xmath163 , thus @xmath67=\frac{e^{u'}_{\max}-e^{u'}_{\min}}{2}=a(x , dx)$ ] .
using eq.([eq : u ] ) we can expand the right side of eq.([eq : defjsm ] ) to the second order @xmath164\}\\ = & 8[1-\cos a(x , dx)]\\ = & 4[t^2db^2+\sin^2(bt)d\theta^2+\sin^2\theta\sin^2(bt)d\phi^2 ] , \endaligned\end{aligned}\ ] ] the maximum qfim is then obtained by comparing coefficients at both sides of eq.([eq : defjsm ] ) , which gives @xmath165 we further show that this @xmath69 dominates all other qfims under free evolution , i.e. , @xmath166 for all other @xmath17(@xmath167 means @xmath168 is positive semi - definite ) . to see this
take any probe state @xmath37 , let @xmath169 , @xmath170 , here @xmath171 , then for any @xmath172(which is always assumed to be sufficiently small so the second order expansion is valid ) , we have @xmath173\\ & \leq \max_{\rho_{sa}}d^2_{bures}[u(x , t)\otimes i_a(\rho_{sa}),u(x+dx , t)\otimes i_a(\rho_{sa})]\\ & = \sum_{ij}\frac{1}{4}j_{ij}^{\max}dx_idx_j , \endaligned\end{aligned}\ ] ] where the last equality is achieved by taking the @xmath37 as the maximally entangled state .
thus @xmath174 for all @xmath38 , i.e. , @xmath175 , @xmath69 thus sets the precision limit for estimating @xmath10 .
this is consistent with previous studies@xcite , however our method makes it easy to incorporate feedback controls .
[ [ measurement - in - the - bell - basis - saturates - the - quantum - cramr - rao - bound ] ] measurement in the bell basis saturates the quantum cramr - rao bound ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ in this section we show that there exists povm saturates the quantum cramr - rao bound in estimating a magnetic field under the optimal sequential feedback scheme , one can also see sec.[sec : sud ] of this supplementary material for an alternative proof . with the maximally entangled state @xmath177 as the probe state
, the output state under the free evolution @xmath178 is @xmath179 which can be rewritten as @xmath180 the probability of the measurement results in the bell - basis @xmath181 can then be easily obtained as @xmath182 based on the probability distribution it is straightforward to calculate the classical fisher information matrix as @xmath183 which is the same as the quantum fisher information matrix .
the quantum cramr - rao bound is thus asymptotically saturable in this case .
this is consistent with previous studies@xcite . under the optimal feedback scheme the total dynamics
is given by @xmath77 where @xmath78 and the feedback controls @xmath184 .
since both the system dynamics and the feedback controls only act on the system , the total dynamics can be written as @xmath185}\otimes i_a$ ] where @xmath186 are some functions of @xmath25 . with the maximally entangled state @xmath177 as the probe state
, the output state is similarly given by @xmath187 it is intuitively clear that better precision of @xmath188 corresponds to better precision of @xmath189 , the measurement in the bell basis thus is also optimal . to see this more formally ,
let @xmath190 and denote @xmath191 as the symmetrical logarithm derivatives of @xmath192 respectively , which are the solutions to the following equations @xmath193 to show the quantum cramr - rao bound is achievable we just need to show @xmath194=0 $ ] @xmath195@xcite .
we first denote @xmath196 as the symmetrical logarithm derivatives of @xmath197 respectively , which are solutions to the following equations @xmath198 it is known that @xmath194=0 $ ] @xmath199@xcite ( we also have shown specifically that the measurement in the bell basis saturates the quantum cramr - rao bound for @xmath188 ) .
now since @xmath200 we get @xmath201 which are just linear combinations of @xmath196 . thus @xmath194=0 $ ] @xmath195 , the quantum cramr - rao bound is then saturable . and as @xmath191 are just linear combinations of @xmath196 , the measurements that saturate the quantum cramr - rao bound for @xmath189 is then the same as the measurements saturate the bound for @xmath188 , in particular the measurement in the bell basis saturates the quantum cramr - rao bound .
this can also be seen in sec.[sec : sud ] of this supplementary material , where we show that the feedback controls essentially shift the parameter values , since the measurement in the bell basis is optimal for all values of the parameters , this measurement is then also optimal under the feedback scheme . for completeness
we include a derivation of the optimal feedback control which follows the treatment in @xcite .
we derive the case of @xmath202 , same strategy works in the general case . when @xmath202 , @xmath203 , here @xmath204 and @xmath205 are feedback controls which can act on system+ancilla .
now @xmath206\times\\ & u_1[u_a(x , t)u^\dagger_a(x , t)]u_a(x+dx , t)\\ = & ( u^\dagger_a(x , t)u_1^\dagger ) [ u^\dagger_a(x , t ) u_a(x+dx , t)](u_1u_a(x , t))\\ & [ u^\dagger_a(x , t)u_a(x+dx , t)].\\ \endaligned\end{aligned}\ ] ] @xmath207 can be chosen as any unitary since it does not change @xmath208 $ ] .
now divide @xmath209 into two parts , @xmath210(u_1u_a(x , t))$ ] and @xmath211 , then @xmath212\leq c[u^\dagger_a(x , t ) u_a(x+dx , t)]\\ & + c[(u^\dagger_a(x ,
t)u_1^\dagger ) [ u^\dagger_a(x , t ) u_a(x+dx , t)](u_1u_a(x , t))]\\ & = 2c[u^\dagger_a(x , t ) u_a(x+dx , t ) ] , \endaligned\end{aligned}\ ] ] the first inequality we used the fact @xmath213 if @xmath214@xcite ; the second equality is based on the fact that @xmath210(u_1u_a(x , t))$ ] has the same eigen - angles as @xmath211 .
one choice of control that saturates the equality is let @xmath215 , as it aligns the eigenvalues of the two parts and the corresponding maximal and minimal eigen - angles add up , in this case @xmath216 ^ 2 $ ] , @xmath208=2c[u_a^\dagger(x , t ) u_a(x+dx , t)]$ ] saturates the equality .
this argument can be easily generalized to the general @xmath124 with @xmath217\leq nc[u_{a}^\dagger(x , t ) u_{a}(x+dx , t)]$ ] and the equality can be saturated with the controls @xmath218 and arbitrary @xmath219 , which for simplicity , can also be taken as @xmath220 .
we consider the estimation of general hamiltonians for @xmath22 dynamics under the aid of ancillary systems and feedback controls , with the probe state taken as the maximally entangled state @xmath221 , where @xmath222 form an orthonormal basis of a d - dimensional hilbert space .
let @xmath223 , here @xmath121 form a basis of @xmath122 and @xmath224 are traceless self - adjoint matrices and chosen as @xmath120 , @xmath123 are the interested parameters , @xmath79 . and
@xmath225 , with the controls taking as @xmath226 .
when @xmath112 , @xmath113 , then @xmath227 , we then have @xmath228[(i+i\sum_{j=1}^{d^2 - 1}x_jf_jt)\otimes i_a]\\ \approx&[i+ i\sum_{j=1}^{d^2 - 1}(x_j-\hat{x}_j)f_jt]\otimes i_a\\ \approx & e^{i\sum_{j=1}^{d^2 - 1}(x_j-\hat{x}_j)f_jt}\otimes i_a , \endaligned\end{aligned}\ ] ] the total dynamics is then @xmath229^n\\ & \approx e^{in\sum_{j=1}^{d^2 - 1}(x_j-\hat{x}_j)f_jt}\otimes i_a\\ & = e^{i\sum_{j=1}^{d^2 - 1}(x_j-\hat{x}_j)f_jt}\otimes i_a . \endaligned\end{aligned}\ ] ] in the asymptotical limit @xmath230 , the controls essentially reduce the problem of estimating a general @xmath10 to estimating @xmath231 . in
the remain part we will then just consider the precision under the dynamics @xmath232 at the point of @xmath233 .
we follow the treatment in @xcite to calculate the quantum fisher information matrix of this dynamics .
the output state in this case is unchanged , @xmath234 .
denote @xmath235 as the symmetric logarithm derivative of the output state with respect to @xmath236 , and @xmath237 , then it is known that @xmath238 with @xmath239@xcite . to calculate @xmath240
we can expand @xmath241 , thus @xmath242}{\partial x_j}|_{x=(0,0,\cdots,0)}\\ = & f_j\otimes i_a|\varphi\rangle t. \endaligned\end{aligned}\ ] ] then @xmath243 where for the second equality we used the fact the reduced state of a maximally entangled state is @xmath244 , for the third equality we used the fact that @xmath224 is traceless .
thus @xmath245 it is then easy to see that @xmath246 the entries of the quantum fisher information matrix is given by @xmath247 , we then have @xmath248 .
since in this case @xmath249 the quantum cramr - rao bound @xmath250 is thus achievable@xcite , we then have @xmath132 .
one measurement that saturates the quantum cramr - rao bound in this case can be taken as the projective measurements on the basis consisting with @xmath251 , @xmath252 , and @xmath253 .
as when @xmath254 , the final state can be approximated as @xmath255\frac{1}{\sqrt{d}}\sum_{p=1}^d|pp\rangle$ ] , the probability under the chosen projective measurement can be calculated as @xmath256 and @xmath257 .
it is then straightforward to verify that when @xmath254 the classical fisher information matrix is @xmath258 , which is the same as the quantum fisher information matrix .
it is also easy to see that when @xmath259 , this measurement can be reduced to the projective measurements in the bell basis by choosing @xmath260 as pauli matrices .
the quantum fisher information matrix under the optimal parallel scheme is given by @xmath261@xcite while in this case @xmath262 , thus @xmath263 which is @xmath0-times bigger than the value obtained under the feedback scheme .
[ [ the - gain - of - the - optimal - feedback - scheme - over - the - optimal - parallel - scheme - for - the - estimation - of - a - magnetic - field ] ] the gain of the optimal feedback scheme over the optimal parallel scheme for the estimation of a magnetic field ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ under the total dynamics @xmath225 , with the controls taking as @xmath264(here @xmath265 denote the estimated value used in the control ) , when @xmath124 is sufficiently large the feedback controls shift the parameter from the true value @xmath266 by the amount of @xmath265 . by preparing the probe state as the maximal entangled state and performing projective measurements in the bell basis @xmath181 we get the measurement results with the probabilities given by @xmath267 here we use @xmath268 to denote the shifted value , i.e. , @xmath269 .
a standard procedure for the estimation of the magnetic field is to take the measurements @xmath12 times , and count the occurrence of each measurement results , say @xmath270 , @xmath271 , here @xmath272 .
one then adjusts the estimation based on these occurrences : from the ratio @xmath273 one can adjust the estimation of @xmath274 , from the ratio @xmath275 adjusting the estimation of @xmath276 and from the ratio @xmath277 adjusting the estimation of @xmath278 .
the data has a multinomial distribution for which one can also obtain the maximum likelihood estimation by solving the following equations@xcite @xmath279 from the solution of the equation one can get the estimator for @xmath280 , and the estimator for the true value is then given by @xmath281 , here @xmath282 .
it is well - known that when @xmath12 is sufficiently large this maximum likelihood estimation saturates the cramr - rao bound@xcite as long as eq.([eq : est ] ) has a unique solution , i , e .
, when @xmath283 are known to belong to some intervals . for example
@xmath284 , which is @xmath285 , has a unique solution if @xmath286 $ ] for some @xmath287 , i.e. , when @xmath288 $ ] .
since we can choose @xmath265 to shift the parameter , without loss of generality we can take the interval as @xmath289 $ ] .
similarly one can get the condition on @xmath276 and @xmath278 as @xmath290 $ ] . here
the main restriction is on the condition for @xmath274 as when @xmath76 increases the interval gets smaller . in practice
one can start with small @xmath76 , which corresponds to a large interval for @xmath274 , to first get some rough estimation , then gradually increase @xmath76 to get more information each round .
methods to resolve the ambiguity by choosing @xmath76 adaptively have also been studied previously@xcite .
we note that the adaptive choice of @xmath76 is needed for both the sequential feedback and the parallel scheme , here we focus on the comparison of the two schemes in any of the round of the adaptive procedure with a chosen @xmath76 .
we first study the precision limit under the sequential feedback scheme assuming @xmath10 has a prior distribution @xmath291(based on the width of @xmath291 the evolution time is assumed to be chosen such that the maximal - likelihood estimation works ) . the average mean square error
is given by @xmath292 ^ 2dkdx.\ ] ] when @xmath12 is sufficiently large , for each @xmath10 the maximum likelihood estimator is approximately unbiased , we thus have @xmath293=\frac{1}{n}tr[j^{-1}(x-\hat{x}_c)],\ ] ] here @xmath294 is the quantum fisher information matrix at @xmath10 under the feedback scheme , which equals to the quantum fisher information matrix(without controls ) at @xmath295(since the feedback controls @xmath264 just shift the parameter by @xmath265 when @xmath124 is sufficiently large ) .
we note that in this case the optimal measurement is independent of @xmath10 , the quantum fisher information matrix thus equals to the classical fisher information matrix at all @xmath10 .
the problem in this case can be reduced to a classical estimation problem and as long as @xmath291 is within certain interval so that eq.([eq : est ] ) has a unique solution the quantum cramr - rao bound is achievable asymptotically . for the optimal parallel scheme we use the performance at the optimal point , which is @xmath296 in this case(this
can be seen from eq.(10 ) in the main text ) , to bound the average mean square error , @xmath297,\ ] ] which just says that the average performance is worse than the performance at the optimal point . , @xmath298 , @xmath299 .
the y - axis , @xmath300}{tr[j^{-1}_{fa}(x-\hat{x}_c)]}$ ] with @xmath301 and @xmath302 denote the qfims under the optimal parallel scheme and the feedback scheme respectively ; the x - axis , @xmath303 , quantifies the deviation of the estimation @xmath265 from the true value . in this case
the quantum fisher information matrix under the feedback scheme leads to better precision than the quantum fisher information matrix under the optimal parallel scheme as long as @xmath304 . ]
we then compare the bounds in eq.([eq : feedback ] ) and eq.([eq : parallelsm ] ) . under the feedback scheme from eq.([eq : jtsm ] ) we have @xmath305=tr[j^{-1}(x-\hat{x}_c)]=\frac{1}{4}[\frac{1}{t^2}+\frac{2\tilde{b}^2}{\sin^2(\tilde{b}t)}]$ ] , here @xmath306 . while under the optimal parallel scheme , from eq.(17 ) of the main text we have @xmath307=\frac{9}{4n(n+2)t^2}\approx \frac{9}{4t^2}$ ] .
it is then easy to see that as long as @xmath308 , i.e. , as long as @xmath309 , the feedback scheme outperforms the parallel scheme .
thus as long as the prior distribution , @xmath291 , is mainly concentrated in the region of @xmath310 , the feedback scheme outperforms the parallel scheme .
this is about the same width for eq.([eq : est ] ) to have a unique solution , which is about @xmath311 , thus it does not require more prior information for the feedback scheme to outperform the parallel scheme than what is already required for the maximum - likelihood estimation to work . the intuition that the sequential feedback scheme can outperform the parallel scheme is because the sequential feedback scheme makes use of the information in the prior distribution by designing the feedback controls , while the parallel scheme , which undergoes the free evolution , ignores that information .
the exact gain of the feedback scheme with a prior distribution should be bigger than the above analysis as the lower bound in the parallel scheme used in the above analysis is quite loose .
we note that the adaptive choice of @xmath76 is needed for both sequential feedback scheme and the parallel scheme in order to resolve the ambiguity .
we expect that the sequential feedback scheme can still outperform the parallel scheme even when the evolution time is not adaptively chosen to resolve the ambiguity , as long as the prior distribution is not close to be a uniform distribution , since some information will then be encoded in the prior distribution that can be used for the design of the feedback controls(if the prior distribution is uniform the sequential feedback scheme does not outperform the parallel scheme@xcite since the prior distribution contains no additional information that can be used to design the feedback controls ) .
if the evolution time is not adaptively chosen , the performance of both schemes will deteriorate in a similar way due to the ambiguity(the parallel scheme may deteriorate faster due to the incompatibility of the optimal probe states and the optimal measurements at different values of the parameters , which however needs further investigation ) .
future research includes quantifying the gain of the sequential feedback scheme exactly under any prior distribution . | measurement and estimation of parameters are essential for science and engineering , where the main quest is to find out the highest achievable precision with given resources and design schemes to attain it .
two schemes , the sequential feedback scheme and the parallel scheme , are usually studied in quantum parameter estimation .
while the sequential feedback scheme represents the most general scheme , it remains unknown whether it can outperform the parallel scheme for any quantum estimation tasks . in this letter
we show that the sequential feedback scheme has a 3-fold improvement over the parallel scheme for hamiltonian parameter estimations on 2-dimensional systems , and an order of @xmath0 improvement for hamiltonian parameter estimation on @xmath1dimensional systems .
we also show that , contrary to the conventional belief , it is possible to simultaneously achieve the highest precision for estimating all three components of a magnetic field , which sets a benchmark on the local precision limit for the estimation of a magnetic field . a pivotal task in science and technology is to find out the highest achievable precision in measuring and estimating parameters of interest with given resources and design schemes to reach that precision@xcite . typically to estimate some parameters
@xmath2 encoded in some dynamics @xmath3 , a probe state @xmath4 is prepared which evolves under the dynamics @xmath5 . by performing positive operator valued measurements(povm ) @xmath6 , on the output state @xmath7
, one gets the measurement result @xmath8 with a probability @xmath9 . with a prior knowledge that the parameter is within some local interval , the variance of any unbiased estimator of @xmath10
is then bounded below by the fisher information matrix @xmath11@xcite , where @xmath12 is the number of times that the procedure is repeated , @xmath13 denotes the covariance matrix of the estimator , and @xmath14 is the fisher information matrix with the @xmath15-th entry given by @xmath16@xcite .
the fisher information matrix can be further bounded by the quantum fisher information matrix(qfim ) @xmath17 , which gives the quantum cramr - rao bound@xcite @xmath18 in multi - parameter estimation the quantum cramr - rao bound is usually not achievable even asymptotically@xcite .
two tradeoffs have to be considered in multi - parameter estimation : the first tradeoff is on the choice of measurements as the optimal measurements for different parameters are usually incompatible@xcite ; the second tradeoff is on the choice of the probe states since the optimal probe states for different parameters are also usually different .
these tradeoffs are usually dealt with by specifying a particular figure of merit taken as @xmath19 $ ] with @xmath20 then optimizing the measurements and probe states based on the figure of merit .
besides the measurements and probe states , one also needs to optimize the schemes that arrange multiple uses of the dynamics to achieve the ultimate precision limit .
two schemes , the sequential feedback scheme and the parallel scheme , as shown in fig.[fig : scheme ] , are usually studied .
the sequential feedback scheme represents the most general scheme , which includes the parallel scheme as a special case when taking the controls as swap gates .
examples have been found in quantum channel discrimination that the sequential feedback scheme can outperform the parallel scheme for the discrimination of two quantum channels @xcite . in quantum parameter estimation
it remains unknown whether the sequential feedback scheme can outperform the parallel scheme .
based on some upper bounds on the precision limit @xcite , it has been shown that the sequential feedback scheme does not lead to higher precision in single - parameter quantum estimation under several dynamics , including the unitary@xcite and dephasing dynamics@xcite .
this has led to a conjecture that in the asymptotical limit the sequential feedback scheme provides no gains over the parallel scheme for quantum parameter estimation@xcite . in this letter
we show that the sequential feedback scheme outperforms the parallel scheme for hamiltonian parameter estimation , here we focus on the estimations of small shifts of the parameters around some known values .
we first study the estimation of the hamiltonian for @xmath21 dynamics , which is a fundamental problem in quantum parameter estimation@xcite and closely related to the estimation of a magnetic field .
it thus has many applications in quantum sensing , data storage , information processing and magnetic resonance , and has implications in quantum gyroscope , quantum reference frame alignments , etc@xcite . by considering the general sequential feedback scheme we obtain the ultimate local precision limit for the estimation of a magnetic field , and
show that in this case the sequential feedback scheme outperforms the parallel scheme with a 3-fold improvement .
we also show that , contrary to the conventional belief that some tradeoffs have to be made in the estimation of different parameters of a magnetic field , the optimal sequential feedback scheme achieves the highest precision for all three parameters of a magnetic field simultaneously , which sets a benchmark on the local precision of estimating multiple components of a magnetic field .
we then show for the estimation of general hamiltonians for @xmath22 dynamics the sequential feedback scheme outperforms the parallel scheme with an order of @xmath0 improvement .
this sheds light on the comparison between the two schemes .
extension to hamiltonian parameter estimation with prior distributions are also discussed in the supplemental material .
we also note that the sequential feedback scheme is more implementable for many current experimental settings since high - fidelity controls on small systems can now be routinely done while accurately preparing entangled states with many particles for the parallel scheme is still very challenging . for the estimation of a @xmath23-dimensional hamiltonian
, we consider the hamiltonian for a spin-1/2 in a magnetic field , which can be written as @xmath24 here @xmath25 represents the magnitude and the directions of a magnetic field , @xmath26 , @xmath27 and @xmath28 are pauli matrices .
the hamiltonian can also be written concisely as @xmath29 $ ] , where @xmath30 .
we are interested in the ultimate local precision limit in estimating @xmath31 , under the aid of ancillary systems .
we first consider the scheme without feedback controls , then extend to the general sequential feedback scheme .
we denote @xmath32 as the free evolution of the hamiltonian with t units of time and @xmath33 as the evolution with an ancillary system , here @xmath34 denotes the identity operator on the ancillary system .
let @xmath35 and @xmath36 , here @xmath37 denotes the initial state of system+ancilla and @xmath38 represents a small shift of the parameter .
the local precision limit of estimating @xmath10 from the output state @xmath7 is related to the bures distance between @xmath7 and @xmath39 as@xcite @xmath40 here the bures distance @xmath41 is defined as @xmath42 with @xmath43 as the fidelity between @xmath44 and @xmath45 , and @xmath46 is the @xmath15-th entry of the qfim @xmath17 . since @xmath47 here @xmath48 , the maximal qfim is thus related to the minimum fidelity @xmath49 . for any @xmath50 unitary @xmath51 ,
we denote @xmath52 as the eigenvalues of @xmath51 with @xmath53 $ ] , @xmath54 .
we call @xmath55 the eigen - angles of @xmath51 and assume @xmath56 are arranged in decreasing order
. it is known that @xmath57 if @xmath58@xcite .
denote @xmath59 then the equation can be written concisely as @xmath60 . since @xmath61 and @xmath62
, we also have @xmath63(we note that this does not mean ancillary system is not useful , the role of ancillary system will be clear later ) . with eq.([eq
: bj ] ) and eq.([eq : maximum ] ) we can then obtain @xmath64\}. \endaligned\end{aligned}\ ] ] if @xmath65 is continuous with @xmath10 , then when @xmath38 is sufficiently small , @xmath66 , @xmath67\rightarrow 0 $ ] , thus up to the second order @xmath68 .
\end{aligned}\ ] ] to ensure there exists a qfim @xmath17 achieves the @xmath69 for all @xmath38 , we need to show the optimal state @xmath37 that achieves the maximum bures distance in eq.([eq : maximum ] ) is independent of @xmath38 . in the supplemental material
we show that any maximally entangle state(which are those states such that the reduced state is completely mixed , i.e. , @xmath70 ) achieves the maximum bures distance in eq.([eq : maximum ] ) for all @xmath38 , @xmath69 thus corresponds to the qfim of any maximally entangled probe state . and
the maximal qfim is given by@xcite @xmath71 furthermore the projective measurement in the bell - basis saturates the quantum cramr - rao bound . in the supplemental material we showed that the distribution of the measurement results in the bell - basis is given by @xmath72 , @xmath73 , @xmath74 and @xmath75 , which has the classical fisher information matrix equals to @xmath69 .
the quantum cramr - rao bound is thus saturable and @xmath69 sets the local precision limit when the dynamics is evolved for @xmath76 units of time .
this is consistent with previous studies@xcite , however our method makes it easy to incorporate feedback controls as we now show .
for the general sequential feedback scheme as in fig.[fig : scheme](a ) , the total evolution can be written as @xmath77 where @xmath78 with @xmath79 , and @xmath80 denote the feedback controls .
it can be shown that @xmath81\leq nc[u_a^\dagger(x , t)u_a(x+dx , t)]$ ] where the equality can be achieved when @xmath82(see supplementary material@xcite ) . in practice
the true value @xmath10 is not known a - priori , the estimated value @xmath83 need to be used and the controls @xmath84 need be updated adaptively .
this , however , does not affect the asymptotical scaling@xcite . from eq.([eq :
maxqfiphi2 ] ) we then have @xmath85\\ & = n^2 \sum_{ij } ( j_1^{\max})_{ij}dx_idx_j , \endaligned\end{aligned}\ ] ] thus @xmath86 here the equality can be saturated asymptotically with the controls @xmath87 . in this case
the feedback controls only act on the system , we can thus write @xmath88}\otimes i_a , \endaligned\end{aligned}\ ] ] for the last equation we used the fact that any @xmath89 can be written as @xmath90 where @xmath91 is a unit vector . this has similar form as the free evolution , thus by following the same line of argument one can show the optimal probe state is any maximally entangled state which has the qfim @xmath92 under the optimal feedback scheme . in this case the measurement in the bell basis also saturates the quantum cramr - rao bound @xmath93(see supplementary material for detail ) , @xmath94 thus quantifies the asymptotical precision limit . to ease comparison with previous results ,
we rewrite the hamiltonian as @xmath95 with @xmath96 , @xmath97 , @xmath98 . in the asymptotical limit the estimation is in the vicinity of the actual value ,
we can thus write @xmath99 it is then easy to get @xmath100 .
this will be taken as the figure of merit for comparison as it is used in previous studies@xcite , which corresponds to take @xmath101 in @xmath19 $ ] under the representation of @xmath102 .
we note that the choice of @xmath101 here is just for the purpose of comparison , the precision limit obtained under the feedback scheme is optimal for any @xmath103as the obtained precision saturates the quantum cramr - rao bound @xmath104 , thus for any choices of @xmath103 it also saturates the lower bound @xmath105\geq tr[(j^{\max}_{n})^{-1}g]$ ] . here
@xmath12 is the number of times that the procedure is repeated , which accounts for the classical effect , for the following we will neglect @xmath12 by assuming the procedure is repeated with the same(sufficiently large ) number of times .
we now compare @xmath106 obtained from different schemes .
under the optimal sequential feedback scheme we have @xmath107 with @xmath92 , thus @xmath108 .
\endaligned\end{aligned}\ ] ] under the parallel scheme the precision has been extensively studied previously @xcite with the highest precision given by @xmath109@xcite , here @xmath110 .
this corresponds to @xmath111 .
\endaligned\end{aligned}\ ] ] compare eq.([eq : scaling ] ) and eq.([eq : parallel ] ) we can see that the optimal sequential feedback scheme has a 3-fold improvement over the optimal parallel scheme . for a given @xmath76 , when @xmath112 , @xmath113 , @xmath114 , the precision limit under the optimal sequential feedback scheme thus reaches @xmath115 note that for the estimation of a single parameter @xmath116 the highest precision one can get within @xmath76 units of time is @xmath117@xcite .
it is conventionally believed that for simultaneous estimation of different parameters of a magnetic field some tradeoffs have to be made on the probe states and measurements , thus not possible to achieve the highest precision for all parameters simultaneously .
while the tradeoffs are indeed unavoidable under the parallel scheme , we showed that , contrary to this conventional belief , the optimal sequential feedback scheme can achieve the highest precision for all three parameters of a magnetic field simultaneously . we next show that for the estimation of general hamiltonian for @xmath22 dynamics the sequential feedback scheme has similar improvement over the parallel scheme
.
given an @xmath22 dynamics aided with ancillary system , @xmath118 , here @xmath119 are traceless self - adjoint matrices and @xmath120 , i.e. , @xmath121 form an orthogonal basis of @xmath122 , @xmath123 are the parameters to be estimated .
we compare three schemes : 1)the independent scheme ; 2)the parallel scheme ; 3)the sequential feedback scheme .
the independent scheme is to divide the @xmath124 uses of the dynamics into @xmath125 groups and use @xmath126 dynamics in each group to estimate one parameter . under this scheme
the variance of each parameter @xmath127 , the summation of variance is then @xmath128 . for the parallel scheme
the minimum summation of variance has been obtained previously as @xmath129@xcite . for the sequential feedback scheme ,
we show that(see supplemental material@xcite ) by taking the maximally entangled state as the probe state and using the optimal feedback control @xmath130 , the quantum fisher information matrix is given by @xmath131 , and the quantum cramr - rao bound can be saturated .
the summation of variance under the optimal feedback scheme is thus given by @xmath132 , which has an order of @xmath0 improvement over the parallel scheme and an order of @xmath133 improvement over the independent scheme . _ discussion and conclusion : _ the comparison between the sequential feedback scheme and the parallel scheme has been a subject of lasting interest in quantum channel discrimination and quantum parameter estimation . in quantum channel discrimination ac@xmath134n@xcite and
@xcite studied the optimal parallel scheme for the discrimination between two unitary dynamics , @xcite then showed the sequential feedback scheme is equivalent to the parallel scheme for the discrimination of unitary dynamics , @xcite showed the sequential feedback scheme can outperform the parallel scheme for discriminating quantum channels with memory effects .
optimal sequential scheme has also been obtained for the discrimination of two general quantum channels@xcite . for single - parameter quantum estimation
the sequential feedback scheme is shown to be equivalent to the parallel scheme under unitary@xcite and dephasing dynamics@xcite , and it has been conjectured that asymptotically the sequential feedback scheme is equivalent to the parallel scheme@xcite . for multi - parameter quantum estimation
@xcite showed the parallel scheme has an order of @xmath135 improvement over the independent scheme for estimating @xmath136 parameters with commutating generators , for general unitary dynamics the optimal parallel scheme has also been studied@xcite which shows the parallel scheme has similar improvement over the independent scheme .
prior of this study a general belief has been that under unitary dynamics the sequential feedback scheme is equivalent to the parallel scheme(while under noisy dynamics the sequential feedback scheme is believed to be either equivalent to the parallel scheme or can only outperform the parallel scheme for channels with special properties ) . here by showing the sequential feedback scheme
has an order of @xmath0 improvement over the parallel scheme for the estimation of @xmath22 dynamics , our study disclosed a unique feature for multi - parameter quantum estimation and deepened the understanding on the relationship between the sequential feedback scheme and the parallel scheme .
our study also sets a benchmark on the local precision limit for the estimation of a magnetic field , which is of practical importance for many applications .
the precision is obtained by optimizing all steps in the procedure of the estimation , thus represents the ultimate precision one can achieve for the estimation of a magnetic field asymptotically .
our study shows that it is possible to achieve the highest precision simultaneously for all three parameters of a magnetic field , contrary to the conventional belief that some tradeoffs have to be made on the precision of different parameters .
this opened the possibility and initiated the study of using feedback controls to counteract the tradeoffs in multi - parameter quantum estimation .
future research includes finding the ultimate precision at the presence of general noises . in the supplemental material we also discussed the possible extension to hamiltonian parameter estimation with a prior distribution , and showed that the feedback scheme gains over the parallel scheme through adaptive choice of the evolution time@xcite .
intuitively the feedback scheme gains over the parallel scheme by utilizing the information encoded in the prior distribution to design the feedback controls , while under the parallel scheme the information is ignored during the evolution stage .
future research includes quantifying the gain of the feedback scheme exactly under any prior distribution .
69 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , , , * * , ( ) .
, , , * * , ( ) . , ,
, * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.96.010401 . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , ( ) , http://arxiv.org/abs/1402.0495 . , * * , ( ) . , * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.113.250801 . , , , * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.112.120405 . , , , * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.109.233601 . , , , * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.106.090401 .
, , , * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.111.113601 . , , , , * * , ( ) , http://link.aps.org/doi/10.1103/physrevx.5.031018 . , , , , , , * * , ( ) , http://link.aps.org/doi/10.1103/physreva.80.052114 . , , , , * * , ( ) , http://link.aps.org/doi/10.1103/physreva.86.053813 . , , , , * * , ( ) , http://link.aps.org/doi/10.1103/physreva.85.041802 . , * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.108.230401 . , * * , ( ) , http://link.aps.org/doi/10.1103/physrevx.2.041006 . ,
_ _ ( , ) . ,
_ _ ( , ) . , _ _ ( , ) . , * * , ( ) . , * * , ( ) . , * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.72.3439 . , , , * * , ( ) . , , , , , ,
* * , ( ) , http://link.aps.org/doi/10.1103/physreva.87.012107 . , , , , , , , , , * * , ( ) . , , , , * * , ( ) , http://link.aps.org/doi/10.1103/physreva.89.023845 .
, , , , * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.111.070403 . , , , * * , ( ) . , * * , ( ) , http://link.aps.org/doi/10.1103/physreva.90.043818 . ,
( ) . , , , ( ) . , , , , , , , , * * , ( )
.
, * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.116.030801 . , * * , ( ) , http://link.aps.org/doi/10.1103/physreva.61.042312 . , , , * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.101.180501 . , , , , * * , ( ) , http://link.aps.org/doi/10.1103/physreva.81.032339 . , * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.86.4160 . , , , * * , ( ) , http://link.aps.org/doi/10.1103/physreva.64.050302 . , , , ,
* * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.93.180503 . , , , * * , ( ) , http://link.aps.org/doi/10.1103/physreva.69.050303 . , , , * * , ( ) , http://link.aps.org/doi/10.1103/physreva.70.030301 . , , , * * , ( ) , http://link.aps.org/doi/10.1103/physreva.72.042338 . , * * , ( ) , http://link.aps.org/doi/10.1103/physreva.65.012316 . , * * , ( ) . , * * , ( ) , http://link.aps.org/doi/10.1103/physreva.69.022303 . , * * , ( ) , http://link.aps.org/doi/10.1103/physreva.70.032310 . , * * , ( ) , http://scitation.aip.org/content/aip/proceeding/aipcp/10.1063/1.1834432 . , * * , ( ) , http://link.aps.org/doi/10.1103/physreva.90.022117 . , , , * * , ( ) .
, , , ( ) . , *
* , ( ) , http://link.aps.org/doi/10.1103/physreva.90.022333 .
see supplementary material .
, * * , ( ) , http://stacks.iop.org/0305-4470/35/i=13/a=307 . , * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.115.110401 . , , , * * , ( ) . , * * , ( ) . , _ _ ( ) . , * * , ( ) . , p. ( ) . , * * , ( ) , http://scitation.aip.org/content/aip/journal/jmp/49/10/10.1063/1.2988130 . , * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.87.177901 . , , , * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.87.270404 . , , , * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.98.100503 . , , , * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.103.210501 . , * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.108.120402 . |
satellites of galaxies are important probes of the dynamics and masses of galaxies .
currently , analysis of satellite dynamics is one of the best methods of estimating the masses within large radii of our galaxy and of the local group ( e.g. , einasto & lynden - bell 1982 ; lynden - bell et al .
1983 ; zaritsky et al . 1989 ; fich & tremaine 1991 ) , as well as the masses of other galaxies ( zaritsky & white 1994 ; zaritsky et al .
although the satellites of the milky way and andromeda galaxy have been studied for a long period of time , their number is still uncertain .
more and more satellites are being discovered ( irwin et al .
1990 ; whiting et al . 1997
; armandroff et al . 1998 ; karachentseva & karachentsev 1998 ) with a wide range of properties ; some of them are relatively large and luminous and have appreciable star formation rates ( e.g. , m33 and the large magellanic cloud ; lmc ) . exempting the strange case of ic10 , which exhibits a high star formation rate ( @xmath13 ; mateo 1998 ) , most of the satellites are dwarf spheroidals and dwarf ellipticals with signs of only mild star - formation of @xmath14 .
the star formation history of the satellites shows remarkable diversity : almost every galaxy is a special case ( grebel 1998 ; mateo 1998 ) .
this diversity makes it very difficult to come up with a simple general model for formation of satellites in the local group . because of the generally low star formation rates , it is not unexpected that the metallicities of the satellites are low : from @xmath15 for draco and and iii to @xmath16 for ngc 205 and pegasus ( mateo 1998 ) .
there are indications that properties of the satellites correlate with their distance to the milky way ( mw ) or andromeda , with dwarf spheroidals and dwarf ellipticals being closer to the central galaxy ( grebel 1997 ) .
overall , about 40 satellites in the local group have been found .
formation and evolution of galaxy satellites is still an open problem .
according to the hierarchical scenario , small dark matter ( dm ) halos should on average collapse earlier than larger ones . to some degree ,
this is supported by observations of rotation curves of dark - matter dominated dwarfs and low - surface - brightness galaxies .
the curves indicate that the smaller the maximum circular velocity , the higher the central density of these galaxies .
this is expected from the hierarchical models in which the smaller galaxies collapse earlier when the density of the universe was higher ( kravtsov et al .
1998 ; kormendy & freeman 1998 ) .
thus , it is likely that the satellites of the mw galaxy were formed before the main body of the mw was assembled .
some of the satellites may have survived the very process of the mw formation , whereas others may have been accreted by the mw or by the local group at later stages .
indeed this sequence forms the basis of the currently popular semi - analytical models of galaxy formation ( e.g. , somerville & primack 1998 , and references therein ) .
there have been a number of efforts to use the local group as a cosmological probe .
peebles et al .
( 1989 ) modeled formation of the local group by gravitational accretion of matter onto two seed masses .
kroeker & carlberg ( 1991 ) found pairs of `` galaxies '' in cosmological simulations and used them to estimate the accuracy of traditional mass estimates .
governato et al .
( 1997 ) studied the velocity field around local group candidates in different cosmological models and blitz ( 1998 ) simulated a group of galaxies and compared their results with the observations of the high - velocity clouds in the local group . nevertheless , despite significant effort , theoretical predictions of the abundance and properties of the satellites are far from been complete .
one of the difficulties is the survival of satellites inside halos of large galaxies .
this numerically challenging problem requires very high - resolution simulations in a cosmological context and has been addressed in different ways . in the classical approach
( e.g. , lin & lynden - bell 1982 ; kuhn 1993 ; johnston et al .
1995 ) , one assumes a realistic potential for the mw , a density profile for the satellites ( usually an isothermal model with a central core ) , and numerically follows a satellite as it orbits around the host galaxy .
this approach lends many valuable insights into the physical processes operating on the satellites and alleviates some of the numerical problems .
it lacks , however , one important feature : connection with the cosmological background .
the host galaxy is implicitly assumed to be stable over many billions of years which may not be realistic for a typical galaxy formed hierarchically .
moreover , the assumed isothermal density profile of the satellite is different from profiles of typical dark matter halos formed in hierarchical models ( navarro , frenk & white 1997 ) .
last but not least , the abundances of the satellites can only be predicted if the formation of the satellites and of the parent galaxy is modelled self - consistently .
thus , more realistic cosmological simulations are necessary . unfortunately , until recently numerical limitations prevented the usage of cosmological simulations to address satellite dynamics .
namely , dissipationless simulations seemed to indicate that dm halos do not survive once they fall into larger halos ( e.g. , white 1976 ; van kampen 1995 ; summers et al .
it appears , however , that the premature destruction of the dm satellites inside the virial radius of larger halos was largely due to numerical effects ( moore , katz , lake 1996 ; klypin et al . 1997 ( kgkk ) ) .
indeed , recent high - resolution simulations show that hundreds of galaxy - size dm halos do survive in clusters of galaxies ( kgkk ; ghigna et al .
1997 ; et al .
1998 ) . encouraged by this success
, we have undertaken a study of the survival of satellites in galaxy - size halos .
dynamically , galactic halos are different from cluster - size halos ( mass @xmath17 ) .
clusters of galaxies are relatively young systems in which most of the satellite halos have had time to make only a few orbits .
galaxies are on average significantly older , enabling at least some of their satellites to orbit for many dynamical times .
this increases the likelihood of the satellite being destroyed either from numerical effects of the simulation or the real processes of dynamical friction and tidal stripping .
the destruction of the satellites is , of course , counteracted by accretion of the new satellites in an ongoing process of galaxy formation .
it is clear , therefore , that to predict the abundances and properties of galactic satellites , one needs to model self - consistently both the orbital dynamics of the satellites and the formation process of the parent halo in a cosmological setting . in this paper we present results from a study of galactic satellite abundances in high - resolution simulations of two popular variants of the cold dark matter ( cdm ) models .
as will be described below , the dissipationless simulations used in our study are large enough to encompass a cosmologically significant volume and , at the same time , have sufficient resolution to make the numerical effects negligible .
the paper is organized as follows . in
2 we present the data that we use to estimate the _ observed _ velocity function of satellites of our galaxy and m31 .
cosmological simulations are presented and discussed in 3 .
we compare the predicted and observed velocity functions in 4 to show that the models predict considerably more lower mass satellites than is actually observed in the local group . in 5 and 6
we discuss possible interpretation and implications of our results and summarize our conclusions .
lllccrcccc 10 & 11 /13 & 13 /15 & 12/14 & sculptor carina sextans leoii & & & & andi - iii , v , vi , cas pegasus 15 & 7 /9 & 7 /8 & 7/8.5 & phoenix fornax leoi ursmin draco sagit lgs3 20 & 2 /3 & 6 /7 & 4/5 & ic1613 30 & 2 /3 & 6 /6 & 4/4.5 & smc ngc6822 ic10 ngc147 ngc185 50 & 1 /1 & 3 /3 & 2/2 & lmc 70 & 0 /0 & 3 /3 & 1.5/1.5 & m33 m32 ngc205
there are about 40 known galaxies in the local group ( mateo 1998 ) .
most of them are dwarf galaxies with absolute magnitudes of @xmath18 .
while more and more galaxies are being discovered , most of the new galaxies are very small and faint making it seem unlikely that too many larger satellites have been missed .
therefore , we have decided to consider only relatively massive satellites with estimated rotational velocity or three - dimensional velocity dispersion of stars larger than 10 km / s . in order to simplify the situation even further
, we estimate the number of satellites _ per central galaxy_. there is a number of arguments why this is reasonable .
first , it makes comparison with cosmological models much more straightforward .
this is justified to some degree by the fact that the satellites in the local group cluster around either the mw or m31 and there are only a few very remote ones of unclear association with a central galaxy .
we also believe that the estimate of the satellite abundance per galaxy is more accurate because it is relatively straightforward to find the volume of the sample , which would be more difficult if we were to deal with the local group as a whole . using published results ( mateo 1998 ) ,
we have compiled a list of satellites of the milky way and of the m31 with estimated circular velocities above the threshold of @xmath19 . in our estimate of abundances , we have not attempted to decide whether a satellite is bound to its central galaxy or not .
satellites have been simply counted if they lie within a certain radius from the center of their parent galaxy .
we have chosen two radii to make the counts .
the counts of dm satellites were made for the same radii .
the radii were chosen rather arbitrarily to be @xmath20 and @xmath21 .
for a hubble constant of @xmath2 ( @xmath22 km / s / mpc ) , which was assumed for our most realistic cosmological model and which is consistent with current observational results , the radii are 286 kpc and 571 kpc .
the smaller radius is close to a typical virial radius of a milky way size halo in our simulations .
the larger radius allows us to probe larger volumes ( and , thus , gives better statistics ) both in simulations and in observations .
unfortunately , observational data may become less complete for this radius . since we can not estimate the luminosities of galaxies associated with dm satellites in dissipationless simulations , we have chosen circular velocity to characterize both the dark halos and the satellite galaxies .
the circular velocity can be estimated for galaxies ( with some uncertainties ) and for the dm halos . for spiral and irregular galaxies we used the rotational velocity , which is usually measured from 21 cm hi observations . for ellipticals and dwarf spheroidals we used observed line - of - sight velocity dispersion of stars , which was multiplied by @xmath23 to give an estimate of .
using our numerical simulations we confirmed that this gives a reasonably accurate estimate of with an error less than @xmath24 .
table 1 lists the number of satellites with larger than a given value ( first column ) for the milky way galaxy ( second column ) and m31 ( third column ) .
the forth column gives the average number of satellites and the fifth column lists names of the satellites in given velocity bin .
figures 4 and 5 , discussed in detail below , present the cumulative circular velocity distribution of the observed satellites in mw and m31 within 286 kpc and 571 kpc radius from the central galaxies . a few special cases should be mentioned .
there are no measurements of velocity dispersion for and i - iii and the other two satellites of m31 , and v and vi , do not have measured magnitudes .
given that they all seem to have the properties of a dwarf spheroidal , we think it is reasonable to expect that they have in the range 10 - 20 km / s .
details of recent measurements of different properties of these satellites of the m31 can be found in armandroff et al .
( 1998 ) and grebel ( 1998 ) .
we also included cas dsph ( grebel & guhathakurta , 1998 ) in our list with in the range @xmath25 .
one satellite ( and ii ) can be formally included in both lists ( mw and m31 ) .
it is 271 kpc from the m31 , but being at the distance of 525 kpc from mw it is should also be counted as the mw satellite . since this is the only such case , we have decided to count it only once as a satellite of m31 .
lllccrcccc scdm & 1.0 & 0.5 & 1.0 & @xmath26 & 650 - 40,000 & 150 & 2.5 & @xmath27 3cdm & 0.3 & 0.7 & 1.0 & @xmath28 & 650 - 40,000 & 450 & 7.5 & @xmath27
to estimate the satellite abundances expected in the hierarchical models , we have run simulations of two representative cosmologies .
parameters of the models and simulations are given in table 2 , where @xmath29 is the density parameter of the matter at @xmath30 , @xmath31 is the rms of density fluctuations on @xmath32 scale estimated by the linear theory at present time using the top - hat filter .
other parameters given in table 2 specify the numerical simulations : mass of dark a matter particle , @xmath33 , defines the mass resolution , number of time - steps at the lowest / highest levels of resolution , size of the simulation box , and the number of dark matter particles
. numbers on resolution refer to the size of the smallest resolution elements ( cells ) in the simulations .
the simulations have been performed using the adaptive refinement tree ( art ) @xmath34-body code ( kravtsov , klypin & khokhlov 1997 ) .
the art code reaches high force resolution by refining the mesh in all high - density regions with an automated refinement algorithm .
the simulation used here was used in kravtsov et al .
( 1998 ) and we refer the reader to that paper for details and tests .
additional tests and comparisons with a more conventional ap@xmath35 m code will be presented in knebe et al .
the cdm simulation differs from the simulations only in the cosmological parameters and size of the simulation box .
our intent was to use the much more realistic model for comparisons with observations , and to use the cdm model to test whether the predictions depend sensitively on cosmology and to somewhat broaden the dynamical range of the simulations . jumping ahead , we note here that results of the cdm simulation are close to those of the simulation as far as the circular velocity function of satellites is concerned .
this indicates that we are dealing with general prediction of hierarchical scenarios , not particular details of the model .
nevertheless , we do expect that some details of statistics and dynamics of the satellites may depend on parameters of the cosmological models .
the size of the simulation box is defined by the requirement of high mass resolution and by the total number of particles used in our simulations .
dm halos can be identified in simulations if they have more than @xmath36 particles ( kgkk ) .
small satellites of the mw and andromeda have masses of @xmath37 .
thus , mass of a particle in the simulation should be quite small : @xmath38 .
therefore , the number of particles in our simulations ( @xmath27 ) dictates the box size of only a few megaparsec across .
this puts significant constraints on our results .
the number of massive halos , for example , is quite small . in the cdm simulation
we have only three halos with circular velocity larger than @xmath39 .
the number of massive halos in the simulation is higher ( eight ) .
the important issue for our study is the reliable identification of satellite halos .
the problems associated with halo identification within high - density regions are discussed in kgkk . in this study
we use a halo finding algorithm called bound density maxima ( bdm ; see kgkk and et al .
the source code and description of the version of the bdm algorithm used here can be found in klypin & holtzman ( 1997 ) .
the main idea of the bdm algorithm is to find positions of local maxima in the density field smoothed at a certain scale and to apply physically motivated criteria to test whether the identified site corresponds to a gravitationally bound halo .
the algorithm then computes various properties and profiles for each of the bound halos and constructs a uniform halo catalog ready to be used for analysis . in this study
we will use the maximum circular velocity as the halo s defining property .
this allows us to avoid the problem of ambiguous mass assignment ( see kgkk for discussion ) and makes it easier to compare the results to observations .
the density maxima are identified using a top - hat filter with radius @xmath40 ( `` search radius '' ) .
the search is performed starting from a large number of randomly placed positions ( `` seeds '' ) and proceeds by moving the center of mass within a sphere of radius @xmath40 iteratively until convergence . in order to make sure that we use a sufficiently large number of seeds
, we used the position of every tenth particle as a seed .
therefore , the number of seeds by far exceeds the number of expected halos .
the search radius @xmath40 also defines the minimum allowed distance between two halos .
if the distance between centers of any of the two halos is @xmath41 , only one halo ( the more massive of the two ) is left in the catalog . a typical value for the search radius is @xmath42 .
we set a lower limit for the number of particles inside the search radius @xmath43 : halos with @xmath44 are not included in the catalog .
we also exclude halos which have less than 20 bound particles and exclude halos with circular velocity less than @xmath45 .
some halos may have significant substructure in their cores due , for example , to an incomplete merger .
such cases appear in the catalogs as multiple ( 2 - 3 ) halos with very similar properties ( mass , velocity , radius ) at small separations .
our strategy is to count these as a single halo .
specific criteria used to identify such cases are : ( 1 ) distance between halo centers is @xmath46 , ( 2 ) their relative velocity in units of the rms velocity of particles in the halos @xmath47 is less than 0.15 , and ( 3 ) the difference in mass is less than factor 1.5 . if all the criteria are satisfied , only the most massive halo was kept in the catalog .
the box size of the simulations clearly puts limitations on sizes and masses of halos . in a few megaparsec box
, one does not find large groups or filaments .
the mean density in the simulation boxes , however , is equal to the mean density of the universe , and thus we expect our simulations to be representative of the field population of galaxies ( galaxies not in the vicinity of massive clusters and groups ) .
the local group and field galaxies are therefore our main targets .
nevertheless , even in the small boxes used in this paper , the number of halos is very substantial .
we find 1000 2000 halos of different masses and circular velocities in each simulation .
this number is large enough for a reliable statistical analysis .
figure 1 presents the velocity distribution function , defined as the number of halos in a given circular velocity interval per unit volume , in two simulations .
the smaller - box simulation is the one that we use in our further analysis . to estimate whether the halo velocity function is affected by the small - box size
, we compare the small - box result with results from the larger , @xmath48 box , simulation used in et al .
the latter followed the evolution of @xmath49 particles and had a mass resolution of @xmath50 . in the small box ,
the total number of halos with @xmath51 and @xmath52 is 1716 ( 1066 ) for the lowest threshold of 20 bound particles .
the numbers change slightly if a more stringent limit of 25 particles is assumed : 1556 ( 1052 ) . in the overlapping range of circular velocities
@xmath53 the velocity function of the small box agrees very well with that of the ( 0.5,-1.5)(13.0,14.0 ) ( -1.2,14.0 ) ( 0.4,2 . )
= 3.5 mm fig .
1. differential circular velocity distribution function of dark matter halos in the model . the solid curve and
the filled circles are results of the small - box ( box - size of @xmath54 ) simulation .
open circles show the corresponding velocity function in larger ( box - size of @xmath55 ) simulation .
error bars correspond to the poisson noise .
the dotted curve is the power - law with the slope of -3.75 motivated by the press - schechter approximation ( see 4 for details ) . large box .
this shows that the lack of long waves in the small - box simulation has not affected the number of halos with @xmath56 . in the range @xmath57 the velocity function
can be accurately approximated by a power law @xmath58 motivated by the press - schechter ( 1974 ) approximation with assumptions of @xmath59 and of the power - law power spectrum with a slope of @xmath60 . at higher circular velocities ( @xmath61 )
the fit overpredicts the number of halos because the above fit neglects the exponential term in the press - schechter approximation . at small
( @xmath62 ) the points deviate from the fit , which we attribute to the incompleteness of our halo catalog at these circular velocities due to the limited mass resolution .
indeed , comparison with the cdm simulation , which has higher mass resolution , shows that the number of halos increases by about a factor of three when the threshold for changes from 20 to 10 .
we thus estimate the completeness limit of our simulations to be @xmath63 for the simulations and @xmath64 for the cdm run .
note that for the issue of satellite abundance discussed below , any incompleteness of the catalogs at these velocities would increase the discrepancy between observations and models .
figure 2 provides a visual example of a system of satellites around a group of two massive halos in the simulation .
the massive halos have @xmath65 and @xmath66 and masses of @xmath67 and @xmath68 inside the central @xmath69 . in figure 3
the more massive halo is shown in more detail . to some extent
the group looks similar to the local group though the distance between the halos is @xmath70 , which is somewhat larger than the distance between the mw and m31 .
yet , there is a significant difference from the local group in the number of satellites . in the simulation ,
there are 281 identified satellites with @xmath71 within 1.5 sphere shown in figure 2 .
the local group contains only about 40 known satellites inside the same radius .
the number of expected satellites is therefore quite large .
note , however , that the total fraction of mass bound to the satellites is rather small : @xmath72 , where @xmath73 is the total mass inside the sphere .
most of the mass is bound to the two massive halos .
there is another pair of massive halos in the simulation , which has even more satellites ( 340 ) , but the central halo in this case was much larger than m31 .
its circular velocity was @xmath74 .
we will discuss the correlation of the satellite abundances with the circular velocity of the host halo below ( see figs .
the fraction of mass in the satellites for this system was also small ( @xmath75 ) .
( 0,8.0)(15.0,22.6 )
( .0,22.6)=9.cm ( 9.5,22.6)=9.3 cm ( 0.0,12.3 ) = 3.5 mm fig . 2. distribution of dark matter particles inside a sphere of the radius of 1.5 ( solid circle ) for a small group of dark matter halos ( similar in mass to the local group ) in the simulation .
the group consists of two massive halos with circular velocities of @xmath76 and @xmath77 ( masses of @xmath67 and @xmath68 inside @xmath69 radius ) and 281 halos with circular velocities @xmath78 inside @xmath79 .
the distance between the halos is @xmath70 . to enhance the contrast ,
we have color - coded dm particles on a grey scale according to their local density : intensity of each particle is scaled as the logarithm of the density , where the density was obtained using top - hat filter with @xmath80 radius .
( 9.6,12.3 ) = 3.5 mm fig .
3. distribution of dark matter particles inside a sphere of the radius of @xmath81 ( solid circle ) centered on the more massive halo shown in figure 2 .
the small box in the figure has size @xmath82 .
the color - coding is similar to that in figure 2 except that the local density was obtained using top - hat filter of @xmath83 radius .
table 3 presents parameters of satellite systems in the simulation for all central halos with @xmath84 .
the first and the second columns give the maximum circular velocity and the virial mass of the central halos .
the number of satellites and the fraction of mass in the satellites are given in the third and fourth columns .
all satellites within @xmath85 from the central halos , possessing more than 20 bound particles , and with @xmath86 were used .
the last two columns give the three - dimensional rms velocity of the satellites and the average velocity of rotation of the satellite systems .
figures 4 and 5 show different characteristics of the satellite systems in the local group ( see 2 ) , and in the and the cdm simulations .
top panels in the plots clearly indicate that the abundance and dynamics of the satellites depend on the circular velocity ( and thus on mass ) of the host halo .
more massive halos host more satellites and the rms velocity of the satellites correlates with host s circular velocity , as can be expected .
the number of satellites is approximately proportional to the cube of the circular velocity of the central galaxy ( or halo ) : @xmath87 .
this means that the number of satellites is proportional to the galaxy mass @xmath88 because the halo mass is related to as @xmath89 .
the number of the satellites almost doubles when the distance to the central halo increases by a factor of two .
this is very different from the local group where the number of satellites increases only slightly with distance .
note that the fraction of mass in the satellites ( see table 3 ) does not correlate with the mass of the central object .
the velocity dispersion decreases with distance , changing by 10% 20% as the radius increases from @xmath90 to @xmath1 .
we would like to emphasize that both the number of satellites and the velocity dispersion have large real fluctuations by a factor of two around their mean values .
( 0,8.0)(15.0,29 . ) ( -1.5,29)=13.cm ( 8.,29)=13.cm ( 0.0,15.3 ) = 3.5 mm fig .
4. properties of satellite systems within @xmath90 from the host halo . _
top panel : _ the three dimensional rms velocity dispersion of satellites versus maximum circular velocity of the central halo .
_ solid _ and _ open _ circles denote and cdm halos , respectively .
the _ solid line _ is the line of equal satellite rms velocity dispersion and the circular velocity of the host halo . _
middle panel : _ the number of satellites with circular velocity larger than @xmath45 versus circular velocity of the host halo .
the _ solid line _ shows a rough approximation presented in the legend .
_ bottom panel : _ the cumulative circular velocity distribution ( vdf ) of satellites .
_ solid triangles _ show average vdf of milky way and andromeda satellites .
_ open circles _
present results for the cdm simulation , while the _ solid curve _ represents the average vdf of satellites in the simulation for halos shown in the upper panels . to indicate the statistics , the scale on the right y - axis shows the total number of satellite halos in the simulation .
note that while the numbers of massive satellites ( @xmath91 ) agrees reasonably well with observed number of satellites in the local group , models predict about five times more lower mass satellites with @xmath92 .
( 9.6,15.3 ) = 3.5 mm fig . 5.
the same as in figure 4 , but for satellites within @xmath1 from the center of a host halo . in the bottom panel
we also show the cumulative velocity function for the field halos ( halos outside of @xmath1 spheres around seven massive halos ) , arbitrarily scaled up by a factor of 75 .
the difference at large circular velocities @xmath93 is not statistically significant .
comparison between these two curves indicates that the velocity functions of isolated and satellite halos are very similar . as for the satellites within central @xmath20 ( figure 4 ) , the number of satellites in the models and in the local group agree reasonably well for massive satellites with @xmath94 , but disagree by a factor of ten for low mass satellites with @xmath95 .
the bottom panels in figures 4 and 5 present the _ cumulative _ velocity distribution function ( vdf ) of satellites : the number of satellites per unit volume and per central object with internal circular velocity larger than a given value of .
note that the vdf is obtained as the unweighted average of the functions of individual halos in the interval @xmath96 .
this was done to improve the statistics .
however , it is easy to check that the amplitude of the vdf corresponds to the satellite abundance around @xmath97 halos .
for instance , the average vdf shown in figure 5 predicts @xmath98 satellites within the radius of @xmath21 , while the upper panel of this figure shows that this is about what we observe for @xmath99 hosts .
llclcc 140.5 & @xmath100 & 9/15&0.053/0.112 & 99.4/94.4 & 28.6/15.0 278.2 & @xmath101 & 39/94&0.041/0.049 & 334.9/287.6 & 29.8/11.8 205.2 & @xmath102 & 27/44&0.025/0.051 & 191.7/168.0 & 20.0/11.3 175.2 & @xmath103 & 5/10&0.105/0.135 & 129.1/120.5 & 41.5/45.2 259.5 & @xmath104 & 24/52&0.017/0.029 & 305.0/257.3 & 97.1/16.8 302.3 & @xmath105 & 37/105&0.055/0.112 & 394.6/331.6 & 39.4/15.7 198.9 & @xmath106 & 24/58&0.048/0.049 & 206.1/169.3 & 17.7/12.1 169.8 & @xmath107 & 17/26&0.053/0.067 & 162.8/156.0 & 9.3/5.0 the right @xmath108-axis in the lower panels of figures 4 and 5 shows the cumulative number of satellites in all host halos in the simulation .
error bars in the plots correspond to the poisson noise .
the dashed curve in figure 5 shows vdf of all non - satellite halos ( halos located _ outside _ @xmath1 spheres around the massive host halos ) .
comparison clearly indicates that the vdf of the satellite halos has the same shape as the vdf of the field halos with the only difference being the amplitude of the satellites vdf .
there are more satellites in the same volume close to large halos , but the fraction of large satellites is the same as in the field .
we find the same result for spheres of @xmath20 radius .
the velocity distribution function can be roughly approximated by a simple power law .
for satellites of the local group the fit at @xmath109 gives @xmath110 and @xmath111 for @xmath112 and @xmath113 , respectively .
for the simulation at @xmath114 we obtain : @xmath115 @xmath116 again , for @xmath112 and @xmath113 , respectively .
this approximation is formally valid for @xmath114 , but comparisons with the higher - resolution cdm simulations indicates that it likely extends to smaller velocities .
the numbers of observed satellites and satellite halos cross at around @xmath117 .
this means that while the abundance of massive satellites ( @xmath118 ) reasonably agrees with what we find in the mw and andromeda galaxies , the models predict an abundance of satellites with @xmath119 that is approximately five times higher than that observed in the local group .
the difference is even larger if we extrapolate our results to 10 . in this case eq.(4 ) predicts that on average we should expect 170 halo satellites inside a @xmath20 sphere , which is 15 times more than the number of satellites of the milky way galaxy at that radius .
although the discrepancy between observed and predicted satellite abundances appears to be dramatic , it is too early to conclude that it indicates a problem for hierarchical models .
several effects can explain the discrepancy and thus reconcile predictions and observations . in this section
we briefly discuss two possible explanations : the identification of the missing dm satellites with high velocity clouds observed in the local group , and the existence of a large number of invisible satellites containing a very small amount of luminous matter either due to early feedback by supernovae or to heating of the gas by the intergalactic ionizing background . as was recently discussed by blitz et al .
( 1998 ) , abundant high - velocity clouds ( hvcs ) observed in the local group may possibly be the observational counterparts of the low - mass dm halos unaccounted for by dwarf satellite galaxies .
it is clear that not all hvcs can be related or associated with the dm satellites ; there is a number of hvcs with a clear association with the magellanic stream and with the disk of our galaxy ( wakker & van woerden 1997 ; wakker , van woerden & gibson 1999 ; and references therein ) .
nevertheless , there are many hvcs which may well be distant ( @xmath120 ; wakker & van woerden 1997 ; blitz et al .
according to blitz et al . , stability arguments suggest diameters and total masses of these hvcs of @xmath121 and @xmath122 , which is remarkably close to the masses of the overabundant dm satellites in our simulations .
the number of expected dm satellites is quite high .
for the pair of dm halos presented in figure 2 , we have identified 281 dm satellites with circular velocities @xmath123 . since the halo catalog is not complete at velocities @xmath124 ( see 4 ) , we expect that there should be even more dm satellites at the limit @xmath64 .
the correction is significant because about half of the identified halos have cirlular velocities below @xmath125 . using eq.(3 )
we predict that the pair should host @xmath126 dm satellites with @xmath127 within @xmath128 . a somewhat smaller number , 640 satellites , follows from eq.(4 ) , if we double the number of satellites to take into account that we have two massive halos in the system .
the number of hvcs in the local group is known rather poorly .
wakker & van woerden s ( 1991 ) all - sky survey , made with 1 degree resolution , lists approximately 500 hvcs not associated with the magellanic stream .
about 300 hvcs have estimated linewidths ( fwhm ) of @xmath129 ( see fig.1 in blitz 1998 ) , the limit corresponding to 3d rms velocity dispersion of @xmath130 .
stark ( 1992 ) found 1312 clouds in the northern hemisphere , but only 444 of them are resolved .
the angular resolution of their survey was 2 degrees , but it had better velocity resolution than the wakker & van woerden compilation .
comparisons of low and high resolution observations indicates that the existing hvc samples are probably affected by selection effects ( wakker et al .
the abundance of hvcs thus depends on one s interpretation of the data .
if we take 1312 hvcs of stark ( 1992 ) , double the number to account for missing hvcs in the southern hemisphere , we arrive at about 2500 hvcs in the local group .
this is more than three times the number of expected dm satellites .
this large number of hvcs also results in a substantial fraction of mass of the local group confined in hvcs . assuming the average masses given by blitz , this naive estimate gives the total mass in hvcs @xmath131 .
if we take mass of the local group to be @xmath132 ( fich & tremaine 1991 ) , the fraction of mass in the hvcs is high : 0.2 - 0.25 .
this is substantially higher than the fraction of mass in dm satellites in our simulations ( @xmath133 ) .
nevertheless , there is another , more realistic in our opinion , way of interpreting the data .
while it is true that wakker & van woerden ( 1991 ) may have missed many hvcs , it is likely that most of the missed clouds have small linear size .
thus , the mass should not be doubled when we make correction for missed hvcs . in this case 500
hvcs ( as in wakker & van woerden sample studied by blitz ) with average dark matter mass of @xmath134 give in total @xmath135 or 0.05 of the mass of the local group .
this is consistent with the fraction of mass in dm satellites which we find in our numerical simulations .
it should be kept in mind that the small hvcs may contribute very little to the total mass in the clouds . as we have shown above ,
the number density of dm satellites is a very strong function of their velocity : @xmath136 .
if the cloud velocity function is as steep as that of the halos , this might explain why changes of parameters of different observational samples produce very large differences in the numbers of hvcs .
the mass of a dm satellite is also a strong function of velocity : @xmath137 . as the result , the total mass in satellites with velocity _ less _ than @xmath138 is @xmath139 .
the conclusion is that the mass is in the most massive and rare satellites .
if the same is true for the hvcs , we should not double the mass when we find that a substantial number of small hvcs were missed in a catalog . to summarize , it seems plausible that observational data on hvcs are compatible with a picture where every dm satellite either hosts a dwarf galaxy ( a rare case at small ) or an hvc .
this picture relies on the large distances to the hvcs and can be either confirmed or falsified by the upcoming observations ( wakker et al .
note , however , that at present the observed properties of hvcs ( mainly the abundances , distances , and linewidths ) are so uncertain that a more quantitative comparison is impossible .
there are at least two physical processes that have likely operated during the early stages of galaxy formation and could have resulted in the existence of a large number of dark ( invisible ) satellites .
the first process is gas ejection by supernovae - driven winds ( e.g. , dekel & silk 1996 ; yepes et al . 1997 ; mac low & ferrara 1998 ) .
this process assumes at least one initial starformation episode , and thus should produce some luminous matter inside the host dm satellites .
indeed , this process may explain the observed properties of the dwarf spheroidal galaxies in the local group ( e.g. , dekel & silk 1996 ; peterson & caldwell 1993 ; hirashita , takeuchi & tamura 1998 ) .
it is not clear whether this process can also produce numerous very low mass - to - light ratio systems missed in the current observational surveys .
it is likely that some low - luminosity satellites have still been missed in observations , since several faint galaxies have been discovered in the local group just during the last few years ( see 1 ) .
what seems unlikely , however , is that observations have missed so many .
this may still be the case if missed satellites are very faint ( almost invisible ) , but more theoretical work needs to be done to determine whether gas ejection can produce numerous very faint systems .
the recent work by hirashita et al . ( 1998 ) shows that this process may be capable of producing very high mass - to - light ratio ( @xmath140 up to @xmath141 ) systems of mass @xmath142 .
another possible mechanism is prevention of gas collapse into or photoevaporation of gas from low - mass systems due to the strong intergalactic ionizing background ( e.g. , rees 1986 ; efstathiou 1992 ; thoul & weinberg 1996 ; quinn , katz & efstathiou 1996 ; weinberg , hernquist & katz 1997 ; navarro & steinmetz 1997 ; barkana & loeb 1999 ) . numerical simulations by thoul & weinberg ( 1996 ) and by quinn et al .
( 1996 ) show that the ionizing background can inhibit gas collapse into halos with circular velocities @xmath143 .
these results are in general agreement with more recent simulations by weinberg et al .
( 1997 ) and navarro & steinmetz ( 1997 ) .
as explained by thoul & weinberg , accretion of intergalactic gas heated by the ionizing background into dwarf @xmath143 systems is delayed or inhibited because the gas has to overcome pressure support and is , therefore , much slower to turn around and collapse .
if the collapse may be delayed until relatively late epochs ( @xmath144 ) , many low - mass dm satellites may have been accreted by the local group without having a chance to accrete gas and form stars .
this would clearly explain the discrepancy between the abundance of _ dark matter _ halos in our simulations and observed luminous satellites in the local group .
more interestingly , a recent study by barkana & loeb ( 1999 ) shows that gas in small ( @xmath145 ) halos would be photoevaporated during the reionization epoch even if the gas had a chance to collapse and virialize prior to that .
these results indicate that the ionizing background , of the amplitude suggested by the lack of the gunn - peterson effect in quasar spectra , can lead to the existence of numerous dark ( invisible ) clumps of dark matter orbiting around the milky way and other galaxies and thus warrants further study of the subject
. it would be interesting to explore potential observational tests for the existence of dark satellites , given the abundances predicted in hierarchical models .
one such feasible tests , examined recently by widrow & dubinski ( 1998 ) , concerns the effects of dm satellites on microlensing statistics in the milky way halo .
we have presented a study of the abundance and circular velocity distribution of galactic dark matter satellites in hierarchical models of structure formation .
numerical simulations of the and cdm models predict that there should be a remarkably large number of dark matter satellites with circular velocities @xmath146 orbiting our galaxy approximately a factor of five more than the number of satellites actually observed in the vicinity of the milky way or andromeda ( see 4 ) .
this discrepancy appears to be robust : effects ( numerical or physical ) would tend to produce more dark matter satellites , not less .
for example , dissipation in the baryonic component can only make the halos more stable and increase their chance to survive .
although the discrepancy between the observed and predicted satellite abundances appears to be dramatic , it is too early to conclude that it indicates a problem for hierarchical models .
several effects can explain the discrepancy and thus reconcile the predictions and observations .
if we discard the possibility that @xmath147 of the local group satellites have been missed in observations , we think that the discrepancy may be explained by ( 1 ) identification of the overabundant dm satellites with the high velocity clouds observed in the local group or by ( 2 ) physical processes such as supernovae - driven winds and gas heating by the ionizing background during the early stages of galaxy formation ( see 5 ) .
alternative ( 1 ) is attractive because the sizes , velocity dispersions , and abundance of the hvcs appear to be consistent with the properties of the overabundant low - mass halos .
these properties of the clouds are deduced under assumptions that they are located at large ( @xmath148 ) distances which should be testable in the near future with new upcoming surveys of the hvcs .
alternative ( 2 ) means that the halos of galaxies in the local group ( and other galaxies ) may contain substantial substructure in the form of numerous invisible clumps of dark matter .
this second possibility is interesting enough to merit further detailed study of the above effects on the evolution of gas in low - mass dark matter halos .
we are grateful to jon holtzman and david spergel for comments and discussions .
this work was funded by the nsf grant ast-9319970 , the nasa grant nag-5 - 3842 , and the nato grant crg 972148 to the nmsu .
our numerical simulations were done at the national center for supercomputing applications ( ncsa ; urbana - champaign , illinois ) . | according to the hierarchical clustering scenario , galaxies are assembled by merging and accretion of numerous satellites of different sizes and masses .
this ongoing process is not 100% efficient in destroying all of the accreted satellites , as evidenced by the satellites of our galaxy and of m31 . using published data ,
we have compiled the circular velocity ( @xmath0 ) distribution function ( vdf ) of galaxy satellites in the local group .
we find that within the volumes of radius of 570 kpc ( @xmath1 assuming the hubble constant . ]
@xmath2 ) centered on the milky way and andromeda , the average vdf is roughly approximated as @xmath3 for @xmath0 in the range @xmath4 .
the observed vdf is compared with results of high - resolution cosmological simulations .
we find that the vdf in models is very different from the observed one : @xmath5 .
cosmological models thus predict that a halo of the size of our galaxy should have about 50 dark matter satellites with circular velocity @xmath6 and mass @xmath7 within a 570 kpc radius .
this number is significantly higher than the approximate dozen satellites actually observed around our galaxy .
the difference is even larger if we consider abundance of satellites in simulated galaxy groups similar to the local group .
the models predict @xmath8 satellites inside a @xmath9 radius , while there only @xmath10 satellites are observed in the local group .
the observed and predicted vdfs cross at @xmath11 , indicating that the predicted abundance of satellites with @xmath12 is in reasonably good agreement with observations .
we conclude , therefore , that unless a large fraction of the local group satellites has been missed in observations , there is a dramatic discrepancy between observations and hierarchical models , regardless of the model parameters .
we discuss several possible explanations for this discrepancy including identification of some satellites with the high velocity clouds observed in the local group , and the existence of dark satellites that failed to accrete gas and form stars due either to the expulsion of gas in the supernovae - driven winds or to gas heating by the intergalactic ionizing background .
# 1#2.5pt ' # 1#1i"13i |
in the past decade , distributed cooperative control for multi - agent systems , particularly the consensus problem , has gained much attention and significant progress has been achieved , e.g. , @xcite@xcite .
almost all studies assume that the information can be continuously transmitted between agents with infinite precision . in practice , such an idealized assumption is often unrealistic , so information transmission should to be considered in the analysis and design of consensus protocols @xcite .
there are two main approaches to handle the communication limitation : event - triggered and quantized control . in
event - triggered ( and self - triggered ) control the control input is piecewise constant and transmission happens at discrete events @xcite@xcite . for instance , @xcite provided event - triggered and self - triggered protocols in both centralized and distributed formulations for multi - agent systems with undirected graph topology ; @xcite proposed a self - triggered protocol for multi - agent systems with switching topologies .
other authors considered systems with quantized sensor measurements and control inputs @xcite@xcite .
the authors of the papers @xcite@xcite combined event - triggered control with quantized communication . for example , @xcite considered model - based event - triggered control for systems with quantization and time - varying network delays ; @xcite presented decentralised event - triggered control in multi - agent systems with quantized communication . when considering event - triggered control in multi - agent systems with quantized communication or sensing , some aspects should be paid special attention to .
firstly , the notion of the solution should be clarified since in some cases the classic or hybrid solutions may not exist .
for instance , @xcite and @xcite used the concept of filippov solution when they considered quantized sensing .
secondly , the zeno behavior must be excluded @xcite .
thirdly , the need of continuous state access for neighbors should be avoided . in @xcite , which is a key motivation for the present paper
, the authors did not explicitly discuss the first aspect and used periodic sampling to exclude the zeno behavior .
they did not give any accurate upper bound of the sampling time , which restricts the application of the results .
inspired by @xcite and @xcite , we propose centralized and distributed self - triggered rules for multi - agent systems with quantized communication or sensing . under these rules ,
the existence of a unique trajectory of the system is guaranteed and the frequency of communication and system updating is reduced .
the main contribution of the paper is to show that the trajectory exponentially converges to practical consensus set .
it is shown that continuously monitoring of the triggering condition can also be avoided .
an important aspect of this paper is that the weakest fixed interaction topology is considered , namely , a directed graph containing a spanning tree .
the proposed self - triggered rules are easy to implement in the sense that triggering times of each agent are only related to its in - degree .
the rest of this paper is organized as follows : section [ sec2 ] introduces the preliminaries ; section [ sec3 ] discusses self - triggered consensus with quantized communication ; section [ sec4 ] treats instead self - triggered consensus with quantized sensing ; simulations are given in section [ sec5 ] ; and the paper is concluded in section [ sec6 ] .
in this section we will review some results on algebraic graph theory @xcite-@xcite and stochastic matrices @xcite-@xcite . for a matrix @xmath0
, the element at the @xmath1-th row and @xmath2-th column is denoted as @xmath3 ; and denote @xmath4)$ ] . for a _ ( weighted ) directed graph _ ( or digraph ) @xmath5 with @xmath6 agents ( vertices or nodes ) , the set of agents @xmath7 , set of links ( edges ) @xmath8 , and the _ ( weighted ) adjacency matrix _ @xmath9 with nonnegative adjacency elements @xmath3 .
a link of @xmath10 is denoted by @xmath11 if there is a directed link from agent @xmath12 to agent @xmath13 with weight @xmath14 , i.e. agent @xmath12 can send information to agent @xmath13 while the opposite direction transmission might not exist or with different weight @xmath15 .
it is assumed that @xmath16 for all @xmath17 , where @xmath18 .
let @xmath19 and @xmath20 denotes the in - neighbors and in - degree of agent @xmath13 , respectively .
the degree matrix of digraph @xmath10 is defined as @xmath21)$ ] .
the _ ( weighted ) laplacian matrix _ is defined as @xmath22 . a directed path from agent @xmath23 to agent @xmath24 is a directed graph with distinct agents @xmath25 and links @xmath26 .
we say a directed graph @xmath10 has a spanning tree if there exists at least one agent @xmath27 such that for any other agent @xmath28 , there exits a directed path from @xmath27 to @xmath28 .
obviously , there is a one - to - one correspondence between a graph and its adjacency matrix or its laplacian matrix . in the following , for the sake of simplicity in presentation ,
sometimes we do nt explicitly distinguish a graph from its adjacency matrix or laplacian matrix , i.e. , when we say a matrix has some graphic properties , we mean that these properties are held by the graph corresponding to this matrix .
a matrix @xmath9 is called a _ nonnegative matrix _ if @xmath29 for all @xmath30 , and @xmath31 is called a _
stochastic matrix _
if @xmath31 is square , nonnegative and @xmath32 for each @xmath1 .
a stochastic matrix @xmath31 is called _ scrambling _ if , for any @xmath1 and @xmath2 , there exists @xmath33 such that both @xmath34 and @xmath35 are positive .
moreover , given a nonnegative matrix @xmath31 and @xmath36 , the @xmath37-matrix of @xmath31 , which is denoted as @xmath38 , and its element at @xmath1-th row and @xmath2-th column , @xmath39 , is @xmath40 if @xmath38 has a spanning tree , we say @xmath31 contains a @xmath37-spanning tree . similarly ,
if @xmath38 is scrambling , we say @xmath31 is @xmath37-scrambling .
a nonnegative matrix @xmath31 is called a _ stochastic indecomposable and aperiodic _ ( sia )
matrix if it is a stochastic matrix and there exists a column vector @xmath41 such that @xmath42 , where @xmath43 is the @xmath6-vector containing only ones . for two @xmath6-dimension stochastic matrices @xmath31 and @xmath44 , they are said to be of the same _ type _ , denoted by @xmath45 , if they have zero elements and positive elements in the same places .
let @xmath46 denotes the number of different types of all sia matrices in @xmath47 , which is a finite number for given @xmath6 . for two matrices @xmath31 and
@xmath44 of the same dimension , we write @xmath48 if @xmath49 is a nonnegative matrix . throughout this paper
, we use @xmath50 to denote the left product of matrices . here , we introduce some lemmas that will be used later . from corollary 5.7 in @xcite , we have [ lem1 ] for a set of @xmath51 stochastic matrices @xmath52 , if there exists @xmath36 and @xmath53 such that @xmath54 and @xmath55 contains a @xmath56-spanning tree for all @xmath57 , then there exists @xmath58 , such that @xmath59 is @xmath60-scrambling
. from lemma 6 in @xcite , we have [ lem2 ] let @xmath61 be @xmath51 matrices with the property that for any @xmath62 , @xmath63 is sia , where @xmath36 is a constant , then @xmath64 is @xmath65-scrambling for any @xmath66 .
( @xcite ) for a real matrix @xmath67 , define the ergodicity coefficient @xmath68 and its hajnal diameter @xmath69 .
[ remark1 ] obviously , if @xmath31 is a stochastic matrix , then @xmath70 . moreover , if @xmath31 is @xmath37-scrambling for some @xmath36 , then @xmath71 .
[ lem3 ] ( @xcite ) if @xmath31 and @xmath44 are stochastic matrices , then @xmath72 .
[ lem4 ] ( @xcite ) for a vector @xmath73^{\top}\in\mathbb{r}^n$ ] , define @xmath74 . for an @xmath51 stochastic matrix @xmath31 , and @xmath75 , then @xmath76 .
it is straightforward to see that for any @xmath77 , @xmath78 .
we consider a set of @xmath6 agents that are modelled as a single integrator : @xmath79 where @xmath80 is the state and @xmath81 is the input of agent @xmath13 , respectively . in many practical scenarios ,
each agent can not access the state of the system with infinite precision .
instead , the state variables have to be quantized in order to be represented by a finite number of bits to be used in processor operations and to be transmitted over a digital communication channel . in this section
, each agent has a self - triggered control input based on the latest quantized states of its in - neighbours . denoting the _ triggering time sequence _ for agent @xmath12 as the increasing time sequence @xmath82
, the control input is given as @xmath83\label{input}\end{aligned}\ ] ] where @xmath84 , @xmath85 is a quantizer . in this paper
, we consider the following _ uniform _ quantizer : @xmath86 ) . in this subsection
, we consider centralized self - triggered control , i.e. , all agents simultaneously trigger at every triggering time . in this case
, the triggering time sequence can be denoted as @xmath87 . from ( [ system ] ) and ( [ input ] ) , we get : @xmath88\label{systemc}\end{aligned}\ ] ] where @xmath89^{\top}$ ] and @xmath90^{\top}$ ] for any @xmath91 . here
we give a rule to determine the triggering time sequence such that all agents converge to practical consensus .
[ thm1 ] assume the communication graph is directed , and contains a @xmath37-spanning tree with @xmath36 .
given the first triggering time @xmath92 , use the following self - triggered rule to find @xmath93 for known @xmath94 , choose an arbitrary @xmath95 , where @xmath96 , @xmath97 , @xmath98 and @xmath99 .
then the trajectory of ( [ systemc ] ) exponentially converges to the consensus set @xmath100 , where @xmath1014(1-\delta')$ ] and @xmath102 . from the self - triggered rule , for any given @xmath92 , the system can arbitrarily choose @xmath103 $ ] for every agent .
similarly , after @xmath94 has been chosen , the system can arbitrarily choose @xmath104 $ ] for every agent .
then , in the interval @xmath105 $ ] , the only solution to ( [ systemc ] ) is @xmath106 . particularly , we have @xmath107 where @xmath108 , @xmath109 and @xmath110 for any @xmath91 .
then @xmath111 obviously , for every @xmath33 , @xmath55 is a stochastic matrix ; @xmath55 has a @xmath112-spanning tree since @xmath113 has a @xmath37-spanning tree and @xmath114 ; @xmath115 since @xmath116_{ii}=1-\delta t_kl_{ii}\ge1-\delta t_kl_{max}\ge\delta'$ ] . then , from lemma [ lem1 ] , for any positive integer @xmath117 , we know that @xmath118 is @xmath60-scrambling for some @xmath119 .
then from remark [ remark1 ] , lemma [ lem3 ] and lemma [ lem4 ] , we have @xmath120 where @xmath121 .
thus @xmath122 for any @xmath123 , there exists a positive integer @xmath33 such that @xmath124 $ ] .
then , we have @xmath125x(t_k)-(t - t_k)lq(x(t_k))\end{aligned}\ ] ] thus , @xmath126 .
hence @xmath127 the proof is completed . in this subsection , we consider distributed self - triggered control .
in contrast to the centralized triggering where all agents trigger at the same time , each agent can now freely choose its own triggering times no matter when other agents trigger .
here we extend theorem [ thm1 ] to distributed such a distributed setup .
[ thm2 ] assume the communication graph is directed , and contains a @xmath37-spanning tree with @xmath36 .
for each agent @xmath13 , given the first triggering time @xmath128 , use the following self - triggered rule to find @xmath129 for known @xmath130 , choose an arbitrary @xmath131 , where @xmath132 , @xmath133 and @xmath134
. then the trajectory of ( [ system ] ) with input ( [ input ] ) exponentially converges to the consensus set @xmath135 , where @xmath136 is a positive constant which can be determined by @xmath137 .
\(a ) ( this proof is inspired by @xcite and @xcite . ) we say the system triggers at time @xmath138 if there exists at least one agent triggers at this time .
let @xmath139 denotes the system s triggering time sequence .
obviously , this is a strictly increasing sequence . for simplicity , denote @xmath140 and @xmath108 .
we first point out the following fact : [ lem5 ] for any agent @xmath13 and positive integer @xmath33 , the number of triggers occurred during @xmath141 $ ] is no more than @xmath142 , where @xmath143 and @xmath144 .
moreover , for any positive integer @xmath33 , every agent triggers at least once during @xmath145 $ ] , where @xmath146 .
the proof of this lemma can be found in @xcite .
let @xmath147^{\top}$ ] with @xmath148 , then , we can rewrite ( [ system ] ) and ( [ input ] ) as @xmath149\label{systemd}\end{aligned}\ ] ] now we consider the evolution of @xmath150 . if agent @xmath13 does not trigger at time @xmath151 , then @xmath152 .
thus @xmath153 if agent @xmath13 triggers at time @xmath151 , then @xmath154 .
assume @xmath155 be the last update of agent @xmath13 before @xmath151 , where integer @xmath156 is the number of triggers which are triggered by other agents between @xmath157 .
then , @xmath158 . noting @xmath159=\bigcup_{m = k - d_{ik}}^{k}(t_m , t_{m+1}]$ ] and ( [ systemd ] )
, we can conclude that there exists a unique solution to ( [ system ] ) .
then @xmath160 @xmath161 if agent @xmath13 triggers at time @xmath151 , then let @xmath162 , @xmath163 , @xmath164 for @xmath165 , @xmath166 for @xmath167 and @xmath168 , and @xmath169 for @xmath170 and @xmath171 .
otherwise , let @xmath169 for all @xmath172 except @xmath173 . obviously , @xmath174 where @xmath175 .
then we can uniformly rewrite ( [ solutiony1 ] ) and ( [ solutiony2 ] ) as @xmath176\label{solutiony3}\end{aligned}\ ] ] denote @xmath177^{\top}\in \mathbb{r}^{n(\tau_1 + 1)}$ ] , @xmath178 , @xmath179 , @xmath180%\in\mathbb{r}^{n(\tau_1 + 1)\times n(\tau_1 + 1)}\end{aligned}\ ] ] and @xmath181%\in\mathbb{r}^{n(\tau_1 + 1)\times n(\tau_1 + 1)}\end{aligned}\ ] ] from ( [ property1 ] ) and ( [ property2 ] ) , we know that @xmath182 is a stochastic matrix .
we can rewrite ( [ solutiony3 ] ) as @xmath183\end{aligned}\ ] ] \(b ) next , we will prove that there exists @xmath184 such that for any @xmath185 , @xmath186 is @xmath187 scrambling , where @xmath188 . from ( [ property1 ] ) and ( [ property2 ] ) , we know that @xmath189 is a nonnegative matrix for any @xmath190 and @xmath33 , and @xmath191 .
hence , @xmath192 .
denote @xmath193%\in\mathbb{r}^{n(\tau_1 + 1)\times n(\tau_1 + 1)}\end{aligned}\ ] ] and @xmath194 .
then , @xmath195 where @xmath196 . from lemma [ lem5 ] , we know that , for any @xmath33 , @xmath197 since each agent triggers at least once during @xmath145 $ ] .
hence , @xmath198 has a @xmath199-spanning tree .
thus , from lemma 7 and its proof in @xcite , we know that there exists @xmath200 such that @xmath201 is @xmath202-sia and has a @xmath202-spanning tree for any @xmath203 $ ] . here
we choose a @xmath204 such that @xmath205 . for any @xmath206 , note @xmath207^{[(\delta_f)^{\frac{1}{\tau_2}}]}\\
= & \prod_{i=(k_1 - 1)\tau_2 + 1}^{k_2\tau_2 } [ e(i)]^{[(\delta_f)^{\frac{1}{\tau_2}}]}\end{aligned}\ ] ] and the first block row sum of @xmath208^{(\delta_f)^{\frac{1}{\tau_2}}}$ ] has a spanning tree since @xmath209 . then from lemma 7 and its proof in @xcite , we know that @xmath210 is sia .
then , from lemma [ lem2 ] , we know that @xmath211 is @xmath212-scrambling .
hence , from ( [ ce ] ) , we can conclude that @xmath213 is @xmath214-scrambling , where @xmath215 .
\(c ) similar to the proof of theorem [ thm1 ] , we can find the @xmath136 and complete the proof or this theorem . in both theorems
[ thm1 ] and [ thm2 ] , the evolutions of @xmath216 obey @xmath217 , where @xmath218 .
there is no zeno behavior in the centralized and distributed self - triggered systems .
note that the triggering times are not dependent on the state , but the triggering rules are related only to the degree matrix .
in this section , we consider the situation that , each agent @xmath13 discretely sense or measures the quantized value of the relative positions between its in - neighbors and itself .
in other words , the only available information to compute the control inputs of each agent are the latest quantized measurements of the relative positions measured by itself : @xmath219 compared to ( [ input ] ) , the advantage of ( [ inputm ] ) is that the input is not affected by other agents triggering . in this subsection , we consider centralized self - triggered consensus rule and denote the triggering time sequence as @xmath87 .
then , we get @xmath220\label{inputmc}\end{aligned}\ ] ] similar to theorem [ thm1 ] , we have the following result . [ thm3 ] under the assumptions and self - triggered rule of theorem [ thm1 ]
, the trajectory of system ( [ system ] ) with input ( [ inputmc ] ) exponentially converges to the consensus set @xmath221 , where @xmath222 is a positive constant which can be determined by @xmath56 and @xmath37 . from the self - triggered rule in theorem [ thm1 ] , for any given @xmath92 , the system can arbitrarily choose @xmath103 $ ] for every agent .
similarly , after @xmath94 has been chosen , the system can arbitrarily choose @xmath104 $ ] for every agent .
then , in the interval @xmath105 $ ] , the only solution to ( [ system ] ) with input ( [ inputmc ] ) is @xmath223 particularly , we have @xmath224 then , @xmath225 , where @xmath226^{\top}$ ] and @xmath227 $ ] . the proof follows similarly to the proof to theorem [ thm1 ] . in this subsection , we consider distributed self - triggered consensus rule .
similar to theorem [ thm2 ] , we have [ thm4 ] under the assumptions and self - triggered rule of theorem [ thm2 ] , the trajectory of ( [ system ] ) with input ( [ inputm ] ) exponentially converges to the consensus set @xmath228 , where @xmath229 is a positive constant which can be determined by @xmath137 .
we omit the proof since it is similar to the proof of theorem [ thm2 ] .
in this section , a numerical example is given to demonstrate the effectiveness of the presented results .
consider a network of seven agents with a directed reducible laplacian matrix @xmath230\end{aligned}\ ] ] which is described by the graph in fig .
[ fig:1 ] .
the initial value of each agent is randomly selected within the interval @xmath231 $ ] in our simulation and the next triggering time is randomly chosen from the permissible range using a uniform distribution .
the uniform quantizing function used here is @xmath232 if @xmath233 .
[ fig:2 ] shows the evolution of @xmath234 under the four self - triggered rules treated in theorems [ thm1]-[thm4 ] with @xmath235 and @xmath236 . in this simulation
, it can be seen that under all self - triggering rules all agents converge to the consensus set with @xmath237 . .
the dots indicate the triggering times of each agent.,scaledwidth=45.0% ] let the quantizer parameter @xmath238 take different values .
[ fig:3 ] illustrates @xmath239 under the four self - triggering rules for different @xmath238 .
the curves show the averages over 100 overlaps .
as expected , the smaller @xmath238 , the smaller is the consensus set . with different @xmath238.,scaledwidth=45.0% ]
in this paper , consensus problems for multi - agent systems defined on directed graphs under self - triggered control have been addressed . in order to reduce the overall need of communication and system updates , centralized and distributed self - triggered rules have been proposed in the situation that quantized information can only be transmitted , i.e. , quantized communication , and the situation that each agent can sense only quantized value of the relative positions between neighbors , i.e. , quantized sensing .
it has been shown that the trajectory of each agent exponentially converges to the consensus set if the directed graph containing a spanning tree .
the triggering rules can be easily implemented since they are related only to the degree matrix .
interesting future directions include considering stochastically switching topologies and more precise expression of the consensus sets .
f. xiao and l. wang , `` asynchronous consensus in continuous - time multi - agent systems with switching topology and time - varying delays , '' _ automatic control , ieee transactions on _ , vol .
53 , no . 8 , pp .
1804 - 1816 , 2008 .
k. you and l. xie , `` network topology and communication data rate for consensusability of discrete - time multi - agent systems , '' _ automatic control , ieee transactions on _ , vol .
2262 - 2275 , 2011 . x. l.
yi , w. l. lu and t. p. chen , `` pull - based distributed event - triggered consensus for multi - agent systems with directed topologies , '' _ neural networks and learning systems , ieee transactions on _ , to be appeared .
d. v. dimarogonas and k. h. johansson , `` stability analysis for multi - agent systems using the incidence matrix : quantized communication and formation control , '' _ automatica _ , vol .
4 , pp . 695 - 700 , 2010 .
h. yu and p. j. antsaklis , `` event - triggered output feedback control for networked control systems using passivity : achieving @xmath240 stability in the presence of communication delays and signal quantization , '' _ automatica _ , vol .
1 , pp . 30 - 38 , 2013 .
e. garcia and p. j. antsaklis , `` model - based event - triggered control for systems with quantization and time - varying network delays , '' _ automatic control , ieee transactions on _ , vol .
2 , pp . 422 - 434 , 2013 .
e. garcia , y. c. cao , h. yu , p. j. antsaklis and d. casbeer , `` decentralised event - triggered cooperative control with limited communication , '' _ international journal of control _ , vol .
9 , pp . 1479 - 1488 , 2013 .
b. liu , w. l. lu and t. p. chen , `` consensus in networks of multiagents with switching topologies modeled as adapted stochastic processes , '' _ siam journal on control and optimization _ , vol .
227 - 253 , 2011 . | the consensus problem for multi - agent systems with quantized communication or sensing is considered .
centralized and distributed self - triggered rules are proposed to reduce the overall need of communication and system updates .
it is proved that these self - triggered rules realize consensus exponentially if the network topologies have a spanning tree and the quantization function is uniform .
numerical simulations are provided to show the effectiveness of the theoretical results . |
the use of symmetries is an important tool to understand the effects of strong correlation in electronic systems .
recently , the @xmath1-symmetry of the antiferromagnetic ( afm ) order parameter has been combined with that of @xmath2-wave superconductivity to form a five - component vector order parameter @xcite .
it has been argued that the low energy sector of the resulting theory exhibits an approximate @xmath0-symmetry which allows to explain certain features such as the vicinity afm order and superconductivity in the phase diagram of the high-@xmath3 materials .
numerical diagonalization studies have been performed and the spectrum of low lying excitations could in fact be classified according to this symmetry .
a complementary approach has been the attempt to construct microscopic electronic systems with manifest @xmath0 invariance and studies of such models to extract the low energy behaviour .
scalapino , zhang and hanke succeeded in constructing a two - chain ladder hamiltonian of this type and studied the strong coupling phase diagram of this system where they were able to identify several distinct phases ( ref . , referred to as szh in the following ) .
the properties of these systems at weak coupling in the metallic regime has been studied by means of bosonization @xcite .
such ladder systems , in particular for magnetic insulators have attracted much attention recently due to the existence of various experimental realizations in materials closely related to the high-@xmath3 substances @xcite .
an interesting observation of ref .
is the existence of an @xmath0 superspin phase which has been studied in a variational approach based on finitely correlated matrix product states similar to the ones used for @xmath4 haldane magnets @xcite .
finitely correlated states have also been considered in electronic systems to describe aspects of the phase diagram of extended hubbard models @xcite and other one - dimensional electronic models @xcite . for @xmath5 spin systems
the variational approach has been generalized to lattices with ladder geometry and proven to give access to large parts of their phase diagram @xcite .
this is the motivation for the present work where we extend the matrix product states originally introduced in ref . to describe the strong coupling physics of the @xmath0 superspin phase .
we construct manifestly @xmath0 invariant many particle wave functions from matrices containing _ all _ 16 electronic states on a given rung of the electronic ladder .
the relative weight of the six different @xmath0 multiplets on a rung is controlled by free parameters which are used to perform a variational study of the zero temperature phase diagram of the ladder at half filling . at strong coupling the results known from ref .
are reproduced within our approach .
furthermore , at weak coupling and sufficiently large interchain hopping amplitude @xmath6 the matrix product state correctly describes the gapped ground state of a band insulator corresponding to a filled fermi sea of electrons with one parity . for intermediate coupling
we find a phase with finite amplitude of the @xmath0 spinor quartets which are essential for the presence of a metallic phase of the ladder .
the compact form of the variational states allows to study various correlation functions of interest . in the following section we present the classification of the electronic states of a two - leg ladder system according to the @xmath0-symmetry and discuss all possible @xmath0 symmetric single rung interactions .
in section [ sec : szh ] we review the szh - model and consider tensor products of rung states to include couplings of neighboring rungs .
section [ sec : ext ] deals with various @xmath0 symmetric extensions of this model and a general construction routine for systems with exact finitely correlated ground states is given .
section [ sec : var ] contains a detailed analysis of the ground state phase diagram of the system in the case of weak and intermediate coupling within a variational approach based on such wave functions .
furthermore we calculate the corresponding correlation functions within this approach .
a summary of our results is given in section [ sec : sum ] .
we consider a two - chain electronic ladder model with canonical creation and annihilation operators @xmath7 , @xmath8 for electrons ( with spin - projection @xmath9 ) on sites @xmath10 of the upper leg and analogous operators @xmath11 for the electrons on the lower leg . in order to discuss the @xmath0 symmetry of the ladder model and to classify all the @xmath12 possible states on a rung according to this symmetry ,
these operators are combined into four - dimensional @xmath0 spinors @xcite @xmath13 and @xmath14 using this definition the ten local generators @xmath15 of the @xmath0-algebra on a single rung @xmath10 are defined as @xmath16 here @xmath17 are ten antisymmetric , @xmath18 matrices ( their explicit form is given in appendix [ app : gam ] ) .
a convenient basis of the hilbert space on a single rung is diagonal in the quadratic casimir charge @xmath19 in addition we choose to diagonalize the total charge @xmath20 and the z - component of the spin @xmath21 .
based on the eigenvalues of @xmath22 the hilbert space can be decomposed into six @xmath0 multiplets : * three @xmath0 singlets ( @xmath23 ) , for @xmath24 see ( [ rmat ] ) @xmath25 * an @xmath0 vector quintet ( @xmath26 ) containing the ferromagnetically polarized state at half filling @xmath27 * two @xmath0 spinor quartets ( @xmath28 ) for an odd number of electrons on a given rung @xmath29 we label the states @xmath30 on a rung by the dimension @xmath2 of the corresponding multiplet ( @xmath31 ) and an additional index @xmath32 .
similarly , we can characterize product states on two rungs ( see sect .
[ sec : szh ] ) .
alternatively , the vector quintet ( [ quin1 ] ) can be constructed from @xmath0 spinors @xmath33 and @xmath34 again , the explicit form of the @xmath35 dirac @xmath36-matrices @xmath37 is given in appendix [ app : gam ] .
any electronic ladder model with a local @xmath0-symmetry on a rung has to preserve the degeneracy of the energy within the states of each single multiplet .
the invariant hamiltonian on a single rung can therefore be written as a sum over projection operators on these states : @xmath38 where @xmath39 because of the hermiticity of @xmath40 .
all @xmath0-symmetric terms on a rung can be expressed using linear combinations of these projection operators , e.g. , the projection operator on the first singlet @xmath41 is @xmath42 with @xmath43 and @xmath44
. a complete classification of these terms is given in appendix [ app : ops ] .
as a simple example we choose @xmath45 which leads to the hubbard - type hamiltonian@xcite with an @xmath0-symmetry introduced by szh @xmath46 \label{szh - hamrung}\end{aligned}\ ] ] where @xmath47 .
this condition on the exchange amplitude guarantees the degeneracy between the states in the @xmath0-quintet and therefore the local @xmath0-symmetry of the system .
we will discuss this model and @xmath0 symmetric extensions in the following sections .
in order to describe an extended quasi - one dimensional electronic system one has to include coupling of neighboring rungs in addition to single rung interactions considered in the previous section . the simplest possible term is an @xmath0 symmetric hopping term between adjacent rungs @xmath48 , \label{h - tpar}\end{aligned}\ ] ] which can be brought into a manifestly @xmath0-symmetric form using the alternating definitions of the spinors ( [ spinor1],[spinor2 ] ) . this hopping term together with the local rung interactions ( [ szh - ham ] ) yield the complete szh - model @xcite .
the ground state phase diagram of this system in the limit of strong coupling ( @xmath49 ) has been determined by szh using perturbation theory ( see fig .
[ fig2 ] ) .
four different phases have been established at half - filling : + in phase @xmath50 ( occuring for @xmath51 ) the model can be mapped onto an ising - like system in a magnetic field : phase @xmath52 ( @xmath53 ) is a cdw phase and @xmath54 ( @xmath55 ) corresponds to the disordered ising phase .
phase @xmath56 is a spin - gap @xmath2-wave phase ( product of rung singlets ) , emerging for @xmath57 and for @xmath58 .
the phase @xmath59 ( @xmath60 ) is the superspin - phase where the @xmath0-quintet is dominant . for a further examination of this superspin phase , szh
have used the finitely correlated wave function @xmath61 ( summation over the index @xmath62 is implied and the trace is taken in the @xmath63 matrix space where the @xmath37 are defined ) . in this form
periodic boundary conditions have been imposed . by
adding many particle interactions to their original hubbard - type hamiltonian this state ( [ szhans ] ) can be made to be the exact ground state of the resulting model .
this state has been argued to capture the essential physics of the superspin phase similar to the rle of the aklt - model as a representative for a haldane - gapped spin-1 chain .
the wave function ( [ szhans ] ) will be the starting point for constructing a generalized matrix product wave function including all 16 states on a rung ( see section [ sec : ext ] ) and later be used for a variational study of the ground state phase diagram of the szh model and its various @xmath0 symmetric extensions beyond strong coupling ( see section [ sec : var ] ) . the hopping term ( [ h - tpar ] )
is one of many possibilities to include interactions between two adjacent rungs of the ladder but the requirement for a local @xmath0-symmetry puts constraints on the explicit form of these terms .
explicit expressions for some of the interaction terms are listed in terms of electron operators in appendix [ app : ops ] . for a classification of these additional interactions
we consider products of wave functions on two neighboring rungs @xmath10 and @xmath64 .
a decomposition into @xmath0-multiplets similar to ( [ sing1 ] [ quar1 ] ) gives 50 different multiplets invariant under the action of the @xmath0 generators @xmath65 .
tensor products containing a singlet factor on one of the rungs are trivial leading to simple product states , e.g. the @xmath0 singlets @xmath66 .
altogether there are nine singlets , 12 quartets and six quintets of this form .
the remaining 169 states are obtained by forming tensor products of quartets ( [ quar1 ] ) and quintets ( [ quin1 ] ) .
the decomposition of these products into irreducible representations of @xmath0 reads @xmath67 ( numbers denote the dimension of the corresponding @xmath0 irrep ) .
for example , one of four @xmath0-singlets in the tensor product of quartet states ( [ quar1 ] ) is @xmath68 similar combinations of the rung states appear in the other states , the casimir charges of the new multiplets are @xmath69 for the decuplets , @xmath70 for the 14-dimensional and @xmath71 for the 16-dimensional representations .
the multiplets can be classified further according to the different eigenvalues of @xmath72 and @xmath73 on their member states . in fig .
[ fig1 ] the state content of the various multiplets is shown . in the following we use this classification of the @xmath0-multiplets to construct ladder systems with exact ground states including different @xmath0 symmetric nearest neighbor interactions .
as mentioned in the introduction the finitely correlated wave functions originally introduced to discuss the spin - liquid phases arising in one - dimensional higher spin heisenberg models @xcite have recently been generalized to more general lattices . in particular ,
ladder models whose ground states are of this form have been constructed @xcite . in these spin systems
the ground state is of the form @xmath74 where @xmath75 is a ( @xmath76 ) matrix containing the different states on a single site or rung @xmath10 ( e.g. spin-1 states for the aklt model , singlet and triplet states for a two - leg @xmath77 ladder , etc . ) .
different properties under translation in the extended direction can be realized by an appropriate choice of the free parameters appearing in @xmath75 ( e.g. an alternation to introduce dimerization @xcite ) . within a transfer matrix approach
it is straightforward to compute various ground state correlation functions for different boundary conditions , periodic ones correspond to taking the trace of the matrix product wave function @xcite . for a further analysis of the szh - model and the construction of @xmath0-symmetric ladder systems with exact ground states in matrix product form
we have extended the wave function ( [ szhans ] ) to include the three @xmath0-singlets ( [ sing1 ] ) and the two @xmath0 spinor quartets ( [ quar1 ] ) @xmath78 now @xmath75 is a @xmath79-matrix and @xmath80 are variational parameters assigning different weights to the multiplets ( [ sing1 ] ) ( [ quar1 ] ) on a rung ( see appendix [ app : gam ] ) . we restrict ourselves to the translational invariant case , where the parameters @xmath81 are chosen to be independent of the rung position @xmath10 . in this case
the matrix product wave function on two neighboring rungs contains two @xmath0 singlets , two quartets , one quintet and one decuplet .
the 14-dimensional and 16-dimensional representations are absent by construction .
the states of the matrix product are linear combinations of the basis in sect .
[ sec : szh ] above , their explicit form is rather complicated .
with respect to the spin-@xmath5 subalgebra the remaining multiplets present in the matrix product contain spin singlet , doublet and triplet states only ( states with total spin polarization @xmath82 are members of the 14 and 16-dimensional representation , see fig .
[ fig1 ] ) .
an immediate consequence is that the ansatz can not be expected to describe the formation of ferromagnetic domains with higher spin states .
an analogous argument holds for higher values of the charge @xmath72 , corresponding to strong local deviations from half filling .
there is a simple way to construct spin ladder systems with matrix product wave functions as ground states @xcite and a generalization to electronic ladder models with an @xmath0-symmetry is straightforward .
the starting point is a general @xmath0-symmetric hamilton operator on two neighboring rungs @xmath83 where @xmath43 are projection operators on all possible @xmath0-multiplets ( see section [ sec : szh ] ) .
the states @xmath84 are product wave functions on two rungs , @xmath32 and @xmath85 label the multiplet , @xmath86 the states in the multiplet and @xmath2 is the corresponding dimension of this irreducible representation .
the hermiticity of @xmath87 requires @xmath88 for the coupling constants , leaving altogether 322 free parameters in the hamiltonian .
a hamiltonian @xmath89 has a finitely correlated ground state @xmath90 with zero energy provided that the following conditions are satisfied @xcite : * @xmath87 has to annihilate all states contained in the matrix elements of the product @xmath91 * all other eigenstates of @xmath87 have positive energy starting with an ansatz for @xmath75 in ( [ ansatz ] ) , one has to identify all multiplets @xmath92 contained in the product wave function @xmath91 .
these multiplets are labelled by indices @xmath93 where the maximum number @xmath94 is the number of multiplets with an equal casimir - charge ( @xmath2 is the dimension of the irreducible representation ) , e.g. @xmath95 if the product wave function on two neighboring rungs contains two independent @xmath0 decuplets . after the determination of the multiplet content of @xmath96
the corresponding parameters @xmath97 in @xmath87 are set to zero to fulfill the first condition .
the remaining operators in ( [ ham - op ] ) will now project on states not included in the matrix product wave function , which leads to zero energy for the ansatz . to satisfy now the second condition the reduced matrices @xmath98 ( @xmath99 ) have to be chosen positive definite ( i.e. positive eigenvalues ) such that ( [ ansatz ] ) will be the lowest energy state of the system . in principle
a general hamiltonian where our ansatz ( [ ansatz ] ) is the exact ground state , can be built by operators projecting on the other remaining @xmath0-multiplets ( e.g. the 14-dimensional and the four 16-dimensional representations ) and it has 249 free coupling constants @xmath98 ( @xmath99 ) .
we give explicit expressions for some of these operators in terms of electron operators in appendix [ app : ops : gs ] .
in general however , the structure of these projection operators is quite complicated making it difficult to motivate these exactly solvable systems on physical grounds .
an examination of the szh model beyond strong coupling can be done by using ( [ ansatz ] ) as a variational wave function .
this wave function leads to the variational energy @xmath100 for the spin and charge interaction on a rung ( see equation [ szh - hamrung ] ) . here
@xmath101 is the parameter corresponding to the @xmath0-quintet , @xmath102 is the weight of the singlet @xmath103 and @xmath104 of the symmetric and antisymmetric linear combinations of the other @xmath0-singlets ( [ sing1 ] ) .
the variational energy corresponding to the hopping term on a rung is @xmath105 \label{hopp - rung}\end{aligned}\ ] ] and between two neighboring rungs ( see equation [ h - tpar ] ) it is @xmath106 .
\label{hopp - bet}\end{aligned}\ ] ] where @xmath107 , @xmath108 and @xmath109 . in the strong coupling limit
one can neglect these hopping terms ( [ hopp - rung ] , [ hopp - bet ] ) .
minimizing the energy with respect to the @xmath81 reproduces exactly the phase diagram calculated by szh within perturbation theory ( fig .
[ fig2 ] ) where the phases are fixed by the largest amplitude @xmath81 of the corresponding state and the crossover is continuous .
the phase @xmath50 is dominated by the bonding singlet state with amplitude @xmath110 , phase @xmath59 is the superspin phase ( @xmath101 ) and phase @xmath56 consists of products of rung singlets ( @xmath102 ) . in this approach with translationally
invariant @xmath81 the crossover between the two ising - phases can not be reproduced .
we now extend this analysis of the phase diagram of the szh model to weak and intermediate coupling .
the band structure of the non - interacting system at half - filling is well known .
for @xmath111 there are two energy bands , given by @xmath112 and two different cases have to be distinguished : for @xmath113 the fermi energy intersects the two bands ( see fig .
[ fig3].a ) and for @xmath114 they are separated by an energy gap ( see fig . [
fig3].b ) . the gapless system ( @xmath115 ) has been studied using bosonization of the low lying modes in the vicinity of the four fermi points to obtain the phase diagram for weak coupling ( @xmath116 ) : lin et al .
@xcite found that at half filling the system is driven towards an integrable so(8)-symmetric gross - neveu model in a weak coupling renormalization group analysis and predicted the occurrence of additional phases compared to the strong coupling case .
the ongoing debate of these results ( see the criticism of ref . ) can not be clarified within the present ansatz : using finitely correlated wave functions always leads to an exponential decay of correlation functions , indicating the existence of an energy gap between the ground state and the first excited state .
for @xmath118 the variational ansatz gives the exact ground state for the non - interacting system ( @xmath119 ) : choosing @xmath120 and @xmath121 for @xmath122 we find @xmath123 which corresponds to complete filling of the modes with energy @xmath124 in ( [ disp0 ] ) , the band insulator .
consequently , we expect the variational approach to give reasonable results for the weak coupling phase diagram in this regime of hopping amplitudes .
the quality of the approach can be measured by the mean deviation @xmath125 of the energy .
for @xmath116 the mean deviation stays small compared to the energy so that the ansatz should give reliable results .
we find that only two phases are present in the weak coupling case ( see fig .
[ fig4 ] ) : the ising phase @xmath50 ( @xmath110 ) and the spin - gap d - wave phase @xmath56 ( @xmath102 ) already known from the strong coupling diagram ( see fig . [ fig2 ] ) .
the superspin phase disappears and also the @xmath0-quartets have no significant weight ( @xmath126 ) as expected for a band insulator .
considering the complete ground state phase diagram ( see fig . [ fig4 ] ) we find an additional phase for intermediate coupling ( @xmath127 ) where the @xmath0-quartets have the largest weight , in particular the rung - symmetric one @xmath128 ( [ eq : quar ] ) .
apart from these , the symmetric singlet state @xmath129 which determines the ground state in phase @xmath50 has a significant weight . due to the resonating structure of the ansatz and the relatively large variational value of @xmath130
the phase boundaries are not very accurate for a more detailed study of this question the present work should be complemented by a numerical approach . as discussed earlier
it is not possible within this approach to determine the position of the crossover line between the two ising phases , or even whether this transition still occurs for the case of weak or intermediate coupling .
the physics in the ground state is determined by ground state correlation functions , which are easily computed from matrix product wave functions .
the matrix product ansatz ( [ ansatz ] ) with the six free parameters @xmath81 represents the ground state for a large class of models .
we have calculated various correlation functions explicitly in the thermodynamic limit ( @xmath131 ) for this variety of models ( a detailed list is given in appendix [ app : cor ] ) and we determined the correlation length and the amplitude of different ground state correlations for the szh - model when we used the ansatz as a variational wave function .
the two - point correlations in matrix product states are always short - ranged ( if not vanishing ) and have the following form @xmath132 they exhibit an exponential decay with the correlation length @xmath133 and amplitude @xmath134 .
as an example we consider the correlation length and amplitude of the expectation value of the spin - spin correlation function @xmath135 ( see fig .
[ fig5 ] ) and of field correlators @xmath136 , @xmath137 ( fig . [ fig6 ] ) for the szh - model on a circle in the @xmath138-@xmath139-plane ( with @xmath140 ) intersecting the phases @xmath50 , @xmath56 and the quartet phase ( see fig . [ fig4 ] ) .
the spin - spin correlation function @xmath141 in fig .
[ fig5 ] is non - vanishing only in the quartet phase but with an extremely small correlation length indicating strong nearest neighbor correlations .
for the electron - electron correlation in fig .
[ fig6 ] with @xmath142 the correlation length @xmath133 is small for all angles @xmath143 but with a very large amplitude @xmath144 except in the quartet phase .
the sharp peak in both diagrams at @xmath145 indicate the crossover of the phases @xmath56 and @xmath50 in fig .
[ fig4 ] where the correlation length diverges . calculating these correlations in the strong coupling limit the phase boundaries in fig .
[ fig4 ] are denoted by very sharp peaks in the electron - electron correlation length @xmath146 with a non - vanishing amplitude .
the spin - spin correlations are zero in the whole phase diagram and give no further hints of an underlying structure in the system .
our variational approach is also suitable to study the phase diagrams of various @xmath0-symmetric extensions of the szh model .
we have considered additional interactions on a single rung and between two neighboring rungs , using the construction routine of section [ sec : states ] and [ sec : ext ] .
all single rung interactions can be constructed using the projection operators of sect .
[ sec : states ] and a detailed list of all possible terms can be found in the appendix [ app : ops ] .
taking into account the operators @xmath147 with the coupling constants @xmath98 leads to the following contributions to the variational energy ( [ erung ] ) , calculated with the ansatz ( [ ansatz ] ) @xmath148 these modifications of the model cause some changes in the ground state phase diagram , e.g. the simple terms like ( a ) and ( j ) will only shift the phase boundaries without changing the general structure of the phase diagram .
other interactions like the pair hopping term ( f ) @xmath149 will dramatically change the phase diagram ( see fig .
[ fig7 ] ) . for small negative values of @xmath150 ( @xmath151 ) the ising - phase @xmath50 of the phase diagram in fig .
[ fig4 ] with the amplitude @xmath110 splits into two singlet phases : a symmetric ( @xmath110 ) and an antisymmetric phase ( @xmath152 ) ( see fig .
[ fig7].a , @xmath153 ) , where the crossover line has the same gradient ( @xmath154 ) .
increasing the amplitude of @xmath150 leads to a pure antisymmetric phase ( see fig . [ fig7].b , @xmath155 ) in @xmath50 and also to a strong change of the shape of the quartet phase . for small positive values of @xmath150 the general structure of the phase diagram
is preserved ( like fig .
[ fig4 ] ) . in the regime of the coupling constants with @xmath156
the quartet phase vanishes ( see fig . [ fig8 ] ) .
other interactions also exhibit strong effects on the phase diagram , e.g. including a quartet term like ( i ) which contains a hopping term on a rung and a bond - charge interaction ( see app .
[ app : ops ] ) . @xmath157 \label{ext2}\end{aligned}\ ] ] leads to different phase diagrams , depending on the coupling constant . for positive values of @xmath158
the quartet phase vanishes with increasing values of the coupling constant ( like in fig .
[ fig8 ] ) until there are only the three known phases ( see fig .
[ fig9].a ) .
for @xmath159 the quartet phase grows ( see fig . [ fig9].b ) , dominated by the symmetric combination ( @xmath160 ) of the states .
the mean deviation in the weak coupling limit in these two special cases ( [ ext1],[ext2 ] ) is small compared to the energy ( calculated on a circle with radius @xmath161 around @xmath119 ) except for the value @xmath162 .
the ansatz also provides very good results in the strong coupling limit ( @xmath163 ) , except for the crossover lines to the superspin phase where the mean deviation is very large .
the same problem occurs in the intermediate coupling regime in the quartet phase where the ansatz is not a good eigenstate of the system .
we expect that including the other interactions on a rung will lead to similar changes in the ground state phase diagram . in most cases
the @xmath0-symmetric interactions between two neighboring terms have a very complex structure but for some of them we can give simple expressions ( see appendix [ app : ops ] ) . for them
we can calculate the corresponding variational energy , e.g. the two - pair hopping term leading to an @xmath0 singlet - singlet transition @xmath164\end{aligned}\ ] ] giving @xmath165 .
the other @xmath0-singlet interactions on two rungs lead to similar expressions , which will change the phase diagram ( fig .
[ fig4 ] ) in the ising phase according to the value of the coupling constant @xmath166 . including an @xmath0-quartet interaction in the szh - model
( see eqn .
( [ sing - quar ] ) ) gives the variational energy @xmath167 the phase diagrams obtained for different values of the coupling constant @xmath168 are very similar to the phase diagram in fig .
the additional interaction has no significant effect except for minor changes of the crossover lines .
we have constructed a large class of electronic ladder models with @xmath0 symmetry having finitely correlated ground states and , consequently , correlation functions exhibiting exponential decay .
these matrix product states have been used to perform a variational study of the ground state phase diagram of the szh model @xcite for @xmath118 .
for vanishing coupling the ground state of the band insulator is found to be in the class of variational states and at strong coupling the phases identified by szh are reproduced . in the intermediate coupling regime signatures of a new phase dominated by local @xmath0-quartets are found , and at weak coupling the @xmath0 superspin phase is absent . within our approach
it is possible to compute various correlations giving further insights into the nature of the phases which have been identified .
finally , we have introduced various @xmath0 symmetric extensions to the szh model and discussed their impact on the phase diagram . in the future
we will include dimerization in the matrix product ansatz for further studies of the ising transition@xcite in phase i and the possibility of spontaneous breaking of translational invariance in exactly solvable models .
this work has been supported in parts by the deutsche forschungsgemeinschaft under grant no .
fr 737/31 .
for the construction of the @xmath0-invariant quantities , we have used the representation of the matrices in ref . .
the five dirac @xmath36-matrices have the following form @xmath169 where @xmath170 are the pauli matrices .
the matrices @xmath17 are defined by @xmath171 $ ] and the matrix @xmath24 is given by @xmath172 using these definitions , it is simple to construct the matrix @xmath75 of the variational wave function ( [ ansatz ] ) .
it has the following structure @xmath173 the three @xmath0-singlets are included in this ansatz only on the main diagonal elements .
@xmath174 from ( [ sing1 ] ) and @xmath175 are the symmetric and an antisymmetric combinations of the two other singlets @xmath176 the quartets enter the matrix @xmath75 in @xmath177 are the symmetric and antisymmetric combinations of ( [ quar1 ] ) @xmath178 they are arranged in the right column and the lowest row of ( [ eq : gx ] ) in such a way that in the product @xmath91 one has @xmath0-singlets on the diagonal only .
we present here a selection of various @xmath0-symmetric terms on a single and on two rungs ( [ app : ops : si],[app : ops : two ] ) .
furthermore , a list of terms is given for which our matrix product ansatz ( [ ansatz ] ) would be the lowest energy state ( [ app : ops : gs ] ) .
we now present all possible @xmath0-symmetric terms on a rung .
their general construction is done in terms of projection operators on the different @xmath0-multiplets . expressed through electronic operators most of them
are already known from the szh - model ( [ szh - ham ] ) and an additional biquadratic exchange . as a shorthand notation
we introduce @xmath179 & \equiv & \ ; u \ ; \left ( ( n_{c\uparrow}(x)-\frac{1}{2 } ) ( n_{c\downarrow}(x)-\frac{1}{2 } ) + ( c \rightarrow d ) \right ) \nonumber\\ & & \quad + \;v \ ; ( n_c(x)-1)(n_d(x)-1 ) + \ ; j \ ; \vec{s_c}(x ) \vec{s_d}(x ) \\ & & \quad + \alpha ( \vec{s_c}(x)\vec{s_d}(x))^2 \nonumber\end{aligned}\ ] ] in addition we find various single electron and pair hopping terms together with bond - charge type interactions .
using the notation of sect .
[ sec : states ] for the projection operators on a rung we obtain the following terms by projection on the singlets @xmath180 \\
\hat{p}_{0,0}^{2,2}&= & \left[\frac{1}{2},-\frac{1}{4},\frac{2}{3},\frac{4}{3}\right ] + \frac{1}{2 } \left [ n_{d\uparrow}n_{d\downarrow } ( 1-n_c ) - c \leftrightarrow d \right ] \\ \hat{p}_{0,0}^{3,3}&=&\left[\frac{1}{2},-\frac{1}{4},\frac{2}{3},\frac{4}{3 } \right]-\frac{1}{2 } \left [ n_{d\uparrow}n_{d\downarrow } ( 1-n_c ) - c \leftrightarrow d \right ] \\ \hat{p}_{0,0}^{1,2}&+&\hat{p}_{0,0}^{2,1}=\frac{1}{\sqrt{2 } } \left [ ( c_{\uparrow}^{\dagger}d_{\uparrow}^{}+h.c . )
n_{d\downarrow } ( n_{c\downarrow}-1 ) + \uparrow \leftrightarrow \downarrow \right ] \\ \hat{p}_{0,0}^{1,3}&+&\hat{p}_{0,0}^{3,1}=-\frac{1}{\sqrt{2 } } \left [ ( c_{\uparrow}^{\dagger}d_{\uparrow}^{}+h.c . ) n_{c\downarrow } ( n_{d\downarrow}-1 ) + \uparrow \leftrightarrow \downarrow \right ] \\ \hat{p}_{0,0}^{2,3}&+&\hat{p}_{0,0}^{3,2}= d_{\uparrow}^{\dagger}d_{\downarrow}^{\dagger}c_{\uparrow}c_{\downarrow } + h.c . \
.\end{aligned}\ ] ] the projection operators on the quartet states read @xmath181 + ( 1-n_{c\uparrow}n_{c\downarrow})n_d + n_{d\uparrow}n_{d\downarrow } ( n_c-2 ) \\
\sum_{\mu=1}^4 \hat{p}_{4,\mu}^{2,2}&= & \left[0,0,-\frac{8}{3},-\frac{16}{3}\right ] + ( 1-n_{d\uparrow}n_{d\downarrow})n_c + n_{c\uparrow}n_{c\downarrow } ( n_d-2 ) \\ \sum_{\mu=1}^4 \hat{p}_{4,\mu}^{1,2}&+&\hat{p}_{4,\mu}^{2,1}= \left [ ( c_{\uparrow}^{\dagger}d_{\uparrow}^{}\!+h.c . )
( 1\!-\!(n_{c\downarrow}\!-n_{d\downarrow})^2)+\uparrow \leftrightarrow \downarrow\right],\end{aligned}\ ] ] and finally , projection on the quintet gives @xmath182 \ .\end{aligned}\ ] ] equivalently , the @xmath0-symmetric expressions on two rungs can be classified .
the choice of the basis on the two - rung system is very important for the structure of the @xmath0-symmetric terms . using the simplest combination the product of an @xmath0-singlet on one rung and another @xmath0 multiplet on the other gives for a projection operator e.g. @xmath183 for the product of an @xmath0-singlet on rung @xmath10 and an @xmath0-quintet on @xmath64 , where @xmath147 is defined in section [ sec : states ] .
the numbers @xmath32 and @xmath85 in @xmath147 depend on the way the different multiplets on the rungs are labelled .
another example is an operator projecting on an @xmath0-singlet on each of the rungs @xmath184 ( see equation ( [ sing1 ] ) for the definition of the wave functions ) .
all projection operators of states consisting of at least one @xmath0-singlet on a rung can be decomposed in the same manner .
for some of these operators a compact representation in terms of electron operators is possible . as an example consider the operator @xmath185 describing pair exchange between two neighboring rungs .
it causes a transition between two @xmath0 singlet states and can be written as @xmath186 other @xmath0 singlet - singlet transitions of this type are @xmath187\ , \\ & & n_c(x)n_{d\uparrow}^{}(x)n_{d\downarrow}^{}(x ) \left [ c^{\dagger}_{\uparrow}(y)c^{\dagger}_{\downarrow}(y ) d^{}_{\uparrow}(y ) d^{}_{\downarrow}(y ) + h.c \right]\ , \nonumber\end{aligned}\ ] ] where @xmath188 similar terms are obtained from projection operators on direct products of an @xmath0-singlet on one and an @xmath0-quartet on the other rung , e.g. @xmath189 + h.c\end{aligned}\ ] ] the projection operators on the remaining @xmath190 states with a structure similar to ( [ quar2 ] ) can not easily be decomposed in this way .
they are significantly more complex , generically their expansion into electronic operators produces complicated bond - charge interaction terms .
still , forming suitable linear combinations of such terms can lead to simpler @xmath0-symmetric terms on two rungs , e.g. the pair hopping term in ( [ h - tpar ] ) or a diagonal hopping term @xmath191.\ ] ] at the end of section [ sec : ext ] we claimed that a general hamiltonian where our ansatz ( [ ansatz ] ) is the ground state of the system is given by @xmath192 the coupling constants have to be chosen such that @xmath193 and the matrix @xmath194 of coupling constants is positive definite .
this implies that @xmath195 is a lower bound on the spectrum and therefore the state ( [ ansatz ] ) having zero energy by construction will be a ground state .
the additional terms in ( [ exactham ] ) are the projection operators on the remaining multiplets not present in the matrix product wave function .
for example the projection operator on one of the @xmath0-singlets not present in this product reads @xmath196\end{aligned}\ ] ] just as @xmath194 above the matrices @xmath197 , coupling the projection operators on the remaining multiplets have to be chosen to be positive definite .
the calculation of expectation values between matrix product states is straightforward using a transfer matrix method ( see e.g. ref . ): to this end we define a @xmath198 transfer matrix @xmath199 on a rung @xmath200 with the indices @xmath201 in terms of @xmath199 the norm of the ground state can be written as @xmath202 where @xmath203 are the eigenvalues of @xmath199 . in the thermodynamic limit ( @xmath131 ) the largest eigenvalue @xmath204 dominates this expression and we obtain @xmath205 . similarly ,
one - point correlators of an operator @xmath206 are @xmath207 and a two - point correlation function reads @xmath208 here @xmath209 are the eigenvectors with eigenvalue @xmath210 of @xmath199 and @xmath211 is the transfer matrix related to the operator @xmath212 . with the matrix ( [ eq : gx ] ) the largest eigenvalue of @xmath199 is given by @xmath213 with @xmath107 , @xmath108 and @xmath109 .
this enables us to calculate the expectation values of any operator acting on a single or two rungs , respectively .
for example , we find @xmath214 for local magnetic moments and @xmath215 \\
\langle c_{\alpha}^{\dagger}(x ) c_{\beta}^{}(x ) \rangle & = & \frac{1}{w } \left [ \frac{w - h_1}{h_2 } ( 2p_4p_5)-4p_2p_3 \right ] , \hspace{1 cm } \alpha\ne\beta \end{aligned}\ ] ] for electronic expectation values . here ,
@xmath216 and @xmath217 . correlations between the total spin on two rungs decay exponentially @xmath218 as expected for finitely correlated states .
spin - spin correlations between individual sites on rungs separated by a distance @xmath219 can be expressed as @xmath220 where the amplitudes @xmath221 and @xmath222 depend on the choice of @xmath223 and @xmath224 , i.e. whether correlators of spins on the same or on different legs of the ladder are considered .
analogously , one can study electronic correlations , e.g.@xmath225 with the amplitudes @xmath226 and @xmath227 . | we construct a family of electronic ladder models with @xmath0 symmetry which have exact ground states in the form of finitely correlated wave functions .
extensions for these models preserving this symmetry are studied using these states in a variational approach . within this approach ,
the zero temperature phase diagram of these electronic ladders at half filling is obtained , reproducing the known results in the weak coupling ( band insulator ) and strong coupling regime , first studied by scalapino , zhang and hanke .
finally , the compact form of the variational wave functions allows to compute various correlation functions for these systems . |
the @xmath0 decay can be described @xcite in terms of the lorentz invariant kinematic variables @xmath6 and @xmath7 defined as @xmath8 @xmath9 here @xmath10 is the charged pion mass , @xmath11 is the charged kaon mass , @xmath12 are the pion four - momenta , the indices @xmath13 correspond to the two pions of the same electric charge ( `` even '' pions ) , and the index @xmath14 to the other ( `` odd '' ) pion .
the experimental dalitz plot distribution of the @xmath0 decay has been up to now experimentally analyzed @xcite in terms of a polynomial expansion in powers of @xmath6 and @xmath7 : @xmath15 here @xmath16 , @xmath17 , @xmath18 are the linear and quadratic slope parameters ( terms proportional to odd powers of @xmath7 are forbidden by bose symmetry , and the variable @xmath7 is defined only up to a sign ) , and @xmath19 is the coulomb factor @xcite : @xmath20 where @xmath21 are the pion charges , @xmath22 is the fine structure constant , and @xmath23 is the relative velocity of the pions @xmath24 and @xmath25 , expressed via the squared invariant mass of the pion pair @xmath26 as @xmath27 among the measurements of the slope parameters performed in the past , the most precise are reported in @xcite based on @xmath28 @xmath29 decays , and in @xcite based on @xmath30 @xmath31 decays .
the primary goal of the na48/2 experiment at cern sps is the search for direct cp - violating charge asymmetries of dalitz plot linear slopes in @xmath0 @xcite and @xmath32 @xcite decays .
however , the large data sample collected also allows a study of the dalitz plot distributions to be performed at a new level of precision , estimating the detector acceptance with a detailed monte carlo ( mc ) simulation .
the parameterization ( [ slopes ] ) of the @xmath0 decay distribution takes into account electromagnetic interactions only in the first approximation @xcite and totally neglects pion rescattering effects @xcite@xcite , which were recently shown by na48/2 to contribute significantly to the distribution of @xmath32 decays @xcite .
the current study aims to measure the dalitz plot slopes within the conventional framework ( [ slopes ] ) , and represents a first step towards a more complete description of the decay distribution .
in particular , the presence of radiative effects and strong final state interactions , both ignored in the current work , imply that the slope parameters obtained , while fully comparable to those defined in the particle data group compilation @xcite , should not be attributed any precise physical significance as far as the parameterization of the weak decay in terms of more fundamental parameters is concerned ; this would require an improved theoretical framework which is just being developed @xcite , and its implementation is postponed for a forthcoming analysis .
na48/2 collected data during two runs in 2003 and 2004 , with about 50 days of efficient data - taking in each run .
this analysis is based on @xmath1 fully reconstructed @xmath0 events , corresponding to 55% of the data sample collected during the na48/2 run of 2003 .
a beam line providing two simultaneous charged beams of opposite signs overlapping in space all along the decay volume was designed and built in the high intensity hall ( ecn3 ) at the cern sps .
the beam line is a key element of the experiment , as it allows decays of @xmath31 and @xmath33 to be recorded at the same time , and therefore leads to cancellation of several systematic uncertainties for the charge asymmetry measurement .
regular alternation of magnetic fields in all the beam line elements was adopted .
the layout of the beams and detectors is shown schematically in fig .
[ fig : beams ] .
the setup is described in a right - handed orthogonal coordinate system with the @xmath34 axis directed downstream along the beam , and the @xmath35 axis directed vertically up .
the beams are produced by 400 gev/@xmath36 protons impinging at zero incidence angle on a beryllium target of 40 cm length and 2 mm diameter . charged particles with momentum
@xmath37 gev/@xmath36 are selected in a charge - symmetric way by an achromatic system of four dipole magnets with zero total deflection ( ` achromat ' ) , which splits the two beams in the vertical plane and then recombines them on a common axis .
then the beams pass through a defining collimator and a series of four quadrupoles designed to produce focusing of the beams towards the detector .
finally the two beams are again split in the vertical plane and recombined in a second achromat , where three stations of a micromega - type @xcite detector form a kaon beam spectrometer @xcite ( not used in the present analysis ) . after passing the cleaning and the final collimators ,
the beams enter the decay volume housed in a 114 m long cylindrical vacuum tank with a diameter of 1.92 m for the first 65 m , and 2.4 m for the rest .
both beams follow the same path in the decay volume : their axes coincide within 1 mm , while the transverse size of the beams is about 1 cm . with @xmath38 protons incident on the target per burst of @xmath39 s duration , the positive ( negative ) beam flux at the entrance of the decay volume is @xmath40 ( @xmath41 ) particles per pulse ,
of which @xmath42 ( @xmath43 ) are @xmath31 ( @xmath33 ) .
the @xmath44 flux ratio is about @xmath45 .
the fraction of beam kaons decaying in the decay volume at nominal momentum is @xmath46 .
the decay volume is followed by a magnetic spectrometer used for reconstructing @xmath0 decays .
the spectrometer is housed in a tank filled with helium at nearly atmospheric pressure , separated from the vacuum tank by a thin ( @xmath47 ) kevlar composite window . a thin - walled aluminium beam pipe of 16 cm outer diameter traversing the centre of the spectrometer ( and all the following detectors )
allows the undecayed beam particles and the muon halo from decays of beam pions to continue their path in vacuum .
the spectrometer consists of four drift chambers ( dch ) : dch1 , dch2 located upstream , and dch3 , dch4 downstream of a dipole magnet .
the magnet has a field integral @xmath48 tm , thus providing a horizontal transverse momentum kick @xmath49 for charged particles .
the dchs have the shape of a regular octagon with a transverse size of about 2.8 m and a fiducial area of about 4.5 m@xmath50 .
each chamber is composed of eight planes of sense wires arranged in four pairs of staggered planes oriented horizontally , vertically , and along each of the two orthogonal @xmath51 directions .
the spatial resolution of each dch is @xmath52 m .
the nominal momentum resolution of the magnetic spectrometer is parameterized as @xmath53 ( @xmath54 expressed in gev/@xmath36 ) .
the measured resolution on the reconstructed @xmath55 invariant mass varied during the running period corresponding to the considered data sample in the range of @xmath56 mev/@xmath57 , depending on dch performance .
the magnetic spectrometer is followed by a plastic scintillator hodoscope ( hod ) used to produce fast trigger signals and to provide precise time measurements of charged particles .
the hodoscope has a regular octagonal shape with a transverse size of about 2.4 m. it consists of a plane of horizontal and a plane of vertical strip - shaped counters .
each plane consists of 64 counters arranged in four quadrants .
each quadrant is subdivided into 4 segments taking part in the trigger logic .
counter widths ( lengths ) vary from 6.5 cm ( 121 cm ) for central counters to 9.9 cm ( 60 cm ) for peripheral ones .
the hodoscope is followed by a liquid krypton electromagnetic calorimeter ( lkr ) , a hadronic calorimeter ( hac ) and a muon detector ( muv ) , all of which are not used in the present analysis .
a detailed description of the components of the na48 detector can be found elsewhere @xcite .
the @xmath0 events are triggered with a two - level system . at the first level ( l1 ) ,
the rate of @xmath58 khz is reduced to @xmath59 khz by requiring coincidences of hits in the two planes of the hod in at least two of the 16 non - overlapping segments . the second level ( l2 )
is based on a hardware system computing coordinates of hits from dch drift times , and a farm of asynchronous microprocessors performing fast reconstruction of tracks and running a selection algorithm , which requires at least two tracks to originate in the decay volume with the closest distance of approach less than 5 cm .
l1 triggers not satisfying this condition are examined further and accepted nevertheless if there is a reconstructed track which is not kinematically compatible with a @xmath60 decay of a @xmath29 having momentum of 60 gev/@xmath36 directed along the beam axis .
the resulting trigger rate is about 10 khz .
* reconstruction and selection * event reconstruction is based entirely on the magnetic spectrometer information .
tracks are reconstructed from hits in dchs using the measured magnetic field map rescaled according to the recorded value of electric current in the spectrometer analyzing magnet .
three - track vertices , compatible with a @xmath0 decay , are reconstructed by extrapolation of track segments from the upstream part of the spectrometer back into the decay volume , taking into account the stray magnetic fields due to the earth s field and parasitic magnetization of the vacuum tank , and multiple scattering in the kevlar window .
the stray field correction is based on a three - dimensional field map measured in the entire vacuum tank , and reduces the amplitude of the observed sinusoidal variation of the reconstructed @xmath61 invariant mass on the azimuthal orientation of the odd pion by more than an order of magnitude to a level below 0.05 mev/@xmath57 .
the event kinematics is calculated using measured momenta and track directions extrapolated to the decay vertex .
the principal selection criteria applied to the reconstructed variables are the following : * total charge of the three pion candidates : @xmath62 ; * transverse momentum : @xmath63 gev/@xmath36 ; * longitudinal vertex position within the decay volume : @xmath64 ; * transverse vertex radius within the beam area : @xmath65 cm ; * kaon momentum within the nominal range : @xmath66 . to improve the resolution on the kinematic variables , and to reduce the impact of differences between data and mc resolutions , the events were passed through a kinematic fitting procedure with three constraints ( constraining the initial kaon direction to by along the @xmath34 axis , and the @xmath61 invariant mass to the kaon mass ) .
events with the quality of the kinematic fit corresponding to probability @xmath67 were rejected .
the fraction of these rejected events increases as a function of deviation of the reconstructed @xmath61 mass from the pdg kaon mass @xmath68 ; in particular , all the events with @xmath69 mev/@xmath57 are outside the signal region , i.e. rejected by the above condition . the geometric acceptance for the @xmath0 decays is mainly determined by the beam pipe traversing the centres of the dchs , and the material in the central region of each dch where central dch wires terminate .
this material defines a region of high dch inefficiency - value of the @xmath0 decay : @xmath70 mev . ] .
this inefficiency together with beam optics performance and variations influences the acceptance , and is difficult to reproduce accurately with a mc simulation . to minimize the effects of this problem
, it is required that the transverse positions of each pion in dch1 and dch4 planes @xmath71 ( @xmath72 )
satisfy the condition @xmath73 cm , where @xmath74 is the position of the momentum - weighted average of the three pions impact points : latexmath:[$\vec r_{0}=\sum_{i=1}^3 ( \vec r_{\pi i}|\vec p_{\pi i}|)/\sum_{i=1}^3 extrapolated from dch1 and dch2 planes ) .
@xmath74 corresponds to the transverse position of the line of flight of the initial kaon .
the value of 18 cm was chosen to exclude safely the inefficient central region taking into account the beam sizes and variations of their average transverse positions .
the described selection criterion costs about @xmath76 of the statistics ; however an appropriate mc description of the experimental conditions is more important than the sample size for the present analysis . as will be shown below in the discussion leading to the results presented in fig .
[ fig : datamc ] , the mc simulation of the experimental conditions reproduces the data distributions to a level of a few parts per mille .
the selection leaves a sample of @xmath1 events , which is practically background free , as @xmath0 is by far the dominant decay mode of the charged kaon with more than one charged particle in the final state .
the fact that backgrounds due to other decays of beam kaons and pions are negligible was also checked with a mc simulation .
the distribution of the reconstructed @xmath55 invariant mass of data events ( before the kinematic fitting ) and its comparison with mc are presented in fig .
[ fig : kmass]a .
the non - gaussian tails of the mass distribution are primarily due to @xmath77 decays in flight , and are well understood in terms of mc simulation .
the ratio of mc to data mass spectra is presented in fig .
[ fig : kmass]b .
it demonstrates the imperfection of resolution description in mc , and a deficit of mc events in the low mass region , which however contains a small fraction of the events , and is mostly outside the signal region .
the dalitz plot distribution of the selected data events ( after the kinematic fitting ) @xmath78 used for the subsequent analysis is presented in fig .
[ fig : kmass]c .
the bin sizes of the dalitz plot distributions used in the analysis are @xmath79 .
@l@r@ ( -2,100 ) ( -30,148)(1)(-30,184)(2)(-195,190)(3 ) ( -176,273)*a * ( -176,79)*b * ( -188,188)(1,-2)8 & ( 7,40 ) ( -30,25)@xmath80(-240,107)@xmath81 ( -195,196)*c * ' '' '' + * correction for trigger inefficiency * to simplify the treatment of the trigger inefficiency , stable trigger performance was the main condition used to select the sample to be used for the analysis
. inefficiencies of both l1 and l2 trigger components were directly measured as functions of ( @xmath82 ) using control data samples of prescaled low bias triggers collected along with the main triggers , which allowed a correction of the observed ( @xmath82 ) distributions , and propagation of the statistical errors of trigger inefficiencies into the result .
the l1 trigger condition requiring a coincidence of hits in two of the 16 non - overlapping hod segments is loose , as there are three charged particles in a fully reconstructed event , providing a rather low ( and stable in time ) inefficiency of @xmath83 for the selected event sample .
however , the l1 inefficiency depends rather strongly on the kinematic variables .
the primary mechanism generating such a dependence is the enhancement of inefficiency for topologies with two pions hitting the same hod segment ; such events preferably belong to the kinematic regions characterized by small relative velocity of a certain @xmath84 pair in the kaon rest frame .
the l2 inefficiency , which is due to local dch inefficiencies affecting the trigger more strongly than the off - line reconstruction due to lower redundancy and trigger timing effects , was measured to be @xmath85 ( the error indicates the maximum size of its variation during the data taking period ) .
it did not exhibit any significant correlation to the kinematic variables , due to relative complexity of the decision - making algorithm .
* monte carlo simulation * a detailed geant - based mc simulation was developed , which includes full detector geometry and material description , simulation of stray magnetic fields , dch local inefficiencies and misalignment , the beam line ( which allows a reproduction of the kaon momentum spectra and beam profiles ) , and @xmath44 relative fluxes .
moreover , time variations of the above effects during the running period were simulated . a production of @xmath86 @xmath0 events distributed according to the matrix element ( [ slopes ] ) with pdg values of the slope parameters @xcite was performed to determine the detector acceptance .
a sample of @xmath87 mc events ( almost 2.5 times larger than the data sample ) passes the selection .
comparison of data and mc distributions in such significant variables as longitudinal decay vertex position and illuminations of dch1 and dch4 planes by pions is presented in fig .
[ fig : datamc ] , and demonstrates that mc simulation reproduces the data distributions to a level of a few units of @xmath88 .
the precision of the data description can be improved by tighter cuts on pion radial positions .
( -240,100)*a * + ( -240,100)*b * + ( -240,100)*c * * fitting procedure * the measurement method is based on fitting the binned reconstructed data distribution @xmath78 presented in fig . [ fig : kmass]c with a sum of four reconstructed mc components generated according to the four terms in the polynomial ( [ slopes ] ) presented in fig .
[ fig : mc3d ] .
let us denote these reconstructed mc distributions as @xmath89 , @xmath90 , @xmath91 , and @xmath92 . to obtain them , the mc sample ( distributed in kinematic variables according to the pdg slope parameters )
was divided into four subsamples , @xmath91 , and @xmath92 samples are of equal sizes , while the @xmath89 sample is 5 times larger than each of the former three . ] , and events in each subsample were assigned appropriate weights depending on the generated ( @xmath82 ) to obtain the desired distributions .
( -371,180)*a * ( -160,180)*b * ( -27,18)@xmath80(-234,18)@xmath80 ( -203,105)latexmath:[$\bf ( -371,180)*c * ( -160,180)*d * ( -27,18)@xmath80(-234,18)@xmath80 ( -203,103)@xmath94 : @xmath95 the sum is evaluated over all the bins of reconstructed @xmath96 distributions with at least 1000 data events , which eliminates the necessity to include non - gaussian behaviour of errors . here @xmath97 is an uncertainty of the number of data events in a given bin ( composed of a statistical part and a trigger efficiency part added in quadrature ) , @xmath98 is a mc population of a bin for given values of @xmath99 : @xmath100 and @xmath101 is its statistical error : @xmath102 here @xmath103 , @xmath104 , @xmath105 and @xmath106 are the normalization constants computed taking into account the numbers of generated events in each of the four mc subsamples , and the integrals of the four terms in ( [ slopes ] ) over the dalitz plot .
the free parameters of the functional ( [ chi2 ] ) are the slope parameters @xmath99 and an overall mc normalization parameter @xmath107 . the minimization yields @xmath108 , corresponding to a satisfactory probability of 7.0% .
the results of the fit and the trigger corrections are presented in table [ tab : fit ] .
the non - zero values of the corrections arise mostly from the l1 trigger inefficiency dependence on kinematic variables , while their statistical errors receive contributions of similar sizes from l1 and l2 trigger efficiency uncertainties .
.fit results with statistical and mc statistical uncertainties , trigger corrections and their statistical uncertainties .
the trigger corrections are included into the resulting values .
[ cols="^,>,^,<,^,^,>,^ , < " , ] an important source of systematic uncertainty is the imperfect description of the resolution in pion momentum , which can be observed in fig . [
fig : kmass]b as a slight disagreement of the shapes of the reconstructed @xmath109 spectra for data and mc in the signal region values is outside the signal region . ] . to evaluate the corresponding effect , two different plausible ways of introducing smearing of the mc resolution
were used : either by increasing the smearing of dch space points from 90@xmath110 m to 100@xmath110 m , or by adding an extra @xmath111 layer of matter in the position of the kevlar window ( the former correction is more realistic ) .
the sizes of the added perturbations are such as to correct for the `` double bump '' shape of the ratio in the signal region .
both methods lead to similar systematic uncertainties on the slope parameters attributed to description of resolution in pion momentum .
these uncertainties are listed in table [ tab : syst ] .
effects due to imperfect knowledge of the magnetic field in the spectrometer magnet were evaluated .
the variation of the magnet current can be monitored with a relative precision of @xmath112 .
smaller variations are continuously controlled with a precision of @xmath113 by the deviation of the measured charge - averaged kaon mass from the nominal pdg value . a time - dependent correction is introduced by scaling the reconstructed pion momenta , decreasing the effect of overall field scale to a negligible level . to account for possible differences between the shape of the field map used for simulation and the true field , variations of the mc field map
were artificially introduced , consistently with the known precision of field measurement of @xmath114 .
the corresponding uncertainties are listed in table [ tab : syst ] .
the transverse positions of dchs and individual wires were controlled and realigned at the level of reconstruction software every 24 weeks of data taking using data collected in special runs in which muon tracks were recorded with no magnetic field in the spectrometer .
this allows an alignment precision of @xmath115 m to be reached . however , time variations of dch alignment on a shorter time scale can bias the measurement .
these variations were measured by the difference between the average reconstructed @xmath61 invariant masses for @xmath31 and @xmath33 decays , and taken into account .
the precision with which these effects are simulated leads to systematic uncertainties presented in table [ tab : syst ] .
the effects due to limited precision of measurement of the stray magnetic field in the decay volume were estimated by variation of the stray field map used for decay vertex reconstruction ; the corresponding systematic effects are presented in table [ tab : syst ] .
the kaon momentum spectra were carefully simulated , and the related residual uncertainties were found to be negligible .
possible differences between data and mc transverse scales were found to have a negligible influence on the result .
the total systematic errors were obtained by summing the above contributions in quadrature , and are presented in table [ tab : syst ] .
the dalitz plot slope parameters of the @xmath0 decays measured with a fraction of na48/2 data sample ignoring radiative effects ( apart from the coulomb factor ) and strong rescattering effects , are : @xmath116 these values are in agreement with the world averages and @xmath33 @xcite ; averaging the pdg data between @xmath31 and @xmath33 decays should take into account correlated systematic uncertainties of the @xmath31 and @xmath33 measurements by the same experiment @xcite . ] , and have an order of magnitude smaller uncertainties .
this is the first measurement of a non - zero value of the quadratic slope parameter @xmath17 .
the compatibility of the measured distribution with the pdg polynomial parameterization @xcite appears still to be acceptable at an improved level of precision ; no significant higher order slope parameters were found .
the measurement of the slope parameters is in agreement with a full next - to - leading order computation @xcite : @xmath117 the whole na48/2 sample suitable for @xmath118 dalitz plot distribution analysis contains at least three times more data ; a more elaborate analysis is foreseen when the corresponding theoretical framework is available .
it is a pleasure to thank the technical staff of the participating laboratories , universities and affiliated computing centres for their efforts in the construction of the na48 apparatus , in operation of the experiment , and in data processing .
_ et al_. ( pdg ) , j. phys . *
g33 * ( 2006 ) 1 .
et al_. , phys . lett .
* b38 * ( 1972 ) 335 .
b. devaux _ et al_. , nucl
b126 * ( 1977 ) 11 .
belkov _ et al_. , hep - ph/9311295 .
j.r . batley _
( na48/2 ) , phys . lett .
* b634 * ( 2006 ) 474 .
( na48/2 ) , phys . lett .
* b638 * ( 2006 ) 22 . + erratum - ibid . *
b640 * ( 2006 ) 297 . n. cabibbo and g. isidori , jhep * 03 * ( 2005 ) 021 . g. colangelo _
et al . _ ,
b638 * ( 2006 ) 187 .
i. scimemi , e. gmiz and j. prades , hep - ph/0602023 , hep - ph/0610364 .
s.r . gevorkyan _
et al . _ , hep - ph/0702154 .
j.r . batley _
( na48/2 ) , phys . lett .
* b633 * ( 2006 ) 173 .
y. giomataris _
et al . _ ,
methods * a376 * ( 1996 ) 29 .
b. peyaud _
et al . _ , nucl .
methods * a535 * ( 2004 ) 247 .
( na48 ) , nima - d-07 - 00056 ( 2007 ) .
f. james and m. roos , comp .
phys . comm .
* 10 * ( 1975 ) 343 .
e. gmiz , j. prades and i. scimemi , jhep * 10 * ( 2003 ) 042 . | the distribution of the @xmath0 decays in the dalitz plot has been measured by the na48/2 experiment at the cern sps with a sample of @xmath1 fully reconstructed events . with the standard particle data group parameterization
the following values of the slope parameters were obtained : @xmath2 , @xmath3 , @xmath4 .
the quality and statistical accuracy of the data have allowed an improvement in precision by more than an order of magnitude , and are such as to warrant a more elaborate theoretical treatment , including pion - pion rescattering , which is in preparation .
european organization for nuclear research + cern - ph - ep/2007 - 002 + 13 february 2007 * measurement of the dalitz plot slope parameters of the @xmath0 decay * the na48/2 collaboration + j.r .
batley , a.j .
culling , g. kalmus , c. lazzeroni , d.j .
munday , m.w .
slater , s.a .
wotton + _ cavendish laboratory , university of cambridge , cambridge , cb3 0he , u.k.@xmath5 _ + r. arcidiacono , g. bocquet , n. cabibbo , a. ceccucci , d. cundy@xmath5 , v. falaleev , m. fidecaro , l. gatignon , a. gonidec , w. kubischta , a. norton , a. maier , m. patel , a. peters + _ cern , ch-1211 genve 23 , switzerland _
+ s. balev , p.l .
frabetti , e. goudzovski@xmath5 , p. hristov@xmath5 , v. kekelidze , v. kozhuharov , l. litov , d. madigozhin , e. marinova , n. molokanova , i. polenkevich , yu .
potrebenikov , s. stoynev , a. zinchenko + _ joint institute for nuclear research , 141980 dubna , russian federation _ + e. monnier@xmath5 , e. swallow , r. winston + _ the enrico fermi institute , the university of chicago , chicago , illinois , 60126 , u.s.a .
_ + p. rubin , a. walker + _ department of physics and astronomy , university of edinburgh , jcmb king s buildings , mayfield road , edinburgh , eh9 3jz , u.k .
_ + w. baldini , a. cotta ramusino , p. dalpiaz , c. damiani , m. fiorini , a. gianoli , m. martini , f. petrucci , m. savri , m. scarpa , h. wahl + _ dipartimento di fisica delluniversit e sezione dellinfn di ferrara , i-44100 ferrara , italy _ + a. bizzeti@xmath5 , m. calvetti , e. celeghini , e. iacopini , m. lenti , f. martelli@xmath5 , g. ruggiero@xmath5 , m. veltri@xmath5 + _ dipartimento di fisica delluniversit e sezione dellinfn di firenze , i-50125 firenze , italy _ + m. behler , k. eppard , k. kleinknecht , p. marouelli , l. masetti , u. moosbrugger , c. morales morales , b. renk , m. wache , r. wanke , a. winhart + _ institut fr physik , universitt mainz , d-55099 mainz , germany@xmath5 _
+ d. coward@xmath5 , a. dabrowski , t. fonseca martin@xmath5 , m. shieh , m. szleper , m. velasco , m.d .
wood@xmath5 + _ department of physics and astronomy , northwestern university , evanston illinois 60208 - 3112 , u.s.a . _ + g. anzivino , p. cenci , e. imbergamo , a. nappi , m. pepe , m.c .
petrucci , m. piccini , m. raggi , m. valdata - nappi + _ dipartimento di fisica delluniversit e sezione dellinfn di perugia , i-06100 perugia , italy _
+ c. cerri , g. collazuol , f. costantini , l. dilella , n. doble , r. fantechi , l. fiorini , s. giudici , g. lamanna , i. mannelli , a. michetti , g. pierazzini , m. sozzi + _ dipartimento di fisica delluniversit , scuola normale superiore e sezione dellinfn di pisa , i-56100 pisa , italy _ + b. bloch - devaux , c. cheshkov@xmath5 , j.b . chze , m. de beer , j. derr , g. marel , e. mazzucato , b. peyaud , b. vallage + _ dsm / dapnia - cea saclay , f-91191 gif - sur - yvette , france _
+ m. holder , m. ziolkowski + _ fachbereich physik , universitt siegen , d-57068 siegen , germany@xmath5 _
+ s. bifani , c. biino , n. cartiglia , m. clemencic@xmath5 , s. goy lopez , f. marchetto + _ dipartimento di fisica sperimentale delluniversit e sezione dellinfn di torino , i-10125 torino , italy _ + h. dibon , m. jeitler , m. markytan , i. mikulec , g. neuhofer , l. widhalm + _ sterreichische akademie der wissenschaften , institut fr hochenergiephysik , a-10560 wien , austria@xmath5 _ + _ submitted to physics letters b. _ |
the measurement of open charm and beauty production in pb
pb collisions at @xmath0 will allow to investigate the mechanisms of heavy - quark production , propagation and hadronization in the hot and dense qcd medium formed in high - energy heavy - ion collisions .
heavy - quark production measurements in pp collisions at @xmath1@xmath2 , besides providing the necessary baseline for the study of medium effects in pb
pb collisions , are interesting _ per se _ , as a test of qcd in a new domain , 57 times above the present energy frontier at the tevatron .
the @xmath3 and @xmath4 production yields assumed as the baseline for alice simulation studies are : for pp collisions at @xmath2 , 0.16 and 0.007 , respectively @xcite ( and lower by about 25% at 10 tev , the envisaged energy of the first long pp run ) ; for the 5% most central pb
pb collisions at @xmath5 , 115 and 4.6 , respectively .
these numbers are obtained from pqcd calculations at nlo @xcite with a reasonable set of parameters @xcite , including nuclear shadowing .
an illustration of the theoretical uncertainty bands , spanning over a factor about 2 , for the d and b meson cross sections will be shown in section [ central ] , along with the expected alice sensitivity .
heavy quark medium - induced quenching is one of the most captivating topics to be addressed in collisions at the lhc . due to the qcd nature of parton energy loss , quarks
are predicted to lose less energy than gluons ( that have a higher colour charge ) and , in addition , the ` dead - cone effect ' is expected to reduce the energy loss of massive quarks with respect to light partons @xcite .
therefore , one should observe a pattern of decreasing suppression when going from the mostly gluon - originated light - flavour hadrons ( @xmath6 or @xmath7 ) to d and b mesons . in terms of the pb pb - to - pp nuclear modification factors of the @xmath8-differential yields : @xmath9 @xcite .
the alice experimental setup , described in detail in @xcite , allows for the detection of open charm and beauty hadrons in the high - multiplicity environment of central pb
pb collisions at lhc energy , where a few thousand charged particles might be produced per unit of rapidity .
the heavy - flavour capability of the alice detector is provided by : tracking system ; the inner tracking system ( its ) @xcite , the time projection chamber ( tpc ) @xcite and the transition radiation detector ( trd ) @xcite , embedded in a magnetic field of @xmath10 t , allow for track reconstruction in the pseudorapidity range @xmath11 with a @xmath8 resolution better than 2% up to @xmath12 and a transverse impact parameter , defined as the track distance of closest approach to the interaction vertex , in the plane transverse to the beams . ]
resolution better than @xmath13 for @xmath14 ( the two innermost layers of the its , @xmath15 and @xmath16 , are equipped with silicon pixel detectors ) .
particle identification system ; charged hadrons are separated via @xmath17 in the tpc and via time - of - flight in the tof detector ; electrons are separated from charged hadrons in the dedicated trd , in the tpc , and in the electromagnetic calorimeter ( emcal ) ; muons are identified in the muon spectrometer covering the pseudo - rapidity range @xmath18 .
simulation studies @xcite have shown that alice has large potential to carry out a rich heavy - flavour physics programme .
the main analyses in preparation are : open charm : fully reconstructed hadronic decays @xmath19 , @xmath20 , @xmath21 , @xmath22 , @xmath23 , @xmath24 ( under study ) in @xmath11 ; single muons and di - muons in @xmath18 .
open beauty : inclusive single leptons @xmath25 in @xmath11 and @xmath26 in @xmath18 ; inclusive displaced charmonia @xmath27 ( under study ) ; b - tagging of jets reconstructed in the tracking detectors and in the emcal ( under study ) . at present , the installation of most of the alice detector is completed and , since december 2007 , the different sub - systems are being commissioned and calibrated with cosmic - ray tracks ( atmospheric muons ) @xcite . in view of the heavy - flavour measurements , a crucial part of the commissioning
is represented by the alignment of the its , that is the determination of the actual position and orientation in space of its 2198 silicon sensors .
the alignment , which has to reach a precision well below @xmath28 in order to guarantee a close - to - design tracking resolution , will be performed using tracks from cosmics and first pp collisions .
the first results @xcite , obtained with cosmics collected during summer 2008 , indicate that the target precision is within reach . on the offline software side ,
an intense activity for the preparation and refinement of all the analysis tools is ongoing . in particular
, the analysis model using the grid distributed computing environment is being tested . as an example , in a recent campaign , more than @xmath29 pp events , corresponding to about 1/10 of the expected yearly statistics , have been simulated by running up to 15,000 simultaneous processes at almost 100 centres worldwide .
the simulated data are now being analyzed remotely by several tens of single users . in the following sections
, we report on a selection of results extrapolated to the expected statistics collected by alice per lhc year .
central ( 05% @xmath30 ) events in 1 month at @xmath31 and @xmath32 pp events in 8 months at @xmath33 , in the barrel detectors ; the forward muon arm will collect about 40 times larger samples of muon - trigger events ( i.e. @xmath34 central events ) ; safety factors are included . ]
@xmath8-differential ( left ) and b @xmath35-differential ( right ) production cross sections in @xmath36 , in pp at 14 tev , compared to nlo pqcd predictions ( mnr @xcite and fonll @xcite ) .
inner error bars represent the statistical errors , outer error bars represent the quadratic sum of statistical and systematic errors .
a normalization error of 5% is not shown.,title="fig:",scaledwidth=42.0% ] @xmath8-differential ( left ) and b @xmath35-differential ( right ) production cross sections in @xmath36 , in pp at 14 tev , compared to nlo pqcd predictions ( mnr @xcite and fonll @xcite ) .
inner error bars represent the statistical errors , outer error bars represent the quadratic sum of statistical and systematic errors .
a normalization error of 5% is not shown.,title="fig:",scaledwidth=42.0% ] among the most promising channels for open charm detection are the @xmath19 ( @xmath37 , branching ratio @xmath38 ) and @xmath20 ( @xmath39 , branching ratio @xmath40 ) decays , reconstructed in the tpc and its , in the rapidity range @xmath36 .
the detection strategy to cope with the large combinatorial background from the underlying event is based on the selection of displaced - vertex topologies , i.e. separation of the decay tracks from the primary vertex and good alignment between the reconstructed d meson momentum and flight direction @xcite .
the accessible @xmath8 range for the @xmath41 is @xmath42@xmath12 in and @xmath10@xmath12 in pp , with statistical errors better than 1520% at high @xmath8 .
similar capability is expected for the @xmath43 .
the systematic errors ( acceptance and efficiency corrections , centrality selection for pb
pb ) are expected to be smaller than 20% .
the production of open beauty at central rapidity , @xmath36 , can be studied by detecting the semi - electronic decays of b - hadrons ( branching ratio @xmath44 ) . given that electrons from beauty have an average impact parameter @xmath45 , it is possible to obtain a high - purity sample with a strategy that relies on electron identification ( tpc and trd ) and impact parameter cut ( to reduce the semi - electronic charm - decay component and reject misidentified @xmath46 and @xmath47 from dalitz decays and @xmath48 conversions ) . as an example , with @xmath49 central events ,
this strategy is expected to allow for the measurement of the b - decay electron @xmath8-differential cross section in the range @xmath50 with statistical errors lower than 15% at high @xmath8 .
similar performance figures are expected for pp collisions . in fig .
[ fig : d0btoe ] we superimpose the simulated results for d@xmath51 @xmath52 and b @xmath53 in pp collisions to the predictions from the mnr @xcite and fonll @xcite calculations .
the comparison shows that alice will be able to perform a sensitive test of the pqcd predictions for c and b production at lhc energy . by comparing to theoretical predictions the expected alice precision for the measurement of the nuclear modification factors @xmath54 and @xmath55 , and for the heavy - to - light ratio @xmath56 , it has been shown @xcite that the charm and beauty measurements described above can be used to test the expected colour - charge and mass dependence of parton energy loss .
charm and beauty production can be measured also in the forward muon spectrometer ( @xmath18 ) by analyzing the single - muon @xmath8 and di - muon invariant - mass distributions @xcite .
the main background to the ` heavy - flavour muon ' signal is @xmath46 and @xmath57 decays . the cut @xmath58 is applied in order to increase the signal - to - background ratio .
then , a technique that performs a simultaneous fit of the single - muon and di - muon distributions with the charm and beauty components , using the predicted shapes as templates , allows to extract a @xmath35-differential cross section for d and b mesons . the expected performance for pp collisions is shown in fig .
[ fig : muons ] .
since only minimal cuts are applied , the statistical errors are expected to be lower than 5% up to muon @xmath59 .
the systematic errors , mainly due to the fit assumptions , are expected to be lower than 20% .
high-@xmath8 single muons could provide the first observation of b - quark energy loss at lhc .
indeed , the single - muon @xmath8 distribution at lhc energies is expected to be dominated by b decays in the range @xmath60 and by w - boson decays above this range . therefore ,
the central - to - peripheral muon nuclear modification factor of @xmath61 would be suppressed in the region dominated by beauty , due to parton energy loss , and would rapidly increase to about one ( binary scaling ) , where the medium - blind muons from w decays dominate @xcite . , using single muons and di - muons , in pp at @xmath62 .
boxes represent the systematic uncertainties .
error bars represent the statistical uncertainties.,title="fig:",scaledwidth=45.0% ] , using single muons and di - muons , in pp at @xmath62 .
boxes represent the systematic uncertainties .
error bars represent the statistical uncertainties.,title="fig:",scaledwidth=45.0% ]
heavy quarks will provide ways to test different aspects of qcd under extreme conditions at the lhc : from the predictions of pqcd at a new energy scale , in pp collisions , to the mechanism of energy loss in a qcd medium , in pb pb .
the alice detectors and analysis tools are being commissioned , with the aim of achieving the excellent tracking , vertexing and particle identification performance that will allow to accomplish the rich heavy - flavour physics program . | alice @xcite will study nucleus nucleus and proton proton collisions at the lhc .
the main goal of the experiment is to investigate the properties of qcd matter at the extreme energy densities that will be reached in pb pb collisions .
heavy quarks ( charm and beauty ) are regarded as powerful tools for this study .
after briefly reviewing the alice heavy - flavour program , we will describe the preparation for the first measurements to be performed with pp collisions . |
let @xmath11 be a graph with @xmath1 vertices and eigenvalues @xmath12 .
the energy of @xmath11 is defined as the sum of the absolute values of its eigenvalues : @xmath13 this concept was introduced in the mathematical literature by gutman in 1978 , @xcite . for further details on this theory
we refer to @xcite and the literature cited in .
the @xmath14-vertex complete graph @xmath15 has eigenvalues @xmath16 y @xmath17 ( @xmath18 times ) .
therefore , @xmath19 . in @xcite , it was conjectured that the complete graph @xmath15 has the largest energy among all @xmath14-vertex graphs @xmath11 , that is , @xmath20 with equality iff @xmath21 . by means of counterexamples ,
this conjecture was shown to be false , see for instance @xcite .
an @xmath14-vertex graph @xmath11 whose energy satisfies @xmath22 is called _
these graphs has been introduced in @xcite in relation with some problems of molecular chemistry .
a simplest construction of a family of hyperenergetic graphs is due to walikar et al , @xcite , where authors showed that the line graph of @xmath15 , @xmath23 is hyperenergetic .
there are a number of other recent results on hyperenergetic graphs @xcite .
there are also some recent results for the hyperenergetic circulant graphs , see for instance @xcite . in this paper
we consider the following family of circulant graphs .
for each @xmath1 and @xmath5 with @xmath24 we define the circulant graph @xmath0 with vertices @xmath25 in which there is an edge between @xmath26 and @xmath14 if the inequation @xmath27 has a solution @xmath28 , see fig .
[ figure_circulant ] .
we give a closed formula for the energy of the circulant graph @xmath0 .
based on that formula we also show that the energy of the graph @xmath29 obtaining by deleting the edges of a perfect matching of a complete graph of index @xmath30 is non - hyperenegetic .
moreover , we show that the graph @xmath29 has minimal energy over the set of regular graphs of degree @xmath31 . finally , we find some bounds and explicit expressions for @xmath32 .
this allows us to conclude that for @xmath33 there is finite number of hyperenergetic graphs of the form @xmath0 ; moreover , for @xmath34 there is a finite number of non - hyperenergetic graphs of the form @xmath0 .
( -1,-1)(8,4)(2,2 ) ( 2,1 ) ( 5,2 ) ( 5,1 ) ( 4,3 ) ( 4,0 ) ( 3,0 ) ( 2,1.5)@xmath35 ( 3.5,3)@xmath36 ( 2.5,2.5)@xmath37 ( 2.5,0.5)@xmath37 ( 4,3)(5,2)(5,1)(4,0)(3,0 ) ( 4,3)(5,1 ) ( 5,2)(4,0 ) ( 5,2)(3,3 ) ( 5,1)(3,0 ) ( 3,0.5)(4,0 )
let us remark the following easy to check facts : * @xmath38 is the @xmath1-cycle graph .
* @xmath9 is the complete graph @xmath39 .
* @xmath40 is @xmath41 the complement of a perfect matching of the edges in a complete graph @xmath39 .
* @xmath42 is @xmath43 the complement of a hamiltonian cycle in a complete graph @xmath44 .
the eigenvalues of @xmath0 can be computed by terms of a finite fourier transform , it turns out that if @xmath45 denotes the @xmath46-th eigenvalue of @xmath0 then ; @xmath47 where @xmath48 thus , the energy @xmath49 for the circulant graph @xmath0 is given by the expression : @xmath50
for small values of @xmath5 , we can split the trigonometric sum defining @xmath49 into positive and negative parts : @xmath51 for each @xmath5 we have a finite number of trigonometric sums along arithmetic sequences .
those sums can be computed explicitly , by means of the following elementary lemma on trigonometric sums .
[ lem : trig_sum ] let @xmath52 be an arithmetic sequence of real numbers with @xmath53 .
the following identities hold : @xmath54 @xmath55 let us take @xmath56 .
we have @xmath57 and @xmath58 .
each sum split as the addition of two geometric sums that are explicitly computed . by undoing the same change of variables ,
we obtain the above identities . for @xmath59
we obtain a closed formula for the energy of the cycle : @xmath60 in the case @xmath61 , we take into account : @xmath62 and thus we obtain : @xmath63.\ ] ] -axis @xmath1 , from @xmath64 to @xmath65 .
@xmath66-axis @xmath67 from @xmath68 to @xmath69 .
( right ) @xmath70-axis @xmath14 , from @xmath64 to @xmath65 .
@xmath66-axis @xmath71 from @xmath72 to @xmath73.,title="fig:",width=256 ] -axis @xmath1 , from @xmath64 to @xmath65 .
@xmath66-axis @xmath67 from @xmath68 to @xmath69 .
( right ) @xmath70-axis @xmath14 , from @xmath64 to @xmath65 .
@xmath66-axis @xmath71 from @xmath72 to @xmath73.,title="fig:",width=256 ] figure [ fi : energy_2 ] illustrates the modular behavior ( mod @xmath74 and mod @xmath75 respectively ) of the energy which is easily seen in formulae and .
analogous formula , of growing complexity , may be computed for other values of @xmath5 . by means of lemma [ lem : trig_sum ]
we can also compute explicitly the energy of the graphs @xmath40 obtaining : @xmath76 @xmath77 let us recall that the graph @xmath40 is isomorphic to @xmath41 ; the graph obtained by deleting a perfect matching from a complete graph of even order @xmath78 .
it is a regular graph of degree @xmath79 .
it is well know that the energy of a regular graph of degree @xmath80 is equal or greater than @xmath81 ( see ( * ? ? ?
this trivial lower bound is reached for the complete graph @xmath82 .
the family @xmath41 gives us another example of regular graphs minimizing energy within each family of regular graphs of fixed even degree .
let us consider the the graph @xmath29 obtained by deleting the edges of a perfect matching of a complete graph of index @xmath30 .
* @xmath83 and thus @xmath29 is non - hyperenergetic . *
@xmath29 minimizes the energy in the family of regular graphs of degree @xmath31 .
a direct consequence of theorem 2 in @xcite is that the limit @xmath84 exists and it is bigger than @xmath85 .
here we compute it explicitly . for each @xmath86 the limit :
@xmath87 has value : @xmath88 note that the integral @xmath89 is the @xmath90-norm of @xmath91 where @xmath92 is the dirichlet kernel . from the triangular inequality
we obtain : @xmath93 where @xmath94 is the lebesgue constant ( see @xcite page 67 ) .
we may evaluate analytically integral .
equation @xmath95 can be solved analytically .
it yields @xmath96 between @xmath2 and @xmath97 we obtain the partition : @xmath98 the function @xmath99 is positive in the first interval , negative in the second interval , and so on .
this allows us to evaluate the integral obtaining , @xmath100 and finally by application of lemma [ lem : trig_sum ] : @xmath101 & \mbox { if $ r$ odd , } \\
\displaystyle \frac{2}{\pi } \sum_{k=1}^r \frac{1}{k}\left [ \frac{1-(-1)^k}{\sin \left(\frac{k\pi}{r+1}\right ) } + \frac{\cos\left(\frac{k \pi}{r}\right)- \cos\left(\frac{(r+1 ) k \pi}{r}\right)}{\sin \left(\frac{k\pi}{r}\right ) } \right ] & \mbox { if $ r$ even . } \end{cases}\ ] ] the first values can be computed explicitly obtaining , @xmath102 and some approximate values are : @xmath103 in particular we have that @xmath104 for @xmath105 and @xmath106 for @xmath107 .
it follows the following result .
[ th : final ] the following statements hold : * for @xmath7 there is a finite number of hyperenergetic graphs of the form @xmath0 .
* for each @xmath107 there is a finite number on non - hyperenergetic graphs of the form @xmath0 .
an example is given by @xmath40 .
-axis @xmath1 , from @xmath108 to @xmath109 .
@xmath66-axis @xmath110 from @xmath111 to @xmath112 .
the horizontal line corresponds to @xmath113 .
dots over the horizontal line represent exceptional hyperenergetic graphs in the family @xmath114 .
( right ) @xmath70-axis @xmath1 , from @xmath115 to @xmath65 .
@xmath66-axis @xmath116 from @xmath117 to @xmath118 .
the horizontal line corresponds to @xmath113 .
dots over the horizontal line represent exceptional hyperenergetic graphs in the family @xmath119.,title="fig:",width=256 ] -axis @xmath1 , from @xmath108 to @xmath109 .
@xmath66-axis @xmath110 from @xmath111 to @xmath112 .
the horizontal line corresponds to @xmath113 .
dots over the horizontal line represent exceptional hyperenergetic graphs in the family @xmath114 .
( right ) @xmath70-axis @xmath1 , from @xmath115 to @xmath65 .
@xmath66-axis @xmath116 from @xmath117 to @xmath118 .
the horizontal line corresponds to @xmath113 .
dots over the horizontal line represent exceptional hyperenergetic graphs in the family @xmath119.,title="fig:",width=256 ] a numerical exploration ( see fig .
[ figura_g4n ] ) allows us to find all the hyperenergetic graphs of the form @xmath0 with @xmath7 . 1 .
there is no hyperenergetic graph of the form @xmath38 or @xmath120 .
2 . there are five hyperenergectic graphs of the form @xmath114 with @xmath121 .
3 . there is much bigger but finite family of hyperenergetic graphs of the form @xmath119 , ilustrated by fig .
[ figura_g4n ] ( right ) . with respect to theorem [ th : final ] ( b )
, we conjecture that the only non - hyperenergetic graph of the form @xmath0 with @xmath107 is the graph @xmath40 .
we want to thank juan pablo rada for introducing us to the topic of graph energy , and sharing his experience with us .
the first author acknowledges universidad nacional de colombia research grant hermes-27984 .
the second author acknowledges the support of universidad de antioquia . | we consider circulant graphs @xmath0 where the vertices are the integers modulo @xmath1 and the neighbours of @xmath2 are @xmath3 .
the energy of @xmath0 is a trigonometric sum of @xmath4 terms . for low values of @xmath5
we compute this sum explicitly .
we also study the asymptotics of the energy of @xmath0 for @xmath6 .
there is a known integral formula for the linear growth coefficient , we find a new expression of the form of a finite trigonometric sum with @xmath5 terms . as an application
we show that in the family @xmath0 for @xmath7 there is a finite number of hyperenergetic graphs . on the other hand
, for each @xmath8 there is at most a finite number of non - hyperenergetic graphs of the form @xmath0 .
finally we show that the graph @xmath9 minimizes the energy among all the regular graphs of degree @xmath10 .
* keywords : * circulant graph , graph energy , finite fourier transform .
2010 * msc * 05c35 ; 05c50 ; 42a05 . |
it has recently been discussed that the holes in the high-@xmath18 cuprates self - organize into quasi - one - dimensional stripes.@xcite the stripe phase is a periodic distribution of antiferromagnetically - ordered spin regions separated by quasi - one - dimensional charged domain walls which act as magnetic antiphase boundaries .
although the relation between the stripe correlations and the mechanism of high-@xmath18 superconductivity is not fully understood yet , it has become clear @xcite that the charge stripes determine the charge transport behavior , at least in the lightly hole doped region : charges can move more easily along the stripes than across the stripes.@xcite given that the stripes indeed affect the basic physical properties such as charge transport , it is desirable to build a comprehensive picture of the roles of stripes in the cuprates .
since the nonuniform charge distribution is expected to induce variations of the local crystal structure,@xcite which disturb phonons , the phonon heat transport is expected to be a good tool capable of detecting the influence of stripes even in the charge - localized region .
thermal conductivity is one of the basic transport properties that provides a wealth of useful information on the charge carriers and phonons , as well as their scattering processes .
it is known that the antiferromagnetic ( af ) insulating compound la@xmath7cuo@xmath6 shows predominant phonon transport at low temperatures , which is manifested in a large phonon peak at 20 25 k in the temperature dependence of both @xmath0-plane and @xmath1-axis thermal conductivities ( @xmath2 and @xmath3);@xcite such phonon peak disappears in sr - doped la@xmath4sr@xmath5cuo@xmath6 ( lsco ) with @xmath10 = 0.10 0.20.@xcite the suppression of the phonon peak is normally caused by the defect scattering and the electron scattering of phonons in doped single crystals ; however , it is also known that the phonon peak re - appears in both @xmath17 and @xmath19 of overdoped la@xmath20sr@xmath21cuo@xmath6,@xcite which can not be explained in this scenario .
furthermore , it was found that in rare earth ( re ) and sr co - doped la@xmath7cuo@xmath6 , such as la@xmath12nd@xmath13sr@xmath14cuo@xmath6 , the phonon thermal conductivity is much more enhanced in the non - superconducting ltt ( low - temperature tetragonal ) phase , compared to that in lsco with the same sr content.@xcite this can not happen if the defect scattering and electron scattering of phonons are the only source of the peak suppression .
based on the fact that in re- and sr - doped la@xmath7cuo@xmath6 systems the phononic thermal conductivity is always strongly suppressed in superconducting samples,@xcite it was proposed that dynamical stripes cause a pronounced damping of phonon heat transport , while static stripes do not suppress the phonon transport so significantly.@xcite it is indeed possible that the static stripes are not effective in scattering phonons , if they only induce _ periodic _ local distortions in the crystal structure . since the spin stripes in lightly doped lsco ( @xmath10 = 0.01 0.05 ) are reported to be static,@xcite one may naively expect that the phonon heat transport is not strongly suppressed in such lightly doped lsco .
however , there has been no measurement of the thermal conductivity of lightly doped lsco single crystals . in this paper
, we report our study of the thermal conductivity of lightly doped la@xmath4sr@xmath5cuo@xmath6
( @xmath10 = 0 0.04 ) single crystals .
such low doping levels are intentionally selected for the reasons of both tracing the evolution of the phonon peak and avoiding significant modifications of the crystal structure and phonon mode , which may complicate interpretations of the data .
it is found that the phonon peak is suppressed significantly even with very small doping concentration , especially for the @xmath1-axis heat transport .
since this result is contrary to the naive expectation mentioned in the previous paragraph , we also study the thermal conductivity of la@xmath7cu@xmath8zn@xmath9o@xmath6 ( lczo ) ( @xmath11 = 0 0.04 ) single crystals , which have similar amount of dopants as lsco .
the important difference between these two systems is that there can not be stripes in lczo ( since there is no carrier ) , while there are static spin stripes in lsco at low temperatures . by comparing the thermal conductivity behaviors in these two systems
, we can elucidate whether the static stripes are responsible for strong scattering of phonons .
the comparison indicate that the stripes , though they are static , indeed damp the @xmath1-axis phonon transport significantly , while their role is minor in the in - plane phonon transport .
the single crystals of lsco and lczo are grown by the traveling - solvent floating - zone ( tsfz ) technique and carefully annealed in flowing pure he gas to remove the excess oxygen.@xcite after the crystallographic axes are determined by using the x - ray laue analysis , the crystals are cut into rectangular thin platelets with the typical sizes of @xmath22 mm@xmath23 , where the @xmath1 axis is perpendicular or parallel to the platelet with an accuracy of 1@xmath24 .
the thermal conductivity @xmath25 is measured in the temperature range of 2 300 k using a steady - state technique;@xcite above 150 k , a double thermal shielding is employed to minimize the heat loss due to radiation , and the residual radiation loss is corrected for by using an elaborate measurement configuration . the temperature difference @xmath26@xmath27 in the sample is measured by a differential chromel - constantan thermocouple , which is glued to the sample using ge vanish .
the @xmath26@xmath27 varies between 0.5% and 2% of the sample temperature . to improve the accuracy of the measurement at low temperatures
, @xmath25 is also measured with one heater , two thermometer " method from 2 to 20 k by using a chip heater and two cernox chip sensors.@xcite the errors in the thermal conductivity data are smaller than 10% , which is mainly caused by the uncertainties in the geometrical factors .
magnetization measurements are carried out using a quantum design squid magnetometer .
it is useful to first establish an understanding of the anisotropic heat transport in undoped crystals .
the temperature dependences of the thermal conductivity measured along the @xmath0 plane and the @xmath1 axis in pure la@xmath7cuo@xmath6 are shown in fig .
1 , which also contains data for lsco and lczo single crystals . for undoped la@xmath7cuo@xmath6 sample
, a sharp peak appears at low temperature , at @xmath2825 k in @xmath17 and at @xmath2820 k in @xmath19 , respectively . in the @xmath1 direction , above the peak temperature
, @xmath3 decreases with increasing @xmath27 approximately following @xmath29 dependence , which is typical for phonon heat transport.@xcite this phonon peak originates from the competition between the increase in the population of phonons and the decrease in their mean free path ( due to the phonon - phonon umklapp scattering ) with increasing temperature .
in contrast to the mainly phononic heat transport in the @xmath1 direction , in the @xmath0 plane another large and broad peak develops at higher temperature ( @xmath28270 k ) , which has been attributed @xcite to the magnon transport in a long - range - ordered af state .
the heat conduction due to magnetic excitations has been observed in many low - dimensional quantum antiferromagnets , such as one - dimensional spin systems cugeo@xmath30,@xcite sr@xmath7cuo@xmath30,@xcite srcuo@xmath7,@xcite and ( sr , ca)@xmath31cu@xmath32o@xmath33,@xcite and also two - dimensional antiferromagnet k@xmath7v@xmath30o@xmath34.@xcite the absence of this high-@xmath27 peak in @xmath19 is obviously due to the much weaker magnetic correlations in the @xmath1 direction compared to that in the cuo@xmath7 plane .
all these behaviors in undoped la@xmath7cuo@xmath6 are consistent with the previously reported data.@xcite it is worthwhile to note that the phononic peak value of @xmath3 ( 27 w / km ) is considerably larger than that of @xmath2 ( 16 w / km ) .
one possible reason for this difference is that the phonon heat transport is intrinsically easier along the @xmath1 axis , which is not so easy to conceive in view of the layered crystal structure of la@xmath7cuo@xmath6 .
another , more likely possibility is that there exists phonon - magnon scattering in the @xmath0 plane that causes additional damping of the phonon peak .
such scattering may also happen in the @xmath1 axis , however , it should be much weaker than in the @xmath0 plane because it is well known that the magnons are good excitations only for the in - plane physics in the la@xmath7cuo@xmath6 system .
sr@xmath5cuo@xmath6 and ( b ) la@xmath7cu@xmath8zn@xmath9o@xmath6 single crystals measured in 0.5 t field applied along the @xmath1 axis.,width=226 ] upon sr or zn doping , there are strong doping dependences in both @xmath2 and @xmath3 .
let us first discuss the change of the high-@xmath27 magnon peak in @xmath17 for different dopants .
this peak is suppressed very quickly upon sr - doping and in fact completely disappears at only 1% doping concentration . on the other hand ,
the suppression of this peak is much slower in the zn - doped case , where the peak , though gradually suppressed , survives to @xmath11 = 0.04 . to clarify the relation between the high-@xmath27 peak and the nel order , the magnetic susceptibility of lsco and lczo is measured from 5 to 350 k in the magnetic field of 0.5 t applied along the @xmath1 axis .
figure 2 shows the temperature dependences of the magnetic susceptibility , where the peak corresponds to the nel transition ( no transition is observed in lsco with @xmath35 0.02 ) .
it is clear ( as has been already reported @xcite ) that sr doping is much stronger than zn doping in destroying the af long - range order , where the former decreases the nel temperature @xmath36 much more quickly .
the reason is related to the fact that holes introduced by sr are mobile and have 1/2 spin , while zn ( zero spin ) brings static spin vacancy in cuo@xmath7 plane .
figure 3 summarizes the doping dependences of the size of the high-@xmath27 peak in @xmath17 [ defined by the difference between the peak value and @xmath2(100 k ) ] , as well as those of the nel temperature @xmath36 , for the two systems .
this result confirms that the magnon peak is quickly diminished as @xmath36 is reduced .
it is useful to note that the disorder in the nel state induced by sr doping appears to be detrimental to the magnon heat transport , because the magnon peak completely disappears at @xmath10 = 0.01 even though the nel order is still established at 240 k for this sr doping .
peak in @xmath17 [ defined by the difference between the peak value and @xmath2(100 k ) ] of la@xmath4sr@xmath5cuo@xmath6 and la@xmath7cu@xmath8zn@xmath9o@xmath6 single crystals .
( b ) doping dependences of the nel temperature @xmath36 for the two cases.,width=207 ] more interesting changes happen in the suppression of the phonon peak at low temperatures .
one can see in fig .
1 that in @xmath17 the peak magnitude decreases more quickly with zn doping than with sr doping ; on the contrary , the peak value in @xmath3 decreases much more quickly with sr doping than with zn doping .
it should be noted that those peculiar differences between sr doping and zn doping can not be due to the additional electronic thermal conductivity in the sr - doped samples : the electronic thermal conductivity @xmath37 can roughly be estimated by the wiedemann - franz law , @xmath38 , where @xmath39 is the electrical resistivity and @xmath40 is the lorenz number ( which can be approximated by the sommerfeld value , @xmath41 w@xmath42/k@xmath43 ) . using the resistivity data for our crystals,@xcite the contribution @xmath44 for la@xmath45sr@xmath46cuo@xmath6 can be estimated to be smaller than 0.25 w / km in the whole temperature region [ dashed line in fig .
1(a ) ] , and samples with lower sr doping should have smaller @xmath44 than this estimate for @xmath10 = 0.04 ; clearly , the electronic contribution can be safely neglected in the lightly sr doped region in the discussion of the @xmath17 behavior , not to mention the @xmath47 behavior .
thus , the changes in the low - temperature peak upon doping must be due to the changes in the phonon transport properties , that is , the scattering process that determines the phonon mean free path .
= 0.12 ) single crystals to that of nd- and eu - doped crystals , in which the phonon peak re - appears .
the data for both ( a ) @xmath17 and ( b ) @xmath47 are shown.,width=245 ] although both the phonon - impurity scattering and the phonon - carrier scattering can contribute to the suppression of the phonon peak , these scattering processes can not be the only mechanisms to determine the sr - doping dependence of the phonon heat transport , because the phonon peak re - appears in overdoped lsco and in re - doped lsco.@xcite figure 4 shows our data for the re - appearance of the phonon peak in re - doped samples ( both nd and eu doped cases ) for @xmath10 = 0.12 ; these data are taken on single crystals , and essentially confirms the polycrystalline data reported by baberski _
_ @xcite the single crystal data of fig .
4 allow us to compare the absolute values of @xmath2 and @xmath3 of the re - doped samples to those of the re - free lsco ; such comparison tells us that the low-@xmath27 peak values of @xmath2 and @xmath3 of the re - doped crystals at @xmath10 = 0.12 are similar to those of the re - free lsco crystals at @xmath10 = 0.04 , despite the factor of three difference in the sr concentrations .
this result clearly indicates that the phonon - impurity scattering and the phonon - carrier scattering are not the only scattering mechanisms to determine the phonon mean free path .
based on the apparent similarity of the doping dependence of the phonon peak in @xmath2 , shown in figs .
1(a ) and 1(b ) , one might naively conclude that the lattice disorders induced by zn and sr in the @xmath0 plane are very similar and that the @xmath17 behavior is essentially explained only by the lattice disorder ; however , this is not likely to be the case , which can be understood by considering the nature of the impurity scattering and the additional charge carriers introduced in lsco .
first , it is important to notice that the low-@xmath27 phonon peak in @xmath2 is suppressed slightly more quickly by zn doping than by sr doping . in the impurity - scattering scenario,@xcite it is difficult to conceive that the phonons are scattered more strongly in lczo than in lsco , because the atomic mass difference between cu and zn is much smaller than that between la and sr .
in addition , the contribution of the phonon - carrier scattering , which exists only in lsco , would cause the phonon peak in lsco to be suppressed more quickly than in lczo .
thus , the difference between the sr - doping dependence and the zn - doping dependence in @xmath17 at low temperature , which is opposite to the naturally expected trend , suggests that there are additional scatterers of phonons in the zn doped samples .
it is most likely that the additional scatterers in lczo are of magnetic origin ; magnons , which cause the high - temperature peak in lczo , are an obvious candidate , and the magnetic disordering around zn atoms may also cause some scattering of phonons through magnetoelastic coupling.@xcite it is useful to note that , as we mentioned in section iii a , magnons are likely to be already responsible for the relatively small phonon peak in @xmath2 ( compared to that in @xmath3 ) in pure la@xmath7cuo@xmath6 .
contrary to the relatively small difference in the phonon peak of @xmath2 between lsco and lczo , the suppression of the phonon peak of @xmath3 is much quicker in lsco than in lczo ; especially , the sharp suppression of the phonon peak from @xmath10 = 0 to 0.01 is quite surprising and is the most striking observation of this work .
apart from the impurity scattering , two different scattering processes of phonons are possibly responsible for the suppression of the @xmath1-axis phonon heat transport .
the first possibility is the magnon - phonon scattering .
this may partly play a role ( particularly in samples where the long - range nel order is established ) in the @xmath1-axis phonon heat transport , although their role is expected to be minor because of the essentially two - dimensional nature of the magnons in the lco system , which means that the magnons can not effectively change the wave vector of the @xmath1-axis phonons .
the second possibility is the scattering of phonons caused by the static stripes that exist only in lsco . in the following , we elaborate on this possibility .
it is known that static spin stripes are formed at low temperatures in lightly doped lsco.@xcite if the charges also form static stripes together with the spins , they certainly cause local lattice distortions .
even if the charges do not conform to the static stripe potentials set by the spins , the spin stripes themselves may well cause local lattice distortions due to the magnetoelastic coupling;@xcite this direct coupling between the spin stripes and the local lattice distortions is in fact very likely , because recently lavrov _ et al . _
found that the spin - lattice coupling is very strong in lightly doped lsco.@xcite it should be noted that the local lattice distortions due to static stripes are _ not _ expected to scatter phonons when the stripes induce only static and _ periodic _ modulation of the lattice ( which is in fact a kind of superlattice ) ; however , possible disordering of stripes in the @xmath1 direction can introduce rather strong scattering of phonons in this direction .
in fact , because of the weak magnetic correlations along the @xmath1 axis , the stripes in neighboring cuo@xmath7 planes are only weakly correlated or even uncorrelated , which is best evidenced by the very short magnetic correlation length @xmath48 in the stripe phase ( usually smaller than the distance between neighboring cuo@xmath7 planes ) in the lightly doped lsco.@xcite we can further discuss the possible role of static stripes in the suppression of the phonon peak in @xmath2 of lightly doped lsco .
although the stripes are much better ordered in the @xmath0 plane than in the @xmath1 direction,@xcite there are two reasons that phonons in the @xmath0 plane are possibly scattered by the stripes .
first , the static spin stripes were reported to be established at about 30 k to 17 k for @xmath10 = 0.01 to 0.04 by the neutron measurements.@xcite these temperatures are very close to the position of the phonon peak ( 20 25 k ) . near the stripe freezing temperature ,
local lattice distortions are expected to be well developed and yet are slowly fluctuating ( or disordered ) , which would tend to scatter phonons .
second , there is no evidence until today that well periodically - ordered static charge stripes are formed in lightly doped lsco ; instead , our charge transport data strongly suggest @xcite that the charge stripes exist in a liquid ( or nematic ) state.@xcite such disordered state of charge stripes would tend to scatter phonons .
thus , it is possible that the static stripes also contribute to the scattering of phonons in the in - plane heat transport in lsco , although their effect appears to be much weaker than that in the @xmath1 direction .
we have measured the @xmath0-plane and the @xmath1-axis thermal conductivities of lightly doped lsco and lczo single crystals .
it is found that the low - temperature phonon peak is significantly suppressed upon sr or zn doping even at very low doping levels , and that the doping dependence show clear differences between the sr and zn dopants , and between @xmath2 and @xmath49 .
the experimental observations can be summarized as follows : ( i ) the phonon peak in @xmath3 decreases much more quickly with sr doping than with zn doping .
( ii ) on the other hand , the phonon peak in @xmath2 is suppressed slightly more quickly with zn doping than with sr doping .
( iii ) at high temperature , the magnon peak in @xmath17 decreases much more quickly with sr doping than with zn doping ; in fact , the magnon peak completely disappears in lsco with @xmath10 = 0.01 , while it is still observable in lczo with @xmath11 = 0.04 .
( iv ) rare - earth ( nd or eu ) doping for lsco ( @xmath10 = 0.12 ) enhances the phonon heat transport in both @xmath2 and @xmath3 , and this is manifested in the re - appearance of the low temperature phonon peak , whose height is similar to that of rare - earth - free lsco with @xmath50 .
based on the peculiar doping dependences of the low - temperature phonon peak , we can deduce the following conclusions for the phonon heat transport at low temperatures : ( i ) in the in - plane direction , disordered static stripes are probably working as scatterers of phonon ( in addition to the sr impurities and the holes ) in lsco , while in lczo magnetic scatterings cause strong damping of the phonon transport , which overcompensates the absence of the stripe scattering .
( ii ) in the @xmath1-axis heat transport , besides the possible contribution of the magnon - phonon scattering , strong disorder of the stripe arrangement along the @xmath1-axis is mainly responsible for the scattering of phonons .
the latter scattering mechanism naturally explains the strong damping of the phonon peak in lsco , while allowing much slower suppression of the phonon peak in lczo where there is no stripe .
therefore , the effect of static spin stripes , which are formed in the lightly doped lsco , appears to be most dramatically observed in the @xmath1-axis phonon heat transport . | both @xmath0-plane and @xmath1-axis thermal conductivities ( @xmath2 and @xmath3 ) of lightly doped la@xmath4sr@xmath5cuo@xmath6 and la@xmath7cu@xmath8zn@xmath9o@xmath6 single crystals ( @xmath10 or @xmath11 = 0 0.04 ) are measured from 2 to 300 k. it is found that the low - temperature phonon peak ( at 20 25 k ) is significantly suppressed upon sr or zn doping even at very low doping , though its precise doping dependences show interesting differences between the sr and zn dopants , or between the @xmath0 plane and the @xmath1 axis .
most notably , the phonon peak in @xmath3 decreases much more quickly with sr doping than with zn doping , while the phonon - peak suppression in @xmath2 shows an opposite trend .
it is discussed that the scattering of phonons by stripes is playing an important role in the damping of the phonon heat transport in lightly doped lsco , in which static spin stripes has been observed by neutron scattering .
we also show @xmath2 and @xmath3 data of la@xmath12nd@xmath13sr@xmath14cuo@xmath6 and la@xmath15eu@xmath16sr@xmath14cuo@xmath6 single crystals to compare with the data of the lightly doped crystals for the discussion of the role of stripes . at high temperature , the magnon peak ( i.e.
, the peak caused by the spin heat transport near the nel temperature ) in @xmath17 is found to be rather robust against zn doping , while it completely disappears with only 1% of sr doping . |
the high - acceptance dielectron spectrometer ( hades ) is a magnetic spectrometer in operation at the heavy ion synchrotron facility sis at gsi darmstadt .
it is designed for di - electron spectroscopy in nucleus - nucleus collisions around 1 - 2 agev to study hot and dense nuclear matter .
di - electron production in c+c collisions at 1 and 2 agev have been reported on @xcite . in the invariant mass region 0.15 gev/@xmath6
< @xmath7 < 0.5 gev/@xmath6 the measured pair yield shows a strong excess above the contribution expected from hadron decays after freeze - out . in this mass range ,
the @xmath8 dalitz decay is well under control , but two other di - electron sources which play an important role are poorly known : the @xmath1 ( 1232 ) resonance dalitz decay ( @xmath2 ) and @xmath9 bremsstrahlung ( @xmath10 ) .
+ for a better understanding of the contribution from such processes , hades studied @xmath0 and @xmath11 interactions at 1.25 gev kinetic energy @xcite , just below the @xmath8 production threshold . due to the small expected contribution for the @xmath9 bremsstrahlung
, the @xmath0 reaction at 1.25 gev is expected to be mostly sensitive to the @xmath1 dalitz decay @xcite .
this process is studied by leptonic inclusive ( @xmath12 ) and exclusive ( @xmath13 ) channel analyses . at the same time
, we measured the hadronic channels ( @xmath14 ) which provide a consistency check with the analysis of leptonic channels as well as a precise and independent control of inputs ( cross section , angular distribution , etc . ) for @xmath1 dalitz decay study .
the pion production is dominated by @xmath1 excitation at @xmath15 gev , therefore , the two hadronic channels are correlated by isospin symmetry : @xmath16 , leading to @xmath17 .
the hades detector , as shown in fig .
[ hades_sideview ] , consists of 6 identical sectors covering the full azimuthal range and polar angles from @xmath18 to @xmath19 with respect to the beam direction .
each sector contains : a ring imaging cherenkov ( rich ) detector used for electron identification ; two sets of mini - drift chambers ( mdc ) with 4 modules per sector placed in front and behind the magnetic field to determine momenta of charged particles ; the time of flight detectors ( tof / tofino ) and the pre - shower detector improving the electron identification . for reaction time measurement , a start detector
is located in front of the target .
an @xmath20 invariant mass resolution at the @xmath21 peak of @xmath22 and a momentum resolution for protons of @xmath23 can be achieved .
the first level trigger is obtained by a fast multiplicity signal coming from the tof / tofino wall , combined with a reaction signal from the start detector , while the second level trigger is made by using the informations from the rich and pre - shower to enrich the lepton candidates signals .
+ in the april 2006 experiment , we used protons at 1.25 gev kinetic energy , with an intensity of about @xmath24 particles / second , and a target filled with liquid hydrogen .
the start detector was not used in this run because of the too high secondaries produced by the start detector itself .
thus a specific algorithm @xcite has been developed to calculate the time of flight and then to identify the charged particles .
the two hadronic channels , @xmath25 and @xmath26 , were simulated at a beam energy of 1.25 gev using the pluto event generator @xcite .
as summarized in table [ tab : let ] , the channels involving @xmath1 and n * resonances are simulated with the cross sections taken from @xcite . for the @xmath1 resonance production , the mass distribution and angular distribution are taken from dmitriev s calculation based on one-@xmath5 exchange model which describes very well the previous experimental data @xcite . for the @xmath1 hadronic decay , we implemented the angular differential cross section @xmath27 which is in agreement with the anisotropies measured by previous experiments @xcite ( @xmath28 is the @xmath5 polar angle in @xmath1 frame with respect to the direction of momentum transfer from @xmath29 to @xmath1 in the total center of mass ) .
theoretical predictions @xcite have been used for the angular distribution of n * production since no data exist .
simulated events were filtered through hades acceptance to compare with the data .
.employed hadronic channels for the @xmath0 reaction at 1.25 gev . [
cols="<,^,^,^",options="header " , ]
the @xmath25 channel is studied using a reconstruction of the undetected neutron .
the reaction was selected first by the charged particle identification based on momentum and reconstructed time of flight @xcite , then a ( @xmath30 ) missing mass cut was imposed around the neutron mass .
this cut efficiently suppresses the background coming from misidentified protons and two-@xmath5 contributions . in the dalitz plot ( fig .[dalitz ] ) , one can clearly see the @xmath31 signal located around @xmath32 ( gev/@xmath6)@xmath33 corresponding to the squared mass of @xmath34 , while the @xmath35 signal located around @xmath36 ( gev/@xmath6)@xmath33 is less pronounced .
the spot appearing in the dalitz plot for ( @xmath37 ) and ( @xmath30 ) invariant masses squared around 2 ( gev / c@xmath33)@xmath33 is due to the @xmath38 final state interaction ( fsi ) .
this is confirmed by a simulation where @xmath38 fsi was implemented using the jost function formalism @xcite .
reaction : @xmath39 against @xmath40 invariant mass squared distributions ( preliminary hades data).,width=240 ] fig . [ invmass ] exhibits the projection on the ( @xmath30 ) ( left ) and ( @xmath37 ) ( right ) invariant masses and a comparison to simulation
. the data and simulation are both normalized to the total @xmath0 elastic cross section which is also measured in this experiment .
error bars include statistical and systematic errors due to event selection ( @xmath41 ) and correction of trigger condition which has been applied for some of the bins . the uncertainty on the normalization to @xmath0 elastic scattering is also considered as a source of systematic error ( @xmath42 ) but not included in the error bars here .
the prominent peak of @xmath43 around 1.23 gev/@xmath6 confirms that most of the @xmath44 are produced via @xmath31 decay , which is consistent with the resonance model .
@xmath45 contribution seems also reasonable and the invariant mass distributions are rather well reproduced by the @xmath46 and @xmath47 simulations . ) and * right : * ( @xmath37 ) invariant mass distributions compared to pluto simulation : total contribution ( red ) , @xmath31 ( blue ) , @xmath35 ( pink ) and n * ( green ) .
both the data and simulations are normalized to the total @xmath0 elastic cross section.,title="fig:",width=240 ] ) and * right : * ( @xmath37 ) invariant mass distributions compared to pluto simulation : total contribution ( red ) , @xmath31 ( blue ) , @xmath35 ( pink ) and n * ( green ) .
both the data and simulations are normalized to the total @xmath0 elastic cross section.,title="fig:",width=240 ] we also looked at the neutron angular distribution in the center of mass system ( fig .
[ angn ] ) which is sensitive to the angular distribution of @xmath1 resonance production since @xmath46 is the dominant process .
the comparison shows a good agreement between data ( black points ) and simulation ( red solid line ) for forward and backward neutron angles , however , an excess around cos@xmath48 is found ( our data are a factor 2 above simulation ) . by adding a non - resonant @xmath5 production contribution with @xmath49 mb the simulation ( red dashed line ) got closer to the data points around @xmath48 , even though
, it can not fully explain the discrepancy .
on the other hand , the cross section assumed here for the non - resonant contribution seems reasonable and can not be larger since it is also constrained by invariant mass distributions .
but one should mention that the majority of statistics is located at forward and backward neutron angles where the yield is in agreement with the simulation and that the overall discrepancy is less than @xmath41 of the total cross section . to minimize the model dependence
, a two dimensional acceptance correction has been performed using the resonance model .
for each ( cos@xmath50 , @xmath43 ) bin , the correction factors are calculated as the ratio of events from simulation in full phase space and in geometrical hades acceptance , then they are applied to correct the data .
the width of the bins is adjusted to optimize the precision of the correction . the systematical error introduced by the acceptance correction procedure
is estimated to @xmath41 .
we finally obtain the cross section @xmath51 mb , in good agreement with previous data @xcite .
center of mass system .
data ( black points ) compared to simulations .
( pluto simulation ( red solid line ) including @xmath31 ( blue ) , @xmath35 ( pink ) and n * ( green ) ; improved pluto simulation ( red dashed line ) with additional non - resonant contribution ( orange dashed line ) ) . *
right : * angular distribution of neutron in center of mass system after acceptance correction .
data ( black points ) compared to pluto simulation ( red line ) .
both data and simulations were normalized to the total @xmath0 elastic cross section.,title="fig:",width=240 ] center of mass system .
data ( black points ) compared to simulations .
( pluto simulation ( red solid line ) including @xmath31 ( blue ) , @xmath35 ( pink ) and n * ( green ) ; improved pluto simulation ( red dashed line ) with additional non - resonant contribution ( orange dashed line ) ) . *
right : * angular distribution of neutron in center of mass system after acceptance correction .
data ( black points ) compared to pluto simulation ( red line ) .
both data and simulations were normalized to the total @xmath0 elastic cross section.,title="fig:",width=240 ] in ( @xmath30 ) center of mass system .
experimental data are presented as black points and compared to simulations with different @xmath1 decay anisotropy parameters ( see text ) .
pluto simulation : @xmath52 ( red solid line ) , pure one-@xmath5 exchange model : @xmath53 ( red dot - dashed line ) and isotropic : @xmath54 ( red dashed line ) . *
right : * angular distribution of @xmath44 in ( @xmath30 ) reference frame after acceptance correction .
data ( black points ) are compared to pluto simulation ( red line).,title="fig:",width=240 ] in ( @xmath30 ) center of mass system .
experimental data are presented as black points and compared to simulations with different @xmath1 decay anisotropy parameters ( see text ) .
pluto simulation : @xmath52 ( red solid line ) , pure one-@xmath5 exchange model : @xmath53 ( red dot - dashed line ) and isotropic : @xmath54 ( red dashed line ) . *
right : * angular distribution of @xmath44 in ( @xmath30 ) reference frame after acceptance correction .
data ( black points ) are compared to pluto simulation ( red line).,title="fig:",width=240 ] the decay angular distribution of the @xmath1 resonance has been studied in this channel as well by looking at the angular distribution of @xmath44 in the ( @xmath30 ) center of mass system . in the simulation , the angular distribution of @xmath5 in the @xmath1 resonance frame according to @xmath55 is assumed . in order to see the sensitivity of the @xmath5 angular distribution to the anisotropy of the @xmath1 decay , and in addition to the decay distribution mentioned above in pluto simulation ( where @xmath52 ) , the distributions expected for pure one-@xmath5 exchange model ( where @xmath53 ) and isotropic decay ( where @xmath54 ) were simulated and filtered through the hades acceptance .
the comparison to the data ( left panel of fig .
[ decay ] ) shows the great sensitivity to the anisotropy of the @xmath1 and the good agreement with anisotropies measured in previous experiments @xcite .
the acceptance correction is done using the same procedure as for the neutron angular distribution mentioned above .
the right panel of fig .
[ decay ] shows the angular distribution of @xmath44 in the ( @xmath30 ) reference frame extrapolated to full solid angle , which fits nicely with the simulation .
the @xmath56 reaction was extracted from @xmath57 missing mass spectrum by subtracting a background under the @xmath58 peak .
the left panel of fig .
[ pppi0 ] presents the dalitz plot of this channel .
the prominent @xmath35 signal ( right panel in fig .
[ pppi0 ] ) is in good agreement with the event distribution obtained from the resonance model with cross section in table [ tab : let ] .
this analysis results in an exclusive @xmath58 cross section extrapolated to 4@xmath5 : @xmath59 mb , in agreement with previous data . the estimated @xmath17 is well reproduced @xcite .
reaction : @xmath60 invariant mass squared distributions with preliminary hades data . *
right : * projection of the dalitz plot onto the @xmath60 invariant mass squared ( black points ) compared to the simulation ( red ) assuming the @xmath35 ( green ) and the n * ( blue ) as intermediate resonances .
both distributions are normalized to the same number of @xmath0 elastic events.,title="fig:",width=211 ] reaction : @xmath60 invariant mass squared distributions with preliminary hades data .
* right : * projection of the dalitz plot onto the @xmath60 invariant mass squared ( black points ) compared to the simulation ( red ) assuming the @xmath35 ( green ) and the n * ( blue ) as intermediate resonances .
both distributions are normalized to the same number of @xmath0 elastic events.,title="fig:",width=211 ]
in summary , the preliminary hades results for hadronic channels @xmath25 and @xmath26 at 1.25 gev have been presented .
both measured @xmath5 production cross sections agree with the previous data and the resonance model within error bars .
the invariant masses and angular distributions show an overall agreement with the one-@xmath5 exchange model although a deviation is observed at large production angles .
consistent with previous data , the @xmath1 resonance decay is clearly anisotropic .
with the large statistics achieved , the spectra could be extrapolated to full phase space using 2-dimensional acceptance correction with a minimized model dependence .
this provides a useful test of the resonance model used for the analysis of the di - lepton channels and is especially important for the ongoing study of @xmath1 dalitz decay , using @xmath61 events .
the hades collaboration gratefully acknowledges the support by bmbf grants 06mt238 , 06tm970i , 06gi146i , 06f-140 , 06fy171 , and 06dr135 , by dfg eclust 153 ( germany ) , by gsi ( tm - krue , tm - fr1 , gi / me3 , of / str ) , by grants ga as cr iaa100480803 and msmt lc 07050 ( czech republic ) , by grant kbn 1p03b 056 29 ( poland ) , by infn ( italy ) , by cnrs / in2p3 ( france ) , by grants mcyt fpa2006 - 09154 , xuga pgid it06pxic296091pm and cpan csd2007 - 00042 ( spain ) , by grant ftc poci / fp /81982 /2007 ( portugal ) , by grant ucy-10.3.11.12 ( cyprus ) , by intas grant 06 - 1000012 - 8861 and eu contract rii3-ct-2004 - 506078 . | in @xmath0 collisions at 1.25 gev kinetic energy , the hades collaboration aimed at investigating the di - electron production related to @xmath1 ( 1232 ) dalitz decay ( @xmath2 ) . in order to constrain the models predicting the cross section and the production mechanisms of @xmath1 resonance , the hadronic channels have been measured and studied in parallel to the leptonic channels .
the analyses of @xmath3 and @xmath4 channels and the comparison to simulations are presented in this contribution , in particular the angular distributions being sensitive to @xmath1 production and decay .
the accurate acceptance corrections have been performed as well , which could be tested in all the phase space region thanks to the high statistic data .
these analyses result in an overall agreement with the one-@xmath5 exchange model and previous data . |
in the state - of - art mobile communication systems , a network operator possesses a spectrum license that provides exclusive transmission rights for a particular range of radio frequencies . spectrum assignment based on dedicated licenses resolves the issues related to inter - operator interference but it also results in low spectrum utilization efficiency .
inter - operator spectrum sharing is envisioned as one of the viable approaches to achieve higher operational bandwidth efficiency and meet the increasing mobile data traffic demand in a timely manner @xcite . in the scenario ,
a limited number of operators share a common resource pool by relying on more flexible and adaptive prioritization policies than is currently possible with dedicated licenses @xcite .
cognitive radio technologies are effective measures to resolve the sharing conflicts over the under vertical spectrum sharing @xcite , where the lessor ( owner ) operator has higher legacy rights over the spectrum than the lessee operator . on the other hand ,
the co - primary or horizontal spectrum sharing scheme conceptualizes the case where authorized operators possess equal ownership on the spectrum being adopted @xcite .
however , _ a priori _ agreements should be made on the spectrum usage with regard to the long term share of an individual operator .
the multilateral use of shared resources in the can , for instance , be achieved with channel allocation schemes originally developed for single - operator systems .
these schemes are in principle applicable to realizes inter - operator spectrum sharing , provided that the operators are willing to exchange information and cooperate honestly . under this requirement , many spectrum sharing algorithms are available in the literature .
they differ related to the domain where inter - operator interference is handled , i.e. time , frequency , and/or space .
the cooperative spectrum sharing schemes require a great deal of network information exchange among the operators , e.g. , interference prices @xcite , channel state information @xcite , etc . , and/or employing a central entity to decide upon the resource allocation . in @xcite , the operator reports the inflicted aggregate interference to the spectrum controller , and on this basis , the controller awards the spectrum pool to the impacted . in @xcite ,
cooperation amongst the operators is realized by broadcasting the spectrum occupancy information , allowing small cells of competitor operators to avoid interference in accessing the spectrum pool .
similarly , operators in @xcite maintain channel occupancy and spectrum reservation matrices for opportunistic access to the shared pool .
although the achievable gains in cooperative schemes are in general high , operators may be reluctant to share proprietary information with their competitors and may also have an incentive to mis - report this information .
finally , information exchange may incorporate excessive inter - operator signaling overhead . in this perspective ,
game theoretic non - cooperative schemes appear to be a more viable option to share spectrum . in these schemes ,
players make decisions independently ; they may still cooperate with competitors but , the cooperation is entirely self - enforcing . in @xcite , operators establish cooperation and play non - zero sum games to share spectrum . however , the choice of utility function , encompassing spectrum pricing is undesirable as it penalizes increased spectrum usage . in @xcite ,
the operators enlist their preferences of partitioning the shared pool and the outcome is established based on a minimum rule .
this method may not work well in scenarios with load variations , in which heavily - loaded and lightly - loaded operators will end up with the same number of orthogonal carriers from the pool . in @xcite , operators model their interactions via repeated games .
a common assumption is that they agree in advance on the spectrum allocation , e.g. , at a in @xcite or at an orthogonal allocation in @xcite and this allocation is maintained under the threat of punishment .
auction - based sharing techniques have been discussed in @xcite , in which operators bid competitively for spectrum access through a spectrum broker .
however , operators may be hesitant in adopting market - driven sharing schemes as they may not want to touch their revenue model . in this paper
, we consider spectrum sharing in a setting where no information is revealed to other operators .
we assume that operators are not willing to monetize spectrum use , keeping spectrum sharing on the level . unlike one - shot games , the proposed scheme takes into account also the history of previous interactions between the operators and entails the benefits of reciprocity .
we illustrate that a repeated game can be set up so that both operators achieve better performance in comparison to a static spectrum allocation . unlike the repeated game models proposed in @xcite
, we do not fix the spectrum allocation but we allow a flexible use of the based on the network load and interference conditions . by employing the proposed scheme in a scenario with two operators , we are able to show that under load asymmetry , both operators can benefit as compared to a scheme where no spectrum coordination is allowed
. the remainder of the paper is organized as follows . in section [ sec : system_model ] , we present the system model .
section [ sec : coordination ] formulates the repeated game for inter - operator spectrum sharing and presents the proposed mechanism for negotiating the utilization of the spectrum pool .
section [ sec : numericals ] demonstrates performance gains with the proposed scheme and finally section [ sec : conclusions ] concludes the paper with a summary and areas for further work .
for simplicity , we concentrate on a spectrum sharing scenario with two small cell operators , operator @xmath0 and operator @xmath1 . each operator has one for dedicated usage .
the operators participate also in a and divide it into @xmath2 of equal bandwidth .
the proposed spectrum sharing scheme will be used to negotiate the utilization of the @xmath2 in the downlink .
we consider a scenario with network load variations . at a particular time
instant , the user distribution is modeled via a with a mean equal to @xmath3 users for operator @xmath4 . given the network state ,
an operator evaluates a network utility function to describe the offered to its users .
it is important to remark that operators need not employ same utility function nor to be aware of each other s utility function . for simplicity
, we assume that both operators maintain a proportionally fair @xcite utility function directly constructed based on the user rates .
@xmath5 where @xmath6 is a particular realization of a with mean @xmath3 , @xmath7 is the transmission rate of the @xmath8-th user of operator @xmath9 on the @xmath10-th calculated as @xmath11 where @xmath12 is the time scheduling weight of the @xmath8-th user scheduled on the @xmath10-th , @xmath1 is the bandwidth of a , @xmath13 is the downlink user sinr and @xmath14 is the sinr efficiency .
we consider downlink transmissions without power control .
the downlink received signal power for the @xmath8-th user on the @xmath10-th is @xmath15 . also , let us denote by @xmath16 the aggregate interference level incorporating both interference from the operator s own network , and from the other operator s interfering .
then , the downlink user sinr is @xmath17 where @xmath18 is the power per of thermal noise and other interference .
note that on the dedicated there is no inter - operator interference .
the scheduling weights , @xmath12 are determined to maximize the utility @xmath19 .
@xmath20 in order to evaluate the effect the opponent operator has on its utility , an operator may ask its users to measure the amount of aggregate interference level they receive from the opponent . this functionality requires that the users are able to separate between their own and the other operator s generated interference .
note that this kind of functionality does not require any signaling between the two operators .
it is assumed that inter - operator interference measurements are ideal .
in small cell deployments it is expected to have changing traffic and interference profiles .
small cell deployments of different operators sharing spectrum in the same geographical area can exploit these fluctuations and achieve mutual benefits by regulating the allocation of .
for instance , let us consider spectrum sharing between two operators with unbalanced traffic loads over a . a lightly - loaded operator can satisfy its with few and
could perhaps stop using some of the from the pool .
an operator that is heavily loaded at that time would not suffer from inter - operator interference on the emptied and would be able to meet its too .
however , there should be an incentive for the lightly - loaded operator to free up some . without going to monetary transactions
, we propose to regulate the allocation of by means of _ spectrum usage favors _ asked and granted by the operators . in a
, a spectrum usage favor refers to the following action @xmath21 an operator asks its competitor for permission to start using a certain number of from the pool on an exclusive basis .
while negotiating for spectrum , the operators should agree about the default utilization of the spectrum pool . in principle , any mac protocol could be applied in the default state .
we consider that both operators utilize all the of the pool .
a spectrum usage favor that is exchanged between the operators necessitates a departure from the default state .
the time period a spectrum favor is valid is agreed between the operators .
e.g. , a favor can be valid in the order of seconds , reflecting the time the network states remain unchanged .
after the validity of a favor expires , the utilization of the spectrum pool falls back to the default state and the operators will begin a new round of negotiations .
since the operators will share spectrum for a long time , an operator taken favors in the past will return them in future to show a cooperative spirit and maintain the exchange of favors with its competitor . monetized compensations or auction schemes are not considered here .
it is a non - trivial task to design such efficient mechanisms for a limited area and a limited time , and to couple operator strategies to the income model of operators . realizing spectrum sharing in the form of favors entails the benefits of reciprocity , circumvents monetary - based spectrum sharing and enables the operators to achieve mutual benefits without revealing ran - specific information to competitors and/or other parties .
these make the considered approach similar to peering agreements in the internet .
we first consider a one - shot game where the operators @xmath0 and @xmath1 are modeled as myopic players .
the game is strategic and non - cooperative .
each player @xmath9 s action or strategy , @xmath22 , is to either ask a favor of @xmath23 ccs denoted by @xmath24 , grant a favor on @xmath23 ccs denoted by @xmath25 or do neither , denoted by @xmath26 .
let @xmath27 denote the set of such actions for player @xmath9 . to specify the outcome of the game , we assume that a favor is exchanged only if one player plays @xmath28 and the other plays @xmath29 with @xmath30 , then the outcome is an exchange of @xmath10
otherwise , no exchange of favors occur .
a trade for @xmath31 still occurs .
we leave such a model for future work . ]
depending on the outcome , operators draw rewards : ( i ) the reward when a player takes a favor is equal to the utility gain when the interference on @xmath10 is eliminated , ( ii ) the reward when a player grants a favor is equal to the utility loss when stopping to use @xmath10 and ( iii ) the reward when a player does not ask nor grant a favor is zero .
the gains and losses thus depend on the current internal state of the player in question , and are not known to the opponent .
in such a game , a final outcome is a @xmath32 , from which no player can improve its utility by deviating unilaterally , i.e. , @xmath33 for every player @xmath34 and every alternative strategy @xmath35 .
it is straightforward to see that the of the formulated one - shot game corresponds to the situation where a player always asks for a favor on @xmath2 to maximize its reward but never grants a favor . as a result
, both operators would utilize all from the pool irrespective of the network load and interference profiles .
the game formulated here differs from the power control game in @xcite in that here the actions are requests of other player powers , and power control is binary , whereas in @xcite , the actions are power allocation profiles across frequency .
also , no information of other operator state is assumed here .
the of the strategic one - shot games discussed here and in @xcite , however , are similar ; full usage of the available spectrum is the only option for a rational player .
since the operators will share spectrum for a long time , the one - shot game described above will be played repeatedly . in a repeated game ,
the action of a player at a stage game depends not only on the current rewards but also in the sequence of previous rewards see , e.g. , @xcite .
repeated games , such as prisoner s dilemma have been well - studied and admit a rich set of equilibrium profiles @xcite including various punishment strategies .
the setting here is even more challenging as the game we are interested in is a _ stochastic game _ meaning that each player s pay - off depends on a random parameter , namely the configuration of the users at that time , and moreover the player s have _ imperfect information _ since they only observe their own user configuration .
we thus have a _ bayesian game_. given this , we focus on a simplified set of stationary threshold policies and characterize an equilibrium among these policies .
negotiations of favors on a single were considered in @xcite .
here , we extend that to a situation with multiple . at each stage of the game
an operator can compute its utility gain and utility loss by asking and granting favors on @xmath10 for @xmath36 .
the of the utility gains when operator @xmath9 gets a favor on @xmath10 is denoted by @xmath37 and similarly , the of utility losses by @xmath38 . in our system setup ,
the randomness is only due to the poisson distribution of the operator s own users ( and the user s corresponding channel gains ) .
therefore the depend only on the network state of the operator s own network .
however , note that if power control and inter - cell interference coordination is employed , these distributions may depend on the state of the opponent operator s network too . at each stage game , the operator @xmath9 first checks whether to ask for a favor on @xmath2 by comparing its immediate utility gain with a threshold @xmath39 .
if the utility gain is less than the threshold @xmath39 , the operator can consider asking for a favor on @xmath40 instead , and so forth .
we assume that an operator always asks for the largest number of for which its utility gain exceeds the corresponding threshold . as a result ,
the probability that operator @xmath9 asks for a favor on @xmath10 is equal to the probability that the utility gain from taking a favor on @xmath41 is less than the corresponding thresholds @xmath42 and also , the utility gain from taking a favor on @xmath10 is higher than the threshold @xmath43 @xmath44 where , to simplify the analysis , it has been assumed that the distributions of utility gains from taking favors on different number of are independent .
similarly , the operator @xmath9 grants a favor on @xmath10 upon being asked , if its immediate utility loss is smaller than a threshold @xmath45 and it has not already requested a favor . taking into account the fact that an operator can not ask and grant a favor at the same stage game , the probability to grant a favor on @xmath10 is @xmath46 where it has been assumed that the distributions of utility gains and utility losses are independent .
we assume that the networks of the operators are similar , and in symmetric relationship with each other , and do not assume a discount of favors . to get preliminary understanding on steady state behavior in such a setting , inspired by @xcite
we thus assume that averaged over long times , operators give and take the same amount of equally valuable favors .
hence , favors would become a rudimentary ran - level spectrum sharing currency .
thus we have @xmath47 where the left - hand side describes the average number of that operator @xmath0 gets a favor on , and the right - hand side the same quantity for operator @xmath1 .
an operator can monitor the probabilities of asking and granting of the opponent and set its own decision thresholds for satisfying the constraint .
however , there may be multiple combinations of thresholds fulfilling the constraint .
we propose to identify the thresholds maximizing an excess utility @xmath48 calculated over the of the one - shot game ( i.e. both operators utilize simultaneously all the @xmath2 ) .
the excess utility for an operator reflects its expected gain from taking a favor penalized by its expected loss from granting a favor . in order to avoid unnecessary complexity in the notation we show how to set the decision thresholds for operator @xmath0 .
similarly , the decision thresholds for operator @xmath1 can be computed .
the excess utility for operator @xmath0 is @xmath49 where @xmath50 and @xmath51 are the average gain and loss in utility on @xmath10 for operator @xmath0 such that , @xmath52 the optimization problem for identifying the decision thresholds is @xmath53 in order to solve this optimization problem , we construct the lagrangian function and solve the system of first - order conditions . @xmath54 where @xmath55 is the lagrange multiplier .
starting with the partial derivative of the lagrangian in terms of @xmath56 and setting it equal to zero allows computing the value of the lagrange multiplier @xmath57 . setting the partial derivative of the lagrangian with respect to @xmath58 equal to zero , and substituting the value of the lagrange multiplier into the resulting equation gives @xmath59 next , starting from @xmath60 one can determine the threshold @xmath61 as a function of the thresholds @xmath62 @xmath63 finally , setting @xmath64 and using the solution for @xmath65
, we end up with @xmath66 the thresholds @xmath67 that may maximize the lagrangian must jointly satisfy equations @xmath21 and also the constraint .
note that the above system of equations does not accept a closed - form solution but it is straightforward to solve numerically .
also , from equations ( [ eq : dervg1],[eq : dervgn ] ) one can deduce that @xmath68 . in the appendix
, we show that the solution satisfying the first - order conditions and the constraint satisfies @xmath69 .
therefore , the proposed method can achieve better performance in comparison to the of the one - shot game .
finally , besides the calculation of the lagrangian at the stationary point , we also compute it at the borders .
the thresholds , either interior or border , maximizing the lagrangian are selected .
in order to assess the performance of the proposed coordination protocol , we consider an indoor deployment scenario in a hall of a single - story building .
the hall is a square with a side of @xmath70 m. the are partitioned into two groups as illustrated in fig .
[ fig : building ] modeling a spectrum sharing scenario with two operators .
the service areas of the operators fully overlap .
a user is connected to the of its home network with the highest received signal level at its location .
we consider a power law model for distance - based propagation pathloss with attenuation constant @xmath71 and pathloss exponent @xmath72 .
the available power budget on a is @xmath73 dbm , the thermal noise power is @xmath74 dbm / hz and the noise figure is @xmath75 db .
the sinr efficiency is @xmath76 .
the bandwidth of a is @xmath77 mhz .
initially , we consider that the consists of two . first , we consider an initialization phase of @xmath78 simulation snapshots ( or equivalently @xmath79 stage games ) . at each stage game ,
the user locations are independently generated according to the and the operators calculate and keep track of their utility gains and utility losses from taking and granting favors over one and two .
we simulate many different network loads so that the distributions of utility gains and utility losses at the end of the initialization can be seen as the steady state distributions over all possible network states .
the different network loads are generated by varying the mean of the used to model the locations of the users . in fig .
[ fig : gainlossdistr ] we depict the distributions of utility gains and losses for an operator at the end of the initialization .
next , we evaluate the performance in terms of the user rate distribution over a finite time horizon of @xmath80 stage games following the initialization phase .
initially , the values of the thresholds are set arbitrarily equal to @xmath81 , @xmath82 , @xmath83 and @xmath84 for both operators .
every @xmath85 stage games , the operators probabilities for asking and granting favors are recomputed considering all stage games . then
, the decision thresholds are updated by solving the optimization problem and so forth .
given the allocation of at each stage of the game , the operators compute and keep track of the user rates .
recall that granted favors are valid only for a particular stage game . at the end of each stage game , the allocation returns to the default state i.e. both operators utilize all the of the .
the performance of the proposed scheme is assessed in comparison with the of the one shot game , which is a static spectrum allocation scheme where both operators utilize all the of the .
first , we consider a scenario with network load asymmetry between the operators .
the mean number of users for the first @xmath86 stage games are @xmath87 and @xmath88 . in the second half of the simulation , the mean
values are reversed . in fig .
[ fig : opra_asy ] , the rate distribution for the users of operator @xmath0 is depicted over the full course of the simulation . in the first @xmath86 stage games , operator @xmath1
can mostly cope with fewer or no due to the lower load and it grants more favors than the operator @xmath0 . in the second half of the simulation , operator
@xmath0 returns the favors .
overall , operator @xmath0 offers better in comparison with the attained without any coordination e.g. it improves its mean user rate by approximately @xmath89 % .
the user rate distribution curves for operator @xmath1 follow the same trend and are not depicted . fig .
[ fig : opra_asy_4scc ] depicts the rate distribution curves for the users of operator @xmath0 when the consists of four .
one can see that the mean user rate increases by approximately @xmath70 % while , the user rate at the @xmath75 % of the distribution increases by more than @xmath85 % .
finally , we show that gains due to coordination can be achieved even in cases with equal mean network loads for the operators . in that case , the proposed protocol takes advantage of the instantaneous network load variations . in fig .
[ fig : opra_sym ] we see that operator @xmath0 improves its mean user rate by approximately @xmath90 % for mean number of users @xmath91 and four in the .
in this paper , we considered co - primary spectrum sharing between two small cell operators deployed in the same geographical area .
we considered a scenario where the operators have equal access rights on a spectrum pool and we proposed a protocol for coordinating the utilization of component carriers from the pool . according to it , an operator may ask for spectrum usage favors from its competitor .
a spectrum usage favor means that the competitor would stop using some component carriers from the pool .
an operator that has few users to serve could perhaps cope with less component carriers and grant the favor .
operators that have taken favors in the past are likely to return these favors in future and reciprocity is maintained .
we formulated the interaction between the operators as a strategic , non - cooperative repeated game . since it is hard to analyze the proposed game and find its nash equilibrium , we resort to a heuristic strategy that uses a threshold - based test to decide whether to ask or grant a favor at each stage game .
the decision thresholds depend on the current network realization and also in the history of previous interactions with the competitor .
we proved that the proposed strategy is strictly better as compared to the case without coordination between the operators .
we illustrated that in an indoor deployment scenario , two operators are both able to offer higher user rates as compared to the case with no coordination , without revealing any operator - specific information to each other .
our results show that a rational operator , knowing that the opponent is rational and has a network with similar characteristics , has incentive to be cooperative . in future works
, operators with non - similar load and network characteristics will be addressed , as well as models where the statistics of the underlying poisson process changes to reflect e.g. variations in load due to time - of - day .
using the fact that the decision thresholds over the distribution of utility losses are related as @xmath92 , one can write write equation as @xmath93 using equation into equation , the excess utility can be read as @xmath94 according to the definition of the probabilities of granting a favor from equation , we note that the last term in equation is equal to the right - hand side of the constraint in equation scaled by @xmath56 . after replacing the last term of equation by the left - hand side of the constraint in equation
, we end up with @xmath95 using the probabilities of ask a favor from equation into the last term of equation and replacing back the decision thresholds , @xmath96 , the first and the last terms of equation can be factorized together resulting to @xmath97 which is always positive since @xmath98 .
this work has been performed in the framework of the fp7 project ict 317669 metis , which is partly funded by the european union .
also , this work was supported in part by the academy of finland funded project smaciw under grant no . | we consider two small cell operators deployed in the same geographical area , sharing spectrum resources from a common pool . a method is investigated to coordinate the utilization of the spectrum pool without monetary transactions and without revealing operator - specific information to other parties .
for this , we construct a protocol based on asking and receiving _ spectrum usage favors _ by the operators , and keeping a book of the favors .
a spectrum usage favor is exchanged between the operators if one is asking for a permission to use some of the resources from the pool on an exclusive basis , and the other is willing to accept that . as a result , the proposed method does not force an operator to take action .
an operator with a high load may take spectrum usage favors from an operator that has few users to serve , and it is likely to return these favors in the future to show a cooperative spirit and maintain reciprocity .
we formulate the interactions between the operators as a repeated game and determine rules to decide whether to ask or grant a favor at each stage game .
we illustrate that under frequent network load variations , which are expected to be prominent in small cell deployments , both operators can attain higher user rates as compared to the case of no coordination of the resource utilization .
[ lsp]limited spectrum pool [ cc]component carrier [ bs]base station [ pdf]probability distribution function [ ne]nash equilibrium [ ran]radio access network [ ppp]poisson point process [ qos]quality - of - service [ mno]mobile network operator co - primary spectrum sharing , repeated games , spectrum pooling . |
it is well known that if quantum effects are taken into account , a threshold energy scale , over which general relativity looses its validity , emerges .
this threshold is called the planck scale . for the 4-dimensional spacetime ,
the planck energy is defined by @xmath0 gev , where @xmath1 , @xmath2 , and @xmath3 are the speed of light , dirac constant , and newton s gravitational constant , respectively .
the circumferential radius of a spherically symmetric black hole of mass @xmath4 is equal to its reduced compton wavelength @xmath5 , and hence such a black hole will behave as a gravitating quantum object whose behavior is unpredictable in the framework of general relativity .
the planck energy , @xmath6 , is much larger than the electroweak scale ( @xmath7 gev ) , and this fact is recognized as the hierarchy problem in the elementary particle physics . the large extra - dimension scenario was proposed as a solution for the hierarchy problem @xcite .
this scenario is inspired by superstring theories , and in this scenario , the fundamental planck energy , @xmath8 , may be as low as @xmath9 gev scale .
the peculiarity of this scenario is that the length scales of the compactification of extra - dimensions can be much larger than the fundamental planck length @xmath10 .
hence , the gravity can be described by a higher - dimensional classical theory ( e.g. , higher - dimensional general relativity ) for the distance scale smaller than the compactification scale and larger than @xmath11 .
also in this scenario , the classical theory of gravity will loose its validity in the super - planckian domains . for non - gravitational interactions ,
a length scale ( e.g. , compton wavelength ) , which is usually explored by a particle scattering , decreases with increasing its energy measured at the center of mass frame .
however , this relation may not hold for the phenomena in which gravity plays an important role .
it is widely believed that collisions of particles with the super - planckian energy scale would produce black holes , and hence , physical processes characterized by the length scale shorter than @xmath12 are hidden inside black holes @xcite .
this is a kind of the cosmic censorship hypothesis for the super - planckian domain , but it is not trivial whether this hypothesis is correct .
the original cosmic censorship hypothesis claims that , roughly speaking , naked singularities are not formed in our universe @xcite .
however , we have to note that even if the cosmic censorship hypothesis is correct , it does not necessarily imply that the super - planckian physics is hidden inside horizons . in the framework of general relativity or in the large extra - dimension scenario , the super - planckian curvature is not a spacetime singularity as long as it is finite , and thus the presence of the super - planckian domains may be irrelevant to the cosmic censorship
. indeed , nakao , harada , and miyamoto ( nhm ) recently suggested , by a simple dimensional analysis , a possibility that visible super - planckian domains may be produced in high - energy particle collisions , if the spacetime dimension is larger than four @xcite . to theoretically explore the phenomena in a higher - dimensional gravity ,
numerical relativity is probably the unique approach . in the past a few years , several implementations for the higher - dimensional numerical relativity have been developed @xcite , and now , it is feasible to explore the nonlinear physics such as high - velocity collision in higher dimensions as in 4 dimensions @xcite . in this paper , we numerically show that in a scattering of two black holes in the framework of 5-dimensional ( 5d ) general relativity , super - planckian domains may be indeed visible .
hereafter , we adopt the natural units @xmath13 and the abstract index notation : latin indices except for @xmath14 denote a tensor type and greek indices denote a component with respect to some basis vectors @xcite . in this paper , we define the fundamental planck energy @xmath8 such that 5d einstein s equation is written as @xmath15
we define a super - planckian domain as a region where 5d general relativity looses its validity . in the case that quantum effects on a spacetime geometry are not very large , quantum corrections to the einstein - hilbert action might be given in the form of scalar polynomials of the riemann tensor .
hence , we adopt the square root of the absolute value of the kretschmann invariant @xmath16 as a reference quantity .
to determine a reasonable threshold value of @xmath16 over which the domain becomes super - planckian , we adopt , as a reference , the 5d schwarzschild - tangherlini ( st ) black hole which is a spherically symmetric vacuum solution in 5d general relativity . by the definition of @xmath8 through eq .
( [ einstein ] ) , the circumferential radius of the st black hole with mass @xmath17 is equal to @xmath18 .
thus , the st black hole with @xmath19 will behave as a gravitating quantum object which actually can not be described by general relativity , because the reduced compton wavelength of this black hole agrees with its circumferential radius .
@xmath16 at the event horizon of this black hole is equal to @xmath20 .
hence , by introducing a dimensionless reference quantity @xmath21 , we can define a super - planckian domain @xmath22 by @xmath23 note that the above condition will be one of sufficient conditions for the appearance of a super - planckian domain , because a domain , in which one of scalar quantities defined from the riemann tensor exceeds an appropriately determined critical value , should be regarded as a super - planckian one @xcite .
hereafter , we consider a scattering of two non - rotating black holes with identical mass @xmath17 . a procedure for setting initial data of the scattering problem with negligible junk radiation was presented in @xcite , which we follow .
when the distance between two black holes is much larger than their gravitational radii @xmath24 , the metric near each black hole in each rest frame is well approximated by that of the st black - hole spacetime , @xmath25 where @xmath26 and @xmath27 . to obtain an approximate metric in the vicinity of each black hole in the _ center - of - mass frame _ of the two - black - hole system ,
we perform coordinate transformations for the above metric twice .
first , a boost transformation , @xmath28 , @xmath29 , @xmath30 , @xmath31 , and @xmath32 , is performed , where the velocity @xmath33 is a positive constant less than unity , and @xmath34 is the lorentz factor .
next , a spatial translation , @xmath35 and @xmath36 , is performed , where @xmath37 and @xmath38 are positive constants , respectively . as a result ,
we obtain two coordinate systems in which the world line of the black hole ( puncture ) is given by @xmath39 , @xmath40 , @xmath41 , and the line element is @xmath42 where @xmath43 , @xmath44 , and @xmath45 with @xmath46 . the components of the extrinsic curvature of the spacelike hypersurface labeled by @xmath47 are @xmath48,\\ k^\pm_{xx}&=&k^{\pm}_{yy}=k^{\pm}_{zz } = \mp\frac{\gamma v(w\mp\ell_w\pm vt)\alpha_\pm\psi_\pm'}{b_0\psi_\pm r_\pm } , \\ k^\pm_{wx}&=&\mp\frac{\gamma v(x\mp\ell_x)b_\pm}{r_\pm}\left[\alpha_\pm '
-\frac{\alpha_\pm}{2}\left\{\ln(\psi_\pm^2-v^2\alpha_\pm^2)\right\}'\right ] , \\
k^\pm_{wy}&=&\mp\frac{\gamma vyb_\pm}{r_\pm}\left[\alpha_\pm ' -\frac{\alpha_\pm}{2}\left\{\ln(\psi_\pm^2-v^2\alpha_\pm^2)\right\}'\right ] , \\
k^\pm_{wz}&=&\mp\frac{\gamma vzb_\pm}{r_\pm}\left[\alpha_\pm ' -\frac{\alpha_\pm}{2}\left\{\ln(\psi_\pm^2-v^2\alpha_\pm^2)\right\}'\right],\end{aligned}\ ] ] and the other components vanish . here
, the prime denotes the ordinary derivative with respect to @xmath49 .
based on the above results , we write the initial data for a scattering of two black holes with initial velocities @xmath50 .
the metric of the spacelike hypersurface is written in the following form @xmath51 where @xmath52.\end{aligned}\ ] ] the extrinsic curvature is written by @xmath53 finally , we set @xmath54 . the unknown functions @xmath55 in eq .
( [ 4-metric ] ) and @xmath56 in eq .
( [ eq13 ] ) should be determined from the conditions that the hamiltonian and momentum constraints are satisfied . however
, if the coordinate separation between two black holes , @xmath57 , is much larger than @xmath58 , @xmath59 and @xmath60 are much smaller than @xmath61 and @xmath62 @xcite .
we choose the initial separation to be sufficiently large and set the small corrections to be zero .
such approximation is acceptable for our present purpose .
numerical simulations were performed using sacra - nd code reported in ref .
@xcite , in which the so - called baumgarte - shapiro - shibata - nakamura formalism @xcite together with the `` moving puncture '' approach @xcite and adaptive mesh refinement algorithm @xcite are employed .
the numerical accuracy is monitored by computing @xmath63-norm of the hamiltonian and momentum constraints .
the convergence of the numerical solution was tested varying the grid resolutions with the grid spacing as @xmath64 , 12/320 , and 10/320 .
we confirmed a reasonable convergence behavior ( see below ) . and
@xmath65 are plotted as functions of the initial speed @xmath33 of each black hole . for @xmath66 , @xmath67 and @xmath65
agree with the critical impact parameter @xmath68 .
, title="fig : " ] [ fig : fig1 ] numerical simulations were systematically performed varying @xmath33 and an impact parameter defined by @xmath69 . for a small velocity @xmath70 , we were always able to determine a critical value of the impact parameter , @xmath71 , for the merger of two black holes .
namely , for @xmath72 , two black holes merge to a single spinning black hole whereas for @xmath73 , two black holes go apart to infinity after one scattering .
the zoom - whirl orbit was never found in 5 dimensions in contrast to the 4d case @xcite , because the law of the gravitational force is modified : note that in the newtonian limit in 5 dimensions , both gravitational and centrifugal forces are proportional to @xmath74 where @xmath75 is the separation of two objects . for a high velocity with @xmath76 , by contrast , we were not able to determine the value of @xmath68 using our current code .
the reason is that for @xmath77 , the numerical simulation crashed soon after the scattering of two black holes occurs .
figure 1 plots @xmath67 and @xmath65 as functions of @xmath33 ( note that for @xmath78 , @xmath79 ) .
however , we were able to confirm that for @xmath80 , two black holes merges , while for @xmath81 , the merger does not happen and two black holes merely go apart to infinity after the scattering . in this paper , we focus on the case @xmath82 .
[ cols="^,^,^ " , ] [ fig : fig2 ] figure 2 plots the time variation of @xmath83 in a scattering process with @xmath84 and @xmath85 .
the value of @xmath83 is shown in the unit of @xmath86 in the @xmath87-@xmath88 plane . in this figure , the apparent horizon of each black hole
is denoted by the solid circles .
when the separation between two black holes is much larger than @xmath58 , no super - planckian domain emerges outside the apparent horizons as long as the mass of each black hole @xmath17 is larger than @xmath8 [ see fig . 2(a ) ] .
by contrast , when the separation is equal to 2@xmath89 , a highly elongated domain with a large value of @xmath90 is formed between two black holes [ see fig .
finally , two black holes are scattered away , and the domain with a large value of @xmath83 disappears [ see fig . 2(c ) ] . at the center of mass as a function of the coordinate separation between two black holes .
the parameters @xmath33 and @xmath91 are the same as those for fig . 2 .
the time proceeds from left to right .
the results with three grid resolutions are plotted.,title="fig : " ] [ fig : fig3 ] figure 3 plots @xmath92 at the center of mass @xmath93 as a function of the coordinate separation between two black holes . in this figure
, the time elapses from left to right
. initial speed of each black hole and the impact parameter are the same as those for fig .
2 . as the separation between two black holes becomes small
, the value of @xmath83 steeply increases .
then , after the black holes slightly goes through the periastron , the maximum value of @xmath83 is reached .
the maximum value @xmath94 in the best resolution run is @xmath95 in this scattering process , the event horizons of these black holes do not merge with each other , because these black holes go away toward infinity separately after this scattering .
hence , by the symmetry of this system , the center of mass is not enclosed by the event horizon .
this fact implies that even if each black hole is a classical object before the scattering , a visible super - planckian domain can emerge in the vicinity of the center of mass in 5d general relativity .
calculated in the scattering of @xmath96 and @xmath97 at the center of mass as a function of @xmath33 .
the results only for @xmath81 are plotted .
the pluses and the crosses show the results of @xmath98 , and the circles and the squares show the results of @xmath99.,title="fig : " ] [ fig : fig4 ] figure 4 plots the maximum values of @xmath83 for the scattering of a fixed impact parameter at the center of mass as a function of @xmath33 .
the results only for @xmath81 are plotted .
this shows that the maximum value of @xmath100 increases steeply with @xmath33 , and thus , the super - planckian domain appears to be always visible in high - velocity collisions .
this also suggests that for @xmath101 , the maximum value of @xmath100 would be much larger than @xmath102 .
probing the short - distance physics would be limited if scattering processes at energies well above the planck scale were hidden behind a black hole horizon
. if this hypothesis is correct , possible productions of small black holes in particle colliders , such as the cern large hadron collider , could be probably the unique opportunity for exploring the nature of the quantum gravity @xcite .
however , as shown in this paper , a visible super - planckian domain can be generated even through a scattering process of two classical black holes in 5d spacetime .
we may say in the practical sense that the cosmic censorship does not hold in 5d spacetime , because a super - planckian domain should be regarded as an effective singularity in the classical gravity .
the visible super - planckian domain is formed even for @xmath103 , i.e. , in a mildly relativistic scattering , and visible super - planckian domains can be generated through gravitational scatterings more efficiently than a guess by nhm . it should be emphasized that such domain could emerge even in the absence of special spacetime symmetry or special assumption of spacetime geometry .
the present result implies that it is necessary to study quantum gravitational effects in the super - planckian domain of no horizon .
although , at present , we do not know what happens in the super - planckian domain , we expect that the semi - classical particle creation in the sub - planckian region could occur around the visible super - planckian domain @xcite . finally , we comment on the scatterings for @xmath104 with @xmath105 for which our numerical simulations do not keep sufficient accuracy probably due to the emergence of a very large spacetime curvature .
we performed simulations for @xmath81 with several grid resolutions .
the maximum value of @xmath83 becomes larger for the finer grid resolution for a given value of @xmath91 .
it also increases with approaching @xmath65 .
these facts suggest that ultra - high spacetime curvature , which is not encompassed in a horizon , may be realized at some impact parameter for @xmath106 , i.e. , the formation of a naked singularity .
we would like to explore this issue in more detail elsewhere .
we thank t. tanaka for helpful duscussions .
kn is grateful to c - m .
yoo , h. ishihara , and colleagues in astrophysics and gravity group at osaka city university for helpful discussions .
ho thanks s. kinoshita and n. sago , and colleagues in astrophysics and cosmology group at yukawa institute for theoretical physics for helpful discussions .
this work was supported by grant - in - aid for scientific research ( 21340051 ) and by grant - in - aid for scientific research on innovative area ( 20105004 ) of japanese mext .
99 n. arkani - hamed , s. dimopoulos and g. dvali , phys .
b * 429 * , 263 ( 1998 ) ; i. antoniadis , n. arkani - hamed , s. dimopoulos and g. dvali , phys . lett .
b * 436 * , 257 ( 1998 ) .
l. randall and r. sundrum , phys .
lett . * 83 * , 3370 ( 1999 ) ; phys .
lett . * 83 * , 4690 ( 1999 ) .
s. b. giddings and s. thomas , phys .
d * 65 * , 056010 ( 2002 ) .
r. penrose , riv .
nuovo cim . *
i * , 252 ( 1969 ) ; reprinted in gen .
. grav . * 34 * , 1141 ( 2002 ) .
k. nakao , t. harada and u. miyamoto , phys .
d * 82 * , 121501 ( r ) ( 2010 ) .
m. shibata and h. yoshino , phys .
d * 81 * , 021501 ( r ) ( 2010 ) : _ ibid _ , * 81 * , 104035 ( 2010 ) . m. zilhao , h. witek , u. sperhake , v. cardoso , l. gualtieri , c. herdeiro and a. nerozzi , phys .
d * 81 * , 084052 ( 2010 ) .
l. lehner and f. pretorius , phys .
lett * 105 * , 101102 ( 2010 ) .
e. sorkin , phys .
d * 81 * , 084062 ( 2010 ) .
h. witek , m. zilhao , l. gualtieri , v. cardoso , c. herdeiro , a. nerozzi and u. sperhake , phy .
d * 82 * , 104014 ( 2010 ) .
u. sperhake , v. cardoso , f. pretorius , e. berti and f.a .
gonzlez , phys .
* 101 * , 161101 ( 2008 ) .
m. shibata , h. okawa and t. yamamoto , phys .
d * 72 * , 101501 ( 2008 ) .
u. sperhake , v. cardoso , f. pretorius , e. berti , t. hinderer and n. yunes , phys .
103 * , 131102 ( 2009 ) .
r. m. wald , _ general relativity _
( the university of chicago press , chicago , 1984 ) . m. shibata and t. nakamura , phys . rev .
d * 52 * , 5428 ( 1995 ) : t.w .
baumgarte and s.l .
shapiro , phys .
d * 59 * , 024007 ( 1998 ) .
m. campannelli , c.o .
loust , p. marronetti and y. zlochower , phys .
lett . * 96 * , 111101 ( 2006 ) ; j. baker , j. centrella , d .-
i choi , m. koppitz and j. van meter , phys .
* 96 * , 111102 ( 2006 ) . t. yamamoto , m. shibata , k. taniguchi , phys . rev .
d. * 78 * , 064054 ( 2008 ) .
s. dimopoulos and g. landsberg , phys .
* 87 * , 161602 ( 2001 ) .
u. miyamoto , h. nemoto , and m. shimano , arxiv:1012.2374 . | it may be widely believed that probing short - distance physics is limited by the presence of the planck energy scale above which scale any information is cloaked behind a horizon .
if this hypothesis is correct , we could observe quantum behavior of gravity only through a black hole of planck mass .
we numerically show that in a scattering of two black holes in the 5-dimensional spacetime , a visible domain , whose curvature radius is much shorter than the planck length , can be formed .
our result indicates that super - planckian phenomena may be observed without an obstruction by horizon formation in particle accelerators . |
thin and flat few layer hbn flakes were prepared by mechanical exfoliation of hbn single crystals on sio@xmath0/si substrates .
the hbn growth method has been previously described@xcite .
exfoliated graphene flakes were then transferred to the hbn using poly(methyl methacrylate ) ( pmma ) as a carrier @xcite .
then cr / au electrodes were deposited using standard electron beam lithography .
the lithography process leaves some pmma resist on the surface of graphene which is cleaned by annealing in argon and hydrogen at 350@xmath4 c for 3 hours @xcite .
the device was then immediately transferred to the stm ( omicron low temperature stm operating at t = 4.5 k in ultrahigh vacuum ( p @xmath36 mbar ) ) .
electrochemically etched tungsten tips were used for imaging and spectroscopy .
all of the tips used were first checked on an au surface to ensure that their density of states was constant .
the di / dv spectroscopy was acquired by turning off the feedback loop and holding the tip a fixed distance above the surface .
a small ac modulation of 5 mv at 563 hz was applied to the tip voltage and the corresponding change in current was measured using lock - in detection .
we also measured di / dv curves with 0.5 mv excitation and observed the same results .
28 geim , a. k. and novoselov , k. s. the rise of graphene . _ nature mater . _
* 6 * , 183 - 191 ( 2007 ) .
castro neto , a. h. , guinea , f. , peres , n. m. r. , novoselov , k. s. , and geim , a. k. the electronic properties of graphene .
phys . _ * 81 * , 109 - 162 ( 2009 ) .
martin , j. _ et al .
_ observation of electron - hole puddles in graphene using a scanning single - electron transistor .
_ nature phys . _ * 4 * , 144 - 148 ( 2008 ) .
deshpande , a. , bao , w. , miao , f. , lau , c. n. , and leroy , b. j. spatially resolved spectroscopy of monolayer graphene on sio@xmath0 .
b _ * 79 * , 205411 ( 2009 ) .
zhang , y. , brar , v. w. , girit , c. , zettl , a. , and crommie , m. f. origin of spatial charge inhomogeneity in graphene .
_ nature phys . _ * 5 * , 722 - 726 ( 2009 ) .
du , x. , skachko , i. , barker , a. , and andrei , e. y. approaching ballistic transport in suspended graphene .
_ nature nanotech .
_ * 3 * , 491 - 495 ( 2008 ) .
bolotin , k. i. , sikes , k. j. , hone , j. , stormer , h. l. , and kim , p. temperature - dependent transport in suspended graphene .
_ * 101 * , 096802 ( 2008 )
. dean , c. r. _ et al .
_ boron nitride substrates for high - quality graphene electronics .
_ nature nanotech . _ * 5 * , 722 - 726 ( 2010 ) .
giovannetti , g. , khomyakov , p. a. , brocks , g. , kelly , p. j. , and van den brink , j. substrate - induced band gap in graphene on hexagonal boron nitride : _ ab initio _ density functional calculations .
b _ * 76 * , 073103 ( 2007 ) .
slawinska , j. , zasado , i. , and klusek , z. energy gap tuning in graphene on hexagonal boron nitride bilayer system .
b _ * 81 * , 155433 ( 2010 ) .
novoselov , k.s .
_ et al . _
two - dimensional atomic crystals .
usa _ * 102 * , 10451 - 10453 ( 2005 ) .
chen , j .- h .
_ charged - impurity scattering in graphene .
_ nature phys . _ * 4 * , 377 - 381 ( 2008 ) .
rossi , e. and das sarma , s. ground state of graphene in the presence of random charged impurities .
lett . _ * 101 * , 166803 ( 2008 ) .
du , x. , skachko , i. , duerr , f. , luican , a. , and andrei , e. y. fractional quantum hall effect and insulating phase of dirac electrons in graphene .
_ nature _ * 462 * , 192 - 195 ( 2009 ) .
bolotin , k. i. , ghahari , f. , shulman , m. d. , stormer , h. l. , and kim , p. observation of the fractional quantum hall effect in graphene .
_ nature _ * 462 * , 196 - 199 ( 2009 ) .
bao , w. _ et al .
_ controlled ripple texturing of suspended graphene and ultrathin graphite membranes .
_ nature nanotech .
_ * 4 * , 562 - 566 ( 2009 ) .
liu , l. , feng , y. p. , and shen , z. x. structural and electronic properties of h - bn . _ phys .
b _ * 68 * , 104102 ( 2003 ) .
watanabe , k. , taniguchi , t. , and kanda , h. direct - bandgap properties and evidence for ultraviolet lasing of hexagonal boron nitride single crystal .
_ nature mater . _ * 3 * , 404 - 409 ( 2004 ) .
ishigami , masa , chen , j. h. , cullen , w. g. , fuhrer , m. s. , and williams , e. d. atomic structure of graphene on sio@xmath0 .
_ nano lett . _ * 7 * , 1643 - 1648 ( 2007 ) .
stolyarova e. _ et al .
_ high - resolution scanning tunneling microscopy imaging of mesoscopic graphene sheets on an insulating surface .
* 104 * , 9209 - 9212 ( 2007 ) .
lui , c. h. , liu , l. , mak , k. f. , flynn g. w. , and heinz , t. f. ultraflat graphene .
_ nature _ * 462 * , 339 - 341 ( 2009 ) .
kim , e. a. and castro neto , a. h. graphene as an electronic membrane .
_ europhys .
lett . _ * 84 * , 57007 ( 2009 ) .
guinea , f. , katsnelson , m. i. , and vozmediano , m. a. h. midgap states and charge inhomogeneities in corrugated graphene .
b _ * 77 * , 0705422 ( 2008 ) .
young , a. f. _ et al .
_ electronic compressibility of gapped bilayer graphene .
preprint at arxiv:1004.5556v2 ( 2010 ) .
choudhury , s. k. and gupta , a. k. tip induced doping effects in local tunnel spectra of graphene .
preprint at arxiv:1008.3606v1 ( 2010 ) .
zhang , l. m. _ et al .
_ determination of the electronic structure of bilayer graphene from infrared spectroscopy .
b _ * 78 * , 235408 ( 2008 ) . taniguchi , t. and watanabe , k. synthesis of high - purity boron nitride single crystals under high pressure by using ba - bn solvent .
. growth _ * 303 * , 525 - 529 ( 2007 ) .
meyer , j. c. _ et al . _ the structure of suspended graphene sheets . _
nature _ * 446 * , 60 - 63 ( 2007 ) .
the authors would like to thank w. bao , z. zhao and c.n .
lau for providing the graphene on sio@xmath0 samples used for the comparison with graphene on hbn .
, a. d. and b.j.l . were supported by u. s. army research laboratory and the u. s. army research office under contract / grant number w911nf-09 - 1 - 0333 and the national science foundation career award dmr-0953784 .
p.j . was supported by the national science foundation under award dmr-0706319 .
j.s .- y . , d.b . and
p. j .- h .
were supported by the u.s .
department of energy , office of basic energy sciences , division of materials sciences and engineering under award # de - sc0001819 and by the 2009 u.s .
office of naval research multi university research initiative ( muri ) on graphene advanced terahertz engineering ( gate ) at mit , harvard and boston unversity .
j.x . and b.j.l .
performed the stm experiments of the graphene on hbn .
a.d . performed the stm experiments of graphene on sio@xmath0 .
j. s .- y .
and d. b. fabricated the devices .
p.j . performed the theoretical calculations .
k.w . and t.t .
provided the single crystal hbn .
j .- h . and b.j.l .
conceived and provided advice on the experiments .
all authors participated in the data discussion and writing of the manuscript .
the authors declare no competing financial interests .
a moir pattern occurs when the atoms in the graphene layer form a super - lattice structure with the atoms in the hbn layer . in this section
, we derive the conditions which lead to a moir pattern and therefore predict the angles and lengths of the moir pattern .
we consider two superimposed hexagonal lattices defined by @xmath37 with the layer index @xmath38 for the graphene and hbn lattices , respectively , and the sublattice index @xmath39 ( a sublattice ) , 2 ( b sublattice ) . in the hbn layer ,
@xmath39 for nitrogen and 2 for boron .
we define @xmath40 as the vector connecting the two sublattices , with lattice spacing @xmath41 .
the vectors lattice vectors associated with the hbn , @xmath42 , are rotated by an angle @xmath43 counterclockwise with respect to the graphene lattice @xmath44 and are longer by a factor @xmath45 .
therefore the hbn lattice vectors are given by @xmath46 an arbitrary point in the graphene lattice can be represented by two integers @xmath47 which correspond to the number of lattice vectors in the @xmath48 and @xmath49 directions respectively .
similarly , a position in the hbn lattice is represented by the integers @xmath50 .
if the graphene and hbn lattice are ab stacked at a given position , they will be ab stacked again when @xmath51 . in terms of @xmath52-coordinates
this gives the conditions @xmath53 @xmath54 a moir pattern occurs when these equations can be satisfied for integer values of @xmath47 and @xmath50 .
for a given ratio of @xmath55 these conditions will only hold for certain special angles . in terms of the values of @xmath47 , the length of the moir pattern can be written as @xmath56 and it is at an angle @xmath57 with respect to @xmath48 . however
, a near commensurate condition can always be found leading to a moir pattern over some finite length .
we have numerically created lattices to reproduce the images in figure 2 .
the results are shown in fig .
[ sifig ] .
the first set of images correspond to the data shown in fig .
2a and 2b in the main text .
it was created by using a hbn lattice with a 1.8% longer lattice constant and rotated by @xmath58 counterclockwise with respect to the graphene lattice . from the fft of the lattice , fig .
[ sifig]b , we see that it matches the experimental figure , fig .
2b very well . to create the shorter moir pattern we must use a larger rotation angle .
we find the best match occurs at @xmath59 . in general , the moir pattern gets shorter as the angle @xmath43 increases . .
( c ) simulated lattice showing a shorter moir pattern .
the scale bar is 2 nm .
( d ) fourier transform of ( c ) showing the atomic lattice as well as the moir pattern .
the scale bar is 10 nm@xmath6 ]
we restrict the interlayer hopping potential to nearest - neighbor and next - nearest - neighbor hopping , and evaluate @xmath61 \ , , \end{aligned}\ ] ] with @xmath62 labelling the nearest and next - nearest neighbor sites to @xmath63 . note that if one is a boron atom , the other one is a nitrogen atom . in this way
, four interlayer couplings between the two sublattices are defined .
the parameters @xmath64 ev and @xmath17 nm are calibrated to fit the interlayer couplings in bilayer graphene@xcite . while @xmath21 should in principle depend on the sublattice index @xmath65 in the hbn layer , this has no influence on our main conclusion , that the graphene spectrum is effectively gapless due to the mismatch between lattices and their different relative orientations , because it does not break sublattice symmetry in the graphene layer .
we consider system sizes up to @xmath66 unit cells , which is more than sufficient to extract the periodicity of the moir patterns shown in figs .
2a and 2c . fig .
[ vmunu]a shows a color plot of @xmath67 for an interlayer rotation angle of @xmath68 corresponding to the sample shown in fig .
2a . the potential has a clear hexagonal periodicity , with a wavelength of about 10 lattice sites , in quantitative agreement with fig .
2a . from that hexagonal pattern ,
the fourier transform @xmath69 of @xmath70 generically exhibits seven peaks , including a large one at @xmath71 , and six secondary peaks at wave vectors @xmath72 , @xmath73 corresponding to the moir pattern .
this is shown in fig .
[ vmunu]b , with hexagonally distributed secondary peaks at positions reflecting the central pattern of fig .
potential for an interlayer rotation of @xmath68 .
the scale bar is 2 nm .
( b ) fourier transform of ( a ) showing a central peak at @xmath71 and six additional peaks due to the moir pattern .
the scale bar is 0.8 nm@xmath6 . ]
the hopping potential induces interlayer coupling which we introduce in the low - energy hamiltonian . we keep only the seven dominant transitions and because the hbn bands are several ev s away from the dirac points and @xmath74 , we truncate the hamiltonian to a 16 @xmath75 16 matrix @xmath76 \delta_{{\bf q},{\bf q } ' } + \left[\sum_{i=0}^6 h_{\rm c , bn}({\bf k}_i ) + h.c . \right ]
\delta_{{\bf q}+{\bf k}_i,{\bf q } ' } \ , , \end{aligned}\ ] ] with all @xmath77 s being @xmath78 matrices , with @xmath80 and parameters @xmath81 ev , @xmath82 ev , @xmath83 ev and @xmath84 ev@xcite . the low - energy graphene dispersion is obtained via diagonalization of @xmath85 .
we find that a gap exists in the spectrum only when @xmath86 for at least one @xmath65 , thereby breaking sublattice symmetry in the graphene layer . at experimentally relevant rotation angles @xmath43 of a few degrees
, we find that @xmath87/ \tilde{v}_{\mu \nu } ( { \bf k}_i ) \lesssim 10^{-4 } $ ] for 150 @xmath75 150 systems , decreasing with size to less than @xmath88 for 600 @xmath75 600 systems .
consequently , we get an upper bound of @xmath89 ev for the excitation gap @xmath90 between the two graphene bands for the largest systems we investigated .
the latter are still smaller than the experimentally observed domains so that such a small gap can not be resolved .
large values for @xmath87/ \tilde{v}_{\mu \nu } ( { \bf k}_i)$ ] are obtained only for small systems or if the lattice mismatch is neglected .
this explains the 50 mev gap reported in refs .
@xcite , which we qualitatively reproduced ( see fig .
3d ) . even without relative rotation of the two layers
, we found that the lattice mismatch is sufficient to effectively close this gap already for systems of sizes 300 @xmath75 300 .
we finally note that the expected different hoppings on the boron and nitrogen atoms induce differences in @xmath91 vs. @xmath92 as well as in @xmath93 vs. @xmath94 .
however , such differences do not break sublattice symmetry and therefore do not open a gap . | graphene has demonstrated great promise for future electronics technology as well as fundamental physics applications because of its linear energy - momentum dispersion relations which cross at the dirac point@xcite .
however , accessing the physics of the low density region at the dirac point has been difficult because of the presence of disorder which leaves the graphene with local microscopic electron and hole puddles@xcite , resulting in a finite density of carriers even at the charge neutrality point .
efforts have been made to reduce the disorder by suspending graphene , leading to fabrication challenges and delicate devices which make local spectroscopic measurements difficult@xcite .
recently , it has been shown that placing graphene on hexagonal boron nitride ( hbn ) yields improved device performance@xcite . in this letter , we use scanning tunneling microscopy to show that graphene conforms to hbn , as evidenced by the presence of moir patterns in the topographic images . however , contrary to recent predictions@xcite , this conformation does not lead to a sizable band gap due to the misalignment of the lattices .
moreover , local spectroscopy measurements demonstrate that the electron - hole charge fluctuations are reduced by two orders of magnitude as compared to those on silicon oxide .
this leads to charge fluctuations which are as small as in suspended graphene@xcite , opening up dirac point physics to more diverse experiments than are possible on freestanding devices .
graphene was first isolated on silicon dioxide because of the ability to image monolayer regions using an optical microscope@xcite .
however , the electronic properties of sio@xmath0 are not ideal for graphene because of its high roughness and trapped charges in the oxide .
these impurity - induced charge traps tend to cause the graphene to electronically break up into electron and hole doped regions at low charge density which both limit device performance and make the dirac point physics inaccessible@xcite . in order to create devices with less puddles ,
the substrate must be removed or changed .
one possibility to get rid of substrate interactions is to suspend graphene@xcite , as shown by the drastic improvement in mobility which has enabled the observation of the fractional quantum hall effect in suspended devices@xcite .
however , the freely suspended monolayers are very delicate , leading to fabrication difficulties as well as strain@xcite . because of these difficulties
, there have been no stm spectroscopy measurements of the local electronic properties of suspended graphene devices .
all of this points to the need for new substrates that offer mechanical support to the graphene without interfering with its electrical properties .
recently , such a candidate substrate has been found with the demonstration of high - quality graphene devices on hexagonal boron nitride ( hbn)@xcite . hexagonal boron nitride has the same atomic structure as graphene , but with a 1.8% longer lattice constant@xcite , and shares many similar properties with graphene except that it is a wide - bandgap electric insulator@xcite . the planar structure of hbn cleaves into an ultra - flat surface and the ionic bonding of hbn should leave it free of dangling bonds and charge traps at the surface resulting in less induced electron - hole puddles in graphene .
indeed , graphene on hbn devices exhibit the highest mobility reported on any substrate , as well as narrow dirac peak resistance widths , indicating reduced disorder and charge inhomogeneity@xcite . to study how the local electronic structure of graphene is affected by the hbn substrate
, we prepare graphene on hbn devices for stm measurements .
a schematic of the measurement set - up showing the graphene flake on hbn with gold electrodes for electrical contact is shown in fig .
[ topography]a .
a typical stm image of the monolayer graphene showing the surface corrugations due to the underlying hbn substrate is shown in fig .
[ topography]b .
this image can be compared with an stm image of monolayer graphene prepared in a similar manner but on sio@xmath0 as shown in fig .
[ topography]c .
it is clear from these two images that the surface corrugations are much larger for graphene on sio@xmath0 as compared to hbn .
this is due to the graphene conforming to the substrate and the planar nature of hbn as compared to the amorphous sio@xmath0 .
figure [ topography]d shows a histogram of the heights in the two images . in both cases , the heights are well described by gaussian distributions with standard deviations of 224.5 @xmath1 0.9 pm for graphene on sio@xmath0 and 30.2 @xmath1 0.2 pm for graphene on hbn .
the values for graphene on sio@xmath0 are similar to previously reported values @xcite while the distribution for graphene on hbn is similar to graphene on mica or hopg@xcite .
reducing the surface roughness is critical for graphene devices because local curvature can lead to electronic effects such as doping@xcite and random effective magnetic fields@xcite .
as the height variation of the graphene on hbn is as flat as hopg , it has reached its ultimate limit of flatness . .
( a ) optical microscope image of the mechanically exfoliated monolayer graphene flake with gold electrodes .
the wiring of the stm tip and back gate voltage is indicated .
( b ) stm topographic image of monolayer graphene on hbn showing the underlying surface corrugations .
the image is 100 nm x 100 nm . the imaging parameters are tip voltage @xmath2 = -0.3 v , tunneling current @xmath3 = 100 pa .
( c ) stm topographic image of monolayer graphene on sio@xmath0 showing significantly increased corrugations .
the imaging parameters are tip voltage @xmath2 = -0.5 v , tunneling current @xmath3 = 50 pa .
( d ) histogram of the height distributions for graphene on sio@xmath0 ( blue squares ) and graphene on hbn ( red triangles ) along with gaussian fits .
[ topography ] ] looking more closely at the topography of the graphene on hbn , we resolve its atomic lattice and also observe longer periodic modulations which result in a distinct moir pattern as seen in figs .
[ fft]a and [ fft]c .
these two images were acquired from different areas of the same graphene flake and show moir patterns of different length due to a shift in the alignment of the graphene with the underlying hbn lattice . in the case of figs .
[ fft]a and [ fft]c , we find a long wavelength modulation of 2.6 nm and 1.3 nm respectively . by examining the fourier transform of these images
, we can learn about the underlying hbn substrate as well as its relative orientation with respect to the graphene lattice .
in both cases , we see two distinct sets of peaks in the fft , there is a set of six points near the center of the images which correspond to the moir pattern and there are six additional points near the edge of the images which correspond to the graphene atomic lattice .
the first observation is that the locations of these points is rotated with respect to each other in the two images . for fig .
[ fft]b , the atomic lattice is rotated by 12.6 @xmath1 1.0@xmath4 from the horizontal . in the case of fig .
[ fft]d , the atomic lattice is rotated by 18.5 @xmath1 0.6@xmath4 . in contrast
the moir pattern is rotated by 29.9 @xmath1 0.1@xmath4 from the horizontal in fig .
[ fft]b and 38.4 @xmath1 0.2@xmath4 from the horizontal in fig .
[ fft]d . from the lengths of the moir patterns and their angles
, we can calculate the orientation of the graphene lattice with respect to the underlying hbn ( see supplementary information ) . in the case of fig .
[ fft]a , we find that the hbn is rotated by -5.4@xmath4 from the graphene . on the other hand , for fig .
[ fft]c , we find that the hbn is rotated by -10.9@xmath4 .
this difference of 5.5@xmath4 matches the difference in the orientation of the graphene lattices in the two images .
therefore , we conclude that the underlying hbn substrate is continuous and the graphene above it sits at two different angles .
atomic force microscopy images show that graphene on hbn tends to form flat regions separated by ridges and pyramids .
as the two stm images were taken from different sides of one of these ridges , it is clear that the graphene can change orientation across these ridges .
= -0.3 v and i@xmath5 = 100 pa .
( b ) fourier transform of ( a ) showing the six graphene lattice points near the edge of the image and the long wavelength moir pattern near the center of the image and around each lattice point .
the scale bar is 10 nm@xmath6 .
the inset is a zoom in around one of the lattice points .
its scale bar is 2 nm@xmath6 .
( c ) stm topography image from another region of the same graphene flake showing a different moir pattern .
the scale bar is 2 nm .
the inset is a zoom in of a 2 nm region and the scale bar is 0.3 nm .
the imaging parameters are v@xmath5 = -0.3 v and i@xmath5 = 100 pa .
( d ) fourier transform of ( c ) showing the atomic lattice as well as the moir pattern .
the scale bar is 10 nm@xmath6 .
the inset is a zoom of the moir pattern and its scale bar is 4 nm@xmath6 .
[ fft ] ] the scanning tunneling microscope is not only able to acquire images of atoms but can also map the local density of states . we have performed scanning tunneling spectroscopy of the graphene on hbn .
figure [ fermivelocity]a shows a typical di / dv spectroscopy curve which is proportional to the local density of states .
the curve is nearly linear in energy with a minimum near zero tip voltage indicating the energy of the dirac point .
the location of this minimum can be varied by applying a back gate voltage , @xmath7 , to the sample .
the voltage on the back gate induces a charge on the graphene of @xmath8 with @xmath9 e/@xmath10v based on a parallel plate capacitor model . in the model
, we have taken 285 nm of sio@xmath0 with a dielectric constant of 3.9 and 14 nm of hbn with a dielectric constant of 3 - 4@xcite .
there is a small uncertainty in the value of @xmath11 of about 1% due to the unknown precise value of the hbn dielectric constant .
figure [ fermivelocity]b plots di / dv as a function of tip voltage and gate voltage .
the white line follows the minimum in the di / dv curves for each value of the gate voltage .
this minimum occurs when the fermi energy of the tip lines up with the dirac point .
we observe that the location of the minimum changes more quickly when the dirac point is near zero tip voltage which is consistent with the linear band structure of graphene .
there is also a dark ridge that occurs at decreasing tip voltage as the gate voltage is increased .
this is due to the effect of the voltage on the tip acting as a local gate and changing the density of electrons in the graphene@xcite .
angle mismatch and the red curve has -10.9@xmath12 .
[ fermivelocity ] ] our local spectroscopy measurements indicate that there is no band gap induced in graphene on hbn , not even locally .
these results disagree with earlier theoretical calculations which predicted the opening of a band gap of order 50 mev when graphene is placed on hbn , because of the breaking of sublattice symmetry@xcite .
this discrepancy is explained by the 1.8% mismatch between the graphene and hbn lattices and the different orientations of the two lattices , which were both neglected in refs .
@xcite . taking them into account
, one expects that , while one of the carbon atoms may sit over a boron ( nitrogen ) atom at one location , this alignment gets lost a few lattice constants away . in large enough systems , carbon atoms should therefore have the same probability to have a boron or a nitrogen atom as nearest neighbor in the hbn layer , regardless of their sublattice index .
we numerically checked the validity of this hypothesis by calculating the interlayer hopping potential @xmath13 $ ] from a carbon atom at @xmath14 in the graphene layer to a boron or nitrogen atom at @xmath15 in a rotated hbn layer .
we restricted ourselves to nearest- and next - nearest - neighbor interlayer hopping and chose the parameters @xmath16 ev and @xmath17 nm to fit known values for these hoppings in graphene bilayers@xcite .
we found that the 1.8% lattice mismatch alone is sufficient to make the hopping strength from a carbon atom to a boron or a nitrogen atom independent of the graphene sublattice index for systems of a few hundred unit cells , going down to a few tens of unit cells when the lattices are misaligned by about one degree .
we incorporated the fourier transform of this hopping potential into the low - energy hamiltonian for graphene on hbn to find the energy - momentum dispersion .
the inter - layer coupling is nonzero only for @xmath18 as well as for six additional vectors @xmath19 associated with the moir pattern .
most importantly , we found that the coupling between the a and b atoms in the graphene lattice with the boron and nitrogen atoms in the hbn are almost identical .
thus sublattice symmetry is restored and a gapless dirac spectrum is recovered , albeit at slightly shifted values of @xmath20 .
this is illustrated in fig .
[ fermivelocity]d .
while @xmath21 depends in principle on whether hopping occurs between a carbon and a boron or nitrogen atom , we note that this does not break sublattice symmetry , and we checked that the spectrum remains gapless , even when this discrepancy is taken into account .
more details of our numerical approach are given in the supplementary information . by determining the energy of the dirac point as a function of gate voltage , we can measure the fermi velocity of electrons and holes in graphene .
figure [ fermivelocity]c shows the energy of the dirac point as a function of gate voltage .
graphene has a linear dispersion relation such that @xmath22 where @xmath23 is the fermi velocity .
since it is a two - dimensional material , the density of electrons is given by @xmath24 where @xmath25 and @xmath26 are the spin and valley degeneracy which are both 2 for graphene .
therefore , the dirac point should depend on gate voltage as @xmath27 .
the red curve is a fit to the data from which we can extract the fermi velocity .
we find that @xmath28 m / s for the electrons and @xmath29 m / s for the holes .
moreover , we observe an asymmetry between the fermi velocity for electrons and holes of about 25% depending on the moir pattern observed .
the shorter moir pattern has a higher fermi velocity for holes while the longer one has a higher fermi velocity for electrons .
the origin of this asymmetry is unclear but may arise due to next - nearest neighbor coupling which are not taken into account in our model .
one of the main advantages to using hbn as a substrate for graphene as compared to sio@xmath0 is the improvement in electronic properties of the graphene which is believed to be due to the lack of charge traps on the hbn surface .
figure [ spatialmap]a shows the topography of graphene on hbn over a range of 100 nm .
note that the height variation is less than 0.1 nm over the range of the image as compared to typical values of nearly 1 nm for graphene on sio@xmath0 .
we have performed di / dv measurements at 1 nm intervals over the entire area of fig .
[ spatialmap]a . for each of these di / dv curves ,
we have found tip voltage of the minimum which corresponds to the dirac point .
the results are plotted in fig .
[ spatialmap]b .
we have done a similar analysis for a 100 nm area of graphene on sio@xmath0 and the results are plotted in fig .
[ spatialmap]c .
the red and blue regions correspond to electron and hole puddles respectively .
it is clear from these two images that the variation in the energy of the dirac point is much smaller on hbn .
the spatial extent of each puddle is also much smaller in the graphene on sio@xmath0 consistent with an increased density of impurities@xcite .
( a ) topography of graphene on hbn .
( b ) tip voltage at the dirac point as a function of position for graphene on hbn .
( c ) tip voltage at the dirac point as a function of position for graphene on sio@xmath0 .
the color scale is the same for ( b ) and ( c ) .
( d ) histogram of the energies of the dirac point from ( b ) as well as a gaussian fit .
the inset shows the same data but also includes the histogram for sio@xmath0 shown in red .
the scale bar in all images is 10 nm .
[ spatialmap ] ] we can further quantify the disorder in the graphene by looking at a histogram of the energy of the dirac point , fig .
[ spatialmap]d .
the main part of the histogram for the dirac point energy on hbn is well fit by a gaussian distribution ( red line ) with a standard deviation of @xmath30 mev .
in addition , there is a small extra bump in the distribution from the hole doped region near the bottom right of fig .
[ spatialmap]b . in comparison ,
the distribution on sio@xmath0 is much broader with a standard deviation of @xmath31 mev .
these distributions in energy can be converted to charge fluctuations using @xmath32 .
we find that the charge fluctuations in graphene on hbn are @xmath33 @xmath34 while they are more than 100 times larger for graphene on sio@xmath0 , @xmath35 @xmath34 .
our measurements for the charge fluctuations on sio@xmath0 are consistent with previous single electron transistor@xcite and stm@xcite measurements which established the presence of electron and hole puddles in graphene on sio@xmath0 .
furthermore , our measurements for the charge fluctuations in graphene on hbn show a very similar value to values extracted from electrical transport measurements in suspended graphene samples@xcite implying that using hbn as a substrate provides a similar benefit to suspending graphene without the associated fabrication challenges and limitations . we have demonstrated that graphene on hbn provides an extremely flat surface that has significantly reduced electron - hole puddles as compared to sio@xmath0 . by reducing the charge fluctuations ,
the low density regime and the dirac point can be more readily accessed .
moreover , hbn allows this low - density regime to be reached in a substrate - supported system which will allow atomic resolution local probes studies of the dirac point physics . |
the evolution of qcd amplitudes with energy is described by the balitsky hierarchy @xcite or , equivalently , by the jimwlk @xcite functional renormalization group equations .
they essentially represent generalizations of the well - known bfkl equation @xcite for the evolution of the two - point function to evolution equations for arbitrary @xmath4-point functions including the non - linear effects due to high gluon density . in the unitarity limit of high parton density
the balitsky hierarchy is not closed : the derivative of any @xmath4-point function with respect to energy ( or rapidity @xmath5 ) involves all @xmath6-point functions ( @xmath7 ) . in the dilute regime , however , the hierarchy can be truncated to obtain closed evolution equations for each @xmath4-point function .
prior work in this field has mostly focused on the evolution of the two - point function and its perturbative unitarization at high energies .
the purpose of this paper is to point out that information on the four - point function could be obtained from two - particle correlations in inelastic high - energy collisions in a certain kinematic regime ( see below ) .
moreover , we argue that the b - jimwlk equation for the four point function can not be factorized as a product of two bfkl two - point functions .
we show that the terms that violate this factorization actually contribute to the correlation function at leading order in @xmath1 .
we consider the correlation of two particles with transverse momenta @xmath8 , @xmath9 ( we shall drop the subscript @xmath10 from now on to avoid cluttering of notation ) and rapidities @xmath11 , @xmath12 , respectively : c(*p*,*q * ) = - .
the brackets denote an average over events and the momentum distributions shall be normalized according to d^2pdy_p & = & n , + d^2pdy_p d^2qdy_q & = & n^2 , where @xmath13 is the total average multiplicity per event . it has been argued in ref .
@xcite that in the high - energy limit ( but fixed @xmath14 , @xmath15 , @xmath11 , @xmath12 ) the leading contribution to @xmath16 is due to diagrams such as the one depicted in fig .
[ fig : twopointc ] . for these diagrams
the hard amplitudes are disconnected but the correlations arise because for either one ( or both ) of the colliding hadrons the ladders in the amplitude and/or the conjugate amplitude connect to the same color source .
these two - point functions are essentially the unintegrated gluon distributions of the hadrons ; they are of order @xmath17 when the transverse momentum in the ladder is below the saturation momentum @xmath18 of the corresponding hadron . or target @xmath19 , respectively , and the light - cone momenta are @xmath20 , @xmath21.,width=264 ] diagrams such as fig . [ fig : twopointc ] should dominate @xmath22 even at high ( but not asymptotically high ) transverse momentum , @xmath23 , provided one considers _
generic _ relative angles @xmath24 ( in particular , away from the region of `` back - to - back '' jets , @xmath25 ) . on the other hand , at leading order in @xmath26 , when @xmath27 the gluon pair should originate from the same ladder ; when the rapidity difference between the two produced gluons and the two beams are smaller than @xmath28 the ladder is dglap - ordered and @xmath22 should approach @xmath29 ( back - to - back dijet ) .
when @xmath30 the delta - function is smeared out by a bfkl - ordered ladder inbetween the produced gluons ( mueller - navelet jets @xcite ) . instead , here we consider the situation where @xmath31 are somewhat larger than but on the order of @xmath18 ; also , @xmath32 should be significantly smaller than the total rapidity window between the two beams , and the relative azimuthal angle @xmath33 .
when @xmath14 and @xmath15 are on the order of a few gev it is necessary to subtract the background of uncorrelated particle pairs to reveal the structure of the correlation function .
we note that two - particle correlations away from the back - to - back regime have recently been measured at the bnl - rhic accelerator at @xmath34 gev ( per colliding nucleon pair ) for proton - proton , deuteron - gold , and gold - gold collisions @xcite . for the former systems
only a narrow peak due to fragmentation of the triggered parton have been observed . for collisions of heavy ions ,
on the other hand , @xmath22 exhibits a `` ridge''-like structure : it is narrow in @xmath35 but extends over several units in @xmath36 .
the absence of measurable correlations in @xmath2 and @xmath37 collisions may be due to the smallness of the saturation momentum @xmath18 for a proton or deuteron at rhic energy .
also , the measurements from rhic might be expected to be rather sensitive to the initial conditions for the evolution equation at moderately small @xmath38 . at the higher energies of cern
s lhc collider , the saturation momentum of a proton measured from the central rapidity region is expected to be on the order of 1 gev and such correlations could be sufficiently strong to provide information about the qcd four - point function at small @xmath0 . the diagrams like the one from fig .
[ fig : twopointc ] arise from factorization of the four - point functions in the field of the projectile / target into products of two - point functions @xcite ( unintegrated gluon distributions ) .
doing so , however , picks up only the leading-@xmath1 contribution to the four - point function .
more generally , @xmath39 is given by & = & ( f_g aa^ f_g^b b^ f_gcc^ f_g^dd^ ) _
i=1 ^ 4 + & & + & = & ( f_g aa^ f_g^b b^ f_gcc^ f_g^dd^ ) _
i=1 ^ 4 + & & [ eq : cpq_fourpoint ] in the second step we have assumed factorization of the wave functions of projectile and target .
@xmath40 denotes the lipatov vertex which satisfies l_(p , k_1)l^(p , k_2 ) & = & - k_1^i ( p - k_1)^j k_2^n ( p - k_2)^m + l_(p , k)l^(p , k ) & = & - ( p - k)^2 . the expression ( [ eq : cpq_fourpoint ] ) is depicted in fig .
[ fig : fourpointc ] . or target @xmath19 , respectively.,width=264 ] here , @xmath41 denotes the color charge density per unit transverse area at a transverse coordinate @xmath42 and @xmath43 is its fourier transform .
its two - point function is related to the unintegrated gluon distribution @xmath44 via [ eq : phi_twopoint ] ( x ) = ( 2)^3 ( k - k ) ( x , k^2 ) . with this normalization one recovers the lo @xmath45-factorization formula for the single - inclusive distribution from the diagram [ fig : single ] with the standard prefactor @xcite : = 4_s d^2k , where @xmath46 is the transverse area of the collision ( note that in our convention @xmath44 is the density of gluons per unit transverse area and it therefore contains a factor of @xmath47 ) .
-factorization at leading order .
the blobs denote the unintegrated gluon distribution of the projectile @xmath48 or target @xmath19 , respectively.,width=264 ] in a mean field ( and large @xmath1 ) approximation one may factorize the four - point functions from eq .
( [ eq : cpq_fourpoint ] ) into products of two - point functions , [ eq:4point_fact_2point ] = ^ab ^cd ( ^2)^2 + ^ac ^bd ( ^2)^2 + ^ad ^bc ( ^2)^2 + , where @xmath49 , and the momentum dependence of the two - point function has been suppressed .
then , one of the nine contractions corresponds to the square of the single - inclusive distribution : contract the first @xmath50 with the third and the second with the fourth , for both projectile and target .
the color factor for this diagram is which will enter once @xmath51 is expressed through @xmath52 via eq .
( [ eq : phi_twopoint ] ) . ] & & f_g aa^ f_g^b b^ f_gcc^ f_g^dd^ + & ~ & f_g aa^ f_g^b b^ f_gcc^ f_g^dd^ ^ac ^bd ^a^c^ ^b^d^ = n_c^2 ( n_c^2 - 1)^2 .
[ eq : singlesquarecolor ] the remaining eight diagrams correspond to a color factor of ( we take fig . [
fig : twopointc ] as an example ) & & f_g aa^ f_g^b b^ f_gcc^ f_g^dd^ + & ~ & f_g aa^ f_g^b b^ f_gcc^ f_g^dd^ ^ac ^bd ^a^b^ ^c^d^ = n_c^2 ( n_c^2 - 1 ) .
[ eq : fig1_color ] thus , two - particle correlations are suppressed by a factor of @xmath53 as compared to uncorrelated production .
for this reason , the leading-@xmath1 _ ansatz _ ( [ eq:4point_fact_2point ] ) may not capture the complete result for @xmath22 .
below , we derive the evolution equation for the four - point function from jimwlk .
we determine the corrections beyond the mean - field and large-@xmath1 approximations to the rhs of ( [ eq:4point_fact_2point ] ) and show that these corrections contribute at the same order in @xmath1 to the correlation function . in this regard
, we should point out that @xmath1 corrections to the two - point function in the dense regime were found to be exceptionally small @xcite .
however , this needs not be true for the four - point function .
in fact , we shall argue below that we do not expect @xmath1 corrections to the four - point function to be anomalously small , even in the dilute regime .
a verification or falsification of this expectation via exact numerical solutions would be very valuable .
in this section we present the equation describing the rapidity evolution of the four - point function @xmath54 obtained from the jimwlk equations , which include terms of subleading order in @xmath1 . in this context
it is more natural to work in coordinate space , so @xmath42 , @xmath55 , @xmath56 , @xmath57 denote transverse coordinates ; the four - point function in momentum space can be obtained by fourier transform .
we also find it preferable to work with the fields @xmath58 rather than the color charge densities @xmath50 ; at leading order and in covariant gauge , they are related in coordinate space by a^(x^+,r ) ^- ( x^+,r ) = - g ^- ( x^+ ) ( x^+,r ) , for a hadron moving at the speed of light in the negative @xmath59-direction . since this field also satisfies @xmath60 , the only non - vanishing field - strength is @xmath61 . in momentum space
we have the relation @xmath62 .
the jimwlk evolution equation for the four - point function to lowest order in the fields can be shown to be ( see appendix ) _ r^a _ |r^b _
|s^d & = & d^2z + & & + . [ eq:4point_evol ] this expression neglects contributions from higher @xmath4-point functions on the rhs ; in the dilute regime , i.e. when the transverse momenta of the produced particles are higher than the saturation momenta of the colliding hadrons , this approximation should be justified . in order to derive the color structure of corrections beyond the large-@xmath1 approximation , we factorize the product of four point functions on the rhs of eq .
( [ eq:4point_evol ] ) into products of two point functions .
this gaussian approximation reduces the evolution equation for the four point function to a product of two bfkl equations ( for the two point function ) plus extra terms which provide corrections to the factorization ( [ eq:4point_fact_2point ] ) .
the result is _
r^a _ |r^b _ s^c _
|s^d = - d^2z [ eq : fact_correction ] where f_0^abcd & & f^abf^cd ^2_r-|r ^2_s-|s + f^adf^bc ^2_r-|s ^2_|r - s + f_1^abcd & & f^abf^cd + & + & f^adf^bc + f_2^abcd & & f^abf^cd .
[ eq : fs ] in ( [ eq : fs ] ) all terms in @xmath63 , @xmath64 and @xmath65 are to be duplicated with the substitutions indicated explicitly in the brackets .
second , all terms in @xmath63 and @xmath64 are to be duplicated again substituting @xmath66 and @xmath67 . then , all terms in @xmath63 and @xmath64 should be duplicated a third time exchanging @xmath68 and @xmath69 .
furthermore , all terms in @xmath65 are to be duplicated while letting @xmath70 and @xmath71 .
finally , the terms obtained in the last substitution ( only ) in @xmath65 should be duplicated exchanging @xmath68 and @xmath72 .
the first term in ( [ eq : fact_correction ] ) provides the leading-@xmath1 contribution to the four - point function .
the second term gives corrections beyond the large-@xmath1 factorization ( [ eq:4point_fact_2point ] ) .
since an analytic solution to the evolution equation for the four point function is not within our reach , a numerical investigation of these terms and their magnitude would be extremely useful . nevertheless , from _ y ~_s n_c ^ab^cd ( ^2)^2 + _ s f^ac f^bd ( ^2)^2 , ^2(y ) ~e^_sn_cy one might expect that , generically , the solution to this equation has the following color structure : & ~ & ^ab ^cd ( ^2)^2 + ^ac ^bd ( ^2)^2 + ^ad ^bc ( ^2)^2 + [ eq:4point_leading ] + & & f^acf^bd ( ^2)^2 + f^abf^cd ( ^2)^2 + f^adf^bc ( ^2)^2 . [ eq:4point_subleading ] ( note that the various two - point functions depend on different coordinates / momenta and
so each of the above terms is distinct . )
the color factors emerging from the products of the kronecker tensors have already been discussed above , eqs .
( [ eq : singlesquarecolor ] ) and ( [ eq : fig1_color ] ) .
however , some of the products of a leading-@xmath1 term from the first line ( [ eq:4point_leading ] ) with a subleading-@xmath1 term from the second line ( [ eq:4point_subleading ] ) also contribute at the same order @xmath73 .
for example , ^ac ^bd f_g aa^ f_g^b b^ f_gcc^ f_g^dd^ f^abf^cd = n_c ^ac ^bd f^abf^cd = n_c^2(n_c^2 - 1 ) .
the other products are worked out in appendix [ sec : colorfactorsapp ] .
this shows that some of the subleading-@xmath1 contributions from the four - point function actually enter @xmath22 at leading order , compare to eq .
( [ eq : fig1_color ] ) .
previous results from the literature @xcite ( also see @xcite ) are therefore not complete .
nevertheless , the correlations described here should still extend over several units in @xmath32 @xcite .
quantitative results for the jimwlk four - point function and for the corresponding two - particle correlations @xmath16 as functions of the transverse momenta @xmath14 , @xmath15 , relative azimuth @xmath35 and relative rapidity @xmath32 remain to be found .
in summary , we have argued that two - particle correlations from high - energy collisions may provide some insight into the qcd four - point function .
this should be the case , in particular , when the transverse momenta of the produced particles are not very much higher than the saturation momenta of the colliding hadrons and when their relative azimuthal angle is sufficiently less than @xmath74 .
the narrow ( in both azimuthal and polar angle ) jet - like fragmentation peak should sit on top of a `` background '' which is broader in the relative rapidity @xmath32 .
if expanded in powers of @xmath1 , the leading contribution to the four - point function is given by the product of two bfkl two - point functions .
however , we find that genuine b - jimwlk subleading-@xmath1 corrections also appear in the correlation function @xmath75 , at leading non - vanishing order in @xmath1 .
the correlations mentioned here represent an interesting opportunity to study the non - trivial structure of the four - point function of the b - jimwlk hierarchy , both theoretically and experimentally .
the jimwlk equation for the two - point function is _ r^a _ s^c & = & d^2x d^2y
^bd_xy _ r^a _ s^c + & = & d^2x d^2y _ r^a _
s^c + d^2x d^2y ^bd_xy _ r^a _
s^c , [ eq:2 ] where [ eq:9 ] ^bd_xy = ^bd .
we start with the first term from eq .
( [ eq:2 ] ) : ^bd_xy & = & ^bd [ eq:4 ] + & = & ^bd .
[ eq:5 ] note that in ( [ eq:4 ] ) the derivative does not act on the @xmath76 on the left because ( x)_bd = ig(x - u ) ( t^b v^_u x)_bd ~(t^b)_be ( v^_u x)_ed = 0 . the first term in the bracket from eq .
( [ eq:5 ] ) vanishes because @xmath77 and @xmath78 . to apply the same argument to the second term from ( [ eq:5 ] ) we rewrite 2 ( x - z)(y - z ) = ( x - z)^2 + ( y - z)^2 -[(x - z)-(y - z)]^2 = ( x - z)^2 + ( y - z)^2 - ( x - y)^2
hence , the second term from ( [ eq:5 ] ) becomes ( x - z ) ^bd = 0 .
the first term again vanishes when @xmath79 while for the other two terms the divergences at @xmath79 cancel and so they vanish due to the color structure .
we can therefore simplify eq .
( [ eq:5 ] ) to ^bd_xy & = & ( x - y ) ^bd .
[ eq:5b ] for the second term from eq .
( [ eq:2 ] ) we need _ r^a _ s^c & = & ^ad(r - y ) _ s^c + ^cd(s - y ) _
r^a + _ r^a _ s^c & = & ^ad(r - y ) ^bc(x - s ) + ^cd(s - y ) ^ab(r - x ) .
( [ eq:2 ] ) turns into _ r^a _ s^c & = &
d^2x d^2y d^2z \ { ( x - y ) ^bd } + & & + d^2x d^2y d^2z ^bd + & = & d^2z \ { + } + & & + d^2z \ { ^ca + ^ac } + & = & d^2z + d^2z ^ac .
[ eq:16 ] to expand the rhs to second order in the fields we need the following expressions : ^ba_s^c & = & ig ( _ z^d - _ r^d ) ^ba_s^c + + & = & ig ( _ z^d - _ r^d ) _ s^c + & = & ig ( _ z^d - _ r^d ) _ s^c + & = & -ig n_c ^ad ( _ z^d - _ r^d ) _ s^c + & = & ig n_c ( _ r^a - _ z^a ) _ s^c and ^ac & = & ig^ac + & & + ( ig)^2 ^ac [ eq:23 ] + & = & g^2 ^ac [ eq:24 ] + & = & g^2 t^b_ae t^d_ec + & = & g^2 f^bea f^dec . in the step from ( [ eq:23 ] ) to ( [ eq:24 ] ) we have used the rapidity ordering of the fields ; hence , only one of the @xmath80 terms can contribute . using these expressions in ( [ eq:16 ] ) gives _
r^a _ s^c & = & - _ s^c + & + & d^2z .
[ eq:27 ] we now take the expectation value on the rhs using [ eq:28 ] _ r^a _
s^c = ^ac ^2_r - s where @xmath81 is essentially the unintegrated gluon distribution function .
this turns eq .
( [ eq:27 ] ) into ( we drop an overall @xmath82 ) ^2_r - s & = & - + d^2z .
[ eq:29 ] to cast this into the familiar form for the bfkl equation we symmetrize the kernel of the first term and rewrite that of the second term in a different form : & = & + + 2 & = & + - .
then , the part of ( [ eq:29 ] ) involving @xmath83 ( the bfkl virtual part ) becomes - d^2z _ r - s^2 = - d^2z _ r - s^2 , where @xmath84 .
the other terms from ( [ eq:29 ] ) where @xmath85 depends on the integration variable @xmath59 ( the bfkl real emissions ) turn into & & d^2z \ { + + } + & = & d^2z \ { - - + + + ( _ r - z^2 + _ s - z^2 - _ 0 ^ 2 ) } + & = & d^2z \ { -2 + ( _ r - z^2 + _ s - z^2 - _ 0 ^ 2 ) } .
the jimwlk equation for the four - point function is _ r^a _ |r^b _ s^c _ |s^d & = & d^2x d^2y ^ef_xy _ r^a _ |r^b _ s^c
_ |s^d + & = & d^2x d^2y _ r^a _ |r^b _ s^c _ |s^d + d^2x d^2y ^ef_xy _ r^a _ |r^b _ s^c _
|s^d , [ eq:37 ] we begin with the first term .
the square bracket can be taken over from eq .
( [ eq:5b ] ) , and may also be written in the form ^ef_xy & = & - f^egf ( x - y ) ^eg .
we also need _ r^a _ |r^b _
s^c _ |s^d & = & ^af_r - y _ |r^b _ s^c _ |s^d + ^bf_|r - y _ r^a _ s^c _ |s^d + ^cf_s - y _ r^a _ |r^b _ |s^d + ^df_|s - y _ r^a _ |r^b _ s^c .
the first term on the rhs of ( [ eq:37 ] ) becomes - d^2z & & \
{ ^eg _ |r^b _ s^c
_ |s^d + ^eg _ r^a _ s^c _ |s^d . + & & .
+ ^eg _ r^a _ |r^b
_ |s^d + ^eg _ r^a _ |r^b _ s^c }
. [ eq:40 ] to linear order in the fields , f^aeg ^eg = ig f^aeg ( _ z^- _ r^ ) ( t^)^eg = -g f^aeg f^eg ( _ z^- _ r^ ) = -g n_c ( _ z^a - _ r^a ) , and so eq . ( [ eq:40 ] ) is equal to & & d^2z \ { ( _ z^a - _ r^a ) _ |r^b _
|s^d + _ r^a ( _ z^b - _ |r^b ) _
s^c _ |s^d .
+ & & . + _ r^a _ |r^b ( _ z^c - _ s^c ) _ |s^d + _ r^a _ |r^b _ s^c ( _ z^d - _ |s^d ) } + & = & -4 _ r^a _ |r^b _ s^c
+ & & + d^2z \ { + + + } [ eq:43 ] this is the final result for the first term from ( [ eq:37 ] ) . in the next section [ sec : gauss ]
we will simplify this further by taking the expectation value with a gaussian weight .
we now turn to the second term from eq .
( [ eq:37 ] ) .
with @xmath86 we find [ eq:44 ] d^2x d^2y ^ef_xy _ r^a _ |r^b _ s^c _ |s^d = ^ab_r|r _ s^c
_ |s^d + ^ac_rs _ |r^b _ |s^d + ^ad_r|s _ |r^b _ s^c + ^bc_|r s _ r^a _ |s^d + ^bd_|r |s _ r^a _ s^c + ^cd_s|s _ r^a _ |r^b .
the expansion of @xmath87 in powers of the field starts at second order . from ( [ eq:9 ] ) , ^ab_xy & = & ^ab + & = & f^eaf^fb . [ eq:46 ]
we note that the two terms @xmath88 exhibit different ordering in rapidity as they arise from the expansion of @xmath89 and @xmath90 to order @xmath91 , respectively . for what follows ,
we combine them into a single @xmath92 term which shows up no matter how rapidities are ordered . using ( [ eq:46 ] ) in ( [ eq:44 ] )
we obtain & & \ { f^eaf^fb _ s^c _ |s^d .
+ & + & f^eaf^fc _ |r^b _
|s^d + & + & f^eaf^fd _
s^c + & + & f^ebf^fc
|s^d + & + & f^ebf^fd _
s^c + & + & .
f^ecf^fd _ r^a _
[ eq:47a ] this is the final result for the second term from ( [ eq:37 ] ) . in the next section
we compute its expectation value for a gaussian weight . to simplify the expressions further
, we assume that the expectation value of the four - point function on the rhs of the evolution equation is taken with a gaussian weight so that it is given by a sum over all possible wick contractions : _ r^a _
|r^b _ s^c _
|s^d & = & _ r^a _ s^c _ |r^b _ |s^d + _ r^a
_ |r^b _ s^c
_ |s^d + _ r^a _ |s^d _ |r^b _ s^c [ eq : a44 ] + & = & ^ac ^bd ^2_r - s ^2_|r-|s + ^ab ^cd ^2_r-|r ^2_s-|s + ^ad ^bc ^2_r-|s ^2_|r - s , [ eq:45 ] where we used ( [ eq:28 ] ) .
the above factorization into two - point functions reproduces the evolution of the four - point function at leading order in @xmath1 .
the lhs of ( [ eq:37 ] ) then becomes _
r^a _ |r^b _
s^c _ |s^d & = & ^ac ^bd ( ^2_|r-|s ^2_r - s + ^2_r - s ^2_|r-|s ) [ eq:50 ] + & + & ^ab ^cd ( ^2_s-|s ^2_r-|r + ^2_r-|r ^2_s-|s ) [ eq:51 ] + & + & ^ad ^bc ( ^2_|r - s ^2_r-|s + ^2_r-|s ^2_|r - s ) .
[ eq:52 ] we continue to analyze only one color channel corresponding to the first term @xmath93 ( the others are similar ) .
the two - point functions satisfy the bfkl equation and therefore ^2_|r-|s ^2_r - s & = & - d^2z \ { + 2 } ^2_|r-|s + ^2_r - s ^2_|r-|s & = & - d^2z \ { + 2 } ^2_r - s .
thus , ( [ eq:50 ] ) turns into ^2_r - s ^2_|r-|s & = & - d^2z \ { ^2_r - s ^2_|r-|s .
+ & & + ^2_|r-|s + & & .
+ ^2_r - s } . [ eq:55 ] at leading order in @xmath1 the evolution of the four - point function should be determined by the bfkl evolution of the two - point function , as given in the previous equation [ plus similar terms from eqs .
( [ eq:51],[eq:52 ] ) ] . in
what follows , we verify this explicitly from the full jimwlk evolution equation , eq .
( [ eq:43 ] ) plus ( [ eq:47a ] )
. however , we also derive the contributions which are subleading in @xmath1 .
forming pairwise contractions on eq .
( [ eq:43 ] ) gives & & d^2z \
{ + & & ^ab ^cd + & & ^ac ^bd + & & .
^ad ^bc } .
[ eq:53 ] all of these terms will cancel against corresponding pieces from eq .
( [ eq:47a ] ) .
we organize the virtual contributions ( where none of the fields depends on the integration variable @xmath59 ) from eq .
( [ eq:47a ] ) according to the various possible contractions .
the term involving @xmath94 is & & d^2z \ { f^eaf^fb ^ec^fd .
+ & + & f^eaf^fc ^ef^bd + & + & f^eaf^fd ^ec^bf + & + & f^ebf^fc ^af^ed + & + & f^ebf^fd ^ac^ef + & + & . f^ecf^fd
^ae^bf } ^2_r - s ^2_|r-|s + & & = d^2z \ { f^caf^db .
+ & + & n_c ^ac^bd + & + & f^caf^bd + & + & f^dbf^ac + & + & n_c ^ac^bd + & + & . f^acf^bd } ^2_r - s ^2_|r-|s + & & = d^2z \ { n_c ^ac^bd .
+ & - & . f^acf^bd } ^2_r - s ^2_|r-|s .
[ eq:56 ] the first term on the first line cancels the corresponding term involving @xmath94 from eq .
( [ eq:53 ] ) .
the rest of that line contributes to the rapidity evolution of the four - point function at order @xmath95 and is seen to match the first line of eq .
( [ eq:55 ] ) , i.e. the `` virtual correction '' part of the leading-@xmath1 bfkl equation for the four - point function .
the second line in ( [ eq:56 ] ) is a correction which is suppressed by a relative factor of @xmath96 . along the same lines
one finds that the contribution from eq .
( [ eq:47a ] ) involving @xmath97 is & & = d^2z \ { n_c ^ab^cd .
+ & - & . f^abf^cd } ^2_r-|r ^2_s-|s .
[ eq:57 ] the first term on the first line will again cancel with a similar term @xmath98 from the first line of eq .
( [ eq:43 ] ) .
finally , we also list the contribution from eq .
( [ eq:47a ] ) involving @xmath99 : & & = d^2z \ { n_c ^ad^bc .
f^adf^bc } ^2_r-|s ^2_|r - s .[eq:58 ] as before , the very first term in this expression will cancel .
to summarize the results obtained so far : the evolution equation for the four - point function in mean - field approximation is given by _
r^a _ |r^b _
|s^d & = & ( ^ac ^bd ^2_|r-|s ^2_r - s + ^ab ^cd ^2_s-|s ^2_r-|r + ^ad ^bc ^2_|r - s ^2_r-|s ) [ eq:62 ] + & - & d^2z \
{ + & & f^acf^bd ^2_r - s ^2_|r-|s + & + & f^abf^cd ^2_r-|r ^2_s-|s + & + & .
f^adf^bc ^2_r-|s ^2_|r - s } . the first line , eq .
( [ eq:62 ] ) , involves only standard bfkl evolution of the two - point functions and provides the leading-@xmath1 contributions to both real emissions and virtual corrections .
these terms can be written in explicit form by using the bfkl equation for the rapidity evolution for the two - point functions ; for example , the first term from that line is given in ( [ eq:55 ] ) .
the remaining lines correspond to the subleading ( in @xmath1 ) contributions to the virtual terms for jimwlk evolution . in the next section , we derive the corresponding real emission terms . we begin with terms which contain only one @xmath59-dependent field , as given by the second and third columns in ( [ eq:47a ] ) .
the first line is given by & & f^caf^fb + & = & f^caf^fb + & & + & = & n_c ^ab ^cd ( ^2_r - z ^2_s-|s + ^2_z-|r ^2_s-|s ) + f^caf^db ( ^2_r - s ^2_z-|s + ^2_z - s ^2_|r-|s ) + f^daf^cb ( ^2_r-|s ^2_z - s + ^2_z-|s ^2_|r - s ) . the second plus third terms from the first line of ( [ eq:47a ] ) thus becomes & - & d^2z ( + - ) + & & .
the remaining terms from ( [ eq:47a ] ) with one @xmath59-dependent field are easily obtained via permutations of color indices and coordinates .
the full result is & - & d^2z + & & \ { ( + - ) .
+ & & + & & + ( + - ) + & & + & & + ( + - ) + & & + & & + ( + - ) + & & + & & + ( + - ) + & & + & & + ( + - ) + & & . } .
adding now the terms from eq .
( [ eq:53 ] ) with one @xmath59-dependent field , and those from eq .
( [ eq:47a ] ) with two @xmath59-dependent fields , leads to the sum of the following two contributions ; the terms proportional to @xmath1 are given by & - & d^2z \
{ + & & ^ab ^cd + & & + ^ac ^bd + & & + ^ad ^bc } these provide the leading-@xmath1 contributions and correspond to bfkl evolution of the product of two two - point functions ; compare to eq .
( [ eq:55 ] ) .
the genuine jimwlk contribution is & - & d^2z \ { + & & f^abf^cd + & + & f^acf^bd + & & f^adf^bc } .
[ eq:70 ] this expression ( [ eq:70 ] ) , which is to be added to the rhs of ( [ eq:62 ] ) , is our final result for the evolution equation of the jimwlk four - point function for a gaussian weight .
it can be rewritten in more compact form by exploiting some of the symmetries of @xmath100 , eqs .
( [ eq : fact_correction],[eq : fs ] ) .
in this appendix , we compute the color factors for the products of a leading-@xmath1 term from the first line ( [ eq:4point_leading ] ) with a subleading-@xmath1 term from the second line ( [ eq:4point_subleading ] ) ; color indices from the target side carry a prime and we need to also include the remaining structure constants from eq .
( [ eq : cpq_fourpoint ] ) . using the following su(@xmath1 ) identities f^abf^cb & = & n_c ^ac , + f^abf^cd & = & ( ^ac^bd - ^ad^bc ) + d^acd^bd - d^add^bc , + d^acd^acr & = & ( n_c-)^r , + f^g aa^ f^gcc^ f^ac = i^3 tr_adj t^a^ t^c^ t^ & = & f^a c , + f^a_a g f^c_gc f^d_cg f^b_ga = tr_adj t^a t^c t^d t^b & = & ^ac^bd + ^ab^cd + ( d^acrd^dbr - d^adrd^bcr + d^abrd^cdr ) , + d^add^bcd^acrd^bdr & = & o(n_c^4 ) , one derives ^ab ^cd ( f_g aa^ f_gcc^ f^ac ) ( f_g^b b^ f_g^dd^ f^bd ) & = & n_c^2 ( n_c^2 - 1 ) , + ^ab ^cd f_g aa^ f_gcc^ f_g^b b^ f_g^dd^ f^ab f^cd & = & f^ab f^cd tr_adj t^a t^c t^d t^b + = f^ab f^cd ( ^ac^bd + ^ab^cd + ( d^acrd^dbr - d^adrd^bcr + d^abrd^cdr ) ) & = & ^bd + f^ab f^cdd^acrd^bdr + & & = ( n_c^2 - 1 ) + ^ab ^cd f_g aa^ f_gcc^ f_g^b b^ f_g^dd^ f^ad f^bc & = & f^ad f^bc tr_adj t^a t^c t^d t^b + = f^ad f^bc ( ^ac^bd + ^ab^cd + ( d^acrd^dbr - d^adrd^bcr + d^abrd^cdr ) ) & = & 0 [ eq : c9 ] + ^ac ^bd f_g aa^ f_g^b b^ f_gcc^ f_g^dd^ f^acf^bd = n_c ^ac ^bd f^acf^bd & = & 0 + ^ac ^bd f_g aa^ f_g^b b^ f_gcc^ f_g^dd^ f^abf^cd = n_c ^ac ^bd f^abf^cd & = & n_c^2(n_c^2 - 1 ) + ^ac ^bd f_g aa^ f_g^b b^ f_gcc^ f_g^dd^ f^adf^bc = n_c ^ac ^bd f^adf^bc & = & - n_c^2 ( n_c^2 - 1 ) + ^ad ^bc ( f_g aa^ f_gcc^ f^ac ) ( f_g^b b^ f_g^dd^ f^bd ) & = & - n_c^2 ( n_c^2 - 1 ) , + ^ad ^bc f_g aa^ f_g^b b^ f_gcc^ f_g^dd^ f^abf^cd & = & f^ab f^cd tr_adj t^a t^c t^b t^d + & = & - = 0 + ^ad ^bc f_g aa^ f_g^b b^ f_gcc^ f_g^dd^ f^adf^bc & = & - n_c^2 ( n_c^2 - 1 ) . we have used the jaxodraw program @xcite to draw the diagrams shown in this paper .
thanks a. kovner for useful discussions .
we gratefully acknowledge support by the doe office of nuclear physics through grant no .
de - fg02 - 09er41620 and from the city university of new york through the psc - cuny research award program , grants 60060 - 3940 ( a.d . ) and 62625 - 40 ( j.j.m . ) .
i. balitsky , nucl .
b * 463 * , 99 ( 1996 ) [ arxiv : hep - ph/9509348 ] .
j. jalilian - marian , a. kovner , l. d. mclerran and h. weigert , phys .
d * 55 * , 5414 ( 1997 ) ; j. jalilian - marian , a. kovner , a. leonidov and h. weigert , nucl . phys .
b * 504 * , 415 ( 1997 ) , phys .
d * 59 * , 014014 ( 1999 ) , phys .
d * 59 * , 014015 ( 1999 ) , phys .
d * 59 * , 034007 ( 1999 ) [ erratum - ibid .
d * 59 * , 099903 ( 1999 ) ] ; a. kovner , j. g. milhano and h. weigert , phys . rev .
d * 62 * , 114005 ( 2000 ) ; a. kovner and j. g. milhano , phys .
d * 61 * , 014012 ( 2000 ) ; e. iancu , a. leonidov and l. d. mclerran , nucl .
phys . a * 692 * , 583 ( 2001 ) , phys .
b * 510 * , 133 ( 2001 ) ; e. ferreiro , e. iancu , a. leonidov and l. mclerran , nucl .
a * 703 * , 489 ( 2002 ) .
l. n. lipatov , sov .
j. nucl .
* 23 * , 338 ( 1976 ) [ yad . fiz . *
23 * , 642 ( 1976 ) ] ; e. a. kuraev , l. n. lipatov and v. s. fadin , sov .
jetp * 45 * , 199 ( 1977 ) [ zh . eksp .
fiz . * 72 * , 377 ( 1977 ) ] ; i. i. balitsky and l. n. lipatov , sov .
j. nucl .
* 28 * , 822 ( 1978 ) [ yad .
* 28 * , 1597 ( 1978 ) ] .
a. dumitru , f. gelis , l. mclerran and r. venugopalan , nucl .
a * 810 * , 91 ( 2008 ) [ arxiv:0804.3858 [ hep - ph ] ] .
a. h. mueller and h. navelet , nucl .
b * 282 * , 727 ( 1987 ) ; predictions for rhic energy have been shown in d. kharzeev , e. levin and l. mclerran , nucl .
a * 748 * , 627 ( 2005 ) [ arxiv : hep - ph/0403271 ] ; c. marquet , nucl .
a * 796 * , 41 ( 2007 ) [ arxiv:0708.0231 [ hep - ph ] ] ; k. tuchin , arxiv:0912.5479 [ hep - ph ] ; + for @xmath2 collisions at lhc see , for example , l. h. orr and w. j. stirling , phys .
b * 436 * , 372 ( 1998 ) [ arxiv : hep - ph/9806371 ] ; c. ewerz , l. h. orr , w. j. stirling and b. r. webber , j. phys .
g * 26 * , 696 ( 2000 ) [ arxiv : hep - ph/9912469 ] ; a. sabio vera and f. schwennsen , nucl .
b * 776 * , 170 ( 2007 ) [ arxiv : hep - ph/0702158 ] ; c. marquet and c. royon , phys .
d * 79 * , 034028 ( 2009 ) [ arxiv:0704.3409 [ hep - ph ] ] ; + angular correlations for very high @xmath8 dijets from @xmath101 collisions at tevatron and @xmath102 interactions at hera , and their relation to ordering of the initial - state showers , have been studied in f. hautmann and h. jung , jhep * 0810 * , 113 ( 2008 ) [ arxiv:0805.1049 [ hep - ph ] ] .
j. adams _ et al . _ [ star collaboration ] , phys . rev
. lett . * 95 * , 152301 ( 2005 ) [ arxiv : nucl - ex/0501016 ] ; j. putschke , j. phys .
g * 34 * , s679 ( 2007 ) [ arxiv : nucl - ex/0701074 ] ; b. i. abelev _ et al .
_ [ star collaboration ] , arxiv:0909.0191 [ nucl - ex ] .
b. alver _ et al . _
[ phobos collaboration ] , arxiv:0903.2811 [ nucl - ex ] .
a. adare _ et al .
_ [ phenix collaboration ] , phys .
c * 78 * , 014901 ( 2008 ) [ arxiv:0801.4545 [ nucl - ex ] ] .
l. v. gribov , e. m. levin and m. g. ryskin , phys .
rept . * 100 * , 1 ( 1983 ) .
k. rummukainen and h. weigert , nucl .
a * 739 * , 183 ( 2004 ) [ arxiv : hep - ph/0309306 ] ; y. v. kovchegov , j. kuokkanen , k. rummukainen and h. weigert , nucl .
a * 823 * , 47 ( 2009 ) [ arxiv:0812.3238 [ hep - ph ] ] .
k. dusling , f. gelis , t. lappi and r. venugopalan , arxiv:0911.2720 [ hep - ph ] .
d. binosi , j. collins , c. kaufhold and l. theussl , comput .
commun . * 180 * , 1709 ( 2009 ) [ arxiv:0811.4113 [ hep - ph ] ] ; | we derive the rapidity evolution equation for the gluon four - point function in the dilute regime and at small @xmath0 from the jimwlk functional equation .
we show that beyond leading order in @xmath1 the mean field ( gaussian ) approximation where the four point function is factorized into a product of two point functions is violated .
we calculate these factorization breaking terms and show that they contribute at leading order in @xmath1 to correlations of two produced gluons as a function of their relative rapidity and azimuthal angle , for generic ( rather than back - to - back ) angles .
such two - particle correlations have been studied experimentally at the bnl - rhic collider and could be scrutinized also for @xmath2 ( and , in the future , also @xmath3 ) collisions at the cern - lhc accelerator . |
using the formula ( due to bill gosper @xcite ) @xmath12 fabrice bellard ( * ? ? ?
* file pi1.c ) found an algorithm for computing the @xmath3-th decimal of @xmath13 without computing the earlier ones .
thus he improved an earlier algorithm due to simon plouffe @xcite .
this formula can be proved in the following way .
we have @xmath14 hence the right hand side of the formula will be @xmath15 but @xmath16
so we get @xmath17 _ 0 ^ 1=\pi .\end{aligned}\ ] ] we then asked if there are other such formulas of type @xmath18 where @xmath2 is a polynomial in @xmath3 . using the same trick as before we have @xmath19 and hence @xmath20 if @xmath2 has degree @xmath21 then @xmath22 where @xmath23 has degree @xmath24 .
it follows with @xmath25 that @xmath26 where @xmath27 is a polynomial in @xmath28 of degree @xmath29 .
we want this integral to be equal to @xmath13 . a good way to get
@xmath13 is to have @xmath30 or arctan@xmath31 after integration .
this means that @xmath32 must have the factor @xmath33 or @xmath34 , that is have a zero at @xmath35 or @xmath36 .
this restricts @xmath9 and @xmath10 and gives the value of @xmath11 . after experimenting with the lll - algorithm we found formulas for @xmath13 in the following cases ( there could be many more ) : [ cols="^,^,<,^ , < " , ] this formula was used in bellard s world record setting computation of the 1000 billionth _
binary _ digit of @xmath0 , being based on the algorithm in @xcite . ]
e.g. , to find the formula in the last case we computed @xmath37 for @xmath38 with @xmath39 digits .
then the lll - algorithm found ( in about three days ) the linear combination between @xmath13 and @xmath40 .
a good check is that the coefficient of @xmath13 is a product of small primes ( actually all less than @xmath41 ) .
for @xmath42 the integrals can be computed by brute force using _
maple_. the higher cases are all symmetric , i.e. , @xmath43 . using the symmetry
@xmath44 one can assume that @xmath30 and @xmath45 each contribute @xmath46 .
so we make the wild assumption that @xmath47 where @xmath48 is a polynomial with @xmath49 .
differentiation with respect to @xmath28 yields @xmath50 or @xmath51 this is a differential equation of first order in @xmath52 with polynomial coefficients . taking the case @xmath53 ,
we have @xmath54 and @xmath55 , so the denominator @xmath56 has degree @xmath57 . we solve the differential equation using _ maple _
s @xmath58 setting the order to @xmath59
. then we find that @xmath52 is a polynomial of degree @xmath60 . in practice one
replaces the coefficients of @xmath61 and @xmath62 by @xmath63 times the denominator in the formula found for @xmath13 .
then @xmath52 will have ( huge ) integer coefficients .
computing the integral in the case @xmath64 , _ maple _ got the terms @xmath65 which can be shown to be equal to @xmath66 in the nonsymmetric case @xmath67 one gets @xmath45 and @xmath68 , but the latter term makes no contribution since it has the same values at @xmath69 and @xmath70 .
when we set this up then , in principle , we could let _
maple _ do the evaluation of @xmath71 ( and thus of @xmath27 ) through , i.e. , let _ maple _ s formal summation routines compute the sum @xmath72 or , to make the point clearly visible , compute a sum of the form @xmath73 where @xmath74 is a certain polynomial in @xmath3 .
however , it turns out that this is very time - consuming .
it is a much better idea to expand @xmath74 as @xmath75 , and then use that @xmath76 for @xmath77 .
see also the approach that we follow in section [ sec : proof ] when we prove our main result .
solving the differential equation for @xmath48 one finds that @xmath48 has the factor@xmath80 . if one divides out the factor @xmath81 , the remaining factor is invariant under the substitution @xmath82 .
hence we make the substitution @xmath83 where @xmath84 then @xmath85 now @xmath86
so @xmath87 and @xmath88 it follows that the left hand side of the differential equation is @xmath89 on the right hand side we have @xmath90 and @xmath91 is for all @xmath8 divisible by @xmath92 hence @xmath93 and @xmath91 and @xmath94 are all polynomials in @xmath95 . finally we get the following differential equation for @xmath96 : @xmath97 if @xmath93 is known then one solves this equation for @xmath96 just as before ( but the degrees are cut in half ) .
but there is a possibility to _ find _ the formula for @xmath13 _ and prove _ it _ in one stroke_. let @xmath98 and assume that @xmath99 and @xmath100 with @xmath101 substituting this into the differential equation we will get a system of linear equations for the @xmath102 s of size @xmath103 . if this system is nonsingular , we can be sure to be able to solve it and thus _ find and prove _ an expansion for @xmath0 of the form .
we went ( again ) to the computer and generated the system of equations for small values of @xmath8 .
let @xmath104 be the matrix of the system .
let further @xmath105 be the least common denominator of the coefficients of @xmath106 .
we obtained the following tables : l|l @xmath8 & @xmath107 + + 1 & @xmath108 + 2 & @xmath109 + 3 & @xmath110 + 4 & @xmath111 + 5 & @xmath112 r|l @xmath8 & @xmath113 + + 1 & @xmath114 + 2 & @xmath115 + 3 & @xmath116 + 4 & @xmath117 + 5 & @xmath118 + 6 & @xmath119 + 7 & @xmath120 + 8 & @xmath121 + 9 & @xmath122 + 10 & @xmath123 from the table it seems obvious " that the determinant of @xmath104 will never vanish as it grows rather quickly in absolute value with @xmath8 .
we also see that the same primes occur in the factorizations of @xmath105 and @xmath124 .
even more striking , it seems that the largest prime factors occurring grow only slowly ( namely approximately linearly ) when @xmath8 increases .
the last fact strongly indicates that there may even be a closed form formula for @xmath125 .
as it turns out , this is indeed the case . in the next section
we will explicitly compute the determinant of an equivalent system of equations , from which it follows that ( see the remark after theorem [ t2 ] ) @xmath126 hence we have the following theorem , which is the main theorem of our paper .
[ t1 ] for all @xmath127 there is a formula @xmath128 where @xmath106 is a polynomial in @xmath3 of degree @xmath129 with rational coefficients .
the polynomial @xmath106 can be found by solving the system of linear equations generated by and the ansatz and .
the denominators of all @xmath102 s divide @xmath105 which is much smaller than @xmath130 .
this means that there must be some miracle occurring at the end when solving the system .
e.g. , when @xmath131 then @xmath132 has about @xmath133 digits but @xmath134 only @xmath135 .
unfortunately , we are not able to offer an explanation for that . in practice
we are only interested in the coefficients of @xmath136 so we try to eliminate @xmath137 first .
this can be done by first avoiding all equations containing @xmath138 , i.e. , the equations coming from the coefficients of @xmath139 , @xmath140 .
this looks very nice theoretically , in particular as the system for @xmath141 is triangular , but in practice the computer breaks down since the rational numbers occuring become very large .
we close this section by listing a few explicit examples .
we have @xmath142 where @xmath143 and @xmath144 we have @xmath145 where @xmath146 and @xmath147 we have @xmath148 where @xmath149 and @xmath150 ( nonsymmetric ) .
we have @xmath151 where @xmath152 and @xmath153
we want to prove that , by making the ansatz and and substituting this into the differential equation ( the polynomial @xmath93 being given by @xmath136 through , @xmath91 being given by @xmath154 ) , the resulting system of linear equations will always have a solution .
in fact , we aim at finding an explicit formula for the determinant of the corresponding matrix of coefficients that allows us to conclude that it can never vanish .
it turns out that for that purpose it is more convenient to set up the system of linear equations in a different , but equivalent way .
this equivalent system will have coefficient matrix @xmath155 ( see ) .
the evaluation of its determinant will be accomplished through eqs .
, , , , and theorem [ t2 ] . to be precise ,
we encode the polynomial @xmath156 ( and , thus , @xmath71 and @xmath27 ) differently .
we claim that @xmath71 has an expansion of the form @xmath157 subject to the single constraint @xmath158 ( as usual , empty poducts have to be interpreted as 1 . ) this is seen as follows .
the polynomial @xmath156 can be written in the form @xmath159 for some coefficients @xmath160 .
hence , we have @xmath161 in the last line , @xmath162 and @xmath163 have to be read as @xmath69 .
it is now a trivial exercise to substitute the coefficients of @xmath164 in the sum in the last line into the left - hand side of and verify the truth of .
now we are ready to set up the system of linear equations .
we make again the ansatz , but we replace by @xmath167 where the @xmath168 , @xmath169 , are subject to ( i.e. , the relation holds when @xmath166 is replaced by @xmath168 ) . clearly , we have to add to the set of equations that result from the differential equation .
the coefficient matrix of the system looks a follows : @xmath170 where @xmath28 is a line vector of @xmath171 zeroes , @xmath172 is the vector of coefficients of , i.e. , @xmath173 , and @xmath174 @xmath175 , where @xmath176 is an @xmath177 matrix and @xmath178 is an @xmath179 matrix , both of which we define below .
we consider the top - most line of @xmath155 ( which is formed out of @xmath28 and @xmath180 ) as row 0 of @xmath155 .
we label the rows of @xmath176 and @xmath178 by @xmath35 running from @xmath70 to @xmath181 .
furthermore , we label the columns of @xmath155 by @xmath182 running from @xmath70 to @xmath183 . following this labelling scheme
, the matrix @xmath176 has nonzero entries only in the four diagonals @xmath184 , @xmath185 , @xmath186 , @xmath187 .
we denote the entries in column @xmath182 on these four diagonals in order @xmath188 , @xmath189 , @xmath190 , and @xmath191 , where @xmath192 to be precise , the @xmath193-entry is @xmath188 , the @xmath194-entry is @xmath189 , the @xmath195-entry is @xmath190 , the @xmath196-entry is @xmath191 , @xmath197 . on the other hand ,
the matrix @xmath178 is composed out of columns , labelled @xmath198 , each of which containing just one nonzero entry . to be precise ,
the nonzero entry of column @xmath199 is located in the @xmath200-st row ( according to our labelling scheme ) , and it is equal to @xmath201 , @xmath169 .
we perform some row operations on @xmath155 , with the effect that the entries of @xmath180 get eliminated .
this is achieved by subtracting @xmath202 times row 1 from row 0 , and @xmath203 times row @xmath204 from row 0 , @xmath205 .
doing this , we must expect changes in row 0 in columns @xmath70 , @xmath129 , @xmath206 , @xmath207 , @xmath208 ,
, @xmath209 .
however , at this point a miracle occurs : the new entries in row 0 in columns @xmath204 , @xmath210 , are still 0 . on the other hand ,
the values of the new entries in row 0 in columns @xmath211 , @xmath212 are @xmath213 after these manipulations we obtain a matrix of the form @xmath214 where @xmath215 and @xmath216 have the same dimensions as before @xmath28 and @xmath180 , respectively , where the only nonzero entries of @xmath215 are in columns labelled by numbers which are divisible by @xmath129 , with the entry in column @xmath211 given by , and where @xmath216 consists only of zeroes .
we have @xmath217 the next step consists in expanding the determinant of @xmath218 with respect to columns @xmath219 ( i.e. , the last @xmath220 columns ) .
since each of these columns contains just one nonzero entry ( which is a power of @xmath221 ) , we have @xmath222 where @xmath223 is the matrix arising from @xmath218 by deleting the last @xmath220 columns and the rows @xmath204 , @xmath224 .
more precisely , the matrix @xmath223 has the following form : @xmath225 where @xmath226 , @xmath212 , is a line vector with @xmath129 entries , all of them being zero except for the last , which is equal to , where @xmath227 is a line vector of @xmath228 zeroes , where @xmath229 and @xmath230 , @xmath212 , are @xmath231 matrices with nonzero entries only in the two main diagonals , and where @xmath232 and @xmath233 are @xmath234 matrices , @xmath233 being upper triangular . to be precise , for @xmath212 we have @xmath235 and @xmath236 and we have @xmath237
the precise form of @xmath232 is without relevance for us .
we do a laplace expansion with respect to the last @xmath228 rows . because of the triangular form of @xmath233 we obtain @xmath238 where @xmath239 instead of @xmath240 we consider a more general matrix .
define the functions @xmath241 it should be noted that these functions specialize to @xmath242 , respectively , if @xmath243 and @xmath244 .
now we define the matrix @xmath245 by @xmath246 where @xmath247 and @xmath248 clearly , we have @xmath249 once having found this theorem , it is not difficult to prove , by working out how the coefficients of @xmath258 resulting from the ansatz are related to the coefficients of @xmath258 resulting from the ansatz . since this is not essential for the proof of theorem [ t1 ] , we leave the details to the reader .
first we show that @xmath259 divides @xmath250 , @xmath260 .
what has to be proved is that @xmath250 vanishes for @xmath261 .
this can be done by showing that for this choice of @xmath262 there is a nontrivial linear combination of the rows of @xmath245 . indeed , if @xmath263 we have @xmath264 as is easy to verify .
next we claim that @xmath265 divides @xmath250 , @xmath266 .
let us first impose the additional restriction that @xmath267 . using the above reasoning
, the claim then follows from the fact that if @xmath268 we have @xmath269 as is again easy to verify . on the other hand , if @xmath270 , then the same argument shows that @xmath271 divides @xmath250 .
it remains to be checked that also @xmath272 divides @xmath250 .
indeed , if @xmath273 then we have @xmath274 whence @xmath272 divides @xmath250 for @xmath260 . clearly , the degree in the @xmath262 s , @xmath275 s , and @xmath272 s of @xmath250 is at most @xmath276 .
but the degree of the right - hand side of is exactly @xmath276 .
therefore we have proved that @xmath277 where @xmath278 is a constant independent of the @xmath262 s , @xmath275 s , and @xmath272 s . in order to determine @xmath278 , we compare coefficients of @xmath279 on both sides of .
we claim that the coefficient of this monomial in @xmath250 is equal to @xmath280 , where @xmath281 is defined exactly in the same way as @xmath245 ( see ) , except that the definitions of the functions @xmath282 are replaced by [ eq : ck8 ] @xmath283 this is seen as follows .
the monomial does not contain any @xmath275 .
therefore , for finding its coefficient in @xmath250 , we may set @xmath284 in @xmath245 for all @xmath11 . in which way may the monomial appear in @xmath250 ( with all @xmath275 equal to 0 ) ?
a typical term in the expansion of @xmath250 is the product of @xmath285 entries of @xmath245 , each from a different row and column .
the monomial contains @xmath286 .
the variable @xmath287 is only found in columns @xmath204 , @xmath288 ( and rows labelled by numbers @xmath289 mod @xmath228 , according to our labelling scheme ) .
therefore in a product of entries ( each from a different row and column ) which produces a term containing @xmath286 _ all _ the entries from columns @xmath204 must be ones containing @xmath287 .
this explains the above definitions and of @xmath290 and @xmath291 , @xmath288 , respectively .
moreover , we _ must _ generate the @xmath292 in from an entry in a column @xmath293 , for some @xmath294 .
( the variable @xmath292 is also found in entries in columns @xmath204 , but these columns are already taken by our choice of entries which contain the @xmath287 s . ) this explains the definitions and of @xmath295 and @xmath296 , @xmath288 , respectively .
next we ask how we can find ( in the remaining columns and rows ) entries which contain @xmath297 .
arguing in an analogous manner , the variable @xmath298 only appears in columns @xmath293 , @xmath299 .
one of these columns is already taken by the entry from which we picked @xmath292 . therefore in _ all _ the remaining ones we must choose entries containing @xmath298 .
this explains the definitions and of @xmath300 and @xmath301 , @xmath288 , respectively .
next we consider the term @xmath302 in .
it must come from two entries in columns @xmath303 , for two different @xmath294 s .
this explains the definitions and of @xmath304 and @xmath305 , @xmath288 , respectively .
etc .
we consider now a more general determinant than @xmath280 , the latter having been defined through the functions in .
replace these functions by @xmath309 let us denote the matrix defined by these functions in the same way as before @xmath281 by @xmath310 .
clearly , @xmath310 specializes to @xmath281 if all @xmath311 are set equal to 1 and all @xmath312 to @xmath313 .
we proceed in a similar way as in the proof of theorem [ t2 ] . in the first step
we show that the product on the right - hand side of divides @xmath315 as a polynomial in the @xmath316 s and @xmath317 s .
then , in the second step , we compare the degrees of the product and @xmath315 . since the degree of @xmath315 turns out to be at most the degree of the product , it then follows that @xmath315 is equal to the product times some constant which is independent of the @xmath311 s and @xmath312 s .
finally , in the third step , this constant is found by computing the leading coefficient of @xmath315 . .
we start by applying several row and column operations to @xmath315 , with the final goal of reducing the size of the determinant .
first , for @xmath318 , in this order , we add @xmath319 times row @xmath35 to row @xmath320 .
( it should be recalled that , according to our labelling scheme , we number the rows of @xmath310 from @xmath69 to @xmath276 . ) thus , we obtain the determinant of the following matrix : @xmath321 where the @xmath322 s , @xmath323 , are defined as earlier , and where the @xmath324 s and @xmath325 s , @xmath323 , are the @xmath231 matrices @xmath326 and @xmath327 next we make " the submatrices @xmath325 , @xmath323 , to diagonal matrices , by subtracting @xmath328 times column @xmath182 from column @xmath329 , @xmath330 , @xmath331 times column @xmath182 from column @xmath329 , @xmath332 , , and @xmath333 times column @xmath182 from column @xmath329 , @xmath334 .
after these operations we obtain the determinant of the matrix @xmath335 where @xmath336 and the @xmath337 s , @xmath338 , are the @xmath231 matrices @xmath339 and @xmath340 where @xmath341 now we eliminate the last columns in @xmath342 for @xmath343 .
we start by eliminating the last column of @xmath344 .
we do this by adding @xmath345 times column @xmath346 to column @xmath285 , @xmath347 .
this makes all the entries in the last column which are in rows @xmath348 zero , whereas the entries in the last column in rows @xmath349 are modified .
next we eliminate these entries in a similar fashion , by using the columns @xmath350 , @xmath347 , etc . in the end all the entries in the last column in rows @xmath351 will be zero , whereas the entries in the last column in rows @xmath352 will have been ( significantly ) modified .
an analogous procedure is applied to eliminate the entries in the last columns of @xmath353 , @xmath354 .
just to mention the first step : we add @xmath355 times column @xmath350 to column @xmath356 , @xmath347 .
this makes all the entries in column @xmath356 which are in rows @xmath357 zero , whereas the entries in the last column in rows @xmath358 are modified .
the advantage after having done all this is that now all the entries in columns @xmath211 , @xmath212 , are zero except for entries in rows @xmath359 .
this fact , and the fact that the submatrices @xmath336 are diagonal matrices ( of rectangular form ) with last column consisting entirely of zeroes , makes it possible to reduce the determinant of the ( new ) matrix significantly . for @xmath360 we may expand the determinant with respect to row @xmath35 , in this order . if the details are worked out , then we see that our original determinant @xmath315 is equal to @xmath361 where @xmath362 is given by , and where @xmath363 is given by @xmath364 clearly , we may extract @xmath365 from column @xmath182 , @xmath366 , and@xmath367 from row @xmath35 , @xmath368 ( still using our nonstandard labelling scheme where the rows are numbered @xmath359 ) , so that we obtain the expression @xmath369 where @xmath370 is given by @xmath371 and where @xmath372 is given by @xmath373 from it is abundantly clear that @xmath374 divides @xmath315 ( which , after all , is equal to ) .
it remains to show that also @xmath375 divides @xmath315 .
let @xmath11 and @xmath376 , @xmath377 , @xmath378 , be given .
we want to show that @xmath379 divides @xmath315 .
we will show the equivalent fact that the rows of @xmath380 are linearly dependent .
the crucial observation , from which this claim follows easily , is that the entries @xmath372 , @xmath366 , in row @xmath35 of the determinant in are given by a polynomial in @xmath182 , @xmath381 say , of degree @xmath382 and with leading coefficient ( i.e. , coefficient of @xmath383 ) equal to @xmath384 this is seen as follows .
the summand in is a polynomial in @xmath385 of multidegree @xmath386 ( i.e. , the degree in @xmath182 is @xmath387 , the degree in @xmath388 is @xmath389 , etc . ) . because of the fact that ( for fixed @xmath390 and varying @xmath391 ) @xmath392 is a polynomial in @xmath391 of degree @xmath393 with leading coefficient ( i.e. , coefficient of @xmath394 ) equal to @xmath395 , successive summation over @xmath396 , @xmath397 , , @xmath388 yields the claimed facts . under the specialization @xmath398 , it is seen by inspection " that with @xmath35 replaced by @xmath399 vanishes , because the summand corresponding to @xmath400 cancels with the summand corresponding to @xmath401 . hence , the polynomial @xmath402 has degree ( at most ) @xmath403 ( instead of @xmath404 ) .
consequently , if @xmath398 then the entries in rows @xmath399 , @xmath405 , , @xmath228 are given by polynomials in @xmath182 ( to wit : @xmath182 denoting the column index of the entries ) of respective degrees @xmath403 ( ! ) , @xmath403 , @xmath406 ,
, @xmath63 , @xmath70 .
these are @xmath407 polynomials , all of degree at most @xmath403 .
it follows that there must be a nontrivial linear combination of these polynomials that vanishes .
hence , the rows @xmath399 , @xmath405 , , @xmath228 are linearly dependent , which , in turn , implies that the determinant in ( and , thus , also @xmath315 ) vanishes for @xmath398 , @xmath377 , @xmath378 . . clearly , the degree of @xmath315 as a polynomial in the @xmath311 s and @xmath312 s is at most @xmath276 , whereas the degree of the product on the right - hand side of is exactly @xmath276 .
hence , we have @xmath408 where @xmath409 is a constant independent of the @xmath311 s and @xmath312 s . .
in order to determine @xmath409 , we determine the coefficient of @xmath410 in the expansion of @xmath411 . by arguments similar to those at the end of the proof of theorem [ t2 ] , it is seen that this coefficient is given by the determinant of the following matrix , which we denote by @xmath412 .
it is defined exactly in the same way as @xmath245 ( see ) , except that the definitions of the functions @xmath282 are replaced by @xmath413 by expanding this determinant with respect to row 0 , we obtain @xmath414 where @xmath415 is , as earlier , given by , and @xmath416 is the matrix arising from @xmath412 by deleting row 0 and column @xmath417 .
let @xmath418 , @xmath419 , be fixed .
we will next compute @xmath420 .
when we built @xmath416 from @xmath412 , we deleted in particular row 0
. therefore we will now switch to the usual labelling scheme for rows and columns of a matrix , i.e. , we will subsequently not only label the columns by @xmath421 but also the rows .
if @xmath422 , then we expand @xmath420 with respect to the last @xmath228 rows and then with respect to the last column .
since these rows and this column contain only one nonzero entry , we obtain some multiple of the determinant of a @xmath423 matrix . if @xmath424 , then we continue by expanding the ( now reduced ) determinant with respect to the last @xmath425 rows and then with respect to the last 2 columns .
we continue in the same manner until we have reduced @xmath420 to the determinant of a @xmath426 matrix , more precisely , until we arrive at @xmath427 where @xmath428 is the matrix @xmath429 with @xmath430 and @xmath336 the @xmath231 matrices @xmath431 and @xmath432 @xmath433 the @xmath434 matrix which arises from @xmath430 by deleting its last column , and @xmath176 the @xmath435 matrix @xmath436 we continue by expanding @xmath428 with respect to the last @xmath437 rows .
thus we obtain @xmath438 for @xmath420 , where @xmath439 is the matrix @xmath440 with @xmath178 the @xmath441 matrix @xmath442 next we prepare for a reduction from the top of @xmath439 .
we subtract 4 times column @xmath182 from column @xmath443 , @xmath444 , @xmath445 , , @xmath446 , in this order .
thus @xmath447 is converted to @xmath448 , where @xmath449 is the matrix @xmath450 with @xmath451 and @xmath452 the @xmath231 matrices @xmath453 and @xmath454 our next goal is to push " the ( nonzero ) entries in columns @xmath455 , @xmath456 , down to rows @xmath457 .
( this is similar to what we did in step 1 when we pushed " all the nonzero entries in columns @xmath458 , @xmath459 up to rows @xmath359 . ) in order to achieve this for the 1-st column , we add @xmath460 to column 1 . similarly , in order to achieve this for the @xmath461-st column , we add @xmath462 to column @xmath206 .
etc . as a result ,
the determinant @xmath448 is converted to the determinant of the matrix @xmath463 where @xmath451 and @xmath452 are as before , @xmath464 is the @xmath465 matrix @xmath466 and @xmath467 , @xmath468 , is a @xmath465 matrix with all entries equal to 0 , except for the entries in the first column . to be precise , the entry in the first column and row @xmath469 of @xmath467 is given by @xmath470
it should be noted that the entries in the first column of @xmath452 ( @xmath452 appearing at the bottom of the matrix , as do the matrices @xmath467 ) is given by with @xmath471 .
now everything is prepared for the reduction .
we expand the determinant of with respect to rows @xmath472 .
this reduces the determinant of to @xmath473 where @xmath474 is a @xmath434 matrix of the form @xmath475 with the @xmath476-entry of @xmath477 being given by , @xmath478 , @xmath479 , and @xmath178 the @xmath480 matrix from above .
if we substitute all this in , we obtain that @xmath420 is equal to @xmath481 the submatrix @xmath178 of @xmath474 is almost diagonal .
subtraction of 4 times row @xmath469 from row @xmath482 in @xmath474 , @xmath483 , will transform it into a completely diagonal matrix ( namely into the matrix on the right - hand side of with all entries @xmath484 replaced by 0 ) . as a side effect
, this will turn the @xmath485-entry of @xmath474 into @xmath486 as is easily seen , the determinant of the in this way modified matrix , @xmath487 say , is @xmath488 or , explicitly , @xmath489 substitution of the above in yields that the determinant @xmath420 is equal to @xmath490 now we substitute this in .
we obtain that @xmath491 is equal to @xmath492 the sum is readily evaluated by means of the chu
vandermonde summation ( see e.g. ( * ? ? ?
5.1 , ( 5.27 ) ) ) , so that we obtain @xmath493 since the coefficient of @xmath410 in the expression on the right - hand side of is exactly the same , we have completed the proof of the lemma . | we show how to find series expansions for @xmath0 of the form @xmath1 , where @xmath2 is some polynomial in @xmath3 ( depending on @xmath4 ) .
we prove that there exist such expansions for @xmath5 , @xmath6 , @xmath7 , for any @xmath8 , and give explicit examples for such expansions for small values of @xmath9 , @xmath10 and @xmath11 . |
recent attempts to measure the primordial @xmath0he abundance report accuracies better than @xmath3 ( izotov & thuan 1998 ; olive , steigman , & skillman 1997 ; peimbert , peimbert , & ruiz 2000 ) .
such accurate measurements are useful for testing the consistency of big bang nucleosynthesis and also for estimating the mean baryon density , @xmath4 ( schramm & turner 1998 ) .
the primordial @xmath0he mass fraction , @xmath5 , is often inferred from the emission lines of blue compact galaxies ( e.g. , peimbert & torres - peimbert 1974 ; pagel , terlevich , & melnick 1986 ) under the implicit assumption that it had the same initial value throughout the universe , including within underdense regions of the intergalactic medium ( igm ) . however , a net difference between the element abundances in overdense and underdense regions could arise if diffusion occurs during the gravitational collapse of massive objects .
different atomic species , because of their different masses , will have different thermal speeds and will therefore experience different pressure forces per unit mass .
consequently , different elements would not flow into an overdense region at the same rate ; instead , heavier elements would obtain higher relative abundances in galaxies than in voids .
diffusion is generally moderated by a frictional force arising from inter - species collisions .
these collisions are most frequent for an ionized gas ( spitzer 1956 ) in which the characteristic cross section for ion - proton collisions , @xmath6 , is far greater than the collision cross
section for the corresponding atoms in a neutral gas , @xmath7 ( here , @xmath8 is the ion charge , @xmath9 is the temperature , @xmath10 is boltzmann s constant , and @xmath11 is the coulomb logarithm ) .
we therefore focus our attention on the period between cosmological recombination and reionization during which the igm was neutral ( loeb & barkana 2001 ) . during this stage
, atomic collision cross sections may be small enough to allow some diffusion to occur . in this paper
we quantify the diffusion of @xmath0he , @xmath1li , and d relative to @xmath12h that occurs as a spherical , slightly overdense region begins to collapse .
we assume a @xmath13cdm cosmology , characterized by densities relative to critical of @xmath14 , @xmath15 , and @xmath16 for matter , vacuum , and baryons ; a hubble constant of @xmath17 ; a @xmath0he primordial mass fraction of @xmath18 ; and primordial @xmath1li and d number densities relative to @xmath12h of @xmath19 and @xmath20 , respectively .
the effects of diffusion are most important where the gas pressure matters , namely around the comoving jeans length scale @xmath21 . for redshifts
@xmath22 , when the baryonic temperature is primarily determined by the adiabatic expansion , the jeans wavenumber is ( barkana & loeb 2001 ) @xmath23^{1/2 } \label{eq : k_j}\ ] ] where @xmath24 , @xmath25 is the mean molecular weight , @xmath26 is the mass of hydrogen and @xmath27 the baryonic temperature at a redshift @xmath28 . in
[ sec : baseq ] , we derive the differential equations used for our analysis .
these are linearized equations which are strictly valid only as long as the overdensities remain small .
however , we assume that reionization occurs before the collapsing region becomes significantly nonlinear and we stop our calculation at reionization because charged particle cross sections suppress any further diffusion . in [ sec : inicond ] , we discuss the initial conditions used for our differential equations . in [ sec : calc ] , we compute both the mean abundance deviation and the typical abundance fluctuation within a collapsing region .
we then use a press - schechter mass function to average the resulting deviations over all collapsed objects in the universe . in [ sec : obs ] , we compare our model to recent measurements of @xmath0he abundances .
finally , we present our conclusions in [ sec : sandc ] .
we first establish our notation and illustrate our assumptions by carrying out a brief derivation of the equations satisfied by density perturbations .
each particle species has a mass density , @xmath29 , and a bulk velocity , @xmath30 , which are related by the continuity equation , @xmath31 where we start by using physical ( proper ) coordinates . since the universe is initially homogeneous , @xmath32 and all particles move with the hubble flow : @xmath33 ( peebles 1993 ) . in this approximation ,
equation ( [ eq : conteq ] ) becomes @xmath34 we next allow for density and velocity perturbations through the definitions @xmath35 , \label{eq : defrhop}\ ] ] @xmath36 \delta_{kronecker}^{ij } .
\label{eq : defvelp}\ ] ] the velocity gradient is decomposed into a symmetric , traceless tensor ( @xmath37 ) , an antisymmetric tensor ( @xmath38 ) , and a trace .
we insert these definitions into equation ( [ eq : conteq ] ) and linearize the perturbations to obtain @xmath39 switching from now on from physical to comoving coordinates , and letting an overdot represent a time derivative , equation ( [ eq : ceqphysco ] ) may be compactly expressed as @xmath40 the density and bulk velocity also satisfy a force equation ; in comoving coordinates , @xmath41 the sum on the right - hand side includes all forces acting on species @xmath42 .
we linearize this equation and make use of equation ( [ eq : ceqcoco ] ) to obtain a new equation which does not explicitly contain the velocity perturbations , @xmath43 .
\label{eq : diffeqt}\ ] ] we now switch independent variables from time to redshift via the friedmann equation , which leads to @xmath44 however @xmath45 only becomes significant for @xmath46 ; in this paper we consider redshifts @xmath47 for which @xmath48 denoting a derivative with respect to @xmath28 by a prime , equation ( [ eq : diffeqt ] ) becomes @xmath49 .
\label{eq : diffeqf}\ ] ] we now obtain expressions for the forces . the force due to gravity , when expressed in terms of the potential , @xmath50 ,
is @xmath51 .
we insert this into equation ( [ eq : diffeqf ] ) and use poisson s equation which gives @xmath52 + \frac{3}{2\omega_m } \left ( 1+z \right)^{-2 } \sum_{species } \omega_i\delta_i , \label{eq : gravity}\ ] ] where the primed sum means a sum over all forces except gravity .
although all types of particles are affected by gravity , we assume that only the baryons ever obtain a non - zero temperature .
each species of baryons experiences a different pressure force which is determined by the negative of its pressure gradient , @xmath53 where we have used @xmath54 for the atomic mass and @xmath55 for the temperature of species @xmath42 . dividing equation ( [ eq : prforce ] ) by the mass density and taking the divergence
, we see that @xmath56 to first order in the perturbations .
we define @xmath57 to be the temperature in the absence of perturbations , and express the effects of perturbations in terms of the small quantity @xmath58 : @xmath59 . inserting this into equation ( [ eq : prlin ] ) and linearizing
once more we see that @xmath60 the temperature of the baryons is determined by two effects , the coupling of free electrons to the background radiation and the adiabatic expansion ( peebles 1993 ) .
this gives @xmath61 where @xmath62 is the free electron fraction , @xmath63 is the thomson cross section , @xmath64 is the stefan - boltzmann constant , @xmath65 is the electron mass , @xmath66 is the speed of light , and @xmath67 is the temperature of the cosmic microwave background ( cmb ) . switching the independent variable to
@xmath28 and linearizing , equation ( [ eq : dtdt ] ) becomes @xmath68 where @xmath69 k is the cmb temperature at present . the mean baryon temperature , @xmath57 ,
is defined to be a solution of @xmath70 this indicates that @xmath57 is independent of baryonic species and so all baryons would be at the same temperature in the absence of perturbations .
we use the computer program recfast ( seager , sasselov , & scott 1999 ) to determine the free - electron fraction as a function of redshift
. then we use equation ( [ eq : tempzero ] ) to compute the mean baryon temperature .
both quantities are shown in figure [ fig : xtz ] .
after linearizing equation ( [ eq : dtdz ] ) , we obtain the equations for the temperature perturbations , @xmath71 where @xmath72 . with this definition of @xmath73 , equation ( [ eq : prtaucontrib ] )
becomes @xmath74 because of the spatial derivatives , we are forced to work in comoving fourier space . however , for simplicity of notation we will continue to simply write @xmath75 and @xmath73 , although from now on we use these symbols to represent the fourier components of the perturbations .
finally we take into account the frictional term arising from inter - species collisions .
physically , this force is non - zero only if two species have different bulk velocities ; thus , the force on one species due to a second species may be written as @xmath76 ( burgers 1969 ) .
since the difference in bulk velocities is already linear in the perturbations , we only need to calculate the coefficient @xmath77 to zeroth order for the linearized equations of motion .
this allows us to calculate @xmath77 under the assumption that all baryons share the same temperature .
in general , the frictional force per unit volume acting on species @xmath78 due to species @xmath79 is obtained by multiplying the number of collisions per unit time per unit volume by the momentum transfer per collision .
we multiply this by the probability that species @xmath78 and
@xmath79 have initial momentum @xmath80 and @xmath81 and that the center of mass scattering angle is @xmath82 . finally , we integrate over all parameters , @xmath83 we assume that the collisions are elastic and that the atoms can be approximated by hard spheres .
we take each species to have a maxwellian velocity distribution . following
the integration equation ( [ eq : collint ] ) becomes @xmath84 here , @xmath85 is the reduced mass , @xmath86 is the classical hard - sphere cross section , @xmath87 , and the integral is a quantum mechanical correction ( massey , burhop , & gilbody 1969 ) . the function @xmath88 is given by @xmath89 .
\label{eq : qmcor}\ ] ] the integration variable @xmath62 is related to @xmath90 by @xmath91 , and the phase shifts are given by @xmath92 and @xmath93 , where the functions @xmath94 are bessel functions of order @xmath95 .
the distance scale @xmath64 represents the sum of the radii of two colliding atoms ; we adopt the values @xmath96 , @xmath97 , and @xmath98 ( clementi , raimondi , & reinhardt 1963 ; slater 1964 ) .
we evaluate the integral in equation ( [ eq : collforce ] ) numerically and plot the results in figure [ fig : xsect ] for each type of inter - species collision .
quantum mechanics becomes more important as the temperature decreases but the correction is never greater than a factor of four . for brevity
, we henceforth denote this integral by @xmath99 .
we now summarize the equations governing each species
. for dark matter equation ( [ eq : diffeqf ] ) becomes @xmath100 for each baryonic species there are also pressure and collisional friction terms , and so @xmath101 where @xmath102 is the comoving wave number . from equation ( [ eq : bdeq ] ) we can easily find the differential equations satisfied by the differences relative to hydrogen , @xmath103 : @xmath104 \\ & & - \frac{h_0}{g } \sqrt{\frac{2 k_b \bar{t}}{\omega_m \pi^3}}\sum_{j\neq h } \frac{\omega_j\sigma^{h - j}_{cl}}{\sqrt{m_h m_j \left ( m_h + m_j \right ) } } \left ( 1+z \right)^{1/2 } i_{h - j}\left ( z \right ) \delta^{\prime}_j \nonumber \\ & & + \frac{h_0}{g } \sqrt{\frac{2 k_b \bar{t}}{\omega_m \pi^3}}\sum_{j\neq i } \frac{\omega_j\sigma^{ij}_{cl}}{\sqrt{m_i m_j \left ( m_i + m_j \right ) } } \left ( 1+z \right)^{1/2 } i_{ij}\left ( z \right ) \left ( \delta^{\prime}_j - \delta^{\prime}_i \right ) \nonumber .
\label{eq : diffdeq}\end{aligned}\ ] ]
we consider dark matter perturbations initially of the spherical top - hat form @xmath105 , which has the fourier components @xmath106 here , @xmath107 and @xmath108 are constants and @xmath109 represents the step function .
we take the initial values of @xmath110 , @xmath111 , @xmath112 and their derivatives to be zero , but we would like to choose the initial values of @xmath113 , @xmath114 , and @xmath115 that correspond to the most rapidly growing mode . to gain some insight into how to make this choice , we consider a simplified problem where baryonic diffusion is negligible .
we approximate the baryonic temperature by the temperature of the background radiation ( valid for @xmath116 ; see figure [ fig : xtz ] ) .
equation ( [ eq : bdeq ] ) for the baryon overdensity becomes @xmath117 we use this equation to express @xmath118 , differentiate , and insert the results into equation ( [ eq : dmdeq ] ) .
the outcome is a fourth order differential equation for @xmath119 .
it has four independent , power law solutions which we write as @xmath120 .
the four @xmath121 are the four arbitrary constants associated with a linear fourth order equation .
the four @xmath122 are given by @xmath123 where @xmath124 is the squared comoving wavenumber normalized by @xmath125 .
the resulting expression for the dark matter overdensity is @xmath126 with the constants @xmath127 given by @xmath128 motivated by this solution , we choose our initial conditions so that the perturbation will grow like the most negative @xmath122 ( i.e. , the third one listed in equation [ eq : inipower ] ) .
we let the three @xmath121 corresponding to all but the largest growth rate be zero , and let the final @xmath121 be determined by the tophat distribution ( equation [ eq : tophatf ] ) .
the remaining initial conditions may now be written in terms of the known @xmath127 and @xmath122 .
we define a new set of variables , @xmath129 @xmath130 @xmath131 these quantities are special because they not only satisfy the same differential equations as the @xmath132 and @xmath133 , but also satisfy simpler initial conditions , @xmath134 , @xmath135 , @xmath136 , and @xmath137 .
the parameters @xmath138 have disappeared from the problem .
therefore we only evolve each normalized @xmath102-mode once ; we obtain the physical overdensity for a given @xmath107 and @xmath108 by multiplying each @xmath102-mode by the appropriate factor from equation ( [ eq : normdeldef ] ) .
we have set @xmath139 following cosmological recombination .
the final redshift is marked by the first of two possible events .
our equations are applicable in the linear regime , so we must stop if the object becomes nonlinear . at a transition to nonlinearity , the density would be greatly enhanced which would result in the suppression of diffusion .
therefore , we expect diffusion to be effective in the linear regime where our equations are valid .
the second stopping condition relates to the reionization of the universe . at that time , large coulomb cross sections between charged particles are able to effectively halt diffusion . in conclusion , we consider the case where the object becomes nonlinear after reionization and adopt @xmath140 ( haiman & loeb 1998 , 1999a , b ; gnedin & ostriker 1997 ; chiu & ostriker 2000 ; gnedin 2000 ) as the stopping point of our diffusion calculation . since the collapsed fraction of the universe is small at @xmath141 ( barkana & loeb 2001 ) , most of the material within galaxies that form much later ( at @xmath142 ) was still in the linear regime at @xmath143 , where our diffusion calculation is adequate .
several results of having carried out this prescription are shown in figure [ fig : nplots ] .
all length scales have been measured in terms of @xmath144 , defined in equation ( [ eq : k_j ] ) .
this expression involves the baryonic temperature which is set by the adiabatic expansion for redshifts less than @xmath145 , @xmath146 ; as discussed in
[ sec : baseq ] . at @xmath147 ,
the jeans wavenumber therefore obtains the value @xmath148^{1/2 } , \label{eq : kjnumber}\ ] ] which is the value used in figure [ fig : nplots ] .
we see that the dark matter overdensity has evolved approximately proportionally to @xmath149 , as might have been expected since gravity is the only force applicable to dark matter . on large scales corresponding to @xmath150 ,
the baryons follow the dark matter ; however , on small scales corresponding to @xmath151 , pressure forces prevent baryonic modes from collapsing .
this leads to the slight decline in dark matter amplification near the jeans scale and the dramatic decline in the amplification of baryonic modes there . in the lower panel of figure 3
, we see that the perturbations of the different baryonic species do not experience exactly the same amplification . at @xmath152 ,
pressure is negligible for all species , and the modes of each are amplified . since hydrogen is the lightest species , it reaches its jeans scale at the lowest @xmath102 . thus , all of the @xmath153 become positive at approximately the same wavenumber .
as @xmath102 increases , we begin to reach the jeans scales of the heavier species .
then the growth of @xmath102-modes of this second species is also suppressed , and the @xmath153 approach zero as @xmath102 gets large .
however , figure [ fig : nplots ] indicates that the overall magnitude of diffusion is not solely determined by the jeans scale of a species , or else the effect would have been largest for @xmath1li .
this is because of the frictional term which serves to suppress diffusion , which is proportional to the square of the radius of an atom and therefore the greatest suppression is associated with the largest atoms .
the normalized overdensities in figure [ fig : nplots ] were obtained by evolving 1025 @xmath102-modes , equally spaced between with @xmath154 . to obtain the physical overdensities , each mode must be multiplied by its normalization factor .
the normalization factor depends on the object assembly radius @xmath108 , and is fairly rapidly varying compared to the normalized overdensities . to get an accurate determination of the physical overdensity , we interpolate the normalized overdensity so that we sample it at a rate larger than the frequency of the normalization factor .
we then use equation ( [ eq : normdeldef ] ) to obtain the physical @xmath102-modes .
one such sampling is shown in figure [ fig : delmnine ] .
the value of @xmath108 used for this plot corresponds to an enclosed mass of @xmath155 .
the relative errors incurred in interpolation are typically @xmath156 , and seldom exceed @xmath157 with the 1025 initial points .
the normalization factor is also proportional to the initial dark matter perturbation , @xmath107 , which we express in terms of the collapse redshift of the object ( @xmath158 ) through the approximate relation we take @xmath159 for the overdensity of collapse and , unless otherwise stated , @xmath160 for the redshift of collapse .
such objects would be just on the fringe of nonlinearity at @xmath140 .
since diffusion within all objects was shut off at @xmath147 , the effects of diffusion would be decreased for objects that collapsed later .
next we quantify the extent to which element abundances in collapsed objects differ from their primordial values .
abundances of @xmath0he are commonly reported in terms of the mass fraction , @xmath161 where we have neglected the contributions of all species but @xmath0he and @xmath12h to the mass of the universe .
after linearizing this equation , we see that @xmath162 in contrast , the abundances of @xmath1li and d are typically reported in terms of their number densities relative to hydrogen . in this case , @xmath163 to see how the element abundances in collapsed objects differ , on average , from their primordial values , we average equations ( [ eq : ymdev])- ( [ eq : lidedev ] ) over the original sphere of comoving radius @xmath108 , @xmath164 after integrating , we see that @xmath165 we carry out this final integration numerically . as described in [ sec : kmodes ] , we evolve 1025 @xmath102-modes between 0 and 100@xmath166 to get the @xmath153 ( where @xmath62 could denote any of @xmath0he , @xmath1li , d ) .
we interpolate to get values for @xmath153 at a fixed number of points , and then multiply each point by its normalization .
we then evaluate the integral in equation ( [ eq : mdelint ] ) with these points .
then we repeat the procedure , but with twice as many interpolated points . if the two results differ by less than a fraction of @xmath167 , we stop and accept the result .
figure [ fig : meandev ] illustrates the dependence of these deviations on the size of the collapsed object . instead of measuring the size of the object by its assembly radius , @xmath108 , we instead measure it in terms of its mass defined by @xmath168 we see that the deviation from the primordial mean shrinks as the collapsed object increases in size approximately according to a power law with slope @xmath169 .
a simple argument yields the same result : assuming that the collapsing region starts out with primordial abundances throughout its volume and that diffusion occurs through the boundary of the region , the deviations from primordial abundances should fall off like @xmath170 , or @xmath171 .
now that element abundances for particular collapsed objects are known , we proceed by averaging these abundances over all objects in the universe that have collapsed by a particular redshift .
this statistic is approximated by using a press - schechter mass function ; for @xmath0he , @xmath172 where @xmath173 is the number density of objects of mass @xmath174 that have collapsed at a given redshift and @xmath175 is the averaged @xmath0he mass fraction within an object of mass @xmath174 .
equation ( [ eq : uniavg ] ) is also appropriate for the @xmath1li and d abundances if we replace @xmath176 with @xmath177 .
note also that there is a minimum mass scale of galaxies in which star formation is possible that we denote by @xmath178 ( abel 1995 ; tegmark et al .
we use a code developed by eisenstein & hu ( 1999 ) and barkana ( 2000 ) to calculate @xmath173 at a redshift of @xmath179 .
the averaged abundances differ from primordial by the amounts shown in figure [ fig : unimean ] .
we also consider a collapse redshift of @xmath180 ; in this case figure [ fig : unimean ] shows that the abundance deviations decrease somewhat .
this is not only due to the resulting decrease in @xmath107 , but also due to the dependence of @xmath173 on collapse redshift . at higher @xmath28
, @xmath173 tends to favor lower mass scales where the abundance deviations are larger ( figure [ fig : meandev ] ) .
the element abundances of collapsed objects can differ from primordial abundances in two ways .
first , the abundances of an object could be uniformly different from the primordial values .
this possibility was already discussed in the previous section .
second , if the element abundances within an object vary greatly from point to point , differences from the primordial abundances may be observed because of the fluctuations .
we consider this latter effect by computing the standard deviation of the abundances within a given object .
the lowest - order formula for this quantity is @xmath181 for @xmath0he and @xmath182 for @xmath1li and d. we have already calculated the mean values of the abundances , so now we only need the mean squares .
we proceed as above , but this time we obtain a double integral that must be evaluated numerically , @xmath183 the integration region is the first quadrant of the @xmath184 plane .
since the integrand is symmetric with respect to the interchange of @xmath102 and @xmath185 , it is equal to the integral @xmath186 we carry out the integrations as with equation ( [ eq : mdelint ] ) and our results are plotted as a function of mass scale in figure [ fig : sigplot ] .
comparing figures [ fig : meandev ] and [ fig : sigplot ] , we see that the relative importance of abundance fluctuations depends on the mass of the object in question . on small mass scales , abundances differ from their primordial values by an approximately uniform amount ; but as the object size increases fluctuations from the mean are more likely to produce a difference from the primordial values .
we also average the standard deviation over all collapsed objects in the universe .
our prescription for this is identical to that given by equation ( [ eq : uniavg ] ) , with @xmath187 taking the place of @xmath175 .
these results are plotted versus the minimum mass scale for galaxies in figure [ fig : siguniavg ] . comparing this to figure [ fig : unimean ]
, we again see that the relative importance of fluctuations increases with increasing mass scale .
though much recent work has focused on uncovering the primordial @xmath1li and d abundances ( e.g. , burles & tytler 1998a , b ; tytler et al . 1999 ; webb et al . 1997 ; ryan et al .
2000 ) , the @xmath0he abundance with a reported uncertainty of about @xmath188 , is still more accurately determined .
however , even this accuracy is at least an order of magnitude coarser than the mean effect of diffusion for low - mass objects .
but , as discussed in
[ sec : calc ] the @xmath0he abundance is also expected to vary from point to point within an object . in this section
, we re - examine the @xmath0he abundance data presented in izotov & thuan ( 1998 ) to determine whether or not such fluctuations are currently detectable . in izotov & thuan ( 1998 )
, the helium abundance is assumed to be linear in metallicity for low metallicity objects .
each data point consists of a measurement of the metallicity , @xmath189 , a measurement of the helium abundance @xmath190 , and estimations for the uncertainties in each , @xmath191 and @xmath192 .
a straight line is fit to the data , the intercept of which reveals the primordial ( zero - metallicity ) helium abundance .
the model we use is slightly different . because of diffusion we can not directly associate @xmath0he abundance measurements of zero - metallicity objects with the primordial @xmath0he abundance .
not only do we expect an object of a particular mass to have a @xmath0he abundance differing from the primordial value as shown in figure [ fig : meandev ] , but also we expect that different measurements of the single object would display an intrinsic scatter about this mean as shown in figure [ fig : sigplot ] . therefore , we fit the data to the following model . if an object has metallicity @xmath193 , the probability that it will have a @xmath0he mass fraction @xmath176 is denoted @xmath194 .
we account for measurement uncertainties by defining @xmath195 to be the probability of measuring @xmath190 if the object actually has abundance @xmath176 , and @xmath196 to be the probability that an object has metallicity @xmath193 if @xmath189 is what is observed .
thus , the total probability of a measurement simultaneously obtaining @xmath190 and @xmath189 is @xmath197 both @xmath193 and @xmath176 are intrinsically positive , but we here assume that the uncertainties in the measurments are small enough so that the probability distributions @xmath198 and @xmath199 may be taken to be gaussians , @xmath200 , \label{eq : zmodel}\ ] ] @xmath201 .
\label{eq : ymodel}\ ] ] as for our model , we assume that @xmath202 .
\label{eq : mcmodel}\ ] ] the parameters to be determined by the data are @xmath203 , which contains the physics of the metallicity dependence ; @xmath204 , which is the mean value for the @xmath0he abundance ; and @xmath205 , which represents the intrinsic scatter about the mean .
we use the integration in equation ( [ eq : probgen ] ) to obtain @xmath206 .
\label{eq : probint}\ ] ] if we determine the metallicity and helium abundance for @xmath207 objects , the probability for the entire measurement to occur is @xmath208 to maximize this probability with respect to the parameters @xmath203 , @xmath204 , and @xmath205 , we minimize the @xmath209-like function @xmath210 .
\label{eq : chitwo}\ ] ] a downhill simplex method and a direction set method yielded almost identical parameters for the minimum @xmath209 : @xmath211 , @xmath212 , and @xmath213 .
we used a monte carlo method to set confidence limits on these results . because of random errors , the observed data points @xmath189 and @xmath190 are not a unique realization of the true values of the quantities
. instead ,
for a given measurement of @xmath189 , there are an infinite number of @xmath190 that could have been measured .
the probabilities for obtaining these hypothetical measurements are assumed to be given by equation ( [ eq : probint ] ) .
we have generated @xmath214 of these hypothetical data sets by substituting the observed values for @xmath203 , @xmath204 , and @xmath205 into equation ( [ eq : probint ] ) .
for each hypothetical data set , we again minimized equation ( [ eq : chitwo ] ) to obtain a new triplet of parameters @xmath203 , @xmath204 , and @xmath205 . figure [ fig : mcplots ] shows the resulting intercepts @xmath204 plotted against the scatters @xmath205 . the uncertainty in our best - fit sigma and intercept are large , and so the observational situation must improve before we could settle on the true parameters .
we have explored the effect of diffusion during the linear growth of density perturbations in the neutral igm before the redshift of reionization and have found that element abundances within collapsed objects differ from their primordial values .
element diffusion leads to two related effects .
first , the mean element abundances of a collapsed object could differ from the primordial values ; and second , diffusion could result in abundance fluctuations within a single object . for objects with a total mass below @xmath215m@xmath216 the former effect dominates ; but for larger objects the latter effect is more important .
the magnitude of the effect is @xmath217@xmath218 , depending on the mass of the object and the element .
we have then used a press - schechter mass function to average these distributions over all collapsed objects in the universe , and again found the net effect to be @xmath219 .
this effect is yet too small to be detected since the best observations of element abundances currently have accuracies of @xmath3 .
if the effect of diffusion on element abundances will eventually be measured with sufficient precision in low - metallicity systems , the results could provide important clues about early structure formation in the igm .
in particular , the mean deviations within objects are sensitive to the minimum mass of a galaxy halo .
also , the overall magnitude of diffusion is sensitive to the redshift of reionization since diffusion is brought to a halt at that time . and as determinations of element abundances get more and more accurate , diffusion will eventually have to be accounted for in order to make comparisons with the predictions of big bang nucleosynthesis . | we consider element diffusion during the linear growth of structure in the intergalactic medium ( igm ) before the redshift of reionization .
the elements produced during big bang nucleosynthesis ( such as d , @xmath0he or @xmath1li ) condensed in collapsing regions more readily than hydrogen . diffusion was most effective when the igm was neutral since the collisional cross section between neutral species is smaller than the corresponding coulomb cross section in an ionized gas .
we find that diffusion led to small deviations in element abundances of up to @xmath2 .
abundance measurements to this level of precision in low metallicity environments could provide information about the thermal and ionization history of the primordial gas during the early formation of structure in the igm . |
we employ a pulsed ti : sapphire laser ( coherent mira 900d ) with a wavelength of 780 nm , mean power 1.3 w , repetition rate of 76 mhz and a pulse width of @xmath45 ps .
most of the laser output is directed into an lbo crystal for frequency doubling .
we obtain up to 300 mw second harmonic ; after subsequent spectral cleaning , about 100 mw remain
. then we implement parametric down - conversion in two periodically poled potassium titanyl phosphate crystals in a type ii spectrally and spatially degenerate , but polarization non - degenerate configuration .
the single photon detection is implemented using spcms by excelitas .
the modes entering the spcms are spectrally filtered and delivered to the detectors by means of single - mode fibers , which ensures proper preparation of the heralded photon mode @xcite . including the filters and fibers , the quantum efficiency of these detectors
is estimated as @xmath46 . without losses ,
the experimental double - coincidence event rate is @xmath47 100 hz , which corresponds to a probability of @xmath48 per pulse and the down - conversion amplitude @xmath49 . when we introduce a 10 db total loss equally distributed between modes c and d , the coincidence rate decreases by a factor of @xmath50 due to the reduction of the equivalent spcm efficiency down to @xmath51 .
the condition @xmath5 therefore holds even in the presence of losses .
we thank j. dowling , y. kurochkin , a. pushkina and a. fedorov for helpful discussions and the russian quantum center for support .
99 ourjoumtsev , a. , ferreyrol , f. , tualle - brouri , r. & grangier p. preparation of non - local superpositions of quasi - classical light states .
_ nature phys . _ * 5 * , 189 - 192 ( 2009 ) .
daki , b. _ et al .
_ quantum discord as resource for remote state preparation .
_ nature phys . _ * 8 * , 666 - 670 ( 2012 ) .
lopaeva , e. d. _ et al .
_ experimental realization of quantum illumination .
_ * 110 * , 153603 ( 2013 ) .
ulanov , a. e. _ et al .
_ undoing the effect of loss on quantum entanglement .
_ nature photon . _
* 9 * , 764 - 768 ( 2015 ) .
simon , c. & irvine , w. t. m. robust long - distance entanglement and a loophole - free bell test with ions and photons .
lett . _ * 91 * , 110405 ( 2003 ) .
moehring , d. l. _ et al .
_ entanglement of single atom quantum bits at a distance .
_ nature _ * 449 * , 68 - 71 ( 2007 ) .
hofmann , j. _ et al .
_ heralded entanglement between widely separated atoms
. _ science _ * 337 * , 72 - 75 ( 2012 ) .
hensen , b. _ et al._. experimental loophole - free violation of a bell inequality using entangled electron spins separated by 1.3 km .
_ nature _ * 526 * , 682 - 686 ( 2015 ) .
kok , p. , lee , h. & dowling , j. p. creation of large - photon - number path entanglement conditioned on photodetection .
a _ _ 65 _ , 052104 ( 2002 ) .
edamatsu , k. , shimizu , r. & itoh , t. measurement of the photonic de broglie wavelength of entangled photon pairs generated by spontaneous parametric down - conversion .
lett . _ * 89 * , 213601 ( 2002 ) .
nagata , t. , okamoto , r. , obrien , j. l. , sasaki , k. & takeuchi , s. beating the standard quantum limit with four - entangled photons .
_ science _ * 316 * , 726 - 729 ( 2007 ) .
dowling , j. p. quantum optical metrology the lowdown on high - n00n states .
_ contemporary physics _ * 49 * , 125 - 143 ( 2013 ) .
chen , p. , shu , c. , guo , x. , loy , m. m. t. , & du , s. measuring the biphoton temporal wave function with polarization - dependent and time - resolved two - photon interference .
lett . _ * 114 * , 010401 ( 2015 ) .
kacprowicz , m. , demkowicz - dobrzaski , r. , wasilewski , r. , banaszek , k. & walmsley , i. a. experimental quantum - enhanced estimation of a lossy phase shift .
_ nature photon .
_ * 4 * , 357 - 360 ( 2010 ) .
rubin , m. a. & kaushik , s. loss - induced limits to phase measurement precision with maximally entangled states .
rev . a _ * 75 * , 053805 ( 2007 ) .
hong , c. k. , ou , z. y. , & mandel , l. measurement of subpicosecond time intervals between two photons by interference .
lett . _ * 59 * , 2044 - 2046 ( 1987 ) .
ourjoumtsev , a. , jeong , h. , tualle - brouri , r. & grangier p. generation of optical schrdinger cats from photon number states .
_ nature _ * 448 * , 784 - 786 ( 2007 )
. andersen , u. l. , neergaard - nielsen , j. s. , van loock , p. & furusawa , a. hybrid discrete- and continuous - variable quantum information .
_ nature phys . _ * 11 * , 713719 ( 2015 ) .
dowling , j. p. , franson , j. d. , lee , h. , milburn , g. j. towards scalable linear - optical quantum computers . _ experimental aspects of quantum computing _ * 3 * , 205 - 213 ( 2005 ) lim , y. l. , beige , a. , & kwek , l. c. repeat - until - success linear optics distributed quantum computing .
lett . _ * 95 * , 1 - 4 ( 2005 ) .
etesse , j. , bouillard , m. , kanseri , b. , & tualle - brouri , r. experimental generation of squeezed cat states with an operation allowing iterative growth . _
lett . _ * 114 * , 1 - 5 ( 2015 ) .
walther , p. _ et al .
_ de broglie wavelength of a non - local four - photon state .
_ nature _ * 429 * , 158161 ( 2004 ) .
mitchell , m. w. , lundeen , j. s. & steinberg , a. m. super - resolving phase measurements with a multiphoton entangled state .
_ nature _ * 429 * , 161164 ( 2004 ) .
pan , j .- w . , bouwmeester , d. , weinfurter , h. & zeilinger , a. experimental entanglement swapping : entangling photons that never interacted , _ phys . rev .
* 80 * , 3891 - 3894 ( 1998 ) .
see supplementary information for details on extending our method to higher - photon number n00n states .
berry , d. w. & lvovsky , a. i. linear - optical processing can not increase photon efficiency . _
* 105 * , 203601 ( 2010 ) .
lvovsky , a. i. & raymer , m. g. continuous - variable optical quantum - state tomography .
phys . _ * 81 * , 299 ( 2009 ) .
kumar , r. _ et al .
_ versatile wideband balanced detector for quantum optical homodyne tomography . _
_ * 285 * , 5259 ( 2012 ) .
lvovsky , a. i. _ et al .
_ quantum state reconstruction of the single - photon fock state .
lett . _ * 87 * , 050402 ( 2001 ) .
huisman , s. r. _ et al .
_ instant single - photon fock state tomography .
_ optics letters _ * 34 * , 2739 - 2741 ( 2009 ) .
huang , k. _ et al .
_ optical synthesis of large - amplitude squeezed coherent - state superpositions with minimal resources .
lett . _ * 115 * , 023602 ( 2015 ) .
lvovsky , a. i. iterative maximum - likelihood reconstruction in quantum homodyne tomography .
b : quantum semiclass . opt . _ * 6 * , s556-s559 ( 2004 ) .
ehek , j. , hradil , z. , knill , e. , & lvovsky , a. i. diluted maximum - likelihood algorithm for quantum tomography .
a _ * 75 * , 1 - 5 ( 2007 ) . jeong , h. _ et al .
_ generation of hybrid entanglement of light .
_ nature photon . _
* 8 * , 564569 ( 2014 ) morin , o. _ et al .
_ remote creation of hybrid entanglement between particle - like and wave - like optical qubits .
_ nature photon . _ * 8 * , 570574 ( 2014 ) .
in the main text , we have shown that projecting modes c and d from alice s and bob s parametric amplifiers onto the two - photon n00n state collapses the counterpart pair of modes , a and b , onto the same state . here
we theoretically discuss the extension of this method to higher photon numbers . the scheme for projecting modes c and d onto the 4-photon n00n state is shown in fig .
overlapping two modes of that state , which we associate with annihilation operators @xmath52 and @xmath53 , on a symmetric beam splitter will transform it into @xmath54 subsequently , we implement photon subtraction in each channel by placing a weakly transmitting beam splitter therein and conditioning on clicks in the transmitted channels .
this produces the 2-photon n00n state @xmath55 which can be detected using the reverse hong - ou - mandel effect .
altogether , a coincidence click of all four spcms projects means that the 4-photon n00n state has been present at the input .
this scheme works perfectly if the detectors are number - discriminating .
otherwise one of the spcms may click in response to more than one photon .
such an event would imply that modes c and d have initially contained five or more photons , and hence have not been in the n00n state .
this situation can be avoided by pumping the parametric amplifiers at low powers so that the probability of producing @xmath56 pairs in the two crystals is negligible compared to that of producing @xmath0 pairs .
a more general scheme , capable of detecting the n00n state with an arbitrary even @xmath0 , involves applying operator @xmath59 to it .
this will result in a two - mode vacuum state , which can be detected directly using spcms .
operator @xmath59 for even values of @xmath0 is realized using the scheme similar to that of kok _ et al .
_ @xcite using decomposition @xmath60where @xmath61 .
operator @xmath62 is implemented , in turn , via the setup shown in fig .
the photons , tapped using weakly reflective beam splitters from both modes , are mixed on a symmetric beam splitter and detected using spcms . in this way , the state of the tapped modes c@xmath63 and @xmath64 is projected onto the two - photon n00n state thanks to the reverse hong - ou - mandel effect , so that one of the original modes loses two photons while the other one stays unaffected .
a phase shift of @xmath65 is applied to mode d@xmath63 in order to account for factor @xmath66 in eq .
( [ cndn ] ) . in order to detect state @xmath3 ,
@xmath67 operators @xmath68 must be applied in sequence as per eq .
( [ cndn ] ) . for low down - conversion amplitude , the probability for modes c and d to contain photons after this sequence is negligibly small ,
so the final double - vacuum detection is in fact not necessary .
however , the probability for this scheme to successfully herald a n00n state decreases exponentially with @xmath0 . | preparing highly entangled quantum states between remote parties is a major challenge for quantum communications @xcite . particularly promising in this context
are the n00n states , which are entangled @xmath0-photon wavepackets delocalized between two different locations , providing measurement sensitivity limited only by the uncertainty principle @xcite . however , these states are notoriously vulnerable to losses , making it difficult both to share them between remote locations , and to recombine them to exploit interference effects . here
we address this challenge by utilizing the reverse version of the hong - ou - mandel effect @xcite to prepare a high - fidelity two - photon n00n state shared between two parties connected by a lossy optical channel .
furthermore , we demonstrate that the enhanced phase sensitivity can be directly exploited in the two distant locations , and we remotely prepare superpositions of coherent states , known as schrdinger s cat states " @xcite . in the current race towards the practical implementation of quantum techniques for information processing and communications , a strong trend is to design loss - tolerant quantum protocols , such as the preparation of non - local superpositions of quasi - classical light states @xcite , discord - assisted remote state preparation @xcite , quantum illumination @xcite , undoing the effect of losses on continuous - variable entanglement @xcite , and the preparation of single qubit entangled states over a long distance @xcite . in this article
we will be interested in n00n states @xmath1 , which are useful in linear - optical quantum computation @xcite , quantum - optical state engineering @xcite , and the preparation of photon - number path entanglement @xcite .
but the most important potential application of these states is as a resource for quantum - enhanced metrology @xcite .
interference measurements with n00n states exhibit super - resolving properties : the number of fringes per wavelength equals @xmath0 , in contrast to a single fringe in the case of coherent states .
this property can be exploited for precise measurement of diverse physical quantities .
widespread application of n00n states for metrology is however precluded by their extreme sensitivity to losses . when exposed to even moderate losses ,
the degree of entanglement , and hence the super - resolution potential of the n00n states dramatically degrades to an extent that eliminates any advantage @xcite . in the present work ,
we address this challenge by developing a technique to losslessly produce n00n states between parties that are separated by a lossy quantum channel .
in addition , using n00n states usually requires to bring back together the two entangled parts , introducing more propagation losses .
but we will show that this second step is not required , and that super - resolution for the optical phase can be obtained remotely , by using homodyne detection @xcite . for n00n state production , we exploit some peculiar properties of the hong - ou - mandel ( hom ) effect @xcite , a well - known quantum interference phenomenon in which two indistinguishable photons that are overlapped on a symmetric beam splitter ( bs ) always emerge in the same output mode , preparing the n00n state @xmath2 in the beam splitter output .
our experiment relies upon the _ reverse hom effect _ , in which the measurements in the two output modes of the bs project them onto single - photon states . because of the time - reversible nature of quantum mechanics , such projection is equivalent to projecting the state of the input onto the two - photon n00n state .
if each of the beam splitter inputs is , in turn , entangled with other modes , these modes become entangled with each other thanks to entanglement swapping @xcite . whereas the original hom setting creates a two - photon n00n state , straightforward extensions will produce @xmath3 states ( see supplementary material @xcite ) .
specifically , consider a setup in which two pairs of modes ( a , c ) and ( b , d ) are prepared in two - mode squeezed states @xmath4 by means of non - degenerate parametric down - conversion .
modes c and d are then mixed on a symmetric bs , outputs of which are subjected to measurement in the photon number basis via single - photon counting modules ( spcms ) , as shown in fig .
[ f1 ] .
+ in the weak - squeezing limit @xmath5 , where @xmath6 is the spcm s efficiency , every spcm click is likely to be caused by no more than one photon .
then , a coincidence click in both spcms correspond to projection on state @xmath7 . due to the unitary nature of the bs operation
, this event assures that modes c and d were initially prepared in the n00n state ( [ eq1 ] ) .
this corresponds to two pairs of photons having been produced in either of two crystals ; therefore , the remaining modes a and b now also share a two - photon n00n state .
[ b ] component amounts to @xmath8 .
( b ) density matrix in the case of a @xmath9-ps mismatch between modes c and d , eliminating hom interference .
the emerged @xmath10 component is shown in gray .
green hats show the statistical error of the reconstruction.,title="fig:",width=288 ] a remarkable feature of this scheme its robustness to losses in channels c and d. indeed , such losses are equivalent to corresponding reduction of the spcm s quantum efficiencies @xcite . if the down - conversion amplitude @xmath11 is sufficiently small , the leading term in the state of channels @xmath12 and @xmath13 conditioned on the spcms coincidence click is still the two - photon n00n state .
one can therefore eliminate the effect of loss , however high it may be , on the purity of the n00n state .
conditioned on coincidence clicks of the two spcms ( see methods ) , we characterize the state of modes a and b by means of homodyne tomography @xcite .
the measured states in both cases are very close to the ideal state ( [ eq1 ] ) that has suffered a @xmath14 loss due to imperfect detection efficiency @xcite [ fig .
[ dm](a ) ] . to illustrate the reverse hom effect
, we measured the behavior of the state in modes a and b as a function of increasing temporal mismatch between modes c and d. as evidenced by fig . 3 , the fraction of the biphoton component @xmath15 in that state exhibits a dip that is characteristic as a signature of the hom effect . the tomographic state reconstruction for the case of complete mismatch is shown in fig . [ dm](b ) .
in addition to a macroscopic biphoton component , this state exhibits no off - diagonal terms because no coherence between modes a and b can emerge in the absence of interference .
[ t ] component in the reconstructed density matrix ( obtained from homodyne tomography , not from photon counting ) .
the red points and red curve correspond to the visibility of the classical interference between the master laser pulses that underwent the same spectral filtering as quantum modes c and d. the horizontal axes for the classical and quantum cases are scaled by a factor of @xmath16 with respect to each other to match the theoretically expected widths of the interference patterns.,title="fig : " ] the enhanced phase sensitivity is manifested by the mean values of observables @xmath17 and @xmath18 , where @xmath19 and @xmath20 are quadrature operators of modes a and b. for the one - photon @xmath21 and two - photon n00n states one has , respectively , @xmath22 and @xmath23 where @xmath24 .
we see that in order to compare between the one- and two - photon n00n states using quadrature measurements , we need to use different observables .
this notwithstanding , the phase dependence of the appropriate observable in each state is as expected : with period @xmath25 for each @xmath0-photon n00n state . an experimental check of this behavior is demonstrated in fig .
[ corr ] .
in agreement with the results of state reconstruction , the second - order interference of the @xmath26 state , generated in the presence of the loss , exhibits the same visibility as without loss . of alice s and
bob s phases for states @xmath27 ( a ) and @xmath26 ( b ) .
enhanced phase sensitivity is evident for the two - photon n00n state .
solid lines are theoretical predictions . ] being path - entangled , the n00n state can also be used for single - mode quantum - state engineering .
associating modes a and b with fictitious observers alice and bob , we consider a setting in which alice , by performing quadrature measurements on her mode , remotely prepares a state in bob s mode .
neglecting inefficiencies , alice s quadrature outcome @xmath28 measured at phase angle @xmath29 brings the bob s mode to state @xmath30 where @xmath31 are fock state wavefunctions , with @xmath32 being hermite polynomials . in this way , by postselecting on specific values of alice s observed quadrature , one can generate arbitrary superpositions of the 0- and 2-photon states , which approximate the even coherent - state superpositions ( css ) @xmath33 , sometimes viewed as a quantum optical implementation of the schrdinger s cat paradox @xcite .
the states of bob s mode for different outcomes of alice s homodyne measurement are displayed in fig .
[ wigs ] .
projection on @xmath34 ( fig .
[ wigs ] , first column ) results in superposition @xmath35 , partially mixed due to losses in alice s channel . after correcting for bob s homodyne detection inefficiency
, this state has fidelity of @xmath36 with the even css state of amplitude @xmath37 , squeezed with parameter @xmath38 .
this value compares favorably with state - of - the - art results @xcite , with the added advantage that our protocol is tolerant to the losses in the optical channel between alice and bob .
effective css amplitudes and approximation fidelities for other values of @xmath28 are shown in fig .
[ wigs ] , right bottom panel . with increasing @xmath28 ,
the two - photon fraction initially increases relative to the vacuum because the two - photon state wavefunction @xmath39 decreases faster than the vacuum wavefunction @xmath40 .
wavefunction @xmath39 changes sign near value of @xmath41 , where ideally a pure two - photon state in bob s mode should be observed . in practice , due to the losses and finite width of alice s post - selection window ,
a phase - insensitive mixture of @xmath42 is produced ( second column ) . in this region ,
the remotely prepared state approximates the css poorly because of the high two - photon component . for higher values of x , this two - photon component in bob s
state reduces again , resulting in increasingly faithful approximation of even csss with decreasing amplitudes ( third column ) . for very low amplitudes ,
this state approximates a weakly - squeezed vacuum state .
this is the case for @xmath43 ( fourth column ) : bob s quadrature spectrum exhibits squeezing by @xmath44 ( without efficiency correction ) .
the corresponding quadrature histogram is shown in the upper right panel of fig .
[ wigs ] .
the protocol developed here addresses the primary challenge in the way of employing nonclassical states of light , particularly the n00n state , for quantum metrology : optical losses . with a conventional fiber channel
, it allows establishing nearly ideal n00n entanglement over tens to hundreds of kilometers , no matter how high the loss in the channel connecting the parties is .
our scheme could be used , for example , for precise phase locking of local frequency standards at alice s and bob s sites . to realize that , alice and bob would employ local oscillators that are referenced to these standards .
then they would both prepare two - mode squeezed states and perform homodyne measurements on one of the modes while sending the other one to a third party located midway between them for a reverse hom measurement .
when that party obtains a coincidence event , it would inform alice and bob via a classical channel , in which case they would record the measured quadrature values .
the recorded data would subsequently be used to estimate the phase difference between the local oscillators , eliminating the contribution of the optical path by closely repeated measurements .
the result would then be fed back to one of the frequency standards .
the scheme can be generalized to higher order n00n states , and as shown above it can also be used for ultra - remote preparation of quantum states of light , with possible applications in quantum cryptography @xcite .
it is therefore a valuable addition to the quantum state engineering toolbox , for both discrete- and continuous - variable degrees of freedom of optical modes @xcite . |
pyrochlores with chemical formula a@xmath0b@xmath0o@xmath1 ( a and b generally stands for rare earth and transition metal ions , respectively ) have attracted surge of scientific interest over last one or two decades @xcite .
these materials have in - built frustration arising from structural geometry where the a and b cations form corner - shared tetrahedra . as a result , many interesting physical phenomena are observed .
most notable examples are spin glass @xcite , spin ice @xcite , spin liquid @xcite , anomalous hall effect @xcite , superconductivity @xcite , kondo - like behavior @xcite , etc . in recent times , ir - based pyrochlores ( a@xmath0ir@xmath0o@xmath1 )
have been at center place for investigation .
in addition to its geometric frustration , these materials exhibit sizable spin - orbit coupling ( soc ) effect due to presence of heavy ir atoms .
however , the electron - electron correlation effect ( @xmath4 ) , which is usually prominent in transition metal oxides , turns out relatively small in these systems due to extended character of 5@xmath5 orbitals of ir .
nonetheless , the soc and electronic correlation effect exhibit comparable energy scale , therefore , these ir - based oxides systems offer an ideal playground where the interesting physics coming out as an interplay between these two effects can be studied with control manner .
the theoretical calculation , indeed , has predicted various exotic topological insulating phases in pyrochlore iridates based on this intriguing interplay between soc and electronic correlation effect @xcite .
moreover , recent study has shown that physical properties in pyrochlore iridates a@xmath0ir@xmath0o@xmath1 largely depend on ionic radius of @xmath6 where the material gradually shifts from its insulating / magnetic behavior to metallic / nonmagnetic one with increasing ionic radius @xcite . within the family of pyrochlore iridates
, the y@xmath0ir@xmath0o@xmath1 has special interest as it has nonmagnetic y@xmath7 residing at a - site therefore , the magnetic properties are mostly determined by the contribution from ir sublattice .
in addition , one can exclude the possibility of @xmath8-@xmath5 exchange interactions which has been discussed to introduce complicated magnetic interactions in ir - based pyrochlores @xcite .
this is also important considering obvious interconnection between magnetic and electronic properties in these materials .
this shows y@xmath0ir@xmath0o@xmath1 is an ideal system to study the intercoupling effect of soc and @xmath4 while neglecting other associated contributions .
this material by nature is insulator where the resistivity increases by couple of orders at low temperature @xcite .
the magnetic state is , however , still debated .
while the neutron diffraction / scattering measurements reveal no sign of magnetic long - range ordering @xcite , the muon spin rotation / relaxation measurement @xcite , on the other hand , has shown well - defined spontaneous oscillations in muon symmetry indicating long - range ordering . considering the in - built frustration in pyrochlore lattice ,
the glass - like behavior could also be a low temperature magnetic state in y@xmath0ir@xmath0o@xmath1 .
note , that y - based other pyrochlores i.e. , y@xmath0mo@xmath0o@xmath1 , y@xmath0ru@xmath0o@xmath1 have shown spin - glass behavior @xcite .
this underlines the fact that there is need for conclusive understanding of magnetic state in y@xmath0ir@xmath0o@xmath1 which is further amplified with the fact that recent calculation proposes topological weyl semimetallic phase in y@xmath0ir@xmath0o@xmath1 based on its ground state magnetic character @xcite .
in this work , we present detailed structural , magnetic , thermodynamic and transport properties of pyrochlore iridate y@xmath0ir@xmath0o@xmath1 .
magnetic measurements show onset of magnetic irreversibility at temperature @xmath2 @xmath3 160 k , however , no sharp peak / cusp is observed at this temperature .
analysis of magnetization data shows no spontaneous moment at low temperature . instead , reasonable magnetic relaxation and aging behavior similar to metamagnetic systems such as , spin - glasse , superparamagnet has been observed .
temperature dependent x - ray diffraction ( xrd ) measurements show there is no crystallographic phase / symmetry transition down to low temperature .
the material is found to be insulating throughout the temperature range where the resistivity follows power law behavior with temperature .
the negative magnetoresistance at low temperature can be explained with quantum interference effect .
polycrystalline samples of y@xmath0ir@xmath0o@xmath1 have been prepared using standard solid state method .
the ingredient powder materials y@xmath0o@xmath9 and iro@xmath0 with phase purity @xmath10 99.99% ( sigma - aldrich ) are taken in stoichiometric ratio .
the mixture are ground well , subsequently pelletized and heated in air at 1000@xmath11c for 96 hours , at 1100@xmath11c for 96 hours and at 1160@xmath11c for 252 hours with intermediate grinding .
the heating and cooling rate has been used 3@xmath11c / min .
the material has been characterized by powder x - ray diffraction ( xrd ) using a rigaku made diffractometer ( model : miniflex600 ) with cuk@xmath12 radiation at room temperature .
the xrd data have also been collected at different temperatures in the range of 300 - 20 k using a panalytical xpert powder diffractomer .
the low temperature is achieved using a helium close cycle refregerator ( ccr ) based cold head where proper temperature stabilization is ensured by waiting sufficiently before collecting data .
data have been collected in the range of 2@xmath13 = 10 - 90@xmath11 with a step of @xmath14 = 0.033@xmath11 and scan rate of 2@xmath11/min .
the collected xrd data have been analyzed using reitveld refinement program ( fullprof ) by young _
et al _ @xcite .
the rietveld analysis shows the material is in single phase except presence of small fraction of unreacted ingredient materials .
it can be noted that after certain heat treatment , no further improvement in terms of unreacted ingredients is observed , rather they are found to grow .
magnetically , the y@xmath0o@xmath9 is diamagnetic and iro@xmath0 is paramagnetic therefore , these residual impurities are expected to influence the magnetic properties of desired material minimally .
note , that similar impurity phases have also been observed in previous studies @xcite .
in general , pyrochlore materials adopt cubic crystallographic phase with _
m _ symmetry @xcite .
there are two sets of oxygen atoms , therefore pyrochlore structures consist of four nonequivalent crystallographic positions ( table i ) . the resulting composition can be written as y@xmath0ir@xmath0o@xmath16o@xmath17 .
the ir@xmath18 cations are six coordinated and they form corner shared iro@xmath16 octahedra with equal length ( @xmath19 ) of ir - o bonds . the eight coordinated ( six o and two o@xmath17 ) y@xmath7 cations form distorted cubes where y - o bonds are equal in length and larger than the y - o@xmath17 ones .
the only variable position in pyrochlore structures is @xmath20 position with o ( table i ) which determines its structural stability .
for instance , @xmath20 = 0.3125 gives perfect iro@xmath16 octahedra while @xmath20 = 0.375 results in perfect cubes around y. the pyrochlore structure can be viewed as interpenetrating two sub - lattices consists of each type of tetrahedra i.e. , y@xmath21o@xmath17 and ir@xmath21@xmath22 , where @xmath22 implies empty center - site ( 8a site ) .
as these tetrahedra s are corner shared so the magnetic atoms sitting at vertices of tetrahedra induces frustration .
the rietveld analysis of our xrd data confirms that y@xmath0ir@xmath0o@xmath1 crystallizes in cubic phase with _
m _ symmetry ( fig .
the unit cell lattice parameter is found to be @xmath23 = 10.2445 .
the bond lengths ( @xmath5 ) and angles ( @xmath24 ) are determined from structural parameters as given in table ii . in iro@xmath16
octahedra , bond angles @xmath25o - ir - o@xmath10 in basal plane are not equal while opposite ones are same in value .
the bond lengths and bond angels for present materials ( table ii ) shows good agreement with the reported values @xcite .
dc magnetization data have been collected using a vibrating sample magnetometer ( ppms , quantum design ) , specific heat have been measured following semi - adiabatic method and electrical transport properties have been measured using a home - built insert fitted with oxford superconducting magnet .
ir@xmath0o@xmath1 is shown at room temperature along with rietveld analysis.,width=302 ]
magnetization as a function of temperature has been measured following zero field cooled ( zfc ) and field cooled ( fc ) protocol after applying magnetic field of 1 koe .
2 shows with decreasing temperature , @xmath26 increases , however , there are two observations : around 160 k there is change in slope in data and below around 50 k the magnetization increases very sharply .
the corresponding @xmath27 shows identical nature of @xmath26 till around 160 k , and below this temperature a bifurcation is observed between both branches of magnetization .
the onset of this irreversibility between zfc and fc magnetization is at temperature @xmath2 @xmath3 160 k. some disagreement has earlier been observed about the value of @xmath2 . while the majority of studies@xcite have shown the @xmath2 in the range of 150 - 160 k accompanied by a feeble kink in @xmath28 around @xmath2 , the another study by disseler _
@xcite shows a higher @xmath2 ( 190 k ) and cusp in @xmath28 at @xmath2 .
our observed @xmath2 @xmath3 160 k agrees well with the majority of studies .
as discussed , the nature of magnetic phase below @xmath2 for this material is debated as both presence as well as absence of long - range ordering has been discussed @xcite . nonetheless , the bifurcation around @xmath2 in fig .
2 is suggestive of spin ordering related to long - range or short - range type spin ordering or even spin - glass like freezing .
the cusp in magnetic susceptibility , which generally signifies the onset of magnetic ordering , is notably absent in fig .
this similar absence of peak has been observed in other studies as well@xcite .
this qualitatively suggests magnetic frustration is quite significant in this material which perhaps prevent the long - range type magnetic ordering .
data are collected following zfc and fc protocol .
inset shows temperature dependent inverse susceptibility ( @xmath29 = @xmath30 ) deduced from zfc data for y@xmath0ir@xmath0o@xmath1 .
the solid line is fitting due to modified curie - weiss behavior ( discussed in text).,width=302 ] in inset of fig .
2 , we show temperature variation of inverse susceptibility ( @xmath29 ) which is deduced from zfc magnetization data @xmath31 measured in 10 koe magnetic field . as evident , above @xmath2 , the @xmath32 shows a linear behavior over a wide temperature range .
the @xmath32 has been fitted to modified curie - weiss ( cw ) behavior @xmath33 = @xmath34 = @xmath35 + @xmath36 , where @xmath35 is the temperature independent susceptibility , @xmath37 is the curie constant and @xmath38 is the curie temperature .
the inset of figure shows fitting of data as represented by solid line in temperature regime between 200 - 300 k. the obtained fitted parameters are @xmath35 = 1.9 @xmath39 10@xmath40 emu / mole , @xmath37 = 1.4 emu k / mole and @xmath38 = -386 k. the sign as well as magnitude of @xmath38 imply a reasonably strong non - ferromagnetic type magnetic interaction .
moreover , the obtained @xmath37 corresponds to effective paramagnetic moment , @xmath41 = 3.35 @xmath42/f.u .
the estimated @xmath41 appears higher than the expected value ( @xmath43 ) which turns out to be 1.73 @xmath42/f.u in case of spin - only value ( @xmath44 = 2 and @xmath45 = 1/2 ) .
these pyrochlore compounds are inherently frustrated due to their geometrical configuration which often lead to exotic magnetic ground state such as , spin liquid , spin ice , spin glass , etc .
the relevant frustration parameter ( @xmath8 ) , which is conventionally defined as ratio @xmath46/@xmath47 , where @xmath47 is the ordering or spin freezing temperature , has been calculated to be @xmath3 2.4 taking @xmath47 = 160 k. this @xmath8 value in present material is significant though much lower than the value ( @xmath10 100 ) usually seen for highly frustrated spin - liquid compounds as in pr@xmath0ir@xmath0o@xmath1 @xcite .
ir@xmath0o@xmath1 at 5 k. ( b ) arrott plot ( @xmath48 vs @xmath49 ) of magnetization data is shown at 5 k.,width=321 ] the magnetic field dependent of magnetization data recorded at 5 k in the field range of @xmath5070 koe are presented at fig .
the @xmath51 data do not show linear behavior , and no sign of saturation is observed till 70 koe .
the moment at 70 koe is found to be 0.095 @xmath42/f.u which is much lower than the expected spin - only value ( @xmath52 ) i.e. , 1 @xmath42/f.u for s = 1/2 .
this low value of magnetization is consistent with other studies @xcite .
a close observation , however , shows a very small hysteresis in @xmath51 plot with coercive field @xmath53 @xmath3 100 oe and remanent magnetization @xmath54 @xmath3 4.7 @xmath39
10@xmath40 @xmath42/f.u . to examine the magnetic state we have analyzed the @xmath51 data in term of arrott plot@xcite ( @xmath48 vs @xmath49 ) as shown in fig .
the significance of arrott plot is that if slope is taken at high field regime and yields an intercept on positive @xmath48 axis , this implies presence of spontaneous magnetization or fm ordering . as evident in fig .
3b , the plot renders intercept at negative @xmath48 axis .
this clearly shows absence of spontaneous magnetization viz - a - viz fm ordering in present y@xmath0ir@xmath0o@xmath1 material .
this finding is important because long - range type antiferromagnet with non - collinear spin ordering ( i.e. , all - in / all - out type ) would exhibit a weak ferromagnetic behavior which will also be manifested in arrott plot similar to sr@xmath0iro@xmath21 @xcite .
nonetheless , the small hysteresis in @xmath51 may possibly originate from short - range type fm ordering or freezing of spins at low temperatures .
4 shows temperature dependence of specific heat ( @xmath37 ) in form of @xmath55 vs @xmath56 for y@xmath0ir@xmath0o@xmath1 .
the specific heat does not exhibit any observable anomaly around temperature @xmath2 .
further , the @xmath57 data do not show any anomaly or upturn at low temperature which negates the schottky - like contribution . at low temperature ,
the @xmath57 data can be well explained with @xmath55 = @xmath58 + @xmath59 , where @xmath58 and @xmath60 are the electronic and lattice coefficients of specific heat , respectively @xcite .
interestingly , the linearity in @xmath55 vs @xmath61 extends to quite high temperature ( @xmath3 17 k ) .
the straight line fitting ( inset of fig .
4 ) yields @xmath58 = 32.0(5 ) mj mol@xmath62 k@xmath63 and @xmath60 = 0.592(5 ) mj mol@xmath62 k@xmath40 . using this @xmath60 , we have calculated debye temperature @xmath64 (= [ ( 12@xmath65)/(5@xmath60)]@xmath66 , where @xmath67 is the molar gas constant and @xmath68 is the number of atoms per formula unit ) to be 330.7 k. the @xmath60 and @xmath64 is consistent with the reported studies .
the @xmath58 , on other hand , is not consistent in literature as both zero@xcite as well as finite@xcite @xmath58 is reported which is quite intriguing considering insulating behavior of this material . using obtained @xmath58
we have calculated density of states ( dos ) at fermi surface @xmath69 with following relation , is shown for y@xmath0ir@xmath0o@xmath1 .
inset shows @xmath55 vs @xmath61 .
the straight line is due to linear fitting to @xmath55 = @xmath58 + @xmath59 ( see text).,width=302 ] @xmath70 @xmath71 where @xmath72 is the boltzmann constant and @xmath73 is the electron - phonon coupling constant which can be considered negligibly small . from eqs . 1 and 2
, we have calculated @xmath69 to be 13.6 states ev@xmath62 f.u@xmath62 . this finite dos is rather surprising for this material with strong insulating phase ( discussed later ) .
this estimated dos is quite consistent with the calculated dos values for y@xmath0ir@xmath0o@xmath1 where major contribution comes from ir 5@xmath5 atoms @xcite .
it can be mentioned that finite @xmath58 and resulting dos has been observed for insulating pyrochlore iridate eu@xmath0ir@xmath0o@xmath1 @xcite .
given that this material is highly insulator , localized picture of dos can explain this insulating behavior .
the magnetic ordering as well as the structural symmetry have been theoretically discussed to play profound role in realizing various exotic electronic phases in the class of pyrochlore iridates . in case of materials lacking magnetic ordering ,
the recent calculation has predicted for various topologically insulated phases which comes out as an interplay between spin - orbit coupling effect and electronic correlation effect @xcite .
the fact that long - range magnetic ordering breaks the time reversal symmetry which has motivated calculation to propose exotic weyl - type semi - metal in ir - based pyrochlores @xcite .
however , the structural symmetry also offers a vital clue for weyl semi - metal as this phase demands breaking of either time reversal symmetry or inversion symmetry .
ir@xmath0o@xmath1 are shown at selective temperatures of 300 , 170 , 150 and 20 k. the plots at 170 , 150 and 20 k are shifted vertically for clarity.,width=302 ] in an aim to understand the evolution of structural symmetry across the magnetic irreversibility temperature @xmath2 we have carried out temperature dependent xrd at various temperatures down to 20 k. fig .
5 shows representative xrd plot of y@xmath0ir@xmath0o@xmath1 taken at 300 , 170 , 150 and 20 k where the temperatures correspond to room temperature pm state ( 300 k ) , just above ( 170 k ) and below ( 150 k ) of @xmath2 and low temperature magnetic state ( 20 k ) .
it is clear in figure that xrd pattern does not exhibit any change with temperature in terms of peak splitting or arising of new peak(s ) .
this primarily implies structural symmetry does not change with temperature or the spin ordering at @xmath2 is not accompanied by any change in structural symmetry .
all the xrd data collected at various temperatures are analyzed with rietveld refinement program which shows the material retains cubic _
fd@xmath15 m _ symmetry down to lowest temperature 20 k. though for the first time we present temperature dependent structural evolution using xrd , the similar structural behavior with temperature has also been previously shown from neutron diffraction for y@xmath0ir@xmath0o@xmath1 @xcite .
table i shows lattice parameters and crystallographic positions of all the atoms of y@xmath0ir@xmath0o@xmath1 obtained from rietveld analysis at two representative temperatures i.e. , 300 and 20 k. it is observed that values do not change significantly over wide temperature range .
.[label ] unit cell parameters and crystallographic positions determined from the rietveld profile refinement of the powder xrd patterns for y@xmath0ir@xmath0o@xmath1 at 300 and 20 k. here o and o@xmath17 refers to two different oxygen position ( discussed in text ) . [
cols="^,^,^ " , ] where @xmath74 is the characteristic relaxation time and the @xmath60 is the stretching exponent with value in the range of 0 @xmath75 1 .
the @xmath60 = 1 recoves usual exponential relaxation behavior where the single energy barrier determines the relaxation behavior .
the stretched exponential behavior , on other hand , originates if the system has multiple energy barrier which renders distribution of relaxation times . the solid line in fig .
7 shows a representative fitting of our data using eq . 3 for @xmath76
= 10@xmath77 s. the parameters ( @xmath74 and @xmath60 ) obtained from fitting are shown in table iii .
the value of @xmath74 turns out very high .
both these characteristic parameters ( @xmath74 and @xmath60 ) agree well with the values for glassy systems . at low temperatures , for instance , the typical values of @xmath74 and @xmath60 are obtained in the range of 10@xmath78 - 10@xmath79 s and 0.3 - 0.4 , respectively for classical spin - glass system i.e. , ag : mn @xcite . while the @xmath74 decreases with wait time the @xmath60 , on contrary , increases .
this magnetic relaxation as well as aging behavior in this material is quite interesting as it primarily suggests that the low temperature ground state in y@xmath0ir@xmath0o@xmath1 is not an equilibrium state and the system relaxes to low energy state unlike long - range type magnetically ordered system .
the decrease of @xmath74 and simultaneous increase of @xmath60 with @xmath76 imply that while aging the system tries to achieve low energy states and multiple energy barriers are gradually removed .
the observed behavior akin to glassy systems in present material is quite intriguing .
the calculation taking only classical heisenberg interaction in pyrochlore lattice with antiferromagnetic interaction indeed show inherent frustration does not favor long - range ordering @xcite .
the inclusion of additional dzyaloshinsky - moriya ( dm ) interaction , however , show a transition to long - range ordered phase at low temperature @xcite .
the sizable spin - orbit coupling effect in iridium oxides may induce dzyaloshinsky - moriya type interaction but this can not be generalized as pr@xmath0ir@xmath0o@xmath1 does not exhibit magnetically ordered phase down to low temperature @xcite .
ir@xmath0o@xmath1 in semilogarithmic plot .
inset shows same resistivity data in ln - ln plot .
the solid line shows straight line fitting.,width=264 ] now we look at the electronic transport behavior in this material .
an insulating behavior is evident for y@xmath0ir@xmath0o@xmath1 in previous studies , however , the resistivity value at low temperature has shown wide variation @xcite .
8 shows resistivity ( @xmath80 ) of this material as a function of temperature .
an insulating behavior is observed throughout the temperature regime , even across @xmath2 .
the figure shows resistivity increases by couple of orders at low temperatures .
in contrast to thermally activated hopping transport which is commonly observed for insulating or semiconducting systems , we find that temperature dependence of resistivity follows a power law behavior as , @xmath81 inset of fig .
8 shows ln - ln plot of @xmath82 data and the fitted straight line confirms the validity of eq .
4 . the straight line fitting yields exponent @xmath68 = 2.987(2 ) .
the similar power law driven electronic transport has also been observed for y@xmath0ir@xmath0o@xmath1 with the exponent @xmath68 = 4 @xcite .
we have also looked into the electronic transport behavior in presence of magnetic field .
the response of electrical resistance in application of magnetic field or the magnetoresistance ( mr ) effect , which is defined as @xmath83 = @xmath84/\rho(0)$ ] , has attracted lot of interest for materials with strong spin - orbit coupling effect .
usually , a positive mr has been observed in such materials which is largely explained employing ` weak antilocalization ' effect .
exemplary compounds are bi@xmath0se@xmath9 @xcite , bi@xmath0te@xmath9 @xcite , au covered mg film @xcite , even in ir - based na@xmath0iro@xmath9 films @xcite . in fact , to our knowledge the present mr investigation is first of its kind in y@xmath0ir@xmath0o@xmath1 , even in family of pyrochlore iridates . the main panel of fig .
9 shows observed mr behavior as a function of magnetic field . as evident , the resistivity decreases in magnetic field exhibiting negative mr behavior .
though at maximum available field of 8 tesla the mr continues to increase without sign of saturation , we find mr of value @xmath3 11% at this field .
upper inset shows quadratic field dependence of mr and lower inset shows linear variation of mr with square of magnetization ( emu / mol ) for y@xmath0ir@xmath0o@xmath1.,width=302 ] from theoretical understanding , the negative mr at low temperatures in weakly disordered systems has been largely explained using ` weak localization ' effect .
the quantum interference ( qi ) of electronic wave functions is the basis of this localization effect and this is treated as quantum correction to the classical drude equation for conductivity . in this scenario , application of magnetic field
tend to destroy the qi effect , hence promotes the conductivity which results in negative mr .
sivan , entin - wohlman and imry@xcite applied a method of critical path analysis to understand the conductivity over many random paths and found a quadratic field dependence of mr .
in fact , this behavior has been experimentally observed in variety kind of systems .
the upper inset of fig .
9 shows mr at 5 k varies linearly with @xmath85 up to field @xmath3 4.3 tesla .
this quadratic field dependence of mr implies in low field regime negative mr originates from weak localization effect .
further , what we find interesting is that above this field ( 4.3 tesla ) , mr is found to vary linearly with @xmath48 ( lower inset of fig .
such correlation between electrical resistivity and magnetization is evident for pyrochlore iridates a@xmath0ir@xmath0o@xmath1 with a = pr and nd ascribing it to kondo - like screening behavior @xcite , but this behavior is observed in present y@xmath0ir@xmath0o@xmath1 in higher field regime . more surprisingly , unlike the previous pr and nd based compounds , in present material with y@xmath7 ion sitting at a - site do not posses any @xmath8 electrons .
we speculate that internal fields originating from more orderly spins in higher fields is responsible for this quadratic dependence of mr with magnetization .
nonetheless , electronic transport behavior in magnetic field in iridate systems needs to be understood thoroughly by employing theoretical calculations as well as studying other similar systems .
in conclusion , detailed structural , magnetic , thermodynamic and transport properties are studied in polycrystalline sample of pyrochlore iridate y@xmath0ir@xmath0o@xmath1 .
the material shows magnetic irreversibility at temperature @xmath2 @xmath3 160 k. however , below @xmath2 no sign of long - range type ferromagnetic ordering is observed .
the absence of anomaly in specific heat around @xmath2 also favors absence of long - range type magnetic ordering .
the specific heat data indicate reasonable dos at fermi level . temperature dependent xrd data shows the material under study retains its structural symmetry down to low temperature . from magnetic relaxation and aging behavior
we infer that y@xmath0ir@xmath0o@xmath1 shows glassy behavior at low temperature .
the system is insulator throughout the temperature regime where the electronic transport mechanism follows power law behavior with temperature .
the insulating behavior suggests perhaps dos are localized .
the negative magnetoresistance in this soc dominated material is quite intriguing which is explained with quantum interference effect .
we are thankful to airf , jnu for low temperature xrd measurements and manoj pratap singh for the help in measurements .
we acknowledge ugc - dae csr , indore for magnetization and resistivity measurements .
we sincerely thank alok banerjee and rajeev rawat for the magnetization , specific heat and resistivity data and fruitful discussions .
we thank kranti kumar and sachin kumar for the helps in measurements .
hk acknowledges ugc , india for bsr fellowship .
m. a. subramanian , g. aravamudan and g. v. subba rao , prog .
solid state chem .
* 15 * , 55 ( 1983 ) .
j. s. gardner , m. j. p. gingras , j. e. greedan , rev .
phys . * 82 * , 53 ( 2010 ) .
m. j. p. gingras , c. v. stager , n. p. raju , b. d. gaulin , and j. e. greedan , phys .
lett . * 78 * , 947 ( 1997 ) .
s. yoshii and m. sato , j. phys .
68 , 3034 ( 1999 ) .
s. t. bramwell , m. j. harris , b. c. den hertog , m. j. p. gingras , j. s. gardner , d. f. mcmorrow , a. r. wildes , a. l. cornelius , j. d. m. champion , r. g. melko , and t. fennell , phys .
lett . * 87 * , 047205 ( 2001 ) .
h. fukazawa , r. g. melko , r. higashinaka , y. maeno , and m. j. p. gingras , phys .
b * 65 * , 054410 ( 2002 )
. j. s. gardner , s. r. dunsiger , b. d. gaulin , m. j. p. gingras , j. e. greedan , r. f. kiefl , m. d. lumsden , w. a. macfarlane , n. p. raju , j. e. sonier , i. swainson , and z. tun , phys .
lett . * 82 * , 1012 ( 1999 ) .
s. nakatsuji , y. machida , y. maeno , t. tayama , t. sakakibara , j. van duijn , l. balicas , j. n. millican , r. t. macaluso and j. y. chan , phys .
* 96 * , 087204 ( 2006 ) .
y. taguchi , y. oohara , h. yoshizawa , n. nagaosa and y. tokura , science * 291 * , 2573 ( 2001 ) .
y. machida , s. nakatsuji , y. maeno , t. tayama , t. sakakibara , and s. onoda , phys .
lett . * 98 * , 057203 ( 2007 ) .
m. hanawa , y. muraoka , t. tayama , t. sakakibara , j. yamaura and z. hiroi , phys .
lett . * 87 * , 187001 ( 2001 ) .
d. pesin and l. balents , nature phys * 6 * , 376 ( 2010 ) .
k. matsuhira , m. wakeshima , y. hinatsu and s. takagi , j. phys .
80 * , 094701 ( 2011 ) .
g. chen and m. hermele , phys .
b * 86 * , 235129 ( 2012 ) .
m. c. shapiro , s. c. riggs , m. b. stone , c. r. de la cruz , s. chi , a. a. podlesnyak and i. r. fisher , phys .
b * 85 * , 214434 ( 2012 ) . m. soda , n. aito , y. kurahashi , y. kobayashi , m. sato , physica b * 329 * , 1071 ( 2003 ) s. m. disseler , c. dhital , a. amato , s. r. giblin , c. de la cruz , s. d. wilson , and m. j. graf , phys . rev .
b * 86 * , 014428 ( 2012 ) .
n. taira , m. wakeshima and y. hinatsu , j. phys . : condens .
matter * 13 * 5527 ( 2001 ) .
x. wan , a. m. turner , a. vishwanath , and s. y. savrasov , phys .
b * 83 * , 205101 ( 2011 ) .
r. a. young , a. sakthivel , t. s. moss and c. o. paiva - santos , _ users guide to program dbws-9411 , atlanta : georgia institute of technology _ ( 1994 ) .
w. k. zhu , m. wang , b. seradjeh , f. yang , and s. x. zhang , phys . rev .
b * 90 * , 054419 ( 2014 ) .
a. arrott , phys .
rev . * 108 * , 1394 ( 1957 ) .
i. n. bhatti , r. rawat , a . banerjee and a. k. pramanik , j. phys . : condens .
matter * 27 * 016005 ( 2014 ) .
e. s. r. gopal , _ specific heats at low temperatures _ ( plenum , new york , 1966 ) .
r. s. singh , v. r. r. medicherla , k. maiti , and e. v. sampathkumaran , phys .
b * 77 * , 201102 , ( 2008 ) .
h. fukazawa and y. maeno , j. phys .
jpn . * 71 * , 2578 ( 2002 ) .
j. j. ishikawa , eoin c. t. @xmath86farrell , and s. nakatsuji , phys .
b * 85 * , 245109 ( 2012 ) .
j. a. mydosh , _ spin glasses : an experimental introduction _ , crc press ( 1993 ) .
r. v. chamberlin , g. mozurkewich , and r. orbach , phys .
lett . * 52 * , 867 ( 1984 ) .
r. hoogerbeets , wei - li luo , and r. orbach , phys . rev .
lett . * 55 * , 111 ( 1985 ) .
j. n. reimers , phys .
b * 45 * , 7287 ( 1992 ) .
m. elhajal , b. canals , r. sunyer and c. lacroix , phys .
b * 71 * , 094420 ( 2005 ) .
j. chen , h. j. qin , f. yang , j. liu , t. guan , f. m. qu , g. h. zhang , j. r. shi , x. c. xie , c. l. yang , k. h. wu , y. q. li , and l. lu , phys .
lett . * 105 * , 176602 ( 2010 ) .
he , g. wang , t. zhang , i - k .
sou , g. k. l. wong , and j - n .
wang , phys .
rev . lett . *
106 * , 166805 ( 2011 ) .
g. bergman , phys .
* 48 * , 1046 ( 1982 ) .
m. jenderka , j. barzola - quiquia , z. zhang , h. frenzel , m. grundmann , and m. lorenz , phys .
b * 88 * , 045111 ( 2013 ) .
u. sivan and o. entin - wohlman , and y. imry , phys .
* 60 * , 1566 ( 1988 ) .
s. m. disseler , c. dhital , t. c. hogan , a. amato , s. r. giblin , c. de la cruz , a. daoud - aladine , s. d. wilson , and m. j. graf , phys .
b * 85 * , 174441 ( 2012 ) . | geometrical frustration and spin - orbit coupling effect together play vital role to influence properties in pyrochlore based iridium oxides .
here we have investigated detailed structural , magnetic , thermodynamic and transport properties of pyrochlore iridate y@xmath0ir@xmath0o@xmath1 .
magnetization data show onset of magnetic irreversibility around temperature @xmath2 @xmath3 160 k , however , no sign of long - range type ferromagnetic ordering is observed below @xmath2 .
specific heat data show no visible anomaly across @xmath2 , and the analysis of data indicate sizable density of states across fermi level .
temperature dependent x - ray diffraction measurements show no change in structural symmetry down to low temperature .
the material , on the other hand , shows significant relaxation and aging behavior similar to glassy dynamics .
the electronic charge transport in this highly insulating system is found to follow power law dependence with temperature .
the material shows negative magnetoresistance which is explained with quantum interference effect . |
evolution of star clusters is driven by a variety of dynamical processes .
two - body relaxation , stellar evolution , evaporation of stars , and tidal truncation have all been shown to play a role .
in addition to the internal effects , the external tidal shocks from the galaxy are very important . @xcite ( 1997 ) found that tidal shocks contribute at least as much as two - body relaxation to the destruction of the current galactic sample of globular clusters .
-body simulations of @xcite ( 1999 ) provide a detailed account of the effect of individual tidal shocks on the stellar energy distribution . in this paper
we describe the implementation of tidal shocks in a fokker - planck code and present new models of globular cluster evolution .
the study of the dynamical evolution of globular clusters has a long and rich history .
@xcite ( 1987 ) , @xcite , and @xcite provide comprehensive reviews of the evolution of isolated and tidally - limited clusters .
two - body relaxation causes loosely - bound stars gain velocity higher than the escape velocity , which leads to their evaporation from the cluster ( @xcite 1938 ; @xcite 1940 ) . even without evaporation ,
clusters would experience a catastrophic collapse driven by the negative heat capacity and the conductive transport of the thermal energy from the inner to the outer parts of the cluster ( @xcite ; @xcite ; @xcite ) .
the run - away process leads to a gravothermal catastrophe , or core collapse .
the collapse can be reversed when a sufficient amount of heating is provided in the core via formation of binaries .
the successive mergers of normal stars and subsequent supernova explosions were also shown to be effective in halting the collapse ( lee 1987 ) .
after the reversal of core collapse , the evolution of the cluster can be characterized by quasi - static expansion with possible gravothermal oscillations .
the tidal field of the galaxy affects cluster dynamics in two ways .
the stellar distribution is effectively cut off at the point where the ambient galactic density exceeds that of stars in the cluster .
@xcite found that clusters with short initial relaxation time dissolve in @xmath4 orbital periods around the galaxy , where @xmath5 is the initial number of stars in the cluster .
tidal effects may also cause cluster rotation as the stars on prograde orbits have higher chances to be ejected than those on retrograde orbits ( e.g. , @xcite ) .
in addition , the time - varying tidal forces cause gravitational shocks when the cluster passes through the disk of the galaxy ( @xcite 1972 ) or comes near the galactic nucleus ( e.g. , @xcite 1987 ) .
tidal shocks increase random motion of stars and reduce cluster binding energy .
finally , @xcite ( 1994 ) and @xcite ( 1997 ) emphasized that the tidal shock relaxation can be very important in accelerating cluster evolution .
recent studies ( @xcite 1997a - c ; @xcite ) show that these effects can lead to substantial evolution of the globular cluster population .
all of the above processes are included in our fokker - planck code .
we discuss the detailed implementation of tidal shocks and demonstrate their effects on the evolution of clusters with a wide range of parameters .
we consider both single - mass models and models including a spectrum of stellar masses .
mass segregation and rapid energy transfer between the stars of different masses lead to an even faster core collapse and destruction of globular clusters .
although we do not include the effects of stellar evolution , they are likely to be important only at the early stages of cluster evolution . @xcite and @xcite showed that the mass loss from the short - lived massive stars can disrupt weakly concentrated clusters ( with @xmath6 ) .
we also ignore the effects of rotation and velocity anisotropy but discuss the latter at the end of the paper .
we first review the theory of tidal shocks in 2 .
the new features of the fokker - planck code are described in 3 .
we start with an illustrative case that demonstrates the importance of tidal shocks and adiabatic corrections in 4 .
more systematic study for a wide range of the parameter space is presented in 5 .
finally , we discuss the implications of our results to the evolution of the galactic star clusters .
the first and second order theory of the compressive ( disk ) shocks and tidal ( bulge ) shocks has been developed by @xcite ( 1972 ) ; @xcite ( 1987 ) ; and @xcite ( 1994 ) and summarized in @xcite ( 1997 ) . below we review the important results for the disk shocking .
the theory of tidal shocking is discussed in detail in @xcite ( 1999 ) . for an alternative linear theory calculation see @xcite . in the disk shocking , the phase - averaged first and second order energy changes of stars with the initial energy per unit mass @xmath7 are @xmath8 where @xmath9 is the maximum vertical gravitational acceleration produced by the disk , @xmath10 is the vertical component of the cluster velocity with respect to the disk , and @xmath11 and @xmath12 are the rms position and velocity of stars of energy @xmath7 . here
@xmath13 is the position - velocity correlation factor , which takes values from @xmath14 to @xmath15 ( see @xcite 1999 ) .
the first order energy change , @xmath16 , causes reduction in the binding energy of the system and leads to evaporation of the marginally bound stars .
the second order change , @xmath17 , causes a much larger energy dispersion which allows additional stars to leave the cluster .
the two effects cooperate and lead to faster dissolution of the cluster .
the adiabatic corrections , @xmath18 and @xmath19 , account for conservation of the adiabatic invariants of stars for which the orbital period in the cluster is short compared to the effective duration of the shock , @xmath20 where @xmath21 is the characteristic scale - height of the disk .
the adiabatic corrections can be approximately described as functions of the only dimensionless parameter , @xmath22 here @xmath23 is the stellar orbital frequency of stars of energy @xmath7 at the rms position @xmath11 .
several approximations have been used to study the adiabatic corrections .
( 1 ) the _ impulse approximation _ is valid for the stars whose motion in the cluster is much slower than the shock , @xmath24 .
this condition applies in the outer regions of the cluster .
there both corrections become unity : @xmath25 ( 2 ) the _ harmonic approximation _ is valid for the stars which oscillate very quickly in a simple harmonic potential and which are not in resonance with the perturbing force .
building on results of @xcite ( 1987 ) , we get @xmath26 here we have modified the second correction from its version in @xcite ( 1997 ) to make it equal unity for @xmath27 .
the numeric coefficients are dictated by the asymptotic form for large values of @xmath28 .
( 3 ) the _ linear theory _ by @xcite includes the possible resonances and predicts a less steep power - law function . in the limit of large @xmath29 ,
the results can be approximated ( @xcite 1999 ) as : @xmath30 \(4 ) @xcite ( 1999 ) performed _
-body simulations _ of the tidal shock , allowing for the self - consistent oscillations of the cluster potential as it relaxes into a new virial equilibrium .
the simple fits to the results are @xmath31 for the `` fast shocks ''
whose durations are comparable to or smaller than the dynamical time at the half - mass radius : @xmath32 . for the `` slow shocks '' ,
the results agree with the predictions of the linear theory ( eq . [ [ eq : wcorr ] ] ) . in the case of bulge
shocking , the spitzer adiabatic corrections involve a combination of bessel functions , as described in @xcite ( 1997 ) .
the weinberg and -body adiabatic corrections are taken to be the same as for disk shocking .
since the -body simulations of @xcite ( 1999 ) include only disk shocks , further study is needed to determine accurately the adiabatic corrections for bulge shocking .
the first and second order energy changes are of comparable importance .
this becomes clear if we define the characteristic shock timescales : @xmath33 @xmath34 [ eq : definetimes ] here @xmath35 is the period of cluster s passage through the galactic disk .
the energy changes @xmath36 and @xmath37 are evaluated at the half - mass radius @xmath38 , and the characteristic energy @xmath39 is given by the virial theorem : @xmath40 ( e.g. , @xcite 1987 ) , where @xmath41 is the mass of the cluster . in the impulse approximation we obtain @xmath42 where @xmath43 is the rms angular velocity of stars at the half - mass radius .
the two timescales are simply related by @xmath44 , so that both processes contribute similarly to the destruction of the cluster .
we model the evolution of globular clusters using an orbit - averaged isotropic fokker - planck code descended from cohn ( 1979 , 1980 ) .
the code has been modified by @xcite and @xcite to include tidal boundary and the three - body binary heating .
stars beyond the tidal boundary do not escape instantaneously but follow instead a continuous distribution function @xmath45 , as described in @xcite .
this accounts for the balance of the internal and external forces at the tidal radius .
therefore , those stars only drift slowly away from the cluster .
the tidal field is assumed to be spherically symmetric , which is a weakness of the one - dimensional code ( for a discussion see @xcite 1995 ) .
we also assume that the cluster fills its roche lobe and require that the average density within the tidal radius remain constant throughout the cluster evolution .
the heating of stars which reverses core collapse is provided by three - body binaries .
they are included explicitly without following their actual formation and evolution , according to the prescription by @xcite .
@xcite argued that tidally - captured binaries are probably more dynamically important for massive clusters . whether tidal capture leads to a hard binary which contributes to stellar ejection after a close encounter or
whether it leads ( more often ) to a merger which causes mass loss following stellar evolution , the outcome is the same .
lee ( 1987 ) showed that the merger of stars would give a nearly identical dynamical effect on the cluster because the massive stars formed in the merger would evolve off rapidly .
both processes cause indirect heating by ejecting mass .
the situation becomes more complicated if we allow for the existence of primordial binaries and massive degenerate stars .
the existence of massive remnant stars , such as neutron stars , could cause three - body binaries to be more important than tidal capture binaries ( @xcite ) .
however , many aspects of the evolution after core collapse are independent of the actual energy source ( @xcite 1961 ; goodman 1993 ) .
we include the effects of tidal shocks by modifying the diffusion coefficients in the fokker - planck equation .
we assume that the first and second order energy changes are known as functions of the energy and position using the results from [ sec : theory ] and @xcite ( 1999 ) .
we now re - derive the fokker - planck equation in order to define the diffusion coefficients corresponding to the shocks .
let @xmath46 be the probability of scattering of a star of energy @xmath7 by the amount @xmath47 $ ] .
-body simulations ( @xcite 1999 ) show that the probability distribution is nearly gaussian : @xmath48 where both @xmath0 and @xmath1 are functions of energy @xmath7 .
let @xmath49 be the number of stars in the energy range @xmath50 $ ] .
the time evolution of @xmath51 , in the absence of two - body relaxation and binary heating , can be described by the following equation @xmath52 where the energy changes by the amount @xmath53 in the time interval @xmath54 . expanding this equation in a series over @xmath53 and @xmath54
, we obtain the usual form of the fokker - planck equation @xmath55 where @xmath56 and @xmath57 are the orbit - averaged diffusion coefficients of the first and second order .
they are defined by @xmath58 and @xmath59 where @xmath60 and @xmath61 are the average rates of change of the energy and its dispersion , respectively , per unit volume at a given location @xmath11 , and @xmath62 is the maximum radius allowed for a star of energy @xmath7 .
given equations ( [ eq : de_av ] ) and ( [ eq : de2_av ] ) for the energy changes and equation ( [ eq : psi ] ) for their distribution function , we define @xmath63 the bulge and disk tidal shocks are applied at the time step when the cluster is at its perigalacticon or crosses the galactic disk , respectively .
the diffusion coefficients are normalized to the current time step , @xmath54 , so as to produce the right amount of heating per event .
the shocks are repeated every orbital period of the cluster .
the time step was chosen such that both the cluster mass and the central density change by no more than 1% between time steps : @xmath64 .
\label{eq : dt}\ ] ] @xcite suggested that such a time step should be sufficient for an accurate calculation of the cluster evolution , both before and after core collapse .
we compare our results without tidal shocking with those of @xcite , who used the same time step for the potential re - computation but broke it down into 32 sub - steps for updating the distribution function ( the `` fokker - planck steps '' ) .
we found a measurable difference in the core collapse times only for very concentrated clusters , with the initial concentration @xmath65 .
for these clusters an even smaller time step may be required , although our results never differ by more than 30% ( cf .
figure [ fig : tcc_rel ] ) .
as described below , in order to incorporate correctly the effects of tidal shocks into the code , the shock diffusion coefficients should be applied separately from the relaxation terms , and the potential should be recomputed accordingly thereafter ( an alternative approach has been suggested by @xcite ) .
it seems awkward , though still possible , to include a set of fokker - planck steps for every potential time step in this scheme , as @xcite did for the isolated clusters .
we chose not to do so and therefore we had the potential updated every time the distribution function was changed .
since our time step ( eq . [ [ eq : dt ] ] ) gives very good accuracy for most of the computed cluster models , we did not change it specifically for high concentration clusters . the resulting error
should in any case be small .
the alternative implementation of tidal shocks in a fokker - planck code has been done by @xcite ( 1997a - c ) .
those authors employ the linear theory formalism of @xcite and calculate the change of the distribution function directly from the boltzmann equation .
that approach is more accurate but is more cumbersome .
since the linear theory expressions are supported by the restricted -body simulations ( weinberg 1994 ) , that method and our derived adiabatic corrections should give the same result .
the work of @xcite has focused more on the destruction time of globular clusters under various assumptions and on predicting the initial cluster distribution . in this paper , in addition to obtaining fitting formulae for the destruction time we present also the detailed evolution of individual clusters and compare it with previous work on isolated clusters .
@xcite ( 1999 ) report self - consistent -body simulations of a single disk shock of various strengths and durations .
the cluster is modeled as a low - concentration king - model ( @xmath66 ) with a realistic number of particles , @xmath67 .
the characteristic disk shocking strength is low and most of the stars remain bound to the cluster .
however , the energy distribution changes in response to the perturbation . the resulting distribution after the virialization phase is accurately fitted by equations ( [ eq : de_av ] ) , ( [ eq : de2_av ] ) , and ( [ eq : numcorr ] ) .
although the result depends on the initial structure of the cluster , the above fits still apply for the more concentrated cluster with @xmath68 ( see @xcite 1999 for more details ) .
the -body modeling of tidal shocks is intrinsically more accurate than the fokker - planck formulation .
the former takes into account the non - linear virialization following the shock and the anisotropic effects induced by disk shocking .
a direct comparison of the resulting energy changes between the two codes is difficult because of the algorithmic differences .
therefore , we have taken the simple approach to reproduce the -body results in our f - p code as closely as possible .
the general procedure for advancing the distribution function ( df ) in time consists of two parts ( cohn 1979 ) .
the first part introduces the first and second order changes of the df as a solution of the fokker - planck equation .
the second part updates the cluster potential by solving poisson s equation with the density following from the new df . during the second step ,
the df is kept constant as a function of the stars adiabatic invariants and therefore it changes as a function of energy .
this procedure is no longer correct when treating tidal shocks which last of the order of the dynamical time of the cluster .
this timescale is short enough that adiabatic invariants may themselves change .
we find that in order to reproduce the -body results , we need to modify the above procedure for tidal shocks . during the time step when the cluster passes through the galactic disk or close to the galactic center ,
we first apply the shock diffusion coefficients and update the df .
then we fix the df as the function of energy and obtain the new potential by solving poisson s equation .
specifically , since the potential is changed and the energy grid is renormalized , we require that in each energy bin the new energy distribution , @xmath69 , scale with the new energy grid , @xmath70 , as @xmath71 where @xmath51 is the energy distribution before the potential recalculation .
we have checked that the above procedure reproduces well the -body results for a range of shock amplitudes .
figure [ fig : ne ] shows the change in the energy distribution after a single shock in a test cluster .
the agreement between the fokker - planck and -body results is reasonably good .
also , figure [ fig : ne ] shows that the previous procedure for updating the df for the fixed adiabatic invariants departs much more strongly from the -body results than does our new prescription .
another effect demonstrated by the -body simulations is a slight decrease in the shock - induced energy dispersion subsequent to and caused by the potential fluctuations following the shock .
this paradoxical `` refrigeration effect '' persistently survived all our tests and seems to be real ( see @xcite 1999 ) .
however , the magnitude of the dispersion decrease is relatively small , which led us to ignore it in the present calculation . once the nature of this effect is understood better
, it should also be included in the models of globular clusters .
recent paper by @xcite ( 1998 ) compares the multi - mass two - dimensional fokker - planck ( f - p ) models of @xcite ( 1995 ) with the -body simulations using a special purpose computer , _
grape-4_. the models include stellar evolution , two - body relaxation , and the tidal truncation for a low concentration ( @xmath133 ) cluster model from @xcite .
the above authors argue that the escape of stars through the tidal boundary in an isotropic f - p code , where the distribution function depends only on the energy , is faster than in the -body models .
to remedy this disagreement , they suggest a new escape criterion based both on the energy and angular momentum of stars .
this `` apocenter '' criterion can only be implemented in the anisotropic f - p code of @xcite .
the old isotropic and the new anisotropic escape criteria lead to significantly different results only when the mass loss rate per relaxation time is high .
this especially affects the early stages of the cluster evolution , when stellar winds from young massive stars drive mass loss on a relatively short timescale .
immediate removal of that mass in the @xcite models lead to the disagreement with the -body simulations ( see @xcite 1998 ) .
since our models do not include stellar evolution , we expect less disagreement .
we run a _ case 0 _ model of the cluster using the parameters from @xcite .
we use 14 mass components from @xmath134 to @xmath135 , distributed according to the specified mass function , @xmath136 . without stellar evolution the time to destruction in our model should be longer than in the -body and the anisotropic f - p models which give @xmath137 .
indeed , we find that our model evolves slower , with @xmath138 .
this indicates that our results represent an upper limit on the destruction times of globular clusters .
we consider now the detailed models of the globular cluster evolution . for the most part , we study single - mass models and then briefly investigate the effect of a spectrum of stellar masses . with the new data on proper motions from _ hipparcos _ , it is now possible to reconstruct full three - dimensional velocities for some of the galactic globular clusters ( @xcite ) and to build realistic models of their evolution .
we use the inferred orbital parameters for ngc 6254 and calculate its orbits in an analytic potential of the galaxy from @xcite .
the orbits are of the rosette type ( figure [ fig : orbits ] ) with a small eccentricity , @xmath72 .
the cluster moves around the galaxy with a period of about @xmath73 yr , during which it passes next to the galactic center every @xmath74 yr and crosses the galactic disk every @xmath75 yr . the period ratio differs from the expected factor of two due to the slow precession of the orbit .
from the computed orbits of we infer that the amplitude of tidal shocks varies approximately as a gaussian function of time .
this satisfies the conditions under which the -body adiabatic corrections were calculated ( see @xcite 1999 ) .
the observed structure of is well characterized by a king model ( @xcite ) with the concentration @xmath76 , or the structural parameter @xmath77 .
the mass of the cluster is @xmath78 , assuming a constant mass - to - light ratio @xmath79 in solar units .
first , we consider in detail a model composed of stars of the same mass , @xmath80
. the important parameters of the model are summarized in table [ tab : param ] .
we implement both disk and bulge shocks in our integration .
the models differ in the way the shocks are included : no tidal shocks , steady tidal field and relaxation only ( _ case 0 _ ) ; tidal shocks with the first order @xmath0 term only ( _ case 1 _ ) ; tidal shocks with the second order @xmath1 term only ( _ case 2 _ ) ; and finally tidal shocks with both @xmath0 and @xmath1 terms ( _ case 3 _ ) .
an important observable of the cluster evolution is the mass - loss due to tidal shocks and due to the direct escape of high - velocity stars .
figure [ fig : m ] shows the run of the cluster mass with time . as is well - known , the mass of a tidally limited cluster goes to zero in an approximately linear fashion .
a convenient choice is to express time @xmath81 in the units of the initial half - mass relaxation time @xmath82 ( eq . [ [ eq : trh ] ] ) .
two - body relaxation leads to destruction of the cluster in about @xmath83 .
note that the numerical problems cause the cluster to lose its stability and dissolve into the sea of background stars before its mass reaches exactly zero .
usually the code fails to re - calculate the cluster potential when the mass falls to about 1%
4% of the initial mass .
however , we can extrapolate through the last few time steps in order to estimate the hypothetical time when the cluster mass vanishes .
the disruption time obtained this way is an overestimate of the real disruption time , but all other estimates would suffer from a personal choice .
thus , in the rest of the paper we use the extrapolation of our calculations as the destruction time @xmath84 . the destruction time is significantly reduced when we include gravitational shocks .
the @xmath0 term alone changes @xmath84 from @xmath83 to @xmath85 and the shock - induced relaxation brings it to about @xmath86 . to access the importance of the two shock terms separately , we turn off the energy shift , @xmath0 , for the _ case 2 _ run .
the second order term leads to an enhanced mass loss relative to the relaxation case for most of the evolution .
this effect is not as strong as the first order effect due to the limiting adiabatic corrections , which we investigate in more detail later . for comparison
, we also show the mass loss that would have been due to the shock - induced relaxation if there were no adiabatic corrections ( the impulse approximation ) .
the latter effect would reduce the destruction time of the cluster by about a half .
an important feature of tidal shocks is the enhanced mass loss at the early stages of the evolution , prior to core collapse . in the pure relaxation case ,
the mass loss is slow until core collapse and linear in time afterwards .
the first theoretical calculations of the probability of stars escape from the cluster through two - body relaxation ( @xcite 1938 ; @xcite 1940 ) predicted the following dimensionless rate @xmath87 where @xmath88 is the half - mass relaxation time ( @xcite ) : @xmath89 here @xmath41 is the current cluster mass , @xmath38 is the half - mass radius , @xmath90 is the average stellar mass , and @xmath91 is the coulomb logarithm , @xmath5 being the number of stars in the cluster .
later calculations of the tidally - truncated cluster by @xcite ( 1961 ) gave a larger value , @xmath92 .
@xcite used monte - carlo simulations for several models of globular clusters and found @xmath93 for @xmath94 , and @xmath95 for @xmath96 , where @xmath97 is the tidal radius of the cluster .
based upon those simulations , @xcite proposed the following fitting formula for the time to disruption : @xmath98 where @xmath82 is the initial relaxation time . for , the disruption time predicted by aho would be about @xmath99 , whereas we see that the cluster dissolves in @xmath83 even without tidal shocks .
in general , we find that equation ( [ eq : tevaho ] ) overestimates the actual disruption time by a factor of several .
we compare the previous results for the escape probability @xmath100 with our fokker - planck calculations in figure [ fig : dm ] . in the relaxation model ( _ case 0 _ ) , the value of @xmath100 rises almost monotonically through core collapse and until destruction of the cluster , reaching and exceeding hnon s estimate only in the late stages of the evolution . on the contrary , in the shock - dominated models
the escape probability is very high in the beginning when tidal shocks efficiently remove stars from the outer parts of the cluster .
the mass loss then slows down and conforms to the relaxation rate soon before core collapse .
then again , @xmath100 rises with time to reach hnon s self - similar value .
this later increase caused primarily by the declining mass of the cluster , since the mass loss rate is almost constant in the post core - collapse cluster .
the cluster structure can be described in terms of the characteristic radii , such as the _ core radius _ , reflecting compactness of the cluster ; the _ tidal radius _ , confining all stars bound to the cluster ; and the _ half - mass radius _ , relating to the global properties of the cluster .
figure [ fig : r ] shows the evolution of these parameters normalized to the initial core radius .
the core radii for all models drop to extremely small values at the point of core collapse .
the core - collapse time @xmath101 is about @xmath102 for the relaxation case , in agreement with @xcite and @xcite .
tidal shocks speed up core collapse significantly : @xmath103 in _ case 1 _ and it is still smaller when we include the shock - induced relaxation term .
the half - mass radius changes only by a factor of two over the whole cluster lifetime , which indicates that core collapse affects only a small fraction of stars .
the tidal radius decreases slowly , reflecting the gradual mass loss from the cluster .
figure [ fig : rho ] illustrates core collapse more clearly .
the central density of the cluster , @xmath104 , rises by nine orders of magnitude before the collapse is halted by a production of three - body binaries in the very dense core of the cluster .
the energy released from the binaries eventually reverses the collapse .
the cluster consequently expands , though the central density still remains very high ( about four orders of magnitude higher than initially ) .
the actual value of @xmath104 at the core bounce and post - collapse phase depends on the specific heating mechanism .
a stronger heating rate gives a lower central density during the post - collapse phase . the cluster concentration , @xmath105 , is closely related to the central density . since the tidal radius @xmath97 decreases slowly with time as @xmath106 , the concentration varies roughly as @xmath107 .
the core collapse time provides a very useful yardstick for cluster evolution . even in the presence of tidal shocks ,
the evolution of the models is similar when the units of time are scaled to the core collapse time .
figure [ fig : rho_cc ] illustrates this point .
the central density in models _ 03 _ varies almost identically with @xmath108 .
therefore , we can use this scaling to parameterize cluster evolution .
the destruction time is an almost constant multiple of the core collapse time : @xmath109 .
the post - collapse evolution of globular clusters depends on the concentration and the remaining mass .
the self - similar solution by @xcite ( 1961 ) for tidally - limited clusters can be expressed ( goodman 1993 ) as follows : @xmath110 where @xmath111 is the time to disruption at time @xmath81 .
figure [ fig : tremain ] shows that the self - similar limit is approached at late stages of the cluster evolution , but for the most part equation ( [ eq : henon ] ) underestimates the time to destruction by about 50% . an important constituent of our models is adiabatic corrections for the tidal shock effects ( eqs .
[ eq : spcorr][eq : numcorr ] ) .
various corrections predict different impact of shocks in the middle parts of clusters and lead to different evolutionary paths .
the amplitude of the corrections depend on the ratio of the perturbation timescale to the dynamical time of stars in the cluster . for , the characteristic time for disk shocking ( eq . [ [ eq : tau ] ] ) is @xmath112 yr , and for bulge shocking ( cf .
@xcite 1999 ) is @xmath113 yr . the half - mass dynamical time of the cluster is initially much shorter , @xmath114 yr , leading to strong suppression of the effects of tidal shocks .
figure [ fig : mad ] compares the spitzer , weinberg , and -body adiabatic corrections for the _ case 3 _ models of . when the shocks are `` slow '' ( @xmath115 ) , as indicated by the true values of @xmath116 and @xmath117 , both _
s1 _ and _ n1 _ models have tidal shock effects suppressed more strongly than does the _ w1 _ model with the less steep adiabatic corrections . for the `` slow '' shocks , the _ w1 _ model gives the correct description of the cluster evolution .
when we decrease the shock timescales by a factor of five , we enter the regime of `` fast shocks '' and the _
model is correct . for @xmath118
, we can estimate the importance of the adiabatic corrections by the taylor expansion of equations ( [ eq : spcorr][eq : numcorr ] ) .
n2 _ model shows the slowest evolution because @xmath119 , whereas for the _ w2 _ model , @xmath120 . even though the first order correction for the _
s2 _ model , @xmath121 , is smaller than @xmath122 , the second order correction is larger , @xmath123 .
this allows the _ s2 _ model to evolve faster than the other two .
finally , when we reduce the shock timescales by a factor of hundred , the impulse approximation applies everywhere in the cluster and all three models are essentially identical .
the evolution in this case proceeds much faster : the cluster is destroyed in half the time of the original model _
thus , adiabatic corrections are critical for calculating correct globular cluster models . for comparison
, we consider another model of , with lower concentration @xmath66 , or @xmath124 .
this is the model used for -body simulations of @xcite ( 1999 ) .
all other parameters are the same as in the previous model .
we keep the observed tidal radius in parsecs , so the core radius and the relaxation time scale in physical units accordingly ( see table [ tab : param ] ) .
therefore , both models have the same average density , @xmath125 .
figure [ fig : m1 ] shows the mass evolution of the low concentration model .
the cluster evaporates much faster than the high concentration model , in units of the initial relaxation time @xmath82 .
clearly , tidal shocks have much stronger effect on less concentrated clusters but the total speed - up in time to destruction , due to shocks , is again about 25% .
notice that the destruction time in years is about the same for the low- and high - concentration models .
figure [ fig : rho1 ] shows that the evolution of the central density is qualitatively similar to the previous model .
however , the effect of the tidal shock relaxation is stronger for the low - concentration model . in a single mass model
, the energy is transferred between the cluster core and the envelope , which requires a conduction mechanism acting on the half - mass relaxation timescale . in a multi - mass model
, the energy exchange occurs also between stars of different mass . a typical timescale for the energy exchange between the two mass species @xmath126 and @xmath127 ( where @xmath128 ) is @xmath129 .
since this is much shorter than the core collapse time for single mass clusters , the overall evolution of multi - mass clusters is expected to be faster .
@xcite ( 1995 ) , for example , considered the evaporation of a multi - mass cluster in a steady tidal field and found that the mass loss rate can more than double relative to the single - component case .
@xcite investigated the effects of stellar evolution on the cluster dynamics . in the future , it will be important to generalize the present calculations by including a realistic mass function and stellar evolution , along with the presently implemented two - body relaxation and tidal shocks . in this paper ,
we present only an illustrative model allowing for the mass spectrum .
we take the previously described model for and include seven mass components , ranging from @xmath130 to @xmath131 .
the number of stars in each component @xmath132 .
thus , the upper mass limit corresponds to the turn - off of the main sequence at 10 billion years ( we also used it in the single - mass models ) , and the lower limit is a reasonable cutoff of the initial mass function . the relaxation time is defined by equation ( [ eq : trh ] ) , where we substitute for @xmath90 the mean stellar mass ( see @xcite 1995 ) .
figure [ fig : m_multi ] shows the mass loss for the multi - mass model .
notice that although the evolution proceeds faster in units of the initial half - mass relaxation time , this timescale expressed in years is longer than that for the single - component model ( table [ tab : param ] ) . as a result ,
the destruction time is roughly 20 gyr in both cases .
the contribution of tidal shocks in the multi - mass model is greater , reducing the time to destruction by about 37% .
the evolution of the isolated single - mass king models ( in dimensionless units ) is determined by the only parameter , the concentration @xmath139 .
inclusion of tidal shocks introduces another independent parameter , for example , the ratio of the half - mass relaxation time to the characteristic shocking time , @xmath140 the parameter @xmath141 is a combination of two variables describing the structure of clusters : the number of stars , @xmath5 , and the characteristic density , @xmath142 .
when the amplitude of the tidal force on the cluster is fixed , the shocking time ( eq . [ [ eq : tsh ] ] ) scales as @xmath143 , and the relaxation time ( eq . [ [ eq : trh ] ] )
scales as @xmath144 , where @xmath145 .
then @xmath146 while the concentration @xmath139 describes the effects of two - body relaxation , the parameter @xmath141 shows the relative importance of tidal shocks . in this section
, we explore the evolution of globular clusters for a wide range of initial parameters @xmath139 and @xmath141 .
all models in this section include fully the relaxation and tidal shock effects ( i.e. , _ case 3 _ ) with the adiabatic corrections for `` slow shocks '' , equation ( [ eq : wcorr ] ) .
we arrange a grid of initial conditions , covering a number of cluster concentrations , from @xmath147 to @xmath148 with the step 0.2 , and a range of shock parameters , @xmath141 .
we compute 15 models per concentration family , equally spaced on the logarithmic scale from @xmath149 to @xmath150 .
all other parameters correspond to the model of described in the previous section . the observed tidal radius of is fixed and the core radius scales accordingly with the concentration .
thus all models have initially the same average density , @xmath151 .
figure [ fig : cbeta ] shows how different regimes of cluster evolution map on the plane of parameters @xmath139 and @xmath141 . in the lower left region of the plot
tidal shocks are weak and evolution is dominated by two - body relaxation ; in the upper left region core collapse proceeds on a relatively short timescale without affecting most of the cluster ; and in the lower right region tidal shocks are strong and cause substantial mass loss before core collapse .
the upper right region is never reached in reality because of extremely strong tidal shocks leading to fast destruction of the clusters .
qualitative evolution in the left region of the plane is characteristic of a tidally - truncated model .
as clusters start to collapse , the relaxation time becomes increasingly shorter and the tidal shock time becomes increasingly longer because the mean density @xmath152 rises ( figure [ fig : rhomean ] ) .
the mass loss is weak at this stage and the number of stars , @xmath5 , stays relatively constant . as a result ,
the parameter @xmath141 decreases ( and moves to the left of the diagram ) until the very advanced stages of core collapse . at that point the value of @xmath141 freezes and the evolutionary tracks are strictly vertical ( the upper part of the diagram ) . after core collapse , the cluster re - expands and its density falls slightly , moving it to the right of the @xmath153 diagram . at a later time , so little mass
is left that two - body relaxation speeds up again and drives clusters to complete dissolution .
figure [ fig : beta ] illustrates the behavior of the parameter @xmath141 with time . on the low end , all lines scale similarly , following the evolution of the mean density . for the large initial values of @xmath141 ,
the early mass loss due to shocks is very important .
this leads us to investigate the right part of the @xmath153 diagram .
the evolution of clusters with strong tidal shocks differs dramatically from the previous case .
an early mass loss caused by the shocks changes the structure of the clusters , removing stars from the outer parts and adding energy dispersion in the core .
this increases the central and the mean density of the clusters .
both core collapse and final destruction proceed much faster .
these effects of tidal shocks leave noticeable ripples in the clusters tracks on the right part of the diagram ( figure [ fig : cbeta ] ) .
tidal shocking is rapidly self - limiting . clusters with large values of @xmath141 quickly lose mass and evolve to @xmath154 .
figure [ fig : beta ] demonstrates that just a first few shocks lower the value of @xmath141 by several orders of magnitude .
every subsequent shock causes the see - saw variations of the shock parameter , with an increasing amplitude towards the late stage of evolution when fewer stars are left .
a few clusters in the right part of the @xmath153 diagram cross their evolutionary tracks on the way to core collapse . at the time of crossing , at least one of the clusters has already suffered a severe mass loss caused by the shocks and has a structure significantly deviating from king models .
the density profiles of the two clusters are similar within the half - mass radii but depart from each other closer to the tidal radius .
those clusters with higher initial concentration and stronger shocks are more severely truncated than the clusters with smaller concentration and weaker shocks .
also , tidal shocks add the velocity dispersion in the core and , therefore , raise the central density so that the ratio @xmath155 , as measured by the concentration @xmath139 , is the same for both clusters . at the crossing point the two parameters , @xmath139 and @xmath141 , no longer uniquely specify cluster s future path .
a third parameter comes into play as we investigate the thermodynamics of clusters . from a thermodynamic point of view
, we can consider stars as particles of gas .
a critical factor determining evolution is the heat flow between the core and the halo of the cluster .
the heat in this case is the velocity dispersion of stars , @xmath156 , whose flux is derived in @xcite ( 1987 , p. 68 ) .
the amount of heat transported through the cluster per unit time , or the `` luminosity '' , is @xmath157 .
the heat conduction facilitates core collapse and proceeds on a relaxation timescale .
let us construct a new dimensionless ratio of the heat transported through the half - mass radius in the relaxation time @xmath88 , to the total reservoir of heat , @xmath158 .
neglecting constant factors of order unity , we define the following parameter @xmath159 which is essentially a logarithmic gradient of the velocity dispersion . for an isothermal sphere , @xmath160 . in general , for clusters on the way to core collapse , this parameter is positive as the heat is transferred from the kinematically hot core to the cold halo and grows as core collapse accelerates .
in contrast , tidal shocks try to reverse the heat flow by heating preferentially the outer parts of the cluster . even though the instantaneous value of @xmath161 is determined only by the density structure , its derivative , @xmath162 ,
is a signature of the relaxation or the tidal shock driven evolution .
the clusters crossing paths on the @xmath163 diagram have similar values of @xmath161 , but the derivatives have opposite sign . for the clusters moving almost vertically on the diagram @xmath164 , whereas for the clusters which cross the diagram horizontally @xmath165 .
therefore , the evolution of the former is dominated by two - body relaxation and of the latter by tidal shocks .
knowing the value of @xmath162 for the galactic globular clusters , we could in principle differentiate the two evolutionary paths .
unfortunately , it is hard to establish the sign of @xmath162 observationally .
another signature of the tidal shock
dominated clusters is a steep density profile and a low number density of stars at the tidal radius .
strong shocks sweep away a large number of stars at once and leave the cluster filling smaller volume in space than the roche lobe imposed by the external tidal field . clusters with a sharp density contrast and a high velocity dispersion of the halo stars might be experiencing strong shocking . most of such clusters are expected to come close to the galactic center but are not necessarily present there now . the mass loss from clusters is strongly enhanced by tidal shocks .
an illustration of this is the fraction of the initial mass remaining at the time of core collapse , @xmath166 .
figure [ fig : mcc ] shows that the larger @xmath141 , the smaller the remaining mass . note , that many models of various initial concentration converge at @xmath167 , giving @xmath168 .
the core collapse time is a strong function of both initial parameters @xmath139 and @xmath141 .
figure [ fig : tcc_rel ] shows @xmath101 for the models without tidal shocks . in units of the initial half - mass relaxation time , the core collapse time can be fitted as follows : @xmath169 where the coefficients @xmath170 are given in table [ tab : fitc ] .
our results are in good agreement with @xcite , except for the clusters with very high concentration . even for them ,
the maximum difference in @xmath101 is less than 30% .
figure [ fig : tcc_sh ] shows the core collapse time for the models including tidal shocks .
the shock induced relaxation speeds up core collapse by adding velocity dispersion in the core and by removing stars in the outer parts of the clusters , thus reducing the relaxation time .
we fit the results assuming , as an _ ansatz _ , the following functional form : @xmath171 the coefficients @xmath172 for models of various concentration @xmath139 are given in table [ tab : fittccbeta ] .
the fit is generally accurate to 20% , with larger errors for models with very high initial concentration .
the time to destruction , @xmath84 , is also strongly affected by tidal shocks . as we have emphasized in the previous section , evolution of clusters with different parameters scales similarly when time is normalized to the core collapse time .
accordingly , we provide the fitting formulae for the scaled destruction time , @xmath173 . for the models without tidal shocks ( figure [ fig : td_cc ] ) , we use the same functional form as equation ( [ eq : tcc_rel ] ) and fit @xmath84 both in units of the initial relaxation time and in units of the core collapse time : @xmath174 where the coefficients @xmath170 are again given in table [ tab :
fitc ] . for the models including tidal shocks ( figure [ fig : td_sh ] ) ,
the scaled destruction time varies very little for the low concentration clusters and increasingly more for the high concentration ones .
interestingly enough , regardless of the initial structure , models with strong shocks converge to an almost constant value @xmath175 .
it follows that the core collapse time is a midpoint in cluster s lifetime .
since the value of @xmath173 is almost constant for @xmath176 and for @xmath177 , we choose the following functional form to fit the results : @xmath178 where the coefficients @xmath170 are given in table [ tab : fittdbeta ] .
this expression allows for a variation in the slope of the fit for large values of @xmath141 , but the coefficient @xmath179 is always close to unity , as expected . in the case of very strong shocks , both the core collapse and the destruction times scale approximately as @xmath180 ( cf . the coefficient @xmath181 in table [ tab : fittccbeta ] ) .
this asymptotic behavior can be understood as follows .
the shock parameter @xmath141 is proportional to the mean energy change due to shocks ( eq . [ [ eq : definetimes ] ] ) , which in turn is proportional to the amplitude of the tidal force squared , @xmath182 .
therefore , the approximate scaling of @xmath101 and @xmath84 indicate that these timescales vary inversely proportional to @xmath183 .
our fokker - planck models provide the most comprehensive calculations of the globular cluster evolution including gravitational tidal shocks .
we find that tidal shocks significantly alter the evolution , in agreement with the earlier studies by @xcite and more recent ones by @xcite ( 1997 ) and @xcite ( 1997a - c ) .
globular clusters appear to be rather fragile with regard to the effects of tidal shocking .
tidal shocks accelerate dynamical evolution and evaporation of clusters by effectively stripping stars in the outer regions and adding energy dispersion in the core .
these effects are self - limiting : as clusters lose mass and become more compact , the relative importance of tidal shocks rapidly diminishes in favor of two - body relaxation . in the case of the galactic globular cluster ,
tidal shocks decrease the destruction time from 24 gyr to 18 gyr .
we examine several types of adiabatic corrections : the harmonic approximation ( of l. spitzer ) , the linear perturbation theory ( of m. weinberg ) , and the results of the -body simulations .
the differences between the various types of adiabatic corrections depend on the regime of `` slow shocks '' or `` fast shocks '' , but overall the corrections are critical for accurate calculation of the evolution . without adiabatic corrections ,
the destruction time of is twice as short .
the evolution of models with a single mass component can be characterized by the king - model concentration parameter , @xmath139 , and the shock parameter , @xmath141 ( eq . [ [ eq : betadef ] ] ) .
the effects of tidal shocks are important when @xmath184 , depending on the concentration .
tidal shocks increase both the mean and the central densities , relative to the relaxation case .
core collapse is accelerated by about 30% for @xmath185 and by a factor of 3 for @xmath186 . in the limit of strong shocks ,
the core collapse time scales approximately as @xmath187 .
the time to destruction is also dramatically reduced , following the same scaling law .
overall , we find that evolution of models proceeds very similarly when time is expressed in units of the core collapse time . in the limit of strong shocks , the destruction time @xmath188 , independently on the initial concentration .
core collapse provides a useful midpoint of cluster evolution .
we also consider an illustrative model with a range of stellar masses . in units of the initial relaxation time
, the evolution proceeds much faster than in the single - mass case , but for both models the time to destruction is comparable when expressed in years .
thus , the models with the same average density evolve in approximately the same time .
a systematic study of multi - mass models is beyond the scope of the present study and requires a knowledge of the initial mass function in star clusters .
the results of our calculations have important implications for the evolution of globular clusters . for the sample of the galactic globular clusters from @xcite ( 1997 ) the typical values of the shock parameter are @xmath189 .
this indicates that the effects of tidal shocks and of two - body relaxation are comparable at present .
since the shock effects quickly saturate , tidal shocks must have been more important in the past .
not only will the current population of clusters be significantly depleted in the future ( @xcite 1997 ) , but also a large fraction of the initial population might have been already destroyed in the galaxy ( @xcite 1997c ) .
present calculations can be improved in several ways . a reasonable mass function and effects of stellar evolution
should be included in order to accurately describe the early phase of cluster evolution .
also , in our treatment of tidal shocks we have neglected small ( but real ) negative relaxation due to cluster oscillations .
the final weakness of our models is the assumption of the isotropic velocity distribution .
radial anisotropy would develop in the outer parts in the absence of tidal shocks . since the stars on radial orbits will be ejected more easily than those on circular orbits , tidal shocks should effectively isotropize orbits even in the outer parts .
@xcite ( 1995,1997 ) and @xcite have developed an accurate method for integrating the anisotropic fokker - planck equation .
we plan to include the effect of tidal shocks in the anisotropic code in order to clarify these points .
we would like to thank kathryn johnston for useful discussions , and the anonymous referee and the scientific editor steven shore for detailed comments .
this work was supported in part by nsf grant ast 94 - 24416 and by korea science and engineering foundation grant no . 95 - 0702 - 01 - 01 - 3 .
lccccc & 1.4 & 0.76 & 96 & & & & 12.9 & 32 & & & & 10.0 & 26 & & & & 11.2 & 32 & & & & 9.7 & 24 & 0.84 & 1.80 & 7.3 & & & & 12.2 & 16.5 & & & & 8.0 & 13.5 & & & & 8.5 & 14.5 & & & & 7.1 & 12 & 1.4 & 2.77 & 15 & & & & 1.76 & 13.5 & & & & 1.59 & 9.5 & & & & 1.60 & 13 & & & & 1.57 & 8.5 lcccc 0.8 & 1.5029 & 0.5539 & 0.0185 & 0.4558 1.0 & 0.1174 & 1.4184 & 1.6004 & 0.4491 1.2 & 0.5124 & 0.8597 & 1.5189 & 0.3840 1.4 & 3.3268 & 0.4680 & -1.0819 & 0.5281 1.6 & 2.9924 & 0.4613 & -1.0366 & 0.5726 1.8 & 1.1447 & 0.3691 & 0.1791 & 0.3614 2.0 & 1.4912 & 0.2996 & -0.7153 & 0.1606 2.2 & 0.8931 & 0.6518 & -0.6721 & 0.2626 2.4 & 0.6413 & 0.5716 & -0.9622 & 0.2167 2.6 & 0.9452 & 0.5916 & -1.4586 & 0.2065 lccc 0.8 & 2.1942 & 1.0376 & 2.3645 1.0 & 0.1315 & 1.2982 & 0.2646 1.2 & 1.6051 & 1.1528 & 2.6816 1.4 & 45.420 & 1.0356 & 74.731 1.6 & 135.52 & 0.9993 & 300.84 1.8 & 188.62 & 0.9932 & 587.29 2.0 & 159.24 & 0.9675 & 788.60 2.2 & 83.444 & 0.9608 & 667.98 2.4 & 40.040 & 1.0309 & 520.84 2.6 & 42.112 & 1.1056 & 862.75 | we present new fokker - planck models of the evolution of globular clusters , including gravitational tidal shocks .
we extend our calculations beyond the core collapse by adopting three - body binary heating .
effects of the shocks are included by adding the tidal shock diffusion coefficients to the ordinary fokker - planck equation : the first order heating term , @xmath0 , and the second order energy dispersion term , @xmath1 . as an example
, we investigate the evolution of models for the globular cluster ngc 6254 . using the _ hipparcos _ proper motions , we are now able to construct orbits of this cluster in the galaxy .
tidal shocks accelerate significantly both core collapse and the evaporation of the cluster and shorten the destruction time from 24 gyr to 18 gyr .
we examine various types of adiabatic corrections and find that they are critical for accurate calculation of the evolution . without adiabatic corrections ,
the destruction time of the cluster is twice as short .
we examine cluster evolution for a wide range of the concentration and tidal shock parameters , and determine the region of the parameter space where tidal shocks dominate the evolution .
we present fitting formulae for the core collapse time and the destruction time , covering all reasonable initial conditions . in the limit of strong shocks
, the typical value of the core collapse time decreases from @xmath2 to @xmath3 or less , while the destruction time is just twice that number .
the effects of tidal shocks are rapidly self - limiting : as clusters lose mass and become more compact , the importance of the shocks diminishes .
this implies that tidal shocks were more important in the past . |
the electromagnetic properties of spin-1 particles ( @xmath3 ) can help us to understand the symmetry structure and interactions of the fundamental particles .
for example , considering @xmath3 as a stable state , the electromagnetic properties of the @xmath0 gauge boson are predicted by the symmetry structure of the standard model , while for the @xmath1 meson several predictions exist , based on effective models of the strong interaction binding the quarks @xcite , and from lattice - qcd @xcite .
however , they are not stable states . therefore , in order to draw definite conclusions , a complete study of the _ additional _ effects due to their instability is mandatory . the proper theoretical description of such states requires to incorporate their unstable features ( parameterized by their finite decay width , @xmath4 ) in an electromagnetic gauge invariant way . to do so ,
several schemes have been developed , we can mention , for example , the so called fermion loop scheme @xcite and the boson loop scheme @xcite ( suitable for the @xmath0 and @xmath1 bosons respectively ) , which consider that @xmath4 is naturally included in the calculations by taking into account the absortive contributions in the electromagnetic vertex and in the propagator . under these schemes ,
the electromagnetic vertex @xmath5 is modified respect to the tree level form in a non - trivial way .
this implies that the electromagnetic structure itself suffers modifications . in the present work we analyze the most general results for the electromagnetic vertices obtained in the loop schemes , which include the mass of the particles in the loops , and extract the expressions for the modified form factors .
in order to exhibit the size of the corrections due to the unstable nature of the particles , we compare our results for the magnetic dipole moment ( mdm ) and electric quadrupole moments of the @xmath0 , @xmath6 and @xmath2 mesons with others computed in the literature for contributions from different nature .
the gauge invariance requirement , along with the modification in the propagator inherent to the schemes , allows to identify the proper complex renormalization factor of the vector field , which keep the electric charge free of radiative correction contributions .
the schemes developed in @xcite for the introduction of the finite width effects , while keeping electromagnetic gauge invariance , are based in two main observations : in quantum field theory the width is naturally included in the imaginary part of the self - energy of the particles and , the ward identity is respected at all orders in perturbation theory .
these facts are exploited in those schemes by including the resummation of the fermion / boson loops in the propagator and the corrections in the electromagnetic vertices .
then , the imaginary part of the fermion / boson loops introduces the tree level width in the gauge boson propagator and , the gauge invariance is not violated since the fermion / boson loops obey the ward identity order by order . at tree level
, the propagator for a vector boson of mass @xmath7 can be set as @xmath8 where @xmath9 and @xmath10 , are the transversal and longitudinal projectors , respectively .
the vertex for the process @xmath11 is defined from the electromagnetic current @xmath12 at tree level , the vertex can be set as that given by the standard model for the @xmath0 boson : @xmath13 these expressions satisfy the ward identity @xmath14 ^{-1 } - \left [ \imath d_0^{\nu \lambda } ( q_2 ) \right ] ^{-1 } ; \label{wardide}\ ] ] upon the inclusion of the finite width of the boson , by considering the loop contributions , the propagator is modified in a generic form as : @xmath15 where @xmath16 and @xmath17 , are the transverse and longitudinal part of the absortive contribution of the self - energy induced by the particles in the loop . similarly , the vertex becomes @xmath18 where @xmath19 contains the loop corrections .
the ward identity relates the loop contributions by requiring to satisfy @xmath20 for a boson like the @xmath0 , the scheme consider that such loops are produced by fermions , while for vector mesons , like the @xmath1 , bosons are the natural particles in the loop . in general ,
the cp conserving electromagnetic vertex can be decomposed into the following lorentz structure @xmath21 where the electromagnetic form factors can be identified as : @xmath22 is the electric charge form factor ( in @xmath23 units ) , @xmath24 is the magnetic dipole moment form factor ( in @xmath25 units ) and the electric quadrupole form factor is @xmath26 ( in @xmath27 units ) . the parameters @xmath28 and @xmath29 are of common use in the literature to refer to the electromagnetic multipoles @xcite .
the static electromagnetic properties of a particle are defined for the case when the particle is on - shell and in the limit of @xmath30 . at tree level ,
for example , the standard model predicts for the @xmath0 to have @xmath31 , @xmath32 and @xmath33 ( @xmath34 and @xmath35 ) , corresponding to @xmath36 , @xmath37 and @xmath38 .
deviations from these values are generically called anomalous and are produced by the inclusion of higher order contributions @xcite . in the present case such contributions are exclusively those required to maintain the electromagnetic gauge invariance , upon the introduction of the finite decay width .
let us identify the modification to the @xmath0 boson form factors introduced by the correction to the electromagnetic vertex .
for that purpouse we consider the explicit expression of the vertex obtained in ref .
@xcite , where the mass of the emitting particles in the loop ( @xmath39 ) and its weak partner ( @xmath40 ) have been considered .
the lorentz structure , transversality and on - shell condition of the boson along with the proper limit for @xmath41 leads to the following expressions for the form factors , defined in equation ( [ vercompletomul ] ) : * electric charge @xmath42 where @xmath43 , @xmath44 and @xmath45 is the electric charge of the radiating particle in the loop , @xmath46 is the partial decay width for the modes corresponding to the particles in the loop , @xmath47 denotes the coupling and @xmath48 . a sum over all the allowed flavors and color degeneracies
is explicitly included .
since the schemes consider the particles in the loop to be on - shell , the flavors include all the leptons and the @xmath49 , @xmath50 , @xmath51 and @xmath52 quarks .
* magnetic dipole moment @xmath53 * electric quadrupole moment @xmath54 * * isospin limit * let us show , just for illustration , the above expressions in the case of @xmath55 .
they become @xmath56 + @xmath57 + @xmath58 where @xmath59 . proceeding along the same lines of the @xmath0 gauge boson
, we obtain the modifications to the @xmath1 meson form factors , taking the complete expressions for the modified vertex computed in @xcite : * electric charge @xmath60 where , @xmath61 , @xmath62 is the effective @xmath63 coupling constant , and @xmath64 and @xmath65 . *
magnetic dipole moment @xmath66 * electric quadrupole moment @xmath67 * * isospin limit * the neutral and charged pions are almost degenerated . taking the isospin limit
the form factors become @xmath68 + @xmath69 + @xmath70 where @xmath71 .
the corrections for the @xmath72 meson follows from the results for the @xmath1 meson by including the two possible channels for the loop contributions : @xmath73 and @xmath74 , with the corresponding masses and partial decay widths .
the chiral limit correction to the vertex is known to be proportional to the tree level , in both the fermion and boson loops corrections @xcite .
therefore , in this limit we can write the modification to the form factors in a generic form as follows : @xmath75
the correction to the vertex seems to induce a modification to the the electric charge .
however , since the ward identity is fulfilled , the modification to the vertex is followed by a modification to the propagator , which produces an exact cancelation of the correction to the electric charge .
let us illustrate this point in more detail , for the sake of clarity we consider the expression in the chiral limit : the modified propagator can be set as @xcite : @xmath76 where @xmath77 , ( @xmath78 ) , i.e. , it can be seen as a renormalization of the vector field by the inclusion of the finite width @xmath79 where @xmath80 . then , the electromagnetic current becomes @xmath81 therefore , gauge invariance requires that @xmath82 , which is indeed the case , and the electric charge does not receive any correction .
note that the modification to the vertex given by equation [ chiral ] implies that none of the multipoles receive corrections in the chiral limit .
an analysis for unstable spin-1/2 particles has also been performed in ref .
@xcite , pointing out to complex renormalization factors as a requirement for properly defined physical quantities .
further considerations on the renormalizability of the wave function can be seen in ref .
+ the proper values of the modifications to the form factors are then found by the expressions given in the previous section divided by @xmath83 . in table
[ multi ] , we present the corresponding results for the @xmath0 gauge boson and the @xmath1 and @xmath2 mesons . recall that the particles in the loops are restricted to be on - shell . for the @xmath0 we have included all the leptons and the @xmath49 , @xmath50 , @xmath51 and @xmath52 quarks , and the numerical values are similar to those obtained in the chiral limit (
the largest @xmath84 mass ratio is only @xmath85 ) . as a reference of the magnitude of the modification
, we recall several results obtained for contributions of different nature : ref .
@xcite finds a correction to the mdm induced by the higgs of @xmath86 , reference @xcite finds a correction from quarks , leptons and higgs loops in the sm of @xmath87 , i.e. the width induced corrections is at least two orders of magnitude smaller than other standard contributions .
note that , due to the renormalization condition , the argument that the finite width correction to the vertex is of order @xmath88 ( @xmath890.03 , 0.19 and 0.06 for the @xmath0 , @xmath1 and @xmath2 respectively ) does not extend to the electromagnetic multipoles . for the @xmath1 meson ,
pions are the only on - shell particles allowed in the loops . in this case
the @xmath90 mass ratio ( @xmath91 ) is not as small as the corresponding for the @xmath84 , and therefore a significant effect from the mass of the particles in the loop is expected .
we can compare our results as shown in table [ multi ] with those shown in table [ tabref ] as obtained from several approaches to qcd : light - front framework with constituent quarks @xcite and covariant formulations based on the dyson - schwinger equations of qcd @xcite . in general these last are about the same or one order of magnitude larger than the finite width induced corrections computed in this work .
recently , lattice calculations of the form factors have become available @xcite and , in particular , the dependence on the pion mass they are able to reproduce has been exhibited . in figure
[ dipoloplot ] , we compare our result for the mdm with lattice calculations as a function of the pion mass @xcite , we also include predictions from the models at the physical mass of the pion .
we observe that the pion mass dependence of our results are mostly flat with a slightly tendency to rise for very large masses .
lattice results are also flat with a tendency to increase for low masses . the corrections for the @xmath72 meson is dominated by the @xmath92 mass ratio ( @xmath93 ) which is very large thus , although the width to mass ratio of the @xmath2 is only @xmath94 , the correction to the multipoles are important . compared with the predictions listed in table [ tabref ] , they can be about the same or one order of magnitude smaller . .@xmath0 gauge boson , and @xmath1 and @xmath2 mesons multipoles including the correction due to its finite width . [ cols="^,^,^,^",options="header " ,
] comparison of the @xmath1 meson mdm as obtained from different approaches .
this work ( solid line ) , ref .
@xcite ( square ) , ref .
@xcite ( triangle up ) , ref .
@xcite ( triangle down ) , ref .
@xcite ( circle ) , ref @xcite ( diamond ) and ref .
@xcite ( square with error bars ) . ] @xmath95 [ @xmath96 + @xcite & @xmath1 : 2 + 0.26 & @xmath1 : 1 + 0.22 + @xcite & @xmath1 : 2 + 0.14 & @xmath1 : 1 + 1.65 + @xcite & @xmath1 : 2 - 0.08 & @xmath1 : 1 - 0.57 + @xcite & @xmath1 : 2 + 0.01 & @xmath1 : 1 - 1.41 + @xcite & @xmath1 : 2 + 0.69 & @xmath1 : 1 + 1.8 + & @xmath72 : 2 + 0.37 & @xmath72 : 1 + 0.96 + @xcite & @xmath1 : 2 + 0.25 & @xmath1 : 1 - 0.75 + & @xmath72 : 2 + 0.14 & @xmath72 : 1 - 0.62 + as a byproduct , the mean square radius can be computed following @xcite as , @xmath97 . for the @xmath1
we obtain a deviation of @xmath98 @xmath99 respect to the normal value ( defined for @xmath1001 and @xmath101 ) , which can be compared , for example , with the one computed in reference @xcite , where they observe a correction of @xmath102 @xmath99 , due to the inclusion of the pion contribution respect to a pure quark- antiquark state . for @xmath2
the deviation is @xmath103 @xmath99 .
the inclusion of the unstable features of spin-1 particles , without breaking the electromagnetic gauge invariance , induces a non trivial modification to the electromagnetic vertex of the particle . in this work
we have extracted the corresponding modifications to the multipole structure of the @xmath0 and @xmath104 mesons .
our numerical results for the @xmath0 gauge boson multipoles shows no substantial deviations from the stable case . for the @xmath1 and @xmath2 mesons
, the mass of the particles in the loop makes a significant effect , pointing out that the unstable nature of the vector mesons can be as relevant as other dynamical effects and should be considered in refinements when accounting for their properties .
the modifications in both the propagator and electromagnetic vertex in combination with the gauge invariance show that the properly defined form factors can be seen as accompanied by a complex renormalization of the vector fields .
the general grounds of the loop schemes for spin-1 particles , to account for the finite decay width in a gauge invariant way , have been invoked to study spin-3/2 particles @xcite . since in this case
the mass ratio between the unstable particle and the ones in the loop can be very large , further studies are desirable to understand at which extend the finite decay width contributes to the multipoles . f. t. hawes and m. a. pichowsky , phys .
c * 59 * 1743 ( 1999 ) .
f. cardarelli , et al .
b * 349 * , 393 ( 1995 ) .
j. p. b. c. demelo and t. frederico , phys .
c * 55 * 2043 ( 1997 ) .
- meoyng choi and chueng - ryong ji , phys . rev .
d * 70 * 053015 ( 2004 ) m. s. bhagwat and p. maris , phys .
c * 77 * 025203 ( 2008 ) .
u. baur and d. zeppenfeld phys .
lett . * 75 * 1002(1995 ) .
e. n. argyres et al .
phys lett .
b * 358 * 339(1995 ) .
m. beuthe , r. gonzales felipe , g. lpez castro and j. pestieau , nucl .
b * 498 * 55(1997 ) .
g. lpez castro and g. toledo snchez , phys .
d * 61 * 033007(2000 ) .
see , for example m. napsuciale , s. rodriguez , e. g. delgado - acosta and m. kirchbach , phys .
d * 77 * , 014009 ( 2008 ) , and references therein .
bardin , a. leike , t riemann , and m. sachwitz , phys .
b * 206 * , 539(1988 ) ; g. lpez castro , j.l .
lucio m. , and j. pestieau , mod .
lett . a * 6 * , 3679 ( 1991 ) .
j. gegelia and s. scherer , arxiv:0910.4280v1 b. a. kniehl and a. sirlin , phys .
b * 530 * , 129(2002 ) , and references therein .
k. j. kim and y - s tsai , phys .
d * 7 * 3710(1973 ) . b. a. arbuzov , eur .
j. c * 61 * , 51 ( 2009 ) [ arxiv:0901.3997 [ hep - ph ] ] .
j. papavassilious , [ arxiv : hep - ph/9504382 ] .
m. a. pichowsky , s. walawalkar and s. capstick , phys .
d * 60 * 054030(1999 ) g. lopez castro and a. mariano , phys .
b * 517 * , 339 ( 2001 ) [ arxiv : nucl - th/0006031 ] ; nucl .
phys . a * 697 * , 440 ( 2002 ) [ arxiv : nucl - th/0010045 ] . | the inclusion of the unstable features of a spin-1 particle , without breaking the electromagnetic gauge invariance , can be properly accomplished by including higher order contributions as done in the so - called fermion loop scheme ( for the @xmath0 gauge boson ) , and the boson loop scheme ( for vector mesons ) .
this induces a non trivial modification to the electromagnetic vertex of the particle , which must be considered in addition to any other contribution computed as stable particles .
considering the modified electromagnetic vertex , we obtain general expressions for the corresponding corrections to the multipoles as a function of the mass of the particles in the loop . for the @xmath0 gauge boson no substantial deviations from the stable case is observed . for the @xmath1 and @xmath2 mesons
the mass of the particles in the loop makes a significant effect , and can be comparable with corrections of different nature . |
in recent years , several unexpected astrophysical observations have failed to find a common interpretation in terms of standard astrophysical or particle physics sources .
a non - exhaustive list of these observations includes the 511 kev gamma - ray signal from the galactic center observed by the integral satellite @xcite , the excess in the cosmic ray positrons reported by pamela @xcite , the total electron and positron flux measured by atic @xcite , fermi @xcite , and hess @xcite , the annual modulation of the dama / libra signal @xcite and the low energy spectrum of nuclear recoil candidate events observed by cogent @xcite .
although there are alternative explanations for some of these anomalies , they could be all explained with the existence of a dark matter weakly interacting massive particle , wimp , belonging to a secluded gauge sector under which the standard model ( sm ) particles are uncharged @xcite .
an abelian gauge field , the @xmath1 boson with mass near the gev scale , couples the secluded sector to the sm through its kinetic mixing with the sm hyper - charge gauge field .
the kinetic mixing parameter , @xmath11 , is expected to be of the order 10@xmath1210@xmath13 @xcite , so that observable effects can be induced in @xmath14)energy @xmath15 colliders @xcite and fixed target experiments @xcite .
the possible existence of a new light boson gauging a new symmetry with a small coupling was in fact already introduced on general grounds in @xcite , and rediscussed in models postulating also the existence of light spin 0 or 1/2 dark matter particles @xcite .
this boson can have both vector and axial - vector couplings to quark and leptons , however axial couplings are strongly constrained by data , leaving room to vector couplings only . the @xmath1 boson can be produced at colliders via different processes : @xmath16 , @xmath17 ( @xmath18-strahlung ) , where @xmath18 is a higgs - like particle responsible for the breaking of the hidden symmetry , and @xmath19 decays , where @xmath20 and @xmath21 are vector and pseudoscalar mesons , respectively . in this work ,
we study the process @xmath22 , using a sample of @xmath23 mesons produced resonantly at the da@xmath0ne collider .
the @xmath1 boson can be observed by its decay into a lepton pair , while the @xmath24 can be tagged by one of its main decays .
an irreducible background due to the dalitz decay of the @xmath23 meson , @xmath25 , is present .
this decay has been studied by the snd and cmd-2 experiments , which measured a branching fraction of @xmath26 and @xmath27 , respectively @xcite .
this corresponds to a cross section of @xmath28 nb , with a di - lepton mass range @xmath29 mev .
for the signal , the expected cross section is expressed by @xcite : @xmath30 where @xmath31 is the @xmath32 transition form factor evaluated at the @xmath1 mass while the following term represents the ratio of the kinematic functions of the involved decays.^2 - 4 m_1 ^ 2 m_3 ^ 2/(m_1 ^ 2-m_2 ^ 2)^2 $ ] . ] using @xmath33 and @xmath34 , a cross section @xmath35 is obtained . despite the small ratio between the overall cross section of @xmath2 and @xmath25 , their different di - lepton invariant mass distributions allow to test the @xmath11 parameter down to @xmath36 with the kloe data set .
the best @xmath1 decay channel to search for the @xmath2 process at kloe is in @xmath15 , since a wider range of @xmath1 boson masses can be tested and @xmath37 are easily identified using a time - of - flight ( tof ) technique .
the @xmath24 can be tagged by the three - pion or two - photon final state , which represent @xmath38 of the total decay rate .
we have used the @xmath39 decay channel , which provides a clean final state with four charged particles and two photons .
the kloe experiment operated from 2000 to 2006 at da@xmath0ne , the frascati @xmath23-factory .
da@xmath0ne is an @xmath15 collider running at a center - of - mass energy of @xmath40 mev , the mass of the @xmath23 meson .
equal energy positron and electron beams collide at an angle of @xmath41 - 25 mrad , producing @xmath23 mesons nearly at rest .
the detector consists of a large cylindrical drift chamber ( dc ) , surrounded by a lead - scintillating fiber electromagnetic calorimeter ( emc ) . a superconducting coil around the emc provides a 0.52 t field .
the beam pipe at the interaction region is spherical in shape with 10 cm radius , it is made of a beryllium - aluminum alloy of 0.5 mm thickness .
low beta quadrupoles are located at about @xmath4250 cm distance from the interaction region .
the drift chamber @xcite , 4 m in diameter and 3.3 m long , has 12,582 all - stereo tungsten sense wires and 37,746 aluminum field wires .
the chamber shell is made of carbon fiber - epoxy composite with an internal wall of @xmath43 mm thickness , the gas used is a 90% helium , 10% isobutane mixture .
the spatial resolutions are @xmath44 m and @xmath45 2 mm .
the momentum resolution is @xmath46 .
vertexes are reconstructed with a spatial resolution of @xmath47 3 mm .
the calorimeter @xcite is divided into a barrel and two endcaps , for a total of 88 modules , and covers 98% of the solid angle .
the modules are read out at both ends by photomultipliers , both in amplitude and time .
the readout granularity is @xmath47(4.4 @xmath48 4.4 ) @xmath49 , for a total of 2440 cells arranged in five layers .
the energy deposits are obtained from the signal amplitude while the arrival times and the particles positions are obtained from the time differences .
cells close in time and space are grouped into energy clusters .
the cluster energy @xmath50 is the sum of the cell energies .
the cluster time @xmath51 and position @xmath52 are energy - weighted averages .
energy and time resolutions are @xmath53 and @xmath54 , respectively . the trigger @xcite uses both calorimeter and chamber information . in this analysis
the events are selected by the calorimeter trigger , requiring two energy deposits with @xmath55 mev for the barrel and @xmath56 mev for the endcaps .
a cosmic veto rejects events with at least two energy deposits above 30 mev in the outermost calorimeter layer .
data are then analyzed by an event classification filter @xcite , which selects and streams various categories of events in different output files .
the analysis of the decay chain @xmath2 , @xmath3 , @xmath57 , has been performed on a data sample of 1.5 fb@xmath5 , corresponding approximately to @xmath58 produced @xmath23 mesons .
the monte carlo ( mc ) simulation of the irreducible background @xmath59 , @xmath3 has been produced with @xmath60 weighted according to vector meson dominance model @xcite , using the form factor parametrization from the snd experiment @xcite . the mc simulation for the @xmath61 decay has been developed according to @xcite , with a flat distribution in @xmath62 .
all mc productions , including all other @xmath23 decays , take into account changes in da@xmath0ne operation and background conditions on a run - by - run basis .
data - mc corrections for cluster energies and tracking efficiency , evaluated with radiative bhabha events and @xmath63 samples , respectively , have been applied . as a first step of the analysis
, a preselection is performed requiring : 1 .
two positive and two negative tracks with point of closest approach to the beam line inside a cylinder around the interaction point ( ip ) , with transverse radius @xmath64 cm and length @xmath65 cm ; 2 . two photon candidates _
i.e. _ two energy clusters with @xmath66 mev not associated to any track , in an angular acceptance @xmath67 and in the expected time window for a photon ( @xmath68 ) ; 3 .
best @xmath69 match to the @xmath24 mass in the pion hypothesis to assign @xmath70 tracks ; for each positive / negative track pair , @xmath71 , is evaluated in the hypothesis that the two tracks belong to charged pions . the track pair with the smaller @xmath72 value
is assigned to and . ]
the other two tracks are then assigned to @xmath37 ; 4 .
loose cuts of about @xmath73 s on @xmath24 and @xmath74 invariant masses ( @xmath75 mev , @xmath76 mev ) . after this selection
, a clear peak corresponding to @xmath59 events is observed in the distribution of the recoil mass to the @xmath15 pair , @xmath77 , as shown in fig .
[ fig : mmiss ] .
the second peak at @xmath78 mev is due to @xmath79 events with a wrong mass assignment .
events in the @xmath80 mev window are retained for further analysis .
a residual background contamination , due to @xmath81 events with photon conversion on beam pipe ( bp ) or drift chamber walls ( dcw ) , is rejected by tracing back the tracks of the two @xmath82 , @xmath83 candidates and reconstructing their invariant mass ( @xmath62 ) and distance ( @xmath84 ) at the bp / dcw surfaces .
as both quantities are small in case of photon conversions , @xmath81 background is removed by rejecting events with : @xmath85 mev and @xmath86 cm , @xmath87 mev and @xmath88 cm .
a further relevant background , originated by @xmath89 decays surviving analysis cuts , has more than two charged pions in the final state and is suppressed using time - of - flight ( tof ) to the calorimeter .
when an energy cluster is connected to a track , the arrival time to the calorimeter is evaluated both using the calorimeter timing ( @xmath90 ) and the track trajectory ( @xmath91 ) .
the @xmath92 variable is then evaluated in both electron ( @xmath93 ) and pion ( @xmath94 ) hypotheses .
events with an @xmath82 , @xmath83 candidate outside a 3@xmath95 s window on the @xmath93 variables are rejected . in fig .
[ fig : bckgreduction ] , the @xmath62 distribution evaluated at different steps of the analysis is shown .
the peaks at @xmath96 mev and @xmath97 mev are due to photon conversions on bp and dcw , respectively .
the tof cut reduces the tail at high @xmath62 values while the conversion cut removes events in the low invariant mass region . the analysis efficiency , defined as the ratio between the number of events surviving analysis cuts and that of all generated events , is shown in fig .
[ fig : effi ] as a function of @xmath62 , ranging between 10% and 20% .
the main contribution to the loss of efficiency is due to preselection cuts , being @xmath98 . in fig .
[ fig : allcuts ] the comparison between data and monte carlo events for @xmath62 and @xmath99 distributions is shown . here
@xmath100 is the angle between the @xmath24 and the @xmath82 in the @xmath15 rest frame .
about 14,000 @xmath101 , @xmath3 candidates are present in the analyzed data set , with a negligible residual background contamination .
as an accurate description of the background is crucial for the search of the @xmath1 boson , its shape is extracted directly from our data .
a fit is performed to the @xmath62 distribution , after applying a bin - by - bin subtraction of the @xmath81 background and efficiency correction .
the parametrization of the fitting function has been taken from ref .
@xcite : @xmath102 with @xmath103 and the transition form factor described by @xmath104 free parameters of the fit are @xmath105 and an overall normalization factor . a good description of the @xmath62 shape
is obtained except at the high end of the spectrum ( see fig . [
fig : fit ] ) , where a residual background contamination from multi - pion events is still present . as mentioned in sec .
[ sec : datasample ] , the @xmath2 mc signal has been produced according to ref .
@xcite , with a flat distribution of the @xmath1 boson invariant mass , @xmath106 .
this sample has been used to evaluate the resolution on the invariant mass as a function of @xmath106 , applying a gaussian fit to the @xmath107 distributions .
results are reported in fig .
[ fig : resomee ] .
the resolution is @xmath108 mev for @xmath109 mev and then improves to 1 mev for higher values . the upper limit on @xmath2 signal as a function of @xmath110
is then obtained in the following way : * mc events are divided in sub - samples of 1 mev width in the range @xmath111 mev ; * for each @xmath110 sub - sample , the average value of the @xmath59 background , @xmath112 , is obtained by fitting the reconstructed @xmath62 spectrum with 5 mev binning , removing five bins centered at @xmath110 ; * for each fit , the maximum variation of @xmath112 events , @xmath113 , is obtained changing by @xmath114 the fit parameters ; * for each @xmath110 value , the signal hypothesis is tested comparing observed data , @xmath112 and mc signal in the five reconstructed bins excluded in ( b ) .
the exclusion plot is obtained applying the cls method @xcite .
a gaussian spread of width @xmath113 on the background distribution is applied while evaluating cls . in fig .
[ fig : ul_nev ] the exclusion plot at 90% c.l . on the number of events for the decay chain @xmath2 , @xmath3 , @xmath4 , is shown .
using eq .
( [ eq : reece ] ) and taking into account the analysis efficiency this result is then reported in terms of the parameter @xmath115 , where @xmath116 is the coupling of the @xmath1 boson to electrons and @xmath117 is the fine structure constant . the opening of the @xmath118 threshold , in the hypothesis that the @xmath1 boson decays only to lepton pairs and assuming equal coupling to and @xmath119 , has been included . in fig .
[ fig : ulcurves ] the smoothed exclusion plot at 90% c.l . on @xmath7
is compared with existing limits from the muon anomalous magnetic moment @xmath120 @xcite and from recent measurements of the mami / a1 @xcite and apex @xcite experiments .
the gray line is where the @xmath1 boson parameters should lay to account for the observed discrepancy between measured and calculated @xmath120 values .
our result greatly improves existing limits in a wide mass range , resulting in an upper limit on the @xmath7 parameter of @xmath8 @ 90% c.l . for @xmath9 mev , thus covering part of the expected @xmath11 range ( see sec . [
sec : intro ] ) .
we exclude that the existing @xmath120 discrepancy is due to a @xmath1 boson with mass ranging between 90 and 450 mev .
we warmly thank our former kloe colleagues for the access to the data collected during the kloe data taking campaign .
we thank the da@xmath0ne team for their efforts in maintaining low background running conditions and their collaboration during all data taking .
we want to thank our technical staff : g. f. fortugno and f. sborzacchi for their dedication in ensuring efficient operation of the kloe computing facilities ; m. anelli for his continuous attention to the gas system and detector safety ; a. balla , m. gatta , g. corradi and g. papalino for electronics maintenance ; m. santoni , g. paoluzzi and r. rosellini for general detector support ; c. piscitelli for his help during major maintenance periods .
this work was supported in part by the eu integrated infrastructure initiative hadronphysics project under contract number rii3-ct-2004 - 506078 ; by the european commission under the 7th framework programme through the research infrastructures action of the capacities programme , call : fp7-infrastructures-2008 - 1 , grant agreement n. 227431 ; by the polish ministery of science and higher education through the grant no .
0469/b / h03/2009/37 . | the existence of a light dark force mediator has been tested with the kloe detector at da@xmath0ne .
this particle , called @xmath1 , is searched for using the decay chain @xmath2 , @xmath3 , @xmath4 .
no evidence is found in 1.5 fb@xmath5 of data .
the resulting exclusion plot covers the mass range @xmath6 mev , setting an upper limit on the ratio between the @xmath1 boson coupling constant and the fine structure constant , @xmath7 , of @xmath8 at 90% c.l . for @xmath9 mev . , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , .
@xmath10 collisions , dark forces , gauge vector boson 14.70.pw |
non - dominated sorting is a combinatorial problem that is fundamental in multiobjective optimization , which is ubiquitous is scientific and engineering contexts @xcite .
the sorting can be viewed as arranging a finite set of points in euclidean space into layers according to the componentwise partial order .
the layers are obtained by repeated removal of the set of minimal elements .
more formally , given a set @xmath0 of @xmath1 points equipped with the componentwise partial order @xmath2 for @xmath3 .
] , the first layer , often called the first pareto front and denoted @xmath4 , is the set of minimal elements in @xmath5 .
the second pareto front @xmath6 is the set of minimal elements in @xmath7 , and in general the @xmath8 pareto front @xmath9 is given by @xmath10 in the context of multiobjective optimization , the @xmath11 coordinates of each point in @xmath5 are the values of the @xmath11 objective functions evaluated on a given feasible solution . in this way
, each point in @xmath5 corresponds to a feasible solution and the layers provide an effective ranking of all feasible solutions with respect to the given optimization problem .
rankings obtained in this way are at the heart of genetic and evolutionary algorithms for multiobjective optimization , which have proven to be valuable tools for finding solutions numerically @xcite .
figure [ fig : example - fronts ] gives a visual illustration of pareto fronts for randomly generated points .
it is important to note that non - dominated sorting is equivalent to the longest chain problem in combinatorics , which has a long history beginning with ulam s famous problem of finding the length of a longest increasing subsequence in a sequence of numbers ( see @xcite and the references therein ) .
the longest chain problem is then intimately related to problems in combinatorics and graph theory @xcite , materials science @xcite , and molecular biology @xcite . to see this connection ,
let @xmath12 denote the length of a longest chain . ] in @xmath5 consisting of points less than or equal to @xmath13 with respect to @xmath2 .
if all points in @xmath5 are distinct , then a point @xmath14 is a member of @xmath4 if and only if @xmath15 . by peeling off @xmath4 and making the same argument
, we see that @xmath16 is a member of @xmath6 if and only if @xmath17 . in general , for any @xmath14 we have @xmath18 this is a fundamental observation .
it says that studying the shapes of the pareto fronts is equivalent to studying the longest chain function @xmath19 .
the longest chain problem has well - understood asymptotics as @xmath20 . in this context , we assume that @xmath21 where @xmath22 are _ i.i.d . _ random variables in @xmath23 and let @xmath24 denote the length of a longest chain in @xmath5 .
the seminal work on the problem was done by hammersley @xcite , who studied the problem for @xmath22 _ i.i.d .
_ uniform on @xmath25 ^ 2 $ ] .
he utilized subadditive ergodic theory to show that @xmath26 in probability , where @xmath27 .
he conjectured that @xmath28 , and this was later proven by vershik and kerov @xcite and logan and shepp @xcite .
hammersley s results were generalized to higher dimensions by bollobs and winkler @xcite , who showed that @xmath29 almost surely , where @xmath30 are constants tending to @xmath31 as @xmath32 .
the only known values of @xmath33 are @xmath34 and @xmath35 .
deuschel and zeitouni @xcite provided another generalization of hammersley s results ; for @xmath22 _ i.i.d .
_ on @xmath25 ^ 2 $ ] with @xmath36 density function @xmath37 ^ 2 \to { \mathbb{r}}$ ] , bounded away from zero , they showed that @xmath38 in probability , where @xmath39 is the supremum of the energy @xmath40 over all @xmath41\to[0,1]$ ] nondecreasing and right continuous . in @xcite
, we studied the longest chain problem for @xmath22 _ i.i.d . _ on @xmath42 with density function @xmath43 . under general assumptions on @xmath44
, we showed that @xmath45 in @xmath46 almost surely , where @xmath47 is the viscosity solution of the hamilton
jacobi equation @xmath48 here @xmath49 and @xmath50 .
in this paper we study a fast numerical scheme for ( p ) , first proposed in @xcite , and prove convergence of this scheme .
we then show how the scheme can be used to design a fast approximate non - dominated sorting algorithm , which requires access to only a fraction of the datapoints @xmath22 , and we evaluate the sorting accuracy of the new algorithm on both synthetic and real data . a fast approximate algorithm for non - dominated sorting has the potential to be a valuable tool for multiobjective optimization , especially in evolutionary algorithms which require frequent non - dominated sorting @xcite .
there are also potential applications in polynuclear growth of crystals in materials science @xcite . here , the scheme for ( p ) could be used to simulate polynuclear growth in the presence of a macroscopically inhomogeneous growth rate .
this paper is organized as follows . in section
[ sec : convergence ] we prove that the numerical solutions converge to the viscosity solution of ( p ) .
we also prove a regularity result for the numerical solutions ( see lemma [ lem : holder ] ) and other important properties . in section [ sec : num ] we demonstrate the numerical scheme on several density functions , and in section [ sec : fast ] we propose a fast algorithm for approximate non - dominated sorting that is based on numerical solving ( p ) .
let us first fix some notation . given @xmath51 we write @xmath52 if @xmath53 and @xmath54 .
we write @xmath55 when @xmath56 for all @xmath57 . for
@xmath58 , @xmath59 and @xmath60 will retain their usual definitions .
for @xmath53 we define @xmath61 = \{z \in { \mathbb{r}}^d \ , : \ , x \leqq z \leqq y\ } , \ \ ( x , y ] = \{z \in { \mathbb{r}}^d \ , : \ , x <
z \leqq y\},\ ] ] and make similar definitions for @xmath62 and @xmath63 . for any @xmath64 and @xmath65 , there exists unique @xmath66 and @xmath67 such that @xmath68
. we will denote @xmath69 by @xmath70 so that @xmath71 .
we also denote @xmath72 and @xmath73 . for @xmath74 , we denote by @xmath75 $ ] the projection mapping @xmath42 onto @xmath76 $ ] . for @xmath77 this mapping
is given explicitly by @xmath78 we say a function @xmath79 is pareto - monotone if @xmath80 we now recall the numerical scheme from @xcite .
let @xmath65 . for a given @xmath77 ,
the domain of dependence for ( p ) is @xmath81 .
this can be seen from the connection to non - dominated sorting and the longest chain problem .
it is thus natural to consider a scheme for ( p ) based on backward difference quotients , yielding @xmath82 where @xmath83 is the numerical solution of ( p ) and @xmath84 are the standard basis vectors in @xmath42 . under reasonable hypotheses on @xmath44 , described in section [ sec : convergence - proof ] , there exists a unique pareto - monotone viscosity solution of ( p ) .
as we wish to numerically approximate this pareto - monotone solution we may assume that @xmath85 for all @xmath57 .
given that @xmath44 is non - negative , for any @xmath86 , there is a unique @xmath87 with @xmath88 satisfying .
hence the numerical solution @xmath89 can be computed by visiting each grid point exactly once via any sweeping pattern that respects the partial order @xmath2 .
the scheme therefore has linear complexity in the number of gridpoints . at each grid point
, the scheme can be solved numerically by either a binary search and/or newton s method restricted to the interval @xmath90.\ ] ] in the case of @xmath91 , we can solve the scheme explicitly via the quadratic formula @xmath92 now extend @xmath89 to a function @xmath93 by setting @xmath94 .
defining @xmath95 we see that @xmath89 is a pareto - monotone solution of the discrete scheme @xmath96 where @xmath97 is defined by @xmath98 here , @xmath99 is the space of functions @xmath100 . in the next section
we will study properties of solutions @xmath89 of ( s ) .
in this section we prove that the numerical solutions @xmath89 defined by ( s ) converge uniformly to the viscosity solution of ( p ) . as in @xcite , we place the following assumption on @xmath101 : * there exists an open and bounded set @xmath102 with lipschitz boundary such that @xmath103 is lipschitz and @xmath104 .
it is worthwhile to take a moment to motivate the hypothesis ( h ) .
consider the following multi - objective optimization problem @xmath105 where @xmath106 with @xmath107 for all @xmath57 , and @xmath108 is the set of feasible solutions .
this formulation includes many types of constrained optimization problems , where the constraints are implicitly encoded into @xmath108 .
if @xmath109 are feasible solutions in @xmath108 , then these solutions are ranked , with respect to the optimization problem , by performing non - dominated sorting on @xmath110 .
thus the domain @xmath111 of @xmath22 is given by @xmath112 .
supposing that @xmath109 are , say , uniformly distributed on @xmath108 , then the induced density @xmath44 of @xmath22 on @xmath42 will be nonzero on @xmath111 and identically zero on @xmath113 .
thus , the constraint that feasible solutions must lie in @xmath108 directly induces a discontinuity in @xmath44 along @xmath114 . in @xcite
we showed that , under hypothesis ( h ) , there exists a unique pareto - monotone viscosity solution @xmath47 of ( p ) satisfying the additional boundary condition @xmath115 the boundary condition is natural for this problem . indeed , since @xmath116 , there are almost surely no random variables drawn outside of @xmath117 .
hence , for any @xmath77 we can write @xmath118^d \ , : \ , y\leqq x } u_n(y).\ ] ] since @xmath19 is pareto - monotone , the maximum above is attained at @xmath119 , and hence @xmath120 . for completeness ,
let us now give a brief outline of the proof of uniqueness for ( p ) .
for more details , we refer the reader to @xcite . the proof is based on the auxiliary function technique , now standard in the theory of viscosity solutions @xcite .
however , the technique must be modified to account for the fact that @xmath44 is possibly discontinuous on @xmath114 , and hence does not possess the required uniform continuity .
a commonly employed technique is to modify the auxiliary function so that only a type of one - sided uniform continuity is required of @xmath44 @xcite .
this allows @xmath44 to , for example , have a discontinuity along a lipschitz curve , provided the jump in @xmath44 is locally in the same direction ( see @xcite for more details ) .
we can not directly use these results because they require coercivity or uniform continuity of the hamiltonian and/or lipschitzness of solutions
none of which hold for ( p ) . our technique for proving uniqueness for ( p ) employs instead an important property of viscosity solutions of ( p)namely that for any @xmath121 , @xmath122 is a viscosity subsolution of ( p ) .
this property , called _ truncatability _ in @xcite , follows immediately from the variational principle @xcite @xmath123 this allows us to prove a comparison principle with no additional assumptions on the hamiltonian .
a general framework for proving convergence of a finite - difference scheme to the viscosity solution of a non - linear second order pde was developed by barles and souganidis @xcite .
their framework requires that the scheme be stable , monotone , consistent , and that the pde satisfy a _ strong uniqueness property
_ @xcite .
the monotonicity condition is equivalent to ellipticity for second order equations , and plays a similar role for first order equations , enabling one to prove maximum and/or comparison principles for the discrete scheme .
the strong uniqueness property refers to a comparison principle that holds for semicontinuous viscosity sub- and supersolutions . the numerical scheme ( s )
is easily seen to be consistent ; this simply means that @xmath124 for all @xmath125 .
the scheme is stable @xcite if the numerical solutions @xmath89 are uniformly bounded in @xmath126 , independent of @xmath127 .
it is not immediately obvious that ( s ) is stable ; stability follows from the discrete comparison principle for ( s ) ( lemma [ lem : discrete - comp ] ) and is proved in lemma [ lem : holder ] .
the monotonicity property requires the following : @xmath128 it is straightforward to verify that ( s ) is monotone when restricted to pareto - monotone @xmath129 .
this is sufficient since we are only interested in the pareto - monotone viscosity solution of ( p ) .
all that is left is to establish a strong uniqueness result for ( p ) .
unfortunately such a result is not available under the hypothesis ( h ) .
since @xmath44 may be discontinuous along @xmath114 , we can only establish a comparison principle for continuous viscosity sub- and supersolutions ( see ( * ? ? ?
* theorem 4 ) ) .
one way to rectify this situation is to break the proof into two steps .
first prove convergence of the numerical scheme for @xmath44 lipschitz on @xmath130 .
it is straightforward in this case to establish a strong uniqueness result for ( p ) .
second , extend the result to @xmath44 satisfying ( h ) by an approximation argument using inf and sup convolutions .
although this approach is fruitful , we take an alternative approach as it yields an interesting regularity property for the numerical solutions . in particular , in lemma [ lem : holder ] we establish approximate hlder regularity of @xmath89 of the form latexmath:[\[\label{eq : holder - est } verify in appendix a , the approximate hlder estimate along with the stability of ( s ) allows us to apply the arzel - ascoli theorem , with a slightly modified proof , to the sequence @xmath89 .
this allows us to substitute the ordinary uniqueness result from @xcite in place of strong uniqueness .
we first prove a discrete comparison principle for the scheme ( s ) .
this comparison principle is essential in proving stability of ( s ) and the approximate hlder regularity result in lemma [ lem : holder ] .
for the remainder of this section , we fix @xmath65 .
[ lem : discrete - comp ] let @xmath132 and suppose @xmath133 are pareto - monotone and satisfy @xmath134.\ ] ] then @xmath135 on @xmath136 $ ] implies that @xmath135 on @xmath76 $ ] .
suppose that @xmath137 } ( u - v ) > 0 $ ] and set @xmath138 and @xmath139\}.\ ] ] since @xmath135 on @xmath140 and @xmath137 } ( u - v ) > 0 $ ] , we must have @xmath141 $ ] , where @xmath142 . by the definition of @xmath143
, there exists @xmath144 $ ] and @xmath145 such that @xmath146 since @xmath145 , we have @xmath135 on @xmath147 $ ] and hence @xmath148 the second inequality above follows from pareto - monotonicity of @xmath149 .
since @xmath150 and @xmath149 are pareto - monotone and @xmath151 we have @xmath152 hence @xmath153 , contradicting the hypothesis . using the comparison principle
, we can establish that numerical solutions of ( s ) satisfy the boundary condition at infinity .
[ prop : boundary_condition ] let @xmath154 be pareto - monotone with @xmath155 on @xmath156 .
suppose that for some @xmath132 we have @xmath157.\ ] ] then we have @xmath158 .
define @xmath159 and fix @xmath77 .
since @xmath150 is pareto - monotone and @xmath160 , we have @xmath161 .
hence @xmath162 .
since @xmath163 on @xmath76 $ ] we have @xmath164\setminus \gamma_h.\ ] ] for @xmath165\cup \gamma_h$ ] we have @xmath166 by assumption .
since @xmath149 is pareto - monotone we have @xmath167 for such @xmath13 , and hence @xmath168 for all @xmath169 .
since @xmath170 on @xmath156 we can apply lemma [ lem : discrete - comp ] to find that @xmath171 on @xmath172 , and hence @xmath173 .
an important consequence of the comparison principle is the following approximate hlder regularity result .
[ lem : holder ] let @xmath174 be pareto - monotone with @xmath155 on @xmath156 .
then for any @xmath175 we have @xmath176^d$ ] .
let @xmath175 and @xmath177^d$ ] .
we first deal with the case where @xmath178 .
set @xmath179 and define @xmath180 by @xmath181 by the concavity of @xmath182 we have @xmath183 for any @xmath184 and hence @xmath185 by the translation invariance of @xmath186 and we have @xmath187 set @xmath188 . for @xmath77 set @xmath189^d)}\sum_{i=1}^d \psi(x - b^i),\ ] ] and note that @xmath190 is pareto - monotone .
let @xmath191 $ ] . then for some @xmath192 we have @xmath193 , and hence @xmath194 .
we therefore have @xmath195^d ) } ( \psi(x - b^k ) - \psi(x - b^k - h e_i))\bigg ) \\ & { } \geq{}s(h , x,{\widehat}{u } ) + \|s(h,\cdot , u)\|_{l^\infty((h , r]^d ) } s(h , x,\psi(\cdot - b^k))\\ & { } \hspace{-2.4mm}\stackrel{\eqref{eq : h - super}}{\geq } { } \hspace{-2.4mm}s(h , x,{\widehat}{u } ) + \|s(h,\cdot , u)\|_{l^\infty((h , r]^d ) } \\ & { } \geq { } s(h , x , u).\end{aligned}\ ] ] suppose now that @xmath196 $ ] .
then since @xmath197 on @xmath198 $ ] we have @xmath199 and hence @xmath200 .
since @xmath201 on @xmath202^d$ ] , we can apply lemma [ lem : discrete - comp ] to obtain @xmath203 on @xmath204^d$ ] , which yields @xmath205^d)}^\frac{1}{d } \sum_{i=1}^d \psi(y_0 - b^i ) \notag \\ & { } \leq { } dr^\frac{d-1}{d}\|s(h,\cdot , u)\|_{l^\infty((h , r]^d)}^\frac{1}{d } \sum_{i=1}^d ( y_{0,i } - x_{0,i } + h)^\frac{1}{d}\notag \\ & { } \leq { } d^2r^\frac{d-1}{d}\|s(h,\cdot , u)\|_{l^\infty((h , r]^d)}^\frac{1}{d } ( |x_0 - y_0|^\frac{1}{d } + h^\frac{1}{d}).\end{aligned}\ ] ] noting that @xmath206 we have @xmath207 , which completes the proof for the case that @xmath178 .
suppose now that @xmath177^d$ ] such that @xmath208 .
set @xmath209 then @xmath210 , @xmath211 , @xmath212 , and @xmath213 . it follows that @xmath214^d)}^\frac{1}{d } ( |x_0 - y_0|^\frac{1}{d } + h^\frac{1}{d}),\end{aligned}\ ] ] which completes the proof .
our main result is the following convergence statement for the scheme ( s ) .
[ thm : conv ] let @xmath44 be nonnegative and satisfy ( h ) .
let @xmath47 be the unique pareto - monotone viscosity solution of ( p ) satisfying .
for every @xmath65 let @xmath93 be the unique pareto - monotone solution of ( s ) . then @xmath215 uniformly on @xmath172 as @xmath216 .
by ( h ) we have that @xmath217 for @xmath218 , and hence @xmath219^d$ ] .
therefore , by proposition [ prop : boundary_condition ] , we have that @xmath89 satisfies . combining this with lemma [ lem : holder ]
we have @xmath220 for all @xmath65 .
similarly , combining with lemma [ lem : holder ] we have latexmath:[\[\label{eq : holder } for every @xmath65 . the estimates in and show uniform boundedness , and a type of equicontinuity , respectively , for the sequence @xmath89 .
by an argument similar to the proof of the arzel - ascoli theorem ( see the appendix ) , there exists a subsequence @xmath222 and @xmath223 such that @xmath224 uniformly on compact sets in @xmath172 . by
, we actually have @xmath224 uniformly on @xmath172 .
since the scheme ( s ) is monotone and consistent , it is a standard result that @xmath150 is a viscosity solution of ( p ) @xcite .
note that @xmath89 is pareto - monotone , @xmath225 on @xmath156 , and @xmath89 satisfies .
since @xmath224 uniformly , it follows that @xmath150 is pareto - monotone , @xmath155 on @xmath226 , and @xmath150 satisfies . by uniqueness for ( p ) ( * ? ? ?
* theorem 5 ) we have @xmath227 . since we can apply the same argument to any subsequence of @xmath89 , it follows that @xmath215 uniformly on @xmath172 . in section [ sec : num ] , we observe that the numerical scheme provides a fairly consistent underestimate of the exact solution of ( p ) .
the following lemma shows that this is indeed the case whenever the solution @xmath47 of ( p ) is concave .
[ lem : conv - below ] let @xmath44 be nonnegative and satisfy ( h ) .
let @xmath47 be the unique pareto - monotone viscosity solution of ( p ) satisfying .
for every @xmath65 let @xmath93 be the unique pareto - monotone solution of ( s ) . if @xmath47 is concave on @xmath172 then @xmath228 for every @xmath65 .
fix @xmath65 . since @xmath47 is concave , it is differentiable almost everywhere .
is pareto - monotone also implies differentiability almost everywhere .
] let @xmath184 be a point at which @xmath47 is differentiable and @xmath44 is continuous .
since @xmath47 is concave we have @xmath229 since @xmath47 is a viscosity solution of ( p ) and @xmath44 is continuous at @xmath13 we have @xmath230 since @xmath231 is continuous , we see that @xmath232 for all @xmath233 $ ]
. now define @xmath234 .
then we have @xmath235,\ ] ] and @xmath236 on @xmath156 .
it follows from lemma [ lem : discrete - comp ] that @xmath237 .
since @xmath47 is pareto - monotone , we have @xmath238 , which completes the proof .
we now present some numerical results using the scheme ( s ) to approximate the viscosity solution of ( p ) .
we consider four special cases where the exact solution of ( p ) can be expressed in analytical form .
let @xmath239 , @xmath240 , @xmath241^d}(x ) \ \ \
\text{and } \ \ \ f_4(x ) { } = { } \left(\sum_{i=1}^d x_i^9\right)^{1-d } \prod_{i=1}^d \left ( 9 x_i^9 + \sum_{i=1}^d x_i^9\right).\end{aligned}\ ] ] here , @xmath242 denotes the characteristic function of the set @xmath243 .
the corresponding solutions of ( p ) are @xmath244 , @xmath245 , and @xmath246 where @xmath247 is the error function defined by @xmath248 , and @xmath249 . the solutions @xmath250 and @xmath251 are special cases of the formula @xmath252 } f(y ) \ , dy\right)^\frac{1}{d},\ ] ] which holds when @xmath44 is separable , i.e. , @xmath253 @xcite .
the solution @xmath254 can be obtained by the method of characteristics .
we chose to evaluate the proposed numerical scheme for @xmath255 because it has non - convex level sets , and then computed @xmath256 via ( p ) . in the probabilistic interpretation of ( p ) as the continuum limit of
non - dominated sorting , non - convex pareto fronts play an important role @xcite .
we computed the numerical solutions for @xmath91 and @xmath257 .
for @xmath91 we used a @xmath258 grid , and for @xmath257 , we used a @xmath259 grid and solved the scheme at each grid point via a binary search with precision @xmath260 .
figures [ fig:2d - sim ] and [ fig:3d - sim ] compare the level sets of the exact solutions to those of the numerical solutions for @xmath91 and @xmath257 , respectively . in figure [ fig:2d - sim ] , the thin lines correspond to the exact solution while the thick lines correspond to the numerical solutions , with the exception of [ fig : nonconvex-2d ] where both are thin lines for increased visibility . in figure
[ fig:3d - sim ] , the darker surfaces correspond to the numerical solution while the lighter surfaces represent the exact solution . for both @xmath91 and @xmath257 , we can see that the level sets of the numerical solutions consistently overestimate the true solution , indicating that the numerical solutions are converging from below to the exact solutions .
we proved in lemma [ lem : conv - below ] that @xmath261 whenever @xmath47 is concave , so this observation is to be expected .
note however , that @xmath254 is not convex , yet the overestimation is still present , indicating that lemma [ lem : conv - below ] may hold under more general hypotheses on @xmath47 .
we also observe that @xmath254 has a shock , which is resolved reasonably well for @xmath91 and @xmath257 , given the grid sizes used .
we show here the results of some numerical experiments concerning the rate of convergence of @xmath215 and @xmath262 .
figure [ fig : pde - rates ] shows @xmath263 ^ 2)}$ ] and @xmath264 versus @xmath127 for the density @xmath265^d}(x)$ ] from the beginning of section [ sec : num ] .
both norms appear to have convergence rates on the order of @xmath266 , and a regression analysis yields @xmath267 for the @xmath126 norm and @xmath268 for the @xmath269 norm .
thus , it is reasonable to suspect an @xmath126 convergence rate of the form @xmath270 for some constant @xmath271 .
we intend to investigate this in a future work .
it is quite natural that the convergence rate for the @xmath269 norm is substantially better than the @xmath126 norm , due to the non - differentiability of @xmath254 at the boundary @xmath272 .
this induces a large error near @xmath272 which has a more significant impact on the @xmath126 norm . to measure the rate of convergence of @xmath273
, we consider the following two norms @xmath274 shows and versus @xmath1 for the same density @xmath275 .
for each @xmath1 the values of and were computed by taking the average over @xmath276 independent realizations .
it appears that both norms decay on the order of @xmath277 , and a regression analysis yields @xmath278 for the @xmath269 norm and @xmath279 for the @xmath126 norm .
these results are in line with the known convergence rates for the longest chain problem with a uniform distribution on @xmath25^d$ ] @xcite .
the results for the other densities @xmath280 and @xmath256 are similar .
we demonstrated the convergence rates on @xmath275 due to the fact that it has many important features ; namely , it is discontinuous , yields non - convex pareto - fronts , and induces a shock in the viscosity solution @xmath254 of ( p ) .
we demonstrate now how the numerical scheme ( s ) can be used for fast approximate non - dominated sorting , and give a real - world application to anomaly detection in section [ sec : real ] .
we assume here that the given data @xmath22 are drawn _ i.i.d . _ from a reasonably smooth density function @xmath44 , and that @xmath1 is large enough so that @xmath281 is well approximated by @xmath282 . in this regime , it is reasonable to consider an approximate non - dominated sorting algorithm based on numerically solving ( p ) .
a natural algorithm is as follows .
since the density @xmath44 is rarely known in practice , the first step is to form an estimate @xmath283 of @xmath44 using the samples @xmath22 . in the large sample regime , this can be done very accurately using , for example , a kernel density estimator @xcite or a @xmath192-nearest neighbor estimator @xcite . to keep the algorithm as simple as possible , we opt for a simple histogram to estimate @xmath44 , aligned with the same grid used for numerically solving ( p ) .
when @xmath1 is large , the estimation of @xmath44 can be done with only a random subset of @xmath22 of cardinality @xmath284 , which avoids considering all @xmath1 samples .
the second step is to use the numerical scheme ( s ) to solve ( p ) on a fixed grid of size @xmath127 , using the estimated density @xmath283 on the right hand side of ( p ) .
this yields an estimate @xmath285 of @xmath47 , and the final step is to evaluate @xmath285 at each sample @xmath22 to yield approximate pareto ranks for each point .
the final evaluation step can be viewed as an interpolation ; we know the values of @xmath285 on each grid point and wish to evaluate @xmath285 at an arbitrary point .
a simple linear interpolation is sufficient for this step .
however , in the spirit of utilizing the pde ( p ) , we solve the scheme ( s ) at each point @xmath22 using the values of @xmath285 at neighboring grid points , i.e. , given @xmath286 for all @xmath57 , and @xmath287 $ ] , we compute @xmath288 by solving @xmath289 where @xmath290 . in
we compute @xmath291 by linear interpolation using adjacent grid points .
figure [ fig : subgrid ] illustrates the grid used for computing @xmath288 .
according to .
the values of @xmath292 and @xmath293 are computed by linear interpolation using adjacent grid points , i.e. , @xmath292 is computed via linearly interpolating between @xmath294 and @xmath295.,scaledwidth=45.0% ] the entire algorithm is summarized in algorithm [ alg : ndom ] .
[ alg : ndom ] fast approximate non - dominated sorting 1 .
select @xmath192 points from @xmath22 at random . call them @xmath296 .
2 . select a grid spacing @xmath127 for solving the pde and estimate @xmath44 with a histogram aligned to the grid @xmath297 , i.e. , @xmath298 3 .
compute the numerical solution @xmath285 on @xmath299^d$ ] via ( s ) .
4 . evaluate @xmath300 for @xmath301 via interpolation . for simplicity of discussion
, we have assumed that @xmath22 are drawn from @xmath25^d$ ] , but this is not essential as the scheme ( s ) can be easily adapted to any hypercube in @xmath42 , and this is in fact what we do in our implementation of algorithm [ alg : ndom ] .
it is important to understand how the parameters @xmath192 and @xmath127 in algorithm [ alg : ndom ] affect the accuracy of the estimate @xmath302 .
we first consider the estimate @xmath303 . by
, we can write @xmath304}(y_i).\ ] ] hence @xmath305 is the average of _ i.i.d . _ bernoulli random variables with parameter @xmath306 } f(y ) \ , dy.\ ] ] by the central limit theorem , the fluctuations of @xmath307 about its mean satisfy @xmath308 with high probability .
let us suppose now that @xmath44 is globally lipschitz .
the following can be easily modified for @xmath44 more or less regular , yielding similar results . then by we
have @xmath309 combining this with we have @xmath310^d\cap h{\mathbb{n}}^d ) } \leq c\left ( \frac{1}{\sqrt{k}h^d } + h\right),\ ] ] with high probability . by the discrete comparison principle ( lemma [ lem : discrete - comp ] ) and we have that @xmath311^d ) } \leq d\|{\widehat}{f}_h - f\|^\frac{1}{d}_{l^\infty([0,1]^d\cap h{\mathbb{n}}^d ) } \leq c\left(k^{-\frac{1}{2d}}h^{-1 } + h^\frac{1}{d}\right),\ ] ] with high probability . based on the numerical evidence presented in section [ sec : pde - rates ] ,
it is reasonable to suspect that @xmath312^d ) } \leq ch^\frac{1}{d}$ ] . if this is indeed the case , then in light of we have @xmath313^d ) } \leq c\left(k^{-\frac{1}{2d}}h^{-1 } + h^\frac{1}{d}\right),\ ] ] with high probability .
the right side of the inequality is composed of two competing additive terms . the first term @xmath314 captures the effect of random errors ( variance ) due to an insufficient number @xmath192 of samples .
the second term @xmath315 captures the effect of non - random errors ( bias ) due to insufficient resolution @xmath127 of the proposed numerical scheme ( s ) .
this decomposition into random and non - random errors is analogous to the mean integrated squared error decomposition in the theory of non - parametric regression and image reconstruction @xcite .
similarly to @xcite we can use the bound in to obtain rules of thumb on how to choose @xmath192 and @xmath127 .
for example , we may first choose some value for @xmath192 , and then choose @xmath127 so as to equate the two competing terms in .
this yields @xmath316 and becomes @xmath317^d ) } \leq ck^{-\frac{1}{2d(d+1 ) } } = ch^\frac{1}{d},\ ] ] with high probability .
notice that steps 1 - 3 in algorithm [ alg : ndom ] , i.e. , computing @xmath302 , require @xmath318 operations . if we choose the equalizing value @xmath316 , then we find that computing @xmath302 has complexity @xmath319 .
thus algorithm [ alg : ndom ] is sublinear in the following sense . given @xmath320 , we can choose @xmath192 large enough so that @xmath321^d ) } \leq \frac{{\varepsilon}}{2c_d},\ ] ] with high probability .
the @xmath269 sorting accuracy of using @xmath302 in place of @xmath19 is then given by @xmath322 with high probability . by the stochastic convergence @xmath323 , and the rates presented in section [ sec : pde - rates ]
, there exists @xmath324 such that for all @xmath325 we have @xmath326 with high probability .
thus , for any @xmath320 there exists @xmath327 and @xmath127 such that @xmath302 is an @xmath328 approximation of @xmath19 for all @xmath325 , and @xmath302 can be computed in constant time with respect to @xmath1 .
we emphasize that the sublinear nature of the algorithm lies in the computation of @xmath302 .
ranking all samples , i.e. , evaluating @xmath302 at each of @xmath22 , and computing the @xmath269 error in of course requires @xmath329 operations . in practice , it is often the case that one need not rank all @xmath1 samples ( e.g. , in a streaming application @xcite ) , and in such cases the entire algorithm is constant or sublinear in @xmath1 in the sense described above .
we evaluated our proposed algorithm in dimension @xmath91 for a uniform density and a mixture of gaussians given by @xmath330 , where each @xmath331 is a multivariate gaussian density with covariance matrix @xmath332 and mean @xmath333 .
we write the covariance matrix in the form @xmath334 , where @xmath335 denotes a rotation matrix , and @xmath336 , @xmath337 are the eigenvalues .
the values for @xmath338 and @xmath339 are given in table [ tab : param ] , and the density is illustrated in figure [ fig : gauss - den ] .
.parameter values for mixture of gaussians density [ cols="<,<,<,<,<",options="header " , ] it is important to evaluate the accuracy of the approximate sorting obtained by algorithm [ alg : ndom ] . in practice ,
the numerical ranks assigned to each point are largely irrelevant , provided the relative orderings between samples are correct . hence a natural accuracy measure for a given ranking
is the fraction of pairs @xmath340 that are ordered correctly . recalling that the true pareto rank is given by @xmath341 , this can be expressed as @xmath342 where @xmath343 if @xmath344 and @xmath345 otherwise .
it turns out that the accuracy scores for our algorithm are often very close to 1 . in order to make the plots easier to interpret visually ,
we have chosen to plot @xmath346 instead of accuracy in _ all _ plots .
unfortunately , the complexity of computing the accuracy score via is @xmath347 , which is intractable for even moderate values of @xmath1 .
we note however that is , at least formally , a monte - carlo approximation of @xmath348 hence it is natural to use a truncated monte - carlo approximation to estimate .
this is done by selecting @xmath1 pairs @xmath349 at random and computing @xmath350 the complexity of the monte - carlo approximation is @xmath329 . in all plots in the paper , we computed the monte - carlo approximation @xmath276 times and plotted means and error bars corresponding to a @xmath351 confidence interval . in all of the figures ,
the confidence intervals are sufficiently small so that they are contained within the data point itself .
we can see in figure [ fig : acc ] that we can achieve excellent accuracy while maintaining a fixed grid and subsample size as a function of @xmath1 .
we also see that , as expected , the accuracy increases when one uses more grid points for solving the pde and/or more subsamples for estimating the density .
we also see that the algorithm works better on uniformly distributed samples than on the mixture of gaussians .
indeed , it is quite natural to expect the density estimation and numerical scheme to be less accurate when @xmath44 changes rapidly . we compared the performance of our algorithm against the fast two dimensional non - dominated sorting algorithm presented in @xcite , which takes @xmath352 operations to sort @xmath1 points .
the code for both algorithms was written in c++ and was compiled on the same architecture with the same compiler optimization flags .
figure [ fig : cputime ] shows a comparison of the cpu time used by each algorithm . for our fast approximate sorting , we show the cpu time required to solve the pde ( steps 1 - 3 in algorithm [ alg : ndom ] ) separately from the cpu time required to execute all of algorithm [ alg : ndom ] , since the former is sublinear in @xmath1 .
it is also interesting to consider the relationship between the grid size and the number of subsamples @xmath192 . in figure
[ fig : grid ] , we show accuracy versus grid size for @xmath353 and @xmath354 subsamples for non - dominated sorting of @xmath355 points . notice that for @xmath353 subsamples , it is not beneficial to use a finer grid than approximately @xmath356 .
this is quite natural in light of the error estimate on algorithm [ alg : ndom ] .
there are certainly other ways one may think of to perform fast approximate sorting without invoking the pde ( p ) .
one natural idea would be to perform non - dominated sorting on a random subset of @xmath22 , and then rank all @xmath1 points via some form of interpolation .
we will call such an algorithm _ subset ranking _ ( in contrast to the pde - based ranking we have proposed ) .
although such an approach is quite intuitive , it is important to note that there is , at present , no theoretical justification for such an approach .
nonetheless , it is important to compare the performance of our algorithm against such an algorithm .
let us describe how one might implement a subset ranking algorithm .
as described above , the first step is to select a random subset of size @xmath192 from @xmath22 .
let us call the subset @xmath296 .
we then apply non - dominated sorting to @xmath296 , which generates pareto rankings @xmath357 for each @xmath358 .
the final step is to rank @xmath22 via interpolation .
there are many ways one might approach this . in similar spirit to our pde - based ranking ( algorithm 1 ) , we use grid interpolation , using the same grid size as used to solve the pde .
we compute a ranking at each grid point by averaging the ranks of all samples from @xmath296 that fall inside the corresponding grid cell .
the ranking of an arbitrary sample @xmath359 is then computed by linear interpolation using the ranks of neighboring grid points . in this way ,
the rank of @xmath359 is an average of the ranks of nearby samples from @xmath296 , and there is a grid size parameter which allows a meaningful comparison with pde - based ranking ( algorithm 1 ) .
figure [ fig : subsample ] shows the accuracy scores for pde - based ranking ( algorithm 1 ) and subset ranking of @xmath360 samples drawn from the uniform and mixture of gaussians distributions .
a grid size of @xmath361 was used for both algorithms , and we varied the number of subsamples from @xmath362 to @xmath363 . notice a consistent accuracy improvement when using pde - based ranking versus subset ranking , when the number of subsamples is significantly less than @xmath1 .
it is somewhat surprising to note that subset ranking has much better than expected performance .
as mentioned previously , to our knowledge there is no theoretical justification for such a performance when @xmath192 is small .
we now demonstrate algorithm [ alg : ndom ] on a large scale real data application of anomaly detection @xcite .
the data consists of thousands of pedestrian trajectories , captured from an overhead camera , and the goal is to differentiate nominal from anomalous pedestrian behavior in an unsupervised setting .
the data is part of the edinburgh informatics forum pedestrian database and was captured in the main building of the school of informatics at the university of edinburgh @xcite .
figure [ fig : traj ] shows 100 of the over 100,000 trajectories captured from the overhead camera .
the approach to anomaly detection employed in @xcite utilizes multiple criteria to measure the dissimilarity between trajectories , and combines the information using a pareto - front method , and in particular , non - dominated sorting .
the database consists of a collection of trajectories @xmath364 , where @xmath365 , and the criteria used in @xcite are a walking speed dissimilarity , and a trajectory shape dissimilarity . given two trajectories
@xmath366 \to [ 0,1]^2 $ ] , the walking speed dissimilarity @xmath367 is the @xmath368 distance between velocity histograms of each trajectory , and the trajectory shape dissimilarity is the @xmath368 distance between the trajectories themselves , i.e. , @xmath369 .
there is then a pareto point @xmath370 for every pair of trajectories @xmath371 , yielding @xmath372 pareto points .
figure [ fig : points ] shows an example of 50000 pareto points and figure [ fig : fronts ] shows the respective pareto fronts . in @xcite , only 1666 trajectories from one day were used , due to the computational complexity of computing the dissimilarities and non - dominated sorting .
the anomaly detection algorithm from @xcite performs non - dominated sorting on the pareto points @xmath373 , and uses this sorting to define an anomaly score for every trajectory @xmath374 .
let @xmath375 and let @xmath376 denote the longest chain function corresponding to this non - dominated sorting .
the anomaly score for a particular trajectory @xmath374 is defined as @xmath377 and trajectories with an anomaly score higher than a predefined threshold @xmath378 are deemed anomalous . using algorithm [ alg : ndom ] , we can approximate @xmath19 using only a small fraction of the pareto points @xmath379 , thus alleviating the computational burden of computing all pairwise dissimilarities .
figure [ fig : acc - real ] shows the accuracy scores for algorithm [ alg : ndom ] and subset ranking versus the number of subsamples @xmath192 used in each algorithm . due to the memory requirements for non - dominated sorting , we can not sort datasets significantly larger than than @xmath380 points .
although there is no such limitation on algorithm 1 , it is important to have a ground truth sorting to compare against . therefore we have used only @xmath381 out of @xmath382 trajectories , yielding approximately @xmath380 pareto points . for both algorithms ,
a @xmath383 grid was used for solving the pde and interpolation . notice the accuracy scores are similar to those obtained for the test data in figure [ fig : acc ] .
this is an intriguing observation in light of the fact that @xmath384 are _ not _ _ i.i.d .
_ , since they are elements of a euclidean dissimilarity matrix .
we have provided theory that demonstrates that , when @xmath22 are _ i.i.d .
_ in @xmath385 with a nicely behaved density function @xmath44 , the numerical scheme ( s ) for ( p ) can be utilized to perform fast approximate non - dominated sorting with a high degree of accuracy .
we have also shown that in a real world example with non-_i.i.d .
_ data , the scheme ( s ) still obtains excellent sorting accuracy .
we expect the same algorithm to be useful in dimensions @xmath257 and @xmath386 as well , but of course the complexity of solving ( p ) on a grid increases exponentially fast in @xmath11 . in higher dimensions , one could explore other numerical techniques for solving ( p ) which do not utilize a fixed grid @xcite . at present , there is also no good algorithm for non - dominated sorting in high dimensions .
the fastest known algorithm is @xmath387 @xcite , which becomes intractable when @xmath1 and @xmath11 are large .
this algorithm has the potential to be particularly useful in the context of big data streaming problems @xcite , where it would be important to be able to construct an approximation of the pareto depth function @xmath19 without visiting all the datapoints @xmath22 , as they may be arriving in a data stream and it may be impossible to keep a history of all samples .
in such a setting , one could slightly modify algorithm [ alg : ndom ] so that upon receiving a new sample , the estimate @xmath303 is updated , and every so often the scheme ( s ) is applied to recompute the estimate of @xmath302 .
there are certainly many situations in practice where the samples @xmath22 are not _ i.i.d .
_ , or the density @xmath44 is not nicely behaved . in these cases , there is no reason to expect our algorithm to have much success , and hence we make no claim of universal applicability .
however , there are many cases of practical interest where these assumptions are valid , and hence this algorithm can be used to perform fast non - dominated sorting in these cases . furthermore , as we have demonstrated in section [ sec : real ] , there are situations in practice where the _ i.i.d . _ assumption is violated , yet our proposed algorithm maintains excellent accuracy and performance .
we proposed a simple _ subset ranking _ algorithm based on sorting a small subset of size @xmath192 and then performing interpolation to rank all @xmath1 samples .
although there is currently no theoretical basis for such an algorithm , we showed that subset ranking achieves surprisingly high accuracy scores and is only narrowly outperformed by our proposed pde - based ranking .
the simplicity of subset ranking makes it particularly appealing , but more research is needed to prove that it will always achieve such high accuracy scores for moderate values of @xmath192 .
we should note that there are many obvious ways to improve our algorithm .
histogram approximation to probability densities is quite literally the most basic density estimation algorithm , and one would expect to obtain better results with more sophisticated estimators .
it would also be natural to perform some sort of histogram equalization to @xmath22 prior to applying our algorithm in order to spread the samples out more uniformly and smooth out the effective density @xmath44 .
provided such a transformation preserves the partial order @xmath2 it would not affect the non - dominated sorting of @xmath22 . in the case that @xmath44 is separable ( a product density ) , one can perform histogram equalization on each coordinate independently to obtain uniformly distributed samples .
we leave these and other potential improvements to future work ; our purpose in this paper has been to demonstrate that one can obtain excellent results with a very basic algorithm .
we thank ko - jen hsiao for providing code for manipulating the pedestrian trajectory database .
we use the following minor extension of the arzel - ascoli theorem in section [ sec : convergence - proof ] .
let @xmath99 be a compact metric space .
we say that a sequence @xmath388 of real - valued functions on @xmath99 is _ approximately equicontinuous _ if for every @xmath320 there exists @xmath389 such that @xmath390 for every @xmath391 .
let @xmath397 . since @xmath388 is approximately equicontinuous there exists @xmath398 such that for all @xmath1 we have @xmath399 forms an open cover of @xmath99 .
since @xmath99 is compact , there exists a finite subcover @xmath400 for some integer @xmath401 . without loss of generality
we may assume that @xmath402 .
now let @xmath403 .
by we have @xmath404 for some @xmath405 and any @xmath406 .
hence we have @xmath407 it follows that @xmath394 is cauchy in @xmath126 , which completes the proof . | non - dominated sorting is a fundamental combinatorial problem in multiobjective optimization , and is equivalent to the longest chain problem in combinatorics and random growth models for crystals in materials science . in a previous work @xcite
, we showed that non - dominated sorting has a continuum limit that corresponds to solving a hamilton
jacobi equation . in this work we present and analyze a fast numerical scheme for this hamilton
jacobi equation , and show how it can be used to design a fast algorithm for approximate non - dominated sorting . |
in statistical mechanics the thermodynamic functions are given in terms of the logarithm of the partition function and its derivatives with respect to volume , temperature , etc .
the mathematical expression of the canonical partition function is an infinite sum over all the states of the system or , under some simplifying assumptions , the states of the individual particles.@xcite some of these infinite sums are commonly calculated approximately by means of the euler - maclaurin summation formula@xcite and most textbooks simply show how to apply it to the cases of interest.@xcite a rigorous derivation of the summation formula may be rather too demanding for inexperienced students and one avoids it in introductory courses on statistical thermodynamics or statistical mechanics .
there is a relatively simple way of deriving the summation formula by means of operator methods@xcite but it also requires some kind of mathematical expertise that students of introductory courses may not posses .
however , the students may feel more confident and confortable if they are shown how to carry out such calculations by means of mathematical methods that they have already learned in a first course on mathematical analysis . in an introductory undergraduate course
we show how to obtain the first two terms of the euler - maclaurin summation formula in an extremely simple way that only requires the students to be familiar with the taylor expansion .
the aim of this paper is to put forward this approach that we deem suitable for pedagogical purposes . in section [ sec : simple_appro ]
we outline the method and apply it to some simple examples in section [ sec : examples ] .
if we decide to show the students how to obtain terms of higher order this simple approach results to be rather cumbersome .
for that reason , in section [ sec : systematic ] we show how to derive the full summation formula by a relatively minor change of strategy that requires some additional mathematical skills .
the reader may choose one or another approach depending on the level of the course .
finally , in section [ sec : conclusions ] we summarize the results and draw conclusions .
the simplest canonical partition functions in statistical mechanics can be expressed as sums of the form @xmath0 where @xmath1 should tend to zero sufficiently fast when @xmath2 because , otherwise , the sum does not converge . in most cases
one can not obtain this sum in closed form and therefore resorts to some kind of approximation valid under certain conditions ; for example , sufficiently high temperature . in order to derive such an approach
we define @xmath3 that satisfies @xmath4 and @xmath5 substituting the taylor expansion @xmath6 for @xmath7 into ( [ eq : s(x)_prop2 ] ) we obtain , after some rearrangement , @xmath8 if we integrate this expression between @xmath9 and @xmath10 and take into account that @xmath11 and all its derivatives vanish at the upper limit we have @xmath12 that can be solved iteratively . in the first step
we omit the derivatives of @xmath11 so that this function is approximately given by @xmath11 @xmath13 .
if we substitute this approximate result into the right - hand side of ( [ eq : s(x)_iter ] ) we have @xmath14 thus , for @xmath15 we obtain a simple approximation to the sum ( [ eq : s ] ) : @xmath16 the simple method just outlined is suitable for introductory courses because it only requires basic knowlegde in mathematical analysis .
this procedure is not suitable for the systematic calculation of the corrections of higher order to the basic formula ( [ eq : s_approx_simp ] ) because it soon becomes rather cumbersome .
however , this result is suitable for most porposes in an introductory course on statistical mechanics because several partitions functions for atomic and simple molecular systems can be easily derived from it.@xcite
under some simplifying assumptions that we will not discuss in this paper statistical mechanics tell us how to express the thermodynamic functions of atomic and molecular systems in terms of one - particle partition functions of the form@xcite @xmath17 where @xmath18 is the volume of the container , @xmath19 the absolute temperature , @xmath20 the @xmath21-th energy level ( assumed to be @xmath22-fold degenerate ) of the particle and @xmath23 the boltzman constant .
let us first consider a particle of mass @xmath24 in a one - dimensional box of length @xmath25 with impenetrable walls which is the starting point for the calculation of the thermodynamic properties of an ideal monoatomic gas.@xcite if the particle has no internal structure its spectrum is only given by the translational degree of freedom : @xmath26 where @xmath27 is the planck constant @xmath28 divided @xmath29 and @xmath30 . upon defining @xmath31 @xmath32
the partition function @xmath33 reads @xmath34 and the approximation ( [ eq : s_approx_simp ] ) yields @xmath35 this approximate expression is accurate for sufficiently small values of @xmath36 or , equivalently , when the thermal de broglie wavelength @xmath37 is much smaller than the box length @xmath25.@xcite the addition of corrections of higher order commonly improves the result of the approximate partition function , but in this case all of them vanish.@xcite figure [ fig : pb ] shows that this approximate expression becomes increasingly more accurate as @xmath36 decreases . the spectrum of a one - dimensional harmonic oscillator is given by @xmath38 where @xmath39 is the frequency of the oscillation and @xmath30 .
the partition function for this degree of freedom is given by @xmath40 where @xmath41 .
in this case the sum @xmath42 is a geometric series that can be calculated exactly and therefore we can derive its small-@xmath36 expansion in a straightforward way : @xmath43 if we define the vibrational temperature @xmath44 then @xmath45 and the euler - maclaurin formula is a good approximation when @xmath46 . the approximate expression ( [ eq : s_approx_simp ] )
yields the first two terms of this series exactly .
the last example is the partition function for a rigid rotor with moment of inertia @xmath47 . in this case , every energy level @xmath48 is @xmath49-fold degenerate and the partition function for the rotational degree of freedom is given by @xmath50 where @xmath51
. the approximate expression ( [ eq : s_approx_simp ] ) yields @xmath52 that is accurate enough if @xmath36 is sufficently small as shown in figure [ fig : rr ] .
it is custommary to define the rotational temperature @xmath53@xcite so that @xmath54 when @xmath55 .
it is worth mentioning that the @xmath36-independent term in equation ( [ eq : s_rr_appro1 ] ) is not exact as shown in the following section .
if we decide to show the students how to obtain expression of higher order it is convenient to proceed in a different way .
if we write the taylor expansion discussed in section [ sec : simple_appro ] as @xmath56 where @xmath57 is the @xmath58-th derivative of @xmath11 with respect to @xmath9 , then equation ( [ eq : s(x)_iter ] ) reads @xmath59 where @xmath60 . instead of trying to solve equation ( [ eq : s(x)_iter2 ] ) iteratively
we propose a solution of the form @xmath61 where @xmath62 . in order to obtain the coefficients
@xmath63 we substitute ( [ eq : s(x)_ansatz ] ) into ( [ eq : s(x)_iter2 ] ) and compare the coefficients of @xmath64 in the left- and right - hand sides ; the result is @xmath65 we thus have @xmath66 and @xmath67 straightforward inspection of the first coefficients @xmath68 @xmath69 suggests that @xmath70 vanish for all @xmath71 . in order to prove this conjecture we consider the function @xmath72 that satisfies @xmath73 if we substitute the taylor expansion @xmath74 into equation ( [ eq : u(x)-u(-x ) ] ) we conclude that @xmath75 for all @xmath71 . if we now expand @xmath76 in a taylor series about @xmath15 and compare the coefficients of @xmath77 in the left- and right - hand sides of the resulting equation we obtain a recurrence relation for the coefficients @xmath78 that is identical with equation ( [ eq : a_n_rec ] ) for the coefficients @xmath68 .
we thus conclude that @xmath79 for all @xmath21
. therefore , equation ( [ eq : s_appro_gen1 ] ) reduces to @xmath80 it is worth noting that the coefficients @xmath81 are related to the brillouin numbers @xmath82@xcite in the following way : @xmath83 .
the operator method leads to this result in a more straightforward way and is also more convenient for the discussion of the radius of convergence of the series.@xcite however , we do not discuss it here because it requires the introduction of functions of operators that may not be suitable for an undergraduate course .
let us apply this more accurate approach to the examples discussed in section [ sec : examples ] .
as pointed out in that section nothing can be done with the particle in a box because all the corrections of higher order vanish .
this surprising result can be explained very easily and applies to all the sums in which @xmath84 because @xmath85 and @xmath86 .
a more rigorous analysis of such problems , based on the poisson summation formula , can be found eslewhere.@xcite the harmonic oscillator is a suitable simple example for testing the summation formula ( [ eq : s_appro_gen2 ] ) because we can calculate the exact expansion as shown in equation ( [ eq : s_exact_exp ] ) . in this case
every new term added to the summation formula ( [ eq : s_appro_gen2 ] ) yields one more term of the small-@xmath36 series ( [ eq : s_exact_exp ] ) as one may easily verify .
the application of the summation formula ( [ eq : s_appro_gen2 ] ) to the partition function for the rigid rotor should be carried out with care .
the reason is that @xmath87 , where @xmath88 is a polynomial function of @xmath36 of degree @xmath89 and @xmath90
. therefore , the summation formula that includes all the terms through @xmath91 only yields the small-@xmath36 series correctly through degree @xmath92 . for this reason the simple expression ( [ eq : s_approx_simp ] ) does not yield the correct @xmath36-independent term that receives contributions from @xmath93 and @xmath94 .
the expansion accurate to @xmath95 is easily shown to be @xmath96 if we add all the terms trough @xmath97 in the summation formula ( [ eq : s_appro_gen2 ] ) as argued above . since
this series does not converge@xcite one should truncate it before the terms start to increase .
as stated in the introduction we show the students the simple method developed in section [ sec : simple_appro ] because it is sufficient for the purposes of our course .
however , if a motivated student wants to learn how to obtain the corrections of greater order that appear in some of the available texbooks on statistical mechanics then one can suggest him or her to try the systematic approach of section [ sec : systematic ] or even the operator method.@xcite in our opinion the present way of deriving the euler - maclaurin summation formula for statistical mechanics applications is more convenient for pedagogical purposes than the traditional one that appears in most textbooks on mathematics or numerical analysis.@xcite such traditional approaches are certainly more rigorous and general but require the students to be more experienced in mathematics . in passing ,
it is worth mentioning that the summation formula given by equations ( [ eq : a_n_rec ] ) and ( [ eq : s_appro_gen2 ] ) is suitable for motivating the students to resort to a computer algebra system . in this way they bypass the tedious algebraic manipulation of the equations that is necessary for the calculation of contributions of large order like those in equation ( [ eq : s_rr_em ] ) and practise programming in any of such useful languages . | we propose a simple pedagogical way of introducing the euler - maclaurin summation formula in an undergraduate course on statistical mechanics .
we put forward two alternative routes : the first one is the simplest and yields the first two terms of the expansion .
the second one is somewhat more elaborate and takes into account all the correction terms .
we apply both to the calculation of the simplest one - particle canonical partition functions for the translational , vibrational and rotational degrees of freedom . |
one - dimensional and quasi - one - dimensional quantum chain models have been intensively investigated for a twofold reason .
on one hand , for the algebraic properties of spin-@xmath9 chain models ( @xmath1 and generalized ladder models ) which , being integrable models , can be solved exactly by the powerful bethe _ ansatz _ technique ( see refs . @xcite and @xcite , respectively ) . on the other hand , for the existence of quasi one - dimensional magnetic systems that exhibit nearly ideal one - dimensional behavior for some interval of temperature .
the thermodynamic properties of those magnetic systems have been described by spin-@xmath0 @xmath1 models for various values of spin , mixed - spin models , ladder models , etc . in particular , there are materials well described by higher spin values of the @xmath1 models such as @xmath10 and @xmath11 ( @xmath12)@xcite , @xmath13 ( @xmath14)@xcite and @xmath15 @xmath16@xcite . motivated by the discovery of these materials , the magnetic and thermodynamic properties of the ferro- and antiferromagnetic isotropic spin-@xmath0 @xmath1 model for higher values of @xmath0 have been studied numerically and analytically ( e.g. , the high temperature expansions ; see ref . @xcite and references therein ) .
the classical limit ( @xmath17 ) of the @xmath1 model is exactly soluble and was originally solved by fisher@xcite , in the isotropic regime , for a null external magnetic field and no single - ion anisotropy term .
its anysotropic case has been solved by joyce @xcite , also in the absence of an external magnetic field and without a single - ion anisotropy term , by writing the helmholtz free energy of the classical @xmath1 model as an integral equation whose solution is the spheroidal wave function .
unfortunately , in the presence of an external magnetic field and/or the single - ion anisotropy parameter , the solution of the integral equation can not be reduced to any known function .
recently , fukushima _ et al._@xcite obtained the specific heat and the magnetic susceptibility per site of a ferromagnetic mixed - spin model , with two kinds of spins , @xmath18 and @xmath0 , arranged alternatively and coupled by a heisenberg - type nearest - neighbor exchange for arbitrary values of @xmath18 and @xmath0 at the isotropic point ( @xmath19 in eq.([hamt1 ] ) ) and in the absence of external magnetic fields .
they obtained numerical results through the exact diagonalization method and the analytical high temperature expansions . for @xmath20
their model becomes a one - dimensional spin-@xmath0 @xmath1 model for arbitrary value of @xmath0 .
their high temperature expansion of the specific heat per site , in the absence of an external magnetic field , goes up to order @xmath21 and their expansion of the magnetic susceptibility , also for a vanishing magnetic field , goes up to @xmath22 with a single - ion anisotropy term in one of the spins . in ref .
@xcite we extended the results of ref .
@xcite , for spin values up to @xmath23 , including a free parameter of anisotropy in the @xmath2-direction , a single - ion anisotropy term and a non - zero external magnetic field in the @xmath3-expansions of the helmholtz free energies up to order @xmath24 . following ref .
@xcite , we also consider the hamiltonian of the anisotropic spin-@xmath0 @xmath1 with a single - ion anisotropy term and in the presence of an external magnetic field , @xmath25 we use the notation : @xmath26 . here
, @xmath27 , @xmath28 and @xmath29 stand for the spin-@xmath0 matrices in the _ i_-th site of the chain and norm @xmath30 ; @xmath31 is the number of sites in the periodic chain ; @xmath32 is the exchange integral ; @xmath33 is the anisotropy constant in the @xmath2-direction ; @xmath34 is the external magnetic field in the @xmath2-axis and @xmath35 is the single - ion anisotropy parameter .
even in one spatial dimension , the solution of the quantum chain model ( [ hamt1 ] ) is more complex if we consider a finite arbitrary spin-@xmath0 ( with @xmath36 ) once the @xmath1 model becomes a non - integrable model and can not be solved by the thermodynamic bethe _ ansatz _ technique . numerical approaches for higher spin values becomes more involved since there are more degrees of freedom to be handled .
certainly , the high temperature expansions are easily manipulated by symbolic computer languages and can be used as a reliable check for numerical calculations in the high temperature regime .
the aim of the present communication is to extend the results of ref .
@xcite to arbitrary values of the spin @xmath0 in hamiltonian ( [ hamt1 ] ) . in doing so ,
we are including in the results of ref .
@xcite the effects of anisotropies in the @xmath2-direction of the @xmath1 model in the presence of an external magnetic field . having a high temperature expansion for the model , for any @xmath0
, we can derive its classical limit in this regime of temperature .
we want to re - obtain the classical results of the heisenberg model by considering the quantum nature of the spin variable , in opposition to the known results in the literature where this limit is derived from a chain of classical spins . from the high temperature expansion of the helmholtz free energy , we can also verify for which values of @xmath0 the anisotropic quantum @xmath1 model , in the presence of an external magnetic field , is well described by its classical version , extending the results of ref .
@xcite . in the present article , comparison of thermodynamic functions for different values of @xmath0
is performed at the same ( high ) @xmath37 , whereas in ref .
@xcite , they were plotted as functions of the `` rescaled '' temperature @xmath38 .
this paper is organized as follows : in ii we use the method developed in ref .
@xcite and an interpolation technique to obtain the high - temperature series expansion of the helmholtz free energy for the quantum spin-@xmath18 xxz chain ( @xmath18 with unitary norm ) up to order @xmath4 . in order to check our analytic @xmath3-expansion , in the first subsection of ii our results for the classical limit ( @xmath39 )
are compared to the well - known isotropic@xcite and anisotropic@xcite classical heisenberg chains . in the second subsection of
ii our results for finite values of @xmath0 are explored to include the effect of the anisotropy parameter @xmath33 , the presence of @xmath40 and @xmath41 in the hamiltonian and the dependence of thermodynamic functions on @xmath0 . in
iii we use pad representatives to enhance our high - temperature expansions for the spin-@xmath18 @xmath1 model , extending them to lower temperatures .
we do so by taking into account the known behavior at @xmath42 . for the specific heat
we use the method presented in ref .
@xcite , whereas for the magnetic susceptibility we apply the dlog - pad approximant@xcite ( antiferromagnetic case ) and the two - point pad approximant@xcite ( ferromagnetic case ) . in
iv we present our conclusions . in appendix
a we show that the trace of powers of @xmath43 can be written as an expansion in the square of the norm of the spin at each site
. finally , in appendix b we present the high - temperature expansion , up to order @xmath44 , of the helmholtz free energy of the spin-@xmath18 @xmath1 model ( @xmath45 ) , with a single - ion anisotropy term and in the presence of an external magnetic field .
we intend to study the thermodynamics of the hamiltonian ( [ hamt1 ] ) for arbitrary spin , including its classical limit ( @xmath46 ) . in this limit ,
the thermodynamic functions diverge and to render them finite we define a rescaled spin operator @xmath47 .
this rescaled spin operator has unitary norm for all values of @xmath0 . rewriting the hamiltonian in terms of @xmath18 and redefining the parameters @xmath48 , @xmath49 and @xmath50
, we obtain @xmath51 this also describes the dynamics of the @xmath1 model of a spin of unitary norm and @xmath52 @xmath2-components .
we continue to use the notation : @xmath53 . here ,
@xmath54 , @xmath55 and @xmath56 stand for the spin-@xmath18 ( @xmath57 ) rescaled matrices in the _ i_-th site of the chain ; @xmath31 is the number of sites in the periodic chain ; and @xmath33 is the anisotropy constant in the @xmath2-direction . in order to evaluate the helmholtz free energy for arbitrary values of spin ,
we first calculate its expansion for a set of spin values using the hamiltonian ( [ hamiltonian ] ) .
we apply the method of ref.@xcite to calculate those @xmath3-expansions for arbitrary values of @xmath58 , @xmath33 , @xmath59 and @xmath60 .
we presented elsewhere@xcite the helmholtz free energy of hamiltonian ( [ hamt1 ] ) for a number of spin values ( semi - integer and integer ) up to @xmath23 .
we use the interpolation method to extend our results for fixed values of @xmath0 to arbitrary spin value . for the sake of obtaining the @xmath3-expansion of any spin-@xmath0 quantum chain up to order @xmath4 , having the expansions presented in ref .
@xcite is not enough . in order to derive the expansion of the helmholtz free energy of the hamiltonian ( [ hamt1 ] ) up to order @xmath61 ( for a survey of the method@xcite , cf .
section 2 of ref .
@xcite ) , we have to calculate normalized traces , at each site , of the products of @xmath62 matrices from the set \{@xmath63 , @xmath64 , @xmath43 } , so that @xmath65 and @xmath66 ranges over all possible powers of @xmath3 , @xmath67 . due to the commutation relation @xmath68=
i s^z$ ] and to the fact that the products @xmath69 and @xmath70 can be written in terms of @xmath43 , @xmath71 and the identity matrix @xmath72 , then the normalized trace at each site is reduced to the calculation of traces of @xmath73 , with @xmath74 . in eq .
we show that the @xmath75 , for even values of @xmath76 , is a polynomial of degree @xmath77 in @xmath0 .
therefore the polynomial in @xmath0 of highest degree at order @xmath61 is of order @xmath78 . in order to carry out the interpolation of the expansion up to order @xmath4
, we calculated the @xmath3-expansion of the helmholtz free energy from the spin @xmath79 up to @xmath80 . later , by inspection of the expansion in @xmath0 and in @xmath3 , we verified that the expansion in @xmath0 can be rewritten in terms of @xmath81 , that is , in terms of the eigenvalue of a constant of motion ( the square of the norm of the spin-@xmath82 at each site ) . in appendix [ appendixa ]
we show , for arbitrary value of spin , that @xmath83 $ ] , @xmath84 , is a polynomial of degree @xmath85 in the parameter @xmath81 .
the thermodynamic properties of the quantum spin-@xmath0 @xmath1 chain , in the high temperature region , can be obtained from its helmholtz free energy .
we present in appendix b the coefficients of the helmholtz free energy @xmath86 derived from the hamiltonian ( [ hamiltonian ] ) , up to order @xmath4 , for any value of @xmath87 .
in particular , the result of appendix b is also valid at the isotropic point of the hamiltonian with @xmath8 and @xmath7 , when the model has rotational symmetry .
the @xmath3-expansion of the helmholtz free energy of hamiltonian ( [ hamt1 ] ) is easily obtained from result ( [ free_e ] ) ( see appendix [ b ] ) .
et al._@xcite obtained the @xmath3-expansion of the specific heat and magnetic susceptibility per site of the latter model for arbitrary spin at the isotropic point and in the absence of an external magnetic field .
our results agree with theirs , up to order @xmath4 .
the range of validity in @xmath3 of our high - temperature expansions obviously depends on the particular values given to the parameters of .
the thermodynamic properties of the chain depend on the sign of the product @xmath88 .
we let @xmath89 and let @xmath33 refer to either the ferromagnetic @xmath90 or the antiferromagnetic @xmath91 phases .
an interesting check to our expansion is to verify if it recovers known classical limits ( @xmath39 or @xmath92 , where @xmath93 ) of the @xmath1 chain .
the classical limit of the @xmath1 chain of the hamiltonian ( [ hamiltonian ] ) with @xmath7 and @xmath8 has been solved by joyce@xcite ; this model ( including the anisotropic constant @xmath33 in the hamiltonian ) is exactly solvable and its helmholtz free energy is given by the radial component of the spheroidal wave function @xmath94 where @xmath95 and @xmath96 is the first radial spheroidal wave function@xcite .
for @xmath97 , eq . becomes the well - known result obtained by fisher@xcite , whose explicit expression is @xmath98 $ ] .
the classical limit of the @xmath3-expansion is obtained by letting @xmath99 . in order to have a shorter @xmath3-expansion of the helmholtz free energy at @xmath99 , we take @xmath8 , and obtain @xmath100 where @xmath101 . from ( [ w_clss ] )
we see that in the absence of the external magnetic field ( @xmath7 ) the helmholtz free energy is an even function of the anisotropic parameter @xmath33 . under this condition ,
the classical limit of the @xmath1 model does not distinguish between the ferromagnetic and the antiferromagnetic phases in the region of high temperatures .
( [ fig_1]@xmath102 ) shows a comparison of our @xmath3-expansion of the helmholtz free energy to the numerical solution of eq .
, for @xmath7 , @xmath8 and @xmath103 .
they are in good agreement in the high temperature region ; the percental error between them at @xmath104 is less than 2.8% .
( [ fig_1]@xmath105 ) performs the same comparison , but at the isotropic point ( @xmath106 ) .
we also have good agreement , but for higher temperatures ; percental error is about @xmath107 for @xmath108 . in conclusion ,
the classical limit of the expansion ( for @xmath99 ) coincides with joyce s solution of the helmholtz free energy of the hamiltonian ( [ hamiltonian ] ) in a larger interval of temperature than that which we call `` high temperature region '' .
the greater the ( finite ) spin , the greater the number of degrees of freedom to be handled , thus the more involved it gets to compute thermodynamic properties of the quantum spin-@xmath82 model .
in contrast , for infinite values of spin , the calculation of those properties for the classical version of the model is simpler , since we can treat the spin as a vector that rotates continuously .
it is interesting to know for which values of @xmath0 the quantum spin model can be well approximated by its classical version , at least in the high temperature region . in ref .
@xcite we compared various thermodynamic functions of the chain model at the heisenberg point ( @xmath109 ) , in the absence of the single - ion anisotropy term and external magnetic fields .
we verified , in the high temperature region , that thermodynamic functions like the magnetic susceptibility per site have a classical behavior , for @xmath110 .
our intention in this section is to extend the results of ref .
@xcite in order to include in this comparison ( quantum @xmath111 classical behavior ) of spin models with anisotropies in the @xmath2-direction ( @xmath112 and @xmath113 ) and in the presence of an external magnetic field .
this analysis can be realized from the expansion ( [ free_e ] ) of @xmath114 .
the referred thermodynamic functions are all derived from the hamiltonian ( [ hamiltonian ] ) .
let @xmath115 and @xmath116 be the specific heat per site for the spin-@xmath18 model ( with unitary norm ) and for the spin-@xmath0 chain ( with norm equal to @xmath30 ) , respectively .
the specific heat per site is calculated from a derivative of the helmholtz free energy ( @xmath117 ) . from the results ( [ b2 ] ) we obtain , [ 5 ]
@xmath118 these results show that @xmath119 is a homogeneous function of zero degree for all temperatures .
the @xmath3- expansion of the specific heat derived from eq .
( [ free_e ] ) satisfies eqs.([5 ] ) identically . from the high temperature expansion of @xmath120 we verify that @xmath121 . in the high temperature region ,
the @xmath1 model presents a tail of the schottky peak@xcite ( @xmath122 ) , for all values of @xmath0 .
[ fig_2 ] shows the specific heat per site as a function of @xmath123 for distinct values of @xmath0 , including the classical limit ( @xmath6 ) of the hamiltonian ( [ hamiltonian ] ) and the relative percental error of this function for various values of @xmath0 in relation to the classical specific heat per site .
we take @xmath124 and @xmath125 .
( [ fig_2]a ) pictures the antiferromagnetic case ( @xmath126 ) and ( [ fig_2]c ) the ferromagnetic case ( @xmath127 ) .
( [ fig_2]b ) and ( [ fig_2]d ) show the relative percental error , i.e. , @xmath128 , of figs .
( [ fig_2]a ) and ( [ fig_2]c ) , respectively , where @xmath129 is the classical limit of the specific heat per site . in both cases
, even the @xmath14 model behaves classically ( within an error smaller than @xmath130 ) up to @xmath131 . within this range of error
, we can also neglect the quantum nature of the spin model with @xmath132 up to @xmath133 .
let @xmath134 and @xmath135 be the magnetic susceptibilities per site of the spin-@xmath18 model ( with unitary norm ) and of the spin-@xmath0 model ( with norm equal to @xmath30 ) , respectively .
the relation between the magnetic susceptibilty per site and the helmholtz free energy is @xmath136 . from the results ( [ b2 ] )
we obtain the relation between @xmath137 and @xmath138 , [ 6 ] @xmath139 equating the r.h.s .
( [ 6a ] ) and ( [ 6b ] ) we obtain that @xmath138 is a homogenous function of degree 1 for all values of temperature .
its @xmath3-expansion satisfies this property . fig .
[ fig_3 ] shows the magnetic susceptibility versus @xmath123 for various finite values of @xmath0 and the classical chain model ( @xmath6 ) .
( [ fig_3]b ) shows the relative percentage error of the magnetic susceptibilty with respect to the classical curve for @xmath140 and @xmath141 . in fig .
[ fig_3 ] we have @xmath142 , @xmath143 and @xmath144 . from fig .
( [ fig_3]b ) we verify that the magnetic susceptibility of the spin-3/2 can be approximated by the classical result up to @xmath145 with an error smaller than @xmath130 .
the classical magnetic susceptibility curve is a good approximation of the spin-2 model up to @xmath146 .
[ fig_4 ] shows the magnetization of the @xmath1 model of the spin with unitary norm versus @xmath147 ( @xmath148 ) .
we call @xmath149 the magnetization derived from hamiltonian ( [ hamiltonian ] ) and @xmath150 the magnetization derived from ( [ hamt1 ] ) . from eqs .
( [ b2 ] ) , we also obtain [ 7 ] @xmath151 from eqs .
( [ 7 ] ) we obtain that @xmath152 is a homogeneous function of degree 1 at all temperatures .
this condition is satisfied by the @xmath3-expansion of the magnetization , derived from eq .
( [ free_e ] ) . fig .
( [ fig_4]a ) shows the magnetization @xmath153 as a function of @xmath147 at @xmath154 .
we choose the same values of constants in the hamiltonian ( [ hamiltonian ] ) as in fig .
( [ fig_3 ] ) , @xmath142 and @xmath155 .
comparison of figs .
( [ fig_3]b ) and ( [ fig_4]b ) shows a closer similarity of quantum and classical magnetization curves ( even for low values of spin such as @xmath12 ) than that of magnetic susceptibility curves , as far as high temperatures are concerned .
the correlation function of spin @xmath2-components between first neighbors is written as @xmath156 , for a given spin @xmath18 . its percental variation ( when @xmath18 varies by half - integer steps ) , shown in fig .
( [ fig_5 ] ) , is defined as @xmath157 for @xmath158 , the function @xmath159 has an expansion in @xmath160 whose leading term is @xmath161 , and its coefficient depends on the temperature . in fig .
( [ fig_5 ] ) we take @xmath142 , @xmath143 and @xmath162 . in the very high temperature of @xmath163 , only the correlation functions for @xmath164 and @xmath165
differ from the classical curve by a difference smaller than @xmath166 .
as we lower the temperature , spins of higher values have correlation functions between first neighbors that differ slightly from the classical ones , but even at @xmath167 , the correlation function for @xmath14 approximates the classical result with an error smaller than @xmath168 .
although there are only seven terms in the high - temperature expansion of the helmholtz free energy of the spin-@xmath18 @xmath1 model ( see appendix ( [ b ] ) ) , different pad approximants permit us to enhance our analytic results and extend them to lower temperatures . in this section
, we shall assume a quantum spin chain with unitary norm .
bernu and misguich @xcite presented an approach to `` interpolate '' the high- and low - temperature behaviors of the specific heat per site , provided that the ground state energy per site of the chain is known .
we refer the reader to their work for details on the method .
the @xmath1 models ( with @xmath169 ) , at the isotropic points ( @xmath170 ) , are among the most studied ones .
for this reason , in this section we calculate the pad approximants at those points , for the sake of comparison of our results to the literature . at @xmath171
the behavior of the specific heat per site is distinct for gap and gapless models .
ferromagnetic and half - odd integer spin chains do not have excitation energy gaps and the specific heat per site around @xmath172 is @xmath173 , where @xmath174 and @xmath175 are integers@xcite .
by contrast , the integer antiferromagnetic spin chains have an energy gap between the ground state energy and the lowest exited state and this thermodynamic function , in the region of @xmath171 , is @xmath176@xcite .
as the spin increases , this energy gap decreases rapidly ; it only vanishes at the classical limit ( @xmath6 ) , though . in ref .
@xcite , yamamoto studied numerically the thermodynamics of the @xmath14 @xmath1 model with @xmath142 , @xmath8 and @xmath7 .
the ground state energy per site of this model for the non - unitary spin chain was obtained in ref .
@xcite ; the ground state energy per site of the unitary spin chain can then be easily obtained , being equal to @xmath177 .
this model exhibits an energy gap @xmath178 @xcite . at very low temperature
it is expected that the specific heat vanish exponentially ( @xmath179 ) as @xmath180 . on the other hand ,
the spin wave theory gives a linear behavior for the specific heat over a finite interval of intermediate low temperatures @xcite . in order to enhance our @xmath3-expansion of @xmath181 to lower temperatures
, we simply use @xmath182 .
the entropy at low temperatures behaves as @xmath183 , where @xmath184 is the ground state energy per site , and it has an essential singularity when @xmath185 . following ref .
@xcite , we construct an auxiliary function @xmath186 , which is analytic in the interval @xmath187 $ ] , and use the pad approximant to fit @xmath188 , so the entropy becomes @xmath189 . by @xmath190
we denote the pad approximant realized over @xmath188 .
the specific heat per site of the model is obtained using the relation @xmath191 , which can be plotted in parametric form @xmath192 ( for details see ref .
@xcite ) . in fig . ( [ fig_6 ] ) ( top )
we compare all `` acceptable '' pad approximants @xmath193 ( i.e. , all approximants so that @xmath194 and that do not possess spurious singularities ) and their average @xmath195 , to the numerical results obtained by the quantum monte carlo method ( qmc)@xcite and the @xmath3-expansion of @xmath196 .
( [ fig_6 ] ) ( bottom ) displays the percental deviation of each pad approximant @xmath197 with respect to @xmath195 .
we define @xmath198 .
the high temperature expansion satisfactorily describes @xmath181 up to @xmath199 .
the solid line corresponds to the percental deviation of @xmath195 with respect to the qmc prediction .
although we have only seven terms in expansion ( [ free_e ] ) of the helmholtz free energy , the percental relative difference of the pad approximants is smaller than @xmath200 for @xmath201 , allowing good precision in calculating the temperature where @xmath196 is maximum . figs .
( [ fig_7 ] ) and ( [ fig_9 ] ) show the specific heat for several values of spin for the antiferromagnetic ( @xmath142 ) and ferromagnetic ( @xmath202 ) cases , respectively , allowing us to observe how the specific heat per site approaches its classical limit as @xmath0 varies . in order to show how the pad approximants enhance the high temperature solution of the specific heat per site , fig .
( [ fig_7 ] ) also displays the @xmath3-expansion of each @xmath203 .
following the spin wave theory , we assume that the specific heat behaves as @xmath204 for the antiferromagnetic@xcite case , and as @xmath205 in the ferromagnetic@xcite case .
we fit the numerical values of the ground state energy per site calculated in ref .
@xcite , for the antiferromagnetic case with @xmath206 and @xmath207 , and extrapolate them to the integer antiferromagnetic spin @xmath0 chain .
our approximate result for this ground state energy per site is @xmath208 . for the ferromagnetic case at the heisenberg point ( @xmath170 ) with @xmath7 and @xmath8 , the ground state energy simply becomes @xmath209 .
( [ fig_7 ] ) shows the specific heat as a function of temperature for @xmath210 , with @xmath8 and @xmath7 .
the curves correspond to the @xmath195 of this thermodynamic function for the following set of spin values , as well as their respective @xmath3-expansions : @xmath211 .
the exact classical curve corresponds to fisher s solution@xcite . from our numerical results ,
we verify that even the pad approximants for @xmath212 deviates less than @xmath130 from the classical result ( @xmath46 ) up to @xmath213 , while for @xmath214 this deviation is about @xmath130 only at @xmath215 . except for the classical limit ( @xmath17 ) , to the best of our knowledge
there are no known results of the specific heat per site for @xmath216 for the whole range of temperature . in order to assure the validity of our results for the specific heat at lower temperatures , for distinct @xmath0 as presented in fig .
( [ fig_7 ] ) , we plotted in fig .
[ fig_8 ] the relative percental difference between the pad approximants without spurious poles and the average pad for several values of spin ( @xmath217 and @xmath218 ) .
for all those values of spin , it can be seen that deviations get larger than @xmath130 for @xmath219 , whereas their respective @xmath3-expansion are good approximations only up to @xmath220 . in fig .
( [ fig_7 ] ) we verify that the curves of the specific heat for the antiferromagnetic case are close to the classical curve up to low temperatures , even for @xmath212 .
comparing its behavior to that of the ferromagnetic case , we plot in fig .
( [ fig_9 ] ) the specific heat for several values of spin , taking @xmath221 , @xmath7 and @xmath222 .
we assume that the specific heat for the ferromagnetic case and small values of @xmath60 still behaves as @xmath223 and the ground state energy per site is @xmath224 .
the first term in the ground state energy corresponds to the case @xmath8 ; the second one is the contribution of the single - ion anisotropy to the ground state energy ; and the last term corresponds to its global shift . in order to simplify our calculations
, we impose that the average energy vanishes at @xmath225 ( otherwise it would be equal to @xmath226 , that is , the value obtained from our high temperature expansion ) . as in the antiferromagnetic case , we do not know of results that would allow comparison regarding the dependence of the specific heat on the temperature in the interval @xmath227 $ ] . in order to carry out a first check of the curves in picture [ fig_9 ]
, we proceed in a similar way as before when testing the antiferromagnetic curves .
( [ fig_10 ] ) shows the relative percental differences of the pad s approximants and the respective @xmath3-expansion of @xmath228 to the average pad for @xmath229 and @xmath230 .
we see that the high temperature series describe well the specific heat of the ferromagnetic chains for @xmath231 . from theses
plots we verify that the extension of the high - temperature results to the region of lower temperatures for the ferromagnetic case is worse than that of the antiferromagnetic case , even though integer - spin ferromagnetic materials are gapless and the behavior of their specific heat about @xmath172 is well described by the spin - wave approach .
it is also possible to write the magnetic susceptibility per site as a function of the ground state energy per site @xmath184 .
its low temperature behavior can be inferred from the extrapolation in a similar fashion as that of the specific heat per site .
unfortunately , convergence is not as much as satisfactory . by applying the method of ref .
@xcite to the magnetic susceptibility , a large number of singularities are found in the pad approximants to this thermodynamic function .
for this reason we use the dlog - pad approximant@xcite to extend the high temperature expansion of the antiferromagnetic susceptibility , incorporating the low - temperature information in a simple way .
we follow ref .
@xcite to calculate the dlog - pad approximant of the magnetic susceptibility per site @xmath232 ; we refer the reader to this article for further details on the method . from ref .
@xcite , we verify that the dlog - pad approximants are independent of the leading term coefficient of the magnetic susceptibility at low temperatures , if it vanishes at @xmath180 . on the other hand ,
if it is non - null at @xmath172 , we need to know the exact coefficient of the leading term of the magnetic susceptibility for @xmath171 in order to calculate the pad approximants at low temperatures .
this also occurs@xcite for @xmath79 . using the qmc method
, yamamoto@xcite obtained the magnetic susceptibility of the antiferromagnetic @xmath14 @xmath1 model ( @xmath233 ) in the whole interval of temperature of a periodic chain with 96 sites . in fig .
( [ fig_11 ] ) we compare the @xmath3-expansion of @xmath234 and its dlog - pad approximants for @xmath14 , with @xmath142 , @xmath8 and @xmath7 with yamamoto s numerical results .
our extension to lower temperatures assumes that the behavior of the magnetic susceptibility around @xmath172 be as @xmath235 with @xmath236 ; the latter value yields the best fit to the qmc results@xcite .
the bottom of fig .
( [ fig_11 ] ) shows the percental difference between the qmc and the average dlog - pad approximant to the magnetic susceptibility for @xmath14 ; it can be seen that @xmath237 up to @xmath238 . from fig .
[ fig_11 ] ( top ) we verify that our high - temperature expansion of @xmath239 fits well the numerical result only up to @xmath240 .
( [ fig_12 ] ) shows the deviation from the average @xmath195 of the dlog - pad approximants to the magnetic susceptibility , for the antiferromagnetic heisenberg model ( @xmath19 and @xmath8 ) for for @xmath165 and @xmath212 , showing the enhanced extention of our high - temperature results to lower temperatures . from ref .
@xcite we obtain the behavior of @xmath241 around @xmath172 ( @xmath242 ) ; for @xmath212 we assume @xmath243 .
the behavior of the magnetic susceptibility at low temperature for the ferromagnetic heisenberg model ( @xmath244 and @xmath8 ) was obtained by takahashi@xcite using the modified spin wave theory for arbitrary spin - s .
the magnetic susceptibility for the unitary spin chain for low temperatures , in units of @xmath66 , is @xmath245 where @xmath246 and @xmath247 is the riemann zeta function .
when the dlog - pad approximant method@xcite is applied , not even the coefficient of the leading term of eq.([chi_low ] ) is used , but only its power .
therefore , a better approximation will be achieved if we take into account all the three terms in ( [ chi_low ] ) ; we do so by applying the two - point pad approximant@xcite method .
we verify from expansion ( [ chi_low ] ) that @xmath248 is singular at @xmath172 .
expansion ( [ free_e ] ) gives the @xmath3-expansion of @xmath134 up to order @xmath4 .
the first terms of @xmath134 for @xmath244 , @xmath8 and @xmath7 are @xmath249 .
we define the auxiliary function @xmath250 .
this auxiliary function is regular at @xmath172 and @xmath251 ( @xmath252 ) ; its only problems are that it has non - integer powers of @xmath3 and it does not have a taylor expansion about @xmath172 . in order to circumvent this drawback , we proceed as in ref .
@xcite and apply the transformation @xmath253 where the parameter @xmath254 is mapped onto @xmath255 $ ] .
the expansions of @xmath256 , around @xmath172 @xmath257 and @xmath225 @xmath258 , are polynomials in @xmath259 and therefore we can write its two - point pad s approximants , connecting the expansions of @xmath260 at @xmath261 and @xmath262 .
( [ fig_13 ] ) shows the average pad of the magnetic susceptibility for the isotropic ferromagnetic @xmath165 model , in the absence of an external magnetic field , obtained from the two - point pad method and using the transformation ( [ transform ] ) .
for all pad approximants , the relative difference to @xmath195 is smaller than @xmath263 in the whole interval of temperature . finally , fig .
( [ fig_14 ] ) shows the relative percental difference of various pad approximants to the @xmath195 , in the case of the ferromagnetic isotropic chain model ( with @xmath8 ) and in the absence of an external magnetic field .
the difference increases with @xmath0 ; however , for @xmath214 the relative difference is smaller than @xmath130 up to @xmath264 .
in this report we present the high temperature expansion of the helmholtz free energy of the @xmath1 chain , in the thermodynamic limit , for an arbitrary spin @xmath0 up to order @xmath4 .
this expansion is analytic in @xmath0 and in @xmath3 and each coefficient of @xmath61 ( @xmath265 ) is exact .
the model includes anisotropy in the @xmath2-direction , single - ion anisotropy and an external magnetic field . in order to obtain the expansion of the helmholtz free energy ( [ free_e ] )
we apply the method developed in ref .
@xcite ; from this expansion we are able to obtain the thermodynamic quantities of the model for arbitrary spin values .
the expansion ( [ free_e ] ) recovers the @xmath3-expansion of the helmholtz free energy presented in ref .
@xcite for several values of spin , and it is an extension to the results of ref .
@xcite , where we include the anisotropy effects .
the series in @xmath0 allows us to obtain the classical limit of the @xmath1 model , but taking into account the quantum nature of the spins . for the particular case @xmath7 and @xmath8 we recover joyce s result for the classical limit of this model@xcite .
we explicitly show in expansion that only the presence of an external magnetic field distinguishes the classical ferromagnetic and antiferromagnetic phases in the regime of high temperatures .
we interpolate the @xmath3-expansion with its low temperature series using the pad approximant of to extrapolate the properties of the specific heat to finite temperatures using the method developed in ref .
only a few terms of the high temperature series are needed to yield good agreement : for low spin values the deviation of all non - singular pad approximants respect to @xmath195 fluctuates around 5% . for the antiferromagnetic magnetic susceptibility we use the dlog - pad@xcite approximants to investigate the finite temperature properties of the spin-@xmath18 quantum xxz chain . in order to plot the magnetic susceptibility @xmath248 as a function of temperature for different spin values
we take the average of all non - singular dlog - pad approximants , since their deviation from the corresponding @xmath195 is less than 2% .
the ferromagnetic case is studied by applying two - point like pad approximants , since we have the first three terms of the low temperature series expansion@xcite .
the magnetic susceptibility is plotted for different spins , in the same way as for the previous case . obviously , if we knew more low temperature information about any physical quantities , it could be also interpolated with our high temperature expansions , by adapting different approximants available in the literature .
our main limitation regarding the study of finite temperature behavior of other parameters of the hamiltonian is the lack of related low temperature information .
o. r. thanks fapemig for financial support .
s.m . de s. and o.r . thank fapemig and cnpq for partial financial support .
thank s cnpq and faperj for partial financial support .
we have that @xmath266 if @xmath267 is odd . for the purposes of this appendix , @xmath267 is taken as an even integer . in order to calculate
@xmath268 we use the result of ref.@xcite that permits us to write this trace as a function of the spin @xmath0 , @xmath269 where @xmath270 , @xmath271 , so that @xmath272 s are the coefficients of bernoulli s polynomial @xmath273 .
they are defined by@xcite @xmath274 on the other hand , we prefer to rewrite eq . as a function of the square of the norm of the spin @xmath82 ( that is , @xmath81 ) , which is a constant of motion , @xmath275 we remind that @xmath276 is equal to the dimension of the hilbert space at each site , that is , @xmath277 , @xmath278 . comparing eqs . and , and after some algebraic manipulation
, we obtain each coefficient @xmath279 of eq . as a combination of the bernoulli numbers@xcite @xmath280 , @xmath281 in eq.([hamiltonian ] ) we rewrite hamiltonian ( [ hamt1 ] ) in terms of the rescaled spin operator @xmath282 . to obtain the results ( [ b2])-([b8 ] ) we need to calculate the powers of the operator @xmath283 . from the results - we are able to write the normalized traces of @xmath283 , that is @xmath284 just as a polynomial of @xmath285 of degree @xmath286 , which reads @xmath287 in terms of the normalized matrix representation ,
the commutation relation becomes @xmath288=\frac{i s^z}{\sqrt{s(s+1 ) } } = i \sqrt{y } \
s^z , \end{aligned}\ ] ] which is useful to evaluate the normalized traces .
we call @xmath114 the helmholtz free energy of the hamiltonian ( [ hamiltonian ] ) , that drives the dynamics of the quantum spin chain with unitary norm
. let @xmath289 be the helmholtz free energy of the hamiltonian where the norm of the spin-@xmath82 is @xmath30 and @xmath270 . in ref .
@xcite we calculated the high temperature expansion of the helmholtz free energy of the hamiltonian ( [ b1 ] ) up to @xmath23 .
the relation between the free energies derived from hamiltonians ( [ hamiltonian ] ) and ( [ b1 ] ) is in this appendix we present the high temperature expansion of the helmholtz free energy @xmath114 , up to order @xmath293 , for arbitrary value of @xmath294 .
this expansion can be written in terms of @xmath285 , that is , | the thermodynamics of the spin-@xmath0 anisotropic quantum @xmath1 chain with arbitrary value of @xmath0 and unitary norm , in the high - temperature regime , is reported . the single - ion anisotropy term and the interaction with an external magnetic field in the @xmath2-direction
are taken into account .
we obtain , for arbitrary value of @xmath0 , the @xmath3-expansion of the helmholtz free energy of the model up to order @xmath4 and show that it actually depends on @xmath5 .
its classical limit is obtained by simply taking @xmath6 . at @xmath7 and @xmath8 ,
our high temperature expansion of the classical model coincides with joyce s exact solution@xcite .
we study , in the high temperature region , some thermodynamic quantities such as the specific heat and the magnetic susceptibility as functions of spin and verify for which values of @xmath0 those thermodynamic functions behave classically .
their finite temperature behavior is inferred from interpolation of their high- and low - temperature behavior , and shown to be in good agreement with numerical results .
the finite temperature behavior is shown for higher values of spin . |
despite decades of continuing effort , the nature of the progenitor systems of type ia supernovae ( sne ) remains unknown .
the most widely accepted theoretical scenarios involve the thermonuclear explosion of a c+o white dwarf ( wd ) destabilized by accretion of material from a binary companion , either another wd ( double degenerate systems , dds ) or a normal star ( single degenerate systems , sds ) .
observations of supernova rates in distant galaxies suggest at least two different ways to produce type ia sne @xcite : a ` prompt ' channel associated with young stellar populations and a ` delayed ' channel associated with old stellar populations . at present , it is not clear what relationship , if any , these channels have with the theoretical scenarios , or even if the proposed scenarios can yield type ia sne at the observed rate @xcite .
most attempts to constrain the fundamental properties of the progenitors have been unsuccessful , including direct identification in optical pre - explosion images @xcite , searches for h stripped from the binary companion in sn spectra @xcite , and searches for prompt emission from circumstellar interaction @xcite .
very recently , the possible identification of the progenitor in a pre - explosion x - ray image of sn 2007on by @xcite has been questioned by post - explosion images @xcite .
campaigns to look for the surviving companion star to the exploded wd in the nearby tycho supernova remnant ( snr ) , known to be of type ia origin , have also produced controversial results @xcite .
this lack of clues to the identity of the progenitors in the observations of type ia sne is sometimes referred to as ` stellar amnesia ' .
one of the most important constraints on the age and nature of the progenitor systems is their metallicity @xmath3 . during stellar evolution , the c , n , and o that act as catalysts for the cno cycle pile up into @xmath4n , which is converted to @xmath5ne in the hydrostatic he burning phase through the reactions @xmath4n(@xmath0,@xmath6)@xmath7f(@xmath8,@xmath9)@xmath7o(@xmath0,@xmath6)@xmath5ne .
the @xmath8 decay of @xmath7f increases the neutron excess of the wd material , defined as @xmath10 ( where @xmath11 is the atomic number and @xmath12 is the mass number ) , resulting in a linear scaling of @xmath13 with metallicity : @xmath14 @xcite .
this can have an observable impact on type ia supernova spectra @xcite , but the predicted effects are hard to detect in practice ( in one case , sn 2005bl , a progenitor of subsolar metallicity has been suggested by * ? ? ?
* ) . in this _
letter _ , we describe a new direct method to measure the metallicity of type ia sn progenitors using mn and cr lines from shocked ejecta in the x - ray spectra of snrs .
nuclear burning in type ia sne happens in different regimes depending on the peak temperature reached by the wd material . from higher to lower temperatures , the regimes are : nuclear statistical equilibrium ( nse ) , incomplete si burning , incomplete o burning and incomplete c - ne burning @xcite . during the explosion itself , electron captures are too slow to change the original value of @xmath13 except in the innermost @xmath16 of ejecta , where nuclear burning happens at higher densities , leading to neutron - rich nse @xcite . outside this region
, an imbalance between free neutrons and protons does not affect the production of the most abundant species , with the possible exception of @xmath17ni at supersolar metallicities @xcite . however , some trace nuclei with unequal numbers of protons and neutrons are very efficient at storing the neutron excess .
the most abundant among these nuclei is @xmath18mn , which is produced during incomplete si burning as @xmath18co .
when normalized to another product of incomplete si burning whose nucleosynthesis is insensitive to the neutron excess ( cr is the ideal choice ) , the yield of mn becomes an excellent tracer of the progenitor metallicity . ) in wds with main sequence progenitor masses of @xmath19 and @xmath20 , respectively ( see [ sub : simmering ] ) .
the horizontal solid and dotted lines represent the estimated @xmath15 ratio in the tycho snr from the _ suzaku _
observations ( @xmath21 , shaded in gray ) , which corresponds to @xmath1 ( see [ sec : zoftycho ] ) . [ fig-1 ] ] this diagnostic can be quantified with type ia sn explosion models .
we have used the model grid from @xcite , complemented with other models calculated with the code described in @xcite .
physical inputs and initial conditions are as in @xcite , except in the cases noted below .
we consider examples of several explosion mechanisms : deflagrations , delayed detonations , and pulsating delayed detonations .
for the present work , we have recalculated the nucleosynthesis of four delayed detonation models ( ddta , ddtc , ddte , and ddtf ) at different metallicities by altering the value of @xmath22 in the pre - explosion wd from the canonical @xmath23 ( corresponding to @xmath24 ) .
we have also calculated one deflagration model starting from a central density twice as large as the other models , and one delayed detonation model with the c to o mass ratio @xmath25 in the wd instead of the usual 1 . in spite of these fundamental differences in explosion mechanisms and initial conditions , we find a very tight correlation between the mn to cr mass ratio outside the neutron - rich nse region in the sn ejecta , @xmath15 , and the progenitor metallicity @xmath3 ( see figure [ fig-1 ] ) .
a power - law fit yields the following relation ( with a correlation coefficient @xmath26 ) : @xmath27 removal of the inner @xmath28 of neutron - rich nse material from this relation is justified because ( 1 ) there is observational evidence that this material does not mix outwards during the explosion @xcite or its aftermath @xcite ; and ( 2 ) the dynamical ages of most ejecta - dominated type ia snrs are too small for the reverse shock to have reached so deep into the ejecta . for illustration ,
we have plotted in figure [ fig-1 ] the core collapse sn models from @xcite , which also show some correlation between @xmath15 and @xmath3 , albeit with a much larger spread and a much shallower slope . in order for eq .
[ eq-1 ] to hold , the value of @xmath13 set by the progenitor metallicity must remain unchanged between the formation of the wd and the sn explosion .
this should be true for dd progenitors if the final merger and runaway happen on dynamical timescales . in slowly accreting wds
, @xmath13 can be modified through electron captures during the so - called ` simmering ' phase of non - explosive c burning that takes place in the @xmath29 yr prior to the explosion @xcite . the impact of this additional neutronization on the @xmath15 ratio will depend on the extent of the convective region over which the neutronized material is mixed ( @xmath30 ) , and its overlap with the portion of the wd that will undergo incomplete si burning where mn and cr are synthesized . during the simmering phase ( left axis , black plots ) and its overlap with the si rich region of ejecta ( right axis , colored plots ) as a function of @xmath3 .
the black plots represent quadratic interpolations for different @xmath31 : @xmath19 ( dashed ) , @xmath32 ( dotted ) , and @xmath20 ( solid ) , with the symbols indicating the values from @xcite ( for @xmath33 at @xmath34 , we have taken the average between @xmath35 and @xmath36 ) .
the colored plots represent the overlap ( in @xmath37 ) between @xmath30 and the si - rich region of ejecta , for very subluminous ( @xmath38 , red ) , mildly subluminous ( @xmath39 , orange ) , and normal , but faint ( @xmath40 , blue ) type ia sne . [ fig-2 ] ] some previous studies @xcite have found large values for @xmath30 by assuming an homogeneous chemical composition for the wd , effectively applying the schwarzschild criterion for convection .
we will adopt the hypothesis that the convective region is limited by the ledoux criterion to the c - depleted core of the wd created during hydrostatic he - shell burning , @xmath41 @xcite . in the absence of a self - consistent picture for convection inside wds
, this must remain an open issue ( see * ? ? ? * for a discussion ) , but we believe that our hypothesis is reasonable given the existence of other processes that can reduce the extent of @xmath30 and limit the mixing of neutronized material .
these include the presence of urca shells @xcite and the long thermal diffusion time scale for buoyant bubbles larger than @xmath42 m @xcite , which prevents them from mixing completely with their surroundings during convection . line .
the k@xmath0 lines of cr and mn are detected at @xmath43 and @xmath44 confidence levels , with fluxes of @xmath45 and @xmath46 respectively @xcite .
[ fig-3 ] ] according to @xcite , si - rich ejecta in type ia sne lie between lagrangian mass coordinates @xmath47 and @xmath48 , where @xmath49 is the light curve width parameter ( defined as the decline in blue magnitude 15 days after maximum ) , which acts as a proxy for sn brightness . in this context , the extent of the overlap between @xmath30 and the si - rich ejecta depends on the main sequence mass @xmath31 and initial metallicity of the wd progenitor ( which determine @xmath41 , * ? ? ?
* ) and the sn brightness ( see figure [ fig-2 ] ) .
this overlap is only significant for subluminous ( @xmath50 ) type ia sne originated by progenitors with either large @xmath31 or low @xmath3 , or both .
@xcite find an upper limit for the increase of @xmath13 during the simmering phase of @xmath51 , which is comparable to the value of @xmath13 in solar material .
the impact of simmering on the @xmath15 ratio can then be estimated by mixing material with @xmath52 ( appropriate for the value of @xmath53 derived by * ? ? ?
* ) into the incomplete si burning region , in a proportion equivalent to the extent of the overlap shown in figure [ fig-2 ] .
the green and orange plots in figure [ fig-1 ] are two examples of such ` simmering - modified ' models for very subluminous
( @xmath54 ) type ia sne , illustrating our conclusion that c simmering will only modify the @xmath15 ratio in the sn ejecta for very subluminous sne , and then only in cases where @xmath31 is large , or @xmath3 is low , or both .
the work in this _
letter _ is motivated by the recent _ suzaku _ detection of mn and cr in the x - ray spectrum of the tycho sn reported by @xcite ( see figure [ fig-3 ] ) .
this observational result opens the possibility of studying the @xmath15 ratio in type ia sn ejecta , which can not be done using optical sn spectra due to the 2.7 yr half - life of @xmath18fe in the decay chain @xmath18co@xmath55fe@xmath55mn .
since mn and cr are synthesized together in the explosion and have very similar electronic structures , it is possible to estimate their mass ratio from the line flux ratio : @xmath56 , where @xmath57 is the ratio of atomic masses , @xmath58 is the line flux ratio , and @xmath59 is the ratio of specific emissivities per ion
. emissivity ratio as a function of @xmath60 and @xmath61 .
the dotted box encompasses the values of @xmath60 and @xmath61 found in the si - rich ejecta of six young type ia snrs by @xcite : @xmath62 ; @xmath63 .
the striped area corresponds to the region of parameter space appropriate for the tycho snr , @xmath64 ; @xmath65 .
[ fig-4 ] ] public data bases for x - ray astronomy do not usually include lines from trace elements like mn and cr , but the value of @xmath59 can be estimated by interpolation along the atomic number sequence from elements with available data @xcite .
we have used the atomdb data base @xcite to retrieve k@xmath0 line emissivities for si , s , ar , ca , fe , and ni ( @xmath66 ) as a function of ionization timescale @xmath60 and electron temperature @xmath61 , the two variables that control the line emission of a plasma in nonequilibrium ionization .
then we performed a spline interpolation to obtain @xmath59 , which we plot in figure [ fig-4 ] , together with the region of the @xmath67 parameter space that is populated by young type ia snrs ( see the appendix in * ? ? ?
the uncertainty in the value of @xmath59 comes from the variation within this region and the error introduced by the interpolation itself , which we have estimated by comparing the true and interpolated values for the @xmath68 ratio . for the larger region of the @xmath67 plane
plotted in figure [ fig-4 ] , we find @xmath69 ( @xmath70 for the smaller region appropriate for tycho ) .
from the _ suzaku _ observation , @xmath71 , which gives @xmath72 .
this mass ratio translates into a metallicity of @xmath73 for the progenitor of the tycho snr ( @xmath1 ) .
the large error bar in this result is dominated by the statistical uncertainties in the x - ray fluxes from the faint mn and cr lines , and it will be reduced in an upcoming , approved deeper _ suzaku_observation . at present , we find a strong indication for a supersolar metallicity ( @xmath74 with the solar value from @xcite ; @xmath2 with the newer value from @xcite ) .
our results are compatible with a solar metallicity , but subsolar values can be discarded with confidence . given this measurement and the fact that tycho s sn was probably either normal or slightly overluminous @xcite
, we do not have to concern ourselves with the impact of neutronization during c simmering in this particular case .
the tycho snr is @xmath75 above the galactic plane at a galactocentric radius of @xmath76 ( assuming a distance of @xmath77 , * ? ? ?
a supersolar metallicity is higher than average for this location , but well within the spread of measured @xmath78}$ ] values @xcite .
this combination of galactrocentric radius and metallicity suggests a young progenitor age ( a few gyr or less ) , which would make tycho a candidate for the ` prompt ' channel of type ia sne , but the scatter in the age - metallicity relations and the uncertainties in our measurement are too large to be more specific on this point . in this _
, we have proposed a new method to measure the metallicity of type ia sn progenitors using mn and cr lines in the x - ray spectra of their snrs .
we have applied it to the tycho snr and obtained a strong indication that its progenitor had a solar or supersolar metallicity .
the main strength of our method is its simplicity : it is based on well - known nuclear physics and observable parameters that are easy to measure .
detection of mn and cr lines in other young type ia snrs in the galaxy should be possible with _ suzaku _ , _ chandra _ , and _ xmm - newton _ observations .
a few additional snrs in the magellanic clouds might be reached with deep exposures , bringing the total of potential targets to perhaps a dozen .
this number would increase significantly with the inclusion of snrs in nearby galaxies like m31 and m33 , which may be accessible to next generation x - ray observatories like _ constellation - x _ and _ xeus_. even with a small sample of objects , we should be able to use this method to verify and complement the indirect metallicity studies of extragalactic type ia sne @xcite and test claims about the metallicity dependence of the type ia sn rate @xcite .
measurements or upper limits below a certain threshold ( @xmath79 ) would also provide interesting constraints on the extent of the convective region in accreting wds .
we conclude by noting that mn and cr lines have already been detected by _
asca _ in the galactic snr w49b @xcite .
the measured line flux ratio also suggests a solar or supersolar progenitor metallicity , but both the age and the sn type of w49b are controversial @xcite .
we defer a detailed discussion of this object and the application of our method to other galactic snrs to a forthcoming publication . | we propose a new method to measure the metallicity of type ia supernova progenitors using mn and cr lines in the x - ray spectra of young supernova remnants .
we show that the mn to cr mass ratio in type ia supernova ejecta is tightly correlated with the initial metallicity of the progenitor , as determined by the neutron excess of the white dwarf material before thermonuclear runaway .
we use this correlation , together with the flux of the cr and mn k@xmath0 x - ray lines in the tycho supernova remnant recently detected by _ suzaku _ @xcite to derive a metallicity of @xmath1 for the progenitor of this supernova , which corresponds to @xmath2 according to the latest determination of the solar metallicity by @xcite .
the uncertainty in the measurement is large , but metallicities much smaller than the solar value can be confidently discarded .
we discuss the implications of this result for future research on type ia supernova progenitors . |
every real lie group @xmath13 carries a distinguished left - invariant torsionfree connection @xmath14 , defined by @xmath15/2\,$ ] for all left - invariant vector fields @xmath16 and @xmath17 . in view of the jacobi identity ,
the curvature tensor of @xmath18 is @xmath19-parallel , and hence so is the ricci tensor of @xmath14 , equal to a nonzero multiple of the killing form @xmath20 .
our convention about @xmath21 reads @xmath22 ^ 2\hskip12pt\mathrm{for\ any } \hskip6ptx\hskip6pt\mathrm{in\ the\ lie\ algebra}\hskip6pt{\mathfrak{g}}\hskip6pt \mathrm{of\ } \,{\hskip.7pt}{{g}}{\hskip.4pt}.\ ] ] thus , if @xmath13 is semisimple , @xmath21 constitutes a bi - invariant , locally symmetric , non - ricci - flat pseuriemannian einstein metric on @xmath23 , with the levi - civita connection @xmath14 .
we denote by @xmath24^{\odot2}{\hskip-1.5pt}\to[{\mathfrak{g}}{\hskip-.7pt}^*]^{\odot2}$ ] a specific multiple of the curvature operator of the metric @xmath20 , acting on symmetric symmetric bilinear forms @xmath25 , so that , whenever @xmath26 , @xmath27({x},{y})\ , = \,2\,\mathrm{tr}\hskip1.6pt[(\mathrm{ad}\,{x}){\hskip.4pt}(\mathrm{ad}\,{y}){\hskip.4pt}\sigma]{\hskip.4pt } , \hskip15pt\mathrm{for}\hskip5pt\sigma:{\mathfrak{g}}\to{\mathfrak{g}}\hskip5pt\mathrm{with}\\ \mathrm{b)}\hskip5pt{\sigma}({x},{y})\,=\,{\beta}(\sigma { x},{y}){\hskip.7pt}. \end{array}\ ] ] see remark [ crvop ] . the same formula ( [ oms ] ) defines the operator @xmath28 in a _ complex _
semisimple lie group @xmath23 , acting on symmetric complex - bilinear forms @xmath29 .
we then identify @xmath28 with the analogous curvature operator for the ( @xmath30-bilinear ) killing form @xmath20 , treating the latter as a holomorphic einstein metric on the underlying complex manifold of @xmath23 .
the structure of @xmath28 in complex simple lie groups is known from the work of meyberg @xcite , who showed that @xmath28 is diagonalizable and described its spectrum . for the reader
s convenience , we reproduce meyberg s theorem in an appendix .
his result easily leads to a similar description of the spectrum of @xmath28 in real simple lie algebras @xmath31 , which we state as theorem [ rspec ] and derive in section [ so ] from the fact that , given any such @xmath31 , @xmath32 see ( * ? ? ?
* lemma 4 on p. 173 ) .
the .4ptalgebra isomorphism types of real simple lie algebras @xmath33 thus form two disjoint classes , characterized by ( [ rcp].a ) and ( [ rcp].b ) . for both real and complex semisimple lie groups @xmath23 ,
studying @xmath28 can be further motivated as follows .
let ` metrics ' on @xmath13 be , by definition , pseuriemannian or , respectively , holomorphic , and @xmath34 denote the set of levi - civita connections of left - invariant einstein metrics on @xmath23 . then ,
as shown in ( * ? ? ?
* remark 12.3 ) , whenever a semisimple lie group @xmath13 has the property that @xmath35 is _ not _ an eigenvalue of @xmath36 , the levi - civita connection @xmath18 of its killing form @xmath21 is an isolated point of @xmath37 .
the converse implication holds except when @xmath13 is locally isomorphic to @xmath38 , with @xmath39
. see ( * ? ? ?
* theorems 22.2 and 22.3 ) . in a real.7pt@xmath40complex lie algebra @xmath31 , we define @xmath41^{\odot2}{\hskip-1.5pt}\to[{\mathfrak{g}}{\hskip-.7pt}^*]^{\wedge4}$ ] by @xmath42,[{z},{z}{\hskip.4pt}'{\hskip.4pt } ] ) + { \sigma}([{y},{z}],[{x},{z}{\hskip.4pt}'{\hskip.4pt}])+{\sigma}([{z},{x}],[{y},{z}{\hskip.4pt}'{\hskip.4pt}]){\hskip.4pt}.\ ] ] thus , @xmath43 is a real.7pt@xmath40complex - linear operator , sending symmetric bilinear forms @xmath44 on @xmath33 to exterior @xmath45-forms on @xmath31 . for the killing form @xmath21 one
has @xmath46,[{z},{z}{\hskip.4pt}'{\hskip.4pt}])={\beta}([[{x},{y}],{z}],{z}{\hskip.4pt}'{\hskip.4pt})$ ] , as @xmath47 is @xmath20-skew - adjoint .
furthermore , by the jacobi identity and ( [ bvw ] ) ( [ oms].a ) , @xmath48 if , in addition , @xmath33 is semisimple , there is also the operator @xmath49^{\otimes4}{\hskip-1.5pt}\to[{\mathfrak{g}}{\hskip-.7pt}^*]^{\otimes2}$ ] with @xmath50,\,\cdot\ , ) { \hskip.7pt}\otimes{\hskip.7pt}{\beta}([y , y{\hskip.4pt}'{\hskip.4pt}],\,\cdot\,){\hskip.4pt},\ ] ] for @xmath51 , where @xmath52 are characterized by @xmath53 . formula ( [ osj ] ) below shows that @xmath54^{\wedge4})\subset[{\mathfrak{g}}{\hskip-.7pt}^*]^{\odot2}{\hskip-1.5pt}$ ] .
our first main result , established in section [ cd ] , relates @xmath28 to @xmath55^{\odot2}{\hskip-1.5pt}\to[{\mathfrak{g}}{\hskip-.7pt}^*]^{\odot2}{\hskip-1.5pt}$ ] , the composite of @xmath43 and the restriction of @xmath56 to the subspace @xmath57^{\wedge4}{\hskip-.7pt}\subset[{\mathfrak{g}}{\hskip-.7pt}^*]^{\otimes4}{\hskip-1.5pt}$ ] .
[ hstrh ] let @xmath58 and @xmath56 be the operators defined by ( [ oms ] ) , ( [ lgg ] ) and ( [ pxp ] ) 1.5ptfor a given semisimple real@xmath59complex lie algebra @xmath60 .
then @xmath61 . next , in section [ pc ]
, we use meyberg s result and theorem [ hstrh ] to obtain the following description of @xmath62 for semisimple lie algebras @xmath31 .
it provides a crucial step in our proof of theorem [ autcf ] ( see below ) .
[ solut ] given a real@xmath59complex semisimple lie algebra @xmath33 with a direct - sum decomposition @xmath63 into simple ideals , @xmath64 , let @xmath43 and @xmath65 denote the operator defined by ( [ lgg ] ) for @xmath33 and , respectively , its analog for the @xmath66th summand @xmath67 . 1 .
@xmath68 , where @xmath69^{\odot2}{\hskip-1.5pt}\subset[{\mathfrak{g}}^{{\hskip-.7pt}*}]^{\odot2}$ ] via trivial extensions .
2 . @xmath70 if @xmath71 .
3 . @xmath72 if @xmath33 is simple and @xmath73 , which happens only when @xmath33 is real and isomorphic to the underlying real lie algebra of @xmath74 , while @xmath62 then consists of the real parts of all symmetric @xmath75-bilinear functions @xmath76 .
@xmath77 whenever @xmath33 is simple and @xmath78 , while @xmath62 is then spanned either by the killing form @xmath20 , or by @xmath79 and @xmath80 .
the former case occurs if @xmath33 is complex , or real of type ( [ rcp].a ) , the latter if @xmath33 is real of type ( [ rcp].b ) , with @xmath81 denoting the killing form of a complex simple lie algebra @xmath82 in ( [ rcp].b ) .
finally , one defines the _ cartan _ _ three - form _
@xmath83^{\wedge3}$ ] of a lie algebra @xmath33 by @xmath84,\,\cdot\,){\hskip.4pt},\hskip12pt\mathrm{where } \hskip7pt{\beta}\hskip7pt\mathrm{denotes\ the\ kil\-ling\ form.}\ ] ] the following result has been known for decades , although no published proof of it seems to exist @xcite . by an _ isomorphism of the cartan _ _ three - forms _ we mean here a vector - space isomorphism of the lie algebras in question , sending one form onto the other .
[ autcf]let @xmath33 be a real@xmath59complex semisimple lie algebra with a fixed direct - sum decomposition into simple ideals , which we briefly refer to as the `` summands '' of @xmath85 . 1 .
if @xmath82 is a real@xmath59complex lie algebra , the cartan forms of @xmath33 and @xmath82 are isomorphic and , in the real case , @xmath33 has no summands of dimension @xmath86 , then @xmath82 is isomorphic to @xmath85 .
if @xmath33 contains no summands of dimension @xmath87 or @xmath88 , then every automorphism of the cartan form of @xmath33 is a lie-.4ptalgebra automorphism of @xmath89 followed by an operator that acts on each summand as the multiplication by a cubic root of @xmath90 .
if @xmath33 is the underlying real lie algebra of a complex simple lie algebra and @xmath91 , then every automorphism of the cartan form of @xmath33 is complex - linear or antilinear .
conversely , if @xmath33 has @xmath92 summands of dimension @xmath87 and @xmath93 summands of dimension @xmath94 , then the lie-.4ptalgebra automorphisms of @xmath89 form a subgroup of codimension @xmath95 in the automorphism group of the cartan form .
we derive theorem [ autcf ] from theorem [ solut ] , in section [ pa ] .
suppose that @xmath33 is the underlying real lie algebra of a complex lie algebra @xmath96 .
we denote by @xmath21 and @xmath97 the killing form and cartan form of @xmath31 , cf .
( [ bvw ] ) and ( [ oeb ] ) , by @xmath43 the operator in ( [ lgg ] ) associated with @xmath31 , and use the symbols @xmath98 for their counterparts corresponding to @xmath96 . obviously , whenever @xmath99 is a symmetric @xmath100-bilinear form , @xmath101 for ( [ lrs].i ) , see also ( * ? ? ? * formula ( 13.1 ) ) . with @xmath33 and @xmath82 as above , it is clear from ( [ lrs].i ) that @xmath102 furthermore , ( [ lrs].i ) also implies , for dimensional reasons , that @xmath103 as the former space obviously contains the latter .
let @xmath33 now be a lie algebra over the scalar field @xmath104 or @xmath105 . a fixed basis of @xmath31 allows us to represent elements @xmath106 of @xmath31 , symmetric bilinear forms @xmath44 on @xmath31 , and the -algebra bracket operation @xmath107\,$ ] by their components @xmath108 , @xmath109 and @xmath110 ( the structure constants of @xmath31 ) , so that @xmath111 and @xmath112^k{\hskip-.7pt}=c_{ij}{^{\phantom i}}{}^kx^{{\hskip.7pt}i}y^{{\hskip.7pt}j}$ ] .
repeated indices are summed over .
the cartan form @xmath97 with ( [ oeb ] ) has the components @xmath113 , where @xmath21 is the killing form .
the definition ( [ bvw ] ) of @xmath20 , its bi - invariance , and the jacobi identity now read @xmath114 in the remainder of this section @xmath33 is also assumed to be semisimple .
we can thus lower and raise indices using the components @xmath115 of the killing form @xmath21 and @xmath116 of its reciprocal : @xmath117 , and @xmath118 . for any @xmath119
, one has @xmath120={c}({x},{y},{z})$ ] , where @xmath97 is the cartan form given by ( [ oeb ] ) .
equivalently , @xmath121 in fact , by successively using the equalities @xmath122 and @xmath123 ( both due to ( [ bij].ii ) ) , then again ( [ bij].ii ) , ( [ bij].iii ) , and ( [ bij].i.4pt.4ptii ) , we get @xmath124 . lowering the index @xmath125
, we obtain ( [ xyz ] ) .
next , we introduce the linear operator @xmath126^{\otimes2}{\hskip-1.5pt}\to[{\mathfrak{g}}{\hskip-.7pt}^*]^{\otimes2}\hskip6pt\mathrm{with } \hskip5pt({{t}}{\hskip-1.5pt}{\sigma})_{ij}{^{\phantom i}}={{t}}_{ij}^{{\hskip.7pt}kl}{\sigma}_{kl}{^{\phantom i}}{\hskip.4pt},\hskip4pt\mathrm{where } \hskip5pt{{t}}_{ij}^{{\hskip.7pt}kl}=2{\hskip.4pt}c_{ip}{^{\phantom i}}{}^kc_{\!j}{^{\phantom i}}{}^{lp}.\ ] ] [ tndom]for @xmath127 and the operator @xmath24^{\odot2}{\hskip-1.5pt}\to[{\mathfrak{g}}{\hskip-.7pt}^*]^{\odot2}$ ] given by ( [ oms ] ) , 1 .
@xmath128 leaves the subspaces @xmath57^{\odot2}$ ] and @xmath57^{\wedge2}$ ] invariant , 2 .
@xmath129 coincides with the restriction of @xmath127 to @xmath57^{\odot2}{\hskip-1.5pt}$ ] .
our claim is obvious from ( [ zee ] ) and the fact that , by ( [ zee ] ) , @xmath130 is the same as @xmath131 in ( [ oms ] ) , except that now @xmath132 need not be symmetric .
[ eigen]for any complex simple lie algebra @xmath33 and @xmath24^{\odot2}{\hskip-1.5pt}\to[{\mathfrak{g}}{\hskip-.7pt}^*]^{\odot2}$ ] with ( [ oms ] ) , 1 .
@xmath129 is diagonalizable , 2 .
@xmath133 is an eigenvalue of @xmath28 with multiplicity @xmath90 , 3 .
@xmath134 is not an eigenvalue of @xmath135 , 4 .
@xmath129 has the eigenvalue @xmath35 if and only if @xmath33 is isomorphic to @xmath136 for some @xmath137 .
this is a special case of meyberg s theorem , stated in the appendix .
[ clssf ] the isomorphism types of all complex simple lie algebras are : @xmath138 , for @xmath139 , @xmath140 ( even @xmath141 ) , @xmath142 with @xmath143 , as well as @xmath144 and @xmath145 . see ( * ? ? ? * and 77 ) .
[ crvop ] the curvature operator of a ( pseudo)riemannian metric @xmath146 on a manifold , acting on symmetric @xmath147-tensors , has been studied by various authors @xcite , @xcite , ( * ? ? ?
* pp . 5152 ) .
it is given by @xmath148 , where @xmath149 in terms of components relative to a basis of the tangent space at any point , the sign convention about the curvature tensor @xmath150 being that a euclidean tangent plane with an orthonormal basis @xmath151 has the sectional curvature @xmath152 .
when @xmath146 is the killing form @xmath21 of a semisimple lie group @xmath23 , treated as a left - invariant metric ( see the lines following ( [ bvw ] ) ) , this operator equals @xmath153 , for @xmath36 with ( [ oms ] ) .
in fact , the description of the levi - civita connection @xmath18 of @xmath21 in the introduction gives @xmath154,{z}]\,$ ] for left - invariant vector fields @xmath155 , that is , @xmath156 .
lemma [ tndom](b ) now implies our claim , as @xmath157 due to ( [ bij].ii ) and ( [ zee ] ) .
we use the component notation of section [ pr ] . according to ( [ lgg ] ) and ( [ pxp ] ) , @xmath158^{\odot2}\hskip4pt\mathrm{and } \hskip5pt{\zeta}\in[{\mathfrak{g}}{\hskip-.7pt}^*]^{\wedge4}{\hskip-.7pt}. \end{array}\ ] ] in any real.7pt@xmath40complex
semisimple lie algebra @xmath31 , for @xmath159 as in section [ pr ] , @xmath160 in fact , the first of the three terms naturally arising on the left - hand side of ( [ ccc ] ) equals @xmath161 since , by ( [ bij].i.4pt.4ptii ) , @xmath162 and @xmath163 .
the other two terms coincide ( as skew - symmetry of @xmath164 in @xmath165 gives @xmath166 ) , and so they add up to @xmath167 , that is , @xmath168 ; the rightmost equality is due to the jacobi identity ( [ bij].iii ) .
the last expression consists of the first term , @xmath169 , cf .
( [ xyz ] ) , equal , by ( [ bij].ii ) and ( [ zee ] ) , to @xmath170 , and the second term , @xmath171 , the two parenthesized factors of which are , for the same reasons , nothing else than @xmath172 and @xmath173 .
this proves ( [ ccc ] ) .
theorem [ hstrh ] is now an obvious consequence of ( [ osj ] ) ( [ ccc ] ) and lemma [ tndom](b ) .
[ rspec]let @xmath28 denote the operator with ( [ oms ] ) corresponding to a fixed real simple lie algebra @xmath60 , and @xmath174 its analog for @xmath175 , chosen so that @xmath33 and @xmath82 satisfy ( [ rcp ] ) . 1 .
@xmath129 is always diagonalizable .
2 . in case
( [ rcp].a ) , @xmath28 has the same spectrum as @xmath135 , including the multiplicities .
3 . in case
( [ rcp].b ) , the spectrum of @xmath28 arises from that of @xmath174 by doubling the original multiplicities and then including @xmath176 as an additional eigenvalue with the required complementary multiplicity .
note that , by lemma [ eigen](c ) , @xmath176 is not an eigenvalue of @xmath135 .
the eigenspace @xmath177 is spanned in case ( [ rcp].a ) by @xmath20 , and in case ( [ rcp].b ) by @xmath178 and @xmath80 , for the killing forms @xmath21 of @xmath33 and @xmath81 of @xmath175 . by (
* lemma 14.3(ii ) and formulae ( 14.5 ) ( 14.7 ) ) , if @xmath33 is of type ( [ rcp].a ) , the complexification of @xmath57^{\odot2}$ ] may be naturally identified with its ( complex ) counterpart @xmath179^{\odot2}$ ] for @xmath96 , in such a way that @xmath174 and the killing form @xmath81 become the unique @xmath75-linear extensions of @xmath28 and @xmath20 . now lemma [ eigen](a).4pt.4pt(b ) and ( [ lbz].ii ) yield ( i ) , ( ii ) and ( iv ) in case ( [ rcp].a ) . for @xmath33 of type ( [ rcp].a ) ,
lemma 13.1 of @xcite states the following .
first , @xmath57^{\odot2}$ ] is the direct sum of two @xmath36-invariant subspaces : one formed by the real parts of @xmath100-bilinear symmetric functions @xmath180 , the other by the real parts of functions @xmath181 which are antilinear and hermitian .
secondly , @xmath28 vanishes on the `` hermitian '' summand , and its action on the `` symmetric '' summand is equivalent , via the isomorphism @xmath182 , to the action of @xmath174 on @xmath100-bilinear symmetric functions @xmath29 . with diagonalizability of @xmath174 again provided by lemma [ eigen](a ) ,
this proves our remaining claims .
( the multiplicities are doubled since the original complex eigenspaces are viewed as real , while the eigenspace @xmath174 for the eigenvalue @xmath183 consists , by ( v ) and ( [ lbz].ii ) , of complex multiples of @xmath184 , the real parts of which are precisely the real linear combinations of @xmath185 and @xmath80 . )
[ realf ] it is well known @xcite that , up to isomorphisms , @xmath186 as well as @xmath187 with @xmath188 and , if @xmath189 is even , @xmath190 , are the only real forms of @xmath191 .
[ thrdm]the only complex , or real , simple lie algebras of dimensions less than @xmath192 are , up to isomorphisms , @xmath193 or , respectively , @xmath194 , @xmath195 , @xmath196 and @xmath74 , the last one being both complex dimensional and real dimensional . consequently , 1 .
a complex simple lie algebra can not be dimensional , 2 .
there is just one isomorphism type of a complex or , respectively , real simple lie algebra of dimension @xmath87 or , respectively , @xmath94 , both represented by @xmath74 , 3 .
@xmath197 for every real or complex simple lie algebra @xmath60 .
according to remark [ clssf ] , in the complex case , only @xmath193 is possible . for real lie algebras
, one can use remark [ realf ] and ( [ rcp ] ) .
[ oneig ] we can now justify the claim , made in ( * ? ? ?
* remark 12.3 ) , that @xmath35 is not an eigenvalue of @xmath28 in any real or complex simple lie algebra except the ones isomorphic to @xmath198 or , for even @xmath189 only , @xmath199 , where @xmath200 .
in fact , by theorem [ eigen ] and parts ( ii ) ( iii ) of theorem [ rspec ] , the only real or complex simple lie algebras in which @xmath28 has the eigenvalue @xmath35 are , up to isomorphisms , @xmath136 for @xmath201 and their real forms . according to remark [ realf ] ,
these are all listed in the last paragraph .
let @xmath202^{\odot2}$ ] and @xmath203 . consequently , by ( [ lgg ] ) , @xmath204,[{z},{z}{\hskip.4pt}'{\hskip.4pt}])+{\sigma}([{y},{z}],[{x},{z}{\hskip.4pt}'{\hskip.4pt } ] ) + { \sigma}([{z},{x}],[{y},{z}{\hskip.4pt}'{\hskip.4pt}])=0\,$ ] for all @xmath205 in @xmath31 .
thus , @xmath204,[{z},{z}{\hskip.4pt}'{\hskip.4pt}])=0\,$ ] whenever @xmath206 and @xmath207 with @xmath208 .
the summands @xmath209 and @xmath210 , being simple , are spanned by such brackets @xmath211\,$ ] and @xmath212 $ ] , and so @xmath209 is @xmath29-orthogonal to @xmath210 . as this is the case for any two summands ,
we obtain ( i ) , the right - to.4pt - left inclusion being obvious .
next , @xmath213 in fact , the second inclusion is obvious from theorem [ hstrh ] ; the first , from theorem [ rspec](iv ) , ( [ lbz].i ) and ( [ lrs].iii ) applied to complex multiples @xmath44 of @xmath184 .
part ( ii ) of theorem [ solut ] is immediate , as @xmath57^{\wedge4}{\hskip-1.5pt}=\{0\}\,$ ] when @xmath214 .
also , if @xmath33 is simple and @xmath73 , lemma [ thrdm](i).4pt-.4pt(ii ) implies that @xmath33 is real and isomorphic to @xmath74 . from ( [ lrs].iii ) , with @xmath215 by ( ii ) , we now get @xmath216 for @xmath217^{\odot2}\}$ ] , where @xmath57^{\odot2}$ ] denotes the space of all symmetric @xmath100-bilinear forms @xmath218 . as @xmath219 uniquely determines such @xmath29 , that is ,
the operator @xmath220 is injective , we thus have @xmath221 .
the second inclusion in ( [ fbe ] ) is therefore an equality , and @xmath222 , for dimensional reasons : @xmath62 contains the subspace @xmath223 of real dimension @xmath224 , equal , in view of part ( a ) of theorem [ eigen ] and theorem [ rspec](iii ) , to @xmath225 $ ] . this yields assertion ( iii ) in theorem [ solut ] .
let @xmath33 now be simple , with @xmath78 . due to theorems [ eigen ] and [ rspec](ii).4pt-.4pt(iii ) ,
@xmath226 is not an eigenvalue of @xmath36 .
thus , @xmath227 , and the inclusions in ( [ fbe ] ) are equalities . in view of theorem [ rspec](iv ) , this completes the proof .
in this section @xmath33 is the underlying real space of a fidimensional complex vector space @xmath96 and @xmath228 is the operator of multiplication by @xmath229 , also referred to as the _ complex structure_. we denote by @xmath81 a fixed nondegenerate @xmath100-bilinear symmetric form on @xmath96 , so that the @xmath230-bilinear symmetric form @xmath231 on @xmath33 is nondegenerate as well .
the same applies to any nonzero complex multiple of @xmath184 .
thus , @xmath21 and @xmath232 constitute a basis of a real vector space @xmath233 of @xmath230-bilinear symmetric forms on @xmath31 .
all nonzero elements of @xmath233 are nondegenerate . as
@xmath81 is @xmath100-bilinear , @xmath234 for all @xmath26 .
we use components relative to a basis of @xmath31 , as in section [ pr ] .
[ dtcpx]the real spaces @xmath33 and @xmath233 uniquely determine the pair @xmath235 up to its replacement by @xmath236 or @xmath237 , with any @xmath238 . for any basis @xmath239 of @xmath240 , replacing @xmath81 by a complex multiple , which leaves @xmath241 unchanged , we assume that @xmath242 .
thus , @xmath243 , where @xmath244 and @xmath245 .
writing the equality @xmath246 as @xmath247 , and then using the reciprocal components @xmath248 , we obtain @xmath249 . now @xmath250 may be defined by declaring the matrix @xmath251 to be the traceless part of @xmath252 , normalized so that @xmath253 . at the same time , fixing any @xmath254 we may assume , as before , that @xmath242 . then @xmath255 and @xmath256 , determine @xmath257 , being its real and imaginary parts .
combined with the last sentence of the preceding paragraph , this completes the proof .
the next fact concerns two mappings , @xmath258 and @xmath259 .
the former sends every nonzero element of @xmath233 ( which , as we know , is nondegenerate ) to its reciprocal .
the latter is the operator of index - lowering via @xmath20 , and takes values in the space of @xmath260-linear endomorphisms of @xmath31 , which include complex multiples of @xmath261 .
we then have @xmath262{\hskip.4pt}_\flat{^{\phantom i}}:{\sigma}\in{\mathcal{p}}\smallsetminus\{0\}\}\ , = \,\{a{\hskip.7pt}\mathrm{id}:a\in{{\mathchoice { \setbox0=\hbox{$\displaystyle\mathrm{c}$ } \hbox{\hbox to0pt{\kern0.4\wd0\vrule height0.9\ht0\hss}\box0 } } { \setbox0=\hbox{$\textstyle\mathrm{c}$}\hbox{\hbox to0pt{\kern0.4\wd0\vrule height0.9\ht0\hss}\box0 } } { \setbox0=\hbox{$\scriptstyle\mathrm{c}$}\hbox{\hbox to0pt{\kern0.4\wd0\vrule height0.9\ht0\hss}\box0 } } { \setbox0=\hbox{$\scriptscriptstyle\mathrm{c}$}\hbox{\hbox to0pt{\kern0.4\wd0\vrule height0.9\ht0\hss}\box0}}}}\smallsetminus\{0\}\}{\hskip.4pt}.\ ] ] namely , under index raising with the aid of @xmath20 , the operators @xmath263 , for @xmath238 , correspond to elements @xmath264 of @xmath265 characterized by @xmath266 .
every such @xmath264 is in turn the reciprocal of @xmath202^{\odot2}$ ] defined by @xmath267 , where @xmath268 ( as @xmath269 ) .
symmetry of @xmath270 , and hence @xmath29 , is obvious from @xmath20-self - adjointness of @xmath271 .
the inverses @xmath272 of our operators @xmath273 range over nonzero complex multiples of @xmath261 as well , and so the resulting symmetric forms @xmath44 act on @xmath274 by @xmath275 , where @xmath276 range over @xmath277 .
therefore @xmath278 , as required .
[ unqdt ] the relation @xmath246 for @xmath231 and @xmath232 shows that , once @xmath279 is fixed , @xmath185 uniquely determines @xmath184 .
similarly , @xmath280 and @xmath279 determine the cartan form @xmath281 of a complex lie algebra @xmath96 , cf .
( [ oeb ] ) .
in fact , @xmath282 .
[ cbdet ] the bracket @xmath107\,$ ] of a real.7pt@xmath40complex semisimple lie algebra is uniquely determined by @xmath97 and @xmath21 via ( [ oeb ] ) .
knowing @xmath97 and the set of nonzero scalar multiples of @xmath20 , rather than @xmath21 itself , makes @xmath107\,$ ] unique up to multiplications by cubic roots of @xmath90 .
such factors must be allowed as multiplying @xmath107\,$ ] by a scalar @xmath283 replaces @xmath21 and @xmath97 with @xmath284 and @xmath285 .
[ cnjug ] in the first sentence of remark [ cbdet ] , treating @xmath97 and @xmath21 formally , we see that in the complex case @xmath286 and @xmath287 determine , via ( [ oeb ] ) , the same bracket @xmath107\,$ ] as @xmath97 and @xmath20 .
for a real.7pt@xmath40complex lie algebra @xmath31 , let the mapping @xmath288^{\wedge3}{\hskip-1.5pt}\times{\hskip-.4pt}{\mathfrak{g}}^{\odot2}{\hskip-1.5pt}\to[{\mathfrak{g}}{\hskip-.7pt}^*]^{\wedge4}$ ] be defined by @xmath289({x},{y},{z},{z}{\hskip.4pt}'{\hskip.4pt})\ , = \,{\mu}({c}({x},{y}),{c}({z},{z}{\hskip.4pt}'{\hskip.4pt})){\hskip.7pt}+{\hskip.7pt}{\mu}({c}({y},{z}),{c}({x},{z}{\hskip.4pt}'{\hskip.4pt})){\hskip.7pt}+{\hskip.7pt}{\mu}({c}({z},{x}),{c}({y},{z}{\hskip.4pt}'{\hskip.4pt}))$ ] , where @xmath290 is treated as a symmetric real.7pt@xmath40complex - bilinear form on @xmath291 , and @xmath292 stands for the element @xmath293 of @xmath291 .
if @xmath33 is also semisimple , the isomorphic identification @xmath294 provided by the killing form @xmath21 induces an isomorphism @xmath57^{\odot2}{\hskip-.7pt}\to{\mathfrak{g}}^{\odot2}{\hskip-1.5pt}$ ] , which we write as @xmath295 .
then , in view of ( [ lgg ] ) and ( [ oeb ] ) , @xmath296 theorem [ autcf ] is a trivial consequence of the following result combined with lemma [ thrdm](ii ) and the fact that , by multiplying a -algebra bracket operation @xmath107\,$ ] by a nonzero scalar , one obtains a -algebra structure isomorphic to the original one .
[ dtrmn]in a real or complex semisimple lie algebra @xmath60 , the cartan form and the vector - space structure of @xmath33 uniquely determine each of the following objects . 1 . the vector subspaces constituting the simple direct summand ideals of @xmath60 .
2 . up to a sign , in the real case , the complex structure , defined as in section [ sf].4pt , of every summand ideal @xmath297 with @xmath298 which is a complex lie algebra , treated as real .
3 . up to multiplications by cubic roots of @xmath90 , the restrictions of the lie-.4ptalgebra bracket of @xmath33 to all such summands of dimensions other than @xmath87 or @xmath94 .
4 . the lie-.4ptalgebra isomorphism types of all summand ideals @xmath297 with @xmath299 . let @xmath97 be the cartan form of @xmath31 . by ( [ fcs ] ) , @xmath300 for the real.7pt@xmath40complex - linear operator @xmath301^{\wedge4}$ ] given by @xmath302 .
then , if one views all @xmath303 as linear operators @xmath304 , 1 . _
the simple direct summands of @xmath33 are precisely the minimal elements , in the sense of inclusion , of the set _
@xmath305 . in fact , @xmath306 consists of the images of those linear endomorphisms @xmath307 which correspond via ( [ oms].b ) to elements @xmath44 of @xmath308 , and have @xmath309 .
to describe all such @xmath310 , we use the four parts of theorem [ solut ] , referring to them as ( i ) ( iv ) . specifically , by ( i ) , our @xmath310 are direct sums of linear endomorphisms @xmath311 of the simple direct summands @xmath312 of @xmath31 , while @xmath311 are themselves subject to just two restrictions : one due to the exclusion of ranks @xmath313 and @xmath314 , the other depending , in view of ( ii ) ( iv ) , on @xmath315 , as follows . if @xmath316 , ( ii ) states that @xmath311 is only required to be @xmath20-self - adjoint ( to reflect symmetry of @xmath317 related to @xmath311 as in ( [ oms].b ) ) .
similarly , it is clear from ( iv ) and ( [ und ] ) that , with a specific the scalar field @xmath318 , @xmath319 in the remaining case , @xmath320 .
then , by ( iii ) and theorem [ solut](iii ) , @xmath311 is complex - linear and @xmath20-self - adjoint , cf .
( [ cxl ] ) and ( [ lrs].i ) , but otherwise arbitrary .
the image @xmath321 of any @xmath322 as above is the direct sum of the images of its summands @xmath311 , and so it can be minimal only if there exists just one @xmath323 with @xmath324 . for this @xmath229 , minimality of @xmath325
implies that @xmath326 .
in fact , in view of the last paragraph , the cases @xmath316 and @xmath327 are obvious ( the former since @xmath328 ) while , if @xmath320 , complex - linearity of @xmath311 precludes not just @xmath329 and @xmath314 , but also @xmath87 from being its real rank .
we thus obtain one of the inclusions claimed in ( e ) : every minimal element of @xmath306 equals some summand @xmath312 .
conversely , any fixed summand @xmath312 is an element of @xmath330 , realized by @xmath310 with @xmath331 and @xmath332 for all @xmath208 , cf .
lemma [ thrdm](iii ) .
minimality of @xmath312 is in turn obvious from ( [ smi ] ) if @xmath333 , while for @xmath334 or @xmath335 it follows from the restriction on @xmath336 combined , in the latter case , with complex - linearity of @xmath337 .
this yields ( e ) . now ( a ) is obvious from ( e ) , as @xmath338 and @xmath339 depend only on @xmath97 and the vector - space structure of @xmath31 . to prove ( b )
( c ) , we fix @xmath323 with @xmath333 .
elements @xmath264 of @xmath340 having @xmath341 correspond , via ( [ oms].b ) followed by the assignment @xmath342 , to endomorphisms @xmath322 of @xmath33 which satisfy ( [ smi ] ) and vanish on @xmath343 for @xmath208 .
any such @xmath270 , now viewed as a bilinear form on @xmath291 , is therefore obtained from a bilinear form @xmath344 on @xmath345 by the trivial extension to @xmath291 , that is , pullback under the obvious restriction operator @xmath346 . if @xmath347
, it is immediate from ( [ smi ] ) that the resulting forms @xmath344 are nonzero multiples of the reciprocal of the killing form of @xmath312 , and remark [ cbdet ] implies ( c ) .
next , let @xmath105 .
we denote @xmath312 treated as a complex lie algebra by @xmath82 and cartan form of @xmath82 by @xmath348 . formula ( [ rec ] )
states that , in view of ( [ smi ] ) , the reciprocals of our @xmath344 are precisely the nonzero elements of the space @xmath233 defined in section [ sf ] .
thus , lemma [ dtcpx ] , ( [ lrs].ii ) and remark [ unqdt ] imply that @xmath97 determines the triple @xmath349 uniquely up to replacements by @xmath350 or @xmath351 , with @xmath238 .
this proves ( b ) , while using remarks [ cbdet ] [ cnjug ] we obtain ( c ) for @xmath352 as well . finally , ( c ) and lemma [ thrdm](i).4pt.4pt(iii ) easily yield ( d ) .
we thank robert bryant and nigel hitchin for helpful comments about theorem [ autcf ] .
for any complex simple lie algebra @xmath31 , the operator @xmath28 with ( [ oms ] ) is diagonalizable .
its systems @xmath353\,$ ] of eigenvalues and @xmath354\,$ ] of the corresponding multiplicities are @xmath355=(2,1,2/n,-2/n)\hskip5pt\mathrm{and}\\ \mathrm{mult}{\hskip.7pt}[{\mathfrak{sl}}_n{^{\phantom i}}]=(1,n^2 - 1,n^2(n-3)(n+1)/4,n^2(n+3)(n-1)/4){\hskip.4pt } , \hskip5pt\mathrm{if\ } \,n\ge4{\hskip.4pt}.\\ \mathrm{spec}{\hskip.7pt}[{\mathfrak{sp}}_n{^{\phantom i}}]=(2,(n+4)/(n+2),-4/(n+2),2/(n+2))\hskip6pt \mathrm{for\ even\ } \,n\ge4{\hskip.4pt},\hskip5pt\mathrm{and}\\ \mathrm{mult}{\hskip.7pt}[{\mathfrak{sp}}_n{^{\phantom i } } ] = ( 1,(n-2)(n+1)/2,n(n+1)(n+2)(n+3)/24,n(n-1)(n-2)(n+3)/12){\hskip.4pt}.\\ \mathrm{spec}{\hskip.7pt}[{\mathfrak{so}}_n{^{\phantom i}}]=(2,(n-4)/(n-2),4/(n-2),-2/(n-2))\hskip5pt \mathrm{if\ } \,n=7\,\mathrm{\ or\ } \,\,n\ge9{\hskip.4pt},\hskip5pt\mathrm{while}\\ \mathrm{mult}{\hskip.7pt}[{\mathfrak{so}}_n{^{\phantom i } } ] = ( 1,(n+2)(n-1)/2,n(n-1)(n-2)(n-3)/24,n(n+1)(n+2)(n-3)/12)\\ \end{array}\ ] ] and , if @xmath33 is one of the exceptional complex lie algebras @xmath356 , @xmath357=(2,(1+w)/6,(1-w)/6),\hskip5pt\mathrm{with } \hskip5pt\mathrm{mult}{\hskip.7pt}[\mathfrak{g}]\hskip5pt\mathrm{equal\ to}\\ ( 1,3d[(d+2)w-(d+32)]/[w(11-w)],3d[(d+2)w+(d+32)]/[w(11+w)]){\hskip.4pt } , \end{array}\ ] ] where @xmath358 and @xmath359^{1/2}{\hskip-1.5pt}$ ] .
this is a result of meyberg @xcite who , instead of our @xmath36 , studied the operator @xmath360 .
( the exponent @xmath361 is missing in @xcite ) .
for @xmath362 , the resulting `` eigenvalue '' @xmath363 of multiplicity @xmath176 should be disregarded .
all isomorphism types of complex simple lie algebras are listed above , cf .
remark [ clssf ] .
the dimensions @xmath364 of @xmath356 are @xmath365 @xcite . the eigenvalues @xmath176 and @xmath35 in ( [ exc ] ) would correspond to @xmath366 or @xmath367 , of which only the latter occurs , for @xmath368 and @xmath369 , in agreement with lemma [ eigen ] .
besse , a.l .
: einstein manifolds .
( 3 ) , vol .
springer , berlin ( 1987 ) bourguignon , j.p . ,
karcher , h. : curvature operators : pinching estimates and geometric examples .
sci . cole norm .
( 4 ) * 11 * , 7192 ( 1978 ) r.2.3ptbryant , private communication ( 2009 ) calabi , e. , vesentini , e. : on compact , locally symmetric khler manifolds . ann . of math .
( 2 ) * 71 * , 472507 ( 1960 ) derdzinski , a. , gal , .r . :
indefinite einstein metrics on simple lie groups .
indiana univ .
j. ( to appear ) . | the killing form @xmath0 of a real ( or complex ) semisimple lie group @xmath1 is a left - invariant pseudo - riemannian ( or , respectively , holomorphic ) einstein metric .
let @xmath2 denote the multiple of its curvature operator , acting on symmetric 2-tensors , with the factor chosen so that @xmath3 .
the result of meyberg @xcite , describing the spectrum of @xmath2 in complex simple lie groups @xmath1 , easily implies that @xmath4 is not an eigenvalue of @xmath2 in any real or complex simple lie group @xmath1 except those locally isomorphic to su(@xmath5 ) , or sl(@xmath6 ) , or sl(@xmath7 ) or , for even @xmath8 only , sl(@xmath9 ) , where @xmath10 and @xmath11 . due to the last conclusion , on simple lie groups @xmath1 other the ones just listed , nonzero multiples of the killing form @xmath0 are isolated among left - invariant einstein metrics .
meyberg s theorem also allows us to understand the kernel of @xmath12 , which is another natural operator .
this in turn leads to a proof of a known , yet unpublished , fact : namely , that a semisimple real or complex lie algebra with no simple ideals of dimension 3 is essentially determined by its cartan three - form .
[ section ] [ thm]lemma [ thm]remark _ _ |
ngc 1275 ( also commonly known as perseus a , 3c 84 , 4c+41.07 , or iras 03164 + 4119 ) is the central galaxy in the perseus cluster , abell 426 .
the cluster abell 426 shows two distinct velocity components : ngc 1275 ( @xmath13 ) belongs to the low - velocity component , which is associated with velocities of 5200 .
the high - velocity component , showing velocities of 8200 is located in front of ngc 1275 to the north and north - west of the nucleus @xcite .
the two velocity components have been interpreted as indicating an ongoing galaxy cluster merger @xcite .
however , x - ray absorption studies by @xcite and @xcite indicate that the separation between both velocity components is large .
@xcite suggest that the high - velocity system is not yet merging with ngc 1275 but is still in the process of falling into the cluster .
the number of classifications for ngc 1275 in the literature fanaroff - riley 1 radio galaxy ( f - r i ) , bl lac object , post - merger galaxy , cooling - flow galaxy , peculiar seyfert 1 , and liner reflect the fact that ngc 1275 exhibits an extraordinarily rich set of interstellar physical phenomena .
it is a star - forming early - type cd galaxy , harbouring an active nucleus which is a point source of hard x - rays @xcite and a very strong radio source ( hence the name perseus a ) .
the radio emission on large - scales and parsec - scales shows a pair of jets @xcite .
the jets are oriented at a position angle of 160@xmath14 @xcite .
using vlbi measurements of the parsec - scale jet , @xcite inferred a likely range of angles to the line of sight of @xmath15 .
the southern jet is the one approaching us .
ngc 1275 is associated with strong thermal x - ray emission , which extends over several arc minutes around the active nucleus and shows a complex bubble - like morphology @xcite , as well as with an extended structure of h@xmath16 filaments @xcite .
these filaments of ionized gas have been studied in detail by @xcite .
they are likely to be stabilized by magnetic fields @xcite and display liner - like spectra which have been analysed by @xcite .
@xcite examined a variety of excitation mechanisms including shocks , ionization by an agn , and stellar photoionization without finding a fully adequate explanation for the observed spectra .
collisional heating by ionizing particles has recently been investigated in greater detail as a possible explanation for the emission characteristics of the filaments @xcite .
ngc 1275 was originally believed to represent the prototypical cooling - flow galaxy , in which the h@xmath16 filaments have formed by cooling of the intracluster medium @xcite .
but the correlation between the x - ray and h@xmath16 morphology suggests that the filaments rather consist of cold gas which is uplifted behind the radio bubbles driven by the radio jet @xcite .
@xcite and @xcite have shown that there is also a tight correlation between the extended co line emission , the h@xmath16 filaments , the extended x - ray emission , and the large radio lobes mapped by @xcite .
likewise , @xcite found a close correspondence between the h@xmath16 filaments and the extended h@xmath3 s(1 ) emission .
a fraction of the molecular gas and much of the thermal x - ray emission are confined in shells around bubbles of non - thermal synchrotron - emitting plasma .
this is clear evidence of the importance of the radio lobes in compressing and cooling the ambient galactic medium .
this ambient galactic medium can be derived from either shock - cooled thermal x - ray plasma or from the debris of a recent galactic merger @xcite .
the total mass of molecular gas in ngc 1275 is large .
based on observations of the nuclear and extended gas , @xcite measured a total mass of cold h@xmath3 of @xmath17 , using the standard co to molecular hydrogen conversion factor . about @xmath18 of this molecular gas
are contained in the inner @xmath19 kpc , a mass that is comparable with the total mass of h@xmath3 in the milky way .
most of the emission arises in an extended east - west structure passing through the nucleus .
@xcite found that the co emission in the central 8 kpc diameter region does not show any evidence of large - scale rotation .
however , the co line profile in the central 4 kpc diameter region is double - peaked , which may indicate inflows , outflows , or a rotating disc @xcite .
the source of the molecular hydrogen in the nuclear region of ngc 1275 is thought to be turbulent infalling gas . whether this gas is carried into the central regions by a galactic merger or a cooling flow
is unclear ( e.g. * ? ? ?
interferometric observations of the co emission in the central 10 kpc by @xcite revealed that most of the gas is associated with radial filaments , which are oriented in the east - west direction ( i.e. perpendicular to the jet axis ) , located between x - ray cavities , and consistent with radially infalling gas flows .
@xcite argued that the absence of any large - scale rotation makes a merger - origin of the gas less likely .
the near - infrared spectrum of the nuclear region was investigated by @xcite who found that the [ ] lines are unusually strong and narrow , indicating that they arise from a region with kinematics different from the nucleus .
it was suggested that these features are caused by shocks .
shocks were also suggested as the cause of the h@xmath3 s(1 ) line emission , first investigated by @xcite .
these molecular shocks were discussed as possible triggers of starburst activity .
@xcite , @xcite , and @xcite determined the excitation temperatures of the near - infrared h@xmath3 lines in the circum - nuclear region .
they found the emission of this @xmath20 k h@xmath3 component to be consistent with thermal excitation from x - ray or shock heating . because the brightest h@xmath3 emission is extended over only the central @xmath21300 pc ,
@xcite concluded that it is more likely to be directly associated with the agn rather than being located in a large - scale cooling flow .
@xcite agreed with this conclusion and further suggested that the h@xmath3 emission is too bright to be part of a cooling flow .
in addition , their hst images show a faint extension of h@xmath3 s(1 ) emission from the nucleus to the southwest as well as a faint feature to the east of the nucleus .
@xcite inferred the properties of the near - infrared h@xmath3 in the centre of ngc 1275 , using observations of the central @xmath22 arcsec@xmath1 with the imaging spectrograph uist ( ukirt imaging spectrometer ) on ukirt ( united kingdom infrared telescope ) taken under seeing - limited conditions of 0.30.4 arcsec .
they suggested that the h@xmath3 emission at around 50 pc from the nucleus arises in a dense clumpy disc , rotating in a plane perpendicular to the radio jet .
they concluded that thermal excitation by the nuclear radiation field is the dominant excitation mechanism of the h@xmath3 emission .
based on the discontinuity in h@xmath3 velocity across the nucleus of around 240 km s@xmath23 , they estimated a mass of @xmath24 m@xmath25 for the black hole at the nucleus , assuming a disc inclination of @xmath26 .
the nuclear [ ] 1.644 emission was found to be spatially resolved , while the nuclear pa@xmath16 emission remained unresolved . according to @xcite , the nuclear disc of hot h@xmath3 would likely be part of the larger - scale h@xmath3 and co structure seen in ngc 1275 .
this , together with the more recent observations of molecular gas in the extended filaments on several 10 kpc scales @xcite , suggests that the fuelling process of the agn in ngc 1275 by interstellar material is being observed over a large range of scales . in this paper
, we analyse the inner circum - nuclear disc of ngc 1275 using observations with nifs ( near - infrared integral - field spectrograph ; * ? ? ?
* ) in conjunction with the adaptive - optics system , altair , on the gemini north telescope .
the nifs data permit a high - spatial - resolution study of the kinematics of the molecular and ionized gas via the h@xmath3 s(1 ) , [ ] 1.644 , 2.058 , and br@xmath5 2.166 lines as well as an examination of the excitation mechanisms of h@xmath3 and [ ] . throughout the paper we assume h@xmath27 km s@xmath23 mpc@xmath23 , @xmath28 , and @xmath29 , which results in a spatial scale of 358 pc arcsec@xmath23 for ngc 1275 @xcite .
the gemini near - infrared integral - field spectrograph ( nifs ) and altair adaptive - optics system on the 8.1 m gemini north telescope on mauna kea in hawaii were used to obtain @xmath30-band and @xmath31-band spectra of ngc 1275 .
nifs has 29 slitlets , each 0.1 arcsec wide , and has a resolution of 0.04 arcsec per spaxel along each slitlet .
the field of view is @xmath32 arcsec@xmath1 , and it delivers a spectral resolving power of @xmath21 5300 , matched to two pixels on the detector , at both @xmath30 and @xmath31 .
this corresponds to a velocity resolution of @xmath21 60 km s@xmath23 .
the @xmath31-band observations of ngc 1275 were obtained on 2005 november 11 during the commissioning of nifs .
the observations were performed using the altair natural guide star system with the nucleus of ngc 1275 as the reference for the adaptive - optics correction .
the nearby star u1275_02199239 was used as the on - instrument wavefront sensor ( oiwfs ) guide star for slow flexure correction .
the instrument was oriented at a position angle of @xmath33 , so that the slitlets were aligned e - w .
the observation consisted of eight exposures of 600 s duration centred on ngc 1275 , interleaved with eight 600 s exposures of an adjacent sky region .
none of these exposures was spatially dithered . the a0v hipparcos star hip 18769 was observed after the ngc 1275 observation to provide correction of telluric absorption and flux calibration from its 2mass magnitudes .
the @xmath30-band observations of ngc 1275 were obtained during nifs science verification observations ( program gn-2006a - sv-124 ) on 2006 february 1 and 2 .
again , altair was used in natural guide star mode and the instrument was set to a position angle of @xmath33 .
two object - sky exposure pairs were obtained with 900 s exposure times in modest seeing , one pair on each night .
hip 18769 was also observed after ngc 1275 on each night .
the data were reduced in a standard way @xcite using a variant of the gemini iraf package .
the object and sky exposures were combined first . then a dark exposure was subtracted from both the combined - object and combined - sky frames .
the combined - sky frame was scaled to match the combined - object frame and subtracted .
the sky - subtracted object frame was then flatfielded , bad pixels were interpolated , and the 29 individual two - dimensional slitlet spectra were extracted from this image .
argon arc spectra were used to define the wavelength calibration and flatfield exposures through a ronchi grating were used to define the spatial calibration along slitlets .
each two - dimensional slitlet spectrum was then geometrically transformed onto a rectilinear spatio - spectral coordinate grid and the resulting 29 two - dimensional spectra combined into a three - dimensional data cube .
a one - dimensional spectrum was extracted from a 0.5 arcsec - diameter circular aperture defined within the standard star data cube , a correction for telluric absorption was defined , and this one - dimensional correction spectrum was applied to all spaxels within the object and standard star data cubes .
similarly , a flux - calibration spectrum was extracted from the telluric - corrected standard star data cube using a 1.5 arcsec - diameter circular aperture with the level set by the 2mass @xmath30 or @xmath31 band magnitude and the blackbody shape defined by its 2mass @xmath34 colour temperature of 11000 k. the final data cubes obtained at @xmath30 and @xmath31 are in the form of multi - extension fits files with 2040 wavelength pixels and @xmath35 spatial pixels ( spaxels ) .
the resulting @xmath30- and @xmath31-band spectra of the core of ngc 1275 , integrated over a 1.0 arcsec - diameter circular aperture , are shown in figs [ f : hsp ] and [ f : ksp ] .
the @xmath31-band spectrum of the core of ngc 1275 is known to be dominated by h@xmath3 emission lines @xcite .
our spectra confirm a large number of h@xmath3 lines , mostly in the @xmath31-band spectrum and some weaker lines in the @xmath30-band spectrum .
the @xmath30-band is dominated by the strong [ ] 1.644 emission .
other weaker [ ] emission lines are also apparent . furthermore , weaker lines of the brackett series and of are seen in the @xmath30- and @xmath31-band .
the flux measurements for the emission lines detected in the nifs spectra are listed in table [ t : flux ] .
llccccl & & ( 3 ) & ( 4 ) & ( 5 ) & ( 6 ) & + & & flux & fwhm & flux & fwhm & + & & ( erg s@xmath23 @xmath36 ) & ( km s@xmath23 ) & ( erg s@xmath23 @xmath36 ) & ( km s@xmath23 ) & + & & & & + @xmath37\ a^4d_{5/2}-a^4f_{9/2}}$ ] & @xmath38 & @xmath39 & @xmath40 & @xmath41 & @xmath42 & + @xmath37\ a^4d_{3/2}-a^4f_{7/2}}$ ] & @xmath43 & @xmath44 & @xmath45 & @xmath46 & @xmath47 & + @xmath48\ a^4d_{7/2}-a^4f_{9/2}}$ ] & @xmath49 & @xmath50 & @xmath51 & @xmath52 & @xmath53 & + @xmath37\ a^4d_{1/2}-a^4f_{5/2}}$ ] & @xmath54 & @xmath55 & @xmath56 & ... & ... & blend + h@xmath3 1 - 0 s(10 ) & 1.6665 & @xmath57 & ... & @xmath58 & ... & blend + @xmath37\ a^4d_{5/2}-a^4f_{7/2}}$ ] & @xmath59 & @xmath60 & ... & @xmath61 & ... & blend + h@xmath3 1 - 0 s(9 ) & @xmath62 & @xmath63 & @xmath64 & @xmath65 & @xmath66 & + @xmath67 & @xmath68 & @xmath69 & ... & @xmath70 & ... & blend + @xmath37\ a^4d_{3/2}-a^4f_{5/2}}$ ] & @xmath71 & @xmath72 & ... & @xmath73 & ... & blend + h@xmath3 1 - 0 s(8 ) & @xmath74 & @xmath75 & ... & @xmath76 & ... & blend + br10 & @xmath77 & @xmath78 & ... & @xmath79 & ... & low s / n + h@xmath3 1 - 0 s(7 ) & @xmath80 & @xmath81 & @xmath82 & @xmath83 & @xmath84 & + + h@xmath3 2 - 1 s(4 ) & @xmath85 & @xmath86 & ... & ... & ... & low s / n + h@xmath3 1 - 0 s(2 ) & @xmath87 & @xmath88 & @xmath89 & @xmath90 & @xmath91 & + @xmath92 & @xmath93 & @xmath94 & @xmath95 & @xmath96 & @xmath97 & + h@xmath3 2 - 1 s(3 ) & @xmath98 & @xmath99 & @xmath100 & @xmath96 & @xmath101 & + h@xmath3 1 - 0 s(1 ) & @xmath102 & @xmath103 & @xmath100 & @xmath104 & @xmath101 & + h@xmath3 2 - 1 s(2 ) & @xmath105 & @xmath106 & @xmath107 & @xmath108 & @xmath109 & + @xmath110 & @xmath111 & @xmath112 & @xmath113 & @xmath114 & @xmath115 & + h@xmath3 3 - 2 s(3 ) & @xmath116 & @xmath117 & @xmath118 & @xmath119 & @xmath120 & + h@xmath3 1 - 0 s(0 ) & @xmath121 & @xmath122 & @xmath118 & @xmath123 & @xmath124 & + h@xmath3 2 - 1 s(1 ) & @xmath125 & @xmath126 & @xmath127 & @xmath128 & @xmath129 & + h@xmath3 2 - 1 s(0 ) & @xmath130 & @xmath131 & ... & ... & ... & low s / n + a knowledge of the @xmath31-band point - spread function ( psf ) is required for analysing the surface - brightness profiles and the kinematics of the emission - line gas in the circum - nuclear region of ngc 1275 .
we use the spatial profile of the standard star , hip 18769 , as our psf reference . obtaining
the psf from separate standard - star observations involves uncertainties , since the observing conditions may have changed between these and the object observations .
the psf could be measured more directly using an unresolved emission component in the @xmath31-band data cube for ngc 1275 .
candidates for this are the @xmath31-band continuum emission or hydrogen emission from the agn broad - line region
. however , fig .
[ f : psf ] shows that the profile of the @xmath31-band continuum of ngc 1275 ( filled squares ) is extended compared to the profile of the standard star .
the same is true for the br@xmath5 2.166 emission of ngc 1275 ( open squares ) , which is integrated over a velocity interval from @xmath132 to @xmath133 after continuum subtraction .
the fact that the br@xmath5 2.166 emission is spatially extended suggests that the integrated line includes emission from outside the agn broad - line region . in order to isolate any emission from the agn broad - line region , it would be necessary to decompose the line profile , which is not practical at the given signal - to - noise ratio . -band profile of the standard star hip 18769 ( plus signs ) and a comparison with the profiles of the @xmath31-band continuum ( filled squares ) and the integrated br@xmath5 emission ( open squares ) of ngc 1275 .
the br@xmath5 emission is integrated over a velocity interval from @xmath132 to @xmath133 .
all profiles are centred on the emission centroids , respectively , and normalized to their peak values .
positive and negative radii show the profile obtained from an azimuthal average over two opposite halves of the emission peak .
the psf fit is based on a moffat profile , as explained in the text.[f : psf ] ] the radial profile of hip 18769 and the corresponding psf fit are shown in fig .
[ f : psf ] .
the psf fit is based on a single moffat function of the form @xmath134 where @xmath135 is the amplitude , @xmath136 is the radius , and fwhm is the full width at half maximum . fitting this function to the standard star profile results in a fwhm of 0.103 arcsec ( 36 pc ) , corresponding to one spatial pixel across the slitlets
s(1 ) emission , ( b ) unsharp mask of the integrated h@xmath3 s(1 ) surface brightness , ( c ) integrated surface brightness of the [ ] 1.644 emission , and ( d ) unsharp mask of the integrated [ ] 1.644 surface brightness .
the images are plotted using an inverted grey - scale .
the location of the respective continuum peak in the @xmath31- and @xmath30-band is indicated with a white cross .
the h@xmath3 s(1 ) and [ ] 1.644 emission lines are integrated over velocity intervals from @xmath132 to @xmath133 and @xmath137 to @xmath138 , respectively .
the halo of negative pixels around the line emission is an artefact of the unsharp masking procedure , where a smoothed copy of the original image is subtracted from the original image ( see text for details ) .
the extended circum - nuclear emission in h@xmath3 is interpreted as a large - scale molecular accretion disc ( panel a ) on which a streamer is superimposed to the south - west of the nucleus ( panel b ) .
the streamer is not seen in the [ ] 1.644 emission ( panel d ) . ]
the circum - nuclear regions of the integrated flux maps in h@xmath3 s(1 ) and [ ] 1.644 show a steeply increasing surface brightness towards the core of the galaxy ( fig .
[ h2]a and c ) . in the analysis presented here ,
this emission is interpreted as tracing the outer parts of an inclined accretion disc around the central active nucleus .
the classical inner accretion disc is unresolved at our spatial resolution of 36 pc .
the integrated h@xmath3 emission is clearly perturbed towards the south - west .
[ h2]b shows an unsharp mask of the integrated h@xmath3 image of fig .
[ h2]a , which is the result of subtracting a smoothed copy of fig .
[ h2]a from fig .
the smoothing is performed by convolving the image with a gaussian function , using a fwhm of 3 pixels .
[ h2]b reveals a feature extending from the nucleus to the west and south , which is likely to be a part of the southwestern emission feature reported by @xcite .
it is uncertain whether this is a stream of gas falling into the centre or being ejected from the centre .
however , @xcite found signs of generally infalling motions in the molecular gas , , in the inner @xmath139 kpc on kiloparsec scales .
since it is likely that these streams of gas continue to smaller scales , we interpret the southwestern feature in our nifs data as an infalling stream of gas settling into the overall accretion disc at small radii . as we will see in sections [ s : kinematics ] and [ s : model ] , the streamer is associated with redshifted velocities and apparently joins the h@xmath3 disc on a retrograde orbit .
the streamer is not detected in [ ] 1.644 emission .
this is obvious from fig .
[ h2]d , which shows the unsharp mask corresponding to the integrated [ ] 1.644 emission .
the lack of [ ] detection in the streamer suggests that the streamer is either predominantly molecular in nature , or not exposed to sufficient internal high - velocity shocks or sufficient x - ray heating from the nuclear continuum source , both of which are possible [ ] excitation mechanisms .
in agreement with the morphological evidence for a circum - nuclear disc from section [ s : morphology ] and the previous indications for a rotating disc based on the lower - resolution data by @xcite , the h@xmath3 s(1 ) 2.122 and [ ] 1.644 emission lines show kinematics indicative of disc rotation in the central @xmath140 arcsec ( @xmath141 pc ) of ngc 1275 .
the kinematic maps , based on a single - gaussian profile fit to the h@xmath3 s(1 ) 2.122 and [ ] 1.644 lines , are shown in figs [ f : h2fit ] and [ f : feiifit ] .
the four panels of each figure show the central velocity ( upper left panel ) , the velocity dispersion ( upper right panel ) , the continuum flux density ( lower left panel ) , and the intensity of the gaussian fit to the line profile ( lower right panel ) .
the scenario of disc rotation is supported by the fact that the redshifted and blueshifted emission of h@xmath3 s(1 ) 2.122 and [ ] 1.644 is symmetrical with respect to the peak in continuum intensity ( indicated by the cross in figs . [
f : h2fit ] and [ f : feiifit ] ) . in addition , the continuum peak coincides with the line intensity peak and the peak in velocity dispersion of both lines . a jet - driven origin of the h@xmath3 and [ ]
kinematics in the circum - nuclear region can be ruled out , because the major kinematic axis of the proposed disc is perpendicular to the radio jets .
the major kinematic axis is found at a position angle of pa=68@xmath14 , while the radio jets are located at pa=160@xmath14 @xcite .
the fact that the two position angles are perpendicular to each other strongly suggests that the observed rotating disc represents the outer parts of an accretion disc and that the jets are ejected along the polar direction . in this case
, a disc inclination of about 45@xmath14 can be inferred from the known angle of the radio jets with respect to the line of sight ( cf . * ? ? ?
furthermore , since the southern radio jet is , by its brightness , inferred to be pointing towards us , we can conclude that the disc is seen in projection with its northern portions closer to us , and that it is rotating in a counter - clockwise direction . on larger scales the kinematics of h@xmath3
s(1 ) 2.122 and [ ] 1.644 is more complex . where detected ,
both lines show a similar line - of - sight velocity field .
an exception to this is the region directly south and west of the blueshifted portion of the disc , where the h@xmath3 velocity field shows indications of redshifted velocities .
the h@xmath3 emission from this region is interpreted as a streamer of molecular gas , which is associated with the morphological feature shown in fig .
the h@xmath3 velocity dispersion is characterized by a secondary peak at about 0.4 arcsec west of the nucleus in the upper right panel of fig .
[ f : h2fit ] . at this position
the observed h@xmath3 line profile splits into separate kinematic components and becomes double - peaked .
since the single - gaussian fit is not an appropriate model for these complex line profiles , we show the velocity slices for the h@xmath3 emission in fig .
[ f : h2vel ] .
the dominant streamer feature is seen as an extension to the south - west , which is associated with redshifted velocities . as argued in section [ s : morphology ] , we suggest that this represents an infalling stream of molecular gas . in this scenario , the gas stream enters the nuclear region on a retrograde orbit with respect to the sense of rotation of the molecular disc .
this is likely to cause shocks and turbulence that lead to energy dissipation in the disc , which may stimulate more rapid gas accretion onto the nucleus of ngc 1275 .
in addition to this prominent streamer , fig .
[ f : h2vel ] shows indications of other streamers which are likely to connect to the filamentary structure around ngc 1275 on larger scales .
a common feature of the kinematic maps in both , h@xmath3 s(1 ) and [ ] 1.644 emission , is the circular region of redshifted velocities at a distance of about 1.2 arcsec from the nucleus to the north - west .
this region is characterized by small velocity dispersions in both lines .
with respect to the core , this region is found at a position angle of @xmath142 , which is close to the projected axis of the radio jet at pa=160@xmath14 @xcite , ( i.e. pa=@xmath143 for the northern radio jet with respect to the core ) . since the projected location of this region roughly coincides with the jet axis , this feature may result from a jet - cloud interaction .
by comparing the h@xmath3 s(1 ) , [ ] 1.644 2.058 and br@xmath5 emission , we find evidence of an ionization structure within the circum - nuclear disc .
s(1 ) emission traces the outer regions of the disc and is likely to form a toroid , rather than continuing all the way to the nucleus .
we use the dust sublimation radius as an estimate for the innermost radius at which h@xmath3 molecules can survive without being dissociated by the nuclear radiation field .
a simple estimate of that radius is given by the distance from the core at which the nuclear bolometric luminosity , @xmath144 , is sufficient for heating the dust to its sublimation temperature , @xmath145 : @xmath146 where @xmath12 is the stefan - boltzmann constant .
we use @xmath147 as bolometric luminosity of ngc 1275 @xcite . for a dust sublimation temperature of @xmath148
k ( see e.g. * ? ? ?
* ) , the sublimation radius is about 0.1 pc .
a larger sublimation radius is possible , if the sublimation temperature is lower : for example , 1 pc corresponds to a local temperature of about 500 k. for the forbidden [ ] 1.644 line , the critical density is of the order of @xmath149 ( see fig . [
f : feiidens ] ) .
therefore , the [ ] 1.644 emission becomes less important relative to the permitted lines in the high - density regions closer to the nucleus .
the 2.058 and br@xmath5 lines , however , are expected to be tracers of the ( presumably photoionized ) inner portions of the disc as well as , possibly , the outer rim of an inflated disc facing the agn .
evidence for such a scenario can be found in fig .
[ f : hefit ] .
this figure shows a single - gaussian fit to the 2.058 and br@xmath5 emission lines in ngc 1275 , in direct analogy to figs [ f : h2fit ] and [ f : feiifit ] for the h@xmath3 s(1 ) and [ ] 1.644 emission lines .
the 2.058 and br@xmath5 emission is slightly extended compared to the @xmath31-band continuum emission of ngc 1275 ( see also fig . [
f : psf ] ) , but more compact than the h@xmath3 s(1 ) and [ ] 1.644 emission .
the orientation of the major kinematic axis of the velocity field in 2.058 and br@xmath5 matches that of the disc in h@xmath3 s(1 ) and [ ] 1.644 , suggesting that the 2.058 and br@xmath5 emission are associated with the same physical structure as the latter two emission lines .
in addition , like h@xmath3 s(1 ) and [ ] 1.644 , 2.058 and br@xmath5 show a peak in velocity dispersion that is centred on the core of ngc 1275 .
however , the velocity dispersion of and br@xmath5 in the core ( @xmath150 ) is larger than that of h@xmath3 s(1 ) ( @xmath151 ) .
the same result can be deduced from the integrated nuclear spectrum in fig .
[ f : ksp ] , where the 2.058 and br@xmath5 lines are shown to be broader than the h@xmath3 lines ( see also fwhm values in table [ t : flux ] ) .
this larger nuclear velocity dispersion of the 2.058 and br@xmath5 emission compared to the h@xmath3 emission indicates that the and br@xmath5 emission arises at least partially from the inner portions of the circum - nuclear disc .
these inner parts of the disc are expected to be dominated by high gas velocities that manifest themselves in the form of broad line profiles as a consequence of psf smearing .
[ f : feiifit ] shows that the [ ] 1.644 emission in the core of ngc 1275 is also characterized by a larger velocity dispersion
i.e. a broader line profile in fig .
[ f : hsp ] and table [ t : flux ] than the h@xmath3 emission .
the velocity dispersion displayed by the nuclear [ ] 1.644 emission is comparable to the value found for the 2.058 and br@xmath5 emission .
this suggests that the core of ngc 1275 on scales smaller than the nifs resolution also contains [ ] 1.644 -emitting gas with large velocities from ordered or turbulent motion , which may be associated with the disc rotation or with jet - gas interactions .
the excitation mechanism for the h@xmath3 and [ ] emission in the centre of ngc 1275 is analysed based on the line flux measurements given in table [ t : flux ] .
@xcite and , more recently , @xcite and @xcite have analysed the h@xmath3 excitation mechanism in the nucleus of ngc 1275 .
all find it to be thermally excited .
@xcite find that the excitation temperature is @xmath152 k from low - excitation transitions , with a higher temperature of @xmath153 k applying to high - excitation lines , while @xcite have used the ratio of the h@xmath3 lines to determine a vibration temperature of @xmath154 k , and a rotational temperature of @xmath155 k. population levels in the spectra integrated over a 1.0 arcsec - diameter aperture ( black symbols ) and a 0.51.0 arcsec - diameter annulus ( red symbols ) centred on the nucleus of ngc 1275 .
relative upper - level population number densities are plotted versus the upper - level energy , expressed as a temperature .
h@xmath3 1 - 0
( _ circles _ ) , 2 - 1 ( _ squares _ ) , and 3 - 2 ( _ triangle _ ) vibrational transitions are shown .
error bars are calculated for a constant flux error of @xmath156 erg s@xmath23 @xmath36 for all lines .
the dashed straight line corresponds to the excitation temperature of 1360 k derived by @xcite for transitions with @xmath157 k , normalized here to the h@xmath3 1 - 0 s(1 ) transition in the 1.0 arcsec - diameter aperture .
the solid straight line shows a least - squares fit to the transitions with @xmath158 k from the 1.0 arcsec - diameter aperture.[f : h2pop ] ] the relative h@xmath3 level populations for the central 1.0 arcsec - diameter aperture centred on the nucleus of ngc 1275 are shown in fig .
[ f : h2pop ] , together with the corresponding measurements for a 0.51.0 arcsec - diameter annulus .
the level populations are calculated from h@xmath3 line fluxes measured from the integrated @xmath30-band and @xmath31-band spectra ( figs [ f : hsp ] and [ f : ksp ] ) and listed in table [ t : flux ] . at densities above @xmath159 @xmath160
, the h@xmath3 gas is expected to be collisionally excited and so will be in thermal equilibrium .
the level populations are then defined by an excitation temperature , @xmath161 , that will be similar to the local gas temperature and can rise to several thousand kelvin through heating by shocks or uv and x - ray irradiation .
the relative level populations are derived by the boltzman equation such that @xmath162 , where @xmath163 is the line flux , @xmath164 is the line wavelength , @xmath165 is the spontaneous emission coefficient , @xmath166 is the statistical weight of the upper level of the transition , and @xmath167 is the energy of the upper level of the transition .
these are plotted in the population diagram in fig .
[ f : h2pop ] versus the upper - level energy .
transitions for a gas in thermal equilibrium lie along a straight line in this diagram with a slope of @xmath168 . as found by @xcite and @xcite , h@xmath3 transitions from the core of ngc 1275 with @xmath157 k are characterized by an excitation temperature of @xmath169 k , while transitions from higher energy levels are characterized by a hotter excitation temperature . the dashed straight line in fig .
[ f : h2pop ] corresponds to the @xmath170 k value found by @xcite for low - energy transitions .
our data are well fit by this temperature , although our low energy data are not as extensive as those of @xcite who measured several q - branch transitions that lie redwards of our spectra .
a least - squares fit to only higher energy transitions ( @xmath171k ) gives an excitation temperature of @xmath172 k ( solid straight line in fig . [ f : h2pop ] ) .
@xcite derive a comparable value of 3100 k from their data . a significant contribution from fluorescent excitation of the h@xmath3 emission
can be ruled out based on the flux ratio of the high - energy h@xmath3 2 - 1 s(1 ) line and the low - energy h@xmath3 s(1 ) line , shown in a pixel - by - pixel comparison in fig .
[ f : h2rat ] .
the ratio is spatially uniform at a value of about 0.08 - 0.09 .
following the diagnostic diagram in fig .
1a of @xcite , this value overlaps with the thermal excitation models but is much lower than the typical value of 0.5 - 0.6 predicted for fluorescent excitation .
furthermore , the lack of spatial variation in the line ratio over the nuclear region contradicts a scenario with fluorescent excitation .
if the higher excitation temperature of the high - energy levels were due to irradiation of the gas by nuclear uv or x - ray photons , this effect would then be expected to be stronger close to the nucleus . in general , no significant change is observed in the behaviour of the high - energy transitions in our data between the 1.0 arcsec - diameter circular aperture and a 0.51.0 arcsec - diameter annulus .
this can be seen in fig .
[ f : h2pop ] , where the low- and high - energy transitions corresponding to the annulus ( red symbols ) display the same slope as those corresponding to the circular aperture ( black symbols ) . 2 - 1 s(1 )
2.248 flux to h@xmath3 s(1 ) 2.122 flux .
each line is integrated over the velocity range @xmath173 .
an average continuum has been subtracted in each case .
the location of the continuum emission peak is shown by a cross .
north is up and east is to the left .
the scale bar indicates 100 pc.[f : h2rat ] ] the likely scenario for the excitation of the h@xmath3 emission in the nuclear region of ngc 1275 is shocks . in fig .
[ f : h2poporion ] , we compare the two straight lines corresponding to the excitation temperatures of the low and high - energy transitions in our h@xmath3 data for ngc 1275 to h@xmath3 data for shocked gas in the orion molecular outflow , taken from @xcite .
the shocked h@xmath3 gas in orion shows the same properties as our h@xmath3 data for ngc 1275 in terms of a higher excitation temperature for higher - energy transitions , presumably arising in post - shock gas having a range of temperatures .
a range of excitation temperatures with higher temperatures for higher - energy transitions is also characteristic of the mid - infrared h@xmath3 excitation diagrams of ngc 1275 and other h@xmath3-luminous radio galaxies , which @xcite find to be consistent with shock - excitation .
therefore , we conclude that shocks are the dominant excitation mechanism for the h@xmath3 emission in the central 1.0 arcsec of ngc 1275 . in the central 1 arcsec of ngc 1275 with h@xmath3 data for shocked gas in the orion molecular outflow from @xcite .
the straight lines are reproduced from fig .
[ f : h2pop ] .
they correspond to the derived excitation temperatures in ngc 1275 of 1360 k for transitions with @xmath157 k ( dashed line ) and 4290 k for transitions with @xmath158 k ( solid line ) .
the data points are taken from @xcite and represent shocked h@xmath3 gas in the orion molecular outflow .
the symbols are the same as in fig .
[ f : h2pop ] .
[ f : h2poporion ] ] [ [ s : res : feiiexc ] ] [ ] ~~ several authors have attempted to understand the nature of [ ] excitation in ngc 1275 ( e.g. * ? ? ?
the ionization potential of fe@xmath174 ( 16.2 ev ) is similar to that of hydrogen , so [ ] emission tends to arise in partially ionized regions .
being due to forbidden transitions between low - energy levels , the intensities of the near - infrared [ ] emission lines are principally dependent on the electron density .
calculations of the statistical equilibrium of collisional and radiative transitions among the lowest 16 levels of have been performed by several authors to study seyfert galaxies ( e.g. , * ? ? ?
* ; * ? ? ?
* ) and young stars ( e.g. , * ? ? ?
* ; * ? ? ?
the energy level diagram for [ ] has been presented by these authors . including only the lowest four terms
is justified at the electron temperatures involved , which are likely to be @xmath175 and leave higher energy levels significantly less populated .
we have repeated this 16-level atom calculation for using level energies from @xcite , radiative transition probabilities from @xcite , ( as listed by * ? ? ?
* ) , and collision strengths from @xcite and @xcite .
the model results for a range of [ ] emission - line combinations are shown in fig .
[ f : feiidens ] along with the empirical ratios and their uncertainties for ngc 1275 measured from our integrated nuclear spectrum ( fig . [
f : hsp ] ) .
the line ratios indicate a density in the [ ] region of @xmath176 .
emission - line ratios based on the a@xmath177p@xmath178-a@xmath177d@xmath179 1.7489 transition .
models are computed for a range of electron temperatures and electron densities , as indicated .
the shaded region corresponds to the measured 1.644 /1.533
ratio and its uncertainty in the integrated @xmath30-band spectrum of ngc 1275 ( fig .
[ f : hsp ] ) .
the dashed line and arrow correspond to the limit placed on the 1.644 /1.749
ratio by the same data.[f : feii1749 ] ] emission - line ratios based on the a@xmath177p@xmath180-a@xmath177f@xmath181 0.8617 transition .
models are computed for a range of electron temperatures and electron densities , as indicated .
the horizontal shaded region corresponds to the measured 1.644 /1.533
ratio and its uncertainty in the integrated @xmath30-band spectrum of ngc 1275 ( fig .
[ f : hsp ] ) .
the vertical dashed line and arrow correspond to the limit placed on the 1.257 /0.862
ratio from @xcite.[f : feii0862 ] ] deriving the electron temperature in the [ ] region is more difficult .
the near - infrared [ ] lines arise from transitions between the lowest three terms ( a@xmath182d , a@xmath177f , and a@xmath177d ) .
the transitions detected in our @xmath30-band spectra all occur between the a@xmath177d and a@xmath177f terms and so have a limited range of upper - level energies .
the @xmath183-band transitions of [ ] ( e.g. , at 1.257 ) share the same upper energy levels , so do not provide temperature discrimination .
several authors have suggested using shorter wavelength transitions from the a@xmath177p term or longer wavelength transitions between the a@xmath177f and a@xmath182d terms to constrain the electron temperature in seyfert galaxies @xcite and young stars @xcite .
this is the approach that we adopt here .
we begin by using near - infrared transitions from the a@xmath177p energy level to constrain the electron temperature .
the a@xmath177p@xmath178-a@xmath177d@xmath179 transition occurs at 1.7489 in the @xmath30-band so could make a nearly reddening - independent indicator with the 1.644 line .
the predicted dependence of this ratio as a function of electron temperature is shown in fig . [
f : feii1749 ] . in practice , the [ ] 1.7489
line coincides in wavelength with the h@xmath3 1 - 0 s(7 ) 1.7480 line , which is strong in ngc 1275 .
consequently , we have assumed that the [ ] 1.7489 line is not detected in our @xmath30-band spectrum and have defined a 3@xmath12 upper limit on the 1.644 /1.749
ratio accordingly .
this is indicated in fig .
[ f : feii1749 ] by the dashed line and solid arrow .
this limit does not provide a useful constraint on the electron temperature .
the [ ] a@xmath177p@xmath178-a@xmath177f@xmath178 0.9474 line , which has been suggested as an electron temperature indicator for young stars @xcite , is similarly overwhelmed by strong [ ] 0.9531 emission in ngc 1275 @xcite .
@xcite for seyfert galaxies and @xcite for young stars have both suggested using the [ ] a@xmath177p@xmath180-a@xmath177f@xmath181 0.8619 line strength as an indicator of electron temperature .
we choose to compare this line to the [ ] 1.257 line .
this avoids reddening and aperture size issues to the greatest extent possible .
our model predictions for the dependence of this line strength on electron temperature are shown in fig .
[ f : feii0862 ] .
these results are in good agreement with the similar calculation of @xcite .
we use the flux of the [ ] 1.257 line from @xcite .
a prominent emission line is also present at a rest - frame wavelength of @xmath184 in the @xmath185/@xmath183-band spectrum of ngc 1275 presented by @xcite .
the feature is at a significantly different wavelength to the 8446 line seen in several others of their objects and to the closest hydrogen recombination lines ( p13 at 0.8667 and p14 at 0.8601 ) , and also the triplet emission at 0.8498 , 0.8542 , and 0.8662 .
we estimate the line flux to be @xmath186 erg s@xmath23 @xmath36 .
if this is a detection of the [ ] line , it indicates that the electron temperature in the [ ] -emitting region is @xmath187 ( fig .
[ f : feii0862 ] ) .
otherwise , this represents an upper limit to the electron temperature .
[ ] 0.8617 emission has been seen previously in the orion nebula and in the seyfert galaxy ngc 4151 @xcite , which strengthens the case for a detection in ngc 1275 .
possible excitation mechanisms for the [ ] emission in ngc 1275 are x - ray heating or fast shocks .
@xcite argue that [ ] excitation by x - ray heating ( i.e. , a power - law photoionizing continuum ) leads to higher electron temperatures ( @xmath188 ) than shock heating ( @xmath189 ) . to the extent that their models are applicable , the elevated electron temperature in ngc 1275 is more indicative of x - ray heating than shock excitation .
the high [ ] 1.644 / br@xmath5 2.166 ratio of @xmath190 in ngc 1275 also supports x - ray heating as the dominant excitation mechanism in the nuclear spectrum integrated over a 0.5 arcsec aperture .
this ratio is a measure of the relative volumes of the partially - ionized region emitting [ ] and the fully - ionized region emitting .
it is expected to be in the range 0.11.4 in starburst regions @xcite , but can have values up to @xmath191 in environments where x - rays penetrate deeply into neutral gas @xcite .
the high value of this ratio in ngc 1275 and the proximity of the [ ] emission to the nucleus suggest that this is the process that excites the [ ] emission in ngc 1275 .
we can also use the @xmath183-band data of @xcite to locate ngc 1275 in the @xmath192 ( [ ] 1.257 ) /@xmath192 ( [ ] 1.03 ) versus @xmath192 ( [ ] 1.533 ) /@xmath192 ( [ ] 1.644 ) diagram of @xcite .
they propose this as a diagnostic of fe depletion on dust grains in the jets of young stars .
according to table 5 in @xcite , the ratio [ ] 1.257 / [ ] @xmath193 in ngc 1275 is 0.5 .
in contrast , galactic herbig - haro objects show a ratio of 3.0 @xcite . these data ,
therefore , suggest that fe is far more depleted in ngc 1275 than in galactic herbig - haro objects .
this is consistent with other estimates for seyfert galaxies ( e.g. , * ? ? ?
* ) where the strength of the [ ] emission is interpreted to be primarily due to the ionization conditions , rather than to significant dust grain destruction in shocks .
the nifs data for the circum - nuclear region of ngc 1275 suggest predominantly shock - excited molecular - hydrogen emission .
this emission traces a rotating disc , which , according to @xcite , is likely to be unstable or to have a clumpy structure , as well as one or several streamer - like features . a possible interpretation of these results is an accretion scenario in which gas loses orbital energy as a result of shocks and turbulent dissipation , leading to a net mass inflow and to a net increase in binding energy of the accretion flow . our model for this scenario
is based on the model suggested for the ionized accretion disc in the centre of m87 by ( * ? ? ?
* see their equations ( 7 ) and ( 8) ) . in this model , at each radius in the accretion flow , the net increase in binding energy is matched by the shock luminosity from radiative losses , @xmath194 : @xmath195 , \label{shocklum1}\ ] ] where @xmath196 is the gravitational potential and @xmath197 is the net mass accretion rate including the inflow rate _ minus _ any mass outflow in a disc wind .
if the fraction of the total shock luminosity that is radiated into the h@xmath3 line is given by @xmath198 , then @xmath199 first , we consider the case that the gravitational potential in equation ( [ shocklum1 ] ) is dominated entirely by the central massive object ( @xmath200 ) . in this case , the net gain in binding energy in the accretion flow strongly depends on the radius and increases steeply towards small radii . combining equations ( [ shocklum1 ] ) and ( [ shocklum2 ] ) , results in @xmath201 if we assume that the net mass accretion rate @xmath202 is approximately constant ( as is likely in the outer parts of the accretion disc ) and also that the ratio @xmath203 for radiation into the line is approximately constant .
this means that the radial variation in surface brightness should follow a steep scaling law with radius , proportional to @xmath204 : @xmath205 the assumption of constant accretion rate and radiative efficiency per h@xmath3 molecule is clearly not valid for the innermost parts of the accretion disc . in these regions , presumably , disc winds lower the net rate of mass accretion , h@xmath3 molecules are dissociated , and the critical density may be exceeded .
all these effects lead to a flattening of the h@xmath3 surface - brightness profile at small radii .
this can be expressed in the form of a core radius so that @xmath206 where @xmath207 is the amplitude and @xmath208 is a core radius .
now , we consider the case that the gravitational potential in equation ( [ shocklum1 ] ) is dominated entirely by an extended stellar mass distribution . in this case
, the enclosed mass at each radius increases with radius .
the corresponding gravitational potential will lead to an accretion - powered h@xmath3 surface - brightness profile that is flatter than the one in equation ( [ sb ] ) . as the agn dominates the near - infrared continuum emission of ngc 1275 in the circum - nuclear region ,
the detailed stellar distribution within the psf of the agn is uncertain .
however , in order to illustrate the accretion - flow scenario in a gravitational potential that is purely dominated by the stellar mass distribution , we refer to the stellar mass - distribution model used by @xcite .
@xcite fitted the stellar surface - brightness profile , obtained from an hst nicmos f160w image of ngc 1275 , at radii @xmath209 arcsec , excluding the innermost agn - dominated radii .
they find that the stellar surface - brightness profile can be modelled with a modified hubble profile with a core radius of @xmath210 arcsec .
for a constant mass - to - light ratio , the modified hubble profile corresponds to a deprojected stellar density of @xmath211^{-3/2 } \label{eq : stelldens}\ ] ] ( cf .
our nifs observations probe radii which are small compared to the above core radius of @xmath210 arcsec . on these scales ,
the stellar density from equation ( [ eq : stelldens ] ) is approximately constant with radius ( @xmath212 ) .
this means that the enclosed stellar mass increases proportional to @xmath213 and that the gravitational potential @xmath214 is proportional to @xmath215 . according to equations ( [ shocklum1 ] ) and ( [ shocklum2 ] ) , this type of gravitational potential results in an accretion - powered h@xmath3 surface - brightness profile that is constant with radius : @xmath216 in order to compare the above accretion - flow model for the h@xmath3 emission with the data for ngc 1275 , we fit the observed h@xmath3 surface - brightness profile with a generalized form of equation ( [ eq : h2profile ] ) , in which the power - law index @xmath16 is included as a fit parameter : @xmath217 for @xmath218 , this function is equal to the model for the h@xmath3 surface - brightness profile in a gravitational potential that is entirely dominated by the central massive object ( equation ( [ eq : h2profile ] ) ) . for @xmath219 ,
this function is equal to equation ( [ eq : h2profilestellar ] ) , which shows the example of an accretion flow in the gravitational potential of an extended stellar mass distribution . in order to take into account psf smearing ,
equation ( [ eq : h2fit ] ) is convolved with the psf model shown in fig . [
f : psf ] before it is fitted to the observed data .
s(1 ) emission of ngc 1275 shown in fig .
[ h2]a . the surface - brightness profile ( small plus signs )
is obtained over an azimuthal range of @xmath220 , avoiding the perturbed region south - west of the nucleus ( see fig . [ h2]a and b ) .
the large diamonds show the linear interpolation of the profile onto a regular grid in radius .
the solid red line shows the fit to the h@xmath3 profile using equation ( [ eq : h2fit ] ) convolved with the psf model shown in fig .
[ f : psf ] .
the fit results in @xmath221 and @xmath222 arcsec.[f : h2prof ] ] the observed h@xmath3 surface - brightness profile for the inner 1 arcsec ( 358 pc ) of ngc 1275 is displayed in fig .
[ f : h2prof ] . the profile ( small plus signs )
is measured over an azimuthal range of @xmath220 ( in order to avoid the streamer to the south - west of the nucleus ) and mapped onto a regular grid in radius via linear interpolation ( large diamonds ) .
if the inclination of the h@xmath3 disc caused significant projection effects , i.e. if the ellipticity of the isophotes were significant , the measurements for each surface brightness level would be distributed over a range of projected radii .
as such projection effects are negligible at @xmath223 arcsec ( 225 pc ) in fig .
[ f : h2prof ] ( small crosses ) , we do not deproject the profile .
more significant projection effects are only seen at @xmath224 arcsec .
but the h@xmath3 distribution on these scales is likely to be more complex than the disc configuration inferred from the rotation pattern for @xmath225 pc .
the fit of the psf - convolved function of equation ( [ eq : h2fit ] ) to the observed h@xmath3 surface - brightness profile results in @xmath222 arcsec and @xmath221 ( red line in fig .
[ f : h2prof ] ) .
this profile is slightly flatter than the profile expected for an accretion flow in a gravitational potential dominated by a central point mass .
we , therefore , conclude that the observed h@xmath3 surface - brightness profile in ngc 1275 is in general agreement with the above accretion - flow model when taking into account that the gravitational potential has contributions from an extended mass distribution in the inner 1 arcsec ( 358 pc ) .
@xcite estimated that the enclosed stellar mass in ngc 1275 is about 1 per cent of the total dynamical mass at @xmath226 pc ( 0.14 arcsec ) .
it is therefore negligible on the scales that are characterized by the regular disc rotation ( see section [ s : kinematics ] ) . as the surface - brightness profile
is fitted up to larger scales , we use the model of @xcite in order to deduce that the enclosed stellar mass increases by a factor of ten from @xmath226 pc ( 0.14 arcsec ) to @xmath227 pc ( 0.3 arcsec ) and by a factor of 100 from @xmath226 pc ( 0.14 arcsec ) to @xmath228 pc ( 0.67 arcsec ) .
in addition to the stellar mass , the molecular gas mass is likely to contribute significantly to the total mass in the centre of ngc 1275 .
we derive a tentative estimate for the amount of molecular gas in the central @xmath2 pc of ngc 1275 by comparing our nifs data to the large - scale h@xmath3 s(1 ) and co observations by @xcite and @xcite .
@xcite report a total h@xmath3 s(1 ) flux of the ngc 1275 nebula ( nucleus plus filaments ) of @xmath229 .
according to our nifs data , of the order of 1 per cent of this flux originate from the central @xmath230 pc . assuming that this flux ratio translates into a mass ratio and that the same ratio applies to the near- , mid - infrared , and cold molecular gas , we estimate the cold molecular gas mass in the central @xmath230 pc to be about 1 per cent of the total cold molecular gas mass of the ngc 1275 nebula ( @xmath17 , * ? ? ?
this results in a cold molecular gas mass of @xmath231 , which is about two orders of magnitude larger than the stellar mass estimate for the same region .
as the h@xmath3 emission in the inner @xmath2 pc of ngc 1275 shows evidence for a rotating disc , the enclosed mass of the disc can be inferred from kinematic modelling . if the potential of the rotating disc is dominated by the central supermassive black hole
, the enclosed mass provides a direct estimate of the black - hole mass .
for ngc 1275 , a black - hole mass estimate of @xmath232 m@xmath233 dex was reported by @xcite , using the observed jump in h@xmath3 velocity across the nucleus as an indication of disc rotation .
their data , based on seeing - limited uist ( ukirt imaging spectrometer ) observations , lack the resolution to reconstruct the full two - dimensional intensity and velocity profiles of the disc .
their black - hole mass estimate is , therefore , based on a single value for the nuclear velocity discontinuity of @xmath234 resolved on a radial scale of 50 pc ( @xmath210.15 arcsec ) .
another estimate for the black - hole mass of ngc 1275 was given by @xcite .
using the @xmath11-@xmath12 relation and a velocity dispersion of @xmath235 , they obtain an indirect mass estimate of @xmath236 .
we have modelled the h@xmath3 emission for the inner @xmath237 pc of ngc 1275 three - dimensionally , using a simulated nifs data cube of a thin , transparent disc of h@xmath3 in keplerian rotation around a central point mass .
the model disc is first created in polar coordinates by assigning each radial and azimuthal sampling point a gaussian line component .
the central wavelength of the line component is given by the circular velocity at that radius , as determined from the keplerian rotation curve .
the line width is set to the nifs spectral resolution of @xmath238 , i.e. the line components are assumed to be intrinsically unresolved .
as discussed in section [ s : accretion ] , the observed h@xmath3 surface - brightness profile does not show any significant projection effects on @xmath2 pc scales .
therefore , we define the line flux at each radius in the face - on disc using equation ( [ eq : h2fit ] ) with the parameters determined from the fit in fig .
[ f : h2prof ] .
the resulting model disc is projected spatially onto an oversampled cartesian grid of pixels .
this step involves summing up all spectra that fall into each pixel for the given input inclination and the input position angle of pa @xmath239 ( see section [ s : kinematics ] ) .
finally , in order to simulate the nifs data , each two - dimensional slice of the model data cube is convolved spatially , using the psf from section [ s : reduction ] , and rebinned onto the rectangular nifs pixel scale .
the output is a full @xmath240 arcsec@xmath1 nifs data cube of an `` observed '' model disc . for the actual comparison between the model and the data via @xmath241 values , we only consider spectral pixels from the central @xmath242 arcsec@xmath1 ( @xmath243 ) around the position of the nucleus of ngc 1275 ( on which the model disc is centred ) .
this @xmath243 section covers the region with the most regular disc - type velocity field in the observations ( see fig . [
f : h2fit ] ) . the surface - brightness profile ( fig .
[ f : h2prof ] ) does not show a sharp cut - off at the edges of the kinematically defined disc but rather a smooth decline in h@xmath3 surface brightness over the full nifs field - of - view .
therefore , we use an intrinsic size of the model disc that is much larger than the @xmath243 region and extends well beyond this region for all projection angles considered here . for numerical purposes , the model disc is also truncated at small radii in order to avoid infinite velocities when the radius in the keplerian disc approaches zero .
we choose an inner cut - off radius of 1 pc .
this corresponds to the minimum radius at which h@xmath3 molecules are likely to survive the nuclear radiation field , as outlined in section [ sec : ionization ] .
[ f : chi2test ] shows the @xmath241 values for a comparison between the observed h@xmath3 s(1 ) lines and a grid of models with varying central mass and intrinsic disc inclination in the central @xmath242 arcsec@xmath1 ( @xmath243 ) around the core of ngc 1275 . in this figure
, the absolute flux scale of each model has been adjusted independently to the value that fits the observed h@xmath3 lines in the central @xmath242 arcsec@xmath1 best . as an example
, the line profiles from one of the best models ( @xmath26 and @xmath244 ) are compared to the observed line profiles in fig .
[ f : chi2plot ] .
it is obvious that the observed line profiles in the spectra to the west of the nucleus show a redshifted wing , which is not present in the models for pure disc rotation .
this redshifted wing is part of the streamer discussed in sections [ s : morphology ] and [ s : kinematics ] .
data for a grid of models with varying central mass and inclination .
the reduced @xmath241 values are calculated based on a three - dimensional comparison between the modelled and observed data cubes .
spectrally , the @xmath241 is computed over a velocity range from @xmath245 to @xmath246 around the rest wavelength of h@xmath3 in ngc 1275 .
spatially , the @xmath241 values are computed for boxes of @xmath247 non - square nifs pixels ( i.e. @xmath242 arcsec@xmath1 or @xmath243 ) centred on the nucleus , as displayed in fig .
[ f : chi2plot ] .
the white contour level shows @xmath248 with respect to the minimum @xmath241 .
the black vertical lines mark the inclination range of @xmath9 assumed for the black - hole mass estimate.[f : chi2test ] ] fig .
[ f : chi2test ] shows that the enclosed mass of the best - fitting model depends on the disc inclination .
this is due to the degeneracy between central mass and inclination in reproducing the same line - of - sight velocities in the keplerian disc .
small @xmath241 values populate a valley extending from a central mass of about @xmath249 at an inclination of @xmath250 to a central mass of about @xmath251 at an inclination of @xmath252 .
the global lower mass limit for the central mass of @xmath253 in fig .
[ f : chi2test ] is robust against a possible underestimate of the psf width in our @xmath31-band observations . by increasing the fwhm of the model psf ,
the best - fitting models at a fixed inclination are shifted toward higher central masses .
this is a consequence of enhanced psf smearing which causes the central velocity gradient in the models to decrease , so that a higher central mass is required in order to fit the observed velocities .
@xcite use an inclination of @xmath254 in their estimate for the dynamical mass in ngc 1275 .
this value is based on the assumption that the disc rotation axis has the same orientation as the axis of the radio jet , which is inclined by @xmath255-@xmath256 with respect to the line - of - sight @xcite . for a similar disc inclination of @xmath9
, our model suggests a best - fitting central mass of @xmath257 within the contour level shown in fig .
[ f : chi2test ] .
this is larger than the @xmath232 m@xmath25 derived by @xcite .
this discrepancy is a consequence of mainly two factors : first , the higher resolution nifs data resolve a larger velocity discontinuity across the nucleus .
secondly , in contrast to the calculation by @xcite , our three - dimensional analysis takes into account psf smearing , which leads to a decrease in the observed velocities close to the nucleus .
our value for the central mass is in agreement with the @xmath11-@xmath12 relation within the rms scatter . assuming a stellar velocity dispersion of @xmath258 for ngc 1275 ( see * ? ? ?
* ) , the updated @xmath11-@xmath12 relation for all morphological types from @xcite , @xmath259 $ ] results in @xmath260 , which is consistent with our value within the rms scatter of 0.43 dex reported by @xcite .
@xcite show that the stellar contribution to the total dynamical mass within @xmath2 pc is negligible and , therefore , argue that the dynamical mass represents the black - hole mass .
however , as shown in section [ s : accretion ] , we estimate that a large reservoir of molecular gas of the order of @xmath231 is present in the central parts of ngc 1275 .
this large gas mass could potentially contribute significantly to the derived enclosed mass .
in fact , if this gas mass is subtracted from the total enclosed mass of @xmath10 , the resulting black - hole mass of @xmath261 is almost exactly the same as the one derived from the @xmath11-@xmath12 relation .
however , as this result involves large uncertainties , we conclude by merely noting that our estimate of the enclosed mass may rather represent an upper limit for the black - hole mass . using _ spitzer _ observations of pure rotational h@xmath3 transitions in
the mid - infrared , @xcite detect warm molecular hydrogen emission in 30 per cent of their sample of @xmath262 3c radio galaxies , including ngc 1275 .
these molecular hydrogen emission galaxies ( mohegs ) are found among radio galaxies of both fanaroff
riley classes and among cool core clusters .
the sample studied by @xcite shows that radio mohegs as a class typically have low star - formation rates despite the large h@xmath3 masses , which indicates that turbulence prevents the gas from efficiently forming stars .
furthermore , most mohegs harbour low - luminosity agn that can not provide sufficient x - ray heating for powering the observed h@xmath3 emission .
the agn in ngc 1275 ( @xmath263 ) belongs to the more x - ray luminous agn among the radio mohegs ( cf .
17 and table 9 in * ? ? ?
this means that x - ray heating may make a stronger contribution in ngc 1275 than in many x - ray faint mohegs .
however , fig .
17 in @xcite shows that the ratio of the observed warm h@xmath3 luminosity to x - ray luminosity for ngc 1275 , i.e. @xmath264 , is still a factor of 7 higher than the maximum ratio of 0.01 derived for an x - ray dissociation region .
therefore , even in ngc 1275 , x - ray heating by the agn is unlikely to be the primary excitation mechanism for h@xmath3 . instead , @xcite
conclude that the most likely excitation mechanism for the h@xmath3 emission in radio mohegs is shocks ( or , less likely , cosmic ray heating ) .
the same mechanism is suggested for ngc 1275 , based on the nifs data discussed in this paper ( section [ s : res : h2exc ] ) .
this raises the question whether these shocks originate from radio - jet feedback , as suggested for the overall sample of radio mohegs by @xcite , or from gas accretion , as discussed for the molecular disc in ngc 1275 in section [ s : accretion ] . in the feedback scenario ,
shocks are driven into dense clouds in a porous interstellar medium by the hot cocoon powered by the radio jet , as for example shown in the simulations by @xcite and @xcite . in the accretion scenario , gas clumps are repeatedly shocked while falling into the centre of the galaxy so that gravitational potential energy is converted into shocks and turbulent heating . in a case study of 3c 326
n , a @xmath265 radio moheg with a very low star - formation rate and an x - ray faint agn , @xcite suggest a tidally - induced accretion flow as the most likely origin of the shocks , while @xcite favour the jet feedback scenario .
simple energy considerations for ngc 1275 show that either of the processes , radio jet feedback or gas accretion , is sufficient to power the observed mid- or near - infrared h@xmath3 emission . for
the radio jet , @xcite report a jet cavity power of @xmath266 , which is almost 3 orders of magnitude larger than the luminosity of the warm h@xmath3 emission ( @xmath267 , cf . * ? ? ?
for the accretion process , we estimate an accretion luminosity of the order of @xmath268 , which is about one order of magnitude larger than the warm h@xmath3 luminosity .
this accretion luminosity is derived by dividing the potential energy of the infalling gas of about @xmath269 by the dynamical free - fall time of @xmath270 yrs , in analogy to @xcite . for this calculation
we assume that the total cold gas mass of @xmath17 @xcite falls back from an average distance of 30 kpc into the potential well corresponding to a stellar mass of @xmath271 .
this stellar mass is derived from the @xmath272-@xmath273 relations at @xmath274 by @xcite , using the @xmath275 from section [ s : model ] as the black - hole mass . as neither jet feedback nor gas accretion stands out as the dominant energy source in ngc 1275 ,
it is likely that the observed properties of gas excitation and kinematics are a consequence of both processes .
rotating molecular discs , as found in the circum - nuclear region of ngc 1275 , or on scales of 3 kpc in 3c 326 n @xcite , suggest that accretion processes play a role . since the radio jet can not impart angular momentum on a gaseous disc , disc rotation is a sign of gas infall .
this gas inflow may be associated with tidal galaxy - galaxy interactions , or with accretion from the halo .
these processes have been discussed for the case of 3c 326 n @xcite . for ngc 1275 ,
it is most likely that the gas inflows consist of gas which falls back from the radio jet bubbles @xcite .
such back flows of gas uplifted during the active jet phase are also seen in simulations @xcite .
tidally - induced gas inflows could result from an interaction with the high - velocity system .
but this scenario is less likely , since the high - velocity system is still at a large distance from ngc 1275 @xcite .
observational evidence suggests that turbulence is a common characteristic of the molecular gas component in radio mohegs .
@xcite found the molecular gas in the disc of 3c 326 n to be highly turbulent with line widths of fwhm @xmath276 .
broad h@xmath3 lines are also characteristic of the sample of radio galaxies with fast ionized / atomic outflows , studied by @xcite .
integrated over the full field - of - view , our nifs data of ngc 1275 show an h@xmath3 s(1 ) line width of fwhm @xmath277 , which is comparable to the above values .
a fraction of this line width is likely to be attributed to ordered streaming motions , which are characterized by small ( @xmath278 ) and largely positive velocities ( see fig .
[ f : h2vel ] ) . since the models for the circum - nuclear disc in ngc 1275 ( section [ s : model ] )
provide a good representation of the data without including any significant intrinsic line width , the velocity dispersion of the h@xmath3 s(1 ) lines are likely to be @xmath278 ( i.e. fwhm @xmath279 ) .
( the total widths of the h@xmath3 lines in the small - aperture spectra , listed in table [ t : flux ] , are much larger than that , but are dominated by the disc rotation . )
the fact that the dispersion - related line width in the disc is smaller than the total integrated line width in the nifs field - of - view suggests that the accretion flow is energetically capable of producing the turbulent motions in the disc .
alternatively , the turbulence may be driven by the radio jet and its hot cocoon .
turbulent heating via radio jet feedback can affect a large volume if the gas is distributed in the form of clumps in a porous interstellar medium @xcite .
turbulent heating of the molecular gas component by the jet interaction is only compatible with the observed disc configurations if the jet is , at the same time , inefficient in expelling the molecular gas from the disc . such a picture
is also suggested by the observations by @xcite , who find that in radio galaxies with fast ionized and atomic outflows the bulk of the mid - infrared h@xmath3 emission does not participate in the outflow , despite showing signs of turbulent heating .
simulations by @xcite , which probe the effects of a jet interaction with an almost keplerian and turbulently supported disc of a warm clumpy interstellar medium in an elliptical galaxy , show that the jet interaction is , indeed , inefficient in driving the gas out of the disc region .
however , there is a significant gas transport from the inner to the outer portions of the disc , which will eventually leave the inner portions depleted . if the disc is porous , the mass transfer rates are lower than in the case of a homogeneous disc of gas
. we estimate the porosity of the circum - nuclear disc in ngc 1275 by comparing the observed mean density of h@xmath3 molecules in the central @xmath2 pc with the density expected for the corresponding gas component from the ideal equation of state .
as shown in section [ s : accretion ] , we tentatively estimate the total molecular gas mass within a radius of @xmath2 pc to be @xmath231 .
the corresponding mean number density of molecules in the central volume is @xmath280 , where @xmath281 is the proton mass and @xmath282 is approximated by the spherical volume enclosed by a radius of 50 pc .
this value is similar to the clump density derived from the ideal equation of state , if we assume that the clumps are composed of cold molecular gas at a mean temperature of @xmath283 k and that they are in pressure equilibrium with the hot ism pressure of @xmath284 , following the scenario by @xcite .
the similarity between the mean observed density and the estimated clump density from the ideal equation of state suggests that the cold molecular gas , if distributed in clumps , is likely to have a high filling factor .
an independent measure of the actual gas density is provided by detection experiments for high - density gas tracers .
@xcite report the detection of hcn(3 - 2 ) in the centre of ngc 1275 , which indicates a gas density of at least @xmath285 .
this value agrees with the mean gas density estimated above .
but , as a lower limit for the actual gas density , it does not rule out the possibility of smaller filling factors
. detections of other high - density gas tracers are needed in order to obtain a better estimate of the filling factor .
the relative importance of accretion and jet feedback in exciting and heating the molecular gas in ngc 1275 and other radio mohegs is poorly constrained with the current data . improved observational parameters for the molecular gas phase
e.g. , from alma observations of high - density tracers together with refined simulations are required in order to explore the effects of the jet interaction with the interstellar medium in these molecular - hydrogen - rich radio galaxies .
the paper is based on near - infrared integral field spectroscopy of ngc 1275 , obtained with the nifs instrument at gemini north at a spatial resolution of @xmath286 arcsec .
we have presented a scenario in which the shock - excited molecular hydrogen , traced via the near - infrared ro - vibrational h@xmath3 lines , enters into the circum - nuclear region in the form of one or more streamers and settles into a turbulent , clumpy or unstable accretion disc close to the nucleus , where energy dissipation via turbulence and shocks drives the further accretion process towards the core .
the observed disc kinematics has been used as a tool to infer a direct estimate for the central mass enclosed by the disc via model data cubes of discs in keplerian rotation .
we have compared the results to other h@xmath3-luminous radio galaxies and discussed the relative importance of jet feedback and accretion as a mechanism for shock - excitation and heating of the molecular hydrogen . in summary ,
the main results presented in this paper are : * the circum - nuclear h@xmath3 s(1 ) and [ ] 1.644 lines in ngc 1275 show evidence for a disc on @xmath226 pc scales , both in morphology and kinematics .
the projected major kinematic axis of this disc is perpendicular to the projected axis of the radio jets . * a possible ionization structure within the disc
is discussed by comparing the h@xmath3 s(1 ) and [ ] 1.644 emission with the 2.058 and br@xmath5 lines .
the latter two lines are only found in a compact region around the core , but their kinematics is characterized by the same disc signature as seen in h@xmath3 s(1 ) and [ ] 1.644 .
it is suggested that both the 2.058 and br@xmath5 emission lines arise at least partially from the inner parts of the disc , because they display a larger central velocity dispersion than the h@xmath3 s(1 ) line .
future higher angular resolution observations are required to fully explore this ionization structure . * the h@xmath3 surface - brightness profile in the inner 1 arcsec around the core is well represented by a psf - convolved profile following an @xmath287 power law with an inner core radius of 0.17 arcsec .
this type of surface - brightness profile is found to be in general agreement with an accretion model in which the line emission matches the net increase in binding energy of gas falling into the combined gravitational potential of a central massive object and an extended mass distribution . * in h@xmath3 , the data reveal an elongated perturbation to the south - west of the nucleus , which is characterized by redshifted velocities .
this perturbation is interpreted as the strongest of several molecular gas streamers , which are likely to be falling into the core . *
the relative line strengths of the ro - vibrational h@xmath3 lines in the central 1 arcsec of ngc 1275 indicate thermal excitation of the molecular hydrogen via shocks . in agreement with previous studies ,
we find excitation temperatures of 1360 k and 4290 k for transitions with @xmath288 k and @xmath289 k , respectively . any significant contribution from fluorescent excitation
can be excluded based on the small ratio of the h@xmath3 2 - 1 s(1 ) 2.248 to the h@xmath3 s(1 ) 2.122 flux and its spatial uniformity over the nuclear region .
instead , the h@xmath3 excitation in ngc 1275 shows the same characteristics as shocked gas in the orion molecular outflow .
* the high electron temperature as well as the high [ ] 1.644 / br@xmath5 2.166 suggest that the [ ] lines in the integrated nuclear spectrum are excited by x - ray heating .
alternatively , the line emission can be excited by fast shocks .
models for a range of [ ] emission - line ratios are presented .
they indicate a density of @xmath8 in the [ ] emitting region .
the electron temperature in this region is less well constraint . using less certain detections of [ ] lines at shorter wavelengths from previous publications , an electron temperature of @xmath290 k is estimated . *
an estimate for the mass enclosed by the circum - nuclear h@xmath3 disc of ngc 1275 is derived by modelling the data via simulated nifs data cubes . assuming a disc inclination of @xmath9 , the enclosed mass is estimated to be @xmath291 .
interpreted as a direct measurement of the black - hole mass , this value is larger than previous black - hole mass estimates , but is consistent with the @xmath11-@xmath12 relation within the rms scatter . a tentative estimate of the molecular gas mass in the central region of ngc 1275 results in @xmath231 , which is a non - negligible fraction of the enclosed mass .
the enclosed mass may , therefore , represent an estimate for the upper limit of the black - hole mass . *
basic energy considerations show that the shock - excitation and turbulent heating of the molecular gas in ngc 1275 could be driven by the jet interaction , the accretion process , or a mixture of both .
an order - of - magnitude estimate of the distribution of the cold molecular gas in the circum - nuclear disc of ngc 1275 indicates a high filling factor .
improved constraints on the properties of the molecular gas phase can be expected from future observations ( e.g. , alma ) .
we are grateful to the anonymous referee for very useful comments and suggestions .
we thank alex wagner for a helpful discussion .
we acknowledge the tireless efforts of the nifs team from the australian national university , auspace , and the gemini observatory who made the nifs instrument reality . this work is based on observations obtained at the gemini observatory , which is operated by the association of universities for research in astronomy , inc .
, under a cooperative agreement with the nsf on behalf of the gemini partnership : the national science foundation ( united states ) , the science and technology facilities council ( united kingdom ) , the national research council ( canada ) , conicyt ( chile ) , the australian research council ( australia ) , ministrio da cincia , tecnologia e inovao ( brazil ) , and the ministerio de ciencia , tecnologa e innovacin productiva ( argentina ) .
j. s. , p. j. m. , and m. a. d. acknowledge the continued support of the australian research council ( arc ) through discovery projects dp0984657 , dp0664434 , and dp0342844 . | we report the results of high spatial and spectral resolution integral - field spectroscopy of the central @xmath0 arcsec@xmath1 of the active galaxy ngc 1275 ( perseus a ) , based on observations with the near - infrared integral field spectrograph ( nifs ) and the altair adaptive - optics system on the gemini north telescope .
the circum - nuclear disc in the inner @xmath2 pc of ngc 1275 is seen in both the h@xmath3 and [ ] lines .
the disc is interpreted as the outer part of a collisionally - excited turbulent accretion disc .
the kinematic major axis of the disc at a position angle of @xmath4 is oriented perpendicular to the radio jet . a streamer - like feature to the south - west of the disc , detected in h@xmath3 but not in [ ] ,
is discussed as one of possibly several molecular streamers , presumably falling into the nuclear region .
indications of an ionization structure within the disc are deduced from the and br@xmath5 emission lines , which may partially originate from the inner portions of the accretion disc .
the kinematics of these two lines agrees with the signature of the circum - nuclear disc , but both lines display a larger central velocity dispersion than the h@xmath3 line .
the rovibrational h@xmath3 transitions from the core of ngc 1275 are indicative of thermal excitation caused by shocks and agree with excitation temperatures of @xmath6 and @xmath7 k for the lower- and higher - energy h@xmath3 transitions , respectively .
the data suggest x - ray heating as the dominant excitation mechanism of [ ] emission in the core , while fast shocks are a possible alternative .
the [ ] lines indicate an electron density of @xmath8 .
the h@xmath3 disc is modelled using simulated nifs data cubes of h@xmath3 emission from inclined discs in keplerian rotation around a central mass .
assuming a disc inclination of @xmath9 , the best - fitting models imply a central mass of @xmath10 . taken as a black - hole mass estimate , this value is larger than previous estimates for the black - hole mass in ngc 1275 , but is in agreement with the @xmath11-@xmath12 relation within the rms scatter .
however , the molecular gas mass in the core region is tentatively estimated to be non - negligible , which suggests that the central mass may rather represent an upper limit for the black - hole mass . in comparison to other h@xmath3-luminous radio galaxies
, we discuss the relative role of jet feedback and accretion in driving shocks and turbulence in the molecular gas component .
[ firstpage ] galaxies : general galaxies : active galaxies : nuclei galaxies : individual ( ngc 1275 ) |
anyons are exotic quasiparticles that exhibit non - trivial exchange statistics and arise as low lying excitations of topological phases of matter .
there is a promising experimental program to observe anyons in condensed matter systems such as fractional quantum hall systems , majorana edge modes of nanowires , and two dimensional spin liquids.@xcite from a theoretical viewpoint , many - body systems of anyons offer a realm of new physics to explore . for example , just as chains of interacting integer spin systems describe different physics from half integer systems,@xcite interacting chains of anyons exhibit properties that depend on the topological charges and braiding and fusion rules of the corresponding anyon model.@xcite subsequently , several paradigmatic lattice models of interacting anyons have been proposed and studied to gain insight into the many - body physics of these particles.@xcite the framework of matrix product states @xcite has played an instrumental role in the study of lattice spin systems in recent decades , especially for systems in one spatial dimension . in particular , the mps forms
the basis of two highly successful simulation algorithms , namely , the density matrix renormalization group ( dmrg ) algorithm @xcite and the time - evolving block decimation ( tebd ) algorithm.@xcite the latter , along with its variations ( often collectively referred to as time - dependent dmrg @xcite ) , allows for efficient simulation of time evolution of lattice systems made of hundreds of sites , and also of systems with infinite size @xcite in the presence of translation invariance . in this paper
we generalize the mps formalism and the tebd algorithm for lattice systems of anyons
. a pure state @xmath2 } } \rangle}$ ] of a lattice made of @xmath3 spins @xmath4 can be expanded in a tensor product basis , @xmath5 } } \rangle } = \sum_{s_1s_2\ldots s_l } \psi^{\mbox{\tiny [ spin]}}_{s_1s_2\ldots s_l } { | s_1 \rangle}\otimes{| s_2 \rangle}\otimes\ldots{| s_l \rangle},\label{eq : spinstate}\ ] ] where @xmath6}}_{s_1s_2\ldots s_l}$ ] are complex coefficients and @xmath7 denotes the local basis for site @xmath8 . as described in ref .
, the mps decomposition of @xmath2 } } \rangle}$ ] corresponds to a decomposition of tensor @xmath6}}_{s_1s_2\ldots s_l}$ ] into a network of tensors that are interconnected as shown in fig .
[ fig : intro1 ] . an open or _
index @xmath9 in the mps labels the local basis @xmath10 for the spin at site @xmath8 .
index @xmath11 corresponds to the schmidt decomposition ( see appendix [ sec : appaa ] ) of state @xmath2 } } \rangle}$ ] , given according to the bipartition @xmath12:[i+1\ldots l]$ ] of the spins by @xmath5 } } \rangle } = \sum_{\mu_i } \lambda^{^{[i]}}_{\mu_i}~~ { | \phi^{^{[1\ldots i]}}_{\mu_i } \rangle } \otimes { | \phi^{^{[i+1 \ldots l]}}_{\mu_i } \rangle}.\label{eq : spinsd}\ ] ] here @xmath13}}$ ] is a diagonal matrix with non - negative diagonal entries @xmath13}}_{\mu_i } \geq 0 $ ] , and @xmath14}}_{\mu_i } \rangle}\}$ ] and @xmath15}}_{\mu_i } \rangle}\}$ ] are orthonormal bases for the two parts of the lattice .
the generalization of the mps formalism to lattice systems of anyons poses an interesting challenge .
a system of anyons is not necessarily completely specified by giving the `` topological charge '' @xmath16 associated with each site of the lattice . for many species of anyons
this information must be supplemented by the outcome of a number of non - local charge measurements , which can not be uniquely associated with single individual anyons .
thus , in contrast to a system of spins , a lattice system of @xmath3 anyons @xmath17 does not in general admit a description in terms of local hilbert spaces associated with the lattice sites , and nor does the global hilbert space admit a tensor product structure as in e.g. eq .
( [ eq : spinstate ] ) and eq .
( [ eq : spinsd ] ) .
instead a basis is introduced by means of a _ fusion tree _ , illustrated in fig .
[ fig : intro2 ] for a lattice of @xmath18 anyons @xmath19 .
a fusion tree specifies a sequence of pairwise fusions of the @xmath3 anyons ( associated with the open edges of the tree ) into a total anyon charge @xmath20 consistent with the _ fusion rules _ of the anyon model .
the fusion rules are a set of constraints on the charge outcomes arising from the fusion of an anyon pair .
if the charges @xmath21 associated with the lattice sites are fixed , an anyonic lattice still possesses non - local degrees of freedom which correspond to the set of possible charges @xmath22 that can appear on the internal edges of the fusion tree in arrangements consistent with the fusion rules.@xcite the lack of a description in terms of local hilbert spaces , and the subsequent necessity of a description in terms of a fusion tree , poses the key challenge in the direct simulation of anyonic systems . } } \rangle}$ ] of a lattice of @xmath18 spins @xmath23 , as described in ref . .
an open index @xmath9 labels a basis for the spin at site @xmath8 . a bond index @xmath11 is associated with a schmidt decomposition ( [ eq : spinsd ] ) of state @xmath2 } } \rangle}$ ] .
for example , here we indicate the schmidt decomposition for the bipartition @xmath24:[345]$ ] where @xmath25}}$ ] is a diagonal matrix with diagonal entries @xmath26}}}_{\mu_2}\geq 0 $ ] , and @xmath27}}_{\mu_2 } \rangle}$ ] and @xmath28}}_{\mu_2 } \rangle}$ ] are the schmidt bases for the two parts as described by the tensors to the left and to the right of index @xmath29 respectively.[fig : intro1],width=302 ] on the other hand , an anyonic lattice of this sort can be described by an enlarged hilbert space that is spanned by the tensor product basis @xmath30 .
this basis treats the intermediate charges @xmath31 as effective spin degrees of freedom , but will in general contain unphysical states which are non - compliant with the fusion rules and whose existence must be suppressed .
this mapping has been employed in dmrg @xcite and monte carlo @xcite studies of certain su(2)@xmath32 anyon models .
direct simulation of anyon systems , without employing a mapping to spins , has also been performed using exact diagonalisation for up to 37 anyons . @xcite more recently , the multi - scale entanglement renormalization ansatz ( mera ) @xcite has been adapted @xcite to the fusion tree description of anyons by using _
charge - conserving _ tensors that explicitly encode the fusion rules of the anyon model .
this formalism offers a much broader avenue for simulation of anyonic lattice systems as it allows direct access to specific physical charge sectors in the hilbert space for larger system sizes . in this paper
we describe how to adapt the mps formalism to the fusion tree description of anyonic lattice systems .
the anyonic mps corresponds to a tensor network which is connected in the same way as the mps for spin systems [ fig .
[ fig : intro1 ] ] , but which is made of charge - conserving tensors .
we also describe how to extend the tebd algorithm to the anyonic mps for the efficient simulation of time evolution under a local , anyonic charge - conserving hamiltonian .
anyons @xmath33 labelled by means of a _
fusion tree_. the latter corresponds to a sequence of pairwise fusing the anyons into a total anyon charge @xmath20 .
we have attached a redundant vacuum charge at the left of the fusion tree for convenience .
the degrees of freedom of the lattice correspond to sets of the @xmath34 intermediate charges @xmath35 that are compatible with the fusion rules of the anyon model.[fig : intro2],width=207 ] the key benefit of the anyonic mps is that , by working directly in the fusion tree description of an anyonic system , it can be applied to study any anyon model given the description of that model in terms of the following parameters ( see appendix [ sec : appa ] ) : 1 .
the set of allowed anyon types or topological charges @xmath36 2 .
the _ quantum dimension _ @xmath37 associated with each charge @xmath16 , analogous to the dimension of an irreducible representation in group theory .
3 . the fusion rules of the anyon model , encoded in the 3-index tensor @xmath38 , where @xmath39 is the number of copies of charge @xmath40 appearing in the fusion product @xmath41 .
4 . the three - index tensor @xmath42 which describes the braiding of two anyons . 5 . the 6-index tensor @xmath43 , which relates different ways to fuse together three anyons via a relationship known as an @xmath44-move . in this paper
, we will assume that all of these data are available for the anyon model of interest .
for example , our method can be used to simulate anyons models with quantum symmetry su(2)@xmath32 . in order to benchmark the anyonic tebd algorithm we study first the ground state of an infinite chain of ising anyons [ described by su(2)@xmath45 and of fibonacci anyons [ described by su(2)@xmath46 subject to antiferromagnetic interaction .
both these models are critical and well studied .
@xcite we then study the real time dynamics of an anyonic hubbard - type model , @xcite and demonstrate that the transport behaviour depends on the presence or absence of topological disorder which our method can accommodate in a straightforward manner .
the rest of the paper is organized as follows .
section [ sec : one ] introduces the schmidt decomposition and matrix product decomposition of pure anyonic states and outlines the generalization of tebd algorithm to the anyonic mps . section [ sec : three ] contains the numerical results .
appendix [ sec : appaa ] recapitulates the derivation of the standard schmidt decomposition for spin systems . the derivation of the anyonic schmidt decomposition presented in sec .
[ ssec : bidecom ] follows the same sequence of steps described in appendix [ sec : appaa ] but adapts each step to the anyonic setting .
the basic terminology and graphical notation pertaining to anyon models as used in this paper is summarised in appendix [ sec : appa ] .
appendix [ sec : atebd ] describes the step - by - step implementation of the anyonic tebd algorithm .
appendix [ sec : general ] describes some straightforward generalizations of the anyonic mps formalism presented in this paper , and also its specialization to study systems with conserved symmetry charges . in this paper
, we essentially follow the graphical notation for anyon models described in refs . and
. however , we find it convenient to rotate the graphical representation of fusion trees in ref .
counterclockwise to mimic the graphical representation of the mps [ fig .
[ fig : intro1 ] ] . exploiting the fact that for any anyonic model there necessarily exists a vacuum charge with trivial fusion rules ,
we sometimes attach this charge to the left of the fusion tree for convenience , as in fig . [
fig : intro2 ] .
when we do so for a fusion tree of @xmath3 anyons the number of intermediate charges ( those appearing on the internal edges ) in the fusion tree is @xmath47 , whereas there would be only @xmath48 intermediate charges if we did not introduce the trivial vacuum charge . for the purpose of clearer demonstration , we have made certain simplifying assumptions in this paper .
first , we assume that the total fusion charge [ @xmath20 in fig .
[ fig : intro2 ] ] assumes only one value ( this condition arises naturally when describing a pure state ) .
second , we assume that each site of the anyonic lattice is described purely by a charge label from the anyon model , with no additional degeneracies or auxiliary degrees of freedom .
third , we restrict to _ multiplicity - free _ anyon models where the components @xmath38 [ eq . ( [ eq : fusionrules ] ) in appendix [ sec : appa ] ] only take values 0 or 1 .
appendix [ sec : general ] describes how the formalism can be generalized in a straightforward way to relax the latter two of these three assumptions . finally , although it is common practice to consider only anyonic states with @xmath49 [ fig .
[ fig : intro2 ] ] , we do not assume any particular value of @xmath20 throughout the paper except in the construction of the anyonic schimdt decomposition presented in sec .
[ ssec : bidecom ] ( where we set @xmath49 for convenience ) and in sec .
[ sec : three ] which presents the numerical results .
our methodology may therefore readily be applied to systems with non - trivial total charge .
consider a one dimensional lattice @xmath50 made of @xmath3 sites that are fixed on a line and populated by anyons @xmath51 .
denote by @xmath52 the hilbert space that describes lattice @xmath50 .
a basis is introduced in @xmath52 by means of a fusion tree ( illustrated in fig .
[ fig : intro2 ] ) .
we may also denote a fusion tree basis by explicitly listing the sequence of fusions in the tree ; for example , we may denote the basis depicted in fig .
[ fig : intro2 ] as @xmath53 in this paper we are interested in states @xmath54 that have a well defined total charge @xmath20 .
state @xmath54 can be expanded as @xmath55 where @xmath56 are complex coefficients and the sum is over all sets of _ compatible _ charges @xmath57 and @xmath58 , namely , sets of charges resulting in a valid fusion tree .
the space @xmath52 decomposes as @xmath59 where @xmath60 is a subspace of states in @xmath52 that have a well defined total charge @xmath20 .
the dimension of subspace @xmath60 is equal to the number of ways @xmath61 in which the total charge @xmath20 can be obtained by fusing together the @xmath3 anyons .
we also refer to @xmath61 as the _ degeneracy _ of total charge @xmath20 in the decomposition of eq .
( [ eq : totspace ] ) .
state @xmath54 can be expanded as @xmath62 in accordance with the decomposition ( [ eq : totspace ] ) .
here @xmath63 denotes an orthonormal basis in the subspace @xmath60 such that @xmath64 , the _ degeneracy index _ of charge @xmath20 , enumerates the different labellings of the fusion tree that are compatible with the given value of charge @xmath20 . in this subsection
we describe generic bipartite decompositions and the schmidt decomposition of an anyonic state @xmath54 .
the latter plays an instumental role in constructing the matrix product decomposition of @xmath54 .
the reader interested mostly in the definition of the anyonic mps and in the implementation of the anyonic tebd algorithm may skip the following technical discussion and proceed directly to sec .
[ ssec : amps ] . consider a bipartition of @xmath50 into sublattices @xmath65 and @xmath66 that consist of anyons @xmath67 and @xmath68 respectively .
denote by @xmath69 and @xmath70 the vector spaces that describe @xmath65 and @xmath66 respectively .
we have @xmath71 where @xmath72 and @xmath73 are the degeneracy spaces [ eq .
( [ eq : totspace ] ) ] of total charges @xmath16 and @xmath74 for sublattices @xmath65 and @xmath66 respectively .
the total space @xmath75 is the tensor product of @xmath69 and @xmath70 , and decomposes according to @xmath76 where @xmath60 is the degeneracy space of total charge @xmath20 in eq .
( [ eq : totspace ] ) and the direct sum is over charges @xmath16 and @xmath74 that are compatible with @xmath20 according to the fusion rules , that is , the set of charges @xmath77 which satisfy @xmath78 . here @xmath79 is the 3-index tensor that encodes the fusion rules of the anyon model , defined according to eq .
( [ eq : fusionrules ] ) in appendix [ sec : appa ] . note that , in general , there may be several @xmath80 that contribute to the degeneracy of total charge @xmath20 in the decomposition ( [ eq : bidecom ] ) . with definite charge @xmath20 .
the coefficients @xmath81 in eq .
( [ eq : bistate0 ] ) are encoded in _
degeneracy tensors _
@xmath82}}_{au } , \lambda_u$ ] and @xmath83}}_{ub}$ ] with components @xmath84}}_{au})_{t_at_u } , ( \lambda_u)_{t_u}$ ] , and @xmath85}}_{ub})_{t_ut_b}$ ] respectively by means of the singular value decomposition ( [ eq : blocksvd ] ) .
charges @xmath86 and @xmath87 fulfill the fusion rules ( [ eq : aub ] ) that are depicted here by the fusion tree . [
fig : bipartite],width=302 ] notice how the introduction of degeneracy spaces allows for a decomposition of the anyonic hilbert space @xmath75 as a direct sum of _ tensor product _ spaces [ eq . ( [ eq : bidecom ] ) ] . in the remainder of this section ,
we describe how this decomposition can be exploited to construct a bipartite decomposition and the matrix product decomposition of the anyonic state @xmath54 in eqs .
( [ eq : genstate ] ) and ( [ eq : sing ] ) .
let @xmath88 and @xmath89 denote an orthonormal basis in @xmath65 and @xmath66 respectively . then in accordance with the decomposition ( [ eq : bidecom ] )
, we can choose a basis @xmath63 [ eq . ( [ eq : sing ] ) ] in the total space @xmath75 that factorizes as @xmath90 where the fusion rule @xmath91 enforces a total charge @xmath20 .
a generic bipartite decomposition of state @xmath54 according to bipartition @xmath92 reads as @xmath93 next , we introduce the schmidt decomposition of @xmath54 according to the bipartition @xmath92 .
our derivation of the anyonic schmidt decomposition follows the same sequence of steps involved in the standard derivation of the schmidt decomposition for spins systems , which is recapitulated in appendix [ sec : appaa ] .
we refer the reader to appendix [ sec : appaa ] as an aid to understanding the following derivation . without loss of generality ,
@xcite we now specialize to trivial total charge ( @xmath49 ) for simplicity .
components @xmath94 in eq .
( [ eq : bistate0 ] ) can be organized as a matrix @xmath95 where the paired indices @xmath96 and @xmath97 label the rows and columns respectively . since @xmath49 ,
the fusion rules @xmath78 imply that charge @xmath74 is the _ dual _ of charge @xmath16 ( denoted as @xmath98 ) and therefore matrix @xmath95 is block diagonal as @xmath99 where @xmath100 is a matrix block with components @xmath81 . notice that we now denote the components of @xmath95 as @xmath81 instead of @xmath94 to explicitly indicate the block structure ( [ eq : biblocks ] ) . in terms of _
charge - conserving _ tensors @xmath82}}$ ] , @xmath101 , and @xmath83}}$ ] , which decompose in terms of degeneracy tensors @xmath82}}_{au}$ ] , @xmath102 , and @xmath83}}_{ub}$ ] respectively according to eq .
( [ eq : allblock ] ) .
[ fig : bipartite1],width=226 ] consider the singular value decomposition of block @xmath100 of eq .
( [ eq : biblocks ] ) , @xmath103}}_{au } \lambda_u \gamma^{^{[\mathcal{b}]}}_{ub},\label{eq : blocksvd}\ ] ] or in terms of components ( see fig . [
fig : bipartite ] ) @xmath104}}_{au})_{t_at_u } ( \lambda_u)_{t_u } ( \gamma^{^{[\mathcal{b}]}}_{ub})_{t_ut_b}.\label{eq : blocksvd1}\ ] ] where @xmath82}}_{au}$ ] and @xmath83}}_{ub}$ ] are unitary matrices , @xmath105}}_{au } ( \gamma^{^{[\mathcal{a}]}}_{au})^\dagger = \gamma^{^{[\mathcal{b}]}}_{ub}({\gamma^{^{[\mathcal{b}]}}_{ub}})^\dagger = \mathbb{i}.\end{aligned}\ ] ] and @xmath102 is a diagonal matrix with non - negative diagonal entries , @xmath106 .
the charges @xmath107 and @xmath74 satisfy the fusion rules : @xmath108 the fusion rule @xmath109 simply implies that @xmath110 , but we have introduced a new label @xmath87 to clearly distinguish the corresponding degeneracy indices @xmath111 and @xmath112 , which are independent of each other in the following discussion .
the fusion rule @xmath113 implies that @xmath114 .
therefore , charges @xmath16 and @xmath74 can be uniquely determined if @xmath87 is specified .
using eq .
( [ eq : blocksvd1 ] ) in eq .
( [ eq : bistate0 ] ) and summing over @xmath111 and @xmath117 we obtain @xmath118}}_{ut_u } \rangle}\otimes { | \phi^{^{[\mathcal{b}]}}_{u^*t_{u^ * } } \rangle } , \label{eq : schmidt0}\ ] ] where we have replaced @xmath110 and @xmath114 using ( [ eq : aub ] ) and defined the orthonormal vectors @xmath119}}_{ut_u } \rangle } = \sum_{t_a } ( \gamma^{^{[\mathcal{a}]}}_{au})_{t_at_u } n_{1a}^u{| at_a \rangle},\\ { | \phi^{^{[\mathcal{b}]}}_{u^*t_{u^ * } } \rangle } = \sum_{t_b } ( \gamma^{^{[\mathcal{b}]}}_{ub})_{t_ut_b } n_{ub}^1{|
bt_b \rangle}. \end{split}\ ] ] we write eq .
( [ eq : schmidt0 ] ) more succintly by introducing the paired index @xmath120 and its dual @xmath121 , @xmath122}}_\mu \rangle}\otimes { | \phi^{^{[\mathcal{b}]}}_{\mu^ * } \rangle}\label{eq : aschmidt}\end{aligned}\ ] ] with concise graphical representation given in fig .
[ fig : bipartite1 ] .
equation ( [ eq : aschmidt ] ) is the anyonic schmidt decomposition .
the similarity between eq .
( [ eq : aschmidt ] ) and eq .
( [ eq : spinsd ] ) is apparent , however , note the distinction : here index @xmath123 is a charge - degeneracy pair [ in accordance with ( [ eq : abdecom ] ) ] , and the schmidt bases in @xmath65 and @xmath66 are labeled by @xmath123 and its dual @xmath124 respectively to constrain the total charge of the bipartite anyonic state @xmath54 to @xmath49 .
the norm of state @xmath54 is given as @xmath125.\label{eq : normm}\ ] ] the anyonic schmidt decomposition is a useful tool to probe bipartite entanglement in an anyonic state @xmath54.@xcite in analogy with spin or bosonic systems ( see appendix [ sec : appaa ] ) , we define the von neumann entanglement entropy @xmath126 of parts @xmath65 and @xmath66 of a pure anyonic state @xmath54 as @xmath127,\ ] ] where @xmath128 is the quantum dimension of charge @xmath87 . more generally , components @xmath56 in eq .
( [ eq : genstate ] ) can be encoded as a _ matrix product state _ [ see fig . [
fig : anyonicmps](i ) ] ; that is , @xmath129}\mu_1}_{1 a_1 } \lambda^{^{[1]}}_{\mu_1 } \gamma^{^{[2]}\mu_2}_{\mu_1 a_2 } \lambda^{^{[2]}}_{\mu_2 } \ldots \lambda^{^{[l-1]}}_{\mu_{l-1 } } \gamma^{^{[l]}a_\mathrm{tot}}_{\mu_{l-1 } a_l}.\label{eq : amps}\ ] ] here index @xmath11 is a charge - degeneracy pair @xmath130 , @xmath13}}$ ] is the diagonal matrix that appears in the anyonic schmidt decompositon according to the bipartition @xmath12:[i+1\ldots l]$ ] and tensors @xmath131}}$ ] relate the schmidt basis for consecutive bipartitions as @xmath132}}_{u})_{t_u } \rangle } = ( \gamma_{ua_i}^{^{[i]}u'})_{t_ut_{u ' } } ( \lambda^{^{[i]}}_{u'})_{t_{u'}}n_{ua_i}^{u'}~~{| ( \phi^{^{[i+1\ldots l]}}_{u'})_{t_{u ' } } \rangle}.\end{aligned}\ ] ] analogous to eq .
( [ eq : allblock ] ) , tensors @xmath101 and @xmath133 can be decomposed in accordance with the fusion rules as [ see fig . [
fig : anyonicmps](ii)-(iii ) ] @xmath134 where @xmath102 is a diagonal _ degeneracy matrix _ with diagonal entries @xmath106 and @xmath135 is a _ degeneracy tensor _ with components @xmath136 .
the decompositions ( [ eq : gamma ] ) imply that @xmath101 and @xmath133 correspond to linear maps that conserve anyonic charge .
we refer to the decomposition eq .
( [ eq : amps ] ) in terms of charge - conserving tensors as _ the anyonic mps_. by working with a fusion tree that mimics the tensor network structure of the mps , manifesting as the visual similarity between fig .
[ fig : anyonicmps ] and fig .
[ fig : intro2 ] , we obtain a direct correspondence between the fusion tree description and the mps description of an anyonic state .
namely , the physical indices and the bond indices of the mps are labelled by charges @xmath57 and @xmath137 that appear on the open and internal edges of the fusion tree respectively .
this means that for given charges @xmath138 the coefficients @xmath56 in eq .
( [ eq : genstate ] ) can be recovered from the anyonic mps by fixing these charges on the respective mps indices , decomposing each charge - conserving tensor according to ( [ eq : gamma ] ) and multiplying together the degeneracy tensors .
next , we explain how the tebd algorithm is adapted to the anyonic mps by ensuring that the fusion constraints encoded in the mps tensors are preserved during time evolution
. matrix product decomposition of state @xmath54 of 5 anyons .
the coefficients @xmath56 in eq . (
[ eq : genstate ] ) are encoded in anyonic charge - conserving tensors @xmath131}}$ ] and @xmath13}}$ ] .
@xmath1 for a fixed charge @xmath87 , each diagonal matrix @xmath101 decomposes into diagonal degeneracy matrix @xmath139 and the fusion tensor @xmath140 ( we have suppressed the vacuum label in the figure ) .
@xmath141 for fixed charges @xmath107 and @xmath142 , each tensor @xmath133 decomposes into a degeneracy tensor @xmath135 and a fusion tensor @xmath143 .
( [ eq : gamma ] ) .
note that the canonical bipartite decomposition of fig .
[ fig : bipartite1 ] may be regarded as a 2-site matrix product decomposition of @xmath54 .
[ fig : anyonicmps ] ] following ref .
we perform a trotter decomposition of @xmath155 in eq .
( [ eq : timeevolve ] ) over a sequence of small time steps @xmath156 , @xmath157}}(\delta t ) \prod_{i\in~odd } u^{^{[i , i+1]}}(\delta t)\right]^n + o(\delta t^2).\ ] ] each 2-site gate @xmath158}}(\delta t)$ ] decomposes as per eq .
( [ eq : localham1 ] ) : @xmath159}}(\delta t)= \bigoplus_p u^{^{[i , i+1]}}_p.\ ] ] the main step of the ( anyonic ) tebd algorithm is to update the mps after applying a 2-site gate @xmath158}}(\delta t)$ ] , as depicted formally in fig . [
fig : mainstep ] . as explained in ref . for a non - anyonic mps ,
this update comprises of certain tensor contractions and a matrix singular value decomposition . for the anyonic mps the goal is to ensure that the updated tensors @xmath160 } } } ' , { \lambda^{^{[i]}}}'$ ] and @xmath161}}}'$ ] are charge - conserving , having a block structure that is compatible with the fusion rules
this is achieved by decomposing the anyonic mps tensors into degeneracy and fusion parts according to eq .
( [ eq : gamma ] ) . the step by step details of how to enact the update of fig .
[ fig : mainstep ] for the anyonic mps is explained in appendix [ sec : atebd ] . } } , \lambda^{^{[i]}}$ ] and @xmath162}}$ ] are locally updated after absorbing the charge - conserving time - evolution gate @xmath158}}$ ] .
[ fig : mainstep ] ] let us denote by @xmath169 the maximum degeneracy associated with any charge @xmath167 that appears on the bond indices of the anyonic mps . when the amount of entanglement in the ground state @xmath54 of @xmath147 is limited , namely , when @xmath170 is bounded and does not scale with system size @xmath3 , the anyonic mps allows for an extremely efficient description of @xmath54 in terms of approximately @xmath171 coefficients .
the maximum degeneracy @xmath170 also controls the computational cpu cost incurred by the anyonic tebd algorithm : the cpu cost scales approximately as @xmath172 , being dominated by the cost of the singular value decomposition step of the algorithm . , eq .
( [ eq:2point ] ) , between sites @xmath8 and @xmath173 ( @xmath174 ) of the energy density @xmath175}$ ] of the ground state of the infinite antiferromagnetic chain of ising anyons and fibonacci anyons , approximated by means of an anyonic mps with @xmath176.[fig : correlators],width=283 ]
to demonstrate the effectiveness of the algorithm , we applied it to the study of two different types of interacting ( quasi- ) one - dimensional models of interacting anyons .
, eq . ( [ eq : rho ] ) , of a block of @xmath194 sites for the ground state of the infinite antiferromagnetic chain of ising anyons and fibonacci anyons , approximated by means of an anyonic mps with @xmath176.[fig : entropy],width=283 ] c| c| c + & _ charges _ & _ degeneracy _ + [ 0.5ex ] & 1 & 100 + & @xmath195 & 100 + odd & @xmath180 & 200 + c| c| c + & _ charges _ & _ degeneracy _ + [ 0.5ex ] & 1 & 76 + & @xmath186 & 124 + & 1 & 76 + & @xmath186 & 124 + we used the anyonic tebd algorithm to approximate the ground state @xmath196 of the two models by means of imaginary time evolution , @xmath197 and imposed the constraint @xmath198 in eq .
( [ eq : chi00 ] ) .
table [ table : charges ] lists the charges @xmath40 with degeneracies @xmath199 that contribute to the even and odd bipartitions of the resulting state with the constraint @xmath176 .
@xcite .
energy per site of an infinite antiferromagnetic chain of ising anyons and fibonacci anyons obtained using an anyonic mps with bond dimensions @xmath200 , [ . ]
indicates the first significant digit that differs from the exact energy .
the exact energies are listed only upto 10 significant digits.[table : gsenergy ] [ cols="^,^,^ " , ] we obtained an accurate approximation of the ground state energy per site , @xcite as listed in table [ table : gsenergy ] . in fig .
[ fig : correlators ] we plot the 2-point correlator @xmath201 of the energy density @xmath202}}$ ] [ eqs .
( [ eq : ising])-([eq : fib ] ) ] for the ground state @xmath196 , @xmath203}}h^{^{[i+r , i+r+1]}}\rangle - \langle h^{^{[i , i+1]}}\rangle \langle h^{^{[i+r , i+r+1]}}\rangle.\label{eq:2point}\ ] ] the expected polynomial decay @xmath204 is reproduced with exponents @xmath205 and @xmath206 .
these exponents are compared with results from conformal field theory . in a cft ,
2-point correlators of a primary field @xmath207 with conformal dimensions @xmath208 decay as @xmath209 where @xmath210 is the complex space - time coordinate and @xmath211 is the conjugate of z ( treated as an independent coordinate ) . the exponents @xmath205 and @xmath206 are consistent with the correlator ( [ eq:2point ] ) receiving dominant contribution from the energy density field @xmath212 ( @xmath213 ) of the ising cft , which predicts @xmath214 , and from the spin field @xmath180 ( @xmath215 of the tricritical ising cft , which predicts @xmath216 . in fig .
[ fig : entropy ] we plot the entanglement entropy @xmath217,\label{eq : rho}\ ] ] of a block of @xmath218 anyons in the ground state @xmath196 , described by the reduced density matrix @xmath219 .
the expected logarithmic scaling @xmath220 for critical ground states is reproduced , and the central charges are approximated as @xmath221 $ ] and @xmath222 $ ] , in excellent agreement with the theoretical results of @xmath188 and @xmath193 respectively .
we have also studied the dynamics of an anyonic hubbard - like model .
this model describes the hopping of mobile anyons on sites of a ladder with two horizontal legs , around islands which are occupied by pinned anyons ( see fig .
[ fig : ladder ] ) .
the ladder is the minimal geometry which can accommodate interactions between anyons mediated purely via braiding , allowing mobile anyons on the ladder to braid around the pinned anyons .
we consider a possibly non - translationally invariant ( disordered ) filling but restrict to the case of one mobile anyon hopping on the ladder .
@xcite without topological disorder , and @xmath1 with topological disorder corresponding to a uniform and a non - uniform filling of the islands of the ladder with pinned anyons respectively .
one mobile anyon hops ( here shown on site @xmath8 ) between vertices of the ladder around islands constaining pinned anyon(s ) .
the hamiltonian is the sum of a hopping term that corresponds to horizontal translation along the two legs of the ladder [ which is associated with clockwise or counterclockwise braiding as described by operators @xmath223 and @xmath224 respectively in eq .
( [ eq : braidop ] ) ] , and a tunnelling between legs with no braiding . ] for a ladder made of @xmath3 sites , indexed by integer position @xmath8 , the system can be described by a hilbert space @xmath225 where @xmath226 is the fusion space of the pinned anyons plus the mobile anyon , and the @xmath3 qutrits @xmath227 ( @xmath228 ) encode the position of the mobile anyon on the ladder : @xmath229 corresponds to the absence of the mobile anyon on site @xmath8 , @xmath230 corresponds to the presence of the mobile anyon on the lower leg at site @xmath8 and @xmath231 corresponds to the presence of the mobile anyon on the upper leg at site @xmath8 .
the hamiltonian is the sum of terms mediating hopping along the length of the ladder and terms mediating tunnelling between the upper and lower legs of the ladder , @xmath232 where @xmath233 ( @xmath234 is the identity on the fusion space of the anyons . ) here @xmath235 are translation operators between sites @xmath8 and @xmath236 , @xmath237 on an open chain we assume reflecting boundary conditions : @xmath238 .
the operators @xmath239 and @xmath240 braid the mobile anyon across the island immediately to the right of site @xmath8 and may be written as @xmath241 where @xmath242 is the number of anyons in the island and the operators @xmath243 are a unitary representation of the @xmath218-strand braid group , @xmath244 acting on the fusion space @xcite of the anyons on the island immediately to the right of site @xmath8 .
note , @xmath245 if @xmath246 .
finally , the projectors @xmath247 act to select out states where the hopping anyon is on the lower or upper leg for @xmath248 and @xmath249 respectively . in order to simulate this model using the anyonic tebd algorithm we mapped the model on the ladder with @xmath3 sites to a one dimensional lattice @xmath50 also with @xmath3 sites . because the anyons on the islands are pinned , and the mobile anyon @xmath250 braids around them _ en masse _ , we may replace each island @xmath8 with a single anyon having the same total charge @xmath251 as all the pinned anyons located in the island . note that in the presence of disorder charge @xmath251 can assume multiple values on some islands .
we describe site @xmath252 of @xmath50 by a basis labelled as @xmath253 where @xmath254 is an anyon charge , @xmath255 is a u(1 ) charge corresponding to the number of mobile anyons at site @xmath252 [ see appendix [ sec : general ] ] and @xmath256 labels the lower ( @xmath257 ) or upper ( @xmath258 ) leg of the ladder .
the correspondence between the description of the model in terms of sites ( [ eq : mydesc ] ) and the ladder system is as follows : 1 .
@xmath259 mobile anyon on lower leg at site @xmath252 of the ladder .
@xmath260 mobile anyon on upper leg at site @xmath252 of the ladder .
@xmath261 mobile anyon anywhere to the right of island @xmath252 of the ladder .
@xmath262 mobile anyon anywhere to the left of island @xmath252 of the ladder .
we treat the pair @xmath263 as a _ composite _
anyonic charge [ see appendix [ sec : general ] ] with degeneracy @xmath264 , and describe lattice @xmath50 by a fusion tree with @xmath3 open edges that are labelled by composite charges .
the fusion rules for the composite charges are given by @xmath265 we truncate the u(1 ) charge @xmath266 to a maximum value of 1 , which imposes the constraint that the total number of mobile anyons on the ladder is equal to 1 .
( [ eq : chi00 ] ) ] .
the couplings in eqs .
( [ eq : htrans])([eq : htun ] ) are set as @xmath286 .
the situation under study comprises a single anyon ( @xmath287 ) , initialised in the middle of a ladder with 100 sites , and permitted to hop around topologically charged islands filled with anyons of the same type ( and with the total charge of all anyons set to @xmath49 ) .
transport of this anyon along the chain is quantified by the variance of the spatial distribution @xmath288 . in the case of disordered topological backgrounds
the variance is averaged over randomly distributed island charge occupations and is denoted @xmath289 . shown
are the variances for ( i ) abelian anyons with @xmath290 exchange statistics around a uniform background of islands ( uniform abelian = ua ) showing ballistic transport , ( ii ) ising anyons around a uniform background of ising anyons ( uniform non - abelian = una ) exhibiting dispersive transport , and ising model anyons with islands having random occupation levels @xmath291 of ising anyons ( random non - abelian = rna ; 50 configurations ) also exhibiting dispersive transport but with a smaller diffusion constant ( slope ) than in the uniform non - abelian case . for the random case
we have also plotted the error bars for one sigma variance .
the space and time axes are scaled so that the una case corresponds to a classical diffusion@xcite with diffusion constant ( slope ) equal to one .
note that for the uniform cases we have @xmath292 .
[ fig:2],width=264 ] we studied the real time dynamics of the anyonic hubbard model using the anyonic tebd algorithm .
our results are plotted in fig .
[ fig:2 ] and were presented in an earlier work .
@xcite in the case of uniform topological backgrounds , i.e. translationally invariant filling of the islands , abelian anyons have ballistic transport as indicated by the variance of the mobile anyons spatial distribution , @xmath293 .
in contrast non - abelian anyons display a dispersive transport , @xmath294 , due to the fact that the different trajectories of the particle become correlated with different fusion environments while braiding , meaning spatial coherences are quickly lost .
@xcite in the presence of topological disorder , the behaviour changes substantially . for abelian anyons ,
the disorder acts to localize the particle while for non - abelian anyons , the transport is still dispersive .
this result is due to the fact that the fusion degrees of freedom become sufficiently entangled with the mobile anyon that the destructive interferences necessary to provide localisation are lost . as described in detail in ref .
the competition between localization and decoherence is a subtle one , and long - time simulations using the anyonic mps were essential to establish this result . in this paper
we have introduced the matrix product decomposition of states of 1d lattice systems of anyons , and have described how to extend the tebd algorithm to the anyonic mps .
we have demonstrated the efficacy of the anyonic tebd algorithm by computing the expected scaling of the ground state entanglement and 2-point correlators for two critical antiferromagnetically coupled infinite chains of non - abelian anyons , our results being in agreement with those previously obtained by authors using other techniques .
our method has the advantages that it is conceptually simple , demands only modest computational power to achieve accurate results , and can be applied to the study of generic anyon lattice models .
the basic data @xmath295 listed in sec .
[ sec : intro ] which characterize an anyon model may also be used to describe fermionic constraints or constraints due to the presence of an onsite global symmetry@xcite where charges correspond to the irreducible representations ( irreps ) of a symmetry group @xmath296 , as illustrated in appendix [ sec : appa ] . by furnishing the data @xmath295 from a symmetry group @xmath296 in this way
, the anyonic mps can also be used to efficiently represent states of a lattice system that are invariant , or more generally covariant , under the action of an onsite abelian or non - abelian global symmetry @xmath296 on the lattice , as this is equivalent to a specialization from the more general anyonic case .
our implementation of anyonic constraints in the mps is closely related to ( and generalizes ) the implementation of global onsite symmetry constraints in tensor network algorithms ( see e.g. refs . and references therein ) .
similarly , the @xmath297symmetric mps described in ( for example ) ref . provides an effective illustration of the present anyonic mps formalism in what will be , for many , a more familiar context .
we hope that the anyonic mps formalism presented in this paper will prove to be a useful tool for studying generic lattice models of interacting anyons .
s.s . acknowledges financial support from the mqns grant by macquarie university grant no .
also thanks robert pfeifer and perimeter institute for hospitality .
thanks the ontario ministry of research and innovation early researcher awards for financial support . g.k.b .
thanks the kitp where part of this work was completed with support from the national science foundation under grant no .
nsf phy11 - 25915 .
this research was supported in part by the arc centre of excellence in engineered quantum systems ( equs ) , project no .
this research was supported in part by perimeter institute for theoretical physics .
research at perimeter institute is supported by the government of canada through industry canada and by the province of ontario through the ministry of research and innovation .
in this appendix we recapitulate the derivation of the schmidt decomposition for spin systems as an aid for understanding the analogous derivation for the schmidt decomposition of anyonic systems that is presented in sec.[ssec : bidecom ] .
consider a pure state @xmath2 } } \rangle}$ ] of a spin system that belongs to a tensor product space @xmath298 .
state @xmath2 } } \rangle}$ ] can be expanded as @xmath5 } } \rangle } = \sum_{ab } \psi^{\mbox{\tiny [ spin]}}_{ab } { | a \rangle } \otimes { | b \rangle},\label{eq : spinbistate}\ ] ] where @xmath299 and @xmath300 denote an orthonormal basis in @xmath69 and @xmath70 respectively .
we can regard @xmath6}}$ ] as a matrix with components @xmath6}}_{ab}$ ] where indices @xmath16 and @xmath74 enumerate the rows and columns .
consider the singular value decomposition of matrix @xmath6}}$ ] , @xmath301}}_{ab } = \sum_{u}\gamma^{^{[\mathcal{a}]}}_{au } \lambda_{u } \gamma^{^{[\mathcal{b}]}}_{ub},\label{eq : spinsvd1}\ ] ] where @xmath82}}$ ] and @xmath83}}$ ] are unitary matrices , @xmath116 } } ( \gamma^{^{[\mathcal{a}]}})^\dagger = \gamma^{^{[\mathcal{b } ] } } ( \gamma^{^{[\mathcal{b}]}})^\dagger = \mathbb{i},\ ] ] and @xmath101 is a diagonal matrix with non - negative diagonal entries @xmath302 . using ( [ eq : spinsvd1 ] ) in ( [ eq : spinbistate ] ) we obtain @xmath5 } } \rangle } = \sum_{abu}\gamma^{^{[\mathcal{a}]}}_{au } \lambda_{u } \gamma^{^{[\mathcal{b}]}}_{ub } ~{| a \rangle } \otimes { | b \rangle}.\ ] ] summing over @xmath16 and @xmath74 we obtain @xmath5 } } \rangle } = \sum_{u } \lambda_{u } ~{| \phi^{^{\mathcal{a}}}_u \rangle } \otimes { | \phi^{^{\mathcal{b}}}_u \rangle},\label{eq : spinsd11}\ ] ] where we have defined vectors @xmath303}}_{au } { | a \rangle},\\ { | \phi^{^{\mathcal{b}}}_u \rangle } & = \sum_{a } \gamma^{^{[\mathcal{b}]}}_{ub } { | b \rangle}.\end{aligned}\ ] ] by construction , vectors @xmath304 and @xmath305 are orthonormal , @xmath306 equation ( [ eq : spinsd11 ] ) is the schmidt decomposition of the bipartite state @xmath2 } } \rangle } \in \mathbb{v}^{(\mathcal{a } ) } \otimes \mathbb{v}^{(\mathcal{b})}$ ] .
the norm of state @xmath6}}$ ] is @xmath307 } } | \psi^{\mbox{\tiny [ spin ] } } \rangle } = \sum_u \lambda^{^2}_u.\ ] ] the schmidt decomposition is a useful tool in quantum information theory to study bipartite entanglement .
the reduced density matrices @xmath308}}$ ] and @xmath309}}$ ] for parts @xmath65 and @xmath66 of state @xmath2 } } \rangle}$ ] in ( [ eq : spinsd11 ] ) are obtained as @xmath310 } } & = \sum_u \lambda^{^2}_u { | \phi^{^{[\mathcal{a}]}}_u \rangle}{\langle \phi^{^{[\mathcal{a}]}}_u |},\\ \rho^{^{[\mathcal{b } ] } } & = \sum_u \lambda^{^2}_u { | \phi^{^{[\mathcal{b}]}}_u \rangle}{\langle \phi^{^{[\mathcal{b}]}}_u |}.\end{aligned}\ ] ] the von - neumman entanglement entropy @xmath126 , @xmath311}}\mbox{log}\rho^{^{[\mathcal{a } ] } } ) = -\mbox{tr}(\rho^{^{[\mathcal{b}]}}\mbox{log}\rho^{^{[\mathcal{b}]}}),\ ] ] of the bipartite state @xmath2 } } \rangle}$ ] is obtained as @xmath312
in this appendix we introduce basic terminology and graphical notation pertaining to anyon models as used in this paper . for those already familiar with graphical notations for states of anyonic systems , the formalism employed in this paper corresponds to that described in ref . , save that it has been rotated 135@xmath313 counterclockwise in order to emphasise the relationship between the tensor network structure of the mps ( fig .
[ fig : intro1 ] ) and the corresponding anyonic fusion tree ( fig .
[ fig : intro2 ] ) .
the decision was made to rotate the fusion tree to match the tensor network , rather than rotating the tensor network to match the fusion tree as in * ? ? ?
* , because the target audience of this paper is primarily intended to be readers with prior experience in conducting simulations using mps and drmg , and thus it was considered desirable to reflect the familiar tensor network configuration as closely as possible using the anyonic model . for readers who are not familiar with the graphical notation for anyonic systems , we summarize the pertinent features below .
an anyon model consists of a finite set @xmath314 of particle types , or charges , @xmath36 .
the set @xmath314 of allowed charges includes a distinguished trivial or vacuum charge @xmath315 .
the local properties of a single anyon are completely specified by its `` charge '' .
however , two ( or more ) anyons with charges @xmath16 and @xmath74 can be _ fused _ together into a total charge @xmath40 which can , in general , take several values , @xmath316 here @xmath38 is the number of times ( or the _ multiplicity _ ) charge @xmath40 appears in the fusion outcome .
we says charges @xmath86 and @xmath40 are _ compatible _ with one another if @xmath317 , and that tensor @xmath79 encodes the _ fusion rules _ of the anyon model . for simplicity ,
we consider anyon models that are multiplicity free , namely , @xmath318 for all charges @xmath16 and @xmath74 .
however , multiplicities can be accommodated into our formalism in a rather straightforward way , as described in appendix [ sec : general ] .
the fusion rules for the vacuum charge satisfy @xmath319 .
two charges @xmath16 and @xmath74 are said to be _ dual _ to one another , denoted as @xmath320 and @xmath321 , if they fuse together to the vacuum . the operation
@xmath322 is an involution , @xmath323 .
consider two anyons @xmath16 and @xmath74 that are fixed on a line .
we denote an orthonormal basis in the total hilbert space as @xmath324 where charge @xmath40 is obtained by fusing @xmath16 and @xmath74 .
the graphical representations of the ket @xmath325 and the corresponding bra @xmath326 as employed in this paper are shown in fig .
[ fig : twoanyons](i)-(ii ) .
their inner product @xmath327 is graphically represented by gluing the diagrams of the ket and bra as shown in fig .
[ fig : twoanyons](iii ) .
the ket @xmath325 , @xmath1 the corresponding bra @xmath328 , @xmath141 the inner product ( braket ) , and @xmath329 braiding two anyons .
[ fig : twoanyons],width=226 ] the ordering of anyons @xmath16 and @xmath74 on the line may be interchanged by braiding them around one another to obtain another basis @xmath330 .
we follow the convention that when braiding @xmath16 counterclockwise around @xmath74 , @xmath330 is related to @xmath325 by a 3-index tensor @xmath281 [ see fig . [
fig : twoanyons](iv ) ] , @xmath331 while braiding @xmath16 clockwise around @xmath74 relates the two bases as @xmath332 where * denotes complex conjugation . for three ( or more )
anyons fixed on a line , different choices of basis are possible corresponding to different ways of performing pairwise fusings of the anyons into a single total charge .
three anyons @xmath86 and @xmath40 can be fused to a total charge @xmath333 by first fusing @xmath16 and @xmath74 into an intermediate charge @xmath334 and then fusing @xmath334 and @xmath40 to total charge @xmath333 .
denote the corresponding basis by @xmath335 .
alternatively , we could first fuse @xmath74 and @xmath40 into @xmath336 and then fuse @xmath16 and @xmath336 into @xmath333 . denote the basis corresponding to this fusion sequence by @xmath337 .
the two bases are related by a unitary transformation given by a 6-index tensor @xmath44 ( see fig . [
fig : threeanyons ] ) , @xmath338 the transformation given in eq .
( [ eq : f ] ) is also known as an _ f - move_. two possible fusion bases for three anyons @xmath86 and @xmath40 .
@xmath141 @xmath44-move , the unitary transformation which implements the change of basis from ( i ) to ( ii ) .
[ fig : threeanyons ] ] let us now consider a one dimensional lattice @xmath50 made of @xmath3 sites that are fixed on a line and populated by anyons @xmath57 that belong to a given anyon model . denote by @xmath75 the hilbert space that describes lattice @xmath50 . generalizing the description for three anyons , we introduce an orthonormal basis in @xmath52 by means of a _
fusion tree_. a fusion tree corresponds to a particular sequence of pairwise fusions of the @xmath3 anyons @xmath57 into a total definite charge @xmath20 by means of @xmath47 intermediate charges @xmath137 .
for example , a possible choice of fusion tree for a lattice made of @xmath18 sites is shown in fig .
[ fig : intro2 ] . different choices of fusion trees correspond to different choice of bases , and are related one another by @xmath44-moves .
when each site of the lattice carries the same charge @xmath16 , the dimension of the hilbert space @xmath75 is found to scale as @xmath339 where @xmath37 is the _ quantum dimension _ of charge @xmath16 , being analogous to the dimension of an irreducible representation of a group . in general this expression only holds approximately for finite @xmath3 , because @xmath37 may be non - integer for anyon models . an anyon model is completely specified by the following set of data : 1 .
the set of allowed charges , @xmath314 .
2 . the quantum dimension @xmath37 of each charge @xmath16 .
3 . the fusion rules , encoded in the multiplicity tensor @xmath38 of eq .
( [ eq : fusionrules ] ) .
4 . the braiding coefficients @xmath42 of eq .
( [ eq : r ] ) .
the @xmath44-move coefficients @xmath43 of eq .
( [ eq : f ] ) . for a consistent anyon model the @xmath44 and @xmath281 coefficients are required to satisfy the _
pentagon _ and _ hexagon _ relations that express associativity of fusion and compatibility of fusion with braiding respectively .
@xcite a broad class of anyon models is described by the quantum symmetry groups su(2)@xmath340 where the set of allowed charges corresponds to the irreducible representations @xmath341 of su(2)@xmath32 and , for instance , the @xmath44 coefficients correspond to the quantum 6-j symbols of the group .
next , we list some simple examples of anyon models . for a more extensive list of anyon models
see also ref . .
the fibonacci anyon model consists of two charges : @xmath342 ( the vacuum ) and @xmath186 .
the quantum dimensions of these charges are @xmath343 the only non - trivial fusion rule is @xmath344 .
that is , all components @xmath38 are zero except @xmath345 the coefficients @xmath42 are non - zero only if @xmath346 .
the non - zero @xmath281-coefficients are @xmath347 the non - trivial @xmath44-move coefficients are @xmath348 where @xmath349 and @xmath350 .
the remaining @xmath44-move coefficients are given by @xmath351 the ising anyon model consists of three charges , @xmath352 and @xmath195 , with quantum dimensions @xmath353 the non - trivial entries in the multiplicity tensor are given by @xmath354 the non - trivial @xmath281-coefficients are @xmath355 the non - trivial @xmath44-move coefficients are @xmath356 where @xmath357 .
the remaining @xmath44-move coefficients are once again given by eq .
( [ eq : trivialf ] ) .
the relevant charge for fermions is the parity @xmath154 of fermion particle number .
charge @xmath154 takes two values , @xmath358 and @xmath179 corresponding to an even or odd number of fermions respectively .
the fusion rules @xmath359 are given by @xmath360 and @xmath361 for all remaining values of @xmath362 and @xmath363 .
+ the non - trivial @xmath281-coefficients are given by @xmath364 the @xmath44-move coefficients are given by @xmath365
in this appendix we explain the step by step implementation of the main update [ fig .
[ fig : mainstep ] ] of the anyonic tebd algorithm . ] that describes fusion @xmath366 of indices @xmath367 and @xmath368 that carry both charge and degeneracy , @xmath369 , @xmath370 and @xmath371 . here
we show the generalized fusion tensor @xmath372}}$ ] ( @xmath373}}$ ] ) that corresponds to the fusion of two incoming ( outgoing ) indices @xmath367 and @xmath368 into an outgoing ( incoming ) index @xmath374 .
the fusion tensors @xmath372}}$ ] and @xmath373}}$ ] satisfy @xmath372 } } \tilde{n}^{\tiny \mbox{[out ] } } = \mathbb{i}$ ] when summed over @xmath367 and @xmath368 .
for fixed charges @xmath375 and @xmath40 , @xmath376}}$ ] decomposes [ eq . ( [ eq : ndecompose ] ) ] into a degeneracy tensor @xmath377 with components @xmath378 and a usual fusion tensor @xmath38 . in the graphical representation of degeneracy tensor @xmath377 the caret points to the total charge ,
that is , when fusing incoming ( outgoing ) indices the caret points to the outgoing ( incoming ) index.[fig : ntensor],width=264 ] first we introduce an important transformation
the _ generalized fusion tensor _ that is required in the algorithm .
a generalized fusion tensor @xmath379 describes fusion of indices @xmath380 that carry both charge and degeneracy , @xmath381 and @xmath371 , and thus generalizes the usual fusion tensor @xmath38 that is defined for fusing charges , @xmath382 of an anyon model according to eq .
( [ eq : fusionrules ] ) .
components @xmath379 are identically zero if charges @xmath16 , @xmath74 , and @xmath40 are incompatible with the fusion rules , i.e. when @xmath383 .
denote by @xmath384 and @xmath385 the degeneracy of charges @xmath86 and @xmath40 respectively , namely , @xmath386 in general there exist multiple choices of @xmath16 and @xmath74 compatible with any given charge @xmath40 .
the total degeneracy @xmath385 of charge @xmath40 then corresponds to the total number of different ways by which charge @xmath40 may be obtained by fusing charges @xmath16 and @xmath74 , with each degenerate copy of @xmath16 and @xmath74 counting as a separate channel .
thus @xmath387 for given values of @xmath16 and @xmath74 , each value of @xmath388 can be associated to a pair @xmath389 so as to construct a one - to - one correspondence .
we encode this association by setting the relevant component @xmath390 .
thus , for fixed charges @xmath86 and @xmath40 , tensor @xmath391 can be decomposed as @xmath392 where @xmath377 is a ( degeneracy ) tensor made of components @xmath393 that encodes the contributions of @xmath80 to the degeneracy of @xmath40 . the graphical representation of the generalized fusion tensor and the decomposition ( [ eq : ndecompose ] ) is shown in fig .
[ fig : ntensor ] .
next we explain how the update depicted in fig .
[ fig : mainstep ] is performed in 4 steps .
_ step 1 _ of the update is depicted in fig .
[ fig : step21 ] .
it corresponds to the reduction of a section of the mps to a charge - conserving matrix @xmath394 , @xmath395 obtained by fusing indices and contracting tensors together as depicted at the top of fig .
[ fig : step21 ] .
since indices carry both charge and degeneracy , we employ generalized fusion tensors @xmath372}}$ ] and @xmath373}}$ ] [ fig .
[ fig : ntensor ] ] to fuse indices as shown in fig .
[ fig : step21 ] .
a fusion vertex @xmath396 is inserted to reverse the orientation of @xmath397 prior to fusion with @xmath398 .
the aforementioned fusion of indices and the index reversal is undone in step 3 .
this can be understood as employing a resolution of identity , @xmath399 where , for example , transformation @xmath400 comprised of the reversal of @xmath397 and the fusion of charges @xmath398 and @xmath401 is enacted in this step and the corresponding @xmath402 is enacted in step 3 .
each block @xmath403 of @xmath394 is determined independently by fixing charge @xmath40 on the open indices and summing over all compatible internal charges and their respective degeneracy indices . for fixed compatible internal charges , the tensors of the mps
decompose into degeneracy and fusion parts according to eq .
( [ eq : gamma ] ) , and as depicted in fig .
[ fig : anyonicmps ] . by construction the generalized fusion tensors @xmath372}}$ ] and @xmath373}}$ ] also decompose into degeneracy and fusion parts as shown in fig .
[ fig : ntensor ] .
thus , for fixed charges on all indices the entire contraction decomposes into two parts as shown in fig .
[ fig : step21 ] , namely , a part made of only degeneracy tensors and a fusion network made of only fusion tensors .
the degeneracy tensors are contracted to obtain a matrix @xmath404 .
the fusion network encodes the fusion contraints on the entire contraction , and vanishes when some of the charges are incompatible with fusion rules . for compatible charges
the fusion network can be replaced by a single fusion vertex @xmath405 multiplied by a factor @xmath406 , @xmath407 by applying @xmath44-moves ( fig .
[ fig : threeanyons ] ) and contracting loops [ fig .
[ fig : twoanyons](iii ) ]
. matrix @xmath404 is then multiplied by the factor @xmath406 .
the above procedure is repeated for all compatible internal charges , and the matrices @xmath404 thus obtained are summed together to obtain the total ( degeneracy ) block @xmath403 . by iterating over all different charges @xmath40 ,
all blocks @xmath403 of matrix @xmath394 are determined .
figure [ fig : step22 ] depicts the remaining steps 2 - 4 of the local update .
the figure shows each step both in terms of the full charge - conserving tensors ( in the left column of each box ) and as it is actually performed by decomposing tensors into degeneracy and fusion parts ( in the right column of each box ) . addressing each in turn : , width=330 ] _
step 3 _ is to reorganize matrices @xmath281 and @xmath413 into three - index tensors @xmath414 and @xmath415 respectively .
this is achieved by contracting @xmath281 and @xmath413 with generalized fusion tensors @xmath373}}$ ] and @xmath372}}$ ] respectively as shown .
these contractions also proceed as before by fixing charges on the indices , decomposing the tensors into degeneracy and fusion parts and then contracting the degeneracy tensors .
the reorganization of matrix @xmath413 also involves inserting a fusion vertex @xmath416 to undo the reversal of @xmath152 enacted in step 1 and to re - orient index @xmath152 as outgoing on the reorganized tensor @xmath415 .
then by applying an @xmath44-move the two fusion vertices can be replaced by a single fusion vertex multiplied by a factor @xmath417 , @xmath418 this factor @xmath417 is multiplied into the reorganized tensor @xmath415 .
_ step 4 _ is to contract tensors @xmath414 and @xmath415 with the inverse matrices @xmath419}}}\right)^{-1}$ ] and @xmath420}}}\right)^{-1}$ ] as shown to obtain the updated tensors @xmath160}}}'$ ] and @xmath161}}}'$ ] and restore the canonical form of the anyonic mps .
finally , we make some remarks pertaining to a practical software implementation of the anyonic tebd algorithm .
first , for an anyon model with a finite number of charges , there are only a finite number of fusion networks that appear in step 1 of the anyonic tebd algorithm .
these fusion networks depend only on the anyon charges and can therefore be enumerated and the factors @xmath406 [ fig .
[ fig : step21 ] ] may be computed _ prior _ to executing the algorithm for the given hamiltonian .
this saves the cpu time that is otherwise expended to compute the same factors @xmath406 in each iteration of the tebd algorithm .
second , each block @xmath403 is determined _ independently _ by fixing different values of charge @xmath40 on the open indices and summing over all compatible internal charges .
this fact can be exploited in the software implementation to parallelize the contractions in step 1 that correspond to different values of @xmath40 .
in this appendix we discuss some generalizations of the anyonic mps formalism that is presented in this paper . the formalism can be readily extended 1 . to study lattice models with global onsite symmetry ( appendix [ ssec : symmetrygroup ] ) , and 2 . to study anyon lattice models with 1 .
auxiliary degrees of freedom on the lattice sites ( appendix [ ssec : supplementary ] ) , and/or 2 .
fusion multiplicities ( appendix [ ssec : fusionmult ] ) . the basic data @xmath295 , listed in the previous section , which characterize an anyon model may also be used to describe properties of a regular symmetry group @xmath296 where the charges correspond to the irreps of @xmath296 .
_ example 1 : _ consider an abelian lie group @xmath421 u(1 ) . charges ( irreps ) of u(1 ) are labelled by integers @xmath422 and have dimension 1 .
the fusion rules are @xmath423 the @xmath281 and @xmath44 coefficients are trivial for u(1 ) ( and for any abelian group ) .
that is , all @xmath281 and @xmath44 coefficients are equal to 1 for compatible charges and equal to 0 otherwise .
_ example 2 : _ consider the non - abelian group @xmath421 su(2 ) .
charges of su(2 ) are labelled by non - negative semi - integers @xmath424 .
the dimension of charge @xmath422 is equal to @xmath425 .
the fusion rules are @xmath426 the @xmath281-coefficients are given by @xmath427 the @xmath44-move coefficients correspond to the 6-j symbols of the su(2 ) . by furnishing the data @xmath295 from a symmetry group @xmath296 , the anyonic mps
can be used to represent states of a lattice system that are invariant or , more generally , covariant under the action of an onsite global symmetry @xmath296 on the lattice .
( an onsite global symmetry means that the symmetry group acts identically on each site of the lattice . ) for example , in the context of lattice spin systems , a global onsite symmetry @xmath428 may correspond to invariance of total spin under an identical rotation of all spins .
an anyonic mps constructed from the fusion data furnished from @xmath429 su(2 ) ( as per example 2 ) describes states of the lattice that have a well defined total spin .
thus , the implementation of anyonic constraints in the mps is closely related to and generalizes , the implementation of global onsite symmetry constraints in the mps , as discussed in ( for example ) ref . . in secs .
[ sec : one ] and [ ssec : heisenberg ] we illustrated our anyonic mps formalism in the context of a lattice system made up of sites that are populated by a single type of anyon .
however , our formalism can also be applied to the case where lattice sites contain supplementary degrees of freedom that may or may not be described by an anyon model . for instance , consider a lattice where each site @xmath8 corresponds to an anyon @xmath151 and a @xmath430dimensional spin .
the anyonic mps formalism can be extended for such lattice systems in a rather straightforward way by treating the spin degree of freedom on each site as the degeneracy of the anyon @xmath151 .
that is , each site of the lattice is described by a basis @xmath431 where the degeneracy index @xmath432 takes values @xmath433 . by making this identification , the anyonic mps formalism as described in this paper can be applied to study anyon @xmath434 spin lattice systems .
more generally , the lattice sites may contain supplementary degrees of freedom that correspond to charges described by a different anyon model or a global symmetry group @xmath296 . in this case each site @xmath8 is described by a basis @xmath435 where @xmath151 is an anyon charge , @xmath436 is another anyon charge ( or a symmetry charge ) and @xmath437 is a degeneracy index .
our formalism can be applied to this scenario by treating the pair @xmath438 as a _
composite _ charge .
the fusion rules and multiplicity tensor @xmath39 , the @xmath281-coefficients , and the @xmath44-move coefficients for the composite charges @xmath439 can be derived from the corresponding data for the individual charges @xmath16 and charge @xmath422 : @xmath440 @xmath441 an application of the anyonic mps to such a scenario is illustrated in sec .
[ ssec : hubbard ] in the context of the anyonic hubbard lattice model where each site is described by a composite charge @xmath439 ; @xmath16 is an anyon charge and @xmath422 is a u(1 ) charge associated with the number of mobile anyons on the site . in certain anyon models , multiplicities @xmath38 of fusion outcomes [ eq .
( [ eq : fusionrules ] ) ] can be greater than 1 .
non - trivial multiplicities can be accomodated in the anyonic mps formalism simply by appending a multiplicity label @xmath442 to the anyon charges output by a fusion .
the @xmath281 and @xmath44 coefficients are then augmented by a multiplicity index , @xmath443 ( strictly , charges @xmath16 , @xmath74 , and @xmath40 are also supplemented by a multiplicity index but the values of the @xmath281 and @xmath44 tensors are independent of the values of these extra indices . )
the treatment of fusion multiplicities is very similar to the treatment of _ degeneracies _ of anyonic charges discussed above in the context of the mps formalism . however , whereas the degeneracy index discussed above enumerates the different labellings of ( a portion of ) the fusion tree which yield the same total charge , the multiplicity index enumerates multiple copies of an output charge generated by a single labelling of the fusion tree .
the most general anyonic tensor therefore carries three labels on each leg corresponding to charge , multiplicity , and degeneracy .
the multiplicity index has been suppressed throughout this paper , and should not be confused with the number index appearing in eq .
( [ eq : mydesc ] ) which is an additional index specific to that model , and specifies the charge of an auxiliary u(1 ) symmetry group corresponding to the number of particles present at a lattice site .
pachos , _ introduction to topological quantum computation _
, cambridge university press ( 2012 ) .
f. d. m. haldane .
93a * , 464 , ( 1983 ) ; f. d. m. haldane .
lett . * 50 * , 1153 , ( 1983 ) .
s. trebst , m. troyer , z. wang , and a.w.w .
ludwig , progress of theoretical physics supplement no . * 176 * , 384 ( 2008 ) .
a. feiguin , s. trebst , a. w. w. ludwig , et al .
lett . * 98 * , 160409 ( 2007 ) .
l. fidkowski , g. refael , n.e .
bonesteel , and j.e .
moore , phys .
b * 78 * , 224204 ( 2008 ) .
d. poilblanc , a. w.w .
ludwig , s. trebst and m. troyer phys .
b * 83 * , 134439 ( 2011 ) . c. gils , e. ardonne , s. trebst et.al . ,
b * 87 * , 235120 ( 2013 ) .
d. poilblanc , a. feiguin , m. troyer , et .
al , phys .
b * 87 * , 085106 ( 2013 ) .
m. fannes , b. nachtergaele , and r. werner , commun.math .
phys . * 144 * , 443 ( 1992 ) . s. ostlund and s. rommer , phys .
lett . * 75 * , 3537 ( 1995 ) .
s. r. white , phys .
lett . * 69 * , 2863 ( 1992 ) .
d. perezgarcia , f. verstraete , m. m. wolf , and j. i. cirac , quantum inf .
* 7 * , 401 ( 2007 ) .
g. vidal , phys .
* 91 * , 147902 ( 2003 ) .
g. vidal , phys .
lett . * 93 * , 040502 ( 2004 ) .
a. j. daley , c. kollath , u. schollwock , and g. vidal , j. stat .
exp . , p04005 ( 2004 ) .
s. r. white and a. e. feiguin , phys .
lett . , * 93 * , 076401 ( 2004 ) .
u. schollwock , j. phys .
soc . jpn.,*74s
* , 246 ( 2005 ) .
g. vidal , phys .
lett . , * 98 * , 070201 ( 2007 ) .
y. shi , l .- m . duan and g. vidal , phys .
a , * 74 * , 022320 ( 2006 ) . g. vidal , phys .
lett . , * 99 * , 220405 ( 2007 ) .
g. vidal , phys .
* 101 * , 110501 ( 2008 ) .
s. trebst , m. troyer , z. wang , a. w. w. ludwig , prog .
. supp . * 176 * , 384 ( 2008 ) .
r. n. c. pfeifer , p. corboz , o. buerschaper , et al .
b * 82 * , 115126 ( 2010 ) .
r. koenig and e. bilgin phys .
b * 82 * , 125118 ( 2010 ) .
v. zatloukal , l. lehman , s. singh , j.k .
pachos , and g.k .
brennen , arxiv:1207.500 .
l. lehman , v. zatloukal , g.k .
brennen , j.k .
pachos , and z. wang , phys .
* 106 * , 230404 ( 2011 ) .
l. lehman , d. ellinas , and g.k .
brennen , journal of computational and theoretical nanoscience , * 10 * , 1634 - 1643 ( 2013 ) .
the canonical bipartite decomposition of a state @xmath54 of anyonic lattice @xmath50 with total charge @xmath20 that may be different from the vacuum can be constructed as follows .
consider the lattice @xmath445 that is obtained by attaching a `` virtual '' anyon with charge @xmath446 to the right of @xmath50 .
consider state @xmath447 on @xmath445 that has total vacuum charge and that is obtained from @xmath54 as + @xmath448}})_{\beta a^*_\mathrm{tot}}^{\beta ' } n_{a_\mathrm{tot } a^*_\mathrm{tot}}^1\psi_{\alpha\beta},\nonumber\ ] ] where we have fused indices @xmath368 and @xmath446 into a total index @xmath449 to obtain a bipartite decomposition of @xmath447 .
next , construct the schmidt decomposition @xmath84 } } , \lambda,\gamma^{[\mathcal{b}]'})$ ] of state @xmath450 as described in sec .
[ ssec : bidecom ] .
the canonical bipartite decomposition of state @xmath54 is given by tensors @xmath84 } } , \lambda , \gamma^{^{[\mathcal{b}]}})$ ] where tensor @xmath83}}$ ] is obtained from tensor @xmath451'}$ ] as @xmath452a_\mathrm{tot}}_{\mu \beta } = \sum_{\beta ' } ( f^{bb^*a_\mathrm{tot}}_{a_\mathrm{tot}})_{1b ' } n_{a_\mathrm{tot } a^*_\mathrm{tot}}^1 ( \tilde{n}^{\tiny \mbox{[out]}})_{\beta a^*_\mathrm{tot}}^{\beta ' } \gamma^{[\mathcal{b } ] ' 1}_{\mu \beta'}. \nonumber\ ] ] k. hikami , ann .
phys . * 323 * , 1729 ( 2008 ) .
r. n. c. pfeifer , arxiv:1310.0373 [ cond-mat.str-el ] ( 2013 ) . when mapping from a system of ising anyons to su(2)@xmath453 , note that the braid tensors differ .
this detail is , however , unimportant for the present study as no braiding is involved in either the definition of the hamiltonian ( [ eq : ising ] ) or in the implementation of the anyonic mps update . c. gomez , m. ruiz - altaba , and g. sierra , _
quantum groups in two dimensional physics _ , cambridge university press ( 1996 ) .
for a @xmath180 charge placed on each site of the lattice the bipartite decomposition of eq .
( [ eq : aschmidt ] ) for a partition @xmath454:[r\!+\!1 \ldots l]$ ] of the chain has only charges @xmath342 and @xmath195 when @xmath218 is even , and only charge @xmath180 when @xmath218 is odd . for a translationally invariant hamiltonian
, the infinite tebd algorithm produces an mps approximation to the ground state that is only invariant under translations by two sites .
that is , the ground state is described by an infinite mps that consists of repeating tensors @xmath455 } , \gamma^{[a ] } , \lambda^{[b ] } , \gamma^{[b]}$ ] .
the ground state energies per site listed in table [ table : gsenergy ] correspond to the averages over even ( @xmath456 ) and odd ( @xmath457 ) bonds of the mps .
this restriction could be relaxed without significant obstacles provided an infinite contact term is present to prohibit two anyons occupying the same position on the ladder .
for open boundary conditions , the 2-site terms at the boundaries differ slightly from the terms in the bulk , with the matrix elements in eq .
( [ eq : hamnew ] ) corresponding to tunneling on sites @xmath342 and @xmath3 being equal to @xmath280 ( @xmath458 ) instead of @xmath459 ( @xmath460 ) . the @xmath218 pinned anyons in an island may be linearly ordered in an arbitrary way to define the action of the @xmath461strand braid group .
an onsite global symmetry means that the symmetry group acts identically on each site of the lattice .
s. singh , h .- q .
zhou , and g. vidal , new j. phys .
* 12 * 033029 ( 2010 ) .
s. singh , r. n. c. pfeifer and g. vidal , phys .
a * 82 * , 050301 ( 2010 ) .
s. singh , r. n. c. pfeifer and g. vidal , phys .
b * 83 * , 115125 ( 2011 ) . s. singh and g. vidal , phys . rev .
b * 86 * , 195114 ( 2012 ) . | matrix product states ( mps ) have proven to be a very successful tool to study lattice systems with local degrees of freedom such as spins or bosons .
topologically ordered systems can support anyonic particles which are labeled by conserved topological charges and collectively carry non - local degrees of freedom . in this paper
we extend the formalism of mps to lattice systems of anyons .
the anyonic mps is constructed from tensors that explicitly conserve topological charge .
we describe how to adapt the time - evolving block decimation ( tebd ) algorithm to the anyonic mps in order to simulate dynamics under a local and charge - conserving hamiltonian .
to demonstrate the effectiveness of anyonic tebd algorithm , we used it to simulate @xmath0 the ground state ( using imaginary time evolution ) of an infinite 1d critical system of ( a ) ising anyons and ( b ) fibonacci anyons both of which are well studied , and @xmath1 the real time dynamics of an anyonic hubbard - like model of a single ising anyon hopping on a ladder geometry with an anyonic flux threading each island of the ladder .
our results pertaining to @xmath1 give insight into the transport properties of anyons .
the anyonic mps formalism can be readily adapted to study systems with conserved symmetry charges , as this is equivalent to a specialization of the more general anyonic case . |
the trend in solar cell fabrication towards ultrathin wafers , a cost effective and environment - friendly use of materials involves the exploration of additional nanostructured layers to enhance intrinsic quantum efficiencies .
nanotechnology has opened up different routes to reduce front surface reflection , @xcite boost internal absorption through up- and down - conversion @xcite and increase light trapping employing particle layers as back reflectors .
@xcite both regular and irregular ( self - assembled ) arrays of both dielectric and plasmonic particles have been investigated for improving light management in 3rd generation solar cells .
@xcite light harvesting has mostly exploited collective and scattering properties , but also field enhancement effects on the nanoparticle surfaces .
while regular particle arrays typically show narrow spectral resonances , randomly distributed particles yield a broader optical response at weaker coupling strengths .
@xcite broadband total absorption can thus also be achieved in a regular particle array combining different particle sizes and shapes to create a spectral overlap of resonances .
this has been demonstrated experimentally for metamaterials in the mid - infrared region @xcite and total absorption was also predicted for textured graphene surfaces @xcite .
we emphasize that in photovoltaic devices it is not enough to reduce the overall device reflectivity or boost its overall absorption : the absorption of the photo - active region must be enhanced and parasitic absorption ( resonant absorption in the textured layer , ohmic heating in metal components etc . ) reduced .
@xcite with this huge optimization task at hand , computational nanophotonics is a key ingredient to study various configurations and material compositions to assure a high density of optical modes in the active region of a solar cell .
@xcite in this article , we investigate regular particle arrays as textured front layers made of unit cells that include more than a single particle .
we demonstrate that already using two differently sized nanodisks of the same material can increase the internal absorption , i.e. the absorption at the active device level , corrected by losses in the particle array itself .
we study the effect of mixing up to four different particle sizes , including aspects of disk height .
the structure is introduced in the next section and general aspects on the computational procedure are made in section 3 .
figure [ fig1 ] illustrates the setup studied : a particle layer with space - modulated permittivity @xmath0 is placed on top of a si wafer ( permittivity @xmath1 and constant thickness @xmath2 m in order to suppress fabry - prot like oscillations in the calculated absorption ) which is supported on a sio@xmath3 substrate with permittivity @xmath4 .
both lower layers are considered to be homogeneous and described by a wavelength - dependent permittivity .
the particle layer on top of them scatters the incoming light ( air @xmath5 , no standard anti - reflection coating used ) into a number of diffraction orders for both reflected and transmitted light and thus potentially increases absorptivity in the photo - active , homogeneous si region underneath by means of an increased optical path length , and to a minor degree through field enhancement effects directly at the interface between the particle layer and the solar cell wafer .
in particular , we want to study regular particle layers that consist not only of the same particles , but different types of disk - shaped particles with equal height @xmath6 either varying in radius or material and combinations of those .
[ fig2](a ) depicts such unit cells , where we use the term 1-type for a unit cell with a single particle , a unit cell with @xmath7 particles is denoted 2-type and a unit cell with @xmath8 particles 4-type , since we restrict ourselves to the symmetric cases shown in fig .
[ fig2](a ) .
the modulation of the particle layer is determined by the periodicity @xmath9 of the unit cell which is also the particle to particle separation for 1-type particle layers .
we exploit here the plane - wave expansion method to calculate the properties of the particle layer and a scattering matrix approach to obtain the overall optical properties of the structure .
@xcite in order to describe the material properties of the particle layer , a fourier transform of the real - space , piecewise defined permittivity within the unit cell of area @xmath10 of the regular structure is made @xmath11 which is analytical for 2d circular shapes , see for instance @xcite .
this allows for a straightforward extension to include several particles within a single unit cell .
while the computational effort is increased by a finite summation over the number of particles in the unit cell , one has to keep in mind that the size of the unit cell increases and thus , in order to obtain results of equal quality , a similarly higher number of plane - waves has to be considered at the same time , which is the main drawback in this approach .
note that this approach does allow for asymmetric , including random,@xcite particle distributions and an arbitrary assignment of material properties , which is , however , not in the scope of this work .
we study mainly the dielectric si and the metal ag taking the permittivity of these materials from tabulated , experimental data .
si has a low refractive index contrast to air and convergence is reached faster , i.e. for a lower number of plane waves taken into account than for metal particle arrays which have large negative permittivity in the visible spectrum .
note furthermore , that averaging over orthogonal polarization directions is not necessary as long as we maintain a symmetric unit cell for the 2d photonic crystal layer as depicted in fig .
[ fig2](a ) .
we calculate the power flux reflected from the outer ( inner ) surface in front ( behind of ) the particle layer and the power flux transmitted into the substrate , normalize to the incoming power flux and obtain from this the overall device absorbance @xmath10 and the absorbance @xmath12 of the photo - active wafer region , which does exclude parasitic ( ohmic ) absorption in the particle layer itself .
this is compared relative to the bare reference cell @xmath13 , see fig .
[ fig2](b ) , for the different configurations discussed in the next section .
we give in fig .
[ fig2](c ) the absolute values and include for the absorbance @xmath12 the contribution of the lowest diffraction order ( i.e. , where the reciprocal lattice vector @xmath14 ) . whilst this is the dominant contribution at long wavelengths , a clear error would be made , neglecting higher order contributions .
two spectral regions can be identified from fig .
[ fig2](c ) which are significant in the assessment of such setups . in one case
, the absorption in the photo - active region is larger than for the reference cell and close to the overall device absorption . in the other case , while the total device absorption reaches up to 100% , the absorption within the wafer region decreases below the one of the reference cell . in this latter case
, the considered particle layer absorbs most of the light ( ohmic heating ) which in return is no longer available for the active region of the solar cell . in the graphs shown throughout this work , we concentrate on the absorption in the photo - active region relative to the absorbance of the reference cell @xmath15 as the quantity of main interest .
furthermore , we compute the short circuit current @xmath16 for the solar cells with particle arrays relative to the bare reference cell @xmath17 as well as the average absorbance within the photo - active region in the considered spectral range ( @xmath18 nm ) @xmath19 note that eq . takes the absorption in the photo - active region rather than the total device absorption into account .
assuming an ideal internal quantum efficiency @xmath20 would yield for the external quantum efficiency @xmath21 with @xmath10 being the total absorption including all effects of losses , reflected or back - scattered light in the total device .
this is the accessible quantity in an experiment , but here , we can also address the losses within the particle array , and thus have a further correction to the @xmath22 , which is the efficiency of carriers collected in the active si wafer region of the device .
thus , we use directly @xmath12 which gives the light absorbed in active region , does account for the losses introduced on employing the particle array and is thus a more suitable choice for a direct quantity to compare with @xmath23 .
this gives us two quantitative values to assess the efficiency enhancement found for the different setups on the overall ( solar ) spectrum .
since these quantities are independent of a specific wavelength , they are ideal to study the influence of various geometrical parameters on them .
we give an overview of optical properties of 1-type particle layers in fig .
[ fig3 ] for disk arrays with lattice period @xmath24 nm in terms of the absorption in the photo - active region relative to the bare si wafer absorbance .
we study the dielectric material si and the metal ag , with permittivities taken from tabulated , experimental data .
an increase of over 50% in the absorbance ( at specific wavelengths ) is seen here , which improves even further for higher disks as known from black silicon @xcite .
note that gaas produces very similar results to si comparing spectra of @xmath15 as a function of both disk height and disk radius ( not shown ) . while keeping constant height @xmath25 nm and @xmath26 nm , and ( c ) , ( d ) scaling the height for these cells with constant @xmath26 nm and @xmath27 nm .
the curves plotted on the left hand side of each contour show the spectra that are further evalutated in table [ tab.si2type ] for si . ]
the spectral dependence becomes more involved with metal disks , as they show a multitude of sharp resonance features ( reflection bands resulting from collective free electron modes ) . here
, a careful tailoring of geometrical properties is needed to achieve the desired improvement in the absorbance inside the photo - active region .
the overall device absorption is readily enhanced mostly via parasitic absorption in the metal particle array .
the resulting reduction in overall reflection has prompted plasmonic particle layers ideal for implementation in photovoltaic devices , but the internal quantum efficiency suffers from these losses which can not be assessed easily in an experiment and are not counterbalanced by adding a standard arc coating .
@xcite this can already be seen for the absorbance at low wavelengths for si in fig . [ fig2](b ) and ( c ) , and we come back to this later on . the main advantage that can be expected from employing metal disks is to enhance the absorbance at lower wavelengths ( below @xmath28 nm in case of al , between @xmath29 nm in case of ag ) in a setup using the particle layer as a back reflector for long wavelengths .
.efficiency enhancement calculated from the short circuit current and average absorption relative to the reference cell for si - si 2-type cells from fig .
[ fig4](a ) and ( c ) for different particle sizes and heights . [ cols="^,^,^,^,^,^,^,^",options="header " , ] similarly fig . [ fig4](b ) and ( d ) show these results using ag as the particle material . here ,
too , spectral regions with an increase in the absorbance @xmath12 can be seen . however , it is not strong enough to yield positive efficiency boosts ( integrated over the whole spectrum ) in the photo - active region . on the other hand , the overall device absorption shown in fig .
[ fig45 ] ( comparing with fig . [ fig4](b ) ) yields an increase in absorbance and short circuit current , which underlines the importance to separate between the benefits of the particle layer and parasitic absorption effects arising from its employment .
the results calculated from the overall absorbance @xmath10 lead to believe that this particle array has a positive effect on the device performance with an increase in both short circuit current @xmath16 and average absorption @xmath30 . the opposite is found experimentally showing a considerable reduction in the internal quantum efficiency at lower wavelengths.@xcite this is summarized in terms of the relative efficiency enhancement in table [ tab.agsi2type ] .
[ fig5](a ) shows 2-type particle arrays combining the two different materials si and ag as a function of disk height .
these mixed unit cells demonstrate the spectral overlap of characteristic features from both 1-type cells ( fig .
[ fig3 ] ) or pure material 2-type cells ( fig .
[ fig4 ] ) : the sharp resonance features ( resulting in increased reflection or parasitic absorption ) of the ag particles and the significant , almost size - independent enhancement due to the si particles . the light management improves in comparison to the pure ag particle layer , but shows no further improvement with respect to the pure si layer . in fig .
[ fig5](b ) the ag disk radius is @xmath31 nm and the si disk radius is changed .
an increase of the efficiency in the mixed 2-type unit cells can be seen only for ag disks of smaller size than the si disks , cp .
table [ tab.agsi2type ] , and we conclude that employing ag is detrimental to the absorbance in the active cell region for all configurations considered here . while the effect of combining different materials is easily demonstrated with this theoretical method , it remains difficult to achieve in experiments .
we furthermore consider 4-type particle layers , where we combine si disk particles of four different disk radii in fig .
[ fig7 ] , namely @xmath32 nm and a periodicity of @xmath33 nm . the efficiency enhancement in this example is @xmath34 within the active region . despite the use of several smaller disks , this constitutes an increase in the efficiency compared to the 2-type cells which underlines the effect of boosting the absorption on providing a multitude of optical modes assisted by the particle layer . the increased difference between lowest order diffraction response and full calculation
show the importance of diffracted light in the increase of absorbance .
the increase in computational time for this example is moderate .
all the simulation results obtained here are without standard arc coating from which an additional increase in the observed absorbance , resulting in a higher short circuit current , can be expected as it counterbalances reflection losses from back - scattering of the particle array .
however , it can not counterbalance losses due to increased absorption within the particle array .
in summary , we presented the absorbance within the photo - active region of a si - wafer with a regular array of nanoparticles on top , using a combination of particles of both different size and materials in order to further enhance the short circuit current ( up to @xmath35 ) and average absorption ( up to @xmath36 ) relative to the bare reference cell .
we assumed the absorption within the additional front layer of disk - shaped particles as parasitic for the enhancement of the photovoltaic effect and only considered the absorbance within the homogeneous photo - active region .
broadband efficiency enhancement in this active region is observed for pure dielectric ( here si ) particle layers .
metal ( here ag ) particle layers yield a much higher overall absorbance , but parasitic absorption losses - mostly due to ohmic heating - lead to a decrease in the internal quantum efficiency which can could be readily demonstrated in the presented theoretical calculations .
thus , only a partial improvement within the spectral range considered could be shown , but no broadband efficiency enhancement of the wavelength - independent short circuit current or average absorption was found .
combining particles of different radii shows an increase in the absorbance compared to single type unit cells for a range of particle size combinations . adjusting the height of the particle layer
can further improve the benefit observed . a multi - type particle layer made of a single material of varying sizes
can be fabricated with e.g. ebl techniques .
the creation of a spectral overlap of characteristic features in the optical properties of particle arrays was demonstrated by combining the different materials si and ag .
only a moderate increase in the computational effort results from the employment of multi - type unit cells and it remains a fast and reliable method . combining a larger number of differently sized particles improves the absorbance further due to a higher density of optical modes provided by the nanostructured front surface layer .
the author thanks n. a. mortensen and j. p. connolly for stimulating discussions and the deutsche forschungsgemeinschaft ( german research foundation ) for financial support through a dfg research fellowship . the author would furthermore like to acknowledge the contribution of the european cost action mp1406 multiscalesolar . | we discuss regular particle arrays as nanostructured front layers for 3rd generation photovoltaic devices .
a rigorous plane - wave method is used to investigate multi - type particle layers combining different radii and configurations .
the absorbance is enhanced compared to the bare wafer and on mixing particles we demonstrate a broadband boost in the absorbance within the photo - active region .
efficiency enhancement in terms of short circuit current is studied for varying geometries .
si multi - type layers achieved up to 33% improvement without yet standard anti - reflection coatings .
metal multi - type layers show strong parasitic absorption and boost the absorbance only in narrow wavelength regions and no efficiency enhancement is observed . 10 h. m. branz , v. e. yost , s. ward , k. m. jones , b. to , and p. stradins , `` nanostructured black silicon and the optical reflectance of graded - density surfaces , '' applied physics letters * 94 * , 231121 ( 2009 ) .
y. liu , t. lai , h. li , y. wang , z. mei , h. liang , z. li , f. zhang , w. wang , a. y. kuznetsov , and x. du , `` nanostructure formation and passivation of large - area black silicon for solar cell applications , '' small * 8 * , 13921397 ( 2012 ) .
f. corts - juan , c. chaverri ramos , j. p. connolly , c. david , f. j. garca de abajo , j. hurtado , v. d. mihailetchi , s. ponce - alcntara , and g. snchez , `` effect of ag nanoparticles integrated within antireflection coatings for solar cells , '' journal of renewable and sustainable energy * 5 * , 033116 ( 2013 ) .
s. fischer , j. c. goldschmidt , p. lper , g. h. bauer , r. brggemann , k. krmer , d. biner , m. hermle , and s. w. glunz , `` enhancement of silicon solar cell efficiency by upconversion : optical and electrical characterization , '' journal of applied physics * 108 * , 044912 ( 2010 ) .
z. yuan , g. pucker , a. marconi , f. sgrignuoli , a. anopchenko , y. jestin , l. ferrario , p. bellutti , and l. pavesi , `` silicon nanocrystals as a photoluminescence down shifter for solar cells , '' solar energy materials and solar cells * 95 * , 12241227 ( 2011 ) .
x. huang , s. han , w. huang , and x. liu , `` enhancing solar cell efficiency : the search for luminescent materials as spectral converters , '' chem .
soc . rev .
* 42 * , 173201 ( 2013 ) . v. e. ferry , l. a. sweatlock , d. pacifici , and h. a. atwater , `` plasmonic nanostructure design for efficient light coupling into solar cells , '' nano lett . * 8 * , 43914397 ( 2008 ) .
a. basch , f. j. beck , t. sderstrm , s. varlamov , and k. r. catchpole , `` combined plasmonic and dielectric rear reflectors for enhanced photocurrent in solar cells , '' appl .
. lett . * 100 * , 2439035 ( 2012 ) .
a. basch , f. beck , t. sderstrm , s. varlamov , and k. r. catchpole , `` enhanced light trapping in solar cells using snow globe coating , '' prog .
photovolt : res
. appl .
* 20 * , 837842 ( 2012 ) .
m. l. brongersma , y. cui , and s. fan , `` light management for photovoltaics using high - index nanostructures , '' nat mater * 13 * , 451460 ( 2014 ) .
y. a. akimov , w. s. koh , s. y. sian , and s. ren , `` nanoparticle - enhanced thin film solar cells : metallic or dielectric nanoparticles ? '' appl .
. lett . * 96 * , 0731113 ( 2010 ) .
p. spinelli , m. a. verschuuren , and a. polman , `` broadband omnidirectional antireflection coating based on subwavelength surface mie resonators , '' nature communications * 3 * , 692 ( 2012 ) . k. r. catchpole and a. polman , `` design principles for particle plasmon enhanced solar cells , '' appl .
. lett .
* 93 * , 1911133 ( 2008 ) .
x. chen , b. jia , y. zhang , and m. gu , `` exceeding the limit of plasmonic light trapping in textured screen - printed solar cells using al nanoparticles and wrinkle - like graphene sheets , '' light sci appl * 2 * , e92 ( 2013 ) .
h. cui , s. pillai , p. campbell , and m. green , `` a novel silver nanoparticle assisted texture as broadband antireflection coating for solar cell applications , '' solar energy materials and solar cells * 109 * , 233239 ( 2013 ) .
c. david , j. p. connolly , c. chaverri ramos , f. j. garca de abajo , and g. snchez plaza , `` theory of random nanoparticle layers in photovoltaic devices applied to self - aggregated metal samples , '' solar energy materials and solar cells * 109 * , 294299 ( 2013 ) .
x. liu , t. tyler , t. starr , a. f. starr , n. m. jokerst , and w. j. padilla , `` taming the blackbody with infrared metamaterials as selective thermal emitters , '' phys .
. lett . * 107 * , 045901 ( 2011 ) . b. zhang , y. zhao , q. hao , b. kiraly , i .-
c .
khoo , s. chen , and t. j. huang , `` polarization - independent dual - band infrared perfect absorber based on a metal - dielectric - metal elliptical nanodisk array , '' opt .
express * 19 * , 1522115228 ( 2011 ) .
s. thongrattanasiri , f. h. l. koppens , and f. j. garca de abajo , `` complete optical absorption in periodically patterned graphene , '' phys .
rev .
lett . * 108 * , 047401 ( 2012 ) .
m. b. dhring , n. asger mortensen , and o. sigmund , `` plasmonic versus dielectric enhancement in thin - film solar cells , '' applied physics letters * 100 * , 211914 ( 2012 ) .
s. e. han and g. chen , `` toward the lambertian limit of light trapping in thin nanostructured silicon solar cells , '' nano lett .
* 10 * , 46924696 ( 2010 ) .
m. g. deceglie , v. e. ferry , a. p. alivisatos , and h. a. atwater , `` design of nanostructured solar cells using coupled optical and electrical modeling , '' nano lett .
* 12 * , 28942900 ( 2012 ) .
d. m. whittaker and i. s. culshaw , `` scattering - matrix treatment of patterned multilayer photonic structures , '' phys . rev .
b * 60 * , 26102618 ( 1999 ) .
v. liu and s. fan , `` s4 : a free electromagnetic solver for layered periodic structures , '' computer physics communications * 183 * , 22332244 ( 2012 ) .
r. guo , l. you , y. zhou , z. shiuh lim , x. zou , l. chen , r. ramesh , and j. wang , `` non - volatile memory based on the ferroelectric photovoltaic effect , '' nat commun * 4 * , ( 2013 ) .
m. kroll , m. otto , t. kebier , k. fchsel , r. wehrspohn , e .- b .
kley , a. tnnermann , and t. pertsch , `` black silicon for solar cell applications , '' proc .
spie 8438 , photonics for solar energy systems iv * 8438 * , 84381784381711 ( 2012 ) . |
while elemental abundances in the photosphere of a - type stars , which are of particular interest because of the chemical anomalies of various types , have been extensively studied by a number of investigators so far , those of the alkali elements ( li , na , and k ) are barely known even nowadays , despite of their importance in stellar spectroscopy .
this is presumably due to the fact that these elements ( with one valence electron weakly bound ) are characterized by fairly low ionization potential ( @xmath12 45 ev ) and most of them are ionized in the atmospheric condition of early - type stars , leaving only a tiny fraction as neutral .
since detecting spectral lines of the dominant ionized species is almost hopeless ( because of the closed - shell configuration requiring vary large excitation energy ) , one is obliged to measure the very weak lines of neutral species ( li i , na i , k i ) . thus , the necessity of using sufficiently high - s / n spectra must have hampered past researchers from investigating the abundances of these elements .
still , sodium has been comparatively well studied ( e.g. , lane & lester 1987 ; varenne & monier 1999 ; gebran et al . 2008 , 2010 ; gebran & monier 2008 ; fossati et al .
2007 , 2008 ) mainly by using subordinate lines ( na i 5682/5688 or 6154/6160 ) of moderate strengths ( i.e. , neither too strong nor too weak ) , which are less affected by the non - lte effect . however , na abundance determinations in a - type stars conducted so far are restricted to sharp - lined late - a stars ( @xmath13 k ) as well as f - type stars , and thus na compositions of early - a through late - b stars ( @xmath14 k ) remain essentially unexplored , except for our previous non - lte investigations for the very bright stars of sirius ( takeda & takada - hidai 1994 ) and vega ( takeda 2008 ; hereinafter refereed to as paper i ) , and the recent lte studies of fossati et al . ( 2009 ) for two sharp - lined late - b stars ( 21 peg and @xmath15 cet ) .
recently , takeda et al .
( 2009 ; hereinafter referred to as paper ii ) carried out an extensive non - lte analysis of na i resonance d lines at 5890/5896 @xmath16 for a large ( 122 ) sample of a - type stars ( 7000 k @xmath17 k ) including rapid as well as slow rotators , with an aim to establish their na abundance behaviors in general .
unfortunately , however , since the d lines are so strong and the resulting abundances are quite sensitive to the microturbulent velocity @xmath18 , it turned out that ambiguities in the choice of @xmath18 ( which is likely to be depth - dependent and difficult to assign a value relevant for the high - forming strong d lines ) prevented from precise sodium abundance determinations .
almost the same situation holds for lithium : determinations of li abundances in a - type stars have been essentially restricted to late - a stars with @xmath13 k ( e.g. , burkhart & coupry 1991a , 1991b , 1995 , 2000 ; burkhart et al .
2005 ; north et al . 2005 ) , while early - a stars ( @xmath14 k ) remain practically untouched . admittedly , this is a very difficult and challenging task , as the li i 6708 line considerably weakens as @xmath4 becomes higher .
the work of paper i may be counted as an exceptional case , where the li abundance of vega could somehow be evaluated from the very weak li i 6708 line ( @xmath19 m@xmath16 ) based on a spectrum of ultra - high s / n ratio .
coupry and burkhart ( 1992 ) reported the @xmath20(li i 6708 ) of @xmath21 peg ( a well - known very sharp - lined early - a star ) to be 1.3 m@xmath16 , which was used in paper i ( cf . appendix a therein ) to argue that the li abundance of @xmath21 peg is almost consistent with the solar - system composition .
however , as will be shown in this paper ( cf . subsection 4.2 ) , we consider that their measurement is an erroneous overestimation and li is definitely underabundant in @xmath21 peg .
this example well illustrates the enormous difficulty of li abundance derivation in hot a - type stars .
investigations of potassium abundances are even more lacking . as for late - a stars ( @xmath13 k ) ,
it is ( to our knowledge ) only the study of fossati et al .
( 2007 ) that reported the k abundances of late - a and am stars in the praesepe cluster , though their neglect of the non - lte effect ( which may be significant for such a resonance line as k i 7699 ) prevented from obtaining reliable results ( cf .
subsection 4.4 therein ) .
meanwhile , almost nothing has been reported so far regarding the potassium abundances in early - a stars , except for paper i where vega s k abundance could be successfully established .
it is thus evident that our current knowledge on the behaviors of alkali - element abundances in a - type stars is considerably insufficient .
given this situation , we decided to carry out a systematic spectroscopic study aiming at establishing the photospheric abundances of na , li , and k for an homogeneous sample of a - dwarfs ( 7000 k @xmath17 k ) based on the spectral data of high quality . since measurements of very weak lines are concerned in the present case ( especially for early - a stars ) , we had to confine ourselves only to sharp - lined stars ( @xmath0 km s@xmath1 ) .
this inevitably leads to the inclusion of a number of am stars ( metallic - lined a stars ) , which are characterized by deficiencies in comparatively light elements ( c , n , o , ca , sc ) as well as excesses of heavier elements ( such as fe - group or neutron - capture elements ) , since am peculiarities are preferably observed in slow rotators ( such as frequently seen in binaries ) .
accordingly , the purpose of this paper is to clarify and discuss the behaviors of na , li , and k abundances in slowly - rotating a - type stars based on the results of our analysis , while paying attention to the possible connection ( if any exists ) between these alkali elements and the appearance / degree of am anomalies .
we selected 21 sharp - lined stars uma ; magnetic ap star ) and hd 40932 ( @xmath22 ori ; double - lined binary ) because of their apparently unusual spectra , despite that these two stars have @xmath23 values less than 50 km s@xmath1 . ] satisfying @xmath0 km s@xmath1 from 122 a - type stars , studied in paper ii , where na i 5890/5896 d lines were analyzed based on the spectra ( @xmath24 and s / n ratios of a few hundreds ) obtained by using boes ( bohyunsan observatory echelle spectrograph ) attached to the 1.8 m reflector at bohyunsan optical astronomy observatory ( boao ) .
we call these 21 stars as `` boao sample . ''
see section 2 of paper ii ( and electronic table e therein ) for more details about these observational data . besides
, in order to augment the spectral data for early - a stars for which very high - quality is required to measure very weak lines of neutral alkalis , we also used the spectra of 7 bright sharp - lined stars ( @xmath15 dra , @xmath25 lyr , @xmath25 cma , @xmath26 gem , 15 vul , 68 tau , and @xmath21 peg ) , which were observed with the hides spectrograph ( high dispersion echelle spectrograph ; izumiura 1999 ) attached to the 1.88 m reflector at okayama astrophysical observatory ( oao ) .
while the data of @xmath25 lyr were taken from the digital atlas ( based on the observations carried out on 2006 may 14 ) published by takeda , kawanomoto , and ohishi ( 2007 ) as in paper i , those for the remaining 6 stars were secured by the observations on 2008 october 4 ( 15 vul , 68 tau , @xmath21 peg ) , october 7 ( @xmath25 cma , @xmath26 gem ) , and october 8 ( @xmath15 dra ) .
these 2008 oao data , which cover the wavelength range of 41007800 @xmath16 with three mosaicked ccds of 2k@xmath274k pixels , have high spectral resolution ( @xmath28 ) and very high s / n ratio ( typically @xmath29 , or even higher for the very bright @xmath25 cma and @xmath26 gem ) .
we call these 7 stars as `` oao sample . ''
note that the following 4 stars are common to both of the boao and oao samples : hd 172167=@xmath25 lyr , hd 48915=@xmath25 cma , hd 47105=@xmath26 gem , and hd 27962=68 tau .
accordingly , the actual ( net ) number of the targets is 24 ( @xmath30 ) .
the list of these program stars is presented in table 1 .
regarding the effective temperature ( @xmath4 ) and the surface gravity ( @xmath31 ) of each program star , we simply adopted the values used in paper ii for 21 boao sample stars ( including 4 oao samples in common ) .
as for the remaining 3 oao stars , those for 15 vul and @xmath21 peg were determined photometrically from strmgren s @xmath32 , @xmath33 , and @xmath34 colors along with the @xmath35 index by using napiwotzki et al.s ( 1993 ) calibration ( as in paper ii ; cf .
subsection 3.1 therein ) , while the values derived by adelman ( 1996 ) were adopted for @xmath15 dra ( since color data were not available for this star ) .
the final values of @xmath4 and @xmath31 are summarized in table 1 .
as can be seen in figure 1 , where the targets are plotted on the theoretical hr diagram , the mass values of the program stars range between @xmath121.5 @xmath36 and @xmath373 @xmath36 .
the model atmospheres for each of the stars were constructed by two - dimensionally interpolating kurucz s ( 1993 ) atlas9 model grid in terms of @xmath4 and @xmath31 , where we exclusively applied the solar - metallicity models as in takeda et al .
( 2008a ) as well as paper ii .
( 50mm,50mm)figure1.eps since we need to clarify the characteristics ( e.g. , parameters such as rotation or atmospheric turbulence , degree of abundance peculiarities ) of our sample stars , a synthetic spectrum fitting analysis based on takeda s ( 1995 ) best - fit solution search algorithm was applied to the 61466163 @xmath16 region , in order to establish the projected rotational velocity ( @xmath23 ) , microturbulent velocity dispersion ( @xmath18 ) , and the abundances of iron and oxygen [ @xmath3(fe ) and @xmath3(o ) ; both are good indicators of am anomaly ; cf .
takeda and sadakane ( 1997 ) ] ( along with that of na as a by - product ) , similarly to what was done in takeda et al .
( 2008a ; cf . section iv therein ) and paper ii ( cf .
subsection 3.2 therein ) . actually , we varied seven free parameters @xmath18 , @xmath23 , @xmath3(o ) , @xmath3(na ) , @xmath3(si ) , @xmath3(ca ) and @xmath3(fe ) to accomplish the best fit , where the non - lte effect was taken into account for o ( as in takeda et al .
2010 ) as well as for na ( as in paper ii ) . the adopted atomic parameters of important o , na , and fe lines relevant to this region are given in table 2 .
the instrumental broadening corresponding to the spectral resolving power was also taken into account , since sharp - lined stars are concerned here .
since we could not find a reasonable @xmath18 solution for the following five boao cases ( i.e. , stabilized at an apparently unreasonable value , or any convergence could not be accomplished ) , we assumed an appropriate @xmath18 value as fixed and repeated the iterations until convergence . @xmath38 km s@xmath1 ( adelman 1996 ; monier 2005 ) for hd 95418 , @xmath39 km s@xmath1 ( caliskan & adelman 1997 ) for hd 43378 , @xmath40 km s@xmath1 ( sadakane 2006 ) for hd 218396 , @xmath41 km s@xmath1 ( from the analytical formula derived by takeda et al .
2008 ) for hd 204188 , and @xmath42 km s@xmath1 ( @xmath18 solution from the oao spectrum ) for hd 48915 .
the finally adopted results of @xmath18 , @xmath23 , @xmath43(fe ) @xmath43(o ) , are presented in table 1 as well as in the electronic table e ( where @xmath43(na ) is also given ) .
how the theoretical and observed spectra match each other with the converged solutions of these parameters is shown in figure 2 .
( 50mm,200mm)figure2.eps in figure 3 , the resulting @xmath3(o ) and @xmath3(fe ) values are plotted against @xmath23 and @xmath4 , and their mutual correlation is also displayed .
we can see from these figures that the underabundance of o and the overabundance of fe ( both are the characteristics of am peculiarities ) are surely anti - correlated ( figure 3c ) and their weak @xmath23-dependence is recognized ( figures 3a , b ) even in such a small range of @xmath23 ( i.e. , larger anomaly for slower rotators ; cf .
takeda & sadakane 1997 ) .
we also notice that two stars ( hd 218396 and hd 172167=vega ) are apparently metal - deficient ( [ fe / h]@xmath44 ) , showing the abundance charateristics of vega - like stars ( cf .
sadakane 2006 ) , which are presumably related to @xmath45 boo stars ( see , e.g. , paunzen 2004 and the references therein ) .
it is interesting that these two appear to follow the o vs. fe anti - correlation trend exhibited by am stars , despite that interaction with interstellar clouds ( cf . kamp & paunzen 2002 ; paunzen et al .
2002 ) is becoming a promising mechanism for explaining the @xmath45 boo phenomenon ( instead of the diffusion process being considered as the standard theory for am stars ) . regarding the microturbulence
, the @xmath18 vs. @xmath4 relation depicted in figure 3e implies that the @xmath18 values resulting from the 61466163 @xmath16 fitting are consistent with the analytical formula proposed by takeda et al .
( 2008a ) ( which was invoked in paper ii ) .
( 80mm,200mm)figure3.eps in the next section ( section 3 ) we will derive not only the abundances ( of na , li , and k ) but also the relevant abundance errors in response to the possible ambiguities in the adopted atmospheric parameters , where we estimate @xmath46 k , @xmath47 dex , and @xmath48% as the typical uncertainties in the absolute values of @xmath4 , @xmath31 , and @xmath18 , respectively . while the internal errors of our color - based @xmath4 and @xmath31 are @xmath49 k ( 2.5% ) and @xmath50 dex according to napiwotzki et al .
cf . section 5 therein ) , we assumed somewhat larger ambiguities than these in view of the comparison with various published values ( cf . section iv - c in takeda et al .
the uncertainty in @xmath18 was estimated from the dispersion 30% ) is regarded as a typical ambiguity in @xmath18 , which can also be recognized by comparing various @xmath18 values published so far ( see , e.g. , figure 1 in coupry & burkhart 1992 ; figure 2 in gebran & monier 2007 ) . yet , apart from this intrinsic dispersion , our @xmath18 values tend to be somewhat larger than the previously reported results ( usually based on @xmath20 values of fe lines of various strengths ) .
for example , our figure 3f suggests that the maximum @xmath18 attained at mid - a type is @xmath12 45 km s@xmath1 , while it is @xmath5 km s@xmath1 in figure 2 of coupry and burkhart ( 1992 ) ; also , our @xmath18 results are systematically higher by @xmath12 1 km s@xmath1 with those of landstreet et al . ( 2009 ) , where 5 stars ( 68 tau , @xmath26 gem , @xmath25 cma , 15 vul , and @xmath21 peg ) are in common with our sample .
we suspect that this may be due to a depth - dependence of @xmath18 decreasing with height ( cf .
section 5 in paper ii ) .
that is , our @xmath18-determination ( based on 61466163 @xmath16 fitting ) tends to reflect the physical condition in deeper layers ( larger @xmath18 ) because of being mainly controled by the deep - forming o i 61558 triplet lines of high excitation , while the conventional method using @xmath20s of high - forming strong lines may yield comparatively lower @xmath18 . ] in the @xmath18 vs. @xmath4 relation ( figure 3f ) .
since resonance lines of alkali elements with low - ionization potentials are concerned in the present case , it is comparatively easy to understand ( at least qualitatively ) how the abundance is affected by changing each parameter .
that is , since almost all atoms are in the ground state of the first - ionized ( closed - shell ) stage , and only a small fraction of them remain neutral , the number population of the ground level of the neutral atoms ( @xmath51 , which is proportional to the line opacity @xmath52 ) is expressed as @xmath53 according to saha s equation ( @xmath54 is the abundance , @xmath55 , @xmath56 is the ionization potential ( @xmath12 45 ev ) , and @xmath57 is the electron density ) .
then , the dependence of the line - opacity ( @xmath52 ) upon @xmath58 ) and @xmath59 may be written as @xmath60 where we put @xmath61 and used the approximation @xmath62 ( @xmath63 1/32/3 depending on @xmath4 ; see , e.g. , gray 2005 ) .
consequently , the abundance ( @xmath54 ) resulting from a given equivalent width is strongly dependent upon @xmath64 as @xmath65 because of the exponential factor ( i.e. , @xmath54 increases as @xmath4 becomes higher ) .
the @xmath31-dependence is somewhat more complicated because the effect of the continuum opacity ( @xmath66 ) has to be taken into consideration . for late a - type stars of
comparatively lower @xmath4 ( @xmath12 70008000 k ) , h@xmath67 opacity is still important and @xmath66 has almost the same @xmath59-dependence as @xmath52 ; thus the abundances turn out rather independent on @xmath59 because of the cancellation in the @xmath68 ratio determining the line strength ( cf .
takeda et al .
meanwhile , for early a stars of higher @xmath4 where paschen continuum opacity of neutral hydrogen is dominant , @xmath59-dependence in @xmath52 directly reflects the line strength and the abundance scales as @xmath69 ( i.e. , @xmath54 decreases as @xmath31 becomes higher ) .
finally , regarding the effect of @xmath18 ( @xmath54 tends to decrease with an increase in @xmath18 ) , the abundance results are practically inert to any change in @xmath18 , since weak lines in the linear part of the curve of growth are concerned in most of the present cases ( though only the na abundances derived from na i 5688 line for late a stars with @xmath70 8000 k exceptionally show some appreciable @xmath18-dependence ; cf . figure 5f ) .
we will show in section 3 that these expected trends can be actually confirmed by actual calculations ( cf . figures 5 , 8 , and 10 ) .
the abundances of na ( @xmath43(na ) ) were already obtained from the fitting analysis in the 61466163 @xmath16 region ( including na i 6154/6160 lines ) as described in subsection 2.3 , which are given in electronic table e. however , while the na i 6154/6160 lines are quite suitable for na abundance determination for late - a stars , they become considerably weak as @xmath4 becomes higher and the accuracy of @xmath43(na ) may be comparatively low at early - a stars , especially for the non - metal - rich case ( actually , @xmath43(na ) of hd 172167=vega in the boao sample could not be determined ; cf .
note 1 in electronic table e ) .
we therefore decided to determine the na abundances from the spectra in the 56805690 @xmath16 region including the na i 5682/5688 lines ( stronger than na i 6154/6160 while not too strong to be seriously sensitive to @xmath18 ) .
as done for the 61466163 @xmath16 region , we applied the fitting procedure to the 56805690 @xmath16 portion of the spectrum , where the telluric lines had been removed by dividing by the spectrum of a rapid rotator ( cf .
paper i , paper ii ) .
we varied five abundance parameters of @xmath3(na ) , @xmath3(si ) , @xmath3(sc ) , @xmath3(fe ) and @xmath3(ni ) to obtain the best fit , while taking into account the non - lte effect for na as in paper ii ( cf .
table 2 for the atomic parameters of na i 5682/5688 ) .
how the theoretical and observed spectra match each other with the converged solutions of these parameters is shown in figure 4 .
the resulting sodium abundances , @xmath71(na ) , are given in table 1 .
besides , as analysis - related quantities , the equivalent widths ( @xmath20 ) inversely computed from the abundance , the corresponding non - lte corrections ( @xmath72 ) , and the abundance variations ( @xmath73 ) in response to each of the perturbations in @xmath4 ( @xmath46 k ) , @xmath31 ( @xmath47 dex ) , and @xmath18 ( @xmath74 ) , which were computed for the na i 5688 line following the procedure described in paper ii ( cf .
subsections 4.3 and 4.4 therein ) , are plotted against @xmath4 in figure 5 .
the ( @xmath20 , @xmath72 , @xmath73 ) results for both na i 5682 and 5688 lines are also presented along with @xmath71(na ) in electronic table e. ( 50mm,200mm)figure4.eps ( 60mm,160mm)figure5.eps the comparison of the two sodium abundances [ @xmath43(na ) and @xmath71(na ) ] derived from different regions / lines is displayed in figure 6 .
we can see from this figure that the agreement is good especially for late - a stars ( red circles ) .
the star showing the largest departure ( @xmath12 0.4 dex ) is vega ( oao sample ) where @xmath43(na ) = 5.63 and @xmath43(na ) = 6.05 were derived , which is due to the low accuracy of @xmath43 caused by the extreme weakness of na i 6154/6160 lines in this star ( cf .
figure 3 in paper i).(na ) = 5.63 ) based on the 61466163 @xmath16 region fitting is considerably different from the @xmath20-based value of 6.26 ( @xmath75 ; cf .
table 1 in paper i ) , despite that both are based essentially on the same lines and the same non - lte corrections .
we understand that the reason for this marked discrepancy is the erroneous underestimation of the 61466163 @xmath16 fitting - based result ( 5.63 ) in this study .
namely , while the synthetic fitting method applied to a spectrum extending over some range ( e.g. , @xmath76 @xmath16 ) is generally useful for simultaneous determinations of several elements at a time , it is not suitable for such a case where _ extremely _ weak line features are under question .
that is , even a slight fluctuation of the continuum level can be a crucial source of error , because an appropriate adjustment of the local continuum is impossible in such a global spectrum synthesis .
accordingly , when one has to derive an abundance from an extremely weak line , the classical line - by - line @xmath20 analysis ( adopted in paper i ) should be preferable .
this is actually the reason why we adopted the classical @xmath20-based abundance determination for li and k , because the lines of these elements become very weak in early - a stars . ]
we hereafter adopt @xmath71(na ) as the representative sodium abundances which we will discuss in subsection 4.1 .
( 50mm,50mm)figure6.eps we derived the abundance of li from the @xmath20 of the merged li i doublet feature at @xmath77 @xmath16 .
since very weak lines are generally concerned , @xmath20 measurements were carried out by the gaussian fitting while comparing the spectrum on the computer screen with the theoretical spectrum synthesized with the known @xmath23 as well as appropriately varied @xmath3(li ) , by which we tried to minimize the possibility of erroneous measurement . in case
we could not detect any suitable feature at the expected position , we gave up determination of @xmath20 .
actually , we were unable to specify @xmath20(6708 ) for about @xmath78% of the targets ( 13 out of the 21 boao sample and 2 out of the 7 oao sample ) .
in such cases , only the upper limit values were derived by the formula @xmath79 ( cf . takeda & kawanomoto 2005 ) . here ,
fwhm was estimated from the root - sum - square of ( i ) the rotational broadening @xmath80 ( cf .
footnote 12 of takeda et al .
2008b ) , ( ii ) the approximate separation of the doublet @xmath81 , and ( iii ) the instrumental broadening @xmath82 ( @xmath83 : spectrum resolving power ) , as fwhm @xmath84 , and @xmath85 was assumed to be 2 according to our experience .
the actual measurements of @xmath20(6708 ) were carried out on the `` smoothed '' spectrum after applying the 9-pixel boxcar function to the original spectrum in order to reduce the effect of noise .
value ( 0.7 m@xmath16 ) of @xmath25 lyr at that time was actually measured in the `` original '' spectrum . in this study , however , the measurement was done on the `` smoothed '' spectrum , which is the reason for the appreciably larger value ( 1.2 m@xmath16 ) derived for this star in spite of the use of the same spectrum data of takeda et al .
this fact implies the difficulty in evaluating @xmath20 of extremely weak line feature , which is subject to considerable uncertainties .
] figure 7 displays how the gaussian fitting for measuring @xmath20 was done in the relevant li i 6708 region of the spectra .
( 80mm,240mm)figure7.eps then , based on such evaluated @xmath20 ( or @xmath86 ) and the model atmosphere along with the non - lte departure coefficients ( appropriately interpolated from the precomputed ones on a model grid ; cf .
takeda and kawanomoto 2005 for more details of the non - lte calculations ) , @xmath3(li ) ( non - lte abundance ) or its upper limit was determined for each star ( see table 2 for the adopted atomic parameters ) as given in table 1 . besides , in the similar manner as in subsection 3.1 , we also computed the corresponding non - lte corrections ( @xmath72 ) , and the abundance variations ( @xmath73 ) in response to perturbations of the atmospheric parameters .
these [ @xmath20 , @xmath72 , @xmath3(li ) , @xmath73 ] results are plotted against @xmath4 in figure 8 , which are also presented in electronic table e. ( 50mm,160mm)figure8.eps regarding the abundance of k , we exclusively invoked the k i line at 7698.97 @xmath16 , since the alternative line of the doublet at 7664.91 @xmath16 tends to be severely influenced by strong telluric lines .
its @xmath20 was measured on the spectrum ( in which telluric lines in the neighborhood had already been removed by dividing by that of a rapid rotator as in paper i ) in the same way as the case of the li i 6708 line ( cf . subsection 3.2 ) . if the line failed to be detected ( 4 stars among the 21 boao sample ) , we estimated the upper limit ( @xmath86 ) following equation ( 1 ) , where fwhm @xmath88 in this case .
figure 9 displays how these measurements were done in the relevant k i 7699 region of the spectra .
( 80mm,240mm)figure9.eps based on such evaluated @xmath20 along with the non - lte departure coefficients ( appropriately interpolated from the precomputed ones on a model grid ; cf .
takeda et al .
1996 for more details of the non - lte calculations ) , the non - lte k abundance ( @xmath3(k ) ) or its upper limit was determined for each star ( with the atomic parameter of the line in table 2 ) , as given in table 1 .
these results of @xmath20 and @xmath3(k ) , along with the corresponding non - lte corrections ( @xmath72 ) and the abundance variations ( @xmath73 ) in response to perturbations of the atmospheric parameters , are plotted against @xmath4 in figure 10 , and are also presented in electronic table e. ( 50mm,160mm)figure10.eps
we are now ready to answer the questions which motivated this study ( cf . section 1 ) : `` how do the abundances of na , li , and k behave in sharp - lined a - type stars ? how are they related with the am peculiarities frequently seen in slow rotators of this @xmath4 range ? '' in figure 11 are plotted the resulting [ @xmath3(na ) , @xmath3(li ) , and @xmath3(k ) ] against @xmath3(fe ) and @xmath23 ( determined in subsection 2.3 ) , both of which are the key quantities closely connected with the am phenomenon . based on this figure as well as figures 5c , 8c , and 10c ( showing the @xmath4-dependence ) ,
the abundance characteristics of these elements are discussed in the following subsections , where we also compare the observed trend with the theoretical prediction from the atomic diffusion theory by richer , michaud , and turcotte ( 2000 ) as well as talon , richard , and michaud ( 2006 ) .
( 80mm,200mm)figure11.eps regarding na , we notice in figure 11a a significant trend of remarkably tight correlation between @xmath3(na ) and @xmath3(fe ) .
actually , a slight @xmath23-dependence in @xmath3(na ) ( similar to that seen for @xmath3(fe ) in figure 3b ) is observed in figure 11a@xmath89 . since @xmath3(fe ) is known to be one of the am indicators in the sense that larger overabundance of fe is accompanied by stronger am stars , this observational fact means that na behaves in accordance with fe ( i.e. , the excess of na is a measure of am anomaly just like that of fe ) . while na abundances of am(+fm ) and normal a(+f ) stars ( though mostly restricted to those with @xmath13 k ) in the field as well as in open clusters have been reported by several recent studies based on na i 5683/5688 or 6154/6161 lines being less sensitive to non - lte effects ( e.g. , lane & lester 1987 ; varenne & monier 1999 ; gebran et al .
2008 , 2010 ; gebran & monier 2008 ; fossati et al .
2007 , 2008 ) , nobody seems to have explicitly mentioned this trend .
however , an inspection of the results of these papers suggests tendencies of ( i ) moderately supersolar [ na / h ] and ( ii ) [ na / h](am ) @xmath90 [ na / h](normal a ) .
therefore , the existence of this positive correlation between [ na / h ] and [ fe / h ] appears almost certain . from a theoretical point of view ,
however , such a conformance behavior between na and fe has never been predicted by any element segregation simulation designed for explaining am peculiarities such as that of richer et al .
( 2000 ) or talon et al .
( 2006 ) , which suggests that [ na / h ] remains almost solar or even slightly subsolar in spite of an appreciable overabundance of [ fe / h ] ( cf . figure 14 of richer et al . or figure 16 of talon et al . ) .
therefore , the current diffusion theory ( considered to be most promising for understanding the am phenomena ) is evidently imperfect at least in terms of na , which remains to be improved . here , the na fe correlation revealed in this study would serve as an important observational constraint . besides , it is interesting that the apparently metal - deficient stars with @xmath3(fe ) @xmath91 6.97.0 ( @xmath25 lyr and hd 218396=hr 8799 ) almost follow the same @xmath3(na)@xmath3(fe ) relation defined by am and normal stars with @xmath92 [ fe / h ] @xmath93 . since these two stars are known to be vega - like stars ( which may be related to so - called @xmath45 boo stars ; see , e.g. , kamp & paunzen 2002 , paunzen et al .
2002 ) , there might be some connection between am and @xmath45 boo anomalies .
( see also the last paragraph of subsection 2.3 . )
according to figures 11b and 11b@xmath89 , we do not see any clear @xmath3(fe)- as well as @xmath23-dependence in @xmath3(li ) which has values around or slightly lower than the solar - system abundance of @xmath94 .
this may suggest that am and normal a - type stars do not show any manifest difference in terms of their surface li abundances .
we recognize , however , a tendency of @xmath4-dependence for late a - type stars in the sense that @xmath3(li ) gradually decreases from @xmath94 ( @xmath95 k ) to @xmath96 ( @xmath97 k ) while such a systematic trend is not seen for early a - type stars ( @xmath14 k ) where @xmath3(li ) tends to be around the solar system abundance of @xmath94 ( figure 8c ) .
this @xmath4-dependent tendency of @xmath3(li ) in late - a stars is also observed in figure 2 of north et al .
( 2005 ) , and is more or less consistent with the theoretical prediction by richer et al .
( 2000 ; cf . figure 14 therein ) . it should be noted , however , that several stars apparently deviate from this general trend and show considerable li - deficiencies ( i.e. , only the upper limit was derived since the li line could not be detected ) . here , especially important
are the following 4 stars : hd 204188 ( 7622 k , @xmath98 ) , hd 141795 ( 8367 k , @xmath99 ) , @xmath21 peg ( 9453 k , @xmath100 ) , and @xmath25 cma ( 9938 k , @xmath101 ) .
what makes such an exceptional depletion of li ?
all these are classified as am stars ( @xmath21 peg is also an am star according to adelman 1988 , though the conventional type of a1 iv is given in table 1 ) , though am alone can not be the condition to trigger such a marked underabundance of li , because there are a number of other am stars that do show normal behavior of li .
actually , most of the present sample of low @xmath23 stars more or less show some degree of chemical peculiarities . in this connection ,
burkhart and coupry ( 1991a ) stated ( based on their study of late a - type stars ) that `` occurrence of li - deficient stars is not exceptional without any clearcut division between normal and am stars . '' yet , as far as early a - type stars are concerned , the fact that li tends to remain almost unaffected in near - normal stars such as @xmath26 gem ( believed to have almost the solar abundances ; e.g. , adelman & philip 1996 , lehmann et al .
2002 ) and @xmath15 dra ( which were previously considered as normal even though later found to have only weak am anomaly ; cf .
sadakane & okyudo 1990 ) might suggest the existence of some connection between the conspicuous li deficit of @xmath25 cma or @xmath21 peg and their am phenomenon . here
, we have to remark that the conclusion of paper i ( the photospheric li abundance of @xmath21 peg almost coincides with that of the solar - system composition ; cf . appendix a therein ) , which was based on the @xmath20(li 6708 ) value of 1.3 m@xmath16 taken from table 1 of coupry and burkhart ( 1992 ) , was not correct .
as can be seen from figure 7 , the li i 6708 line is not detectable even on the spectrum of s / n @xmath29 ; and we derived the upper limit of its equivalent width as @xmath20(6708 ) @xmath102 m@xmath16 ( cf . the electronic table e ) following equation ( 1 ) .
we suspect , therefore , that coupry and burkhart ( 1992 ) misidentified / mismeasured this line , since the quality of their spectrum with typical s / n of @xmath103 ( cf .
section 1 therein ) is evidently insufficient for measuring such an extremely weak line .
our revised conclusion for @xmath21 peg is @xmath3(li ) @xmath104 , which means that its photospheric li abundance is at least by @xmath105 dex deficient compared to the solar - system value . regarding @xmath3(k )
, we can observe ( from figure 10c ) a trend in terms of @xmath4 just similar to the case of @xmath3(li ) : @xmath3(k ) of late a - type stars appears to systematically decrease from @xmath8 ( @xmath9 k ) to @xmath10 ( @xmath11 k ) , while such a systematic trend is not seen for early a - type stars ( @xmath14 k ) where @xmath3(k ) is almost solar around @xmath8 ( an exception is @xmath21 peg which shows a clearly subsolar abundance of @xmath106 ) .
meanwhile , it also appears that @xmath3(k ) s of late a - type stars are weakly dependent upon @xmath3(fe ) as well as @xmath23 in the sense that k gets more deficient as the am anomaly becomes stronger ( cf .
the red circles in figures 11c and 11c@xmath89 ) .
accordingly , the deficiency of k ( particularly seen in late - a stars of @xmath13 k ) may be controlled by two factors ( @xmath4 and the degree of am peculiarity ) .
richer et al s ( 2000 ) simulation based on the theory of atomic diffusion ( along with radiative acceleration ) predicts a small overabundance of k at higher @xmath4 ( @xmath107 k ) region , whereas a marginal underabundance is expected at the lower @xmath4 side ( @xmath108 k ) ( cf .
figure 14 therein ) .
such a predicted trend of positive @xmath109(k)/@xmath110 gradient is in qualitative agreement with what we found from figure 10c . as mentioned in section 1 , k - abundance determinations of a - type stars
have so far been tried only by fossati et al .
( 2007 ) in their lte analysis of the k i 7699 line for late - a stars in the praesepe cluster .
according to their results for 5 stars , we see @xmath111[k / h]@xmath112 @xmath113 ( cf .
figure 10 and table 3 therein ) .
considering that the typical ( negative ) non - lte corrections amount to @xmath12 0.30.4 dex at the relevant @xmath4 range ( cf .
figure 10b ) , we may state that their results would turn out to be mildly subsolar when the non - lte effect is taken into consideration , and become more consistent with our conclusion .
motivated by the fact that our current understanding on the abundances of alkali elements in a - type stars is considerably insufficient , we conducted a comprehensive non - lte analysis to establish the photospheric abundances of na , li , and k for 24 selected sharp - lined a - type stars ( @xmath0 km s@xmath1 , 7000 k @xmath2 10000 k ) , many of which are am stars showing different degree of chemical peculiarity . for this purpose
, we primarily invoked the spectra of moderately high - dispersion ( @xmath24 ) and high s / n ratio ( typically a few hundreds ) obtained with boes at boao / korea , though spectra of much higher quality ( @xmath28 and s / n @xmath114 ) secured with hides at oao / japan were additionally used for 7 apparently bright stars ( mostly of early a - type ) .
we first carried out spectrum fitting analyses applied to the 61466163 @xmath16 region and derived @xmath18 , @xmath23 , @xmath3(o ) , and @xmath3(fe ) , from which the degree of am characteristics for each star could be confirmed .
the abundances of sodium , which were determined by the spectrum fitting method applied to the region comprising na i 5682/5688 lines , revealed a significant trend that @xmath3(na ) tightly scales with @xmath3(fe ) , which means that na becomes enriched similarly to fe in accordance with the degree of am phenomenon .
this result should be regarded as important because it seriously contradicts the prediction from the atomic diffusion theory . regarding lithium , @xmath3(li ) derived from the weak li i line at 6708 @xmath16 showed a tendency of @xmath4-dependence for late a - type stars in the sense that @xmath3(li ) gradually decreases from @xmath94 ( @xmath95 k ) to @xmath96 ( @xmath97 k ) , while @xmath3(li ) for early a - type stars ( @xmath14 k ) tends to be around the solar system abundance of @xmath94 .
however , several stars apparently deviating from this general trend and showing marked underabundances do exist , though what triggers such a li deficiency is not clear .
the abundances of potassium derived from the k i 7699 line revealed a tendency somewhat similar to the case of lithium : an apparent @xmath4-dependence ( and possibly with a weak anti - correlation with am peculiarity ) exists for late - a stars where @xmath3(k ) tend to be mildly subsolar systematically decreasing from @xmath8 ( @xmath9 k ) to @xmath10 ( @xmath11 k ) , whereas those for most early - a stars remain near - solar around @xmath12 5.05.2 . when these observational facts are compared with the theoretical prediction of richer et al .
( 2000 ) or talon et al .
( 2006 ) based on their atomic diffusion calculations , a serious discrepancy is found for na ( almost no anomaly or slightly subsolar tendency is predicted , while an enrichment of na just like fe is suggested from our analysis ) , though the simulated results appear to qualitatively reproduce the general trends observed for li and k. further improvement and development on the theoretical side would be desirably awaited to explain the results concluded in this study , which may serve as important constraints for any theory aiming to account for the origin of chemical anomalies in am stars .
this research has made use of the simbad database , operated by cds , strasbourg , france .
i. han acknowledges the financial support for this study by kicos through korea
ukraine joint research grant ( grant 07 - 179 ) .
lee acknowledges the astrophysical research center for the structure and evolution of the cosmos ( arsec , sejong university ) of the korea science and engineering foundation ( kosef ) through the science research center ( src ) program .
adelman , s. j. 1996 , mnras , 280 , 130 adelman , s. j. 1988 , mnras , 230 , 671 adelman , s. , & philip , a. g. d. 1996 , mnras , 282 , 1181 anders , e. , & grevesse , n. 1989 , geochim .
acta , 53 , 197 burkhart , c. , & coupry , m. f. 1991a , a&a , 249 , 205 burkhart , c. , & coupry , m. f. 1991b , mem .
soc . astron . it . ,
62 , 91 burkhart , c. , & coupry , m. f. 1995 , mem .
astron . it . , 66 , 357 burkhart , c. , & coupry , m. f. 2000 , a&a , 354 , 216 burkhart , c. , coupry , m. f. , faraggiana , r. , & gerbaldi , m. 2005 , a&a , 429 , 1043 caliskan , h. , & adelman , s. j. 1997 , mnras , 288 , 501 coupry , m. f. , & burkhart , c. 1992 , a&as , 95 , 41 esa 1997 , the hipparcos and tycho catalogues , esa sp-1200 , available from nasa - adc or cds in a machine - readable form ( file name : hip_main.dat ) flower , p. j. 1996 , apj , 469 , 355 fossati , l. , bagnulo , s. , landstreet , j. , wade , g. , kochukhov , o. , monier , r. , weiss , w. , & gebran , m. 2008 , a&a , 483 , 891 fossati , l. , bagnulo , s. , monier , r. , khan , s. a. , kochukhov , o. , landstreet , j. , wade , g. , & weiss , w. 2007 , a&a , 476 , 911 fossati , l. , ryabchikova , t. , bagnulo , s. , alecian , e. , grunhut , j. , kochukhov , o. , & wade , g. 2009 , a&a , 503 , 945 gebran , m. , & monier , r. 2007 , in convection in astrophysics , ed . f. kupka , i. w. roxburgh , & k. l. chan , proc .
239 ( cambridge : cambridge university press ) , 160 gebran , m. , & monier , r. 2008 , a&a , 483 , 567 gebran , m. , monier , r. , & richard , o. 2008 , a&a , 479 , 189 gebran , m. , vick , m. , monier , r. , & fossati , l. 2010 , a&a , 523 , 71 girardi , l. , bressan , a. , bertelli , g. , & chiosi , c. 2000 , a&as , 141 , 371 gray , d. f. 2005 , the observation and analysis of stellar photospheres , 3rd ed .
( cambridge : cambridge university press ) , ch.9 grevesse , n. , & noels , a. 1993 , in origin and evolution of the elements , ed . n. prantzos , e. vangioni - flam , & m. cass ( cambridge : cambridge university press ) , 15 hoffleit , d. 1982 , the bright star catalogue , 4th ed .
( new haven , connecticut : yale university observatory ) izumiura , h. 1999 , in proc .
4th east asian meeting on astronomy , observational astrophysics in asia and its future ed .
p. s. chen ( kunming : yunnan observatory ) , 77 kamp , i. , & paunzen , e. 2002 , mnras , 335 , l45 kurucz , r. l. 1993 , kurucz cd - rom , no .
13 ( harvard - smithsonian center for astrophysics ) kurucz , r. l. , & bell , b. 1995 , kurucz cd - rom , no
. 23 ( harvard - smithsonian center for astrophysics ) landstreet , j. d. , kupka , f. , ford , h. a. , officer , t. , sigut , t. a. a. , silaj , j. , strasser , s. , & townshend , a. 2009 , a&a , 503 , 973 lane , m. c. , & lester , j. b. 1987 , apjs , 65 , 137 lehmann , h. , andrievsky , s. m. , egorova , i. , hildebrandt , g. , korotin , s. a. , panov , k. p. , scholz , g. , & schnberner , d. 2002 , a&a , 383 , 558 monier , r. 2005 , a&a , 442 , 563 napiwotzki , r. , schnberner , d. , & wenske , v. 1993 , a&a , 268 , 653 north , p. , betrix , f. , & besson , c. 2005 , in element stratification in stars , 40 years of atomic diffusion , eds . g. alecian , o. richard , and s. vauclair , eas publication series , vol .
17 , p.333 ( eas ) paunzen , e. 2004 , in the a - star puzzle , proc .
224 , eds .
j. zverko , j. @xmath115i@xmath116@xmath117ovsk@xmath118 , s.j .
adelman , w.w .
weiss ( cambridge : cambridge university press ) , p.443 paunzen , e. , iliev , i. kh . ,
kamp , i. , & barzova , i. s. 2002 , mnras , 336 , 1030 richer , j. , michaud , g. , & turcotte , s. 2000 , apj , 529 , 338 sadakane , k. 2006 , pasj , 58 , 1023 sadakane , k. , & okyudo , m. 1990 , pasj , 42 , 317 smith , v. v. , lambert , d. l. , & nissen , p. e. 1998 , apj , 506 , 405 takeda , y. 1995 , pasj , 47 , 287 takeda , y. 2008 , mnras , 388 , 913 ( paper i ) takeda , y. , han , i. , kang , d .- i . , lee , b .-
, & kim , k .-
2008a , jkas , 41 , 83 takeda , y. , kambe , e. , sadakane , k. , & masuda , s. 2010 .
pasj , 62 , 1239 takeda , y. , kang , d .-
, han , i. , lee , b .-
, & kim , k .- m .
2009 , pasj , 61 , 1165 ( paper ii ) takeda , y. , kato , k. , watanabe , y. , & sadakane , k. 1996 , pasj , 48 , 511 takeda , y. , & kawanomoto , s. 2005 , pasj , 57 , 45 takeda , y. , kawanomoto , s. , & ohishi , n. 2007 , pasj , 59 , 245 takeda , y. , ohkubo , m. , & sadakane , k. 2002 , pasj , 54 , 451 takeda , y. , & sadakane , k. 1997 , pasj , 49 , 367 takeda , y. , sato , b. , & murata , d. 2008b , pasj , 60 , 781 takeda , y. , & takada - hidai , m. 1994 , pasj , 46 , 395 talon , s. , richard , o. , & michaud , g. 2006 , apj , 645 , 634 van leeuwen , f. 2007 , hipparcos , the new reduction of the raw data ( berlin : springer ) varenne , o. , & monier , r. 1999 , a&a , 351 , 247 rrccccrrrrccclhd # & hr # & name & sp.type & @xmath4 & @xmath31 & @xmath18 & @xmath23 & @xmath119 & @xmath120 & @xmath121 & @xmath122 & @xmath123 & remark + + 195725 & 7850 & @xmath124 cep & a7iii & 7816 & 3.74 & 4.45 & 52.0 & 7.60 & 8.47 & 3.27 & 6.35 & 4.81 & + 95418 & 4295 & @xmath35 uma & a1v & 9489 & 3.85 & @xmath1252.50 & 46.8 & 7.73 & 8.45 & ( @xmath1264.34 ) & 6.45 & ( @xmath1265.47 ) & + 27819 & 1380 & @xmath127tau & a7v & 8047 & 3.95 & 3.93 & 46.5 & 7.45 & 8.78 & 3.35 & 6.23 & 4.88 & h + 43378 & 2238 & 2 lyn & a2vs & 9210 & 4.09 & @xmath1252.10 & 45.4 & 7.34 & 8.70 & ( @xmath1264.07 ) & 6.13 & ( @xmath1265.18 ) & + 218396 & 8799 & @xmath128 & a5v & 7091 & 4.06 & @xmath1253.00 & 41.4 & 6.90 & 8.88 & 2.82 & 5.83 & 5.01 & vega - like star + 84107 & 3861 & 15 leo & a2iv & 8665 & 4.31 & 3.57 & 38.1 & 7.50 & 8.62 & ( @xmath1263.44 ) & 6.21 & 4.93 & + 33204 & 1670 & @xmath128 & a5 m & 7530 & 4.06 & 4.14 & 36.4 & 7.76 & 8.46 & 3.11 & 6.47 & 4.72 & h + 204188 & 8210 & @xmath128 & a8 m & 7622 & 4.21 & @xmath1253.90 & 35.3 & 7.50 & 8.74 & ( @xmath1262.77 ) & 6.24 & 4.75 & + 141795 & 5892 & @xmath54 ser & a2vm & 8367 & 4.24 & 4.04 & 34.4 & 7.72 & 8.17 & ( @xmath1263.16 ) & 6.46 & 5.02 & + 173648 & 7056 & @xmath129lyr & a4 m & 8004 & 3.90 & 4.73 & 32.4 & 7.74 & 8.31 & 3.21 & 6.47 & 4.82 & + 27628 & 1368 & 60 tau & a3 m & 7218 & 4.05 & 4.74 & 32.3 & 7.74 & 8.43 & 3.03 & 6.46 & 4.75 & h + 28546 & 1428 & 81 tau & a5 m & 7640 & 4.17 & 3.63 & 28.1 & 7.74 & 8.58 & 3.11 & 6.47 & 4.86 & h + 172167 & 7001 & @xmath25 lyr & a0va & 9435 & 3.99 & 1.85 & 21.9 & 6.96 & 8.63 & ( @xmath1263.84 ) & 6.00 & ( @xmath1265.38 ) & vega - like star + 60179 & 2891 & @xmath25 gem & a1v & 9122 & 3.88 & 2.33 & 19.7 & 7.52 & 8.42 & ( @xmath1263.71 ) & 6.15 & 5.10 & + 95608 & 4300 & 60 leo & a1 m & 8972 & 4.20 & 2.93 & 17.2 & 7.84 & 8.19 & ( @xmath1263.43 ) & 6.46 & 5.32 & + 48915 & 2491 & @xmath25 cma & a1vm & 9938 & 4.31 & @xmath1252.53 & 16.5 & 7.89 & 8.41 & ( @xmath1264.01 ) & 6.52 & ( @xmath1265.19 ) & + 27749 & 1376 & 63 tau & a1 m & 7448 & 4.21 & 3.90 & 12.7 & 7.97 & 8.25 & 3.09 & 6.56 & 4.50 & h + 33254 & 1672 & 16 ori & a2 m & 7747 & 4.14 & 3.07 & 12.2 & 7.99 & 8.18 & ( @xmath1262.53 ) & 6.60 & 4.66 & h + 72037 & 3354 & 2 uma & a2 m & 7918 & 4.16 & 2.48 & 11.6 & 7.89 & 7.98 & ( @xmath1262.55 ) & 6.53 & 4.66 & + 47105 & 2421 & @xmath26 gem & a0iv & 9115 & 3.49 & 1.81 & 10.9 & 7.55 & 8.76 & ( @xmath1263.50 ) & 6.35 & 4.90 & + 27962 & 1389 & 68 tau & a2iv & 8923 & 3.94 & 3.44 & 10.4 & 7.73 & 8.54 & ( @xmath1263.23 ) & 6.46 & 5.22 & h + + 182564 & 7371 & @xmath15 dra & a2iiis & 9125 & 3.80 & 3.43 & 27.3 & 7.83 & 8.48 & 3.47 & 6.65 & 4.98 & + 172167 & 7001 & @xmath25 lyr & a0va & 9435 & 3.99 & 1.90 & 21.9 & 6.97 & 8.63 & 3.21 & 6.05 & 5.13 & vega - like star + 48915 & 2491 & @xmath25 cma & a1vm & 9938 & 4.31 & 2.53 & 16.6 & 7.89 & 8.42 & ( @xmath1263.18 ) & 6.59 & 5.15 & + 47105 & 2421 & @xmath26 gem & a0iv & 9115 & 3.49 & 2.49 & 11.1 & 7.48 & 8.74 & 3.17 & 6.33 & 5.19 & + 189849 & 7653 & 15 vul & a4iii & 7870 & 3.62 & 5.16 & 10.7 & 7.34 & 8.65 & 3.03 & 6.31 & 4.60 & + 27962 & 1389 & 68 tau & a2iv & 8923 & 3.94 & 3.99 & 10.5 & 7.69 & 8.54 & 3.17 & 6.46 & 5.10 & h + 214994 & 8641 & @xmath21 peg & a1iv & 9453 & 3.64 & 3.07 & 6.0 & 7.63 & 8.52 & ( @xmath1262.94 ) & 6.46 & 4.74 & + note .
+ in columns 1 through 6 are given the hd number , hr number , star name ( with constellation ) , spectral type ( taken from hoffleit 1982 ) , effective temperature ( in k ) , and logarithmic surface gravity ( in cm s@xmath130 ) .
columns 7 through 10 show the results derived from 61466163 @xmath16 region fitting : the microturbulent velocity ( in km s@xmath1 ) , projected rotational velocity ( in km s@xmath1 ) , ( lte ) abundance of fe , and ( non - lte ) abundance of o. the non - lte abundances of li , na , and k finally resulting from this investigation are presented in columns 1113 .
all abundance results are expressed in the usual normalization of @xmath3(h ) = 12.00 . in each of the boao ans oao samples ,
the objects are arranged in the descending order of @xmath131 . in the remark in column 14 ,
hyades cluster stars are denoted with `` h '' .
+ @xmath125regarding these @xmath18 values , literature values were adopted unlike the others , because the @xmath18-determination from the 61466163 @xmath16 region fitting did not work successfully .
see subsection 2.3 for more details .
cclrldesig . & species & @xmath132 & @xmath133 ( ev ) & @xmath134 + li 6708 & li i & 6707.756 & 0.00 & @xmath135 + & li i & 6707.768 & 0.00 & @xmath136 + & li i & 6707.907 & 0.00 & @xmath137 + & li i & 6707.908 & 0.00 & @xmath138 + & li i & 6707.919 & 0.00 & @xmath139 + & li i & 6707.920 & 0.00 & @xmath137 + o 6156 & o i & 6155.961 & 10.74 & @xmath140 + & o i & 6155.971 & 10.74 & @xmath141 + & o i & 6155.989 & 10.74 & @xmath138 + o 6157 & o i & 6156.737 & 10.74 & @xmath142 + & o i & 6156.755 & 10.74 & @xmath137 + & o i & 6156.778 & 10.74 & @xmath143 + o 6158 & o i & 6158.149 & 10.74 & @xmath144 + & o i & 6158.172 & 10.74 & @xmath145 + & o i & 6158.187 & 10.74 & @xmath146 + na 5682 & na i & 5682.633 & 2.10 & @xmath147 + na 5688 & na i & @xmath148 & 2.10 & @xmath149 + na 6154 & na i & 6154.226 & 2.10 & @xmath150 + na 6160 & na i & 6160.747 & 2.10 & @xmath151 + si 5684 & si i & 5684.484 & 4.95 & @xmath152 + si 5688 & si ii & 5688.817 & 14.19 & + 0.40 + si 6155 & si i & 6155.134 & 5.62 & @xmath153 + s 7696 & s i & 7696.758 & 7.87 & @xmath154 + k 7699 & k i & 7698.974 & 0.00 & @xmath155 + ca 6162 & ca i & 6162.173 & 1.90 & + 0.10 + sc 5684 & sc ii & 5684.202 & 1.51 & @xmath141 + fe 5686 & fe i & 5686.524 & 4.55 & @xmath156 + fe 6147 & fe ii & 6147.741 & 3.89 & @xmath157 + fe 6149 & fe ii & 6149.258 & 3.89 & @xmath157 + fe 6150 & fe ii & 6150.098 & 11.45 & @xmath158 + fe 6157 & fe i & 6157.725 & 4.08 & @xmath151 + fe 6705 & fe i & 6705.101 & 4.61 & @xmath159 + fe 6708 & fe ii & 6708.885 & 10.91 & @xmath160 + ni 5682 & ni i & 5682.198 & 4.11 & @xmath161 + note .
+ all data are were taken from kurucz & bell s ( 1995 ) compilation , except for those of li i which were adopted from smith et al .
( 1998 ) as in takeda and kawanomoto ( 2005 ) .
+ @xmath125since this line comprises two close components of stronger @xmath162 ( at 5688.205 @xmath16 ) and weaker @xmath163 ( at 5688.194 @xmath16 ) , we adopted @xmath164 as the sum of these two . regarding the li i lines
, we considered only the lines of @xmath165li , since we neglected @xmath166li in our analysis . | the abundances of alkali elements ( li , na , and k ) were determined from the li i 6708 , na i 5682/5688 , and k i 7699 lines by taking into account the non - lte effect for 24 sharp - lined a - type stars ( @xmath0 km s@xmath1 , 7000 k @xmath2 10000 k , many showing am peculiarities to different degrees ) , based on high - dispersion and high - s / n spectral data secured at boao ( korea ) and oao ( japan ) .
we found a significant trend that @xmath3(na ) tightly scales with @xmath3(fe ) irrespective of @xmath4 , which means that na becomes enriched similarly to fe in accordance with the degree of am peculiarity .
regarding lithium , @xmath3(li ) mostly ranges between @xmath5 and @xmath6 ( i.e. , almost the same as or slightly less than the solar system abundance of 3.3 ) with a weak decreasing tendency with a lowering of @xmath4 at @xmath7 k , though several stars exceptionally show distinctly larger depletion .
the abundances of potassium also revealed an apparent @xmath4-dependence in the sense that @xmath3(k ) in late - a stars tends to be mildly subsolar ( possibly with a weak anti - correlation with @xmath3(fe ) ) systematically decreasing from @xmath8 ( @xmath9 k ) to @xmath10 ( @xmath11 k ) , while those for early - a stars remain near - solar around @xmath12 5.05.2 .
these observational facts may serve as important constraints for any theory aiming to explain chemical anomalies of a - type stars . |
let @xmath10 be an interval ( possibly infinite ) .
the sobolev space @xmath11 for integer @xmath9 is defined as the completion of the space of complex functions @xmath12 such that the norm @xmath13 is finite . by the sobolev embedding theorem ,
if @xmath14 , @xmath15 , then @xmath16 is absolutely continuous . similarly , one can define the homogeneous sobolev space @xmath17 as the space of functions of finite seminorm @xmath18 in fact , given any @xmath19 , the expression @xmath20 gives a banach - space norm on @xmath17 . if @xmath21 is a finite interval , then @xmath22 , and this banach norm is equivalent to @xmath23 . in the sequel ,
let @xmath24 stand either for @xmath11 or @xmath17 .
if @xmath25 is any set , we can define the trace space @xmath26 there is the following natural choice of a norm ( or seminorm ) in the trace space : @xmath27 throughout this note we will restrict ourselves to the case where @xmath28 is an increasing sequence : @xmath29 as is known @xcite , the following questions are of relevance in the study of the trace space .
[ problem1 ] given some @xmath30 , when does @xmath31 extend to a function @xmath32 satisfying @xmath33 ?
can one take this extension to depend linearly on the data @xmath31 ?
[ problem2 ] find an explicit simple formula for a norm ( or seminorm ) equivalent to @xmath34 .
observe that in our case , @xmath35 is a sequence space .
hence , we are searching for an intrinsic characterization of the sequences in @xmath35 .
the solution of problem [ problem2 ] can help us solve problem [ problem1 ] , because if we have a formula for an equivalent norm in the trace space , we can characterize the trace space as the space of sequences having finite norm .
this does not address the question of the linear dependence of the extension . to solve this , it is usual to consider the next problem .
[ problem3 ] does there exist a linear operator @xmath36 which is bounded and satisfies @xmath37 , for all @xmath38 ?
if so , can one give a simple formula for one such @xmath39 ?
an operator having these properties is usually called a bounded linear extension operator . in the case when @xmath40 is a sequence , the problem of finding an @xmath41 such that @xmath33 can be understood as an interpolation problem ( we say that such an @xmath41 interpolates the data @xmath31 ) .
a bounded linear extension operator provides a function @xmath42 interpolating the data @xmath31 and satisfying @xmath43 here the first inequality comes from the definition of the trace norm .
hence , a bounded linear operator gives an interpolating function @xmath42 having optimal norm up to a constant factor .
because of this property , we will refer to these operators as quasi - optimal interpolators . in applications
, it is important to find interpolators @xmath39 with small norm . the construction of a quasi - optimal interpolator given by a simple formula can also allow us to solve problem [ problem2 ] , as shows that if we put @xmath44 , then @xmath45 is an equivalent norm to @xmath34 . in this note
we solve problems [ problem1][problem3 ] for the sobolev spaces @xmath46 , @xmath47 for @xmath48 ; @xmath49 .
we use the standard notation @xmath50 for the divided differences : @xmath51 we will always assume that @xmath52.\ ] ] this is a general enough case . to see this , let @xmath53 and let @xmath54 be the right hand side of
. assume , for instance , that @xmath55 , @xmath56 and that we are dealing with the space @xmath47 ( the other cases are very similar ) .
observe that @xmath57 , so that @xmath58 clusters at @xmath59 .
assume that @xmath60 $ ] is such that @xmath61 .
using the fact that @xmath62 , @xmath63 , are continuous , and the mean value theorem for divided differences ( see ) , one sees that the values of @xmath64 are completely determined by @xmath31 .
hence , to find a quasi - optimal interpolating function in @xmath47 , we first find a quasi - optimal interpolating function @xmath60 $ ] . then we have to extend this @xmath41 to a quasi - optimal @xmath65 .
thus , we need to define @xmath66 , preserving the continuity of @xmath67 , @xmath68 .
since the values @xmath69 are fixed by @xmath31 , one can even construct @xmath70 beforehand .
moreover , finding a quasi - optimal @xmath71 with given @xmath72 is an easy problem .
define @xmath73 in this paper we show that @xmath74 note that the constants do not depend on @xmath75 .
let @xmath76 be the interpolation steps .
we say that the steps are uniformly bounded if there is a constant @xmath77 such that @xmath78 then we also show that if holds , then @xmath79 in fact , we prove a general result ( for any @xmath9 ) which relates quasi - optimal interpolation in @xmath80 and @xmath81 ( see theorem [ theorem - relation ] ) . using this theorem , will follow from .
we conjecture that and are also true for @xmath82 .
see section [ open - questions ] for a discussion of this and some other open questions . in the case
@xmath83 , the term @xmath84 in the expression for @xmath85 can be eliminated , thus giving a simpler expression .
we will show in proposition [ simpler ] that for @xmath83 , @xmath86 is equivalent to @xmath87 from the expression for @xmath85 , thus obtaining a expression similar to
. however , if the interval @xmath21 is large enough in comparison with @xmath77 , this can be done .
see section [ simplification ] for these results .
spline interpolators are a useful class of interpolators because they provide computationally simple formulas .
we will say that @xmath39 is a spline interpolator of degree @xmath88 if there is a decomposition of @xmath21 into closed intervals @xmath89 which intersect only at their endpoints , and such that for any @xmath31 , the restriction of @xmath42 to each interval @xmath90 is a polynomial of degree at most @xmath88 . in this paper
we construct quasi - optimal spline interpolators @xmath91 , @xmath92 ( for @xmath48 respectively ) given by simple expressions .
extension by smooth functions dates back to whitney @xcite . in 1934 , he solved problem [ problem1 ] for the space @xmath93 .
brudnyi and shvartsman extended whitney s results to the space @xmath94 in @xcite .
fefferman made more progress in @xcite , where he solved problem [ problem1 ] for the spaces @xmath95 and @xmath96 .
he also proved the existence of bounded linear extension operators .
recently , luli generalized these results to @xmath97 in @xcite .
the concept of the depth of an extension operator appears in some of these works . a linear extension operator @xmath98 is said to be of bounded depth if there is some @xmath99 such that @xmath100 and for each @xmath101 , @xmath102 is nonzero for at most @xmath103 different values of @xmath104 .
if @xmath39 has bounded depth , its depth is the smallest integer @xmath103 such that this holds .
it is easy to see that the spline interpolators constructed in this paper have bounded depth .
in fact , @xmath91 has depth @xmath105 and @xmath92 has depth @xmath106 .
fefferman and klartag have addressed the problem of computing efficiently the extension function and the norm in the trace space in @xcite .
see the expository paper @xcite .
however , their algorithm does not give explicit simple formulas . in @xcite
, luli constructs a bounded depth interpolator in @xmath6 by pasting interpolating polynomials using partitions of unity .
israel gives a result for @xmath107 in @xcite .
shvartsman has obtained in @xcite results for the non - homogeneous sobolev space @xmath108 , both in @xmath109 and in metric spaces . in a recent paper @xcite , fefferman , israel and luli
extend israel s result to @xmath110 .
they show that a linear extension operator can be constructed such that it has assisted bounded depth . in general , one can not hope to construct extension operators of uniformly bounded depth in @xmath111 , as they show in @xcite . in the recent preprint @xcite
, shvartsman gives a generalization of the whitney extension theorem for @xmath110 .
we remark that although our setting is less general than those mentioned above , we will obtain a very simple expression for the norm in the trace space and explicitly construct a simple quasi - optimal interpolator .
we believe that these results can be of interest in numerical analysis . in many numerical methods , such as the finite element methods and the galerkin methods , an approximation for the solution of an equation in an infinite dimensional function space
is searched in some finite dimensional subspace .
the linear spaces of quasi - optimal splines we construct ( corresponding to a finite number of interpolation nodes ) may be good candidates for these finite dimensional spaces .
the interpolator @xmath91 is just the piecewise linear interpolator .
this means that @xmath112 is an affine function in each interval @xmath113 $ ] .
together with the condition @xmath114 , this completely determines @xmath91 . the interpolator @xmath91 is given by the following formula : @xmath115 the interpolator @xmath92 is a spline interpolator of degree 3 .
it is defined as follows .
let @xmath116 given @xmath117 , we construct cubic polynomials on each of the intervals @xmath118 $ ] and @xmath119 $ ] such that the resulting cubic spline @xmath120 is @xmath121 and satisfies @xmath122
. put @xmath123 the conditions @xmath124 determine @xmath120 , since @xmath120 is a piecewise cubic polynomial on each of the intervals @xmath125 $ ] , @xmath126 $ ] .
observe that , restricted to the interval @xmath127 $ ] , @xmath120 only depends on @xmath128 , @xmath129 , @xmath130 .
.,width=453 ] see figure [ fig : phi2 ] for a typical graph of @xmath120 .
the interpolation nodes @xmath131 are shown as @xmath132 s and the auxiliary nodes @xmath133 are shown as @xmath134 s .
we also have the following explicit formula for @xmath120 .
define @xmath135 and observe that @xmath136 satisfies the boundary conditions @xmath137 then we have @xmath138 by using the identities @xmath139 and , it is easy to see that defines the same spline as .
in the sequel , we will use the notation @xmath140 , which means that the norms @xmath141 and @xmath142 are equivalent , i.e. , there are constants @xmath143 and @xmath144 such that , for all @xmath31 , @xmath145 the first result solves problems [ problem1][problem3 ] for @xmath47 , @xmath46 , @xmath48 .
[ main ] let @xmath48 , and @xmath15 .
define @xmath86 , @xmath146 from .
then the following statements are true : a. [ t1-a ] the seminorm @xmath147 gives an equivalent seminorm in @xmath17 : @xmath148 where the equivalence constants depend only on @xmath4 .
moreover , @xmath149 is quasi - optimal and @xmath150 . b. [ t1-b ] if holds , then the norm @xmath151 gives an equivalent norm in @xmath11 : @xmath152 the equivalence constants depend only on @xmath153 .
moreover , @xmath154 is quasi - optimal and @xmath155 .
if the assumption in this theorem is not fulfilled , we can always use the following trick .
first fix some @xmath156 .
let @xmath157 be the collection of all the intervals @xmath158 $ ] such that @xmath159 .
given a function @xmath31 on @xmath40 , proceeding as if were true , one obtains an interpolating function @xmath41 on @xmath21 .
however , @xmath160 is not comparable with @xmath161 .
now one chooses @xmath162 functions @xmath163 such that @xmath164 , @xmath165 , @xmath166 if @xmath167 and @xmath168 .
put @xmath169 .
then one can see that @xmath170 will be comparable with @xmath161 .
the second result relates quasi - optimal interpolation on @xmath80 and @xmath81 .
it will be used to deduce part ( [ t1-b ] ) of the theorem above from part ( [ t1-a ] ) .
it could also be of help in attacking the problem for @xmath82 .
[ theorem - relation ] let @xmath9 be arbitrary and @xmath15 .
suppose that holds .
the following statements are true : a. [ trr - a ] assume that the following equivalence of norms holds : @xmath171 then the following equivalence of norms also holds : @xmath172 the equivalence constants depend only on @xmath153 and the equivalence constants in .
b. [ trr - b ] if @xmath173 is a quasi - optimal interpolator , then @xmath174 is also a quasi - optimal interpolator .
the proofs of theorems [ main ] and [ theorem - relation ] will be given at the end of the next section .
we start by proving that @xmath175 this will follow from a few easy estimates and is valid for all @xmath9 . [ f - lower - bound ] let @xmath176 , and @xmath177 be a closed interval with @xmath178 .
let @xmath179 , @xmath180 .
then , for any @xmath181 , @xmath182 where @xmath143 is a constant @xmath183 . by the mean value theorem for divided differences ( see , for instance , ( * ? ? ?
* chapter 3.2 ) ) , there is some @xmath184 such that @xmath185 it is an easy consequence of the sobolev embedding theorem that there is some universal constant @xmath186 such that if @xmath187 $ ] and @xmath188 $ ] , then @xmath189 } \geq c_0 |g^{(k)}(\eta)|.\ ] ] let @xmath190 be the affine transformation taking @xmath191 $ ] to @xmath192 . if @xmath181 , put @xmath193 and compute @xmath194}^p \geq c_1^pc_0^p|g^{(k)}(\varphi^{-1}(\xi))|^p = c_1^pc_0^p|j|^{kp}|f^{(k)}(\xi)|^p \\ & \geq c_1^pc_0^p[(r-1)!]^{-p}|j|^{kp}|f(x_0,\ldots , x_{k+1})|^p \end{split}\ ] ] where @xmath195 .
the lemma holds with @xmath196 .
let us now recall the following result from @xcite .
[ df - lower - bound ] let @xmath9 , @xmath197 , @xmath198 $ ] , @xmath199 .
then @xmath200^{-p}\|f\|_{l_p^r(j)}^p.\ ] ] with these two lemmas , now we can prove the following proposition , which will give .
[ lower ] let @xmath9 .
put @xmath201 and define @xmath202 and @xmath85 from .
the following statements hold : a. [ norm - lower - homog ] if @xmath203 , then @xmath204 for some constant @xmath143 depending only on @xmath4 . b. [ norm - lower ] if holds and @xmath205 , then @xmath206 for some constant @xmath207 depending only on @xmath153 . if @xmath203 , we have @xmath208 } \geq
\frac{[(r-1)!]^p}{r}\sum_{n\in\mathbb{z}}(\lambda_{n+r } - \lambda_n)|f(\lambda_n,\ldots , \lambda _ { n+r } ) |^p,\ ] ] by lemma [ df - lower - bound ] .
this implies ( [ norm - lower - homog ] ) , with @xmath209 .
to prove ( [ norm - lower ] ) , fix @xmath210 and use lemma [ f - lower - bound ] to obtain @xmath211 } \geq \frac{c''^p}{r}\sum_{n\in\mathbb{z } } ( \lambda_{n+r } - \lambda_n)^{kp+1}|f(\lambda_n,\ldots,\lambda_{n+k})|^p.\ ] ] by summing these inequalities over @xmath212 together with ( [ norm - lower - homog ] ) , we obtain ( [ norm - lower ] ) .
our goal now is to prove theorem [ theorem - relation ] .
we will need the following lemma , which is related to friedrichs inequality .
[ easy - lemma ] let @xmath177 be an interval of finite length . if @xmath213 and @xmath214 , then @xmath215 , \ ] ] for some constant @xmath216 .
we have , for @xmath217 , @xmath218 so that @xmath219 repeated application of this inequality yields @xmath220 where the last inequality comes from hlder s inequality . using hlder s inequality for sums , we see that @xmath221.\ ] ] now the lemma follows by integrating this inequality .
now we can give the proof of theorem [ theorem - relation ] .
let @xmath222 and fix @xmath223 . by the mean value theorem for divided differences , we can choose @xmath224 $ ] such that @xmath225 by lemma [ easy - lemma ] , @xmath226 } & \leq c_0^p \left [ ( \lambda_{n+r } - \lambda_n)^{rp}\|f\|^p_{l^r_p[\lambda_n,\lambda_{n+r } ] } + \sum_{j=0}^{r-1 } ( \lambda_{n+r}-\lambda_n)^{jp+1}\frac{|f(\lambda_n,\ldots,\lambda_{n+j})|^p}{j ! } \right ] \\ & \leq c_1^p \left [ \|f\|^p_{l^r_p[\lambda_n,\lambda_{n+r } ] } + \sum_{j=0}^{r-1 } ( \lambda_{n+r}-\lambda_n)^{jp+1}|f(\lambda_n,\ldots,\lambda_{n+j})|^p \right ] , \end{split}\ ] ] where @xmath227 comes from lemma [ easy - lemma ] and @xmath228 ( see ) . by summing over @xmath229 , using @xmath33 , and by the definition of @xmath85
, we get @xmath230.\ ] ] now assume that holds .
given @xmath231 , choose @xmath232 with @xmath33 and @xmath233 . then , @xmath234 by proposition [ lower ] .
applying , we have @xmath235 hence , by definition of the trace norm , one obtains @xmath236 the reverse inequality comes from proposition [ lower ] , so that this proves ( [ trr - a ] ) . to prove ( [ trr - b ] ) , assume that @xmath237 is quasi - optimal and let @xmath238 be the norm of @xmath39 as an operator @xmath239 .
observe that @xmath240 .
also , if @xmath241 , @xmath42 is locally in @xmath81 , because @xmath242 .
hence , @xmath39 will make sense as an operator taking @xmath243 into @xmath46 if we can prove that @xmath244 is finite whenever @xmath245 .
if @xmath231 , we use and proposition [ lower ] to obtain @xmath246 here we have used @xmath247 , which is trivial .
this shows that @xmath248 is quasi - optimal .
now we can give the proof of theorem [ main ] .
observe that for any @xmath249 , the function @xmath250 is locally in @xmath80 , because it is of class @xmath251 and piecewise @xmath252 .
hence , to check that @xmath253 , it is enough to see that @xmath254 is finite for any @xmath241 .
therefore , we only need to prove that @xmath255 then we will have @xmath256 where the last inequality comes from proposition [ lower ] , so ( [ t1-a ] ) will follow . using theorem [ theorem - relation ] , ( [ t1-b ] ) follows from ( [ t1-a ] ) , because the fact that @xmath257 is contained in its proof
. inequality will follow from some easy computations using the formulas given in section [ definition - interp ] . the analogous estimate for @xmath46
can also be obtained directly by the same means instead of appealing to theorem [ theorem - relation ] .
we first deal with the case @xmath258 .
we compute @xmath259}$ ] using formula : @xmath260}^p = ( \lambda_{n+1}-\lambda_n)|f(\lambda_n,\lambda_{n+1})|^p.\ ] ] by summing over @xmath223 , we obtain @xmath261 for the case @xmath262 , we compute @xmath263}$ ] using formula .
we have , for @xmath264 $ ] , @xmath265 making the change of variables @xmath266 , we compute @xmath267 } = h_n\|q''\|^p_{l^p[0,1/2 ] } \leq 2^{p-1}\|q''\|^p_{l^1[0,1/2 ] } h_n \leq m^p h_n,\ ] ] where we have used jensen s inequality , and @xmath268}$ ] is some universal constant . hence , @xmath269 } = m^ph_n|f(\lambda_{n-1},\lambda_n,\lambda_{n+1})|^p.\ ] ] proceeding analogously , we obtain @xmath270 } = m^ph_{n-1}|f(\lambda_{n-1},\lambda_n,\lambda_{n+1})|^p.\ ] ] by summing inequalities and over @xmath271 and using @xmath272 , we find that @xmath273 this proves the theorem .
in this section , we examine whether the expression for @xmath85 given in can be simplified by eliminating the terms corresponding to @xmath275 . throughout the section
we will assume that holds .
the next proposition proves that the answer is affirmative for @xmath276 .
its proof reduces to doing some easy but somewhat lengthy estimates on the divided differences .
[ simpler ] let @xmath262 and assume that holds .
define @xmath85 and @xmath277 from and .
then we have the equivalence of norms @xmath278 the equivalence constants depend only on @xmath77 .
first , we have @xmath279 hence , @xmath280,\ ] ] by hlder s inequality for sums .
adding over @xmath223 , we obtain @xmath281 to prove the reverse inequality , we fix @xmath223 . since @xmath282 , we have either @xmath283 or @xmath284 .
assume that @xmath283 .
the other case is similar and easier . by the triangle inequality and the inequality @xmath285 by applying inequalities , ,
, one gets @xmath286 hence , we have @xmath287 , \end{split}\ ] ] where in the last inequality we have used and . in the case @xmath284 , one obtains @xmath288 .
\end{split}\ ] ] in fact , is not necessary in this case . for every index @xmath289
, we have either or . by summing these inequalities over @xmath289
, we obtain @xmath290 .
now we give an example that shows that , in general , for any @xmath82 , one can not eliminate all the terms @xmath275 from the expression of @xmath85 .
assume that @xmath291 and suppose that one wishes to prove the following equivalence of norms : @xmath292 for some coefficients @xmath293 not depending on @xmath31 , and with equivalence constants depending only on @xmath294 . taking @xmath295 we see that the coefficients @xmath293 must be in @xmath296 and satisfy @xmath297 .
then , the next example shows that can not hold in some cases .
let @xmath82 .
fix a small @xmath298 and let @xmath28 be a strictly increasing sequence such that @xmath299 , @xmath300 , @xmath301 as @xmath302 and @xmath303 as @xmath304 .
this sequence @xmath40 satisfies with @xmath305 . put @xmath306/[h(2+h ) ]
$ ] and consider the interpolation problem given by @xmath307 .
then @xmath308 solves this problem .
now observe that @xmath309 , @xmath310 and @xmath311 for all @xmath223 .
hence , the right hand side of is less or equal than @xmath312 .
it follows from and proposition [ lower ] that @xmath313 hence , the left hand side of is greater or equal than @xmath314 .
this is a contradiction for small enough @xmath315 .
however , the former example is in some sense pathological because it does not satisfy the property that one can choose @xmath316 points @xmath317 which are `` well separated '' .
it turns out that this kind of condition is the only thing one needs to be able to eliminate the terms @xmath275 in the expression for @xmath85 .
this always happens if the interval @xmath21 is large enough .
in fact , one can prove the following proposition .
let @xmath9 , @xmath15 and assume that @xmath318 _ _ ( _ _ see _ ) _ .
define @xmath277 from .
then theorem [ theorem - relation ] remains true if one replaces @xmath85 with @xmath277 .
we will only give a sketch of the proof .
first one can prove that if @xmath192 is an interval with @xmath319 , and @xmath320 , @xmath321 satisfy @xmath322 for @xmath323 , then latexmath:[\[\|f\|^p_{l^p(j ) } \leq c(r , k)^p\left [ \|f\|^p_{l^r_p(j ) } + \sum_{k=1}^{r+1 } now one covers @xmath21 with intervals @xmath90 of length @xmath325 in such a way that the interiors of any three of these intervals do not intersect .
one obtains the above inequality for @xmath326 . by averaging the above estimate among all possible choices of @xmath327 satisfying @xmath322 for @xmath328 and summing over all indices @xmath329
, one gets @xmath330 \leq c^p\left [ \|f\|^p_{l^r_p(j ) } + \|f\|^p_{\text{simp},w}\right],\ ] ] where @xmath331 and @xmath332 .
now one can repeat the arguments of the proof of ( [ trr - a ] ) in theorem [ theorem - relation ] , replacing @xmath85 by @xmath277 and using this last inequality instead of .
here , one must use the inequalities @xmath333 , which is trivial , and @xmath334 , which is proved in the same way as ( [ norm - lower ] ) in proposition [ lower ] .
then , it is easy to see that all the steps made after in the proof of ( [ trr - a ] ) in theorem [ theorem - relation ] remain true when we replace @xmath85 with @xmath277 .
the first open questions are whether one can generalize our results on @xmath335 to a higher number of derivatives . in particular , we ask the following .
[ quest-1 ] are and true for @xmath82 ?
theorem [ theorem - relation ] shows that if is true for some values of @xmath336 , then is also true for the same values of @xmath336 .
suppose that the answer to question [ quest-1 ] is affirmative .
can one give explicit quasi - optimal spline interpolators for @xmath82 ?
as before , theorem [ theorem - relation ] shows that it is enough to prove that some interpolator is quasi - optimal for @xmath80 .
another matter is whether one can give similar results in higher dimension , i.e. on @xmath337 , @xmath338 .
fefferman , israel and luli show in @xcite that it is impossible to obtain bounded depth interpolators for @xmath338 .
hence , one can not obtain simple interpolators like @xmath91 and @xmath92 for a general configuration of points . however , it might be possible to obtain nice formulas and explicit interpolators if one imposes some kind of regularity conditions on the configuration of points
. the conditions should be mild enough to allow the usual cases which appear in applications .
one possible condition is the minimum angle condition , which is widely used in numerical analysis .
it means that in the triangular mesh formed by the points , the interior angles of every triangle are uniformly bounded form below .
see @xcite for a review on this and other usual geometric conditions in numerical analysis . hence
, we ask the following question
. can one give simple formulas for the trace space norm and quasi - optimal explicit interpolators in @xmath339 and @xmath340 if one imposes some regularity conditions on the allowed configurations of points ?
the author would like to thank dmitry yakubovich for posing this problem and for all his helpful advice during the completion of this work . | let @xmath0 be a strictly increasing sequence . for @xmath1 , we give a simple explicit expression for an equivalent norm on the trace spaces @xmath2 , @xmath3 of the non - homogeneous and homogeneous sobolev spaces with @xmath4 derivatives @xmath5 , @xmath6 .
we also construct an interpolating spline of low degree having optimal norm up to a constant factor .
a general result relating interpolation in @xmath7 and @xmath8 for all @xmath9 is also given . |
the study of nuclei around double shell closures plays a key role in testing the predictive power of the shell model . in this context , of special interest are nuclei in the close vicinity to @xmath5pb , @xmath0sn and @xmath1sn , which may be assumed to be good closed cores .
experimentally , a large body of data is available in the @xmath5pb region , whereas it is a very hard task to obtain information on @xmath0sn and @xmath1sn neighbors . in recent years
, however , substantial progress has been made to access the regions of shell closures off stability , which has paved the way to spectroscopic studies of nuclei in the neighborhood of both @xmath0sn and @xmath1sn .
these data offer the opportunity for testing the basic ingredients of shell - model calculations , especially the matrix elements of the effective interaction , well away from the valley of stability .
this may support confidence in the predictions of calculations , which may , in turn , stimulate efforts to gain more experimental information on these nuclei . during the last few years ,
we have studied [ 1,2 ] several nuclei around @xmath0sn and @xmath1sn within the framework of the shell model employing realistic effective interactions derived from modern nucleon - nucleon ( @xmath6 ) potentials . in these studies
we have been mainly concerned with nuclei having few identical particles or holes .
the main aim of this paper is to report on some results of our current work in these regions , which we have obtained starting from the cd - bonn free @xmath6 potential @xcite . in particular
, we shall consider the four odd - odd nuclei @xmath0sb , @xmath2sb , @xmath3 in and @xmath4ag .
the motivation for this choice lies in the fact that these nuclei are a source of information on the effective neutron - proton interaction in the @xmath0sn and @xmath1sn regions .
more precisely , we focus attention on proton - neutron hole and neutron - proton hole multiplets in the former and latter region , respectively . to place this work in its proper perspectives ,
a brief historical digression is in order .
more than thirty years ago the study of particle - hole multiplets in the @xmath5pb region was the subject of great experimental and theoretical interest [ 4 - 7 ] .
in particular , the spectrum of @xmath5bi was extensively studied through pick - up and stripping reactions and several proton - neutron hole multiplets were identified [ 4,5 ] .
it turned out that a main feature of all multiplets with more than two members is that the states with minimum and maximum @xmath7 have the highest excitation energy , while the state with next to the highest @xmath7 is the lowest one .
on the theoretical side , a very good agreement with experiment was obtained in the shell - model study of ref .
@xcite , where particle - hole matrix elements deduced from the hamada - johnston @xmath6 potential @xcite were employed .
despite these early achievements , little work along these lines has been done ever since . by considering the new data which are becoming available in the @xmath0sn and @xmath1sn regions and the prospect of spectroscopic studies of unstable nuclei opened up by the development of radioactive ion beams
, we feel that the time is ripe to revive theoretical interest in this subject and perform shell - model calculations making use of a modern @xmath6 potential and improved many - body methods for deriving the effective interaction . in sec .
2 we give an outline of the theoretical framework in which our realistic shell - model calculations have been performed . in sec .
3 we present several calculated particle - hole multiplets and show that all the available experimental data are well reproduced by our calculations .
4 presents some conclusions of our study .
in our calculations for the sb isotopes we assume that @xmath0sn is a closed core and let the valence proton and neutron holes occupy the five single - particle levels @xmath8 , @xmath9 , @xmath10 , @xmath11 , and @xmath12 of the 50 - 82 shell .
similarly , for @xmath3 in and @xmath4ag we assume these orbits are occupied by the valence neutrons outside the @xmath1sn core while for the proton holes the model space includes the four orbits @xmath13 , @xmath14 , @xmath15 , and @xmath16 of the 28 - 50 shell . for the sb isotopes the single - proton and single neutron - hole energies
have been taken from the experimental spectra [ 9 - 11 ] of @xmath17sb and @xmath18sn , respectively .
the only exception is the proton @xmath19 which was taken from ref .
@xcite , since the corresponding single - particle level is still missing in @xmath17sb . for @xmath3 in and @xmath4ag the single - particle and single - hole energies can not be taken from experiment , since no spectroscopic data are yet available for @xmath20sn and @xmath21 in .
therefore , we have taken them from ref .
@xcite and ref .
@xcite , respectively , where they were determined by an analysis of the low - energy spectra of the sn isotopes with @xmath22 and of the @xmath23 isotones with @xmath24 .
as already mentioned in the introduction , in our shell - model calculations we have made use of a realistic effective interaction derived from the cd - bonn free nucleon - nucleon potential @xcite .
this high - quality @xmath6 potential , which is based upon meson exchange , fits very accurately ( @xmath25/datum @xmath26 ) the world @xmath6 data below 350 mev available in the year 2000 .
the shell - model effective interaction @xmath27 is defined , as usual , in the following way . in principle
, one should solve a nuclear many - body schrdinger equation of the form @xmath28 with @xmath29 , where @xmath30 denotes the kinetic energy .
this full - space many - body problem is reduced to a smaller model - space problem of the form @xmath31 here @xmath32 is the unperturbed hamiltonian , @xmath33 being an auxiliary potential introduced to define a convenient single - particle basis , and @xmath34 denotes the projection operator onto the chosen model space .
a main difficulty encountered in the derivation of @xmath27 from a modern @xmath6 potential , such as cd - bonn , is the existence of a strong repulsive core which prevents its direct use in nuclear structure calculations .
this difficulty is usually overcome by resorting to the well known brueckner @xmath35-matrix method . here ,
we have made use of a new approach @xcite which has proved to be an advantageous alternative [ 15,16 ] to the use of the above method .
the basic idea underlying this approach is to construct a low - momentum @xmath6 potential , @xmath36 , that preserves the physics of the original potential @xmath37 up to a certain cut - off momentum @xmath38 . in particular ,
the scattering phase shifts and deuteron binding energy calculated by @xmath37 are reproduced by @xmath36 .
the latter is a smooth potential that can be used directly as input for the calculation of shell - model effective interactions .
a detailed description of our derivation of @xmath36 can be found in ref .
@xcite , where a criterion for the choice of the cut - off parameter @xmath38 is also given .
we have used here the value @xmath39 @xmath40 .
once the @xmath36 is obtained , the calculation of the matrix elements of the particle - particle and hole - hole interaction is carried out within the framework of a folded - diagram method , as described , for instance , in refs .
[ 14,17 ] .
it should be pointed out that the effective interaction in the proton - neutron channel has been explicitly derived in the particle - hole formalism .
a description of the derivation of the particle - hole effective interaction is given in ref .
we report here some selected results of our study of nuclei in the regions of shell closures off stability .
the calculations have been performed using the oxbash shell - model code @xcite .
we start by considering the sb isotopes .
the most appropriate system to study the proton - neutron interaction is @xmath0sb which , having one proton valence particle and one neutron valence hole , plays in the @xmath0sn region the role played by @xmath5bi in the pb region .
experimental information on this nucleus is provided by the studies of refs .
[ 19 - 21 ] .
two @xmath41-decaying isomers with @xmath42 and @xmath43 are known , which originate from the @xmath44 @xmath45 and @xmath44 @xmath46 configurations , respectively .
while the relative energy of these two states has not been determined , there are indications that the @xmath43 state is located about 200 kev above the @xmath47 ground state @xcite . as regards the positive - parity states , besides the other three members of the @xmath44 @xmath45 multiplet , some states originating from the @xmath44 @xmath48 , @xmath49 @xmath45 , and @xmath44 @xmath50 configurations have been identified .
for the negative - parity states , very little experimental information is presently available . in a very recent study @xcite , several new transitions feeding the @xmath43 isomer have been revealed . in particular , a @xmath51 state at 1.025 mev
has been identified which may be attributed to the @xmath44 @xmath46 configuration .
this makes very interesting the study of @xmath2sb , for which five low - lying states with @xmath52 and @xmath51 have been identified [ 22,23 ] and interpreted as members of the same multiplet .
several calculated multiplets for @xmath0sb are reported in fig . 1 and compared with the existing experimental data
. note that in fig .
1(a ) all energies are relative to the @xmath47 state while in fig .
1(b ) they are relative to the @xmath43 state ( our calculations predict that the @xmath43 state lies 226 kev above the @xmath47 ground state ) . as regards the comparison between theory and experiment , we see that the calculated energies are in good agreement with the observed ones .
in fact , the discrepancies are all in the order of few tens of kev , the only exceptions being the @xmath53 and @xmath51 states of the @xmath49 @xmath45 and @xmath44 @xmath46 multiplets , which come about 300 kev above and 200 kev below their experimental counterparts , respectively . the calculated @xmath44 @xmath46 multiplet for @xmath2sb is reported and compared with the experimental data in fig .
we see that the agreement between experiment and theory is remarkably good , thus providing further support to our predictions for @xmath0sb .
it should be noted that both the experimental and calculated energies are relative to the @xmath43 state , which has been experimentally observed to be the ground state .
our calculations predict @xcite for the ground state @xmath54 , the @xmath43 state lying at 92 kev excitation energy . as regards the wave functions of the states reported in fig .
2 , we find that they are indeed dominated by the @xmath55 configuration @xmath44 @xmath46 while the remaining two neutron holes give rise to a zero - coupled pair occupying the @xmath56 , @xmath57 , and @xmath58 levels .
it is evident from figs . 1 and 2 that a main feature of the calculated multiplets is that the states with minimum and maximum @xmath7 have the highest excitation energy and are well separated from the other states , for which the splitting is relatively small .
this pattern is in agreement with the experimental one for the @xmath44 @xmath45 multiplet in @xmath0sb and the experimental data available for the other multiplets also go in the same direction .
it should be pointed out that this behavior is quite similar to that exhibited [ 6,7 ] by the multiplets in the heavier particle - hole nucleus @xmath5bi . also , it is worth noting that in all of our calculated multiplets the state of spin ( @xmath59 ) is the lowest , in agreement with the early predictions of the brennan - bernstein coupling rule @xcite .
we turn now to the @xmath1sn region . in this region
the counterpart of @xmath5bi and @xmath0sb is @xmath1 in , for which studies of excited states are at present out of reach .
we have therefore considered the two neighboring odd - odd isotopes @xmath3 in and @xmath4ag , focusing attention on the @xmath60 multiplet for which some experimental information is available . in figs . 3 and 4 we report the results of our calculations for @xmath3 in and @xmath4ag , respectively , and compare them with the experimental data @xcite .
we see that the agreement between experiment and theory is of the same quality as that obtained in the @xmath0sn region , the largest discrepancy being 130 kev for the @xmath61 state in @xmath4ag .
the pattern of the calculated multiplets turns out to be similar to that of the multiplets in the sb isotopes , but with the states with minimum and maximum @xmath7 less separate from the other ones .
actually , for @xmath3 in we find that the energy of the @xmath62 state is about the same as that of the @xmath63 state . in this connection
, it should be mentioned that for the former state the percentage of configurations other than those having a @xmath64 proton hole and a @xmath65 neutron is about 50% . a detailed discussion of the structure of the calculated states will be included in a forthcoming publication .
we have presented here some results of a shell - model study of nuclei in the close vicinity to doubly magic @xmath0sn and @xmath1sn focusing attention on the particle - hole multiplets , which are a direct source of information on the neutron - proton effective interaction . in our calculations
we have employed a realistic effective interaction derived from the cd - bonn @xmath6 potential .
this has been done using a new approach @xcite to shell - model effective interactions instead of the usual brueckner @xmath35-matrix method .
we have shown that our results are in very good agreement with the available experimental data .
it should be stressed that in our calculations no use has been made of any adjustable parameter .
in fact , for the single - neutron and single proton - hole energies in the @xmath1sn region , which are not available from experiment , we have adopted the values determined in our earlier studies [ 13,14 ] on sn isotopes and @xmath23 isotones . in summary
, we may conclude that the results of the present study of particle - hole multiplets in the @xmath0sn and @xmath1sn regions are consistent with those obtained in our previous studies of nuclei with identical valence particles and holes in the same regions @xcite .
this confirms that realistic shell - model calculations are able to describe with quantitative accuracy the spectroscopic properties of nuclei in regions of shell closures off stability , giving confidence in their predictive power . on the experimental side ,
it is of key importance to gain more information on nuclei in these regions .
this is certainly a main challenge of nuclear structure experiments with radioactive ion beams
. 99 a. covello , l. coraggio , a. gargano , and n. itaco , _ acta phys .
* b32 * , 871 ( 2001 ) , and references therein .
l. coraggio , a. covello , a. gargano , n. itaco , and t. t. s. kuo , _ phys .
* c66 * ( 2002 ) , and references therein .
r. machleidt , _ phys .
_ * c63 * , 024001 ( 2001 ) .
j. r. erskine , _ phys .
rev . _ * 135 * , b110 ( 1964 ) .
w. p. alford , j. p. schiffer , and j. j. schwartz , _ phys .
* 21 * , 156 ( 1968 ) .
t. t. s. kuo , _ nucl .
phys . _ * a122 * , 325 ( 1968 ) .
m. moinester , j. p. schiffer , and w. p. alford , _ phys .
rev . _ * 179 * , 984 ( 1969 ) .
t. hamada , i. d. johnston , _ nucl .
phys . _ * 34 * , 382 ( 1962 ) .
m. sanchez - vega , b. fogelberg , h. mach , r. b. e. taylor , a. lindroth , j. blomqvist , a. covello , and a. gargano , _ phys .
rev . _ * c60 * , 024303 ( 1999 ) . b. fogelberg and j. blomqvist , _ phys .
lett . _ * b137 * , 20 ( 1984 ) . b. fogelberg and j. blomqvist , _ nucl .
* a429 * , 205 ( 1984 ) .
f. andreozzi , l. coraggio , a. covello , a. gargano , t. t. s. kuo , and a. porrino , _ phys .
rev . _ * c56 * , r16 ( 1997 ) .
f. andreozzi , l. coraggio , a. covello , a. gargano , t. t. s. kuo , z. b. li , and a. porrino , _ phys .
rev . _ * c54 * , 1636 ( 1996 ) .
l. coraggio , a. covello , a. gargano , n. itaco , and t. t. s. kuo , _ j. phys . _ * g26 * , 1697 ( 2000 ) . s. bogner , t. t. s. kuo , l. coraggio , a. covello , and n. itaco , _ phys . rev . _
* c65 * , 051301 ( 2002 ) .
a. covello , l. coraggio , a. gargano , n. itaco , and t. t. s. kuo , in _ challenges of nuclear structure _
, proceedings of the seventh international spring seminar on nuclear physics , maiori , 2001 , ed .
a. covello , world scientific , singapore 2002 , p.139 .
a. covello , f. andreozzi , l. coraggio , a. gargano , t. t. s. kuo , and a. porrino , _ prog .
phys . _ * 38 * , 165 ( 1997 ) .
b. a. brown , a. etchegoyen , and w. d. m. rae , _ the computer code oxbash _
, msu - nscl , report no . 524 . c. a. stone , s. h. faller , and w. b. walters , _ phys .
* c39 * , 1963 ( 1989 ) . h. mach , d. jerrestam , b. fogelberg , m. hellstrm , j. p. omtvedt , k. i. erokhina , and v. i. isakov , _ phys .
* c51 * , 500 ( 1995 ) .
p. bhattacharyya _
et al _ , _ phys .
rev . _ * c64 * , 054312 ( 2001 ) .
w. b. walters and c. a. stone , in _ proceedings of international workshop on nuclear fission and fission product spectroscopy _ , seyssins , france , edited by h. faust and g. fioni , ill report n@xmath66
94fa05 t , 1994 , p.182 . j. genevey , j. a. pinston , c. foin , m. rejmund , h. faust , b. weiss , _ phys . rev . _
* c65 * , 034322 ( 2002 ) . m. h. brennan and a. m. bernstein , _ phys . rev .
_ * 120 * , 927 ( 1960 ) .
data extracted using the nndc on - line data service from the ensdf database , file revised as of october 18 , 2002 . | we report here on a shell - model study of nuclei close to doubly magic @xmath0sn and @xmath1sn focusing attention on particle - hole multiplets . in our study
we make use of realistic effective interactions derived from the cd - bonn nucleon - nucleon potential .
we present results for the four nuclei @xmath0sb , @xmath2sb , @xmath3 in and @xmath4ag .
comparison shows that the calculated results are in very good agreement with the experimental data available for these nuclei far from stability .
this supports confidence in the predictions of our calculations which may stimulate , and be helpful to , future experiments . |
the electronic structure of the high - temperature ( high - t@xmath14 ) superconductors continues to remain a challenging topic in the physics of strongly correlated transition metal oxides .
recent studies have elucidated important features involving electron - phonon coupling and charge - order @xcite , as well as two dimensionality of the magnetic excitations in the high - t@xmath14 cuprates @xcite .
the zhang - rice singlet ( zrs ) , is a basic starting point for understanding the electronic structure of the cuprates .
the zrs was originally postulated @xcite for a doped hole distributed over the oxygen sites in a cuo@xmath15 plaquette , which gets bound to the cu@xmath10 ( @xmath16 ) state and forms a singlet due to cu - o hybridization .
it is the lowest energy electron removal state .
since a photoemission process also removes an electron ( adds a hole ) , it was nicely shown that the zrs can be probed by valence band photoemission spectroscopy even for insulating copper oxides @xcite .
polarization dependent @xcite and spin - resolved @xcite valence band photoemission have established the zrs feature as the lowest energy state in insulating and superconducting copper oxides .
it was also predicted that cu @xmath0 core - level photoemission spectra should exhibit a low binding energy feature derived from the zrs , even for the parent insulating la@xmath8cuo@xmath3 @xcite .
while the theoretical prediction was made very early , only recent studies with hard x - ray photoemission spectroscopy ( hx - pes ) could identify the zrs feature in the cu @xmath0 core - level spectra of the parent la@xmath8cuo@xmath3 @xcite .
hx - pes has developed significantly in the last few years @xcite mainly due to its ability to overcome the high - surface sensitivity of conventional photoemission spectroscopy .
hx - pes results in much higher kinetic energies of emitted electrons and hence longer mean free paths ( typically 50 - 150 @xmath17 ) , naturally leading to bulk - sensitive measurements @xcite . its application
to strongly correlated systems has led to several significant results , making it an fundamental technique to reveal the true bulk electronic structure @xcite . in the last few years , we have reported bulk sensitive hx - pes with @xmath18=5.95 kev of transition metal ( tm ) @xmath0 core - level for various tm compounds ( pure and cr doped v@xmath4o@xmath19 @xcite , hole - doped lamno@xmath19 ( lsmo , lbmo ) @xcite and high - tc cuprates @xcite ) . unlike conventional soft x - ray ( sx ) pes , tm @xmath0 core - level hx - pes spectra have shown additional well - screened satellites with significantly large intensity in addition to @xmath20 and @xmath21 features , where @xmath22 represents a ligand hole .
these satellites were located at an energy position few ev lower in binding energy than the @xmath23 and @xmath24 main line , respectively . in the case of
pure and cr doped v@xmath4o@xmath19 , clear changes in v @xmath0 hx - pes were observed across the metal - insulator ( mott - hubbard ) transition @xcite .
the structure and position of the main peak and the high and low binding energy satellite structures were explained well by the cluster model including intra - atomic multiplet structure and a metallic screening channel derived from coherent states at the fermi energy .
the low binding energy satellite structure was attributed to the screening from the coherent state at the fermi level .
this has been independently confirmed by direct measurements of the valence band coherent feature and core - level spectra of v@xmath4o@xmath19 @xcite .
a similar interpretation based on a mott - hubbard transition has been used to explain the ru @xmath25 core - level spectra for a series of ru oxides @xcite . for hole - doped lamno@xmath19 ,
a similar well - screened low binding energy feature in mn @xmath0 spectra was observed @xcite .
the feature showed a noticeable increase with decreasing temperature on entering the ferromagnetic metal phase , indicating that the origin of this feature is strongly related to the ferromagnetic metal phase @xcite .
new features at low binding energy are also seen in high - t@xmath14 cuprates @xcite .
but its spectral shape is quite different in electron - doped nd@xmath26ce@xmath27cuo@xmath3 ( ncco ) and hole - doped la@xmath26sr@xmath27cuo@xmath3 ( lsco ) systems .
it was clearly shown that lsco and ncco exhibit new bulk character electronic states , which are strongly suppressed in the surface . a recent theoretical study @xcite offers an alternative interpretation for well - screened features in the cu @xmath0 , mn @xmath0 , and ru @xmath25 core - level spectra based on a non - local screening scenario with doping , magnetic and orbital ordering for the cuprates , manganites and ruthenates , respectively .
we believe in so far as the additional screening channel is required to explain the low binding energy feature , our picture is not different from ref . @xcite .
in particular , we believe the non - local screening channel in the multi - site cluster model @xcite would develop into quasiparticle states at the fermi level ( @xmath28 ) as obtained in dynamical mean - field theory calculations @xcite . in the spirit of the non - local screening channel or screening from quasiparticle states at @xmath28 , we have used a single - site cluster model @xcite , with the screening from the states at @xmath28 but also including intra - site multiplets which are necessary for the observed structure in the core - level spectra . in a previous study ,
we have presented results of cu @xmath0 hx - pes for bi@xmath4sr@xmath8cacu@xmath8o@xmath9 ( bi2212 ) @xcite .
we found that the spectral weight in the @xmath29 binding energy satellite increases considerably in hx - pes compared with sx - pes results .
this enhancement of the satellite intensity in bi2212 is quite different as opposed to the other tm compounds mentioned above .
v@xmath4o@xmath19 , lsmo and ncco have shown huge changes at the @xmath30 binding energy to the main peak , but bi2212 shows a change in the @xmath31 binding energy satellite .
this fact would naively indicate an increase of @xmath32 weight in the bulk since the satellite is generally assigned to the @xmath33 state .
the fitted parameter values indicated that hybridization strength was reduced in the bulk sensitive hx - pes as against the surface sensitive sx - pes .
as previously mentioned @xcite , a decrease of the hybridization strength in the bulk is surprising . in general , the different atomic environment and reduced co - ordination , often conspire to reduce hybridizations and screenings at the surface . in order to shed new light on the above - mentioned problem , we extend our previous work to study the cu - o layer - dependence , doping dependence and photon energy dependence by using hx - pes .
we report the cu @xmath0 core - level spectra in single layer la@xmath1sr@xmath2cuo@xmath3 ( lsco , @xmath34 = 0.0 , 0.145 and 0.26 ) , la doped bi@xmath4sr@xmath5la@xmath6cuo@xmath7 ( bi2201 ) and bilayer bi2212 .
our main purpose is to resolve the puzzling observation of increase in the spectral intensity of the @xmath32 satellite in bi2212 by comparative measurements . from x - ray photon
energy dependent measurements , the results show that the increased intensity in the satellite region is associated with bi @xmath13 core - level spectral intensity , which is usually neglected in sx - pes measurements .
the corrected @xmath12 satellite intensity is independent of the doping content or number of cu - o layers .
the results suggest a correlation of the relative intensities of the zrs feature and the hole - doping induced high binding energy spectral changes in the main peak with superconductivity .
the present paper is organized as follows .
the experimental details are briefly described in sec .
the results and discussion on cu - o layer dependence , doping dependence , and photon energy dependence are given in sec .
conclusions are drawn in sec .
hx - pes measurements were performed in a vacuum of 1 @xmath35 10@xmath36 torr at undulator beam line bl29xu , spring-8 @xcite using a scienta r4000 - 10kv electron analyzer .
the instrumentation details are described in ref .
the energy width of incident x - ray was 70 mev , and the total energy resolution , @xmath37e was set to @xmath38 0.4 ev . all spectra were corrected by subtracting a shirley - type background .
sample temperature was controlled to @xmath39k during measurements .
single crystal bi2212 and bi2201 were peeled with a scotch tape and measured at 30 k. the lsco samples were measured for surfaces fractured @xmath40-@xmath41 .
the @xmath34 = 0.0 composition was measured at room temperature to avoid charging while the metallic compositions ( @xmath34 = 0.145 and 0.26 ) were measured at 30 k. a reference measurement on a metallic bi - based non - cuprate oxides ba@xmath42k@xmath43bio@xmath19 ( bkbo ) was measured at 30 k for an @xmath40-@xmath41 fractured surface .
the fermi level of gold was measured to calibrate the energy scale .
( color online ) ( a ) cu @xmath0 core - level hx - pes spectra of bi2212 , bi2201 and bi @xmath13 hx - pes of bkbo ( filled circles ) with a photon energy @xmath44 kev .
the line is fit to the bi @xmath13 hx - pes data as explained in the text .
( b ) cu @xmath0 core - level hx - pes spectra of bi2212 and bi2201 with a photon energy @xmath45 kev .
all cu @xmath0 spectra were corrected by subtracting a shirley - type background . ]
in fig . 1(a ) we present the cu @xmath0 core - level hx - pes spectra of bi2212 and bi2201 with photon energy @xmath44 kev .
the spectrum of bi2212 is consistent with our previous report @xcite .
it consists of two prominent sets of structures and the main peaks at @xmath38 934 and 953 ev , separated by approximately 19 ev , originating from the spin - orbit splitting of the @xmath0 core - level , @xmath23 and @xmath24 .
each prominent structure itself consists of two peaks , main peak ( @xmath46 state ) and satellite ( @xmath33 state ) nearly 7 ev above two main peaks .
the spectral weight in the high binding energy satellite of @xmath23 at 940 ev is significantly higher in hx - pes as compared with the previous sx - pes @xcite and basically no change is seen over the @xmath24 satellite energy range . in the previous paper @xcite , we had carried out the cluster model calculation of cu @xmath0 hx - pes for bi2212 . to obtain the significant increase in the satellite intensity at 940 ev ,
a reduced hybridization strength was needed ( by a factor of @xmath47 ) compared to that in the sx - pes , implying a decrease of the hybridization strength in the bulk . since a smaller hybridization
is expected on the surface compared to that in the bulk , this fact is quite unusual . from a comparison between bi2212 and bi2201
, we found that the satellite intensity of bi2201 is higher than that of bi2212 , as shown in fig .
1(a ) . even more surprisingly , its relative weight is almost comparable to the main line .
this huge enhancement can not be explained by the reduced hybridization in the bulk .
thus , we are led to another possibility for elucidating our observations .
one possibility would be the photo - ionization cross section effect of bi @xmath13 core - level , because the binding energy of bi @xmath13 core - level is very close to the @xmath33 satellite energy of @xmath23 .
the bi @xmath13 contribution in cu @xmath0 core - level spectra of the bi - based cuprates is generally neglected .
this is a good approximation , because the small bi @xmath13 photo - ionization cross section leads to negligible intensity as far as the sx - pes is concerned .
in fact , the cross section ratio @xmath48 in soft x - ray region is at most @xmath49 @xcite .
the increase of the x - ray photon energy will give an increase of the cross section ratio @xmath48 .
the @xmath50 ratio is very large and is approximately 1 in hard x - ray region .
this means that bi @xmath13 contribution can by no means be neglected .
moreover , the composition ratio of bi : cu is 2:2 for bi2212 . thus , we expect comparable contributions from the bi @xmath13 core - level respect to the cu @xmath0 spectra and was overlooked by us in the previous study .
the considerable enhancement of the satellite intensity of bi2201 with respect to bi2212 can be similarly understood by comparing the composition ratio between bi2201 and bi2212 .
the composition ratio of bi : cu is 2:1 for bi2201 and consequently , the relative contribution from the bi @xmath13 is enhanced by a factor of 2 with respect to bi2212 . since , with further increase of the x - ray photon energy , the photo - ionization cross section ratio @xmath50 increases further , we expect the satellite intensity to be increased with higher value of @xmath18 compared to @xmath44 kev photon energy . to confirm this , we performed hx - pes measurements by using 7.93 kev x - ray photon energy . figure 1(b ) shows the cu @xmath0 core - level hx - pes spectra for bi2212 and bi2201 using photon energy @xmath45 kev .
again , the intensities at 940 ev for both bi2212 and bi2201 are clearly enhanced and they are much stronger than those with @xmath44 kev , as we expected . in particular , for bi2201 , the intensity in the satellite region exceeds that of the main line and the satellite peak dominates over the main peak , confirming the bi 4s contribution . to correct for the bi @xmath13 contribution in the cu @xmath0 hx - pes spectra of the b - based cuprates
, we measured the bi @xmath13 core - level data taken from a single crystal bkbo in fig .
1(a ) . since the measured signal - to - noise
( s / n ) ratio of bkbo spectrum is not sufficient for the present correction procedure , the bi @xmath13 spectrum of bkbo was fitted with a lorenztian ( hwhm=4 ev ) .
the result of the fit is illustrated by a line in fig .
this lorenztian line shape was used to fit the cu @xmath0 hx - pes spectra for bi2212 and bi2201 with @xmath44 kev .
then , we obtained the cu @xmath0 line shape by subtracting the best fit to the bi @xmath13 component , the only free parameter being the intensity ratio . figure 2(a ) shows the corrected cu @xmath23 hx - pes spectra of bi2212 and bi2201 compared with sx - pes data .
thus , satisfactory corrections were obtained by subtracting the bi @xmath13 signal .
the obtained ratio of the integrated intensities of the subtracted bi @xmath13 lorenztian contribution between bi2212 and bi2201 is 1:2 , which is consistent with the relative bi contribution between two materials .
the resulting cu @xmath0 hx - pes for bi2212 and bi2201 are quite similar to those of sx - pes for bi2212 .
( color online ) ( a ) comparison between experimental cu @xmath23 sx - pes from ref .
@xcite and cu @xmath23 hx - pes after subtracting bi @xmath13 contribution .
( b ) difference between cu @xmath0 hx - pes spectra of bi2201 and that of bi2212 .
the dip feature at 931 ev is the zrs feature .
the broad low intensity feature at higher binding energy is due to the poor s / n ratio in the bi2201 spectrum . ]
( color online ) ( a ) cu @xmath23 hx - pes spectra of la@xmath1sr@xmath51cuo@xmath15 for several doping with a photon energy @xmath44 kev .
( b ) difference spectra between several cu @xmath23 hx - pes spectra and reference spectrum of lco . ] having subtracted the bi @xmath13 contribution , we now concentrate on the detailed structure in the main peak of cu @xmath23 . the bi2212 spectrum [ fig .
2(a ) ] shows clear changes compared to bi2201 and the additional spectral weight is seen at the lower binding energy ( nearly 931 ev ) .
this is particularly evidenced well by the difference spectra between bi2201 and bi2212 in fig .
2(b ) . using the multi - site cluster model @xcite
, it was shown that the cu @xmath0 spectrum has a low binding energy zrs feature due to the non - local screening .
thus , this feature is assigned to the zrs feature , indicating that the zrs feature is enhanced in the bilayer bi2212 compared to the single layer bi2201 . in order to better understand the character of the main line , we resort to a comparison with the lsco cu @xmath0 spectra obtained with hx - pes .
cu @xmath23 hx - pes spectra of lsco as a function of doping are shown in fig .
the spectra of lsco with @xmath34 = 0.0 and 0.145 are the same as those of our previous report @xcite .
we observed a systematic change in spectral shape on hole - doping . to see these spectral features
clearly we plot the difference spectra with respect to the lco reference spectrum in fig .
the observed spectral changes in our data are consistent with previous works on the yba@xmath4cu@xmath19o@xmath52 @xcite .
the shoulder at lower energy to the main peak @xmath46 in lsco can again be similarly attributed to the zrs feature , consistent with earlier work @xcite . as discussed in our earlier study , the zrs peak in @xmath34
= 0.0 occurs at the lower binding energy to the @xmath46 state .
the spectral intensity in the zrs feature reduces on increasing @xmath34 from @xmath34 = 0.0 to 0.145 , but does not change for @xmath34 = 0.26 . note
that fig .
3(a ) shows another feature , @xmath53 , a broad feature centered at 934.5 ev ( feature @xmath54 ) which is enhanced as the hole - doping increases .
in contrast to the zrs feature , the feature @xmath54 grows with increasing @xmath34 up to @xmath34 = 0.26 .
we suggest two possible explanations for this broad feature @xmath54 . in lsco , since we dope holes due to sr content @xmath34 , it can lead to the cu@xmath55 character state which grows with doping .
another possible explanation is the non - local screening effects .
very recently , okada and kotani developed the multi - site cluster model calculation for the doping dependence @xcite and a similar enhancement as observed for the high binding energy feature was already apparent in their calculation .
however the relation of non - local screening effects with cu@xmath55 configurations is not clear .
we can not conclude from our experiments alone if the feature @xmath54 reflects the non - local screening effect or simply an increase in cu@xmath55 configurations with hole - doping , or a coupling of both effects .
a comparison of lsco spectra indicates that the zrs feature intensity between @xmath34 = 0.145 ( t@xmath14=36 k ) and 0.26 ( non superconducting ) does not change much but the intensity of the feature @xmath54 beyond optimal doping increases . in terms of its superconducting @xmath56
, it then suggests that an increase in the feature @xmath54 content is primarily responsible for the reduction in @xmath56 .
on the other hand , for bi2201 ( @xmath56=28 k ) and bi2212 ( @xmath56=90 k ) , the high binding energy feature does not seem to change while the zrs feature shows a small but definite increase in intensity for bi2212 .
this suggests that the retention or increase of the zrs feature is more important for higher tc in the hole doped cuprates .
in conclusion , the aim of the present paper has been to understand the origin of the experimentally observed satellite at 940 ev in cu @xmath0 hx - pes of bi based high - t@xmath14 superconductors and in particular their evolution with x - ray photon energy and the composition change .
we investigated the photon energy dependence of the cu @xmath0 core - level photoemission spectra for bi2212 and bi2201 by using hard x - ray photon energy .
we exploited the sensitivity of the relative photo - ionization cross section to clarify the bi @xmath13 contribution from the measured cu @xmath0 hx - pes spectra .
the results from bi based cuprates can be interpreted satisfactorily only by including the bi @xmath13 core - level contribution .
thus the observed enhancement of the satellite is then due to the increase of the photo - ionization cross section of bi @xmath13 core - level rather than the reduction of the hybridization strength in the bulk .
the corrected spectra lead to a systematic picture on the basis of the doping dependence and cu - o layers dependence of zrs feature and hole - doping induced high binding energy spectral changes in the main cu @xmath0 core - level line .
the hx - pes experiments reported here have benefited tremendously from the efforts of dr .
d. miwa of the coherent x - ray optics laboratory riken / spring-8 , japan and we dedicate this work to him .
a. lanzara , p. v. bogdanov , x.j .
zhou , s.a .
kellar , d.l .
feng , e.d .
lu , t. yoshida , h. eisaki , a. fujimori , k. kishio , j .-
shimoyama , t. noda , s. uchida , z. hussain , and z .- x .
shen , nature * 412 * , 510 ( 2001 ) .
m. taguchi , a. chainani , k. horiba , y. takata , m. yabashi , k. tamasaku , y. nishino , d. miwa , t. ishikawa , t. takeuchi , k. yamamoto , m. matsunami , s. shin , t. yokoya , e. ikenaga , k. kobayashi , t. mochiku , k. hirata , j. hori , k. ishii , f. nakamura , and t. suzuki , phys .
lett . * 95 * , 177002 ( 2005 ) .
k. kobayashi , m. yabashi , y. takata , t. tokushima , s. shin , k. tamasaku , d. miwa , t. ishikawa , h. nohira , t. hattori , y. sugita , o. nakatsuka , a. sakai , and s. zaima , appl .
. lett . * 83 * , 1005 ( 2003 ) .
y. takata , k. tamasaku , t. tokushima , d. miwa , s. shin , t. ishikawa , m. yabashi , k. kobayashi , j. j. kim , t. yao , t. yamamoto , m. arita , h. namatame , and m. taniguchi , appl .
. lett . * 84 * , 4310 ( 2004 ) .
a. chainani , t. yokoya , y. takata , k. tamasaku , m. taguchi , t. shimojima , n. kamakura , k. horiba , s. tsuda , s. shin , d. miwa , y. nishino , t. ishikawa , m. yabashi , k. kobayashi , h. namatame , m. taniguchi , k. takada , t. sasaki , h. sakurai , and e. takayama - muromachi , phys .
b * 69 * 180508(r ) ( 2004 ) .
k. horiba , m. taguchi , a. chainani , y. takata , e. ikenaga , d. miwa , y. nishino , k. tamasaku , m. awaji , a. takeuchi , m. yabashi , h. namatame , m. taniguchi , h. kumigashira , m. oshima , m. lippmaa , m. kawasaki , h. koinuma , k. kobayashi , t. ishikawa , and s. shin , phys .
lett . * 93 * , 236401 ( 2004 ) .
n. kamakura , m. taguchi , a. chainani , y. takata , k. horiba , k. yamamoto , e. ikenaga , h. namatame , m. taniguchi , a. awaji , a. takeuchi , k. tamasaku , y. nishino , d. miwa , t. ishikawa , k. kobayashi , and s. shin , europhys .
68 * , 557 ( 2004 ) .
m. taguchi , a. chainani , n. kamakura , k. horiba , y. takata , m. yabashi , k. tamasaku , y. nishino , d. miwa , t. ishikawa , s. shin , e. ikenaga , t. yokoya , k. kobayashi , t. mochiku , k. hirata , and k. motoya , phys .
b * 71 * , 155102 ( 2005 ) .
h. tanaka , y. takata , k. horiba , m. taguchi , a. chainani , s. shin , d. miwa , k. tamasaku , y. nishino , t. ishikawa , e. ikenaga , m. awaji , a. takeuchi , t. kawai , and k. kobayashi , phys .
b * 73 * , 094403 ( 2006 ) .
g. panaccione , m. altarelli , a. fondacaro , a. georges , s. huotari , p. lacovig , a. lichtenstein , p. metcalf , g. monaco , f. offi , l. paolasini , a. poteryaev , m. sacchi , and o. tjernberg phys .
. lett . * 97 * , 116401 ( 2006 ) .
y. takata , m. yabashi , k. tamasaku , y. nishino , d. miwa , t. ishikawa , e. ikenaga , k. horiba , s. shin , m. arita , k. shimada , h. namatame , m. taniguchi , h. nohira , t. hattori , s. sdergren , b. wannberg , and k. kobayashi , nucl .
a * 547 * , 50 ( 2005 ) . | we have performed a detailed study of cu @xmath0 core - level spectra in single layer la@xmath1sr@xmath2cuo@xmath3 , la doped bi@xmath4sr@xmath5la@xmath6cuo@xmath7 ( bi2201 ) and bilayer bi@xmath4sr@xmath8cacu@xmath8o@xmath9 ( bi2212 ) high - temperature superconductors by using hard x - ray photoemission ( hx - pes ) .
we identify the cu@xmath10 derived ( i ) the zhang - rice singlet ( zrs ) feature , ( ii ) the @xmath11 ( ligand screened ) feature , ( iii ) the @xmath12 satellite feature , as well as the hole - doping derived high binding energy feature in the main peak . in bi - based cuprates , intensities of the @xmath12 satellite features seem to be strongly enhanced compared to la@xmath1sr@xmath2cuo@xmath3 . from x - ray photon energy dependent measurements , it is shown that the increased intensity in the satellite region is associated with bi @xmath13 core - level spectral intensity .
the corrected @xmath12 satellite intensity is independent of the doping content or number of cu - o layers .
our results suggest a correlation of the relative intensity of zrs feature and hole - doping induced high binding energy spectral changes in the main peak with superconductivity . |
the sequence of binomial coefficients is one of the most well - studied , frequently - used , and generally significant sequences in all of mathematics .
it is interesting , therefore , that the analysis of repeated binomial coefficients ( coefficients which occur more often than the trivial two times which every number occurs ) has only received sustained attention in the past 50 years . clearly , many numbers occur three and four times ; these are what fill up the inside of pascal s triangle . however , the only other high multiplicities known to occur6 and 8are rare , and the patterns in which they appear are not well understood . however , some scattered progress has been made .
it has been shown by abbott _
[ 1 ] that the function @xmath7 which counts the number of ways of representing @xmath6 as a binomial coefficient has average order and normal order 2 .
all solutions to @xmath8 were found by avanesov [ 2 , referenced in 10 ] , and kiss [ 10 ] established the more general result of finiteness of solutions to @xmath9 , for @xmath10 a prime .
@xmath11 and @xmath12 have also been completely solved by de weger [ 17 ] .
more recently , bugeaud _ et al . _
[ 4 ] have found all solutions to @xmath13 using an improvement of the mordell - weil sieve , which is applicable to finding integral points on all hyperelliptic curves . perhaps the most striking result was found by lind [ 11 ] , who showed that if @xmath14 and @xmath15 ( where @xmath16 is the @xmath17-th fibonacci number ) , then @xmath18 .
david singmaster [ 16 ] also provided a proof of this , and noted that his result provides an infinite family of numbers with multiplicity at least 6 .
the first member of this family3003is also the only known number with multiplicity 8 .
singmaster [ 15 ] also made the following dramatic conjecture , the study of which has been an important feature of subsequent work on repeated binomial coefficients : if @xmath7 denotes the number of times @xmath6 occurs in pascal s triangle , @xmath19 .
there has been no direct attempt at proving the existence of such a finite upper bound on the number of ways @xmath6 may be represented as a binomial coefficient .
bounds on @xmath7 in terms of @xmath6 were obtained first by singmaster [ 15 ] , then by abbott _
[ 1 ] , and then by kane [ 8 ] .
currently the best unconditional bound is @xmath20 , obtained by kane [ 9 ] via an argument relating integer solutions of @xmath21 to derivatives of a function implicitly defined in terms of the @xmath22-function .
conditional on cramr s conjecture about small gaps between prime numbers , abbott _
[ 1 ] obtained @xmath23 .
the purpose here will be to provide information about generalizations of the equation solved by lind [ 11 ] and singmaster [ 16 ] : @xmath18 .
this is equivalent to @xmath24 .
however , to our knowledge , no studies of equations of the general form @xmath25 have been made .
we will present two independent results about the solutions to such equations ; one describes where solutions may occur , and the other asserts the finiteness of solutions ( in most cases ) . we will also provide a rationale for why considering such equations may provide a powerful framework for proving the singmaster conjecture itself .
first , we make the following basic proposition about the location of repeats . let @xmath26 , @xmath27 , @xmath2 , and @xmath3 be natural numbers such that @xmath28 ; let @xmath29 be the positive number defined by @xmath30 .
if @xmath1 , we have @xmath31 .
first , note the elementary fact that any entry in pascal s triangle may be written as the sum of the two entries above it : @xmath32 .
this process may be iterated to obtain a representation of @xmath33 as a sum of binomial coefficients of any row number less than @xmath34 .
for example , with two and three iterations , we obtain , respectively , @xmath35 @xmath36 that the coefficients appearing in the @xmath37-th such iterate correspond to the binomial coefficients of row @xmath37 follows from the observation that when generating the coefficient for the next term with @xmath38 , one adds together the coefficients of the current terms with @xmath38 and @xmath39 .
this is exactly the process which ordinarily generates the binomial coefficients in pascal s triangle .
we therefore may always write the equation @xmath1 as @xmath40 next , we note that if @xmath41 and @xmath34 and @xmath42 are large , the ratios of successive binomial coefficients @xmath43 , @xmath44 , and so on , are strictly decreasing , and are close to being constant .
specifically , @xmath45 .
suppose we call the ratios @xmath46 , @xmath47 , @xmath48
@xmath49 , @xmath50 , @xmath51 and so on
. then we may rewrite ( 1 ) as @xmath52 when @xmath26 and @xmath27 are very large in comparison to @xmath2 and @xmath3 , all the @xmath53 are approximately the same ( because of the expression for the ratio of successive binomial coefficients ) , and hence by the binomial theorem are all approximately the ( positive ) solution of @xmath54 .
\(2 ) would be true if @xmath55 .
however , because of the strict decrease mentioned above , we have @xmath56 .
suppose , then , that @xmath57 . then all the @xmath53 are , and the right side of ( 2 ) has experienced a proportional decrease from @xmath58 which is the product of all the proportional decreases in the individual @xmath53 .
however , the left side can not have experienced so great a decrease from @xmath59 , since no term has decreased proportionally more than the right side , and there is one term ( the constant , 1 ) which has not decreased at all .
thus the equation ( 2 ) can no longer be true .
we apply the same argument to find that @xmath60 can not be greater than @xmath29 .
writing out @xmath61 and @xmath62 yields the inequality in the proposition .
we must impose the condition @xmath28 , because in reformulating equation ( 1 ) as equation ( 2 ) , we have assumed that we may divide through by the leftmost term @xmath63 , which is nonzero iff @xmath28 . if @xmath64 , the equation @xmath1 has finitely many solutions in natural numbers @xmath26 , @xmath27 .
any such equation can be written as the equation of an algebraic curve @xmath65 @xmath66 in @xmath26 and @xmath27 .
for example , the equation @xmath67 , which singmaster [ 16 ] solved , corresponds to the curve @xmath68 .
this means that the proof of the theorem is reduced to the well - studied problem of determining whether an algebraic curve has an infinity of lattice points . for an individual curve , the standard approach to such
a problem is to determine that the curve is irreducible and has genus greater than 0 .
if so , then by siegel s theorem [ 7 , p. 353 ]
the set of lattice points on the curve is finite .
however , determining the genus alone requires an analysis of singularities [ 7 , p. 72 ] .
this , in turn , amounts to solving the system where the three partial derivatives of the homogeneous version of the curve are simultaneously equated to zero [ 5 , p. 19 ] ; a task which appears to be unattackable for general curves of this complexity . we therefore will not attempt to prove that siegel s theorem is directly applicable to these curves , but
instead will make use of the following criterion given by nagell ( originally due to maillet ) [ 12 , p. 264 ] . a unicursal [ genus 0 ] curve passes through an infinity of lattice points if and only if there exists a parametric representation of the form @xmath69 where @xmath34 is a natural number , and where @xmath70 , @xmath71 , and @xmath72 are integral polynomials in @xmath6 satisfying one of the following conditions : + 1 .
either @xmath73 with @xmath74 or @xmath75 ; @xmath70 and @xmath71 are both of degree @xmath34 ; + 2 .
@xmath76 is irreducible , and @xmath77 , @xmath78 ; @xmath70 and @xmath71 are both of degree @xmath79 ; the form @xmath80 can represent for integral values of @xmath81 and @xmath82 a certain integer @xmath83 such that @xmath84 divides all the coefficients of both @xmath70 and @xmath71 .
+ nagell goes on to state that shortly after maillet gave this criterion , siegel proved that it applies to all curves ( i.e. not just unicursal ones ) . to apply this criterion to our curves
, it will be necessary to consider the limiting behavior as @xmath85 .
we may assume the curves have points of arbitrarily large @xmath27 ; if they did not , they could not have arbitrarily large @xmath26 either ( since clearly @xmath86 ) and so could only pass through finitely many points with natural @xmath26 , @xmath27 , which are the only points which matter for the theorem about binomial coefficients .
qualitatively , it is clear that @xmath87 must be such that the highest total degree terms in the equation of @xmath65 almost cancel each other out as @xmath88 . formally if @xmath89 is the @xmath34-th term with total degree @xmath90 , @xmath91 because otherwise , for large @xmath27 , the value of @xmath92 would be the only @xmath93 term in @xmath65 .
there would be other terms @xmath94 , @xmath95 , and so on , but even a direct sum of these is not @xmath93 , and they clearly are not all summed together .
what is the sum of the highest total degree terms in ( 3 ) ?
the degree @xmath90 terms from the first product , @xmath96 , are simply the terms of @xmath97 . there is only one degree @xmath90 term in the second product ; it is @xmath98 . equation ( 4 ) then becomes @xmath99 @xmath100 therefore , if we take @xmath101 , we must have @xmath102 . now , consider the form of @xmath87 if there exists a parametric representation as described in nagell s [ 12 , p. 264
] criterion .
if @xmath75 , @xmath103 , and @xmath27 goes to @xmath104 as @xmath6 does .
this means that @xmath105 , which has a constant , rational value , because @xmath106 and @xmath107 are integral polynomials of the same degree in @xmath6 .
this can not be the case , because a simple application of the rational root test shows that @xmath108 is irrational . by nagell s criterion
, it must then be that @xmath26 and @xmath27 are given by rational functions of @xmath6 with numerator and denominator polynomials of the same degree .
then @xmath27 can only approach infinity when @xmath6 approaches one of the roots of the denominator , @xmath72 , i. e. when @xmath6 approaches either a rational or quadratic irrational number @xmath109 .
we thus have that @xmath110 .
@xmath111 is clearly the quotient of two quadratic irrationals ; because @xmath112 and @xmath113 have the same input @xmath109 , the number under the radical in both quadratic irrationals is the same .
this means that the quotient is itself a quadratic irrational , by rationalization of denominators .
we will now show that if @xmath114 , @xmath108 can not be a quadratic irrational , and hence no parametrization of the type described can exist . for convenience , we will work with the equation @xmath115 , instead of the original @xmath116 ; the former is shifted 1 unit to the left , and obviously has quadratic zeros iff the original does .
if @xmath34 and @xmath37 are such that @xmath117 and @xmath118 , the polynomial @xmath119 has no real roots of degree 2 . we will attack this by showing that it is impossible for @xmath120 to have a quadratic factor with positive discriminant .
we begin by noting that since @xmath121 is primitive , by gauss lemma [ 13 , p. 49 ] , it suffices to consider quadratic factors with integral coefficients .
since the first and last terms of @xmath121 have magnitude 1 , any such factor must be of the form @xmath122 or @xmath123 , with @xmath124 .
also note that by descartes sign test , @xmath121 has exactly 1 positive root . make the substitution @xmath125 , generating the new polynomial @xmath126 , which has been shifted 1 unit to the right .
substituting @xmath127 for @xmath26 in @xmath128 , we see that regardless of the parity of @xmath34 and @xmath37 , there are no sign changes .
thus , @xmath128 has no negative roots .
we conclude that @xmath121 has no negative roots smaller than @xmath129 .
we have @xmath130 , and @xmath131 , depending on whether @xmath34 is even or odd .
this means that any quadratic factor @xmath132 must take the values @xmath133 at @xmath134 and @xmath135 .
if @xmath136 , we have the following four cases : @xmath137 @xmath138 @xmath139 @xmath140 which yield , respectively , @xmath141 , @xmath142 , @xmath143 , and @xmath143 .
the values obtained by the same process for @xmath144 are , in order , @xmath145 , @xmath143 , @xmath141 , and @xmath141 .
the four cases where the first and last terms of @xmath132 have the same sign generate only two polynomials with distinct roots , as do the cases where they have opposite signs .
the complete list of quadratic factors of @xmath146 to be considered is thus @xmath147 the first has no real roots .
the second has a root @xmath148 , which is less than @xmath129 ; therefore it can not be a factor , by the sign test performed earlier .
neither can the last , because of the root @xmath149 .
the only possible quadratic factor is thus @xmath150 .
we observe that if @xmath151 , this _ is _ a factor , as can be seen from writing @xmath152 , adding one term to the other side , and taking roots .
it is then easily seen that no other polynomials @xmath146 can share this factor , for if they did , the difference @xmath153 must also share the factor , which it clearly does not .
this proves the lemma , and by extension the theorem .
the results obtained here describe instances where the numbers in a particular configuration " in the triangle are the same .
the most basic instance of this , the configuration " was shown by singmaster [ 16 ] and lind [ 11 ] to occur infinitely many times ; in fact , precisely when @xmath34 and @xmath42 are certain expressions given by fibonacci numbers .
we have shown that configurations such as and can occur only finitely many times , if at all .
but we have _ not _ shown , for example , that and occur finitely many times , because in those cases the difference in @xmath42-values is equal to the difference in @xmath34-values , and the associated polynomial @xmath154 has the quadratic irrational roots @xmath155 and @xmath156 . however , the assertion that solutions are finite is still nothing more than asserting that a certain subclass of the curves studied are irreducible and have fewer than the maximum allowable number of singularities ( @xmath157 [ 7 , p. 72
] , barring the possibility of non - ordinary singularities ) , something which seems very likely . we will now analyze one of the higher - degree analogues to the curve @xmath68 in order to illustrate the validity of this idea . as we will see
, the reason this is difficult in general is because of the necessity of computing a grbner basis to determine that the polynomial and its two partial derivatives share no common zeros . in the case of the next curve with possibly infinite lattice points ( the curve with @xmath158 , @xmath159 ; defined by @xmath160 )
, we may mechanically compute the grbner basis [ 13 , pp .
221 , 237 ] for the system @xmath161 [ 5 , p. 19 ] to see that there are no affine singularities .
if we then homogenize coordinates , and consider the system @xmath162 [ 5 , p. 19
] at @xmath163 , we see that the only solution must be @xmath164=[0 : 0 : 0]$ ] , which is not a valid point [ 7 , p. 12
this is because @xmath165 , @xmath166 , and @xmath167 are all homogeneous polynomials in @xmath26 and @xmath27 when @xmath163 . by the same argument we have used to determine @xmath87 , any homogeneous polynomial in two variables represents the union of some ( possibly complex ) lines through the origin .
none of these lines are the same , and thus the only solution to this system is @xmath168 .
we conclude that @xmath169 has no singularities .
we may therefore apply the genus - degree formula without subtraction of additional terms : @xmath170 [ 7 , p. 72 ] .
that @xmath169 is irreducible follows immediately from its being nonsingular , for by bzout s theorem [ 7 , p. 84 ] , any hypothetical components of @xmath169 must intersect somewhere in the complex projective plane , and thus create a singularity in @xmath169 . by siegel s theorem [ 7 , p. 353
] , therefore , the set of lattice points is finite . in the proposition
, we have shown that if a certain configuration occurs entirely within the triangle , the smooth function giving the ratio of one binomial coefficient to the preceding one must take a value @xmath29 ( 1 plus the root of the associated polynomial ) between the beginning " of the configuration and the end " .
this is essentially a precise way of stating that all the occurrences of a particular configuration have approximately the same ratio @xmath171 .
the restriction @xmath172 , which was algebraically necessary to avoid dividing by zero , corresponds to requiring that the configuration is not cut off " by the edge of the triangle
. all currently known nontrivial repetitions ( excluding singmaster s [ 16 ] ) occur so close to the side of the triangle that the proposition does not apply ; however , it is still satisfied , because the ratios on the edge are very large and change very rapidly .
it is easily seen that there can not be more than @xmath2 of the cutoff " cases , because for each @xmath27 , there is clearly at most one @xmath26 where @xmath173 . in the case of singmaster
s infinite family , the ratio @xmath155 is always less than @xmath174 , and greater than @xmath175 .
this can also be seen as a direct result of working out ratios of the given expressions involving fibonacci numbers ( @xmath14 and @xmath15 ) .
it works out that the ratios @xmath174 and @xmath175 are ratios of successive pairs of fibonacci numberssuccessive continued fraction convergents to @xmath155 . in this sense ,
the coefficient repetition occurs at all the
best possible " approximations to @xmath155 .
it is tempting to think that this is somehow necessary for repetitions to occur , and then to try and disprove the existence of _ any _ other repeats deep " in the triangle by proving that convergents to the other , non - quadratic ratios can not occur sequentially in this way .
this seems plausible because , even without invoking the more rapid continued fraction convergence properties of higher degree algebraic numbers , we have that the maximum difference between consecutive continued fraction convergents with first denominator @xmath176 is less than @xmath177 [ 6 , p. 152
] , which is very often less than the difference @xmath178 between consecutive coefficient ratios . however , there is no such obvious argument for the necessity " of continued fraction convergence .
the most obvious extension of our work would be to show that the curve defined by @xmath68 is the only one of this family of curves which passes through infinitely many lattice points , i. e. to extend the theorem to the case when @xmath179 and @xmath180 . to do that , an entirely different argument to the one used in this paper
would be necessary , since we have relied on the fact that the limiting ratio of @xmath26 to @xmath27 in most cases is not quadratic . if @xmath179 , it is quadratic , and there is no apparent way to prove that nagell s [ 12 , p. 264
] criterion can not be satisfied .
it is possible that the symmetry of the polynomial defining the curve when @xmath179 allows a simple algebraic manipulation of the system where it and its two partial derivatives are set equal to 0 , such that an inconsistency is derived . as we have seen from the previous consideration of @xmath181 , we may work with this system , instead of @xmath162 , because the absence of singular points at infinity follows easily for all these curves .
another idea would be to try and use the fact that @xmath68 is nonsingular in order to establish the non - singularity of the higher - degree curves by induction .
if we designate the first large product in the equation of one of our curves as @xmath182 , and the second as @xmath183 , we may write the curve as @xmath184 .
if @xmath184 is the equation of a curve with @xmath185 , then the equation of the curve with @xmath186 is @xmath187 ; in other words , when @xmath34 increases by 1 , the equation is multiplied " by a shifted version of @xmath188 .
if this kind of multiplication " of two nonsingular curves could be shown to preserve non - singularity , we would have an induction argument to extend the theorem . * * * the conclusions we have reached here are significant in their own right : they are , to our knowledge , the first fundamental results established concerning equations of the general form @xmath1 , where @xmath26 and @xmath27 vary .
this is fundamentally different than considering @xmath189 and allowing , say , @xmath42 and @xmath37 to vary ( an area in which some progress in bounding and tabulating solutions has already been made [ 2 , referenced in 10 ] [ 10 ] [ 17 ] [ 4 ] ) .
however , it is hardly debatable that the most ambitious and important goal in the examination of repeated coefficients is the proof of singmaster s [ 15 ] conjecture . so far
, the most pointed attacks on the conjecture have resulted only in an increasingly tight series of logarithmic boundsan approach which _ a priori _ seems unlikely to yield the desired @xmath190 result .
kane [ 8 ] has stated that his methodwhich initially yielded @xmath191 , and then was improved by a factor of @xmath192 [ 9]probably can not be further extended .
it may be more fruitful , therefore , to cease considering @xmath7 as a function to be bounded , and instead only to try analyze when particularly high multiplicities of @xmath6 occur .
we have not done this ; our concern has simply been with a certain type of nontrivial repetition .
however , a large part of the value of our exploration lies in the fact that the algebraic curves we have used would seem to provide a good basis for pursuing this .
notice , for instance , that a coefficient occurring six times simply corresponds to an integral intersection between two of our curves beyond a certain @xmath26-value ( the degree of the higher degree curve ) .
a multiplicity of eight corresponds to three curves intersecting at the same point , and so on .
furthermore , any set of the curves has at least some easily calculable number of these common intersections , because of the trivial integral points near the origin which they all share .
these correspond to repetitions in the negative triangle .
each large - multiplicity integral intersection between these curves also corresponds to a large integral point on a curve of much lower degree ; specifically , if @xmath193 curves with highest degree @xmath34 intersect at @xmath194 , there is an integral point on a curve of degree at most @xmath195 with greater @xmath26 and @xmath27 coordinates than @xmath196 .
the singmaster [ 15 ] conjecture would be proved by bounding the number of these curves which can share a common intersection beyond a given @xmath26-value ( although this statement is stronger than is necessary ; the conjecture only considers integral intersections ) .
the nave way to do this would be to take a general set of some number of these curves , shift them left sufficiently far , and try to show via nullstellensatz manipulations [ 13 , p. 22 ] ( generating other polynomials in the same ideal ) that there could not be a common intersection in the first quadrant .
the difficulty , of course , lies in working generally with curves of arbitrary complexity
. it should be noted , however , that because of the fact that the nullstellensatz deals with all intersections , not just integral ones , this strategy is not equivalent to simply manipulating general binomial coefficients themselves . even if the task is still seemingly difficult , we are able to utilize a more powerful tool . _ figs . 1 - 3
. several of the curves we have considered .
the first nontrivial intersection occurs between @xmath197 , @xmath141 , and @xmath198 , @xmath159 .
it corresponds to @xmath199 . _
singmaster s curve : @xmath200 , @xmath141 + + _
+ + _ fig .
@xmath200 , @xmath159 _
+ + _ fig .
@xmath201 , @xmath145 _ advanced tools of algebraic geometry are also possibly applicable to this reformulation of the conjecture , although a major strengthening of current knowledge would certainly be necessary first .
if a general effective form of siegel s theorem [ 7 , p. 353 ] were known , it would be possible to bound the height " of integral points on these curves ( their coordinate size , essentially ) . however , the currently known effective methods for genus 1 curves , such as baker s [ 3 , p. 45
] method , generate bounds too large ( triple exponential ) to be useful , even if they were generalized .
more desirable ( and more difficult ) would be an effective schmidt subspace theorem , as this would result in an effective form of a corollary [ 14 , p. 5
] on simultaneous approximation of algebraic numbers : let @xmath202 , @xmath203 be algebraic numbers such that @xmath204 , @xmath202 ,
@xmath203 are linearly independent over the rationals .
then for any @xmath205 there are only finitely many integers @xmath206 ,
@xmath207 , @xmath176 with @xmath208 such that @xmath209 + if we could find the ratios @xmath210 where the various algebraic numbers @xmath29 associated with a set of our curves are simultaneously approximated , we could find the intersection point .
unfortunately , we have not yet provided a requirement that the approximations to @xmath29 be as close as is dictated in the corollary .
the singmaster conjecture remains as paul erds once described it [ cited in 15 , p. 385
] : a very hard " problem .
the intent here has been only to introduce a novel form for viewing repeated binomial coefficient problems to which the well - developed tools of 20th century mathematics are at least somewhat applicable . whether this method can yield a truly new understanding of such an antique , basic
, elementary part of mathematics , remains to be seen .
[ 1 ] abbott , h. l. ; erds , p. ; hanson , d. _ on the number of times an integer occurs as a binomial coefficient_. amer .
monthly * 81 * ( 1974 ) : 256 - 261 .
mr0335283 ( 49 # 65 ) + [ 2 ] avanesov , .
t. _ solution of a problem on figurate numbers_. ( russian ) acta arithm .
* 12 * ( 1966/1967 ) : 409 - 420 . mr0215784 ( 35 # 6619 ) + [ 3 ] baker , a. _ transcendental number theory_. cambridge university press , london - new york , 1975. mr0422171 ( 54 # 10163 ) + [ 4 ] bugeaud , y. ; mignotte , m. ; siksek , s. ; stoll , m. ; tengely , sz . _ integral points on hyperelliptic curves_. algebra & number theory * 2 * ( 2008 ) , no . 8 : 859 - 885. arxiv:0801.4459 + [ 5 ] coolidge , j. l. _ a treatise on algebraic plane curves_. dover publications , new york , 1959 [ 1931 ] .
mr0120551 ( 22 # 11302 ) + [ 6 ] hardy , g. h. ; wright , e. m. _ an introduction to the theory of numbers_. 3rd .
edn . , oxford university press , new york , 1954 [ 1938 ] .
mr0067125 + [ 7 ] hindry , m. ; silverman , j. h. _ diophantine geometry : an introduction_. graduate texts in mathematics , * 201 * , springer , new york , 2000. mr1745599 ( 2001e:11058 ) + [ 8 ] kane , d. m. _ new bounds on the number of representations of t as a binomial coefficient_. integers : electronic j. of combinatorial number theory * 4 * ( 2004 ) , # a07 : 1 - 10 .
mr2056013 + [ 9 ] kane , d. m. _ improved bounds on the number of ways of expressing t as a binomial coefficient_. integers : electronic j. of combinatorial number theory * 7 * ( 2007 ) , # a53 : 1 - 7. mr2373115 ( 2008m:05014 ) + [ 10 ] kiss , p. _ on the number of solutions of the diophantine equation @xmath211binom(x , p ) = @xmath211binom(y , 2)_. fib . quart . *
26 * ( 1988 ) , no . 2 : 127 - 129 .
mr0938585 (89f:11050 ) + [ 11 ] lind , d. a. _ the quadratic field q(@xmath212 ) and a certain diophantine equation_. fib . quart .
* 6 * ( 1968 ) , no .
3 : 86 - 93. mr0231784 (38 # 112 ) + [ 12 ] nagell , t. _ introduction to number theory_. 2nd edn .
, chelsea publishing company , new york , 1964 [ 1951 ] .
+ [ 13 ] prasolov , v. v. _ polynomials _ , trans .
d. leites .
algorithms and computation in mathematics , * 11 * , springer , berlin , 2004 .
mr2082772 ( 2005f:12001 ) + [ 14 ] schlickewei , h. p. _ the mathematical work of wolfgang schmidt_. in schlickewei , h. p. ; tichy , r. f. ; schmidt , k. d. , eds .
_ diophantine approximation : festschrift for wolfgang schmidt_. springer , vienna - new york , 2000 : 1 - 14 .
+ [ 15 ] singmaster , d. _ how often does an integer occur as a binomial coefficient ? _
monthly * 78 * ( 1971 ) , no . 4 : 385 - 386. mr1536288 + [ 16 ] singmaster , d. _ repeated binomial coefficients and fibonacci numbers_. fib
* 13 * ( 1975 ) , no . 4 : 295 - 298 .
mr0412095 ( 54 # 224 ) + [ 17 ] de weger , b. m. m. _ equal binomial coefficients : some elementary considerations_. j. number theory * 63 * ( 1997 ) , no . 2 : 373 - 386. mr1443768 ( 98b:11027 ) | we consider the problem of characterizing solutions in @xmath0 to the equation @xmath1 in terms of @xmath2 and @xmath3 .
we obtain one simple result which allows the determination of a ratio in terms of @xmath2 and @xmath3 which the ratio @xmath4 must approximate .
we then add to the understanding of the infinite family of repeated coefficients discovered by d. singmaster , by using fundamental results from diophantine geometry to prove that in the case @xmath5 , solutions to @xmath1 are finite .
finally , we make some observations about the potential utility of equations of the form @xmath1 in proving singmaster s conjecture , which is the main unsolved problem in the area of repeated binomial coefficient study .
we remark that this approach to the conjecture is markedly different from previous approaches , which have only established logarithmic bounds on a function which counts the number of representations of @xmath6 as a binomial coefficient . |
locally flatly embedded closed 2-manifolds in the 4-space @xmath2 are called _ surface links_. it is known that any oriented surface link can be deformed to the closure of a simple surface braid , that is , a simple branched covering of the 2-sphere ( @xcite ) . as surface knots of genus one which can be made from classical knots , there are _ spun @xmath0-knots , turned spun @xmath0-knots , symmetry - spun tori _ and _ torus @xmath0-knots_. consider @xmath2 as obtained by rotating @xmath3 around the boundary @xmath4 . then a _ spun @xmath0-knot _ is obtained by rotating a classical knot ( @xcite ) , a _ turned spun @xmath0-knot _ by turning it once while rotating ( @xcite ) , a _ symmetry - spun torus _ by turning a classical knot with periodicity rationally while rotating ( @xcite ) , and a _
torus @xmath0-knot _ is a surface knot on the boundary of a neighborhood of a solid torus in @xmath2 ( @xcite ) .
symmetry - spun tori include spun @xmath0-knots , turned spun @xmath0-knots and torus @xmath0-knots .
we call the link version of a symmetry - spun torus , a spun @xmath0-knot , and a turned spun @xmath0-knot a _ symmetry - spun @xmath0-link _ , a _ spun @xmath0-link _ and a _ turned spun @xmath0-link _ respectively .
now we consider surface links in the 4-space which can be deformed to simple branched coverings of a trivial torus knot , which we define as _ torus - covering - links _
( @xcite ) . by definition ,
a torus - covering - link is described by a _
torus - covering - chart _ , which is a chart on the trivial torus knot .
torus - covering - links include symmetry @xmath0-links ( and spun @xmath0-links , turned spun @xmath0-links , and torus @xmath0-links ) .
a torus - covering - link has no 2-knot component .
each component of a torus - covering - link is of genus at least one . in this paper
we deform the torus - covering - link associated with a torus - covering - chart of degree @xmath5 to the closure of a simple surface braid
. then we obtain its ( surface link ) chart description , which is of degree @xmath6 , and gives an upper estimate of its braid index ( theorem [ thm3 - 1 ] ) .
in particular , we show that the turned spun @xmath0-knot of the torus @xmath1-knot has the braid index four ( corollary [ cor3 - 3 ] ) .
the author would like to thank professors takashi tsuboi and elmar vogt for suggesting this problem , and professor akio kawauchi for advising about titles .
a locally flatly embedded closed 2-manifold in @xmath2 is called a _ surface link_. a surface link with one component is called a _ surface knot_. a surface link whose each component is of genus zero ( resp .
one ) is called a _ @xmath7-link _ ( resp . _ @xmath0-link _ ) . in particular
a surface knot of genus zero ( resp .
one ) is called a _ @xmath7-knot _ ( resp .
_ @xmath0-knot _ ) .
an orientable surface link @xmath8 is _ trivial _ ( or _ unknotted _ ) if there is an embedded 3-manifold @xmath9 with @xmath10 such that each component of @xmath9 is a handlebody .
an oriented surface link @xmath8 is called _ pseudo - ribbon _ if there is a surface link diagram of @xmath8 whose singularity set consists of double points and _ ribbon _ if @xmath8 is obtained from a trivial 2-link @xmath11 by 1-handle surgeries along a finite number of mutually disjoint 1-handles attaching to @xmath11 . by definition ,
a ribbon surface link is pseudo - ribbon .
two surface links are _ equivalent _ if there is an ambient isotopy or an orientation - preserving diffeomorphism of @xmath12 or @xmath2 which deforms one to the other . a compact and oriented 2-manifold @xmath13 embedded
properly and locally flatly in @xmath14 is called a _ braided surface _ of degree @xmath5 if @xmath13 satisfies the following conditions : a. @xmath15 is a branched covering map of degree @xmath5 , b. @xmath16 is a closed @xmath5-braid in @xmath17 , where @xmath18 are 2-disks , and @xmath19 is the projection to the second factor . a braided surface @xmath13
is called a _ surface braid _ if @xmath16 is the trivial closed braid .
moreover , @xmath13 is called _ simple _ if every singular index is two .
two braided surfaces are _ equivalent _ if there is a fiber - preserving ambient isotopy of @xmath20 rel @xmath21 which carries one to the other .
there is a theorem which corresponds to alexander s theorem for classical oriented links .
any oriented surface link can be deformed by an ambient isotopy of @xmath2 to the closure of a simple surface braid .
there is a _
chart _ which represents a simple surface braid .
let @xmath5 be a positive integer , and @xmath22 be a graph on a 2-disk @xmath23 .
then @xmath22 is called a _
surface link chart _ of degree @xmath5 if it satisfies the following conditions : a. @xmath24 .
b. every edge is oriented and labeled , and the label is in @xmath25 . c. every vertex has degree @xmath26 , @xmath27 , or @xmath28 .
d. at each vertex of degree @xmath28 , there are six edges adhering to which , three consecutive arcs oriented inward and the other three outward , and those six edges are labeled @xmath29 and @xmath30 alternately for some @xmath29 .
e. at each vertex of degree @xmath27 , the diagonal edges have the same label and are oriented coherently , and the labels @xmath29 and @xmath31 of the diagonals satisfy @xmath32 .
vertices of degree @xmath26 and @xmath28 are called a _ black vertex _ and a _
white vertex_. a black vertex ( resp .
white vertex ) of a chart corresponds to a branch point ( resp .
triple point ) of the simple surface braid associated with the chart . an edge without end points is called a _
loop_. an edge whose end points are black vertices is called a _ free edge _ , and a configuration consisting of a free edge and a finite number of concentric simple loops such that the loops are surrounding the free edge is called an _
oval nest_. an _ unknotted chart _ is a chart presented by a configuration consisting of free edges .
a trivial oriented surface link is presented by an unknotted chart ( @xcite ) .
a _ ribbon chart _ is a chart presented by a configuration consisting of oval nests .
a ribbon surface link is presented by a ribbon chart ( @xcite ) .
a chart with a boundary represents a simple braided surface .
there is a notion of _ c - move equivalence _ ( @xcite ) between two charts of the same degree .
the following theorem is well - known : [ thm1 - 6]two charts of the same degree are c - move equivalent if and only if their associated simple braided surfaces are equivalent .
we give the definition of torus - covering - links in @xmath2 ( cf .
. [ def2 - 1 ] first , embed @xmath33 into @xmath2 naturally , and identify @xmath33 with @xmath34 , where @xmath35 and @xmath36 for @xmath37 $ ] and @xmath38 $ ] .
let us consider a surface link @xmath13 embedded in @xmath33 such that @xmath39 is a simple braided surface .
we call such a surface link a _ torus - covering - link_. a torus - covering - link @xmath13 can be described by a chart on the trivial torus knot , i.e. by a chart @xmath40 on @xmath41 with @xmath42 and @xmath43 .
let us denote the classical braids described by @xmath44 and @xmath45 by @xmath46 and @xmath47 respectively .
we will call @xmath40 a _ torus - covering - chart with boundary braids @xmath46 and @xmath47_. let @xmath48 be the number of black vertices in the torus - covering - chart @xmath40 . then let us consider the case @xmath49 . in this case
the torus - covering - link associated with @xmath40 is determined by the boundary braids @xmath46 and @xmath47 .
we will call such a @xmath40 a torus - covering - chart _ without black vertices and with boundary braids @xmath46 and @xmath47_. * remark . * in the case @xmath50 , the boudary braids @xmath46 and @xmath47 are commutative . by definition ,
torus - covering - links contain symmetry - spun tori ( and spun @xmath0-knots , turned spun @xmath0-knots and torus @xmath0-knots ) . as we stated in theorem
[ thm1 - 6 ] , if there are two surface link charts of the same degree , their associated surface links are equivalent if and only if their charts are c - move equivalent .
it follows that if two torus - covering - charts are c - move equivalent , their associated torus - covering - links are equivalent .
a torus - covering - link has no 2-knot component .
in particular , if a torus - covering - chart has no black vertices , then each component of the associated torus - covering - link is of genus one .
let @xmath40 be a torus - covering - chart of degree @xmath5 and with the trivial boundary braids .
let @xmath8 be the surface link associated with @xmath40 by assuming @xmath40 to be a surface link chart .
then the torus - covering - link associated with the torus - covering - chart @xmath40 is obtained from @xmath8 by applying @xmath5 trivial 1-handle surgeries .
( example 2.2 in @xcite ) 1 .
let @xmath40 be a torus - covering - chart of degree 2 without black vertices and with boudary braids @xmath51 and @xmath52 ( the trivial braid ) .
then the torus - covering - knot associated with @xmath40 is the spun @xmath0-knot of a right - handed trefoil .
2 . let @xmath40 be a torus - covering - chart of degree 2 without black vertices and with boudary braids @xmath51 and @xmath51 ( or @xmath51 and @xmath53 ) .
then the torus - covering - knot associated with @xmath40 is the turned spun @xmath0-knot of a right - handed trefoil .
3 . let @xmath40 be a torus - covering - chart without black vertices and with boudary braids @xmath54 and @xmath55 .
then the torus - covering - knot associated with @xmath40 is a symmetry - spun torus .
for a ( classical ) braid @xmath55 , let @xmath56 be the braid obtained from @xmath55 by adding @xmath57 ( resp .
@xmath58 ) trivial strings before(resp.after ) @xmath55 , and + @xmath59 , @xmath60 , + @xmath61 . +
* let @xmath62 be garside s @xmath62 for the braid group @xmath63
. then @xmath64 ( cf .
@xcite ) .
[ thm3 - 1]let @xmath40 be a torus - covering - chart of degree @xmath5 with boudary braids @xmath65 and @xmath66 .
then the torus - covering - link associated with @xmath40 can be described by a surface link chart @xmath67 of degree @xmath6 as in fig .
[ fig3 - 1 ] , where @xmath68 is a chart describing the simple braided surface as follows : @xmath69 where @xmath70 means an isotopy transformation and @xmath71 a hyperbolic transformation along bands corresponding to the @xmath72 s ( fig .
[ fig3 - 2 ] ) , and @xmath73 is the orientation - reversed mirror image of @xmath68 , and @xmath74 is the braid obtained from the classical braid @xmath66 by taking its mirror image and reversing all the crossings ( fig .
[ fig3 - 3 ] )
. * remark . *
the surface link chart @xmath67 is of degree @xmath6 and well - defined , for the edges representing @xmath75 have labels at most @xmath76 and the edges representing @xmath77 have labels at least @xmath78 .
note the 1-handle chart @xmath68 has @xmath6 black vertices .
( step 1 ) let us consider a trivial torus knot @xmath0 in @xmath2 as the result of 1-handle surgeries of @xmath79 along @xmath80 , where @xmath81 ( resp .
@xmath82 ) is a 2-sphere in @xmath83 with a positive ( resp.negative ) orientation such that @xmath81 contains @xmath82 , and @xmath84 ( resp .
@xmath85 ) is a 1-handle attaching to the two spheres trivially in a neighborhood of the north ( resp .
south ) pole ( fig .
[ fig3 - 4 ] ) .
( step 2 ) deform the two 1-handles and the inner sphere @xmath86 by an ambient isotopy of @xmath2 to make @xmath82 have a positive orientation as in fig .
[ fig3 - 5 ] , where the 1-handle @xmath84 is as in fig .
[ fig3 - 6 ] , which has a double point curve with a branch point for each end , and the other 1-handle @xmath85 is the orientation - reversed mirror image of @xmath84 .
( step 3 ) slide the 1-handle @xmath85 to the neighborhood of the north pole to make both @xmath84 and @xmath85 be in the neighborhood of the north pole , and cut off the two southern hemispheres to obtain the surface braid and the surface link chart of degree @xmath7 as in fig .
[ fig3 - 7 ] .
( step 4 ) now , consider the trivial torus knot @xmath0 as the torus - covering - link associated with @xmath40 ( of degree @xmath5 ) by drawing @xmath40 on @xmath0 ( fig .
[ fig3 - 8 ] ) .
let us denote the @xmath5 1-handles corresponding to @xmath84 ( resp .
@xmath87 ) by @xmath88 ( resp .
@xmath89 ) .
then @xmath88 can be deformed to the 1-handle braided surface as in fig .
[ fig3 - 2 ] , and @xmath90 to the orientation - reversed mirror image of @xmath88 , and the surface braid will be as in fig .
[ fig3 - 9 ] .
hence we obtain the surface link chart @xmath67 of degree @xmath6 as in fig .
[ fig3 - 1 ] .
the _ braid index _ of an oriented surface link @xmath8 is the minimum degree of simple closed surface braids in @xmath2 which are equivalent to @xmath8 .
kamada showed in @xcite that surface links whose braid index is at most three are indeed ribbon , and shima showed in @xcite that the turned spun @xmath0-knot of a non - trivial classical knot is not ribbon . hence we obtain the following corollary : [ cor3 - 3 ] let @xmath13 be the torus - covering - link associated with a torus - covering - chart of degree @xmath5 .
then the braid index of s is equal or less than @xmath6 . in particular ,
the braid index of the turned spun @xmath0-knot of the torus @xmath1-knot is four . the torus - covering - chart @xmath40 describes the turned spun @xmath0-knot of the right - handed trefoil .
its 1-handle chart is as in fig .
[ fig3 - 11 ] , and the surface link chart @xmath67 obtained from @xmath40 is as in fig .
[ fig3 - 12 ] . | we consider surface links in the 4-space which can be deformed to simple branched coverings of a trivial torus knot , which we call torus - covering - links .
torus - covering - links contain spun @xmath0-knots , turned spun @xmath0-knots , symmetry - spun tori and torus @xmath0-knots . in this paper
we study the braid indices of torus - covering - links .
in particular we show that the turned spun @xmath0-knot of the torus @xmath1-knot has the braid index four . |
since it is rare to find a pair of biological species in nature which meet precise prey - dependence or ratio - dependence functional responses in predator - prey models , especially when predators have to search for food ( and therefore , have to share or compete for food ) , a more suitable general predator - prey theory should be based on the so - called ratio - dependent theory ( see @xcite ) .
the theory may be stated as follows : the per capita predator growth rate should be a function of the ratio of prey to predator abundance , and so should be the so - called predator functional response .
such cases are strongly supported by numerous field and laboratory experiments and observations ( see , for instance , @xcite ) .
denote by @xmath1 and @xmath2 the population densities of prey and predator at time @xmath3 respectively . then the ratio - dependent type predator - prey model with michaelis - menten type functional response
is given as follows : [ 1.1 ] & = & rn(1-)- , + & = & p , where @xmath4 and @xmath5 are positive constants .
in , @xmath6 denotes a mortality function of predator , and @xmath5 and @xmath7 the prey growth rate with intrinsic growth rate and carrying capacity in the absence of predation , respectively , while @xmath8 and @xmath9 are model - dependent constants . from a formal point of view
, this model looks very similar to the well - known michaelis - menten - holling predator - prey model : [ 1.2 ] & = & rn(1-)- , + & = & p .
indeed , the only difference between models ( [ 1.1 ] ) and ( [ 1.2 ] ) is that the parameter @xmath10 in ( [ 1.2 ] ) is replaced by @xmath11 in ( [ 1.1 ] ) .
both terms @xmath11 and @xmath10 are proportional to the so - called searching time of the predator , namely , the time spent by each predator to find one prey .
thus , in the michaelis - menten - holling model the searching time is assumed to be independent of predator density , while in the ratio - dependent michaelis - menten type model it is proportional to predator density ( _ i.e. _ , other predators strongly interfere ) .
predators and preys are usually abundant in space with different densities at difference positions and they are diffusive .
several papers have focused on the effect of diffusion which plays a crucial role in permanence and stability of population ( see @xcite , and the references therein ) . especially in @xcite the effect of variable dispersion rates on turing instability was extensively studied , and in @xcite the dynamics of ratio - dependent system has been analyzed in details with diffusion and delay terms included .
cavani and farkas ( see @xcite ) have considered a modification of when a diffusion was introduced , yielding : [ 1.3 ] & = & rn(1-)-+d_1,x(0,l),t>0 , + & = & p+d_2,x(0,l),t>0 , where the specific mortality of the predator is given by @xmath12 which depends on the quantity of predator . here ,
the positive constants @xmath13 and @xmath14 denote the minimal mortality and the limiting mortality of the predator , respectively . throughout the paper , the following natural condition @xmath15 will be assumed , and we will consider the case of the constant diffusivity , @xmath16 , @xmath17 .
the advantage of this model is that the predator mortality is neither a constant nor an unbounded function , but still it is increasing with the predator abundance . on the other hand , combining and ,
many authors ( see @xcite , for instance ) have studied a more general model as follows : & = & rn@xmath18-+d_1,x(0,l ) , t>0 , + & = & p+d_2,x(0,l ) , t>0 , with the specific mortality of the predator somewhat restricted in the form @xmath19 in this paper we consider a ratio - dependent reaction - diffusion predator - prey model with michaelis - menten type functional response and the specific mortality of the predator given by ( [ 1.4 ] ) instead of .
we study the effect of the diffusion on the stability of the stationary solutions .
also we explore under which parameter values turing instability can occur giving rise to non - uniform stationary solutions satisfying the following equations : [ full1 ] & = & rn(1-)-+d_1,x(0,l),t>0 , + & = & p+d_2,x(0,l),t>0 , assuming that prey and predator are diffusing according to fick s law in the interval @xmath20.$ ] we are interested in the solutions @xmath21 fulfilling the neumann boundary conditions @xmath22 and initial conditions @xmath23for simplicity , we nondimensionalize the system ( [ full1 ] ) with the following scaling @xmath24 and letting @xmath25for the sake of simplification of notations , dropping tildes , the system takes the form [ 2.1 ] & = & n(1-n)-+d_1,x(0,l),t>0 , + & = & p+d_2,x(0,l),t>0 .
set @xmath26 where @xmath27.\]]then the system with the boundary conditions takes the form @xmath28 clearly , in case the predator and prey are spatially homogeneous , the spatially constant solution @xmath29 of ( [ 2.3 ] ) , fulfilling the boundary conditions obviously , satisfies the kinetic system @xmath30
in this section we will study the system ( [ 2.1 ] ) without diffusion , _
i.e. _ , [ kinetic ] & = & n(1-n)- , + & = & p . in particular , we will focus on the existence of equilibria and their local stability .
this information will be crucial in the next section where we study the effect of the diffusion parameters on the stability of the steady states .
the equilibria of the system ( [ kinetic ] ) are given by the solution of the following equations @xmath31the system has at least one equilibrium with positive values .
this is the point of intersection of the prey null - cline @xmath32and the predator null - cline @xmath33thus , denoting the coordinates of a positive equilibrium by @xmath34 , these coordinates satisfy @xmath35 the jacobian matrix of the system ( [ kinetic ] ) linearized at @xmath34 is @xmath36 where @xmath37 and _ 1 & = & -+ , _ 2= , + _ 3&= & , _ 4 = + .
the characteristic equation is given by @xmath38 recall that @xmath34 is locally asymptotically stable if @xmath39 , which is equivalent to have @xmath40 and @xmath41 .
for this , we will assume that @xmath42 due to , we see that @xmath43 .
if @xmath44 , then the two conditions in hold .
in this section we will investigate in turing instability and bifurcation for our model problem .
we will also study pattern formation of the predator - prey solutions . before studying the stability of equilibrium solutions
, we will discuss about the local existence and uniqueness of solution for a given ratio - dependent reaction - diffusion predator - prey model . applying the criteria for the local existence of solution ( see @xcite ) to the nonlinear parabolic systems ( [ 2.3 ] ) , we see that there exists a unique local solution of the given system .
let @xmath45 be a bounded region in @xmath46 with smooth boundary @xmath47 and @xmath48 denotes the unit outward normal to @xmath49 .
then morgan considered in reference ( @xcite ) essentially of the form [ slparabolic ] _
t(x , t)&=&d(x , t ) + f((x , t ) ) , x , t>0 , + & = & 0 , x , t>0
, + ( x,0 ) & = & _ 0(x),x , where @xmath50 @xmath51 is a locally lipschitz continuous function , @xmath52 is an @xmath53 diagonal matrix with diagonal entries @xmath54 , and @xmath55 is bounded and measurable .
then the following theorem holds @xcite : [ thm1 ] under the assumptions on stated above , there exists @xmath56 and @xmath57 such that + ( i ) ( [ slparabolic ] ) has a unique classical solution @xmath58 on @xmath59 which can not be continued to @xmath60 for any @xmath61 , and + ( ii ) @xmath62 for all @xmath63 .
+ moreover , if @xmath64 , then @xmath65 as @xmath66 for some @xmath67 .
defining @xmath68 and @xmath69 in our model , theorem [ thm1 ] implies local existence and uniqueness .
more precisely , there exists @xmath56 and @xmath70 and @xmath71 such that + ( i ) ( [ slparabolic ] ) has a unique classical solution @xmath72 on @xmath73\times [ 0,t_{\max})$ ] which can not be continued to @xmath60 for any @xmath61 , and + ( ii ) @xmath74 and @xmath75 for @xmath76 .
+ moreover , if @xmath64 , then either @xmath77 or @xmath78 as @xmath79 we say that the equilibrium @xmath34 is turing unstable if it is an asymptotically stable equilibrium of the kinetic system ( [ kinetic ] ) but is unstable with respect to solutions of ( 2.1 ) ( see @xcite ) .
an equilibrium is turing unstable means that there are solutions of ( [ 2.3 ] ) that have initial values @xmath80 arbitrarily closed to @xmath81 ( in the supremum norm ) but do not tend to @xmath81 as @xmath82 tends to @xmath83 .
we linearize system ( [ 2.1 ] ) at the point @xmath34 : setting @xmath84 the linearized system assumes the form @xmath85 while the boundary conditions remain unchanged : @xmath86 the linear boundary value problem - can be solved in several ways . in particular , the fourier s method of separation of variables
assumes that solutions can be represented in the form @xmath87 with @xmath88 @xmath89\rightarrow \mathbb r.$ ] then @xmath90 and @xmath91 the eigenvalues of the boundary value problem ( [ 2.10 ] ) are @xmath92 with corresponding eigenfunctions @xmath93 clearly , @xmath94 .
these eigenvalues are to be substituted into ( [ 2.9 ] ) . denoting by @xmath95 and @xmath96 the two linearly independent solutions of ( [ 2.12 ] ) associated with @xmath97 , the solution of the boundary value problem ( [ 2.7])-([2.8 ] )
is obtained in the form @xmath98 where @xmath99 is to be determined according to the initial condition @xmath100 for instance , if @xmath101 for @xmath102 , @xmath103 = \frac{1}{l}\int\limits_{0}^{l}\bv(x,0)dx,\ ] ] @xmath104 = \frac{2}{l}\int\limits_{0}^{l}\bv(x,0)\cos \frac{k\pi x}{l}\,dx\qquad , k=0,1,2,\cdots.\ ] ] set @xmath105 according to casten and holland @xcite , if both eigenvalues of @xmath106 have negative real parts for all @xmath107 , then the equilibrium @xmath34 of ( [ 2.3 ] ) is asymptotically stable ; if at least one eigenvalue of a matrix @xmath106 has positive real part , then @xmath108 is unstable . recalling ( [ 2.6 ] ) , the trace and determinant are given by [ 2.17 ] b_j&=&_1-_4-_j(d_1+d_2 ) , + b_j&=&_2_3-_1_4 + _
j\{d_1_4-d_2_1 } + _ j^2d_1d_2 .
notice that implies that @xmath109 therefore the eigenvalues of @xmath110 have negative real parts if @xmath111 which is guaranteed in case @xmath112 notice that @xmath113 for all sufficiently large @xmath107 if @xmath114 since the eigenvalues @xmath115 is monotonic increasing with its limit @xmath116 therefore , one has the following theorem : [ thm2 ] assume that and
. then the equilibrium point @xmath34 of is asymptotically stable if @xmath117 while it is turing unstable if @xmath118 or if there exist a positive integer @xmath119 such that @xmath120 for a nonnegative real parameter @xmath121 consider the reaction - diffusion system to find @xmath122 such that @xmath123 where @xmath52 is a non - negative diagonal matrix depending smoothly on @xmath121 and @xmath124 is a smooth function .
suppose is equipped with the neumann boundary condition @xmath125 assume further that for some @xmath126 we have @xmath127 for all @xmath128 _ i.e. _ @xmath81 is a parameter - independent constant stationary solution of .
we say that @xmath81 undergoes a turing bifurcation at @xmath129 if the solution @xmath81 is asymptotically stable for @xmath130 while it is unstable for @xmath131 ( or vice versa , _
i.e. _ the regions for asymptotical stability and instability may be exchanged ) , and in some neighborhood of @xmath132 the problem ( [ 3.1])-([3.2 ] ) has non - constant stationary solution ( _ i.e. _ solution which does not depend on time but depends on space . ) with @xmath133 fixed , regarding @xmath134 as the parameter @xmath121 , we will consider the linearized system ( [ 2.7])-([2.8 ] ) as a parameter - dependent problem in the setting . notice that @xmath135 is clearly a solution for ( [ 2.7])-([2.8 ] )
. then the condition for a turing bifurcation for the linearized system ( [ 2.7])-([2.8 ] ) is given as follows : [ thm3 ] suppose that @xmath136 and @xmath137 + ( i ) if @xmath138 then the zero solution of the linear problem ( [ 2.7])-([2.8 ] ) is asymptotically stable for all @xmath139 + ( ii ) if @xmath140 then the zero solution of the linear problem ( [ 2.7])-([2.8 ] ) undergoes a turing bifurcation at @xmath141 \(i ) rewriting ( [ 2.17b ] ) as @xmath142 we see from and that @xmath143 for all @xmath102 if @xmath144 holds , since @xmath145 forms a monotone increasing sequence ( [ 3.3 ] ) .
therefore , the zero solution of ( [ 2.7])-([2.8 ] ) is asymptotically stable under such conditions .
\(ii ) suppose @xmath146 satisfies ( [ 3.4 ] ) and choose @xmath147 as given in .
then @xmath148 clearly , we have @xmath149 for @xmath150 , and @xmath151 for @xmath152 . in both cases
@xmath153 again by casten and holland @xcite as quoted just after formula ( [ 2.14 ] ) , the zero solution is asymptotically stable for @xmath150 , and it is unstable for @xmath152 . if @xmath154 , one eigenvalues of @xmath155 becomes zero and the other is @xmath156 , which is negative . denote the eigenvector corresponding to the zero eigenvalue by @xmath157 , _
i.e. _ @xmath158 as we can see from ( [ 2.9])-([2.12 ] ) the function @xmath159 \cos \frac{\pi x}{l},\ ] ] is a spatially non - constant stationary solution of the linearized problem ( [ 2.7])-([2.8 ] ) .
this implies that the zero solution undergoes turing bifurcation at @xmath160 .
this completes the proof . in the remaining part of this section we will extend the latter result about the turing bifurcation of the zero solution of the linearized system to the non - linear problem ( [ 2.3 ] ) . for this
we need the following : [ thm4 ] let @xmath161 and @xmath162 be banach spaces , @xmath163 an open subset of @xmath164 , and @xmath165 such that @xmath166 denote by @xmath167 and @xmath168 the linear operators obtained by differentiating @xmath169 with respect to its first variable only and the first and second variables at @xmath170 , respectively .
assume that the following conditions hold : \(i ) the kernel of @xmath171 , the subspace @xmath172 of @xmath161 is a one - dimensional vector space spanned by @xmath173 ; \(ii ) the range of @xmath171 , the subspace @xmath174 of @xmath162 has codimension @xmath175 , _ i.e. _ @xmath176=1 $ ] ; \(iii ) @xmath177 .
let @xmath178 be an arbitrary closed subspace of @xmath161 such that @xmath179 @xmath180\oplus z;$ ] then there is a @xmath181 and @xmath182-curve @xmath183 such that ; @xmath184 @xmath185 @xmath186 for @xmath187 furthermore , there is a neighborhood of @xmath188 such that any zero of @xmath169 either lies on this curve or is of the form @xmath189 the idea of the proof is to introduce a new parameter @xmath190 which enables to apply immediately the implicit function theorem for the function @xmath191 defined by @xmath192see , for the details of the proof of the theorem , pp .
172173 of @xcite . in what follows the role of the space @xmath161 will be played by @xmath193;\mathbb r^{2}):\text { } \bv_{x}(0)=\bv_{x}(l)=\b0\right\ } \label{3.7}\ ] ] with the norm @xmath194 } | \partial^{\alpha } \mathbf f(x)|,$ ] where @xmath195 denotes the usual vector or matrix - norm , while @xmath196,\mathbb r^{2})$ ] with the norm @xmath197}| \mathbf f(x)|.$ ] however , in choosing the subspace @xmath178 of @xmath161 we shall use the orthogonality induced by the inner product @xmath198dx,\quad\text{for } \bv=(v_{1},v_{2})^t , \bw=(w_{1},w_{2})^t.\ ] ] [ thm5 ] suppose that @xmath40 and @xmath137 \(i ) if ( [ 3.3 ] ) holds , then the constant solution @xmath199 of the nonlinear problem ( [ 2.3 ] ) is asymptotically stable .
\(ii ) if @xmath200 is not parallel to the second eigenvector @xmath201 of @xmath155 and @xmath146 satisfies ( [ 3.4 ] ) , then at @xmath154 the constant solution @xmath81 undergoes a turing bifurcation .
\(i ) follows immediately from the asymptotic stability of the zero solution of the linear problem _ _ ( [ 2.7])-([2.8])_. _ \(ii ) as in the proof of ( i ) of theorem [ thm3 ] , we have that @xmath81 is asymptotically stable for @xmath202 , while it is unstable for @xmath203 .
we have to show the existence of a stationary non - constant solution in some neighborhood of the critical value @xmath204 such a stationary solution @xmath81 satisfies the following system of second - order partial differential equations @xmath205 we consider ( [ 3.9 ] ) as an operator equation on the banach space @xmath161 given by ( [ 3.7 ] ) , and we apply theorem [ thm3 ] with @xmath206 as the bifurcation parameter .
set @xmath207 . then ( [ 3.9 ] )
assumes the equivalent form @xmath208 where @xmath209 is the jacobian matrix of @xmath210 evaluated at @xmath81 and @xmath211 denote the left hand side of ( [ 3.10 ] ) by @xmath212 where @xmath213 is a one - parameter family of operators acting on @xmath161 and taking its elements into @xmath196;\mathbb r^{2}).$ ] clearly , @xmath213 is a @xmath214 mapping .
the spectrum of the linear operator @xmath215 consists of the eigenvalues @xmath216 of the matrices @xmath106 given by ( [ 2.14 ] ) with its corresponding eigenfunctions are @xmath217 @xmath218 where @xmath219 is given by ( [ 2.12 ] ) and @xmath220 is the eigenvector of the matrix @xmath106 corresponding to the eigenvalues @xmath216 ( see ) .
now , all matrices @xmath221 are to be taken at @xmath154 .
as it can be seen from the proof of theorem [ thm3 ] and from for @xmath222 for all nonnegative integer @xmath107 except @xmath223 all @xmath216 have negative real parts . for @xmath224
one eigenvalue @xmath225 is equal to @xmath226 and the other @xmath227 is negative .
the eigenfunction corresponding to @xmath228 is @xmath229 thus , the null - space of the operator @xmath230 is a one - dimensional linear space spanned by @xmath231 owing to the orthogonality and completeness of the eigenfunction system of the operator @xmath232 , the range of this operator is given by @xmath233;\mathbb r^{2}):\text { the eigenfunction expansion of } \\ & & \qquad \bw\text { does not contain } \cos \frac{\pi x}{l } \big\ } \cup \span\big\{\by_{21 } \cos \frac{\pi x}{l } \big\},\end{aligned}\]]so that the codimension of @xmath174 is one .
let @xmath234 then @xmath235 .\]]clearly , @xmath236.\]]under the assumption @xmath237 @xmath238 , we see that @xmath239 does not belong to @xmath240 fulfilling the condition ( iii ) of theorem [ thm4 ] .
letting @xmath241 which is a closed subspace of @xmath162 , we verify that all the hypotheses of theorem [ thm4 ] hold ; moreover , @xmath242 is a bifurcation point , and there exist a @xmath243 a function @xmath244 such that for @xmath245 @xmath246 is a solution of ( [ 3.10 ] ) with @xmath247 , @xmath248 , @xmath249 and @xmath250 the corresponding solution of ( [ 3.9 ] ) , _ i.e. _ the non - constant stationary solution of the nonlinear parabolic system ( [ 2.3 ] ) is @xmath251 ( corresponding to the choice @xmath252 ) , _ i.e. _ [ 3.25 ] n(x ) & = & + s_1+o(s^2 ) , + p(x ) & = & + s_2+o(s^2 ) .
since @xmath190 is considered to be small here , this solution is called as a _ small amplitude pattern . _
because of theorem [ thm4 ] _ ( [ 2.3 ] ) _ has no other stationary solution apart from @xmath34 and in a neighborhood of @xmath253 in the linear case ( by theorem [ thm3 ] ) for the function @xmath206 holds : @xmath254 and a corresponding one parameter family of solutions is @xmath255
the reaction - diffusion equations ( [ 2.3 ] ) are solved numerically using the forward euler method in time , the centered difference method in space .
this numerical scheme gives a stable solution under a certain that stasisfies the cfl ( courant - friedrichs - lewy ) condition .
the details are as follows .
consider the computational domain @xmath256 $ ] and the mesh size @xmath257 and the time step size @xmath258 which will be determined later in .
set @xmath259 denote by @xmath260 and @xmath261 the numerical approximation of @xmath262 , @xmath263 , respectively for @xmath264 and @xmath265 then , given initial data @xmath266 the numerical scheme is to solve [ eq : njpj ] n_j^k+1&=&n_j^k + t n_j^k + t d_1 , + p_j^k+1&=&p_j^k + tp_j^k + t d_2 for @xmath267 iteratively for @xmath268 on the boundaries @xmath269 where neumann condition holds , we used a three - point interpolation scheme to guarantee the second - order accuracy in space as follows : [ eq : bc ] n_2^k - 4 n_1^k + 3 n_0^k = 0 ; & & p_2^k - 4 p_1^k + 3 p_0^k = 0 ; + n_n_h-2^k - 4 n_n_h-1^k + 3 n_n_h^k = 0 ; & & p_n_h-2^k - 4 p_n_h-1^k + 3 p_n_h^k = 0 .
we will then establish the the positivity of the numerical solutions and boundedness for the numerical prey solution under certain conditions on @xmath270 .
suppose that @xmath271 for @xmath272 then , for @xmath273
@xmath274 + \delta t d_1\frac{n_{j-1}^k - 2n_j^k + n_{j+1}^k}{h^2 } \nonumber \\ & \leq & n_j^k + { \delta t } n_j^k \left[1-n_j^k \right ] + 2\frac{\delta t d_1}{h^2}(1- n_j^k)\nonumber \\ & = & n_j^k + { \delta t } ( 1-n_j^k ) \left[n_j^k + \frac{2d_1}{h^2}\right ] \nonumber \\ & \leq & n_j^k + { \delta t } ( 1-n_j^k ) ( 1+\frac{2d_1}{h^2 } ) \nonumber \\ & = & \left[1-\delta t(1+\frac{2d_1}{h^2})\right]n_j^{k } + \delta t(1+\frac{2d_1}{h^2})\nonumber \\ & \leq & 1-\delta t(1+\frac{2d_1}{h^2 } ) + \delta t(1+\frac{2d_1}{h^2 } ) \leq 1 \label{est : nj}\end{aligned}\ ] ] provided @xmath275 for @xmath224 , owing to the boundary condition , @xmath276 is given by @xmath277 + \delta t d_1\frac { 4 ( - n_{1}^k + n_2^k)}{3h^2}.\end{aligned}\ ] ] hence , the same analysis as above yields , instead of ,
@xmath278 n_1^k + \delta t ( 1+\frac{4d_1}{3h^2 } ) \le 1\end{aligned}\ ] ] provided @xmath279 analgously , one gets @xmath280 + \delta t d_1\frac { 4 ( - n_{n_h-1}^k + n_{n_h-2}^k)}{3h^2},\end{aligned}\ ] ] and therefore @xmath281 n_{n_h-1}^k + \delta t ( 1+\frac{4d_1}{3h^2 } ) \le 1\end{aligned}\ ] ] provided @xmath279 on the other hand , suppose that @xmath282 for @xmath272 then , for @xmath283 @xmath284 + \delta t d_1\frac{n_{j-1}^k - 2n_j^k + n_{j+1}^k}{h^2}\nonumber \\ & \geq & n_j^k + \delta t n_j^k(1-n_j^k ) - \delta t \alpha n_j^k -2\frac{\delta t d_1}{h^2}n_j^k\nonumber\\ & \ge&(1+\delta t-\delta t\alpha)n_j^k-\delta t n_j^k - 2\frac{\delta t d_1}{h^2}n_j^k\nonumber\\ & = & \left[1-\delta t \alpha-\delta t\frac{2d_1}{h^2}\right]n_j^{k } \geq 0,\end{aligned}\ ] ] provided @xmath285 next for @xmath224 , by using , the procedure to get the estimate leads to @xmath286n_1^{k } \geq 0\end{aligned}\ ] ] provided @xmath287 similarly , under the same conditions , one obtains @xmath288n_{n_h-1}^{k } \geq 0.\end{aligned}\ ] ] next , suppose that @xmath289 and @xmath290 for @xmath291 recalling , one then obtains , for @xmath273 p_j^k+1&=&p_j^k + tp_j^k + t d_2 + & & p_j^k - tp_j^k - p_j^k + & & ( 1-t - ) p_j^k 0 [ est : pj ] provided @xmath292 . for @xmath224 and @xmath293 , taking into account of the boundary condition , one gets p_j^k+1 ( 1-t - ) p_j^k 0 j = 1j = n_h-1 provided @xmath294 . collecting all the above results , we are now in a position to state the following theorem : let @xmath295 for @xmath296 suppose that @xmath297 then the numerical solutions @xmath260 and @xmath261 obtained iteratively by and satisfies that @xmath298 numerically a steady state is declared to reach when either the @xmath299 or @xmath300-norm difference is less than a given tolerance value .
the @xmath299 and @xmath300-norm differences are defined as follows : @xmath301}{\max } \left|\bu_{steady}(x , k\delta t)-\bu_{h}(x , k\delta t)\right|,\end{aligned}\ ] ] where @xmath302 are given by with @xmath303 terms neglected and @xmath304 is the piecewise linear interpolation of the numerical solution @xmath305 set @xmath306 .
the unique positive equilibrium is @xmath307 if we fix @xmath308 for the length of the habitat the interval ( [ 3.4 ] ) becomes @xmath309 in the following figure [ fig : d1d2 ] , stability regions , the mean prey - predator diffusion coefficients , @xmath146 and @xmath206 , are plotted .
we tested our model in the cases of @xmath310 = ( 0.005,0.2 ) and @xmath310 = ( 0.005,0.32 ) , which are in the stable and unstable regions with varying @xmath311 , respectively . in these cases ,
the critical value for turing bifurcation @xmath312 is @xmath313 .
figure [ fig : merge_np ] shows the numerical prey and predator solutions , @xmath314 and @xmath315 , with respect to time at a specified fixed point @xmath316 . as shown in figure [ fig : merge_np ] , for @xmath310 = ( 0.005,0.2 ) , the equilibrium solution @xmath34 is asymptotically stable and for @xmath317 , the equilibrium solution @xmath34 is unstable . for the simulation in the case of @xmath318 we used the spatial mesh size @xmath319 and the time step size @xmath320 determined by the .
the iteration was run until the time equals to 1000 , with approximately @xmath321 iterations . in the case of @xmath322
the mesh size @xmath257=0.005 and the time step size @xmath270= 0.0000375 were used , which were alsothe . in this case also the simulation was done until the time equals to 1000 , with approximately @xmath323 iterations . in figure
[ fig : d202 ] , in case of @xmath318 the prey and predator solutions are plotted with respect to number of iterations and space .
we clearly see that as time goes to infinity , the solution converges to the equilibrium solution @xmath34 . in the lower figure in figure
[ fig : d202 ] , in case of @xmath324 where @xmath134 is in unstable region , the prey and predator solutions are plotted with respect to number of iterations and space .
we clearly see that as time goes to infinity , the solution shows the deviation from the equilibrium solution @xmath34 . in figure
[ fig : varying - s ] , for the values near @xmath312 , @xmath310 = ( 0.005,0.27 ) and @xmath310 = ( 0.005,0.272 ) are considered . by varying @xmath190 values from 0.05 to 0.4
, the prey predator solution has a _ small amplitude pattern _ which we expected in the theory . in figure
[ fig : merge - as - preda ] and figure [ fig : merge - as - prey ] , we have plotted the prey and predator solutions and their small amplitude patterns with respect to number of iterations and space by changing @xmath190 values . near the @xmath312 , in case of @xmath325 we use the mesh sizes @xmath326 and ran our simulation until the number of iteration is approximately @xmath327 . in case of @xmath328 we have used with the mesh sizes @xmath329 and @xmath330 . again
our runs were continued until the number of iteration was approximately @xmath327 . in figure
[ fig : merge - as - preda ] and figure [ fig : merge - as - prey ] , the axis scale in @xmath331 has been used as that of the case of @xmath332 which has a bigger amplitude pattern . comparing the solutions in figure [ fig :
merge - as - preda ] and figure [ fig : merge - as - prey ] with the non - constant stationary solution ( [ 3.25 ] ) , we clearly observe that as time goes to infinity the prey and predator solutions converge to non - constant stationary solution ( [ 3.25 ] ) which confirms that @xmath34 undergoes a turing bifurcation .
system ( [ 2.1 ] ) describes the dynamics of a ratio - dependent predator - prey interaction with diffusion .
prey quantity grows logistically in the absence of predation , predator mortality is neither a constant nor an unbounded function , but it is increasing with the predator abundance and both species are subject to fickian diffusion in a one - dimensional spatial habitat from which and into which there is no migration .
it is assumed that the system without diffusion has a positive equilibrium and under certain conditions it is asymptotically stable .
we show that analytically at a certain critical value a diffusion driven ( turing type ) instability occurs , _ i.e. _ the stationary solution stays stable with respect to the kinetic system ( the system without diffusion ) .
we also show that the stationary solution becomes unstable with respect to the system with diffusion and that turing bifurcation takes place : a spatially non - homogenous ( non - constant ) solution ( structure or pattern ) arises . a first order approximation of this pattern is explicitly given . a numerical scheme that preserve the positivity of the numerical solutions and the boundedness of prey solution
is introduced .
numerical examples are also included .
research partially supported by the bk21 mathematical sciences division , seoul national university , kosef ( abrl ) r14 - 2003 - 019 - 01002 - 0 , and krf-2007-c00031 .
upper : the prey / predator solution pattern @xmath333 when @xmath334 with varing @xmath190 .
( @xmath133:0.005,@xmath134:0.27,@xmath312:0.271 ) lower : the prey / predator solution pattern @xmath335 when @xmath336 .
( @xmath133:0.005,@xmath134:0.272,@xmath312:0.271),title="fig:",width=377 ] 0.3 cm upper : the prey / predator solution pattern @xmath333 when @xmath334 with varing @xmath190 .
( @xmath133:0.005,@xmath134:0.27,@xmath312:0.271 ) lower : the prey / predator solution pattern @xmath335 when @xmath336 .
( @xmath133:0.005,@xmath134:0.272,@xmath312:0.271),title="fig:",width=377 ] upper left : the predator solution pattern @xmath337 when @xmath338 .
upper right : the predator solution pattern @xmath337 when @xmath339 .
( @xmath133:0.005,@xmath134:0.27,@xmath312:0.271 ) lower left : the predator solution pattern @xmath337 when @xmath340 .
upper right : the predator solution pattern @xmath337 when @xmath341 .
( @xmath133:0.005,@xmath134:0.272,@xmath312:0.271),title="fig:",height=226 ] 0.3 cm upper left : the predator solution pattern @xmath337 when @xmath338 .
upper right : the predator solution pattern @xmath337 when @xmath339 .
( @xmath133:0.005,@xmath134:0.27,@xmath312:0.271 ) lower left : the predator solution pattern @xmath337 when @xmath340 .
upper right : the predator solution pattern @xmath337 when @xmath341 .
( @xmath133:0.005,@xmath134:0.272,@xmath312:0.271),title="fig:",height=226 ] 0.3 cm upper left : the predator solution pattern @xmath337 when @xmath338 .
upper right : the predator solution pattern @xmath337 when @xmath339 .
( @xmath133:0.005,@xmath134:0.27,@xmath312:0.271 ) lower left : the predator solution pattern @xmath337 when @xmath340 .
upper right : the predator solution pattern @xmath337 when @xmath341 .
( @xmath133:0.005,@xmath134:0.272,@xmath312:0.271),title="fig:",height=226 ] 0.3 cm upper left : the predator solution pattern @xmath337 when @xmath338 .
upper right : the predator solution pattern @xmath337 when @xmath339 .
( @xmath133:0.005,@xmath134:0.27,@xmath312:0.271 ) lower left : the predator solution pattern @xmath337 when @xmath340 .
upper right : the predator solution pattern @xmath337 when @xmath341 .
( @xmath133:0.005,@xmath134:0.272,@xmath312:0.271),title="fig:",height=226 ] upper left : the prey solution pattern @xmath342 when @xmath338 .
upper right : the prey solution pattern @xmath342 when @xmath339 .
( @xmath133:0.005,@xmath134:0.27,@xmath312:0.271 ) lower left : the prey solution pattern @xmath342 when @xmath340 .
upper right : the prey solution pattern @xmath342 when @xmath341 .
( @xmath133:0.005,@xmath134:0.272,@xmath312:0.271),title="fig:",height=226 ] 0.3 cm upper left : the prey solution pattern @xmath342 when @xmath338 .
upper right : the prey solution pattern @xmath342 when @xmath339 .
( @xmath133:0.005,@xmath134:0.27,@xmath312:0.271 ) lower left : the prey solution pattern @xmath342 when @xmath340 .
upper right : the prey solution pattern @xmath342 when @xmath341 .
( @xmath133:0.005,@xmath134:0.272,@xmath312:0.271),title="fig:",height=226 ] 0.3 cm upper left : the prey solution pattern @xmath342 when @xmath338 .
upper right : the prey solution pattern @xmath342 when @xmath339 .
( @xmath133:0.005,@xmath134:0.27,@xmath312:0.271 ) lower left : the prey solution pattern @xmath342 when @xmath340 .
upper right : the prey solution pattern @xmath342 when @xmath341 .
( @xmath133:0.005,@xmath134:0.272,@xmath312:0.271),title="fig:",height=226 ] 0.3 cm upper left : the prey solution pattern @xmath342 when @xmath338 .
upper right : the prey solution pattern @xmath342 when @xmath339 .
( @xmath133:0.005,@xmath134:0.27,@xmath312:0.271 ) lower left : the prey solution pattern @xmath342 when @xmath340 .
upper right : the prey solution pattern @xmath342 when @xmath341 .
( @xmath133:0.005,@xmath134:0.272,@xmath312:0.271),title="fig:",height=226 ] * figure 1 @xmath133 and @xmath134 plot , from equation ( [ 3.5 ] ) * figure 2 left : the prey solution at @xmath343=0.25 with respect to time , the constant line represents @xmath344 and the two solid lines represent two different @xmath134 values . right : the predator solution at @xmath343=0.25 with respect to time , the constant line represents @xmath345 and the two solid lines represent two different @xmath134 values . * figure 3(@xmath133:0.005,@xmath134:0.2,@xmath312:0.271 ) upperleft : the prey solution @xmath342 with respect to time and space when @xmath334 .
prey pattern shows the convergence to the equilibrium solution @xmath314 as time increases .
upperright : the predator solution @xmath337 with respect to space when @xmath334 .
predator pattern shows the convergence to the equilibrium solution @xmath315 as time increases.(@xmath133:0.005,@xmath134:0.32,@xmath312:0.271 ) lowerleft : the prey solution @xmath342 with respect to time and space when @xmath336 .
prey pattern shows the deviation from the equilibrium solution @xmath314 as time increases .
lowerright : the predator solution @xmath337 with respect to space when @xmath336 .
predator pattern shows the deviation from the equilibrium solution @xmath315 as time increases . *
figure 4 left : the prey / predator solution pattern @xmath333 when @xmath334 with varing @xmath190 .
( @xmath133:0.005,@xmath134:0.27,@xmath312:0.271 ) right : the prey / predator solution pattern @xmath335 when @xmath336 .
( @xmath133:0.005,@xmath134:0.272,@xmath312:0.271 ) * figure 5 upper left : the predator solution pattern @xmath337 when @xmath338 .
upper right : the predator solution pattern @xmath337 when @xmath339 .
( @xmath133:0.005,@xmath134:0.27,@xmath312:0.271 ) lower left : the predator solution pattern @xmath337 when @xmath340 .
upper right : the predator solution pattern @xmath337 when @xmath346 .
( @xmath133:0.005,@xmath134:0.272,@xmath312:0.271 ) * figure 6 upper left : the prey solution pattern @xmath342 when @xmath338 .
upper right : the prey solution pattern @xmath342 when @xmath339 .
( @xmath133:0.005,@xmath134:0.27,@xmath312:0.271 ) lower left : the prey solution pattern @xmath342 when @xmath340 .
upper right : the prey solution pattern @xmath342 when @xmath341 .
( @xmath133:0.005,@xmath134:0.272,@xmath312:0.271 ) | ratio - dependent predator - prey models have been increasingly favored by field ecologists where predator - prey interactions have to be taken into account the process of predation search . in this paper
we study the conditions of the existence and stability properties of the equilibrium solutions in a reaction - diffusion model in which predator mortality is neither a constant nor an unbounded function , but it is increasing with the predator abundance . we show that analytically at a certain critical value a diffusion driven ( turing type ) instability occurs , _
i.e. _ the stationary solution stays stable with respect to the kinetic system ( the system without diffusion ) .
we also show that the stationary solution becomes unstable with respect to the system with diffusion and that turing bifurcation takes place : a spatially non - homogenous ( non - constant ) solution ( structure or pattern ) arises . a numerical scheme that preserve the positivity of the numerical solutions and the boundedness of prey solution will be presented .
numerical examples are also included
. [ section ] [ theorem]lemma [ theorem]proposition [ theorem]remark [ theorem]corollary [ theorem]definition # 1#2#1_#2 i @xmath0 ) 00 reaction - diffusion system , population dynamics , bifurcation , pattern formation .
35k57 , 92b25 , 93d20 . |
in what follows , all manifolds , maps and fiber bundle structures will be assumed to be differentiable .
suppose that @xmath0 is any locally trivial fiber bundle such that the fibers are of even dimension but the base space @xmath1 could have dimension in either parity .
let @xmath2 be the fibration map and @xmath3 the structure group .
let us denote such a fiber bundle by the symbol @xmath4 if there exist a symplectic structure @xmath5 on the fiber @xmath6 and all transition maps are symplectomorphisms of @xmath7 ( i.e. , @xmath8 ) , then the fiber bundle in ( [ eqn : fiber_bundle ] ) is called a _
symplectic fibration_. note that , for a symplectic fibration , each fiber @xmath9 is equipped with a symplectic structure @xmath10 obtained from pulling back @xmath5 using a local trivialization around @xmath11 , and this construction is independent of the choosen trivialization as transition maps lies in @xmath12 .
+ in order to motivate the reader , let s recall some related work in even dimensions . given a symplectic fibration as above suppose that the total space @xmath0 admits a symplectic form @xmath13 .
( in particular , @xmath1 is assumed to be even dimensional . )
then @xmath13 is said to be _ compatible with _
the fibration @xmath2 if it restricts to a symplectic form on every fiber @xmath14 .
note that the compatibility condition implies that @xmath15 symplectically embeds into @xmath16 for each @xmath17 .
this definition is motivated by the result which states that on a locally trivial fiber bundle as in ( [ eqn : fiber_bundle ] ) with a connected base if @xmath0 admits a symplectic form @xmath13 which restricts to a symplectic form on every fiber , then @xmath2 is a symplectic fibration compatible with @xmath13 .
conversely , due to a result of thurston , for a symplectic fibration with a connected symplectic base if the symplectic form @xmath18 represents a cohomology class in @xmath19 , then one can construct a symplectic form on @xmath0 which is compatible with @xmath20 . for more details about symplectic topology and symplectic fibrations with even dimensional base ,
we refer the reader @xcite .
+ in this note , we consider similar approaches to build `` compatible '' contact structures on the total spaces of exact symplectic fibrations ( see below for definitions ) .
in particular , the base spaces will be odd dimensional and admit contact structures .
here we only consider co - oriented contact structures , that is , those which appear as kernels of certain globally defined @xmath21-forms : a _ positive contact form _ on a smooth oriented @xmath22-dimensional manifold @xmath23 is a @xmath21-form @xmath24 such that @xmath25 ( i.e. , @xmath26 is a positive volume form on @xmath23 ) .
the hyperplane field ( of rank @xmath27 ) @xmath28 of a contact form @xmath24 is called a _ positive ( co - oriented ) contact structure _ on @xmath23 .
the pair @xmath29 ( or sometimes @xmath30 ) is called a _
contact manifold_. we say that two contact manifolds @xmath29 and @xmath31 are _ contactomorphic _ if there exists a diffeomorphism @xmath32 such that @xmath33 .
two contact structures @xmath34 on a @xmath23 are said to be _ isotopic _ if there exists a @xmath21-parameter family @xmath35 ( @xmath36 ) of contact structures joining them ( gray s stability ) .
note that isotopic contact structures give contactomorphic contact manifolds by gray s theorem .
more details about contact topology can be found in @xcite .
+ through out the paper , we will assume the fibers of any fiber bundle to be noncompact or compact with nonempty boundaries , ( and , therefore , the symplectic structures on them will be exact ) . in section [ sec : contact_symplectic_fibrations ] , we will show the existence of a compatible contact structure on a fiber bundle with exact symplectic fibers , compact contact base and structure group contained in the group of exact symplectomorphisms of the fiber .
one should note that hajduk - walczak @xcite showed the existence of contact structures on some of such fiber bundles which are covered by the corresponding theorem of the present paper .
therefore , one can say that the related results of this note generalize the corresponding statements in @xcite . after constructing contact structures on total spaces of bundles , in section [ sec : bundle_contact_structures ] we also discuss how they are affected if we vary the contact structures on the bases in their isotopy classes .
+ fiber sum operation on fiber bundles or more generally fiber connected summing of manifolds along their submanifolds is a well known technique to obtain new fiber bundles and manifolds out of given ones .
the results from @xcite and @xcite show that under suitable conditions , fiber sum process can be done in the symplectic category .
geiges @xcite proved that an analog result holds also in contact category under certain conditions . in section [ sec : fiber_connected_sum ] , by following the fiber connected sum description given in @xcite we will see that under suitable assumptions one can take the fiber connected sum of two fiber bundles ( equipped with compatible contact structures ) along their fibers which results in another fiber bundle whose total space admiting a compatible contact structure which agrees with the original ones away from the region where we take fiber connected sum , and whose base is the contact connected sum of the original contact bases . _ acknowledgments .
_ the author would like to thank tbitak , the scientific and technological research council of turkey , for supporting this research , and also yildiray ozan for helpful conversations .
a contact manifold @xmath37 is called _ strongly symplectically filled _ by a symplectic manifold @xmath7 if there exists a liouville vector field @xmath38 of @xmath5 defined ( at least ) locally near @xmath39 such that @xmath38 is transversally pointing out from @xmath40 and @xmath41 on @xmath40 .
in such a case , we also say that @xmath42 is the _ convex boundary _ of @xmath7 .
+ an _ exact symplectic manifold _ is a noncompact manifold @xmath6 ( or a compact manifold with boundary ) , together with a symplectic form @xmath5 and a @xmath21-form @xmath43 satisfying @xmath44 . if @xmath6 is a compact manifold with boundary , then we also require that the @xmath5-dual vector field @xmath38 of @xmath43 defined by @xmath45 should point strictly outwards along @xmath46 , in particular , @xmath47 is a contact form which makes @xmath46 the convex boundary of @xmath6 . since @xmath43 determines both @xmath5 and @xmath38 , it suffices to write an exact symplectic manifold as a pair @xmath48 . the @xmath21-form @xmath43 is called an _ exact symplectic structure _ on @xmath6 .
since any symplectic form determines an orientation , any exact symplectic manifold @xmath48 is ( symplectically ) oriented .
an _ exact symplectomorphism _ @xmath49 between two exact symplectic manifolds is a diffeomorphism such that @xmath50 is exact .
+ let us start with defining the analog of the above compatibility notion ( given in the introduction ) in odd dimensions .
[ def : compatibility ] let @xmath51 be a fiber bundle with odd dimensional base such that the total space @xmath0 admits a contact structure @xmath52 .
suppose that the fiber space @xmath6 is a noncompact manifold or a compact manifold with boundary .
then @xmath52 is said to be _ compatible with _
@xmath20 if there exists a contact form @xmath24 for @xmath52 such that @xmath24 restricts to an exact symplectic structure on every fiber of @xmath20 .
given an exact symplectic manifold @xmath48 , one can define the group @xmath53 of exact symplectomorphisms which is a subgroup of the symplectic group @xmath54 also if @xmath55 , one can consider the subgroup @xmath56 of exact symplectomorphisms which are identity near @xmath46 .
+ we can now formally introduce our main objects of interest : [ def : contact_symplectic_fibration ] a _ contact symplectic fibration _ is a fiber bundle @xmath57 with exact symplectic fiber @xmath48 and the structure of group of exact symplectomorphisms ( i.e. , @xmath58 ) such that the base space @xmath1 is compact and admits a co - oriented contact structure . from the definition one might understand that the phrase `` contact symplectic fibration '' refers to a fiber bundle with a compact contact base and an exact symplectic fibers , and also that the adjective `` contact '' in the phrase only emphasizes that the base space is contact .
however , the main result of this section ( theorem [ thm : existence_on_contact_symp_fib ] ) shows that the word `` contact '' has , indeed , a more global meaning .
+ given a fiber bundle @xmath51 , if the base and/or fiber spaces are manifolds with boundaries , then the boundary of @xmath0 consists of two parts : the _ vertical boundary component _ @xmath59 , and the _ horizontal boundary component _
@xmath60 where @xmath61 is the fiber over @xmath11 .
[ thm : existence_on_contact_symp_fib ] the total space of a contact symplectic fibration @xmath51 admits a ( co - oriented ) contact structure @xmath62 which is compatible with @xmath20 .
moreover , if @xmath6 has a boundary and @xmath63 , then the contact form @xmath64 is equal to a product form on some collar neighborhood of @xmath65 .
note that this theorem can be regarded as the analog of thurston s theorem mentioned above in odd dimensions . before proving the theorem ,
let us see some examples of contact symplectic fibrations appeared under a different name in the literature . to this end
, we need to recall exact symplectic fibrations from @xcite . here for our purposes we define them in a slightly different way .
[ def : exact_symplectic_fibration ] an _ exact symplectic fibration _ is a fiber bundle @xmath66 equipped with a @xmath21-form @xmath67 on @xmath0 such that each fiber @xmath68 with @xmath69 is an exact symplectic manifold . an exact symplectic fibration is said to be _ trivial near horizontal boundary _ if @xmath55 and the following triviality condition near @xmath70 is satisfied : choose a point @xmath11 and consider the trivial fibration @xmath71 with the form @xmath72 which is the pullback of @xmath73 , respectively .
then there should be a fiber - preserving diffeomorphism @xmath74 between neighborhoods @xmath75 of @xmath70 in @xmath0 and @xmath76 of @xmath77 in @xmath78 which maps @xmath70 to @xmath77 , equals the identity on @xmath79 , and @xmath80 lemma 1.1 of @xcite implies that for an exact symplectic fibration @xmath81 the structure group @xmath82 falls into @xmath83 where @xmath84 is the restriction of the @xmath21-form @xmath67 on @xmath0 to any fixed fiber @xmath85 .
the same result also implies that if we have an exact symplectic fibration which is trivial near horizontal boundary , then its structure group is contained in @xmath86 . also as discussed after lemma 1.1 , using the fact that @xmath87 is nondegenerate on @xmath68 for any @xmath11
, one can define a preferred ( or compatible ) connection on any exact symplectic fibration .
moreover , one can proceed in the other direction as well , that is , any fiber bundle with structure group contained in @xmath83 ( resp .
@xmath86 ) and equipped with a compatible connection admits a structure of exact symplectic fibration ( resp . a structure of exact symplectic fibration which is trivial near horizontal boundary ) .
putting these observations together for exact symplectic fibrations we obtain [ lem : bundle_structure ] any exact symplectic fibration @xmath88 is equipped with a fiber bundle structure with transition maps contained in @xmath89 , that is , @xmath90 , where @xmath91 is the restriction of @xmath67 to any fixed fiber @xmath85 .
if it is trivial near horizontal boundary , then @xmath63 . as an immediate consequence of this lemma , we have any exact symplectic fibration with a compact base which admits a co - oriented contact structure can be equipped with a structure of a contact symplectic fibration . for completeness
let us provide the following elementary fact which will be used in the proof of theorem [ thm : existence_on_contact_symp_fib ] : [ lem : contact_str_on_m^odd_cross_exact ] the product of a contact manifold with an exact symplectic manifold admits a contact structure .
let @xmath92 be any contact manifold , and @xmath93 an exact symplectic manifold .
consider the product @xmath94 and the standard projections given by @xmath95 consider the pull - back forms @xmath96 , @xmath97 and set @xmath98 then by the binomial expansion formula and using the facts @xmath99 and @xmath100 , we compute @xmath101 therefore , @xmath102 is a volume form on the product manifold @xmath94 because @xmath103 and @xmath104 are volume forms on @xmath1 and @xmath6 , respectively .
hence , @xmath64 is a contact form on @xmath94 . after the above observations
, one can prove theorem [ thm : existence_on_contact_symp_fib ] in the following way : given a contact symplectic fibration as in definition [ def : contact_symplectic_fibration ] , as the transition maps are contained in @xmath83 we can glue the forms @xmath43 on each locally trivial piece of the fibration @xmath105 to get a smooth @xmath21-form , say @xmath67 , on @xmath0 which restricts to an exact symplectic structure on each fiber of @xmath20 .
then by following the lines in the proof of lemma [ lem : contact_str_on_m^odd_cross_exact ] , it is easy to check that the pull - back form @xmath106 on @xmath0 defines contact structure which is compatible with @xmath20 . for the purpose of the subsequent parts of the paper we now prove theorem [ thm : existence_on_contact_symp_fib ] in a different perspective .
let @xmath51 be any given contact symplectic fibration with @xmath107 and @xmath108 . by the fiber bundle structure ,
we know that up to diffeomorphism the total space @xmath0 is obtained by patching the trivial @xmath6-bundles , and the transition maps are contained in the structure group @xmath58 where @xmath48 is any fixed fiber of @xmath20 .
more precisely , there is an open cover @xmath109 of @xmath1 and a collection of diffeomorphisms @xmath110 such that the diagram @xmath111 commutes .
moreover , by restricting @xmath112 to any fiber @xmath113 and then projecting onto @xmath6-factor , we get a smooth map @xmath114 , and the transition maps @xmath115 given by @xmath116 using paracompactness , one can find another open cover @xmath117 of @xmath1 such that for any index @xmath118 the closure @xmath119 is contained in @xmath120 for some @xmath24 . since @xmath1 is compact ( by assumption )
, there is a finite subcover @xmath121 of @xmath117 . here
one can assume that @xmath122 s are listed in a specific order so that @xmath123 whenever @xmath122 and @xmath124 belong to the same connected component of @xmath1 .
indeed , the argument below can be done independently for each connected component .
therefore , without loss of generality @xmath1 will be assumed to be connected .
now consider the collection @xmath125 of compact sets defined by @xmath126 clearly , @xmath127 forms a partition for the base space @xmath1 , that is , we have @xmath128 for any @xmath129 , @xmath130 , and each union @xmath131 intersects with @xmath132 along a smooth compact hypersurface @xmath133 observe that for each @xmath134 we have the map @xmath135 obtained by restricting the corresponding @xmath136 .
more precisely , from the construction of @xmath117 , we know for each @xmath137 and @xmath138 with @xmath139 that @xmath140 and @xmath141 for some @xmath142 .
so one can define the map @xmath143 by patching the restrictions of @xmath144 to the subsets @xmath145 together . as a result ,
up to diffeomorphism we obtain the description given by @xmath146 where each @xmath147 is the smooth gluing map defined by @xmath148 for each @xmath137 the fact that @xmath149 implies that there exists a smooth function @xmath150 such that @xmath151 . in accordance with the smooth gluing map ( or rule ) @xmath152 , one can patch all these functions together which yields a smooth function @xmath153 .
+ next , in order to construct a compatible contact structure on @xmath0 , we will use the above description as follows : by assumption @xmath1 admits a contact structure , say @xmath154 . for any real number @xmath155 , the @xmath21-form @xmath156 is contact ( indeed , the manifolds @xmath157 and @xmath158 are contactomorphic , and their contact hyperplane distributions coincide everywhere on @xmath1 ) .
also from lemma [ lem : contact_str_on_m^odd_cross_exact ] we know that for any @xmath155 the @xmath21-form @xmath159 is a contact form on @xmath94 , and so , in particular , on each @xmath160 .
( for simplicity , we will denote the pull - back of @xmath154 ( resp .
@xmath43 ) on any product space still by @xmath154 ( resp .
by @xmath43 ) .
+ for a sufficiently small @xmath161 consider the tubular neighborhood @xmath162 $ ] of @xmath163 in @xmath164 , and let @xmath165 \times f \longrightarrow \r$ ] be a smooth cut - off function defined by @xmath166 where @xmath167 \to \r$ ] is the smooth cut - off function such that @xmath168 on @xmath169 for some @xmath170 and @xmath171 for all @xmath172 near @xmath173 as in figure [ fig : cut_off_function ] .
+ ] for a real number @xmath174 , consider the smooth @xmath21-form @xmath175 on @xmath162 \times f$ ] . by construction
we can smoothly glue this form with the contact form @xmath176 on @xmath177 \times f$ ] .
also observe that @xmath178 is in accordance with the gluing map @xmath179 because for any fiber @xmath6 over the gluing region @xmath180 ( where @xmath181 ) we have @xmath182 .
therefore , @xmath178 is , indeed , a smooth @xmath21-form on @xmath183 .
moreover , on the region @xmath162 \times f$ ] we compute + @xmath184 + where @xmath185 are constants detemined by the binomial expansion formula for @xmath186 , and we use the facts @xmath187 and @xmath188 . observe that if @xmath189 is chosen large enough , the first term on the right hand side is dominant over all other terms since it has the highest degree of @xmath189 .
therefore , for @xmath189 large enough @xmath190 is a volume form on @xmath162 \times f$ ] , and hence @xmath178 is a contact form on @xmath183 .
+ now we want to extend @xmath178 to a contact form on @xmath191 : for @xmath161 sufficiently small , consider the tubular neighborhood @xmath192 $ ] of @xmath193 in @xmath194 such that @xmath195 \subset w^3 $ ] and @xmath196 \subset w^1 \cup w^2 $ ] . take a smooth cut - off function @xmath197 \times f \longrightarrow \r$ ] defined by @xmath198 where @xmath199 \to \r$ ] is the smooth cut - off function such that @xmath200 on @xmath201 $ ] for some @xmath170 and @xmath202 for all @xmath172 near @xmath203 as in figure [ fig : cut_off_function_2 ] .
+ for a real number @xmath204 , we define the smooth @xmath21-form @xmath205 on @xmath192 \times f$ ] where @xmath206 . on the region
@xmath192 \times f$ ] , similar to the computation given in equation ( [ eqn : contact_condition ] ) we have ] @xmath207 which implies that if @xmath208 is chosen large enough , then @xmath209 is contact on @xmath192 \times f$ ] . moreover , the functions @xmath210 and @xmath211 are identical on their overlaping domains because @xmath212 and @xmath213 are constructed using the same transition functions on their common domains . as a result ,
if we choose @xmath214 , then by gluing with @xmath178 we obtain a smooth contact form @xmath209 on @xmath191 .
+ next , we can repeatedly follow exactly the same lines of the last extension argument for the remaining parts of the description of @xmath0 given in equation ( [ eqn : description ] ) , and eventually obtain a contact form on @xmath0 .
+ more precisely , for each @xmath215 and @xmath161 small enough , consider the smooth cut - off function @xmath216 \times f \longrightarrow \r$ ] defined by @xmath217 . here , similar to above , we assume that @xmath218 \subset w^{i+1}$ ] and @xmath219 \subset w^1 \cup \cdots \cup w^i$ ] .
suppose that the contact form @xmath220 is already constructed on the union @xmath221 , and it extends @xmath222 .
note that one needs to choose the constants @xmath223 ( @xmath224 ) to be @xmath225 to perform this extension
. then one can similarly check as in equation ( [ eqn : contact_condition_2 ] ) that the form @xmath226 is contact on @xmath227 \times f$ ] if @xmath228 is large enough .
moreover , one can see @xmath229 as the extension of @xmath230 to the union @xmath231 once we do the reassignments @xmath232 finally , we set @xmath233
. then by the construction @xmath64 restricts to an exact symplectic structure on each fiber of @xmath20 .
so the contact form @xmath64 defines a contact structure @xmath234 on @xmath0 which is compatible with the fibration map @xmath105 .
+ in order to prove the second statement , observe that if @xmath235 , then for any transition map @xmath236 used in the above construction and for any @xmath237 we have @xmath238 near @xmath46 ( in particular , @xmath239 ) .
therefore , the smooth maps @xmath150 in the construction are all constant functions near @xmath46 , and so the corresponding functions @xmath153 take constant values near @xmath240 ( i.e. , @xmath241 near @xmath240 ) . as a result ,
the description of the contact form @xmath242 on @xmath0 given above takes the form @xmath243 near @xmath240 .
from the previous section any contact symplectic fibration admits a ( co - oriented ) contact structure .
here we study the flexibility of such contact structures . more specifically , we are interested in how they react if we change the contact structure on the base in its isotopy class .
before we proceed let us first fix some terminonology and notation .
let @xmath51 be a contact symplectic fibration with the contact base @xmath157 and the structure group @xmath58 .
the contact structure @xmath234 on @xmath0 which is defined by the contact form @xmath244 constructed as in the proof of theorem [ thm : existence_on_contact_symp_fib ] , is called a _
bundle contact structure for _
_ associated to _
@xmath154 , and will be denoted by @xmath245 .
note by definition bundle contact structures are compatible with the fibration maps .
next we show that isotopic contact structures on the base space produce isotopic bundle contact structures on the total space .
[ thm : isotopic_to_isotopic ] let @xmath51 be a contact symplectic fibration .
suppose co - oriented contact structures @xmath246 , @xmath247 , on @xmath1 are isotopic to each other
. then the corresponding bundle contact structures @xmath248 on @xmath0 are isotopic through bundle contact structures on @xmath0 .
in particular , @xmath249 and @xmath250 are contactomorphic
. since @xmath251 are isotopic , there exists a smooth family @xmath252\}$ ] of contact structures on @xmath1 joining them .
consider a smooth family @xmath253\}$ ] of contact forms on @xmath1 such that @xmath254 and @xmath255 for all @xmath256 .
since @xmath257 , we have @xmath258 for some smooth function @xmath259
. then we obtain the ( modified ) smooth family @xmath260\ } , \quad \mu_t:=\left(1-t+\dfrac{t}{h}\right)\alpha_t\ ] ] of contact forms on @xmath1 joining @xmath261 and @xmath262 and with the property @xmath263 for all @xmath256 .
+ consider the bundle contact structure @xmath264 on @xmath0 associated to @xmath265 given by the contact form @xmath266 constructed as in the proof of theorem [ thm : existence_on_contact_symp_fib ] . in order to simplify the notation
, we will write @xmath267 for @xmath268 .
let us consider @xmath269 \}$ ] .
for each @xmath256 the choice of @xmath267 can be done in such a way that @xmath267 s varies continuously with respect to @xmath256 .
this is possible since @xmath270\}$ ] is a smooth @xmath21-parameter family .
therefore , the set @xmath271 can be realized as the range set of a continuous function on the compact set @xmath272 $ ] , and hence there exists an upper bound , say @xmath273 , for @xmath271 .
we set @xmath274 , and consider three smooth @xmath21-parameter families : @xmath275\mu_0 + \beta + f_{r-1 } d \psi_{r-1 } , \quad t \in [ 0,1 ] , \\
\nonumber \lambda^2_t & : = & k\mu_t + \beta + f_{r-1 } d \psi_{r-1 } , \quad t \in [ 0,1],\\ \nonumber \lambda^3_t & : = & [ ( 1-t)k+tk_1]\mu_1 + \beta + f_{r-1 } d \psi_{r-1 } , \quad t \in [ 0,1].\end{aligned}\ ] ] as @xmath276\geq k_0 $ ] ( resp . @xmath277 and @xmath278\geq k_1 $ ] ) for each @xmath256 , the construction in theorem [ thm : existence_on_contact_symp_fib ] implies that @xmath279 ( resp .
@xmath280 and @xmath281 ) is a contact form on @xmath0 for any @xmath282 $ ] .
now we define a concatenation by the rule : @xmath283 \\ \lambda^2_{3t-1 } & \textrm{if } \quad t \in [ 1/3,2/3]\\ \lambda^3_{3t-2 } & \textrm{if }
\quad t \in [ 2/3,1 ] \end{cases}.\ ] ] observe that @xmath284 and @xmath285 , and so @xmath286\}$ ] is a smooth @xmath21-parameter family of contact forms on @xmath0 joining @xmath287 and @xmath178 . as a result
, we obtain a smooth @xmath21-parameter family @xmath288\}\ ] ] of ( bundle ) contact structures on @xmath0 joining @xmath289 and @xmath290 . hence , @xmath291 and @xmath292 are isotopic by gray s stability theorem ( see , for instance , theorem 2.2.2 of @xcite ) .
in this section we prove that the fiber connected summing of two contact symplectic fibrations along their fibers results in another contact symplectic fibration , and , in particular , the sum admits a ( compatible ) bundle contact structure and it fibers over the contact connected sum of the original contact bases .
we first recall the fiber connected sum operation in a slightly more general situation than the ones described in @xcite .
namely , we also allow the ambient manifolds to be noncompact or compact with boundary .
[ def : fiber_connected_sum ] for an oriented manifolds @xmath293 , let @xmath294 , be two codimension@xmath295 embeddings of an oriented manifold @xmath296 , and for each @xmath137 denote by @xmath297 the normal bundle of @xmath298 in @xmath299 ( which is an oriented @xmath300bundle over @xmath298 ) .
let @xmath301 be a fiber - orientation - reversing bundle isomorphism covering the composition @xmath302 .
after picking a riemannian metric and normalization , one can assume that @xmath303 is norm - preserving .
for an interval @xmath304 , where @xmath305 , define the subset @xmath306 .
furthermore , for a given @xmath161 consider the orientation - preserving diffeomorphism ( or identification ) @xmath307 given explicitly by the rule @xmath308 where @xmath309 denotes the fiber component of @xmath303 . then the _ fiber connected sum _
@xmath310 of @xmath311 , @xmath312 _ along _ their submanifolds @xmath313 is the oriented manifold given as the quotient space [ thm : fiber_conn_sum_along_convex_hyper ] let @xmath317 be two contact symplectic fibrations with @xmath318 .
for an exact symplectic manifold @xmath48 , assume that there exist embeddings @xmath319 , @xmath320 of the fibers @xmath321 and @xmath322 ( which are distinct if @xmath323 and @xmath324 ) of @xmath325 and @xmath326 , respectively , and for each @xmath327 the total space @xmath328 is constructed by regarding @xmath329 . then if @xmath330 is odd ( resp . even ) , then one can form the fiber connected sum @xmath0 of @xmath331 and @xmath332 ( resp . @xmath331 and @xmath333 ) along @xmath334 s using some canonical fiber - orientation - reversing bundle isomorhism @xmath301 between normal bundles of @xmath334 s such that @xmath335 ( resp .
@xmath336 ) admits a contact structure @xmath52 which restricts to a bundle contact structure on each piece @xmath337 ( resp . on @xmath338 and @xmath339 ) .
moreover , there exists a contact symplectic fibration @xmath340 where @xmath341 if @xmath330 is odd and @xmath342 if @xmath330 is even , such that @xmath343 ( resp .
@xmath344 ) is the contact connected sum of the original contact bases @xmath345 ( resp .
@xmath346 ) , @xmath90 , @xmath347 on their overlaping domains , and @xmath52 is a bundle contact structure for @xmath20 .
first note that each @xmath348 admits a contact form @xmath349 from which we can construct a contact form @xmath350 on @xmath328 using the pull - back form @xmath351 by taking @xmath352 large enough .
( see the proof of theorem [ thm : existence_on_contact_symp_fib ] ) .
also in the case when @xmath330 is even , since the orientation on @xmath353 is opposite of the one on @xmath354 , one should consider the contact form @xmath355 in order to define a positive contact structure on @xmath353 . by assumption
both @xmath331 and @xmath332 can be equipped with a fiber bundle structure such that structure groups @xmath356 .
consider the decomposition of each @xmath328 ( as in equation ( [ eqn : description ] ) ) given by @xmath357 where we consider the decomposition of each @xmath348 ( as in equation ( [ eqn : partition ] ) ) given by @xmath358 from differential topology point of view we are free to choose the fibers @xmath359 and @xmath360 along which the fiber connected sum is performed because different choices of fibers produces diffeomorphic total spaces which fibers over diffeomorphic bases .
therefore , by changing the fiber @xmath361 ( if necessary ) , we may assume that the embedding @xmath362 maps @xmath6 onto a fiber @xmath363 which lies in @xmath364 for some @xmath365 .
assume that each @xmath334 is the fiber over the point @xmath366 , that is , @xmath367 .
+ by darboux theorem ( see , for instance , theorem 2.5.1 of @xcite ) , for each @xmath138 there exist a neighborhood @xmath368 of @xmath369 which is contained in @xmath370 and coordinates @xmath371 ( where @xmath372 and @xmath373 ) ) on @xmath374 such that @xmath375 by a further change of coordinates on @xmath376 ( see example 2.1.3 of @xcite for details ) given by @xmath377 , for each @xmath138 we get the local description @xmath378 on @xmath374 . here
@xmath379 , @xmath380 ) , and @xmath381 are the polar coordinates on the @xmath382-plane .
+ now consider the tubular neighborhood @xmath383 of @xmath334 in @xmath328 , denote by @xmath384 the normal bundle of @xmath334 in @xmath328 , and let @xmath385 be the bundle projection .
the tubular neighborhood theorem implies that ( by taking a smaller @xmath374 if necessary ) we can write @xmath386 and @xmath387 .
then from the construction of @xmath350 , by taking @xmath374 small enough , one can guarantee that it is located far enough from @xmath388 , and so @xmath389 on @xmath390 . combining this with equation ( [ eqn : local_description ] )
, we obtain the local description @xmath391 where for simplicity ( as in the proof of theorem [ thm : existence_on_contact_symp_fib ] ) we still write @xmath392 ( resp .
@xmath43 ) for the pull - back form @xmath393 ( resp .
@xmath394 ) on @xmath390 .
we will form the fiber connected sum of @xmath331 , @xmath332 along @xmath395 using the ( canonical ) bundle isomorhism @xmath301 of the normal bundles @xmath396 which is given by observe that in both cases @xmath303 is fiber - orientation - reversing , and it covers the composition @xmath400 .
moreover , @xmath401 where @xmath402 denotes the radial coordinate on @xmath368 , and so with respect to the standard metric on @xmath403 , @xmath303 is norm preserving . in particular , for @xmath161 chosen sufficiently small , @xmath303 is a bijection between the subsets @xmath404 let @xmath405 be the orientation - preserving diffeomorphism ( obtained from @xmath303 ) corresponding to the identification ( given in ( [ eqn : identification ] ) ) constructing @xmath406 .
that is , @xmath407 where is the fiber component of @xmath303 .
consider the function @xmath410 which is used in defining the gluing map @xmath411 . since @xmath412 is the ( only ) fixed point of @xmath413
, @xmath411 identifies the trivial @xmath27-sphere bundles @xmath414 where @xmath415 is the @xmath27-sphere in @xmath416 of radius @xmath412 . for each @xmath138 , @xmath417 deformation retracts onto @xmath418 , and so up to diffeomorphism we may regard @xmath0 as where @xmath420 identifies the boundaries @xmath421 via the restriction map @xmath422 on @xmath423 , we compute @xmath424 this shows that if @xmath330 is odd , then @xmath425 , i.e. , @xmath426 is a contactomorphism , and so by gluing @xmath350 s along @xmath418 s we obtain a contact form , say @xmath64 , on @xmath427 .
if @xmath330 is even , the above computation suggests that by switching the orientations on @xmath354 ( and so on @xmath332 ) and assuming the ( positive ) contact form @xmath209 on @xmath333 is constructed using the positive contact form @xmath428 on @xmath353 , we can make @xmath426 a contactomorphism ( because we will have @xmath425 for this case as well ) , and therefore , one can glue @xmath350 s along @xmath418 s and obtain a contact form @xmath64 on @xmath429 . + by construction the contact structure @xmath430 on @xmath0 restricts to the original bundle contact structure on each piece @xmath337 ( if @xmath330 is odd ) , or on each piece @xmath338 and @xmath339 ( if @xmath330 is even ) .
moreover , in both cases during the gluing process ( along @xmath418 s ) only the base coordinates ( that is , the coordinates on @xmath431 s ) are nontrivially identified , and so one can smoothly glue the maps @xmath432 together , and obtain a fiber bundle projection @xmath20 on @xmath0 with fibers @xmath6 . on the other hand , the gluing map @xmath426 is identity on each fiber @xmath6 over @xmath180 , and , therefore , the structure group @xmath82 of @xmath20 is contained in @xmath83 . + next observe that at the base level , @xmath433 pulls back @xmath392 to itself ( if @xmath330 is odd ) , and to @xmath434 ( if @xmath330 is even ) .
thus , constructing the total space @xmath0 upstairs corresponds to taking the contact connected sum of @xmath435 s ( if @xmath330 is odd ) , or of @xmath436 with @xmath437 ( if @xmath330 is even ) downstairs as claimed .
+ finally , consider the reassignments @xmath438 where @xmath439 .
then using these new choices of @xmath440 we can carry out all the constructions above in the same way , and corresponding conclusion are still be correct .
for instance , in the case that @xmath330 is odd , one can see for each @xmath137 that @xmath441 and @xmath442 define isotopic ( constant isotopy ) contact structures on @xmath443 , and so , theorem [ thm : isotopic_to_isotopic ] , the bundle contact structures @xmath444 on @xmath445 are isotopic .
hence , the corresponding contact structures @xmath52 s ( kernels of @xmath64 s ) on @xmath427 constructed above are isotopic .
+ now it is straight forward to check that when @xmath330 is odd ( resp .
@xmath330 is even ) the contact structure @xmath446 on @xmath427 ( resp .
@xmath429 ) is the bundle contact structure for @xmath20 associated to @xmath447 ( resp . | we consider certain type of fiber bundles with odd dimensional compact contact base , exact symplectic fibers , and the structure group contained in the group of exact symplectomorphisms of the fiber .
we call such fibrations `` contact symplectic fibrations '' . by a result of hajduk - walczak , some of these admit contact structures which are `` compatible '' ( in a certain sense ) with the corresponding fibration structures .
we show that this result can be extended to get a compatible contact structure on any contact symplectic fibration , and also that isotopic contact structures on the base produce isotopic contact structures on the total space .
moreover , we prove that the fiber connected summing of two contact symplectic fibrations along their fibers results in another contact symplectic fibration which admits a compatible contact structure agreeing with the original ones away from the region where we take fiber connected sum , and whose base is the contact connected sum of the original contact bases . |
statistical mechanics @xcite is a theory developed at the end of nineteenth century to deal with physical systems from an atomistic point of view . in principle
the properties of bulk matter , which may contain @xmath0 atoms , can be worked out from the motion of atoms following the basic equations of newtonian mechanics or quantum mechanics . however , such detailed information is not available or not really necessary .
a probabilistic point of view , in the form of statistical ensemble is taken .
the theory is extremely economical and successful in dealing with _ equilibrium _ problems .
there are a number of equivalent formulations of the theory .
but statistical mechanics in a nutshell is the following concise formula that connects statistical mechanics with thermodynamics .
first the partition function is defined as @xmath1 where @xmath2 is a `` state '' of the system ; the summation is carried over all possible states .
@xmath3 is absolute temperature and @xmath4 is the boltzmann constant ( @xmath5 joule / kelvin ) .
the free energy of the system at temperature @xmath3 is given by @xmath6 free energy is a useful thermodynamic quantity in dealing with phase transitions .
other macroscopic observables are averages of the corresponding microscopic quantities over the boltzmann - gibbs weight @xmath7 , @xmath8 the following three quantities are perhaps the most important ones , internal energy @xmath9 , heat capacity @xmath10 , and entropy @xmath11 , where the average is over the gibbs probability density , @xmath12 .
the essential task in statistical mechanics is to do the multi - dimensional integrations ( for continuous systems ) or summations ( for discrete systems ) . to fix the notation and language
, we briefly introduce the basics of monte carlo method @xcite .
consider the computation of a statistical average , @xmath13 where the probability density obeys @xmath14 , and @xmath15 .
suppose that we can generate samples @xmath16 according to the probability @xmath17 , then the integral can be estimated from an arithmetic mean over the samples , @xmath18 the random variable @xmath19 has mean @xmath20 and standard deviation @xmath21 , @xmath22 being decorrelation time .
thus in the limit of large number of samples , the estimate converges to the exact value .
the most general method to generate @xmath2 according to @xmath17 is given by a markov chain @xcite with a transition probability @xmath23 , satisfying the conditions stated below .
this is the probability of generating a new state @xmath24 given that the current state is @xmath2 .
such process converges if @xmath25 the equilibrium distribution of the markov chain satisfies @xmath26 in constructing a monte carlo algorithm , it is convenient to consider a much stronger condition , the detailed balance @xmath27 one of the most famous and widely used monte carlo algorithms is the metropolis importance sampling algorithm @xcite .
it takes a simple choice of the transition matrix : @xmath28 where @xmath29 , and @xmath30 is a conditional probability of choosing @xmath24 given that the current value is @xmath2 , and it is symmetric , @xmath31 . usually @xmath32 unless @xmath24 is in some `` neighborhood '' of @xmath2 .
the diagonal term of @xmath33 is fixed by the normalization condition @xmath34 .
in order to facilitate the discussion we first introduce the ising model .
ising model is an interacting many - particle model for magnets .
a state consists of a collection of variables @xmath35 taken on only two possible values @xmath36 and @xmath37 , signifying the spin up and spin down states .
the spins are on a lattice .
the energy of the state @xmath38 is given by @xmath39 where @xmath40 is a constant which fixes the energy scale , the summation is over the nearest neighbor pairs .
when temperature @xmath3 is specified , the states are distributed according to @xmath41 in a local monte carlo dynamics ( metropolis algorithm ) , one picks a site @xmath42 at random , i.e. , choosing a site with equal probability ( this specifies or realizes @xmath30 ) .
then , the energy increment if the spin is flipped , @xmath43 , is computed , which has the result @xmath44 .
the flip is accepted with probability @xmath45 . if the flip is rejected , the move is also counted and the state remains unchanged .
one monte carlo step is defined as @xmath46 moves ( trials ) for a system of @xmath46 spins , such that each spin is attempted to flip once on average .
the local algorithm of metropolis type has some salient features : ( 1 ) it is extremely general .
less assumption is made of the specific form of probability distribution .
( 2 ) each move involves @xmath47 operations and @xmath47 degrees of freedom .
( 3 ) the dynamics suffers from critical slowing down .
the correlation time @xmath22 diverges as a critical temperature is approached .
we shall elaborate more on this in the following .
the statistical error , using estimator eq . ( [ estimator ] ) , is given by @xmath48 where @xmath49 is the variance of the observable , @xmath50 is the number of monte carlo steps , and @xmath22 is the decorrelation time .
we can take the point of view that the above equation defines @xmath22 , i.e. , @xmath51 perhaps it is appropriate to call @xmath22 decorrelation time , since @xmath52 is sometimes called correlation time in the literature .
the decorrelation time @xmath22 is the minimum number of monte carlo steps needed to generate effectively independent , identically distributed samples in the markov chain .
the smallest possible value for @xmath22 is 1 , which represents independent sample by every step .
the usual integrated autocorrelation time @xcite differs from our definition by a factor of 2 , @xmath53 .
the critical slowing down manifests itself by the fact that @xmath54 at the critical temperature @xmath55 where a second order phase transition occurs . here
@xmath56 is the linear dimension of the system ( @xmath57 in @xmath58 dimensions ) . for the local algorithms for many models and in any dimensions , @xmath59
this suggests bad convergence , specially for large systems . at a first - order phase transition , where some thermodynamic variables change discontinuously ,
the situation is even worse @xmath22 diverges exponentially with system sizes . for the two - dimensional ising model , a phase transition occurs @xcite at @xmath60 .
the magnetization is non - zero below this temperature and becomes zero above @xmath55 .
in addition , in the limit of large system ( @xmath61 ) , heat capacity per spin and fluctuation of magnetization diverge .
these intrinsic properties make computer simulation near critical point very difficult .
cluster algorithms @xcite overcome this difficulty successfully .
for example , for the two - dimensional ising model , the dynamical critical exponent defined in @xmath62 is reduced from 2.17 for the single - spin flip @xcite to much small value for the cluster algorithm @xcite .
it turns out that a precise characterization of the swendsen - wang dynamical critical exponent is very difficult , due to weak size dependence of the decorrelation time .
table 1 represents a recent extensive calculation , based on the definition eq .
( [ tau - def ] ) for the total energy , rather than the usual method of extracting information from time - dependent auto - correlation functions . in this calculation ,
the variance of the sum of @xmath50 consecutive energies ( c.f . eq .
( [ estimator ] ) ) are computed explicitly .
good convergence to the limiting value is already achieved typically for @xmath63 .
an @xmath64 extrapolation is used to get more accurate estimates of the limit . from this calculation
, we find for the two - dimensional ising model , the convergence to asymptotics are slow .
it appears that the divergence is slightly faster than @xmath65 .
if we fit to power law @xmath66 , using two consecutive sizes of @xmath56 and @xmath67 , the exponent @xmath68 decreases from 0.35 to 0.21 .
it is not clear whether @xmath68 will convergence to a finite value or continue decreasing to 0 in the limit of @xmath61 .
the nonlocal cluster algorithm introduced by swendsen and wang for the the ising model ( and more generally the potts model ) goes as follows : ( 1 ) go over each nearest neighbor pair and create a bond with probability @xmath69 that is , if the two nearest neighbor spins are the same , a bond is created between them with probability @xmath70 ; if spin values are different , there will be no bond .
( 2 ) identify clusters as a set of sites connected by zero or more bonds ( i.e. , connected component of a graph ) .
relabel each cluster with a fresh new value @xmath36 or @xmath37 at random .
we note that each monte carlo step per spin still takes @xmath47 in computational cost .
the method is applicable to models containing ising symmetry , i.e. , the energy is the same when @xmath35 is changed to @xmath71 globally .
the algorithm is based on a mapping of ising model to a random cluster model of percolation . specifically , we have @xcite @xmath72 \\ & = & \sum_{\{n\ } } p^b ( 1 - p)^{nd - b } 2^{n_{\mathrm{c } } } , \end{aligned}\ ] ] where
@xmath73 , @xmath74 , @xmath75 is the kronecker delta , and @xmath76 is the number of bonds , @xmath58 is the dimension of a simple hypercubic lattice , and @xmath77 is the number of clusters .
@xmath35 is spin on the site and @xmath78 is the bond variable between the sites @xmath42 and @xmath79 .
it is evident that the moves in the swendsen - wang algorithm preserves configuration probability of the augmented model containing both the spins and bonds .
a single cluster variant due to wolff @xcite is very easy to program .
the following c code generates and flip one cluster : .... # define z 4 # define l 8 double p = 0.7 ; int s[l*l ] ; void flip(int i , int s0 ) { int j , nn[z ] ; s[i ] = - s0 ; neighbor(i , nn ) ; for(j = 0 ; j < z ; + + j ) if(s0 = = s[nn[j ] ] & & drand48 ( ) < p ) flip(nn[j ] , s0 ) ; } .... in this single cluster version , a site @xmath42 is selected at random .
the value of the spin before flip is @xmath80 .
it flips the spin and looks for its neighbors .
if the value of the neighbor spins are the same as @xmath80 , with probability @xmath81 a neighbor site becomes part of the cluster .
this is performed recursively .
it turns out that single cluster is somewhat more efficient than the original swendsen - wang , particularly in high dimensions . for dimensions greater than or equal to 4 ,
swendsen - wang dynamics gives the dynamic critical exponent @xmath82 , while wolff single cluster is @xmath83 @xcite .
it is easy to see why the single cluster algorithm of the above works .
let us consider a general cluster flip algorithm with two bond probabilities : @xmath84 will be the probability of connecting two parallel spin sites ; @xmath85 the probability of connecting anti - parallel sites . consider the transition between two configurations @xmath86 and @xmath87 , characterized by flipping a cluster @xmath88 .
a cluster is growing from a site @xmath42 until the perimeter of the cluster is not connected to the outside .
the transition probabilities can be written down as @xmath89 note that the bond configuration probabilities are the same in the interior of the clusters .
the difference occurs at the boundary @xmath90 , where parallel spins ( @xmath91 ) in @xmath86 becomes anti - parallel spins ( @xmath92 ) in @xmath87 , or vice versa .
detailed balance requires that @xmath93 where @xmath94 is the number of parallel spins on boundary of @xmath88 in configuration @xmath86 ; and @xmath95 is the number of anti - parallel spins on boundary of @xmath88 in configuration @xmath86 .
since we have @xmath96 , we obtain @xmath97 although the algorithm is valid for any @xmath98 , it is most efficient at @xmath99 , the coniglio - klein @xcite bond probability value
. a quite large number of statistical models can be treated with cluster algorithms , with varied success .
excellent performance has been obtained for ising model , potts models , and antiferromagnetic potts models @xcite , xy model and general @xmath100 models @xcite , @xmath101 field - theoretic model @xcite , some regularly frustrated models @xcite , six - vertex model @xcite , etc .
cluster algorithms are proposed for hard sphere fluid systems @xcite , quantum systems @xcite , microcanonical ensembles @xcite , conserved order parameters @xcite , etc . invaded cluster algorithm @xcite and other proposal @xcite
are excellent methods for locating critical points .
the cluster algorithms are also used in image processing @xcite .
the cluster algorithms do not help much in temperature - driven first - order phase transition the slow convergence has been shown rigorously @xcite .
models with frustration , spin glass being the archetype , do not have efficient cluster algorithms , although there are attempts @xcite with limited success .
breakthrough in this area will have a major impact on the simulation methods .
in this and the following sections , we discuss a class of monte carlo simulation approaches that aim at an efficient use of data collected , and sampling methods that enhance rare events .
the computation of free energy , eq . ( [ free - energy ] ) , poses a difficult problem for monte carlo method .
a traditional method is to use thermodynamic integration , e.g. , @xmath102 based on the relation @xmath103 , where @xmath104 .
if we can estimate the density of states ( the number of states with a given energy @xmath105 for discrete energy models ) , then we can compute free energy , as well as thermodynamic averages .
the result is obtained as a function of temperature @xmath3 , rather than a single datum point for a specific value of @xmath3 , as in standard monte carlo simulation .
this idea has been pursued over the last decade by ferrenberg and swendsen @xcite , berg et al @xcite , lee @xcite , oliveira et al @xcite , and wang @xcite .
consider the following decomposition of summations over the states @xmath106 where @xmath107 is the microcanonical ensemble average , @xmath108 since the state space is exponentially large ( @xmath109 for the ising model with @xmath46 spins ) , and the range of @xmath105 is typically of order @xmath46 , if @xmath110 can be computed accurately , the task is done .
the canonical average of @xmath86 is related to the microcanonical average through @xmath111 and free energy is computed as @xmath112 ferrenberg and swendsen @xcite popularized a method which in a sense is to compute the density of states ( up to a multiplicative constant ) in a range close to a given simulation temperature .
this method is generalized as multiple histogram method to combine simulations at differential temperatures , to get the whole energy range @xcite .
we discuss here only the single histogram method for its simplicity . during a normal canonical simulation at fixed temperature @xmath113
, we collect the histogram of energy , @xmath114 , which is proportional to probability distribution of energy , @xmath115 the constant @xmath116 is related to the partition function , @xmath117 , where @xmath50 is the total number of samples collected . from the above equation , we find @xmath118 . with this information , we can compute the free energy difference between temperature @xmath113 and a nearby temperature @xmath3 .
similarly , moments of energy can be computed after the simulation , through histogram reweighting , @xmath119 the range of @xmath105 that the histogram data can be collected at a fixed temperature is limited by the energy distribution , which for the canonical distribution away from critical point , is of order of @xmath120 .
the whole range of energy @xmath105 is of order @xmath46 .
this limit the usefulness of single histogram method .
the multicanonical monte carlo method has been shown to be very effective to overcome supercritical slowing down , reducing the relaxation time from exponential divergence with respect to system size to a power , at the first - order phase transitions @xcite .
multicanonical ensemble flattens out the energy distribution , so that the computation of the density of states @xmath110 can be done for all values of @xmath105 .
a multicanonical ensemble is defined to following the probability density for the states as @xmath121 such that the energy histogram @xmath114 ( @xmath122 ) is a constant . from the histogram samples obtained by a simulation with the weight of state at energy @xmath105 as @xmath123 , the density of state
can be computed @xcite from @xmath124 . however , unlike canonical simulation where @xmath125 is given , in a multicanonical simulation , @xmath123 is unknown to start with .
berg proposed an iterative method to compute the weight in a parametrized form @xmath126 , starting with no information , @xmath127 . a new estimate at iteration @xmath128
is then based on the results of all previous iterations .
we refer to references @xcite for details .
the flat histogram algorithm offers an efficient bootstrap to realize the multicanonical ensemble , while transition matrix monte carlo utilizes more data that can be collected in a simulation to improve statistics .
we start from the detailed balance equation for some given dynamics : @xmath129 by summation over the states @xmath130 of fixed energy @xmath105 , and @xmath131 of fixed energy @xmath132 , and assuming that the probability of the state is a function of energy only , @xmath133 , we get @xmath134 where the transition matrix in the space of energy is defined as @xmath135 the matrix @xmath3 has a number of interesting properties : it is a stochastic matrix in the sense of @xmath136 and @xmath137 ; the stationary solution of @xmath3 is the energy distribution @xmath138 ; the dynamics associated with @xmath3 is considerably faster than that of @xmath33 @xcite .
we specialize to the case of single - spin - flip dynamics for the ising model .
the transition matrix @xmath33 for the spin states consists of a product of two factors , the probability of choosing a spin to flip @xmath139 , and the flip rate @xmath140 .
we have @xmath141 less the two configurations @xmath130 and @xmath131 differ by one spin , in this case , the value of @xmath30 is @xmath142 , where @xmath46 is the number of spins in the system . using these results , we can rewrite the transition matrix as @xmath143 the diagonal elements are determined by normalization . substituting eq.([t - ssf ] ) into eq.([t - balance ] ) , using the relation between @xmath144 and @xmath123 , we obtain @xmath145 this is known as broad histogram equation @xcite which forms the basis for the flat histogram algorithm presented below . additionally , this equation also gives us a way of computing the density of states @xmath110 by the quantity @xmath146 obtained from spin configurations generated from any distribution @xmath147 .
the quantity @xmath148 is the number of ways that the system goes to a state with energy @xmath149 , by a single spin flip from state @xmath130 .
the angular brackets indicate a microcanonical average : @xmath150 the following algorithm @xcite generates flat histogram in energy and realizes the multicanonical ensemble . 1 .
pick a site at random .
2 . flip the spin with probability @xmath151 where the current state @xmath130 has energy @xmath105 , the new state @xmath131 has energy @xmath152 .
3 . accumulate statistics for @xmath153 .
4 . go to 1 .
we note by virtue of eq .
( [ broad - histo - eq ] ) , the flip rate is the same as that in multicanonical simulation with a weight @xmath154 and metropolis acceptance rate .
while in multicanonical sampling , the weight is obtained through several simulations iteratively , the quantities @xmath153 is much easier to obtain , through a single simulation .
this quantity serves a dual purpose
it is used to construct a monte carlo algorithm ( used as input ) , and at the same time , it is used to compute the density of states ( output of the simulation ) .
clearly , this is circular unless approximation is made .
we have considered replacing the exact microcanonical average by an accumulative average , over the history of simulation generated so far , i.e. , @xmath155 where @xmath156 is the sequence of states generated with the algorithm given above ; @xmath114 is the number of samples accumulated at the energy bin @xmath105 . in case the data for computing the flip rate
is not available , we simply accept the move to explore the new state . a more rigorous way of doing simulation is to iterate the simulation with constant flip rate .
for example , after the first simulation , we compute a first estimate to the density of states . in a second simulation , we perform multicanonical simulation la lee @xcite . the data collected in the second run for @xmath157 will be unbiased .
it is found that even with a single simulation , the results converge to the exact values ( for @xmath153 and @xmath110 ) for sufficiently long runs , even though a rigorous mathematical proof of the convergence is lacking .
wang and landau @xcite proposed recently a new algorithm that works directly with the density of states @xmath110 .
the simulation proceeds with the flip rate @xmath158 , but the value of the density of states is updated after every move by @xmath159 and letting @xmath160 for convergence .
excellent results were obtained .
a careful comparison with flat histogram method is needed . in metropolis algorithm ,
moves are sometimes rejected .
this rejection is important for realizing the correct stationary distribution . in 1975
bortz , kalos , and lebowitz @xcite proposed a rejection - free algorithm .
it is still based on metropolis flip rate , but the waiting due to rejection is taking into account by considering all possible moves .
the bortz - kalos - lebowitz n - fold way algorithm for the ising model goes as follows : 1 .
compute the acceptance probability for one attempt of a move @xmath161 2 .
pick an energy change @xmath162 according to probability @xmath163 flip a site belonging to @xmath162 with probability 1 .
the site is choosing from the @xmath148 sites with equal probability .
one n - fold - way move is equivalent to @xmath164 moves in the original dynamics .
thus thermodynamic averages are weighted by @xmath164 , i.e. , @xmath165 , where summation is over every move @xmath166 . in order to implement step 2 efficiently ,
additional data structure is needed so that picking a spin in a given class characterized by @xmath162 is done in @xmath47 in computer time .
combining n - fold way and flat histogram algorithm is easy , since the important quantity @xmath148 is already computed in flat histogram algorithm .
the flip rate @xmath167 is given by formula ( [ flat - histo - rate ] ) . in the flat histogram algorithm , the probability that the energy of the system is @xmath105 is a constant , i.e. , @xmath168 the averages in the second line of the above equation refer to samples generated in an n - fold way simulation .
in equal - hit algorithm ( ensemble ) @xcite , we require that the number of `` fresh configurations '' generated at each energy is a constant .
more precisely equal - hit ensemble is _ defined _ by @xmath169 one possible choice of the flip rate is @xmath170 where @xmath171 is the inverse total acceptance rate @xmath164 arithmetic averaged over the n - fold way samples at energy @xmath105 .
the histogram @xmath114 generated in the equal - hit algorithm depends on the precise dynamics ( the rate @xmath167 ) used .
since there are many possible choices of the rate , such `` equal - hit ensemble '' is not unique . while eq .
( [ broad - histo - eq ] ) gives us a way of obtaining the density of states , there are more equations than unknowns .
we consider two optimization methods .
the first method is based on the transition matrix itself .
we define @xmath172 .
symbols with hat being unknown , and @xmath173 the monte carlo estimate , consider @xmath174 subject to @xmath175 , @xmath176 , and @xmath177 . the last constraint needs more explanation .
we assume that the energy level is equally spaced ( as in the ising model ) .
consider three energy levels , @xmath105 , @xmath178 , @xmath179 . if we write down three equations of type ( [ broad - histo - eq ] ) , for transitions from @xmath105 to @xmath178 , @xmath178 to @xmath179 , and @xmath179 to @xmath105 , we can cancel the density of states by multiplying the three equations together .
this leaves the last equation above , and it is known as ttt identity .
it can be shown that multiple t identities ( four or more ) are not independent , and they need not put in as constraints . for ising model there is also one additional symmetry constraint , @xmath180 .
when the solution for @xmath3 is found , we can use any of the energy detailed balance equation to find density of states @xmath110 .
the ttt identity guarantees that the answer is unique whichever detailed balance equation is used .
the second method is based on optimization directly with variable @xmath110 , actually @xmath181 , by @xmath182 subject to , for the ising model , @xmath183 , where @xmath46 is the total number of spins in the system .
in addition , we can put in the known fact that the ground states are doubly degenerate , @xmath184 .
, calculated by transition matrix monte carlo method for a @xmath185 two - dimensional ising model with a total @xmath186 monte carlo steps and the first @xmath187 discarded , using the flat - histogram algorithm and n - fold way .
the insert is error in @xmath188 with respect to exact results . ]
figure 1 shows one of the simulation results of the density of states , using the second method .
the errors by comparison with exact known values @xcite are presented in the insert of the figure .
the density of states @xmath110 is determined to an accuracy of better than 2 percents in a matter of few minutes of computer time .
the flat - histogram dynamics is used to study spin glasses @xcite .
the dynamic characteristics is quite similar to multi - canonical method of berg .
the study of lattice polymer and protein folding is under way . for related ideas and approaches ,
see refs.@xcite .
the phenomenon of critical slowing down can be effectively dealt with by cluster algorithms for a large class of statistical mechanics models .
we reported new and very accurate results for the decorrelation times of the swendsen - wang dynamics .
large size asymptotic behavior is analyzed . for super - critical slowing down occurring in first - order phase transitions ,
multicanonical ensemble simulation and flat - histogram or equal - hit algorithms are very effective .
since the latter algorithms and associated transition matrix method is efficient in computing the density of states , this method can also be useful for general counting problems by monte carlo method .
the author thanks prof .
r. h. swendsen for many of the work discussed here .
he also thank z. f. zhan , t. k. tay , and l. w. lee for collaborations .
this work is supported by a nus research grant r151 - 000 - 009 - 112 and singapore - mit alliance .
sokal , a. d. : in `` computer simulation studies in condensed matter physics : recent developments '' , eds : d. p. landau , k. k. mon , and h .- b .
schttler , springer proceedings in physics , vol * 33 * , ( 1988 ) 6 . | the basic problem in equilibrium statistical mechanics is to compute phase space average , in which monte carlo method plays a very important role .
we begin with a review of nonlocal algorithms for markov chain monte carlo simulation in statistical physics .
we discuss their advantages , applications , and some challenge problems which are still awaiting for better solutions .
we discuss some of the recent development in simulation where reweighting is used , such as histogram methods and multicanonical method .
we then discuss the transition matrix monte carlo method and associated algorithms .
the transition matrix method offers an efficient way to compute the density of states .
thus entropy and free energy , as well as the usual thermodynamic averages , are obtained as functions of model parameter ( e.g. temperature ) in a single run .
new sampling algorithms , such as the flat histogram algorithm and equal - hit algorithm , offer sampling techniques which generate uniform probability distribution for some chosen macroscopic variable . |
the recent discovery of superconductivity around 30 k in alkaline iron selenides @xmath7fe@xmath1se@xmath2 ( @xmath8 k , rb , cs , tl ) @xcite has generated considerable excitement in the condensed matter physics community because the parent compounds of these materials are antiferromagnetic ( afm ) insulators @xcite instead of being afm metals as the iron arsenide superconductors @xcite .
because of their metallic nature , band structure calculations for iron arsenides have predicted the presence of the hole - like fermi surfaces at the @xmath9 point and electron - like fermi surfaces at the @xmath10 points in the brioullion zone using an orthorhombic full lattice unit cell @xcite . as a consequence
, fermi surface nesting and quasiparticle excitations between the hole and electron pockets can give rise to static afm spin - density - wave order at the in - plane wave vector @xmath11 @xcite .
indeed , neutron diffraction experiments have confirmed the @xmath12 afm order in the parent compounds of iron arsenide superconductors , and doping to induce superconductivity suppresses the static afm order @xcite .
in addition , angle resolved photoemission ( arpes ) measurements @xcite have identified the expected hole and electron pockets in superconducting iron arsenides , thus providing evidence for superconductivity arising from the sign reversed electron - hole inter - pocket excitations @xcite .
if fermi surface nesting and electron - hole pocket excitations are essential ingredients for magnetism and superconductivity in fe - based superconductors @xcite , alkaline iron selenide superconductors should behave differently from iron arsenides since arpes measurements on these materials reveal only electron fermi surfaces at @xmath10 points and no hole fermi surface at @xmath9 point @xcite .
indeed , recent transmission electron microscopy @xcite , x - ray and neutron diffraction experiments @xcite have confirmed that the fe vacancies in @xmath7fe@xmath1se@xmath2 form a @xmath4 superlattice order as shown in fig .
1(a ) @xcite .
furthermore , a block - type afm structure with a large moment aligned along the @xmath13-axis [ figs .
( 1a ) and ( 1b ) ] has been proposed for both superconducting and insulating @xmath7fe@xmath1se@xmath2 based on rietveld analysis of neutron powder diffraction data @xcite .
in stark contrast to other fe - based superconductors , where optimal superconductivity generally occurs in the absence of a static afm order @xcite , the large moment afm order is believed to co - exist with superconductivity microscopically @xcite and the superconducting phase develops without much affecting the afm order @xcite . if magnetic moments up to 3.3 @xmath14 per fe indeed coexist with optimal superconductivity microscopically in @xmath7fe@xmath1se@xmath2 as suggested in powder neutron diffraction @xcite and muon rotation experiments @xcite , the electronic phase diagram in this class of materials will be much different than the other fe - based superconductors @xcite .
since these new materials pose a major challenge to the current theories of superconductivity @xcite , it is important to confirm the proposed magnetic structure in single crystals and determine its relationship with superconductivity . in this article
, we present comprehensive neutron diffraction measurements on powder and single crystals of nonsuperconducting and superconducting rb@xmath3fe@xmath1se@xmath15 .
we used neutron polarization analysis to separate the magnetic from nuclear scattering . from the rietveld analysis of the neutron powder diffraction data on nonsuperconducting rb@xmath16fe@xmath17se@xmath2 @xcite
, we confirm the previously reported @xmath4 fe vacancy order with @xmath18 space group @xcite . since rietveld analysis of the powder diffraction pattern can not conclusively separate the proposed block afm structure from the quaternary collinear afm structure with the @xmath19 space group @xcite , we used four circle single crystal diffractometer to measure bragg peaks associated with each afm structure , and confirmed the proposed block afm structure @xcite . for superconducting rb@xmath20fe@xmath21se@xmath15 ( @xmath22 k ) , we find that in addition to the @xmath4 block afm structure , the sample exhibits a quasi - two - dimensional @xmath6 superlattice distortion associated with wave vectors @xmath23 , where @xmath24 integers .
these results suggest that lattice structures in superconducting rb@xmath3fe@xmath1se@xmath15 are more complicated than the pure @xmath4 superlattice unit cell , consistent with nanoscale phase separation seen by transmission electron microscopy @xcite and x - ray diffraction experiments @xcite .
-axis , respectively .
the solid and dashed squares are the top views of the tetragonal cell ( red ) and the @xmath4 superlattice cell ( blue ) , respectively .
( c ) the expected bragg peak positions in the @xmath25 $ ] tetragonal unit cell notation for the left chirality .
the red circles with @xmath26 are magnetic peaks and the black circles are nuclear peaks with @xmath27 .
the brown squares are nuclear peaks for both even and odd @xmath28 . the green star at ( 0.5,0.5 )
rlu is not an expected bragg peak position for the present structure .
the inset is a schematic of the tetragonal and superlattice unit cell in reciprocal space .
( d ) the expected bragg peaks from both left and right chiralities . ( e , f ) the same expected bragg peaks in @xmath29 $ ] superlattice unit cell notation . ]
we have carried out neutron diffraction experiments at the bt-1 powder diffractometer and bt-7 thermal triple - axis spectrometer at the national institute for standard and technology center for neutron research .
we have also performed additional measurements at hb-1a triple - axis spectrometer and hb-3a four circle single crystal diffractometer at the high - flux isotope reactor , oak ridge national laboratory .
our experimental setup for the bt-1 powder diffraction measurements was described previously @xcite . for bt-7 measurements , we used polarized neutron scattering to separate the magnetic from nonmagnetic scattering processes @xcite . in previous powder diffraction measurements on @xmath7fe@xmath1se@xmath2 near @xmath30 @xcite
, the iron atoms were found to form an ordered vacancy structure with a @xmath31 superlattice unit cell .
although a block afm spin structure with space group @xmath32 [ fig . 1(a ) and fig .
1(b ) ] was identified @xcite , powder rietveld analysis can not conclusively distinguish the block afm structure from a stripe - like afm structure with @xmath19 space group [ fig .
4(b ) ] @xcite .
we have therefore used the hb-3a single crystal diffractometer to measure all the accessible bragg peaks , and including the nonequivalent magnetic reflections with the same momentum transfer that are fully overlapped in the powder diffraction experiments , thus providing more information to separate these two magnetic structures .
hb-3a uses a vertically focusing si(2,2,0 ) monochromator with fixed wavelength of 1.536 @xcite .
the hb-1a triple - axis spectrometer has horizontal collimation @xmath33 with fixed incident beam energy of @xmath34 mev .
fe@xmath17se@xmath2 at 480 k and its comparison with the expected neutron powder patterns for various proposed magnetic structures of @xmath7fe@xmath1se@xmath2 @xcite .
the plus and minus signs denote spin directions parallel and antiparallel to the @xmath13-axis , respectively .
the data and model fitted difference plots are shown below .
the weighted @xmath35 factors for each model are listed as follows : ( a ) ferromagnetic or fm .
( b ) checkerboard afm .
( c ) collinear afm .
( d ) bi - collinear afm .
( e ) cell - in - collinear afm .
( f ) cell - in - collinear afm .
none of the models can fit the observed neutron powder diffraction pattern . ]
we begin our discussion by specifying the real and reciprocal space notations used in this article .
figures 1(a ) and 1(b ) show the left and right chiralities of the proposed block afm structures , respectively .
the blue dashed lines show the structural and magnetic unit cells for the @xmath4 fe vacancy structure , while the red solid lines are the @xmath42 symmetry tetragonal unit cell suitable for doped bafe@xmath2as@xmath2 @xcite .
the @xmath43 signs indicate the fe moment directions parallel and anti - parallel to the @xmath13-axis , respectively . for easy comparison with previous work in iron pnictides ,
we define wave vector @xmath44 in @xmath45 as @xmath46 reciprocal lattice units ( rlu ) , where @xmath47 are lattice parameters for tetragonal unit cell of iron pnictides @xcite .
the bragg peaks in the @xmath4 superlattice unit cell can be indexed as @xmath48 rlu , where @xmath49 and @xmath50 for the nonsuperconducting rb@xmath16fe@xmath17se@xmath2 , @xmath51 and @xmath52 for the superconducting rb@xmath20fe@xmath21se@xmath15 .
the rb , fe , and se compositions are determined from inductively coupled plasma atomic emission spectroscopy analysis .
fe@xmath17se@xmath2 at 480 k. ( a ) quaternary collinear afm structure and ( b ) block afm structure .
the @xmath53 factors for these two magnetic structures are 7.0% and 6.95% , respectively .
both of these models were proposed earlier @xcite . ] in iron pnictides , the fe moments are in the fe plane along the @xmath54-axis direction @xcite .
the magnetic bragg peaks occur at @xmath55_t$ ] ( @xmath56 ) positions in tetragonal unit cell notation . for @xmath7fe@xmath1se@xmath2 with the block afm structure in figure 1 , the magnetic peaks from left chirality are expected at @xmath57 when @xmath58 is even , @xmath59 @xmath58 is odd , @xmath60 as shown in figs .
1(c ) and 1(e ) .
if one considers both left and right chiralities , the magnetic peaks will double and occur at @xmath61 when @xmath58 is even , @xmath59 @xmath58 is odd , @xmath60 .
the nuclear bragg reflections can be indexed the same way , but with @xmath62 and when @xmath58 is even , @xmath63 ; @xmath58 is odd , @xmath64 .
the squares in figs .
1(c)-1(f ) indicate nuclear bragg peak positions in tetragonal and superlattice unit cell notation .
the conversion of miller indices between tetragonal and superlattice unit cell for left chirality is as follows : @xmath65 for right chirality , the conversion is : @xmath66 our single crystals of rb@xmath0fe@xmath1se@xmath2 were grown using the bridgeman method .
first , fe@xmath67se@xmath2 powders were prepared with a high - purity powder of selenium ( alfa aesar , 99.99% ) and iron ( alfa aesar , 99.9% ) as described in ref .
the fe@xmath67se@xmath2 and rb ( alfa aesar , 99.75% ) were then mixed in appropriate stoichiometry and were put into an alumina crucible .
the crucible was sealed in an evacuated silica ampoule .
the mixture was heated up to 1030@xmath68c and kept over 3 h. afterward the melt was cooled down to 730@xmath68c with a cooling rate of 6@xmath68c / h , and finally the furnace was cooled to room temperature with the power shut off .
well - formed black crystals were obtained which could be easily cleaved into plates with flat shiny surfaces .
we have also grown rb@xmath0fe@xmath1se@xmath2 single crystals using flux method as described in ref .
@xcite . for bt-1 powder diffraction measurements , we ground @xmath69 grams of single crystals of rb@xmath16fe@xmath17se@xmath2 and rb@xmath20fe@xmath21se@xmath15 . for the experiment on hb-3a
, we used a @xmath700.5 gram single crystal from the same batch of rb@xmath16fe@xmath17se@xmath2 . for hb-1a unpolarized and bt-7 polarized neutron scattering measurements , we used @xmath71 gram single crystals of rb@xmath16fe@xmath17se@xmath2 and rb@xmath20fe@xmath21se@xmath15 with less than 1@xmath68 mosaic aligned in the @xmath72_t$ ] zone in tetragonal notation . to separate the magnetic order from nonmagnetic scattering processes , we performed neutron polarization analysis , where the neutron spin flip ( sf ) scattering for polarization direction parallel to the scattering plane ( hf ) gives pure magnetic scattering @xcite . in the bt-7 setup , the spin polarization direction in the incident beam
could be changed via a flipper and the spin polarization direction for the scattered beam was fixed .
the neutron sf magnetic scattering corresponds to flipper on , while the nuclear coherent scattering is with flipper off , which corresponds to nonspin flip ( nsf ) scattering .
a horizontal guide field was directed along the in - plane momentum transfer ( hf configuration ) , and the flipping ratio of @xmath7022 was obtained in this hf field configuration using an incident energy of 14.7 mev .
a pyrolytic graphite ( pg ) filter was placed in the incident beam direction to suppress @xmath73 scattering .
a position - sensitive detector ( psd ) was used with @xmath74 collimations .
for all the polarized measurements , the sample was at room temperature and aligned in both the @xmath72_t$ ] and @xmath75_t$ ] zones to reach the desired reciprocal space by tilting the sample goniometer .
we first discuss our neutron powder refinement results on rb@xmath16fe@xmath17se@xmath2 with the goal of determining the magnetic structure of the system . in previous theoretical work
@xcite , eight possible magnetic structures have been proposed for the @xmath4 iron vacancy superlattice unit cell .
figure 2 summarizes the comparison between the observed neutron diffraction intensity and calculated intensity for six suggested magnetic structures .
as we can see from the figure , all six magnetic models fail to describe the observed spectrum
. fe@xmath17se@xmath2 at 30 oe with @xmath76 plane measured using a commercial squid .
no superconductivity was observed .
( b ) magnetic structure in the @xmath19 space group representation .
( c ) the experimental data collected from hb-3a .
the data plotted are from one chiral domain . in single crystal diffraction experiments
, we compared several peaks from lef and right chiral domains and found them to be the same .
we used one domain for the data collection , since this can save half of the beam time for a point detector diffractometer .
the radius of the circles are proportional to the intensity of the bragg peaks .
the two peaks enclosed in the green ellipse have quite different intensities for the @xmath32 and @xmath19 afm models .
( d ) the rocking curves for bragg peaks @xmath77 ( red circle ) and @xmath78 ( blue square ) .
the black and green solid lines are the simulation of the expected magnetic intensities for @xmath79 in @xmath32 and @xmath19 models , respectively , where the intensity of the @xmath80 peak are normalized to the value of the experiment .
error bars represent one standard deviation .
( e ) the simulations of the expected bragg peak intensity with the @xmath32 model .
( f ) identical simulation with the @xmath19 model . ] in previous neutron powder diffraction work @xcite , it has been suggested that the block afm structure fig .
3(b ) and the quaternary collinear afm structure in fig .
3(a ) can both fit the observed neutron diffraction spectra @xcite .
our rietveld analysis on rb@xmath16fe@xmath17se@xmath2 for both afm structures shown in fig .
3 confirms this result .
although the block afm structure is thought to be more energetically favorable @xcite , the afm structure shown in fig .
3(a ) is not conclusively ruled out @xcite .
( a ) schematics of the reciprocal space probed in the @xmath81 $ ] plane .
the scans are along the arrow direction which depends on the lower goniometer arc tilting angles @xmath82 . the observed scattering intensity in sf and nsf channels
are plotted in the @xmath83_t$ ] plane of reciprocal space .
( b ) magnetic peak @xmath84 appears in the sf ( flipper on ) channel and disappears in the ( c ) nsf channel .
( d),(e ) magnetic peak @xmath85 in the sf and nsf channels , respectively .
( f - i ) nuclear peaks @xmath86 and @xmath87 show no observable scattering in the sf channel and the entire peak appears in the nsf channel .
the color bars indicate the strength of the neutron scattering intensity . ] to conclusively determine the magnetic structure of the @xmath4 superlattice , we carried out neutron diffraction experiments on an as - grown single crystal of nonsuperconducting rb@xmath16fe@xmath17se@xmath2 .
the zero field cooled ( zfc ) magnetic susceptibility measurements on the sample indicate no bulk superconductivity [ fig .
4(a ) ] . as discussed in previous work @xcite , the block afm structure in fig .
3(b ) can be described by the space group @xmath32 , while the afm structure in fig .
4(b ) has a space group of @xmath19 .
figures 4(e ) and 4(f ) show the expected afm bragg peak intensities in the @xmath88 scattering plane for the @xmath32 and @xmath19 space groups , respectively .
while the intensity of the @xmath89 reflection is weaker than that of the @xmath79 peak in the block afm structure , @xmath89 reflection should be stronger for the quaternary collinear afm structure .
comparison of the mapping of the bragg peaks in the @xmath88 scattering plane in fig .
4(c ) with these two models in figs . 4(e ) and 4(f ) confirms that the block afm structure with space group @xmath32 is correct .
the background subtracted raw data for @xmath89 and @xmath79 bragg peaks are shown in fig .
4(d ) , which again confirm the block afm structure @xcite . fe@xmath21se@xmath15 indicating @xmath5 k. ( b ) susceptibility measurements with 20 mt in - plane field on a 14.2 mg rb@xmath20fe@xmath21se@xmath15 single crystal confirm the superconducting transition temperature @xmath5 k. the left axis is the calculated volume magnetic susceptibility indicating a @xmath90 superconducting volume fraction .
the right axis show magnetic susceptibility in units of emu ( 1 emu@xmath91 ) .
( c ) neutron powder diffraction pattern at 500 k showing the refinement results for the sample including the @xmath4 superlattice distortion .
the superlattice induced peaks are marked with arrows .
( d ) the neutron refinement results in the paramagnetic state at 550 k. ] to further establish the magnetic nature of the proposed block afm structure , we have measured all bragg peaks in fig .
1(d ) by polarized neutrons , where sf and nsf scattering correspond to pure magnetic and pure nuclear scattering , respectively , in the hf configuration .
all measurements were done at room temperature allowing easy tilting of the samples to access different bragg peaks .
figure 5 summarizes the reciprocal space probed and the raw sf and nsf scattering for different bragg peaks in the tetragonal unit cell notation .
initially , we aligned the sample in the @xmath72_t$ ] zone as shown in the dashed line of fig .
the @xmath92_t$ ] and @xmath93_t$ ] axes are aligned along the lower and upper arc axes of the goniometer , respectively . by rotating the lower arc of the goniometer by angles @xmath82 as shown in fig .
5(a ) , we can access magnetic bragg peaks @xmath84 and @xmath85 associated with the block afm structure [ fig .
5(a ) ] . figures 5(b)-5(e ) reveal that the expected magnetic bragg peaks only appear in the sf channel , and there are no features in the nsf channel .
therefore , @xmath84 and @xmath85 bragg reflections are are entirely magnetic in origin with no nuclear component .
to access @xmath86 and @xmath87 peaks , we realigned the sample to the @xmath75_t$ ] zone .
figures 5(f)-5(i ) show that @xmath86 and @xmath87 peaks appear entirely in the nsf channel thus revealing their nuclear origin .
these results conclusively establish the magnetic nature of the block afm structure with @xmath32 space group for the nonsuperconducting rb@xmath16fe@xmath17se@xmath2 .
fe@xmath21se@xmath15 on hb-1a .
( a ) elastic @xmath94 $ ] scans at different temperature .
the scattering is above background scattering at all @xmath28-values .
the background scattering is determined from @xmath95 $ ] scans .
all the peaks are centered at @xmath96 ( not exactly at 0.5 due possibly to a small crystal misalignment on hb-1a ) .
the strong @xmath28-modulation disappears at 640 k. ( b ) rocking curves of the ( 0.51,0.51,3 ) reflection at 2 k and 540 k. ( c ) temperature dependence of the peak intensity for the ( 0.51,0.51,3 ) reflection shows a first - order - like transition above 480 k as marked by the arrow .
( d ) temperature dependence of the @xmath97 $ ] scans shows that the @xmath98 lattice distortion disappears at 495 k , below the tetragonal to @xmath4 superlattice reflection .
( e ) widths of the ( 2,2,0 ) nuclear reflection indicates a second - order - like structural transition at 513 k. the solid lines are guides to the eye . ]
( a ) the @xmath94 $ ] scans below and above @xmath99 .
the background are from @xmath95_t$ ] scans and solid lines are guides to the eye .
( b - g ) elastic @xmath72_t$ ] scans at @xmath100 below and above @xmath101 .
the lattice distortion is not affected by the superconductivity . ]
although the block afm structure for nonsuperconducting rb@xmath16fe@xmath17se@xmath2 is now firmly established , it is still unclear how the static afm order co - exists with superconductivity . in previous work @xcite
, it was argued that the block afm order with huge moments in @xmath7fe@xmath1se@xmath2 microscopically coexists with superconductivity .
however , recent x - ray diffraction measurements have provided compelling evidence for nanoscale phase separation in k@xmath102fe@xmath103se@xmath2 @xcite . to check how superconducting @xmath7fe@xmath104se@xmath15 differs from the nonsuperconducting samples
, we prepared a single crystal rb@xmath20fe@xmath21se@xmath15 , where transport measurement shows @xmath5 k , metallic behavior below 250 k , and semiconducting characteristics above [ fig .
although magnetic susceptibility confirms the superconducting transition at @xmath5 k , we estimate that the superconducting volume fraction in our sample is only around @xmath105 [ fig .
6(b ) ] . to determine the precise crystal lattice structure and atomic compositions , we carried out neutron powder diffraction measurements on bt-1 .
rietveld analysis of the powder diffraction data at 550 k using the @xmath42 space group fits the data well [ fig .
6(d ) ] . at 500 k , fe vacancies order into a @xmath4 superlattice structure as shown in fig .
1(a ) @xcite and the powder diffraction pattern can be well described by the space group @xmath18 .
we have searched extensively for structural and magnetic peaks in superconducting rb@xmath20fe@xmath21se@xmath2 .
in addition to confirming the @xmath4 afm peaks at identical positions as the nonsuperconducting rb@xmath16fe@xmath17se@xmath2 , we find a set of new peaks at wave vectors @xmath106 where @xmath107 . along the @xmath13-axis ,
these peaks are broad and lorentzian - like , and centered at integer @xmath28 positions .
they disappear on warming from 445 k to 640 k [ fig .
7(a ) ] , suggesting that they are associated with either a magnetic phase transition or structural lattice distortion not related to the known @xmath4 superlattice structure .
the broad nature of the scattering along the @xmath13-axis indicates that they are quasi two - dimensional .
figure 7(b ) plots rocking curve scans at wave vector @xmath108 , which again show the disappearance of the low - temperature peak at 540 k. to determine the phase transition temperature associated with the @xmath106 peaks and compare those with the tetragonal ( @xmath42 ) to the @xmath4 superlattice ( @xmath18 ) transition , we carefully measured the intensity of the @xmath109 peak and the width of the @xmath110 bragg peak .
while the @xmath109 peak shows an abrupt first - order - like phase transition and disappears above 480 k [ fig .
7(c ) ] , the bragg peak width on the @xmath110 reflection shows a second - order - like phase transition at 513 k [ fig .
7(e ) ] . figure 7(d ) shows the @xmath97 $ ] scans at different temperatures , which display no peak at 496 k , thus confirming that the phase transition temperature for the @xmath106 peaks happens at a lower temperature than that of the tetragonal - to-@xmath4 superlattice distortion .
$ ] bragg peak for rb@xmath20fe@xmath21se@xmath15 at room temperature .
the exerpimental configuration is the same with that for fig .
while the nsf scattering on the left panel shows clear nuclear scattering centered around the @xmath111 $ ] positions , there is no evidence of sf scattering in the same reciprocal region .
this means that there is no elastic magnetic scattering at the same wave vector as the iron pnictides such as lafeaso @xcite . ] to see if the two - dimensional @xmath106 scattering responds to the formation of the superconductivity , we have carried out in - plane and @xmath13-axis scans below and above @xmath5 k on rb@xmath20fe@xmath21se@xmath2 .
we find that these peaks do not exhibit any changes across @xmath101 [ fig .
8(a ) ] . figure 8(b)-(g ) show elastic scans along the @xmath72_t$ ] direction at @xmath28-values of @xmath28=0,1,2,3,1.5,2.5 . for all @xmath28-values ,
we find peaks centered at @xmath112 rlu , confirming the two - dimensional nature of the scattering . in previous x - ray diffraction experiments on cs@xmath0fe@xmath113se@xmath2 ,
pomjakushin _ et al .
_ @xcite have also found peaks at @xmath106 .
this means that a portion of the signal we observe at @xmath114 must be due to a structural distortion .
to determine if there is any additional magnetic component in the @xmath114 scattering , we performed neutron polarization analysis . figures 9(a ) and 9(b ) show mappings of the reciprocal space in the @xmath72 $ ] zone for the nsf and sf scattering , respectively . while one can see a clear rod of scattering centered at @xmath115 along the @xmath111 $ ] direction in the nsf channel , the sf scattering is featureless in the entire probed range .
since the @xmath114 rod - like scattering is not compatible with the @xmath4 superlattice structure , our data suggest that rb@xmath20fe@xmath21se@xmath2 is phase separated and exhibits two structural transitions , one at 513 k and the second one at 480 k. fe@xmath116se@xmath15 at 480 k. they are fit using a combination of @xmath117 and @xmath118 phases .
the purple trace indicates the intensity difference between the observed ( green solid line ) and calculated ( red crosses ) structures .
the left inset is a schematic of the @xmath119 phase which can induce nuclear peaks at @xmath55_t$ ] ( @xmath120 ) . ] in an attempt to determine the lattice structure associated with the @xmath111 $ ] reflections , we carried out detailed rietveld analysis on the neutron powder diffraction pattern .
since the @xmath111 $ ] reflections are much weaker in the powder pattern , we were able to fit the powder diffraction pattern with a combination of @xmath121% @xmath4 superlattice ( @xmath18 ) and rb@xmath20fe@xmath122se@xmath2 fe - vacancy model in the inset of fig . 10 ( @xmath119 ) @xcite .
however , such an fe - vacancy model with @xmath119 space group will not be able to explain the quasi two - dimensional rod scattering we observe in the triple - axis measurements .
therefore , it remains unclear what crystalline lattice distortion gives rise to the observed superlattice reflections , although we know such scattering enlarges the nuclear unit cell by @xmath6 .
we note that a recent x - ray study on superconducting rb@xmath0fe@xmath1se@xmath2 samples also found a @xmath6 superlattice structure @xcite .
using single crystal neutron diffraction and neutron polarization analysis , we have confirmed that the block afm structure in fig . 1 is the only possible magnetic structure for insulating afm rb@xmath16fe@xmath17se@xmath2 @xcite .
although we have also found a similar afm structure for the superconducting rb@xmath20fe@xmath21se@xmath2 , careful analysis of the diffraction spectra reveals another structural phase transition associated with enlarged unit cell for superconducting rb@xmath20fe@xmath21se@xmath2 . in previous arpes measurements on superconducting @xmath7fe@xmath1se@xmath2 @xcite
, different groups have reached the same conclusion concerning the electron - like fermi surfaces at @xmath10 points .
however , there have been debates concerning the origin of the observed weak electron pockets near the @xmath123 point @xcite . in principle
, the electron pockets near the @xmath124 point can arise from band folding if there exists a @xmath98 structural or magnetic phase transition @xcite .
our observation of the quasi two - dimensional @xmath114 superlattice reflections suggests that the observed electron fermi surfaces near the @xmath123 point may indeed be due to band folding instead of a surface state .
since the block afm structure in insulating rb@xmath16fe@xmath17se@xmath2 can not arise from fermi surface nesting , we speculate that the @xmath6 lattice distortions in superconducting rb@xmath20fe@xmath21se@xmath2 may be associated with the metallic portion of the sample . in this picture , the superconducting phase in @xmath7fe@xmath1se@xmath2 may be mesoscopically phase separated from the nonsuperconducting phase , where superconductivity and afm order are intertwined in a very short length scale and live in separate regions . theoretically , it has been suggested that the @xmath125fe@xmath126se@xmath2 phase is a semiconductor with a low energy band gap @xcite .
so with electron or hole doping , such a phase would become nonmagnetic and superconducting .
although we have no direct proof that the superconducting portion of the sample is associated with the @xmath114 superlattice distortion , systematic neutron scattering measurements are currently underway to investigate the relationship of such phase to the block afm order and superconductivity .
we are grateful to jiangping hu and tao xiang for helpful discussions .
work at iop is supported by the ministry of science and technology of china ( 973 project no .
2010cb833102 , 2010cb923002 , 2011cba00110 ) , and chinese academy of sciences .
the single crystal growth and neutron scattering effort at ut is supported by u.s .
doe bes under grant no .
de - fg02 - 05er46202 ( p.d . ) .
work at ornl neutron scattering facilities are supported by the scientific user facilities division , office of basic energy sciences , u.s .
department of energy .
part of the work at ut is also supported by the u.s .
nsf - oise-0968226 ( p. d. ) .
the work at zju is supported by by natural science foundation of china ( grant no .
10974175 ) , the ministry of science and technology of china ( 973 project no .
2011cba00103 ) .
j. g. guo , s. f. jin , g. wang , s. c. wang , k. x. zhu , t. t. zhou , m. he , and x. l. chen , phys .
b * 82 * , 180520(r ) ( 2010 ) . a. krzton - maziopa , z. shermadini , e. pomjakushina , v. pomjakushin , m. bendele , a. amato , r. khasanov , h. luetkens , k. conder , j. phys . : condens .
matter * 23 * , 052203 ( 2011 ) .
m. h. fang , h. d. wang , c. h. dong , z. j. li , c. m. feng , j. chen , h. q. yuan , europhys . lett . *
93 * , 47004 ( 2011 ) .
a. f. wang , j. j. ying , y. j. yan , r. h. liu , x. g. luo , z. y. li , x. f. wang , m. zhang , g. j. ye , p. cheng , z. j. xiang , x. h. chen , phys .
b * 83 * , 060512 ( 2011 ) .
l. h@xmath127ggstr@xmath128 m _ et al .
_ , j. magn .
magn . mater .
* 98 * , 37 ( 1991 ) . y. kamihara , t. watanabe , m. hirano , and h. hosono , j. am .
soc . * 130 * , 3296 ( 2008 ) . c. de la cruz , q. huang , j. w. lynn , j. li , w. ratcliff ii , j. l. zarestky , h. a. mook , g. f. chen , j. l. luo , n. l. wang , pengcheng dai , nature ( london ) * 453 * , 899 ( 2008 ) .
d. c. johnston , advnaces in physics * 59 * , 803 ( 2010 ) . i. i. mazin , d. j. singh , m. d. johannes , and m. h. du , phys . rev
* 101 * , 057003 ( 2008 ) .
et al . _ , epl * 83 * , 27006 ( 2008 ) .
h. ding _ et al_. , epl * 83 * , 47001 ( 2008 ) .
k. kuroki , s. onari , r. arita , h. usui , y. tanaka , h. kontani , and h. aoki , phys .
* 101 * , 087004 ( 2008 ) .
a. v. chubukov , physica c * 469 * , 640 ( 2009 ) .
f. wang , h. zhai , y. ran , a. vishwanath , and d .- h .
lee , phys .
* 102 * , 047005 ( 2009 ) .
y. zhang , l. x. yang , m. xu , z. r. ye , f. chen , c. he , j. jiang , b. p. xie , j. j. ying , x. f. wang , x. h. chen , j. p. hu , d. l. feng , nature materials * 10 * , 273 ( 2011 ) . t. qian , x .-
wang , w .- c .
jin , p. zhang , p. richard , g. xu , x. dai , z. fang , j .-
chen , and h. ding , phys .
* 106 * , 187001 ( 2011 ) .
d. x. mou _
et al . _ ,
lett . * 106 * , 107001 ( 2011 ) .
z. wang , y. j. song , h. l. shi , z.w .
wang , z.chen , h. f. tian , g. f. chen , j. g. guo , h. x. yang , j. q. li , physical review b * 83 * , 140505(r ) ( 2011 ) .
p. zavalij , wei bao , x. f. wang , j. j. ying , x. h. chen , d. m. wang , j. b. he , x. q. wang , g. f. chen , p .- y .
hsieh , q. huang , and m. a. green , phys . rev .
b * 83 * , 132509 ( 2011 ) .
j. bacsa , a. y. ganin , y. takabayashi , k. e. christensen , k. prassides , m. j. rosseinsky , j. b. claridge , chem .
* 2 * , 1054 ( 2011 ) .
w. bao , q. huang , g. f. chen , m. a. green , d. m. wang , j. b. he , x. q. wang , y. qiu , chinese phys . lett . * 28 * , 086104 ( 2011 ) .
pomjakushin , d. v. sheptyakov , e. v. pomjakushina , a. krzton - maziopa , k. conder , d. chernyshov , v. svitlyk , z. shermadini , phys .
b * 83 * , 144410 ( 2011 ) .
f. ye , s. chi , wei bao , x. f. wang , j. j. ying , x. h. chen , h. d. wang , c. h. dong , minghu fang , arxiv : 1102.2882 .
wei bao , g. n. li , q. huang , g. f. chen , j. b. he , m. a. green , y. qiu , d. m. wang , j. l. luo , arxiv : 1102.3674 .
pomjakushin , e. v. pomjakushina , a. krzton - maziopa , k. conder , z. shermadini , j. phys . : condens .
matter * 23 * , 156003 ( 2011 ) .
jun zhao , q. huang , c. de la cruz , shiliang li , j. w. lynn , y. chen , m. a. green , g. f. chen , g. li , z. li , j. l. luo , n. l. wang , pengcheng dai , nature materials * 7 * , 953 ( 2008 ) .
z. shermadini , a. krzton - maziopa , m. bendele , r. khasanov , h. luetkens , k. conder , e. pomjakushina , s. weyeneth , v. pomjakushin , o. bossen , a. amato , phys .
lett . * 106 * , 117602 ( 2011 ) .
j. w. lynn and pengcheng dai , physica c * 469 * , 469 - 476 ( 2009 ) .
i. i. mazin , physics * 4 * , 26 ( 2011 ) .
m. y. wang , c. fang , d - x yao , g. t tan , l. w. harriger , y. song , t. netherton , c. l. zhang , m. wang , m. b. stone , w. tian , j. p. hu , pengcheng dai , arxiv : 1105.4675
. j. q. li , y. j. song , h. x. yang , z. wang , h. l. shi , g. f. chen , z.w .
wang , z.chen , h. f. tian , arxiv : 1104.5340
. y. j. song , z. wang , z. w. wang , h. l. shi , z. chen , h. f. tian , g. f. chen , h. x. yang , j. q. li , arxiv:1104.4844 .
a. ricci , n. poccia , b. joseph , g. arrighetti , l. barba , j. plaisier , g. campi , y. mizuguchi , h. takeya , y. takano , n. l. saini , a. bianconi , superconductor science and technology * 24 * , 082002 ( 2011 ) .
a. ricci , n. poccia , g. campi , b. joseph , g. arrighetti , l. barba , m. reynolds , m. burghammer , h. takeya , y. mizuguchi , y. takano , m. colapietro , n. l. saini , a. bianconi , arxiv : 1107.0412 . j. w. lynn , y. chen , s. chang , y. zhao , s. chi , w. ratcliff , b. g. ueland , r. w. erwin , nist journal of research , ( to be published )
. o. j. lipscombe , leland w. harriger , p. g. freeman , m. enderle , chenglin zhang , miaoying wang , takeshi egami , jiangping hu , tao xiang , m. r. norman , and pengcheng dai , phys .
b * 82 * , 064515 ( 2010 ) .
b. c. chakoumakos , h. cao , f. ye , a. d. stoica , m. poovici , m. sundaram , w. zhou , j. s. hicks , g. w. lynn , and r. a. riedel , j. appl .
* 44 * , 655 ( 2011 ) .
x. w. yan , m. gao , z. y. lu , and t. xiang , phys .
b * 83 * , 233205 ( 2011 ) .
v. tsurkan , j. deisenhofer , a. g@xmath129nther , h .- a .
krug von nidda , s. widmam , and a. loidl , arxiv : 1107.3932v1 .
x. w. yan , m. gao , z .- y .
lu , and t. xiang , phys .
* 106 * , 087005 ( 2011 ) . | neutron diffraction has been used to study the lattice and magnetic structures of the insulating and superconducting rb@xmath0fe@xmath1se@xmath2 . for the insulating rb@xmath3fe@xmath1se@xmath2 ,
neutron polarization analysis and single crystal neutron diffraction unambiguously confirm the earlier proposed @xmath4 block antiferromagnetic structure . for superconducting samples ( @xmath5 k )
, we find that in addition to the tetragonal @xmath4 superlattice structure transition at 513 k , the material develops a separate @xmath6 superlattice structure at a lower temperature of 480 k. these results suggest that superconducting rb@xmath3fe@xmath1se@xmath2 is phase separated with coexisting @xmath6 and @xmath4 superlattice structures . |
quantum groups have become an indispensable tool in many areas of mathematical physics and mathematics . in a broad class of quantum theoretical models
it has turned out that finding a relation to a quantum group is the key for obtaining exact information about the spectrum or the correlation functions .
so far most of the vast amount of work devoted to quantum group theory and their physical applications was concerned with quantum groups that can be studied in a purely algebraic manner .
this is the case e.g. if the relevant representations are highest weight representations , as is often assumed .
however , in physical applications to quantum theoretical models the choice of a scalar product on the space of states usually determines the hermiticity relations for the representatives of the quantum group generators . in many cases like those corresponding to the so - called _ non - compact _ quantum groups it turns out that the corresponding unitary representations are always infinite - dimensional and generically neither of highest nor lowest weight type . in order to exploit the information provided by the appearance of such a quantum group it is clearly important to have efficient mathematical tools for analyzing the corresponding representation theory .
unfortunately there are comparatively few results about the representation theory of non - compact quantum groups .
this seems to be an important obstacle for making further progress in many quantum integrable models .
nevertheless there is an interesting example that was first studied independently in @xcite , @xcite and @xcite .
these references were considering a particular class of infinite dimensional unitary representations of @xmath0 .
this class of representations , henceforth denoted @xmath3 , @xmath4 , was the also first to be used in a concrete physical application : understanding the tensor products of the above - mentioned representations was crucial for obtaining exact results on the quantum liouville theory @xcite@xcite .
a rather remarkable duality phenomenon was observed in @xcite,@xcite and @xcite .
the representations in question are simultaneously representations of the two quantum groups @xmath5 and @xmath6 with deformation parameters @xmath7 and @xmath8 respectively .
this duality turns out to be deeply related to the quantum field theoretical self - duality of liouville theory under the change of the coupling constant @xmath9 into @xmath10 @xcite .
moreover , it is this duality under @xmath11 that allows one to cover the so - called _ strong - coupling _ regime where @xmath12 by analytic continuation of the results for real values of @xmath9 @xcite@xcite@xcite@xcite .
the results of the present paper clarify the origin of this phenomenon to a certain extent .
given the operators @xmath13 representing one of the two algebras , say @xmath5 , one may obtain the representatives of the second algebra @xmath6 as _ nonpolynomial _ operator functions of the operators @xmath13 .
this is found to be consistent with the respective co - products . in particular , restricting attention to only one of the two algebras does not lead to any degeneracy as is sometimes suggested in the literature .
faddeev has proposed to unite the quantum groups @xmath5 and @xmath6 into some enlarged object for which he has coined the name `` modular double '' .
the proposal of @xcite as refined in @xcite amounts to defining it as the product of @xmath14 and @xmath15 , where , roughly speaking , @xmath14 is obtained from @xmath5 by adjoining a sign to the center generated by the casimir of @xmath5 .
we feel that this definition for the modular double has a disadvantage , though .
most representations of @xmath16 are simply tensor products of representations of the two factors .
the representations @xmath3 on the contrary are distinguished by the fact that they do _ not _ factorize as a tensor product of representations for @xmath5 and @xmath6 .
this is what makes the duality under @xmath11 a nontrivial statement .
since the category formed by the representations @xmath3 is _ closed _ under tensor products @xcite it seems natural to look for the group - like object that contains the interesting representations @xmath3 _
only_. we therefore propose to look for a definition of the `` modular double '' that excludes the representations of @xmath17 which factorize trivially .
this definition should still capture the duality phenomenon mentioned above .
as we have indicated , this naturally leads us to consider nonpolynomial functions of the generators .
the basic objects underlying our approach to the modular double will be an algebra @xmath18 of _ bounded _ operators , a coproduct @xmath19 on @xmath18 , an invariant integration ( haar - measure ) on @xmath18 and the r - operator proposed in @xcite .
we are going to establish the main properties satisfied by these objects , which are all _ self - dual _ under @xmath11 .
the algebra @xmath18 can be thought of as being generated from operators that represent @xmath5 in a similar ( in fact , closely related ) way as the algebra of bounded operators on @xmath20 is related to the usual quantum mechanical position and momentum operators @xmath21 and @xmath22 .
our point of view is inspired by the one of woronowicz @xcite , which has stimulated the development of a theory for noncompact quantum groups in a @xmath23-algebraic framework , see e.g. @xcite and references therein . however , although we believe that our results represent substantial progress towards a proof that the modular double fits into such a @xmath23-algebraic framework , it was not our aim to actually carry out such a proof here .
we also clarify the relation between the r - operator proposed in @xcite and the calculus of clebsch - gordan and racah - wigner - coefficients of @xcite .
establishing this relation is important for the following reason . in @xcite
it was shown that certain families of representations of the virasoro algebra and of the quantum group @xmath0 behave equivalently under the respective product operations ( fusion and tensor product ) .
together with the results of the present paper it follows that the respective _ braiding _ operations are equivalent as well . to be specific we will mostly consider the case that the deformation parameter is of the form @xmath7 , where @xmath24 .
however , our results will carry over to the `` strong coupling regime '' @xmath12 , see the remarks in subsection [ strong ] .
@xmath0 is a hopf - algebra with @xmath25=\frac{k^2-k^{-2}}{q - q^{-1}};\\ & \text{star - structure:}\quad k^*=k , \qquad e^*=e , \qquad f^*=f \ , .
\end{aligned}\ ] ] the center of @xmath0 is generated by the @xmath26-casimir @xmath27 compared to the definition used in @xcite we have redefined @xmath28 .
this will allow us to realize @xmath29 by positive operators . in the present paper we will study a one - parameter class @xmath30 , @xmath31 , of representations of @xmath32 .
they are constructed as follows : the representation will be realized on the space @xmath30 of entire analytic functions @xmath33 that have a fourier - transform @xmath34 which is meromorphic in @xmath35 with the possible poles contained in @xmath36 where @xmath37 .
the representation of @xmath0 on @xmath3 is then defined by choosing the representatives @xmath38 for @xmath39 to be the following finite difference operators @xmath40 where @xmath22 and @xmath21 are self - adjoint operators satisfying @xmath41=(2\pi i)^{-1}$ ] . by embedding @xmath3 as a dense subspace into the hilbert space @xmath20 one obtains a unitary representation of @xmath0 generated from the self - adjoint operators @xmath42 , @xmath43 and @xmath44 @xcite
. we will find it convenient to formulate our results in a `` universal '' setting .
let us define @xmath47 .
the algebra @xmath48 of bounded operators on @xmath46 is generated by two pairs @xmath49 , @xmath50 satisfying @xmath51=(2\pi i)^{-1}$ ] . the action of @xmath0 on @xmath46 is defined by @xmath52 this representation of @xmath0 on @xmath46 is reducible : the operator @xmath53 commutes with @xmath42 , @xmath43 , @xmath44 and determines the representation of the casimir via @xmath54 the action of ( [ efk2 ] ) on an eigenspace of @xmath55 reduces to the action ( [ efk1 ] ) on @xmath3 upon identification @xmath56 and @xmath57 .
this means that @xmath46 decomposes into the representations @xmath3 as follows : @xmath58 the representations @xmath3 and @xmath59 are unitarily equivalent @xcite : there exists a unitary operator @xmath60 such that @xmath61 for all @xmath62 .
the operator @xmath63 defines an operator @xmath64 if one considers the operator function @xmath65
. it will sometimes be convenient to consider instead of @xmath46 a space @xmath45 in which @xmath3 and @xmath59 are identified : @xmath66 @xmath45 is of course isomorphic to @xmath67 .
it is important to also consider nonpolynomial functions of the operators @xmath42 , @xmath43 , @xmath44 .
let us first note that standard functional calculus for positive selfadjoint operators allows one to consider complex powers of the generators such as @xmath68 , @xmath69 .
the following result offers a partial explanation for the phenomenon of modular duality .
\(i ) the operators @xmath70 , @xmath71 , @xmath72 obtained by replacing @xmath11 in ( [ efk1 ] ) , @xmath73 generate a representation of @xmath74 with @xmath75 .
the generators @xmath70 , @xmath71 , @xmath72 commute with the operators @xmath76 , @xmath77 , @xmath78 on @xmath3 .
+ ( ii ) for @xmath79 we have @xmath80 being operator functions of @xmath76 , @xmath77 , @xmath78 , the operators @xmath70 , @xmath71 , @xmath72 do * not * commute with @xmath76 , @xmath77 , @xmath78 in the usual sense ( commutativity of the spectral projections ) .
we shall now define an algebra of bounded operators that can be considered as operator functions of @xmath42 , @xmath43 , @xmath44 . to begin with ,
let @xmath81 , @xmath4 be a family of bounded operators on @xmath20 such that @xmath82 a bounded operator @xmath83 on @xmath46 can be defined for each such family @xmath84 by means of ( [ chdecomp ] ) .
these operators @xmath83 form a subalgebra @xmath85 of the algebra of all bounded operators on @xmath45 . let @xmath86 be the @xmath23 subalgebra obtained as the completion of @xmath85 w.r.t . the operator norm .
this algebra can be thought of as being generated from the unbounded elements @xmath22 , @xmath21 , @xmath55 .
however , there is no canonical way to define @xmath87 as a function of @xmath42 , @xmath43 , @xmath44 .
the center of the algebra of bounded operators generated from @xmath42 , @xmath43 , @xmath44 should be generated from operator functions of the casimir , or equivalently @xmath88 , cf .
( [ casimir ] ) .
this is closely related to the fact that the representations @xmath30 and @xmath59 are unitarily equivalent .
elements of the `` true '' algebra @xmath89 should therefore commute with the operator @xmath90 which establishes the equivalence between @xmath30 and @xmath59 , @xmath91 this amounts to considering only those elements of @xmath86 that leave @xmath45 invariant .
a co - product is defined on @xmath0 via @xmath92 in the following we shall adopt the convention to denote @xmath93 it follows from ( * ? ? ?
* theorem 2 ) that @xmath94 , @xmath95 and @xmath96 are self - adjoint and positive and therefore generate a representation of @xmath0 on @xmath97 .
the following lemma proven in section 3 establishes consistency of the co - product with modular duality : [ codual ] the co - product ( [ de])([conv ] ) acts on the dual part of the modular double as follows .
@xmath98 a representation of the co - product on the algebra @xmath18 can be defined by means of the clebsch - gordan maps defined in @xcite .
these maps yield a three parameter family of maps @xmath99:{{\mathcal p}}_{s_{{\mathfrak 2}}}{\otimes}{{\mathcal p}}_{s_{{\mathfrak 1}}}{\rightarrow}{{\mathcal p}}_{s_{{\mathfrak 3}}}$ ] that satisfy the intertwining property @xmath100\;\circ ( \pi_{s_{{{\mathfrak 2}}}}{\otimes}\pi_{s_{{{\mathfrak 1}}}})\circ{\delta}(x)\,=\ , \pi_{s_{{\mathfrak 3}}}(x)\circ{{\mathsf c}}[s_3|s_2,s_1]\ ; \,\ ] ] and extends to a two - parameter family of unitary operators @xmath101:l^{2}({{\mathbb r}}^2)\;{\rightarrow}\ ; { { \mathcal h}}$ ] .
let us introduce the operators @xmath102 , @xmath103 on @xmath97 respectively .
the identification ( [ chdecomp ] ) allows us to consider @xmath104 $ ] as a unitary operator @xmath105 where the operators @xmath106 , @xmath50 are realized on the spaces @xmath107 as multiplication operators .
for each element @xmath108 we may now define @xmath109 by @xmath110 since @xmath111 is unitary and @xmath13 is bounded we clearly have boundedness of @xmath112 .
[ coass ] the coproduct @xmath19 is coassociative on @xmath18 , i.e. @xmath113 the antipode consistent with ( [ de ] ) is defined as an anti - automorphism of @xmath0 such that @xmath114 the action of the antipode on nonpolynomial functions of @xmath42 , @xmath43 and @xmath44 can be introduced by means of @xmath115
the fact that @xmath21 is shifted by an imaginary amount means that @xmath116 is not defined on all of @xmath18 .
this unboundedness of the antipode is not unexpected @xcite .
let us first note that the decomposition ( [ chdecomp ] ) induces a family of projections @xmath117 .
we shall often use the shorthand notation @xmath118 .
define linear functionals @xmath119 and @xmath120 on dense subsets @xmath121 and @xmath122 of @xmath18 respectively by @xmath123 where the measure m is defined by @xmath124 [ invthm ] ( i ) the haar - measures @xmath119 and @xmath120 are left and right invariant respectively , @xmath125 where we assume @xmath126 to be taken from the respective domains of definition .
+ ( ii ) for any @xmath127 , the haar - measures satisfy , respectively @xmath128 where @xmath129 is the co - unit , and the left and right @xmath26-adjoint actions are defined as @xmath130 and @xmath131 if @xmath132 .
we believe that the triple @xmath133 that we have defined above constitutes a somewhat more satisfactory definition of the modular double , although more work is needed to show that it fits into the axiomatics for noncompact quantum groups of @xcite .
the self - duality under @xmath11 is manifest in this formulation .
it also becomes clear that the modular double can not be considered as a deformation of a classical group : the haar - measure has no classical limit @xmath134 due to the factor @xmath37 that appears in the definition of @xmath135 . to begin with
, we introduce the special function @xmath136 that will be used to define the r - operator .
it may be defined via ( recall that @xmath37 ) @xmath137 let us furthermore introduce an anti - self - adjoint element @xmath138 such that @xmath139 .
define @xmath140 where @xmath141 . as we will explain below ( see corollary [ rprorep ] ) , @xmath142 coincides with the r - operator proposed by l. faddeev in @xcite .
notice that the property ( [ gcc ] ) implies that @xmath143 for @xmath144 .
this makes @xmath142 manifestly unitary .
[ t1 ] the operator @xmath142 has the following properties : @xmath145 the r - operator allows us to introduce the braiding of tensor products of the representations @xmath30 . specifically , let the operator @xmath146 be defined by @xmath147 , where @xmath148 is the operator that permutes the two tensor factors .
property ( i ) from theorem [ t1 ] implies as usual that @xmath149 .
the operator @xmath142 can clearly be projected to an operator @xmath150 on @xmath151 .
the action of this operator admits a representation by means of a distributional kernel : [ tint ] let @xmath152 be a fourier transform of @xmath153 .
the action of the @xmath142-operator on admits the following representations as an integral operator in `` coordinate '' and `` momentum '' space respectively : @xmath154 with the kernels given by @xmath155 @xmath156 the functions @xmath157 and @xmath158 are close relatives of the function @xmath136 that will be defined in subsection [ specfkt ] below , and @xmath159 is taken as a short notation for the distribution @xmath160 . in order to demonstrate that the r - operator we are considering here indeed deserves to be called `` universal
'' we are now going to show that the usual r - matrix for highest weight representations of @xmath0 can be extracted from the analytic properties of the matrix elements given in theorem [ tint ] . as a preparation
let us consider the representation of @xmath0 on the dual space @xmath161 of @xmath3 . an interesting class of elements of @xmath161 is furnished by the ( complexified ) delta - functionals @xmath162 , @xmath163 the @xmath162 are well - defined for all @xmath164 , and the action of @xmath0 is realized by @xmath165_q{\delta}_{k+ib}\\ { { \mathsf f}}_s^t{\delta}_{k}\;=\;&-\big[{{\textstyle \frac{q}{2b } } } + { { \textstyle \frac{i}{b } } } ( k+s)\big]_q{\delta}_{k - ib } \end{aligned}\qquad { { \mathsf k}}_s^t{\delta}_{k}\;=\;e^{-\pi bk}{\delta}_{k},\ ] ] where @xmath166_q \equiv \frac{\sin(\pi b^2 t)}{\sin(\pi b^2)}$ ] is the standard definition of a q - number , and the superscript `` @xmath167 '' on the generators indicates transposition .
let us restrict attention to the set @xmath168 of functionals @xmath169 for which @xmath170 is an element of @xmath171 .
it is easy to verify that ( [ deltatrf ] ) realizes a highest weight representation on @xmath168 .
[ t2 ] the action of @xmath172 on @xmath173 is given by @xmath174_q } \
, \bigl((q - q^{-1}){{\mathsf e}}{\otimes}{{\mathsf f}}_s\bigr)^n \
, q^{{{\mathsf h}}{\otimes}{{\mathsf h}}_s } \ , .
\end{aligned}\ ] ] [ t3 ] the clebsch - gordan maps @xmath99 $ ] diagonalize the r - operator in the following sense : @xmath175 \,{{\mathsf b}}_{s_{{{\mathfrak 2}}}s_{{{\mathfrak 1}}}}^{}\,=\ , { \omega}(s_3|s_2,s_1 ) \ , { { \mathsf c}}[s_3|s_2,s_1 ] \,,\ ] ] with eigenvalue @xmath176 given as @xmath177 we would finally like to point out that our results carry over to the strong coupling regime @xmath178 .
this is almost obvious for those results whose proof relies mainly on the properties of the special functions @xmath136 , @xmath158 and @xmath157 .
in this case the operators @xmath42 , @xmath43 and @xmath44 are _ normal _ ( as follows from eqn .
( [ efw ] ) below ) , and the hermitian conjugation acts as @xmath179 concerning the results that rely on @xcite one may note that they all amount to certain identities between distributions that are defined by a standard analytic regularization in terms of the meromorphic functions @xmath136 , @xmath158 and @xmath157 . the relevant analytic properties underlying the validity of these identities all remain intact upon analytically continuing from the case of real @xmath9 to @xmath12 .
the double gamma function @xmath180 was introduced and studied by barnes @xcite . in what follows we will be dealing with ( recall that @xmath37 ) @xmath181 this function is closely related to the remarkable special functions introduced under the names of `` quantum dilogarithm '' in @xcite and `` quantum exponential function '' in @xcite .
@xmath158 is a meromorphic function that has the following properties @xcite : @xmath182 \zeta_b \ , e^{\pi i x(x - q ) } & \text { for } \ \im(x){\rightarrow}-\infty \end{array } \right . \ , , \\
\label{gan } \begin{array}{r } g_b(x ) \ \text{has poles at } \\ [ -0.25 mm ] g_b(x ) \ \text{has zeros at } \end{array } & & \begin{array}{l } x = -nb - mb^{-1 } \\ [ -0.25 mm ] x = q+nb+mb^{-1 } \end{array } \ \
n , m \in { { { \mathbb z}}}^{\geq 0 } \,,\end{aligned}\ ] ] where @xmath183 . by proposition 5 in @xcite ,
the @xmath184-function admits for @xmath185 the following infinite product representation @xmath186 we are also going to use two other functions that are closely related to @xmath158 , namely @xmath187 the representation ( [ gg ] ) for @xmath188 introduced above follows easily from the integral representation for the double gamma function introduced in @xcite . for the reader
s convenience we shall also list the relevant properties of @xmath157 that follow from ( [ gdual])([gan ] ) . @xmath189 \label{wcc } \text{complex conjugation } & & \overline{w_b(x ) } = w_b(-\bar{x } ) \,,\\
\label{wan } \begin{array}{r } w_b(x ) \ \text{has poles at } \\
[ -0.25 mm ] w_b(x ) \ \text{has zeros at } \end{array } & & \begin{array}{l } x = - i \ , ( \frac{q}{2 } + nb+mb^{-1 } ) \\ [ 0.25 mm ] x = i \ , ( \frac{q}{2 } + nb+mb^{-1 } ) \end{array } \ \
n , m \in { { { \mathbb z}}}^{\geq 0 } \,.\end{aligned}\ ] ] notice that @xmath190 if @xmath191 is real .
hence @xmath192 is unitary if @xmath13 is a self - adjoint operator .
[ uvg ] let @xmath193 and @xmath194 be self - adjoint operators such that @xmath195 = 2 \pi i$ ] .
let @xmath196 be a function on the positive real axis and let @xmath197 .
then we have @xmath198 where @xmath199 , @xmath200 .
it is convenient to introduce @xmath201 and @xmath202 .
observe that @xmath203= \frac{1}{2\pi i}$ ] ; so that we have @xmath204 for any function @xmath205 that is bounded and analytic in the strip @xmath206 . using the baker - campbell - hausdorff formula for weyl - type operators and the properties of the function @xmath207 , we may calculate as follows : @xmath208 the last expression is therefore unitarily equivalent to the positive self - adjoint operator @xmath209 .
our claim ( [ phiab ] ) follows by applying the standard functional calculus of self - adjoint operators .
relation ( [ qf ] ) can be proven along the same lines taking into account that , thanks to self - duality ( [ wdual ] ) , @xmath207 obeys also the equation @xmath210 therefore , for @xmath211 we have @xmath212 another way to prove relation ( [ qf ] ) in lemma [ uvg ] is to use the @xmath9-binomial formula ( [ bbin ] ) that we derive in appendix . when @xmath167 approaches the value @xmath213 , the @xmath9-binomial coefficient ( [ bbc ] ) vanishes unless @xmath214 takes special values determined by ( [ gan ] ) .
furthermore , for @xmath215 the @xmath9-binomial coefficient has nonvanishing residues only at @xmath216 and @xmath217 .
the contributions from these two poles yield the two terms on the r.h.s . of ( [ qf ] ) .
similar consideration for @xmath167 approaching @xmath218 , @xmath219 shows that the @xmath9-binomial coefficient has nonvanishing residues at @xmath220 .
therefore can be represented as sum of @xmath221 terms which is analogous to the formula in the compact case .
the proven lemma leads to useful representations for the generators and the r - operator of @xmath32 . for brevity
, we denote @xmath222 and @xmath223 , whereas @xmath224 and @xmath225 will stand for their counterparts with @xmath9 replaced by @xmath226 .
[ rprod ] @xmath227 and @xmath228 admit the following representation : @xmath229 @xmath142 may be represented as follows : @xmath230 in lemma [ uvg ] , we can identify @xmath231 and @xmath232 , where @xmath233 ( with convention that @xmath234 ) . then , as seen from the definition ( [ efk2 ] ) , we have @xmath235 for @xmath236 and @xmath237 for @xmath238 . therefore , ( [ efw ] ) is just a particular case of ( [ phiab ] ) .
furthermore , for functions @xmath196 defined on @xmath239 we have @xmath240 in particular , we can take @xmath241 ( recall that @xmath143 for @xmath144 ) . for this choice of @xmath196 , the representation ( [ rpr ] ) for @xmath142 follows immediately from the definition ( [ rdef ] ) .
[ efbb ] for @xmath197 we have @xmath242 notice that , if @xmath243 and @xmath244 are identified as in the proof of lemma [ rprod ] , then for @xmath236 and @xmath245 for @xmath238 .
thus , relations ( [ efdual ] ) are a particular case of ( [ qf ] ) .
this proves the relations ( [ dualeqs ] ) from lemma 1 .
[ rprorep ] the definition of the @xmath142-operator proposed in @xcite , @xmath246 which is valid for @xmath247 , @xmath248 , coincides with our definition ( [ rdef ] ) .
@xmath249 @xmath227 and @xmath228 admit the following representation : @xmath250 where @xmath251 .
( [ efphi ] ) for @xmath252 yield @xmath253 \ , w_b(-{{\mathsf p}}_2 ) = 2\pi b({{\mathsf x}}_{{\mathfrak 1}}-{{\mathsf x}}_{{\mathfrak 2 } } ) + i b \partial_t w_b(t)\bigm|_{t={{\mathsf p}}_2 } \ , \frac{1}{w_b({{\mathsf p}}_2 ) } \end{aligned}\ ] ] and , analogously , @xmath254 .
exponentiating these relations , we obtain ( [ expef ] ) . alternatively , eqs .
( [ expef ] ) can be derived from ( [ efk2 ] ) with the help of the baker - campbell - hausdorff formula .
observe also that @xmath255 commutes with @xmath111 and @xmath138 .
we may now give the proof of lemma [ codual ] .
in lemma [ uvg ] , let us choose @xmath256 and @xmath257 .
in view of ( [ expef ] ) this implies the identification @xmath258 and @xmath259 .
therefore , we have @xmath260 the relation for @xmath261 in ( [ ded ] ) is proven similarly .
[ eft ] let the powers @xmath262 and @xmath263 with @xmath264 be defined on @xmath45 in the sense of ( [ efphi ] ) .
then we have @xmath265 = [ \alpha]_q \ , [ 2 { { \mathsf h}}+ \alpha -1 ] _ q \ , { { \mathsf f}}^{\alpha-1 } \ , , \quad [ { { \mathsf e}}^{\alpha } , { { \mathsf f } } ] = [ \alpha]_q \ , [ 2 { { \mathsf h}}- \alpha + 1 ] _ q \ , { { \mathsf e}}^{\alpha-1 } \,,\ ] ] where the @xmath26-numbers are defined as in theorem [ t2 ] .
@xmath266 & { \stackrel{(\ref{efw})}{= } } \bigl [ w_b({{\mathsf p}}_2 ) \ , \frac{e^{2\pi b({{\mathsf x}}_{{\mathfrak 1}}-{{\mathsf x}}_{{\mathfrak 2}})}}{2\sin\pi b^2 } \ , w_b(-{{\mathsf p}}_2 ) \ , , \ , w_b({{\mathsf p}}_1 ) \
, \frac{e^{2\pi \alpha b({{\mathsf x}}_{{\mathfrak 2}}-{{\mathsf x}}_{{\mathfrak 1}})}}{(2\sin\pi b^2)^\alpha } \ , w_b(-{{\mathsf p}}_1 ) \bigr ] \\ & { \stackrel{(\ref{fe})}{= } } \bigl ( \frac{w_b({{\mathsf p}}_1 + ib)\ , w_b({{\mathsf p}}_2 ) } { w_b({{\mathsf p}}_1)\ , w_b({{\mathsf p}}_2 - i b ) } - \frac{w_b({{\mathsf p}}_1 + ( 1-\alpha ) ib)\ ,
w_b({{\mathsf p}}_2 + \alpha ib ) } { w_b({{\mathsf p}}_1 - \alpha ib)\ , w_b({{\mathsf p}}_2 + ( \alpha-1 ) ib ) } \bigr ) \\ & \phantom{=== } \times \ , w_b({{\mathsf p}}_1 ) \ , \frac{e^{2\pi ( \alpha-1 ) b({{\mathsf x}}_{{\mathfrak 2}}-{{\mathsf x}}_{{\mathfrak 1 } } ) } } { ( 2\sin\pi b^2)^{\alpha+1 } } \ , w_b(-{{\mathsf p}}_1 ) \\ & { \stackrel{(\ref{wfunrel})}{= } } \bigl ( \bigl[{{\textstyle \frac{q}{2b } } } - { { \textstyle \frac{i { { \mathsf p}}_1}{b } } } \bigr]_q \ , \bigl[{{\textstyle \frac{q}{2b } } } + { { \textstyle \frac{i { { \mathsf p}}_2}{b } } } \bigr]_q - \bigl[{{\textstyle \frac{q}{2b } } } -\alpha - { { \textstyle \frac{i { { \mathsf p}}_1}{b } } } \bigr]_q \ , \bigl[{{\textstyle \frac{q}{2b } } } -\alpha + { { \textstyle \frac{i { { \mathsf p}}_2}{b } } } \bigr]_q \bigr ) \ , { { \mathsf f}}^{\alpha-1 } \\ & = [ \alpha]_q \ , \bigl[{{\textstyle \frac{q}{b } } } -\alpha + { { \textstyle \frac{i}{b } } } ( { { \mathsf p}}_2 - { { \mathsf p}}_1 ) \bigr]_q \ , { { \mathsf f}}^{\alpha-1 } = [ \alpha]_q \ , \bigl [ { { \textstyle \frac{1}{ib } } } ( { { \mathsf p}}_2 - { { \mathsf p}}_1 ) + \alpha -1 \bigr]_q \ , { { \mathsf f}}^{\alpha-1 } \\ & { \stackrel{(\ref{efk2})}{= } } [ \alpha]_q \ , \bigl [ 2 { { \mathsf h}}+\alpha -1 \bigr]_q \ , { { \mathsf f}}^{\alpha-1 } \ , . \end{aligned}\ ] ] in the fourth line we used the definition of @xmath26-number ; in the fifth line we used the identities @xmath267_q [ y]_q - [ x-\alpha]_q [ y-\alpha]_q = [ \alpha]_q [ x+y-\alpha]_q$ ] and @xmath268_q = -[t]_q$ ] . the second formula in ( [ eft ] ) is derived analogously . [ qexp ]
let @xmath193 and @xmath194 be self - adjoint operators such that @xmath195 = 2 \pi i$ ] .
then for the function @xmath136 defined in ( [ gg ] ) we have @xmath269 where @xmath270 , @xmath271 and @xmath272 .
furthermore , ( [ qexp ] ) @xmath273 ( [ pent ] ) . in the literature ,
( [ qexp ] ) and ( [ pent ] ) are often referred to as the quantum exponential and the quantum pentagon relations .
they also hold for the function @xmath274 which is the compact counterpart of @xmath136 . for @xmath275 ,
the quantum exponential relation has been known since long time @xcite and the quantum pentagon relation was found in @xcite . since ( [ qexp ] ) and ( [ pent ] )
are equivalent , it suffices to prove one of them .
proofs of the quantum pentagon relation were given in @xcite and @xcite .
nevertheless , we find it instructive to give another proof of the quantum exponential relation in appendix b since it will allow us to introduce the notion of @xmath9-binomial coefficients .
[ rint ] @xmath142 and @xmath276 may be decomposed into powers of @xmath277 as follows : @xmath278 where the integration contour goes above the pole at @xmath279 . by lemma 15 in @xcite
( see also @xcite@xcite ) we have : @xmath280 the function @xmath281 has a pole at @xmath216 and is analytic in the upper half - plane .
the integration contour in ( [ bbeta ] ) goes above this pole . considering the asymptotics of ( [ bbeta ] ) for @xmath282 and @xmath283 , using the properties ( [ grefl ] ) and ( [ gasym ] ) , and making a change of variables , we obtain the following fourier transformation formulae @xmath284 lemma [ rint ] follows if we put here @xmath285 and compare the result with the definition ( [ rdef ] ) .
we now come to the proof of theorem [ tint ] .
consider the product representation of the @xmath142-operator of lemma [ rprod ] projected to @xmath151 by means of the reduction described in subsection [ ch ] , @xmath286 it is now easy to compute the `` matrix elements '' of @xmath287 on the states @xmath288 , where @xmath289 .
taking into account that @xmath290 and @xmath291 , we find @xmath292 which gives us the kernel ( [ ker2 ] ) of the `` momentum '' representation in theorem [ tint ] . the kernel of the `` coordinate '' representation ( [ io1 ] ) can be obtained as a fourier transform of ( [ skk1 ] ) : @xmath293 the remaining integrations are performed by using relation ( [ bbeta ] ) three times .
the result of this straightforward but tedious calculation is given by ( [ ker1 ] ) .
first , it is straightforward to write out @xmath294 and @xmath295 in terms of the clebsch - gordan maps @xmath99 $ ] : @xmath296 where we have introduced @xmath297\cdot\big({{\rm id}}{\otimes}{{\mathsf c}}[s_{21}|s_2,s_1]\big),\\ { { \mathsf c}}_{(32)1}(s_{32})\;\equiv\ ; & { { \mathsf c}}[s_4|s_{32},s_{1}]\cdot\big({{\mathsf c}}[s_{32}|s_3,s_2]{\otimes}{{\rm id}}\big ) .
\end{aligned}\ ] ] proposition 7 in @xcite is equivalent to @xmath298 where @xmath299 are the b - racah - wigner coefficients introduced in @xcite .
it follows that @xmath300 exchanging the integrations over @xmath301 and @xmath302 , @xmath303 , and using formula ( 89 ) from @xcite , @xmath304 yields the claim . we shall consider the left invariant haar measure @xmath119 only , the proof for the case of @xmath120 being completely analogous .
a few preparations are in order .
the elements of @xmath121 can be represented as integral operators : if a vector @xmath305 is realized by a function @xmath306 , then @xmath307 in terms of the kernel @xmath308 one may write the defintion of @xmath119 as @xmath309 the distributional matrix elements of an operator @xmath310 are always of the form @xmath311 by using an analogous notation for operators in @xmath312 one may represent the distributional matrix elements of @xmath313 as @xmath314}\big)^ * { \big [ \begin{smallmatrix } s_3 & s_2 & s_1 \\ k_3 ' & k_2 ' & k_1 ' \end{smallmatrix}\big ] } .
\end{aligned}\ ] ] in order to make the justification for the following manipulations more transparent , we are going to employ the following regularization for the distributions involved : @xmath315}\;=\;\lim_{{\epsilon}\downarrow 0}\ ; { \big [ \begin{smallmatrix } s_3 & s_2 & s_1 \\ k_3 & k_2 & k_1 \end{smallmatrix}\big]}_{{\epsilon}},\qquad { \big [ \begin{smallmatrix } s_3 & s_2 & s_1 \\ k_3 & k_2 & k_1 \end{smallmatrix}\big]}_{{\epsilon}}\;=\ ; e^{-{\epsilon}\sum_{i=1}^3|k_i|}{\delta}_{{\epsilon}}(k_3-k_2-k_1 ) { c_{s_3}{\big [ \begin{smallmatrix } s_2 & s_1 \\ k_2 & k_1 \end{smallmatrix}\big]}},\ ] ] where @xmath316 is a symmetric regularization of the delta - distribution .
let us furthermore note that it suffices to check the invariance property on a dense subset @xmath317 of the domain of @xmath119 .
consider the matrix element @xmath318 where @xmath319 , @xmath320 are smooth functions with compact support .
assuming that @xmath308 has exponential decay w.r.t . @xmath170 and
@xmath321 it is not difficult to show that the matrix element @xmath322 will also have exponential decay w.r.t .
@xmath323 , @xmath324 that can be made as large as one likes by choosing the subset @xmath325 appropriately .
combining ( [ haarrep ] ) and ( [ corep ] ) leads to the following representation for the distributional matrix elements of @xmath326 . @xmath327}\big)^
* { \big [ \begin{smallmatrix } s_3 & s_2 & s_1 \\ k_3 ' & k_2 ' & k_1 \end{smallmatrix}\big ] }
. \end{aligned}\ ] ] we are going to use the following result : the following equation holds as an identity between tempered distributions : @xmath328}\big)^ * & { \big [ \begin{smallmatrix } s_3 & s_2 & s_1 \\ k_3 ' & k_2 ' & k_1 \end{smallmatrix}\big ] } \;=\;\\ & = \ ; { \delta}(k_3'-k_3){\delta}(k_2'-k_2 ) .
\end{aligned}\ ] ] the proof of the proposition will be based on the following important symmetries of the clebsch - gordan kernel : the clebsch - gordan kernel @xmath329}_{{\epsilon}}$ ] has the following symmetries .
@xmath330}_{{\epsilon}}\big)^*\;=\ ; e^{+\pi q ( k_1-k_3)}e^{-\pi i h_{s_2 } } { \big [ \begin{smallmatrix } s_1 & -s_2 & s_3 \\ -k_1 & -k_2 & -k_3 \end{smallmatrix}\big]}_{{\epsilon}},\\ \big({\big [ \begin{smallmatrix } s_3 & s_2 & s_1 \\ k_3 & k_2 & k_1 \end{smallmatrix}\big]}_{{\epsilon}}\big)^*\;=\ ; e^{-\pi q ( k_2-k_3)}e^{-\pi i h_{s_1 } } { \big [ \begin{smallmatrix } s_2 & s_3 & -s_1 \\ -k_2 & -k_3 & -k_1 \end{smallmatrix}\big]}_{{\epsilon}}. \end{aligned}\ ] ] one may verify directly that for @xmath331 , @xmath332 , @xmath333}\big)^*\;=\ ; e^{-\pi i h_{s_2}}{\big [ \begin{smallmatrix } s_1 & -s_2 & s_3 \\ x_1^*-i\frac{q}{2 } & x_2^ * & x_3^*-i\frac{q}{2 } \end{smallmatrix}\big]},\\ \big({\big [ \begin{smallmatrix } s_3 & s_2 & s_1 \\ x_3 & x_2 & x_1 \end{smallmatrix}\big]}\big)^*\;=\ ; e^{-\pi i h_{s_1}}{\big [ \begin{smallmatrix } s_2 & s_3 & -s_1 \\ x_2^*+i\frac{q}{2 } & x_3^*+i\frac{q}{2 } & x_1^ * \end{smallmatrix}\big]}. \end{aligned}\ ] ] the lemma follows by taking the fourier - transformation of ( [ cgcxsymm ] ) , taking into account that our regularization is compatible with the symmetry ( [ cgcksymm ] ) . with the help of equation ( [ cgcksymm ] )
we may rewrite the left hand side of ( [ dist_id ] ) as follows : @xmath334}_{{\epsilon}}\big)^ * { \big [ \begin{smallmatrix } s_1 & -s_2 & s_3 \\ -k_1 & -k_2 ' & -k_3 ' \end{smallmatrix}\big]}_{{\epsilon}}\ ] ] the proposition now follows by using the fourier - transform of ( * ? ? ?
* corollary 1 ) . inserting ( [ dist_id ] ) into ( [ inv_m_el ] )
yields @xmath335 recognizing the definition of the haar - measure on the right hand side completes the proof of the left invariance property of @xmath119 . to prove the property ( ii ) in theorem 2
we observe that definition ( [ hm ] ) can be rewritten as @xmath336 now it is straightforward to verify the first formula in ( [ hinv2 ] ) for @xmath337 using the definition of the adjoint action , the relations ( [ def ] ) , and the cyclicity of trace . for instance , @xmath338 , @xmath339 .
further , we notice that @xmath340 .
together with the linearity of trace this implies that ( [ hinv2 ] ) extends to any element of @xmath0 .
let us adopt the following notations : @xmath341 , @xmath342 , and @xmath343 .
first , we compute with the help of lemma [ eft ] : @xmath344 \ , q^{{{\mathsf h}}{\otimes}{{\mathsf h } } } { \stackrel{(\ref{eft})}{= } } - q^{{{\mathsf h}}{\otimes}{{\mathsf h } } } \bigl ( [ i t]_q [ 2 { { \mathsf h}}_2 + it -1]_q \ , { { \mathsf e}}_1^{i t } \ , { { \mathsf f}}_2^{i t-1 } \bigr ) q^{{{\mathsf h}}{\otimes}{{\mathsf h } } } \,.\end{aligned}\ ] ] next , we find @xmath345_q \ , { { \mathsf e}}_1^{i t+1 } \ , { { \mathsf f}}_2^{i t } \bigr ) \ , q^{{{\mathsf h}}{\otimes}{{\mathsf h } } } \,.\end{aligned}\ ] ] let us write down the integral representation ( [ rint1 ] ) of @xmath142 in the following form @xmath346 where @xmath347 . observe that ( [ gfunrel ] ) implies that @xmath348 satisfies the following functional equation : @xmath349_q \ , \rho(t - i ) = ( 2 i\ , q^{i t } \ , \sin \pi b^2 ) \ , \rho(t ) \,.\ ] ] adding ( [ ia ] ) with ( [ ib ] ) , we derive @xmath350_q \ , { { \mathsf e}}_1^{i t+1 } \ , { { \mathsf f}}_2^{i t } \\ & \qquad\qquad\qquad\qquad - [ i t]_q [ 2 { { \mathsf h}}_2 + it -1]_q \ , { { \mathsf e}}_1^{i t } \ , { { \mathsf f}}_2^{i t-1 } \bigr ) \ , q^{{{\mathsf h}}{\otimes}{{\mathsf h } } } \\ & = \int_{{{\mathbb r}}}dt\ , q^{{{\mathsf h}}{\otimes}{{\mathsf h } } } \ , \bigl ( ( 2 i \ , q^{i t } \ , \sin \pi b^2 ) \ , \rho(t ) \\ & \qquad\qquad\qquad\qquad - [ i t+1]_q \ , \rho(t -i ) \bigr ) [ 2 { { \mathsf h}}_2 + it]_q \ , { { \mathsf e}}_1^{i t+1 } \ , { { \mathsf f}}_2^{i t } \ , q^{{{\mathsf h}}{\otimes}{{\mathsf h } } } { \stackrel{(\ref{req})}{= } } 0 \ , . \end{aligned}\ ] ] thus , we have proven ( [ i ] ) for @xmath351 .
the proof for @xmath43 goes along the same lines with the help of the second formula in ( [ eft ] ) . and
for @xmath44 the proof is trivial because @xmath352 commutes with @xmath353 .
+ _ * property ( ii ) * _ recall that the rescaled generators @xmath354 and @xmath355 were introduced before lemma [ rprod ] . to prove the first formula in ( [ ii ] ) , we use the quantum exponential relation ( [ qexp ] ) from lemma [ qexp ] with identification and , @xmath356 the second formula in ( [ ii ] ) is proved in the same way .
+ _ * property ( iii ) * _ first , we derive @xmath357 this means that , acting with @xmath358 on the r.h.s . of ( [ rint1 ] ) , we obtain the r.h.s . of ( [ rint2 ] ) .
thus , we have proven the first formula in ( [ ii ] ) .
the second formula is verified analogously . finally ,
acting with @xmath359 on the last line in ( [ sr ] ) and performing similar manipulations , we find that @xmath360 which together with ( [ sr ] ) implies the last formula in ( [ iii ] ) .
we first need to discuss the analytic continuation of @xmath361 to the values @xmath362 , @xmath363 . to begin with ,
one may trivially perform e.g. the integral over @xmath364 to get an expression of the form @xmath365 where @xmath366 .
the analytic continuation of ( [ r - act ] ) to @xmath362 can be defined by deforming the contour of integration over @xmath324 , @xmath367 , in ( [ r - act ] ) into @xmath368 plus a sum of small circles around the poles from the factor @xmath369 in @xmath370 that lie between @xmath367 and @xmath371 .
we are now in the position to take the limit @xmath372 .
the factor @xmath373 that appears in @xmath370 , cf .
( [ ker2 ] ) , makes most of the terms vanish except for the terms from the poles at @xmath374 , @xmath375 .
the resulting expression is of the following form .
@xmath376 where @xmath377 and @xmath378 . it remains to calculate the relevant residues .
it is easy to derive from ( [ gdual])-([grefl ] ) that @xmath379 hence @xmath380 .
in fact , using the modular property of the dedekind @xmath381-function , it is straightforward to compute the limit directly for the product representation ( [ gprod ] ) ( as was done in @xcite ) ; which yields @xmath382 hence , taking into account the properties ( [ gdual])-([grefl ] ) , we find that @xmath383 where @xmath384 and @xmath385 .
to complete the proof of theorem [ t2 ] is now the matter of a straightforward calculation using the functional relation ( [ wfunrel ] ) , formula ( [ gres ] ) , as well as ( [ deltatrf ] ) .
let us first note that the left hand side of ( [ rdiag ] ) satisfies the intertwining property @xmath386\;{{\mathsf b}}_{s_{{{\mathfrak 2}}}s_{{{\mathfrak 1}}}}^{}\circ ( \pi_{s_{{{\mathfrak 2}}}}{\otimes}\pi_{s_{{{\mathfrak 1}}}})\circ{\delta}(x)\,=\ , \pi_{s_{{\mathfrak 3}}}(x)\circ{{\mathsf c}}[s_3|s_1,s_2]\ ; { { \mathsf b}}_{s_{{{\mathfrak 2}}}s_{{{\mathfrak 1}}}}^ { } \,.\ ] ] a unitary operator that maps @xmath387 and satisfies ( [ intertw ] ) must be proportional to @xmath99 $ ] .
this is a consequence of the analysis used to prove theorem 2 in @xcite .
it follows that there _ exists _ a function @xmath176 such that the statement of theorem [ t3 ] holds .
we are left with the task to calculate @xmath176 explicitly . to this aim
let us note that theorem [ t3 ] is equivalent to an identity between meromorphic functions . to write this identity down ,
let us assume that @xmath151 is realized by functions @xmath388 .
@xmath389 $ ] is then realized as an integral operator : @xmath390\tilde{\psi}\big)(k_3)\;=\;\int_{{{\mathbb r}}}dk_2dk_1 \;{\big [ \begin{smallmatrix } s_3 & s_2 & s_1 \\ k_3 & k_2 & k_1 \end{smallmatrix}\big]}\tilde{\psi}(k_2,k_1).\ ] ] the explicit expression for the distributional kernel @xmath329}$ ] can be found in appendix a. for the moment it will be enough to note that it can be factorized as @xmath391}\;=\ ; { \delta}(k_1+k_2-k_3)\,{c_{s_3}{\big [ \begin{smallmatrix } s_2 & s_1 \\ k_2 & k_1 \end{smallmatrix}\big]}},\ ] ] where @xmath392}}$ ] is a meromorphic function by lemma 20 of @xcite .
it is then easy to see that theorem [ t3 ] is equivalent to the identity @xmath393}\ ; \tilde{{{\mathsf r}}}_{s_2s_1}(k_2,k_1|k'_2,k'_1)\;=\ ; { \omega}(s_3|s_2,s_1){\big [ \begin{smallmatrix } s_3 & s_2 & s_1 \\ k_3 & k_2 ' & k_1 ' \end{smallmatrix}\big]}.\ ] ] in order to see that equation ( [ diagcomp ] ) is indeed equivalent to an identity between meromorphic functions let us note that ( [ cgcfact ] ) and ( [ ker2 ] ) allow one to split off the distributional factors .
what remains on the left hand side is a convolution of two meromorphic functions , so is itself meromorphic ( cf .
lemma 3 in @xcite ) .
let us note that both sides of ( [ diagcomp ] ) have a pole at @xmath394 . in the case of the right hand side
this is a consequence of lemma 20 of @xcite .
concerning the left hand side of ( [ diagcomp ] ) one may as in the proof of lemma 3 of @xcite identify the above - mentioned pole as the consequence of the pinching of the contour of integration by a collision of two poles of the integrand .
first we have the pole of @xmath395}$ ] at @xmath396 .
second let us note that the factor @xmath397 appearing in @xmath398 produces a pole at @xmath399 if one takes into account that @xmath398 has support only for @xmath400 .
the residue of the resulting pole on the left hand side of ( [ diagcomp ] ) is simply given by the product of the relevant residues of @xmath395}$ ] and @xmath398 respectively .
the equality of the residues of the two sides of equation ( [ diagcomp ] ) implies the following identity : @xmath401}}\ ; e^{\pi qk_1 } e^{2\pi is_2k_1}\;=\ ; { \omega}(s_3|s_2,s_1)\operatorname*{res}_{k_2=-s_2+i\frac{q}{2 } } { c_{s_3}{\big [ \begin{smallmatrix } s_2 & s_1 \\ k_2 & k_1 \end{smallmatrix}\big]}}\ ] ] by evaluating the relevant residues we may therefore calculate @xmath176 .
@xmath402 } } \;=\ ; & e^{-\frac{\pi i}{2}(h_{s_3}-h_{s_2}-h_{s_1})}e^{-\frac{\pi}{2}qk_1 } e^{-\pi is_2k_1 } \frac{w_b(k_1-s_1)w_b(s_3+s_1-s_2)}{w_b(k_1+s_3-s_2+i\frac{q}{2})},\\ 2\pi i\!\!\!\!\!\!\!\operatorname*{res}_{k_2=-s_2+i\frac{q}{2}}{c_{s_3}{\big [ \begin{smallmatrix } s_2 & s_1 \\ k_2 & k_1 \end{smallmatrix}\big ] } } \;=\ ; & e^{+\frac{\pi i}{2}(h_{s_3}-h_{s_2}-h_{s_1})}e^{+\frac{\pi}{2}qk_1 } e^{+\pi is_2k_1 } \frac{w_b(k_1-s_1)w_b(s_3+s_1-s_2)}{w_b(k_1+s_3-s_2+i\frac{q}{2})}. \end{aligned}\ ] ] in order to exhibit the singular behavior of @xmath403}}$ ] near @xmath396 one may deform the contour of integration in ( [ cgcft ] ) into the union of a small circle around the pole of the integrand at @xmath404 and a contour that separates the pole at @xmath404 from all the other poles in the _ upper _ half - plane , approaching asymptotically @xmath405 .
the contribution from the residue of the pole at @xmath404 exhibits the pole at @xmath396 explicitly , whereas the rest is nonsingular . similarly , to analyze the singular behavior of @xmath392}}$ ] near @xmath396
one needs to deform the contour in ( [ cgcft ] ) into a small circle around the pole at @xmath406 together with a contour separating that pole from all the other poles in the _ lower _ half plane .
it is then straightforward to calculate the values of the corresponding residues from ( [ cgcft ] ) .
define a distributional kernel @xmath407}$ ] ( the `` clebsch - gordan coefficients '' ) by an expression of the form @xmath408}\;\equiv \ ; \lim_{{\epsilon}\downarrow 0}\ ; { \big [ \begin{smallmatrix } s_3 & s_2 & s_1 \\ x_3 & x_2 & x_1 \end{smallmatrix}\big]}_{{\epsilon}}^{},\ ] ] where the meromorphic function @xmath407}_{{\epsilon}}^{}$ ] is defined as @xmath409}_{{\epsilon } } \;= \ ; & e^{-\frac{\pi i}{2}(h_{s_3}-h_{s_2}-h_{s_1})}\\ & \qquad{\times}d_b({\sigma}_{32};y_{32}+i{\epsilon})d_b({\sigma}_{31};y_{31}+i{\epsilon})d_b({\sigma}_{21};y_{21}+i{\epsilon } ) , \end{aligned}\ ] ] @xmath410 , the distribution @xmath411 is defined in terms of the function @xmath412 as @xmath413 and the coefficients @xmath414 , @xmath415 , @xmath416 are given by @xmath417 it is often useful to consider the fourier - transform of the b - clebsch - gordan symbols defined by @xmath418}\;=\ ; \int_{{{\mathbb r}}}dx_3dx_2dx_1\;e^{2\pi i \sum_{l=1}^3 k_lx_l } { \big [ \begin{smallmatrix } s_3 & s_2 & s_1 \\ x_3 & x_2 & x_1 \end{smallmatrix}\big]}\ ] ] the distribution @xmath329}$ ] can be factorized as @xmath315}\;=\ ; { \delta}(k_1+k_2-k_3){c_{s_3}{\big [ \begin{smallmatrix } s_2 & s_1 \\ k_2 & k_1 \end{smallmatrix}\big]}},\ ] ] where @xmath392}}$ ] is a meromorphic function . a straightforward calculation using ( * ?
? * lemma 15 ) yields the following expression : @xmath419}}\;=\ ; \frac{e^{-\frac{\pi i}{2}{\beta}_{21}{\beta}}e^{\frac{q}{2}\pi(k_1-k_2)}e^{\pi i(k_1s_2-k_2s_1)}}{w_b({\sigma}_{32 } ) w_b({\sigma}_{31})w_b({\sigma}_{21 } ) } \int\limits_{{{\mathbb r}}+i0}ds\;e^{-\pi s{\beta } } \prod_{l=1}^3\frac{w_b(s+r_l)}{w_b(s+s_l ) } , \end{aligned}\ ] ] where we used the abbreviations @xmath420 , @xmath421 and @xmath422 the analytic properties of @xmath329}$ ] can be summarized as follows : [ cgpoles ] ( lemma 20 in @xcite ) @xmath329}$ ] depends meromorphically on all of its arguments , with poles at @xmath423 , @xmath424 , @xmath332 only .
the definition ( [ gg ] ) and the property ( [ gfunrel ] ) imply that the function @xmath136 obeys the following functional equation @xmath425 for a pair of weyl - type variables , @xmath426 , a consequence of ( [ gfun ] ) is @xmath427 it is now obvious that @xmath428 these relations allow us to prove the equivalence of ( [ qexp ] ) and ( [ pent ] ) stated in lemma [ qexp ] .
for instance , let us show that ( [ pent ] ) together with the first relation in ( [ guv ] ) implies ( [ qexp ] ) .
introduce @xmath429 .
notice that @xmath430 where @xmath431 so that @xmath432= 2\pi i$ ] .
then we have @xmath433 the inverse implication , ( [ pent ] ) @xmath434 ( [ qexp ] ) , is proven similarly .
now we want to prove ( [ qexp ] ) .
first , we represent @xmath435 in an integral form : @xmath436 where we introduced @xmath437 .
computing the integral over @xmath438 with the help of ( [ bbeta ] ) , we derive an analogue of the binomial formula : @xmath439 \tau \end{array } \!\!\ !
\bigr)_b } \ ,
u^{i(t-\tau ) } \ , v^{i\tau } \,,\ ] ] where the @xmath9-binomial coefficient is given by @xmath440 \tau \end{array } \!\!\ !
} = \frac{e^ { 2\pi i b^2 \tau ( t - \tau ) } \ , g_b(q+i b t ) } { g_b(q+ ibt - ib \tau ) \ , g_b(q + ib \tau ) } \,.\ ] ] we see that the function @xmath184 is a @xmath9-analogue of the factorial .
the @xmath9-binomial coefficients satisfy the @xmath26-pascal identity : @xmath441 \tau \end{array } \!\!\ !
= q^{-2 i\tau } \ , { \bigl ( \!\!\ ! \begin{array}{c }
t \\[-0.6 mm ] \tau \end{array } \!\!\ !
\bigr)_b }
+ { \bigl ( \!\!\ ! \begin{array}{c }
t \\[-0.6 mm ] \tau+i \end{array } \!\!\ !
\bigr)_b }
= { \bigl ( \!\!\ !
t \\[-0.6 mm ] \tau \end{array } \!\!\ !
+ q^{2i(\tau -t+i ) } \ , { \bigl ( \!\!\ ! \begin{array}{c }
t \\[-0.6 mm ] \tau+i \end{array } \!\!\ !
\bigr)_b}\ ] ] which can be easily verified with the help of ( [ gfunrel ] ) . using the @xmath9-binomial coefficients and the integral representation ( [ gg1 ] ) of @xmath136
, we derive the quantum exponential relation : @xmath442 \tau \end{array } \!\!\ !
\bigr)_b } \ , \frac { e^{-\pi i b^2 t^2}}{g_b(q+ibt ) } u^{i(t-\tau ) } v^{i\tau } \\ & { \stackrel{(\ref{bbc})}{= } } b^2 \ !
r}}dt \ , d\tau \ , \frac { e^{-\pi i b^2 ( t-\tau)^2 -\pi i b^2 \tau^2 } } { g_b(q+ib(t-\tau ) ) g_b(q+ib\tau ) } \ , u^{i(t-\tau ) } v^{i\tau } \\ & = b \ !
r}}dt \ , \frac { e^{-\pi i b^2 t^2 } } { g_b(q+ibt ) } \ , u^{it } \
r}}d\tau \ , \frac { e^{-\pi i b^2 \tau^2 } } { g_b(q+ib\tau ) } \ , v^{i\tau } { \stackrel{(\ref{gg1})}{= } } g_b(u ) \ , g_b(v ) \ , .
\end{aligned}\ ] ] this completes the proof of lemma [ qexp ] .
we are grateful to l.faddeev for useful comments .
we thank a.volkov for providing us with a draft version of [ v ] prior to its publication .
was supported by alexander von humboldt foundation . j.t . was supported by dfg sfb 288 .
a part of this work was carried out during a.b.s visit to the department of mathematics , university of york .
s. kharchev , d. lebedev and m. semenov - tian - shansky : _ unitary representations of @xmath443 , the modular double , and the multiparticle @xmath26-deformed toda chains _ , comm .
* 225 * ( 2002 ) 573609 [ hep - th/0102180 ] m.p .
schtzenberger : _ une interprtation de certaines solutions de lquation fonctionnelle : _ , c. r. acad .
paris * 236 * ( 1953 ) 352353 ; + j. cigler : _ operatormethoden fr @xmath26-identitten _ , monatsh .
* 88 * ( 1979 ) 87105 | a certain class of unitary representations of @xmath0 has the property of being simultanenously a representation of @xmath1 for a particular choice of @xmath2 .
faddeev has proposed to unify the quantum groups @xmath0 and @xmath1 into some enlarged object for which he has coined the name `` modular double '' .
we study the r - operator , the co - product and the haar - measure for the modular double of @xmath0 and establish their main properties . in particular it is shown that the clebsch - gordan maps constructed in @xcite diagonalize this r - operator .
msc : + keywords : non - compact quantum groups , modular double |
in modern theoretical physics particle interactions are described by gauge theories .
these theories are constructed by demanding that symmetries in the laws of physics should be local , rather than global , in character .
the clearest expositions of this principle are contained in quantum theory , where one initially constructs a lagrangian containing a global symmetry . in order to promote this to a local symmetry ,
the derivatives appearing in the lagrangian are modified so that they are unchanged in form by local transformations .
this is achieved by the introduction of fields with certain transformation properties ( ` gauge fields ' ) , and these fields are then responsible for inter - particle forces .
the manner in which the gauge fields couple to matter is determined by the ` minimal coupling ' procedure , in which partial ( or directional ) derivatives are replaced by covariant derivatives .
this is the general framework that has been applied so successfully in the construction of the standard model of particle physics , which accounts for the strong , weak and electromagnetic forces .
but what of gravity : can general relativity be formulated as a gauge theory ?
this question has troubled physicists for many years @xcite . the first work that recovered features of general relativity from a gauging argument was due to kibble @xcite , who elaborated on an earlier , unsuccessful attempt by utiyama @xcite .
kibble used the 10-component poincar group of passive infinitesimal coordinate transformations ( consisting of four translations and six rotations ) as the global symmetry group . by gauging this group and constructing a suitable lagrangian density for the gauge fields , kibble arrived at a set of gravitational field equations though not the einstein equations .
in fact , kibble arrived at a slightly more general theory , known as a ` spin - torsion ' theory .
the necessary modifications to einstein s theory to include torsion were first suggested by cartan @xcite , who identified torsion as a possible physical field .
the connection between quantum spin and torsion was made later @xcite , once it had become clear that the stress - energy tensor for a massive fermion field must be asymmetric @xcite .
spin - torsion theories are sometimes referred to as einstein cartan kibble
sciama theories .
kibble s use of passive transformations was criticised by hehl _
et al . _
@xcite , who reproduced kibble s derivation from the standpoint of active transformations of the matter fields .
_ also arrived at a spin - torsion theory , and it is now generally accepted that torsion is an inevitable feature of a gauge theory based on the poincar group .
the work of hehl _ et al . _
@xcite raises a further issue . in their gauge theory derivation hehl _
are clear that ` _ coordinates and frames are regarded as fixed once and for all , while the matter fields are replaced by fields that have been rotated or translated _ ' .
it follows that the derivation can only affect the properties of the matter fields , and not the properties of spacetime itself . yet , once the gauge fields have been introduced , the authors identify these fields as determining the curvature and torsion of a riemann cartan spacetime .
this is possible only if it is assumed from the outset that one is working in a riemann
cartan spacetime , and not in flat minkowski spacetime .
but the idea that spacetime is curved is one of the cornerstone principles of general relativity . that this feature must be introduced _ a priori _ , and is not derivable from the gauge theory argument , is highly undesirable
it shows that the principle of local gauge invariance must be supplemented with further assumptions before general relativity is recovered .
the conclusions are clear : classical general relativity must be modified by the introduction of a spin - torsion interaction if it is to be viewed as a gauge theory , and the gauge principle alone fails to provide a conceptual framework for general relativity as a theory of gravity . in this paper
we propose an alternative theory of gravity that is derived from gauge principles alone .
these gauge fields are functions of position in a single minkowski vector space .
but here we immediately hit a profound difficulty .
parameterising points with vectors implies a notion of a newtonian ` absolute space ' ( or spacetime ) and one of the aims of general relativity was to banish this idea .
so can we possibly retain the idea of representing points with vectors without introducing a notion of absolute space ?
the answer to this is yes we must construct a theory in which points are parameterised by vectors , but the physical relations between fields are independent of where the fields are placed in this vector space
. we must therefore be free to move the fields around the vector space in an arbitrary manner , without in any way affecting the physical predictions . in this way our abstract minkowski vector space will play an entirely passive role in physics , and what will remain are a set of ` intrinsic ' relations between spacetime fields at the same point . yet , once we have chosen a particular parameterisation of points with vectors , we will be free to exploit the vector space structure to the full , secure in the knowledge that any physical prediction arrived at is ultimately independent of the parameterisation .
the theory we aim to construct is therefore one that is invariant under arbitrary field displacements .
it is here that we make contact with gauge theories , because the necessary modification to the directional derivatives requires the introduction of a gauge field .
but the field required is not of the type usually obtained when constructing gauge theories based on lie - group symmetries .
the gauge field coupling is of an altogether different , though very natural , character .
however , this does not alter the fact that the theory constructed here is a gauge theory in the broader sense of being invariant under a group of transformations .
the treatment presented here is very different from that of kibble @xcite and hehl _ et al . _
these authors only considered infinitesimal translations , whereas we are able to treat arbitrary finite field displacements .
this is essential to our aim of constructing a theory that is independent of the means by which the positions of fields are parameterised by vectors .
once we have introduced the required ` position - gauge ' field , a further spacetime symmetry remains .
spacetime fields are not simply scalars , but also consist of vectors and tensors .
suppose that two spacetime vector fields are equated at some position .
if both fields are then rotated at a point , the same intrinsic physical relation is obtained .
we therefore expect that all physical relations should be invariant under local rotations of the matter fields , as well as displacements .
this is necessary if we are to achieve complete freedom from the properties of the underlying vector space we can not think of the vectors representing physical quantities as having direction defined relative to some fixed vectors in minkowski spacetime , but are only permitted to consider relations between matter fields .
achieving invariance under local rotations introduces a further gauge field , though now we are in the familiar territory of yang mills type interactions ( albeit employing a non - compact lie group ) .
there are many ways in which the gauge theory presented here offers both real and potential advantages over traditional general relativity .
as our theory is a genuine gauge theory , the status of physical predictions is always unambiguous any physical prediction must be extracted from the theory in a gauge - invariant manner . furthermore , our approach is much closer to the conventional theories of particle physics , which should ease the path to a quantum theory .
one further point is that discarding all notions of a curved spacetime makes the theory conceptually much simpler than general relativity .
for example , there is no need to deal with topics such as differentiable manifolds , tangent spaces or fibre bundles @xcite .
the theory developed here is presented in the language of ` _ geometric algebra _ ' @xcite .
any physical theory can be formulated in a number of different mathematical languages , but physicists usually settle on a language which they feel represents the ` optimal ' choice . for quantum field theory
this has become the language of abstract operator commutation relations , and for general relativity it is riemannian geometry . for our gauge theory of gravity there seems little doubt that geometric algebra is the optimal language available in which to formulate the theory .
indeed , it was partly the desire to apply this language to gravitation theory that led to the development of the present theory .
( this should not be taken to imply that geometric algebra can not be applied to standard general relativity it certainly can @xcite .
it has also been used to elaborate on utiyama s approach @xcite . ) to us , the use of geometric algebra is as central to the theory of gravity presented here as tensor calculus and riemannian geometry were to einstein s development of general relativity .
it is the language that most clearly exposes the structure of the theory .
the equations take their simplest form when expressed in geometric algebra , and all reference to coordinates and frames is removed , achieving a clean separation between physical effects and coordinate artefacts . furthermore , the geometric algebra development of the theory is entirely self - contained .
all problems can be treated without ever having to introduce concepts from other languages , such as differential forms or the newman
penrose formalism .
we realise , however , that the use of an unfamiliar language may deter some readers from exploring the main physical content of our theory which is of course independent of the language chosen to express it .
we have therefore endeavoured to keep the mathematical content of the main text to a minimum level , and have included appendices describing methods for translating our equations into the more familiar languages of tensor and spinor calculus .
in addition , many of the final equations required for applications are simple scalar equations .
the role of geometric algebra is simply to provide the most efficient and transparent derivation of these equations .
it is our hope that physicists will find geometric algebra a simpler and more natural language than that of differential geometry and tensor calculus .
this paper starts with an introduction to geometric algebra and its spacetime version the spacetime algebra .
we then turn to the gauging arguments outlined above and find mathematical expressions of the underlying principles .
this leads to the introduction of two gauge fields . at this point
the discussion is made concrete by turning to the dirac action integral .
the dirac action is formulated in such a way that internal phase rotations and spacetime rotations take equivalent forms .
gauge fields are then minimally coupled to the dirac field to enforce invariance under local displacements and both spacetime and phase rotations .
we then turn to the construction of a lagrangian density for the gravitational gauge fields .
this leads to a surprising conclusion .
the demand that the gravitational action be consistent with the derivation of the minimally - coupled dirac equation restricts us to a single action integral .
the only freedom that remains is the possible inclusion of a cosmological constant , which can not be ruled out on theoretical grounds alone .
the result of this work is a set of field equations that are completely independent of how we choose to label the positions of fields with a vector @xmath0 .
the resulting theory is conceptually simple and easier to calculate with than the metric - based theory of general relativity .
we call this theory ` _ _ gauge theory gravity _ _ ' ( gtg ) .
having derived the field equations , we turn to a discussion of measurements , the equivalence principle and the newtonian limit in gtg .
we end part i with a discussion of symmetries , invariants and conservation laws . in part
ii we turn to applications , concentrating mainly on time - dependent spherically - symmetric systems .
we start by studying perfect fluids and derive a simple set of first - order equations that describe a wide range of physical phenomena .
the method of derivation of these equations is new and offers many advantages over conventional techniques .
the equations are then studied in the contexts of black holes , collapsing matter and cosmology .
we show how a gauge can be chosen that affords a clear , global picture of the properties of these systems . indeed , in many cases one can apply simple ,
almost newtonian , reasoning to understand the physics .
for some of these applications the predictions of gtg and general relativity are identical , and these cases include all present experimental tests of general relativity . however , on matters such as the role of horizons and topology , the two theories differ .
for example , we show that the black - hole solutions admitted in gtg fall into two distinct time - asymmetric gauge sectors , and that one of these is picked out uniquely by the formation process .
this is quite different to general relativity , which admits eternal time - reverse symmetric solutions . in discussing differences between gtg and general relativity , it is not always clear what the correct general relativistic viewpoint is
. we should therefore be explicit in stating that what we intend when we talk about general relativity is the full , modern formulation of the subject as expounded by , for example , hawking & ellis @xcite and dinverno @xcite .
this includes ideas such as worm - holes , exotic topologies and distinct ` universes ' connected by black holes @xcite . after studying some solutions for the gravitational fields we turn to the properties of electromagnetic and dirac fields in gravitational backgrounds .
for example , we give field configurations for a charge held at rest outside a black hole . we show how these field lines extend smoothly across the horizon , and that the origin behaves as a polarisation charge .
this solution demonstrates how the global properties of the gravitational fields are relevant to physics outside the horizon , a fact that is supported by an analysis of the dirac equation in a black - hole background .
this analysis also provides a quick , though physically questionable , derivation of a particle production rate described by a fermi dirac distribution with the correct hawking temperature .
we end with a discussion of the implications of gauge - theory gravity for cosmology .
a study of the maxwell and dirac equations in a cosmological background reveals a number of surprising features .
in particular , it is shown that a non - spatially - flat universe does not appear homogeneous to dirac fields fermionic matter would be able to detect the ` centre ' of the universe if @xmath1 .
thus the only homogeneous cosmological models consistent with gtg are those that are spatially flat .
( this does not rule out spatially - flat universes with a non - zero cosmological constant . )
a concluding section summarises the philosophy behind our approach , and outlines some future areas of research .
there are many reasons for preferring geometric algebra to other languages employed in mathematical physics .
it is the most powerful and efficient language for handling rotations and boosts ; it generalises the role of complex numbers in two dimensions , and quaternions in three dimensions , to a scheme that efficiently handles rotations in arbitrary dimensions .
it also exploits the advantages of labelling points with vectors more fully than either tensor calculus or differential forms , both of which were designed with a view to applications in the intrinsic geometry of curved spaces .
in addition , geometric algebra affords an entirely _ real _ formulation of the dirac equation @xcite , eliminating the need for complex numbers .
the advantage of the real formulation is that internal phase rotations and spacetime rotations are handled in an identical manner in a single unifying framework . a wide class of physical theories
have now been successfully formulated in terms of geometric algebra .
these include classical mechanics @xcite , relativistic dynamics @xcite , dirac theory @xcite , electromagnetism and electrodynamics @xcite , as well as a number of other areas of modern mathematical physics @xcite . in every case
, geometric algebra has offered demonstrable advantages over other techniques and has provided novel insights and unifications between disparate branches of physics and mathematics .
this section is intended to give only a brief introduction to the ideas and applications of geometric algebra .
a fuller introduction , including a number of results relevant to this paper , is set out in the series of papers @xcite written by the present authors .
elsewhere , the books by doran & lasenby @xcite , hestenes @xcite and hestenes & sobczyk @xcite cover the subject in detail .
a number of other helpful introductory articles can be found , including those by hestenes @xcite , vold @xcite , and doran & lasenby @xcite .
the conference proceedings @xcite also contain some interesting and useful papers .
geometric algebra arose from clifford s attempts to generalise hamilton s quaternion algebra into a language for vectors in arbitrary dimensions @xcite .
clifford discovered that both complex numbers and quaternions are special cases of an algebraic framework in which vectors are equipped with a single associative product that is distributive over addition . with vectors represented by lower - case roman letters ( @xmath2 , @xmath3 ) , clifford s ` geometric product '
is written simply as @xmath4 .
a key feature of the geometric product is that the square of any vector is a scalar .
now , rearranging the expansion @xmath5 to give @xmath6 where the right - hand side of is a sum of squares and by assumption a scalar , we see that the symmetric part of the geometric product of two vectors is also a scalar .
we write this ` inner ' or ` dot ' product between vectors as @xmath7 the remaining antisymmetric part of the the geometric product represents the directed area swept out by displacing @xmath2 along @xmath3 .
this is the ` outer ' or ` exterior ' product introduced by grassmann @xcite and familiar to all who have studied the language of differential forms .
the outer product of two vectors is called a _ bivector _ and is written with a wedge : @xmath8 on combining and we find that the geometric product has been decomposed into the sum of a scalar and a bivector part , @xmath9 the innovative feature of clifford s product lies in its mixing of two different types of object : scalars and bivectors .
this is not problematic , because the addition implied by is precisely that which is used when a real number is added to an imaginary number to form a complex number . but why might we want to add these two geometrically distinct objects ?
the answer emerges from considering reflections and rotations .
suppose that the vector @xmath2 is reflected in the ( hyper)plane perpendicular to the unit vector @xmath10 .
the result is the new vector @xmath11 the utility of the geometric algebra form of the resultant vector , @xmath12 , becomes clear when a second reflection is performed .
if this second reflection is in the hyperplane perpendicular to the unit vector @xmath13 , then the combined effect is @xmath14 but the combined effect of two reflections is a rotation so , defining the geometric product @xmath15 as the scalar - plus - bivector quantity @xmath16 , we see that rotations are represented by @xmath17 here the quantity @xmath18 is called the ` _ reverse _ ' of @xmath16 and is obtained by reversing the order of all geometric products between vectors : @xmath19 the object @xmath16
is called a _
rotor_. rotors can be written as an even ( geometric ) product of unit vectors , and satisfy the relation @xmath20 .
the representation of rotations in the form has many advantages over tensor techniques . by defining @xmath21
we can write @xmath22 which relates the rotor @xmath16 directly to the plane in which the rotation takes place .
equation generalises to arbitrary dimensions the representation of planar rotations afforded by complex numbers .
this generalisation provides a good example of how the full geometric product , and the implied sum of objects of different types , can enter geometry at a very basic level .
the fact that equation encapsulates a simple geometric relation should also dispel the notion that clifford algebras are somehow intrinsically ` quantum ' in origin .
the derivation of has assumed nothing about the signature of the space being employed , so that the formula applies equally to boosts as well as rotations . the two - sided formula for a rotation will also turn out to be compatible with the manner in which observables are constructed from dirac spinors , and this is important for the gauge theory of rotations of the dirac equation which follows .
forming further geometric products of vectors produces the entire geometric algebra .
general elements are called ` multivectors ' and these decompose into sums of elements of different grades ( scalars are grade zero , vectors grade one , bivectors grade two _ etc .
multivectors in which all elements have the same grade are termed _ homogeneous _ and are usually written as @xmath23 to show that @xmath24 contains only grade-@xmath25 components .
multivectors inherit an associative product , and the geometric product of a grade-@xmath25 multivector @xmath23 with a grade-@xmath26 multivector @xmath27 decomposes into @xmath28 where the symbol @xmath29 denotes the projection onto the grade-@xmath25 component of @xmath30 .
the projection onto the grade-0 ( scalar ) component of @xmath30 is written @xmath31 .
the ` @xmath32 ' and ` @xmath33 ' symbols are retained for the lowest - grade and highest - grade terms of the series , so that @xmath34 which are called the interior and exterior products respectively . the exterior product is associative , and satisfies the symmetry property @xmath35 two further products can also be defined from the geometric product .
these are the scalar product @xmath36 and the commutator product @xmath37 the scalar product is commutative and satisfies the cyclic reordering property @xmath38 the scalar product and the interior product coincide when acting on two homogeneous multivectors of the same grade .
the associativity of the geometric product ensures that the commutator product satisfies the jacobi identity @xmath39 finally we introduce some further conventions .
throughout we employ the operator ordering convention that , _ in the absence of brackets , inner , outer , commutator and scalar products take precedence over geometric products_. thus @xmath40 means @xmath41 , not @xmath42 .
this convention helps to eliminate unwieldy numbers of brackets .
summation convention is employed throughout , except for indices that denote the grade of a multivector , which are not summed over .
natural units ( @xmath43 ) are used except where explicitly stated . throughout
we refer to a lorentz transformation ( _ i.e. _ a spatial rotation and/or boost ) simply as a ` rotation ' .
of central importance to this paper is the geometric algebra of spacetime , the _ spacetime algebra _ @xcite . to describe the spacetime algebra it is helpful to introduce a set of four orthonormal basis vectors @xmath44 , @xmath45 ,
satisfying @xmath46 the vectors @xmath44 satisfy the same algebraic relations as dirac s @xmath47-matrices , but they now form a set of four independent basis vectors for spacetime , not four components of a single vector in an internal ` spin - space ' . the relation between dirac s matrix algebra and the spacetime algebra
is described in more detail in appendix [ app - dirac ] , which gives a direct translation of the dirac equation into its spacetime algebra form .
a frame of timelike bivectors @xmath48 , @xmath49 is defined by @xmath50 and forms an orthonormal frame of vectors in the space relative to the @xmath51 direction .
the algebraic properties of the @xmath48 are the same as those of the pauli spin matrices , but they now represent an orthonormal frame of vectors in space , and not three components of a vector in spin - space .
the highest - grade element ( or ` pseudoscalar ' ) is denoted by @xmath52 and is defined as : @xmath53 the pseudoscalar satisfies @xmath54 . since we are in a space of even dimension , @xmath52 _ anti_commutes with odd - grade elements , and commutes only with even - grade elements . with these definitions , a basis for the 16-dimensional spacetime algebra is provided by @xmath55 geometric significance is attached to the above relations as follows .
an observer s rest frame is characterised by a future - pointing timelike ( unit ) vector .
if this is chosen to be the @xmath51 direction then the @xmath51-vector determines a map between spacetime vectors @xmath56 and the even subalgebra of the spacetime algebra via @xmath57 the ` relative vector ' @xmath58 can be decomposed in the @xmath48 frame and represents a spatial vector as seen by an observer in the @xmath51-frame .
since a vector appears to an observer as a line segment existing for a period of time , it is natural that what an observer perceives as a vector should be represented by a spacetime bivector .
equation embodies this idea , and shows that the algebraic properties of vectors in relative space are determined entirely by the properties of the fully relativistic spacetime algebra .
the split of the six spacetime bivectors into relative vectors and relative bivectors is a frame - dependent operation different observers see different relative spaces .
this fact is clearly illustrated with the faraday bivector @xmath59 .
the ` space - time split ' @xcite of @xmath59 into the @xmath51-system is made by separating @xmath59 into parts that anticommute and commute with @xmath51 .
thus @xmath60 both @xmath61 and @xmath62 are spatial vectors in the @xmath51-frame , and @xmath63 is a spatial bivector .
equation decomposes @xmath59 into separate electric and magnetic fields , and the explicit appearance of @xmath51 in the formulae for @xmath61 and @xmath62 shows how this split is observer - dependent .
the identification of the algebra of three - dimensional space with the even subalgebra of the spacetime algebra necessitates a convention that articulates smoothly between the two algebras .
for this reason relative ( or spatial ) vectors in the @xmath51-system are written in bold type to record the fact that they are in fact spacetime bivectors .
this distinguishes them from spacetime vectors , which are left in normal type .
further conventions are introduced where necessary .
many of the derivations in this paper employ the vector and multivector derivatives @xcite . before defining these , however , we need some simple results for vector frames .
suppose that the set @xmath64 form a vector frame .
the reciprocal frame is determined by @xcite @xmath65 where @xmath66 and the check on @xmath67 denotes that this term is missing from the expression .
the @xmath64 and @xmath68 frames are related by @xmath69 an arbitrary multivector @xmath70 can be decomposed in terms of the @xmath68 frame into @xmath71 suppose now that the multivector @xmath59 is an arbitrary function of some multivector argument @xmath72 , @xmath73 .
the derivative of @xmath59 with respect to @xmath72 in the @xmath24 direction is defined by @xmath74 from this the multivector derivative @xmath75 is defined by @xmath76 this definition shows how the multivector derivative @xmath75 inherits the multivector properties of its argument @xmath72 , as well as a calculus from equation .
most of the properties of the multivector derivative follow from the result that @xmath77 where @xmath78 is the projection of @xmath24 onto the grades contained in @xmath72 .
leibniz rule is then used to build up results for more complicated functions ( see appendix [ app - results ] ) .
the multivector derivative acts on the next object to its right unless brackets are present ; for example in the expression @xmath79 the @xmath75 acts only on @xmath24 , but in the expression @xmath80 the @xmath75 acts on both @xmath24 and @xmath70 . if the @xmath75 is intended to act only on @xmath70 then this is written as @xmath81 , the overdot denoting the multivector on which the derivative acts . as an illustration , leibniz rule can be written in the form @xmath82 the overdot notation neatly encodes the fact that , since @xmath75 is a multivector , it does not necessarily commute with other multivectors and often acts on functions to which it is not adjacent .
we frequently make use of the derivative with respect to a general vector to perform a range of linear algebra operations . for such operations
the following results are useful : @xmath83 where @xmath10 is the dimension of the space ( @xmath84 for all the applications considered here ) .
further results are given in appendix [ app - results ] .
the derivative with respect to spacetime position @xmath0 is called the _ vector derivative _ , and is of particular importance
. suppose that we introduce an arbitrary set of spacetime coordinates @xmath85 .
these define a coordinate frame @xmath86 the reciprocal frame is denoted @xmath87 , and from this we define the vector derivative @xmath88 by @xmath89 ( the importance of the vector derivative means that it is sensible to give it a unique symbol . ) the vector derivative inherits the algebraic properties of a vector in the spacetime algebra .
the usefulness of the geometric product for the vector derivative is illustrated by electromagnetism . in tensor notation ,
maxwell s equations are @xmath90 } = 0.\ ] ] these have the spacetime algebra equivalents @xcite @xmath91 but we can utilise the geometric product to combine these into the single equation @xmath92 the great advantage of the @xmath88 operator is that it possesses an inverse , so a first - order propagator theory can be developed for it @xcite .
this is not possible for the separate @xmath93 and @xmath94 operators .
geometric algebra offers many advantages over tensor calculus in developing the theory of linear functions @xcite .
a linear function mapping vectors to vectors is written in a sans - serif font , @xmath95 , where @xmath2 is an arbitrary arguments and plays the role of a placeholder . the argument of a linear function
is usually assumed to be a constant vector , unless stated otherwise .
the adjoint function is written with an overbar , @xmath96 , and is defined such that @xmath97 it follows that @xmath98 we will frequently employ the derivative with respect to the vectors @xmath2 , @xmath3 _ etc . _ to perform algebraic manipulations of linear functions , as in equation .
the advantage is that all manipulations are then frame - free . of course ,
the @xmath99 and @xmath2 vectors can easily be replaced by the sum over a set of frame vectors and their reciprocals , if desired .
a symmetric function is one for which @xmath100 . for such functions @xmath101
it follows that for symmetric functions @xmath102 which is equivalent to the statement that @xmath100 .
linear functions extend to act on multivectors via @xmath103 so that @xmath104 is now a grade - preserving linear function mapping multivectors to multivectors .
in particular , since the pseudoscalar @xmath52 is unique up to a scale factor , we can define @xmath105 viewed as linear functions over the entire geometric algebra , @xmath104 and @xmath106 are related by the fundamental formulae @xmath107 which are easily derived ( * ? ? ?
* chapter 3 ) .
the formulae for the inverse functions are found as special cases of , @xmath108 a number of further results for linear functions are contained in appendix [ app - results ] .
these include a coordinate - free formulation of the derivative with respect to a linear function , which proves to be very useful in deriving stress - energy tensors from action integrals .
in this section we identify the dynamical variables that will describe gravitational interactions .
we start by reviewing the arguments outlined in the introduction .
the basic idea is that all physical relations should have the generic form @xmath109 , where @xmath2 and @xmath3 are spacetime fields representing physical quantities , and @xmath0 is the spacetime position vector .
an equality such as this can certainly correspond to a clear physical statement .
but , considered as a relation between fields , the physical relationship expressed by this statement is completely independent of where we choose to think of @xmath0 as lying in spacetime . in particular , we can associate each position @xmath0 with some new position @xmath110 and rewrite the relation as @xmath111 , and the equation still has precisely the same content .
( a proviso , which will gain significance later , is that the map @xmath112 should be non - singular and cover all of spacetime . )
a similar argument applies to rotations .
the _ intrinsic _ content of a relation such as @xmath109 at a given point @xmath113 is unchanged if we rotate each of @xmath2 and @xmath3 by the same amount .
that is , the equation @xmath114 has the same physical content as the equation @xmath115 .
for example scalar product relations , from which we can derive angles , are unaffected by this change .
these arguments apply to any physical relation between any type of multivector field .
the principles underlying gauge theory gravity can therefore be summarised as follows : the physical content of a field equation must be invariant under arbitrary local displacements of the fields .
( this is called position - gauge invariance . )
the physical content of a field equation must be invariant under arbitrary local rotations of the fields .
( this is called rotation - gauge invariance . ) in this theory predictions for all measurable quantities , including distances and angles , must be derived from gauge - invariant relations between the field quantities themselves , not from the properties of the background spacetime . on the other hand , quantities that depend on a choice of ` gauge ' are not predicted absolutely and can not be defined operationally .
it is necessary to indicate how this approach differs from the one adopted in gauge theories of the poincar group .
( this is a point on which we have been confused in the past @xcite . )
poincar transformations for a multivector field @xmath116 are defined by @xmath117 where @xmath118 @xmath16 is a constant rotor and @xmath119 is a constant vector .
transformations of this type mix displacements and rotations , and any attempt at a local gauging of this spacetime symmetry fails to decouple the two @xcite .
furthermore , the fact that the rotations described by poincar transformations include the displacement ( with @xmath120 ) means that the rotations discussed under point ( ii ) above are not contained in the poincar group . as a final introductory point , while the mapping of fields onto spacetime positions is arbitrary , the fields themselves must be well - defined .
the fields can not be singular except at a few special points .
furthermore , any remapping of the fields in the must be one - to - one , else we would cut out some region of physical significance . in later sections we will see that general relativity allows operations in which regions of spacetime are removed .
these are achieved through the use of singular coordinate transformations and are the origin of some potential differences between gtg and general relativity .
we now examine the consequences of the local symmetries we have just discussed . as in all gauge theories
we must study the effects on _ derivatives _ , since all non - derivative relations already satisfy the correct requirements .
we start by considering a scalar field @xmath121 and form its vector derivative @xmath122 .
suppose now that from @xmath121 we define the new field @xmath123 by @xmath124 and @xmath112 is an arbitrary ( differentiable ) map between spacetime position vectors .
the map @xmath112 should not be thought of as a map between manifolds , or as moving points around ; rather , the function @xmath112 is merely a rule for relating one position vector to another within a single vector space .
note that the new function @xmath123 is given by the old function @xmath125 evaluated at @xmath126 . we could have defined things the other way round , so that @xmath127 is given by @xmath121 , but the form adopted here turns out to be easier to implement in practice .
if we now act on the new scalar field @xmath128 with @xmath88 we form the quantity @xmath129 . to evaluate this we first look at the directional derivatives of @xmath128 , @xmath130 where @xmath131 and the subscript on @xmath132 records that the derivative is now with respect to the new vector position variable @xmath126 .
the function @xmath95 is a linear function of @xmath2 and an arbitrary function of @xmath0 . if we wish to make the position - dependence explicit we write this as @xmath133 . in general , any position - dependent linear function with its
position - dependence suppressed is to be taken as a function of @xmath0 . also as stated previously the argument of a linear function should be assumed to be constant unless explicitly stated otherwise .
from we see that @xmath134 and it follows that @xmath135 the bracketed term on the right - hand side , @xmath136 , is the old gradient vector @xmath137 evaluated at @xmath126 instead of @xmath0 .
this tells us how to modify the derivative operator @xmath88 : we must introduce a new linear function that assembles with @xmath88 in such a way that the @xmath106 field is removed when the full object is displaced .
the resulting object will then have the desired property of just changing its position dependence under arbitrary local displacements .
we therefore introduce the _ position - gauge field _
@xmath138 , which is a linear function of @xmath2 and an arbitrary function of position @xmath0 . as usual
this is abbreviated to @xmath139 when the position - dependence is taken as a function of @xmath0 . under the displacement @xmath140 , @xmath139 is defined to transform to the new field @xmath141 , where @xmath142 this ensures that @xmath143 transforms as @xmath144 this transformation law ensures that , if we define a vector @xmath145 by @xmath146 then @xmath145 transforms simply as @xmath147 under _ arbitrary _ displacements .
this is the type of behaviour we seek .
the vector @xmath145 can now be equated with other ( possibly non - differentiated ) fields and the resulting equations are unchanged in form under arbitrary repositioning of the fields in spacetime .
henceforth , we refer to any quantity that transforms under arbitrary displacements as @xmath148 as behaving _ covariantly _ under displacements .
the position gauge field @xmath149 enables us to form derivatives of covariant objects which are also covariant under displacements .
when we come to calculate with this theory , we will often fix a gauge by choosing a labelling of spacetime points with vectors . in this way we remain free to exploit all the advantages of representing points with vectors .
of course , all physical predictions of the theory will remain independent of the actual gauge choice .
the @xmath149-field is not a connection in the conventional yang
mills sense .
the coupling to derivatives is different , as is the transformation law .
this is unsurprising , since the group of arbitrary translations is infinite - dimensional ( if we were considering maps between manifolds then this would form the group of diffeomorphisms ) .
nevertheless the @xmath149-field embodies the idea of replacing directional derivatives with covariant derivatives , so clearly deserves to be called a gauge field .
a remaining question is to find the conditions under which the @xmath149-field can be transformed to the identity
. this would be the case if @xmath149 was derived entirely from a displacement , which would imply that @xmath150 and hence @xmath151 but , from the definition of @xmath95 , it follows that @xmath152 and hence that @xmath153 so , if the @xmath139 field can be transformed to the identity , it must satisfy @xmath154 an arbitrary @xmath149-field will not satisfy this equation , so in general there is no way to assign position vectors so that the effects of the @xmath149-field vanish . in the light of equations and it might seem more natural to introduce the gauge field as @xmath155 , instead of @xmath143 .
there is little to choose between these conventions , though our choice is partially justified by our later implementation of the variational principle .
we now examine how the derivative must be modified to allow rotational freedom from point to point , as described in point ( ii ) at the start of this section . here
we give an analysis based on the properties of classical fields . an analysis based on spinor fields
is given in the following section .
we have already seen that the gradient of a scalar field is modified to @xmath156 to achieve covariance under displacements . but
objects such as temperature gradients are certainly physical , and can be equated with other physical quantities .
consequently vectors such as @xmath156 must transform under rotations in the same manner as all other physical fields .
it follows that , under local spacetime rotations , the @xmath149-field must transform as @xmath157 now consider an equation such as maxwell s equation , which we saw in section [ o - gc ] takes the simple spacetime algebra form @xmath158 .
once the position - gauge field is introduced , this equation becomes @xmath159 where @xmath160 ( the reasons behind these definitions will be explained in section [ max ] . the use of a calligraphic letter for certain covariant fields is a convention we have found very useful . ) the definitions of @xmath161 and @xmath162 ensure that under local rotations they transform as @xmath163 any ( multi)vector that transforms in this manner under rotations and is covariant under displacements is referred to as a _ covariant _
( multi)vector .
equation is covariant under arbitrary displacements , and we now need to make it covariant under local rotations as well . to achieve this we replace @xmath143 by @xmath164 and focus attention on the term @xmath165 . under a position - dependent rotation
we find that @xmath166 since the rotor @xmath16 satisfies @xmath20 we find that @xmath167 hence @xmath168 is equal to minus its reverse and so must be a bivector .
( in a geometric algebra the bivectors form a representation of the lie algebra of the rotation group @xcite . )
we can therefore write @xmath169 to construct a covariant derivative we must therefore add a connection term to @xmath170 to construct the operator @xmath171 here @xmath172 is a bivector - valued linear function . to make the linear argument explicit we write @xmath173 so that @xmath174 is a linear function of @xmath2 with an arbitrary @xmath0-dependence .
the commutator product of a multivector with a bivector is grade - preserving so , even though it contains non - scalar terms , @xmath175 preserves the grade of the multivector on which it acts . under local rotations
the @xmath170 term in @xmath175 can not change , and we also expect that the @xmath175 operator be unchanged in form ( this is the essence of ` minimal coupling ' ) .
we should therefore have @xmath176 but the property that the covariant derivative must satisfy is @xmath177 and , substituting into this equation , we find that @xmath172 transforms as @xmath178 in coordinate - free form we can write this transformation law as @xmath179 of course , since @xmath180 is an arbitrary function of position , it can not in general be transformed away by the application of a rotor .
we finally reassemble the derivative with the @xmath181 term to form the equation @xmath182 the transformation properties of @xmath139 , @xmath161 , @xmath162 and @xmath180 ensure that this equation is now covariant under rotations as well as displacements . to complete the set of transformation laws , we note that under displacements @xmath180 must transform in the same way as @xmath183 , so that @xmath184 it follows that @xmath185 as required for covariance under local translations .
general considerations have led us to the introduction of two new gauge fields : vector - valued linear function @xmath138 and the bivector - valued linear function @xmath186 , both of which are arbitrary functions of position @xmath0 .
this gives a total of @xmath187 scalar degrees of freedom .
the @xmath139 and @xmath180 fields are incorporated into the vector derivative to form the operator @xmath188 , which acts covariantly on multivector fields .
thus we can begin to construct equations whose intrinsic content is free of the manner in which we represent spacetime positions with vectors .
we next see how these fields arise in the setting of the dirac theory .
this enables us to derive the properties of the @xmath175 operator from more primitive considerations of the properties of spinors and the means by which observables are constructed from them .
first , though , let us compare the fields that we have defined with the fields used conventionally in general relativity . to do this , we first construct the vectors
@xmath189 from these , the metric is defined by @xmath190 note that the inner product here is the standard spacetime algebra inner product and is not related to a curved space metric . similarly , the @xmath149-field can be used to construct a _ vierbein _
, as discussed in appendix c. a vierbein in general relativity relates a coordinate frame to an orthonormal frame .
but , while the @xmath149-field can be used to construct such a vierbein , it should be clear that it serves a different purpose in gtg it ensures covariance under arbitrary displacements .
this was the motivation for the introduction of a vierbein in kibble s work @xcite , although only infinitesimal transformations could be considered there .
furthermore , the @xmath149-field is essential to enable a clean separation between field rotations and displacements , which again is not achieved in other approaches .
we now rederive the gravitational gauge fields from symmetries of the dirac action .
the point here is that , once the @xmath149-field is introduced , spacetime rotations and phase rotations couple to the dirac field in essentially the same way . to see this , we start with the dirac equation and dirac action in a slightly unconventional form @xcite .
we saw in section [ outline ] that rotation of a multivector is performed by the double - sided application of a rotor .
the elements of a linear space that is closed under single - sided action of a representation of the rotor group are called _
spinors_. in conventional developments a matrix representation for the clifford algebra of spacetime is introduced , and the space of column vectors on which these matrices act defines the spin - space .
but there is no need to adopt such a construction .
for example , the even subalgebra of the spacetime algebra forms a linear space that is closed under single - sided application of the rotor group .
the even subalgebra is also an eight - dimensional linear space , the same number of real dimensions as a dirac spinor , and so it is not surprising that a one - to - one map between dirac spinors and the even subalgebra can be constructed .
such a map is given in appendix a. the essential details are that the result of multiplying the column spinor @xmath191 by the dirac matrix @xmath192 is represented in the spacetime algebra as @xmath193 , and that multiplication by the scalar unit imaginary is represented as @xmath194 .
it is easily seen that these two operations commute and that they map even multivectors to even multivectors . by replacing dirac matrices and column spinors by their spacetime algebra
equivalents the dirac equation can be written in the form @xmath195 which is now representation - free and coordinate - free . using the same replacements , the free - particle dirac action becomes @xmath196 and , with the techniques of appendix [ app - results ] , it is simple to verify that variation of this action with respect to @xmath197 yields equation with @xmath24=0 .
it is important to appreciate that the fixed @xmath51 and @xmath198 vectors in and do not pick out preferred directions in space .
these vectors can be rotated to new vectors @xmath199 and @xmath200 , and replacing the spinor by @xmath201 recovers the same equation .
this point will be returned to when we discuss forming observables from the spinor @xmath197 .
our aim now is to introduce gauge fields into the action to ensure invariance under arbitrary rotations and displacements .
the first step is to introduce the @xmath149-field . under a displacement @xmath197 transforms covariantly , @xmath202 where @xmath110 .
we must therefore replace the @xmath88 operator by @xmath143 so that @xmath203 is also covariant under translations . but this on its own does not achieve complete invariance of the action integral under displacements .
the action consists of the integral of a scalar over some region .
if the scalar is replaced by a displaced quantity , then we must also transform the measure and the boundary of the region if the resultant integral is to have the same value . transforming the boundary
is easily done , but the measure does require a little work .
suppose that we introduce a set of coordinates @xmath85 .
the measure @xmath204 is then written @xmath205 where @xmath206 by definition , @xmath204 is already independent of the choice of coordinates , but it must be modified to make it position - gauge invariant . to see how , we note that under the displacement @xmath207 , the @xmath208 frame transforms to @xmath209 it follows that to ensure invariance of the action integral we must replace each of the @xmath210 by @xmath211 .
thus the invariant scalar measure is @xmath212 these results lead us to the action @xmath213 which is unchanged in value if the dynamical fields are replaced by @xmath214 and the boundary is also transformed .
having arrived at the action in the form of we can now consider the effect of rotations applied at a point .
the representation of spinors by even elements is now particularly powerful because it enables both internal phase rotations and rotations in space to be handled in the same unified framework . taking the electromagnetic coupling first
, we see that the action is invariant under the global phase rotation @xmath215 ( recall that multiplication of @xmath191 by the unit imaginary is represented by right - sided multiplication of @xmath197 by @xmath216 . )
the transformation is a special case of the more general transformation @xmath217 where @xmath16 is a constant rotor .
similarly , invariance of the action under spacetime rotations is described by @xmath218 in both cases , @xmath197 just picks up a single rotor . from the previous section
we know that , when the rotor @xmath16 is position - dependent , the quantity @xmath219 is a bivector - valued linear function of @xmath210 . since @xmath220 with a similar result holding when the rotor acts from the right , we need the following covariant derivatives for local internal and external rotations : @xmath221 for the case of ( internal ) phase rotations , the rotations are constrained to take place entirely in the @xmath216 plane .
it follows that the internal connection takes the restricted form @xmath222 where @xmath24 is the conventional electromagnetic vector potential and @xmath223 is the coupling constant ( the charge ) .
the full covariant derivative therefore has the form @xmath224 and the full invariant action integral is now ( in coordinate - free form ) @xmath225 the action is invariant under the symmetry transformations listed in table [ g - dtab1 ] .
c|cccc & + .__the symmetries of the action integral . _ _ [ cols="^ " , ] & + + at first sight , the equations of table [ cos - tab2 ] do not resemble the usual friedmann equations .
the friedmann equations are recovered straightforwardly , however , by setting @xmath226 with this substitution we find that @xmath227 and that the @xmath228 and density equations become @xmath229 recovering the friedmann equations in their standard form @xcite .
the intrinsic treatment has therefore led us to work directly with the ` hubble velocity ' @xmath230 , rather than the ` distance ' scale @xmath231 .
there is a good reason for this .
once the weyl tensor is set to zero , the riemann tensor reduces to @xmath232 and we have now lost contact with an intrinsically - defined distance scale .
we can therefore rescale the radius variable @xmath25 with an arbitrary function of @xmath119 ( or @xmath25 ) without altering the riemann tensor .
the hubble velocity , on the other hand , is intrinsic and it is therefore not surprising that our treatment has led directly to equations for this . among the class of radial rescalings a particularly useful one is to rescale @xmath25 to @xmath233 .
this is achieved with the transformation @xmath234 so that , on applying equation , the transformed @xmath149-function is @xmath235 the function reproduces the standard line element used in cosmology .
we can therefore use the transformation to move between the ` newtonian ' gauge developed here and the gauge of .
this is useful for later sections , where the maxwell and dirac equations are solved in a cosmological background described by .
the differences between these gauges can be understood by considering geodesic motion .
a particle at rest with respect to the cosmological frame ( defined by the cosmic microwave background ) has @xmath236 . in the gauge of such
a particle is not moving in the flatspace background ( the distance variable @xmath25 is equated with the comoving coordinate ) . in the newtonian gauge ,
on the other hand , comoving particles are moving outwards radially at a velocity @xmath237 , though this expansion centre is not an intrinsic feature .
of course , attempting to distinguish these pictures is a pointless exercise , since all observables must be gauge - invariant .
all that is of physical relevance is that , if two particles are at rest with respect to the cosmological frame ( defined by the cosmic microwave background ) , then the light - travel time between these particles is an increasing function of time and light is redshifted as it travels between them .
the utility of the newtonian gauge in cosmology has been independently discovered by other authors @xcite .
here we illustrate its advantages for dust models ( @xmath238 ) . setting @xmath239 to zero implies that @xmath240
so @xmath241 and @xmath242 we are therefore left with a single first - order differential equation for @xmath243 .
explicit solutions of this equations are often not needed , as we can usually parameterise time by the density @xmath244 . if we now look at the trajectories defined by the fluid , these have @xmath245 which implies that @xmath246 it follows that @xmath247 and hence that @xmath248 the fluid streamlines form a family of spacetime curves spreading out from the origin at the initial singularity ( when @xmath249 ) .
the newtonian gauge therefore describes an expanding universe in a very simple , almost naive manner .
since all points in a homogeneous cosmology are equivalent , we can consider ourselves to be located at @xmath250 .
the newtonian gauge then pictures our observable universe as a ball of dust expanding outwards radially from us .
while the picture provided by the newtonian gauge has no physical reality of its own , it does have some heuristic merit and can provide a useful aid to one s physical intuition .
for example , consider the familiar relationship between angular size and redshift .
an initially surprising feature of this relationship is that , beyond a certain redshift , angular sizes stop decreasing and start increasing .
this result is easily understood in the newtonian gauge .
consider the photon paths shown in figure [ cos - fig1 ] .
( these paths are for a @xmath251 and @xmath252 universe , though the comments are applicable more generally . )
suppose that at some finite time @xmath253 we receive a photon from the distant past .
this photon must have followed part of a path which begins on the origin at @xmath120 . before a certain time in the past , therefore , photons received by us must have initially travelled outwards before turning round and reaching us .
the angular size of an observed object is then that appropriate to the actual @xmath25 coordinate of the object when it emitted the photons . since the value of @xmath25 decreases before a certain time , angular sizes appear larger for objects that emitted photons before this time .
a radial null geodesic has a trajectory @xmath254 for these trajectories @xmath255 and the condition that @xmath256 reduces to @xmath257 since @xmath230 is positive in a cooling universe , the distance at which photons ` turn round ' is defined by @xmath258 ( in @xcite we mistakenly referred to this distance as a particle horizon . ) for @xmath259 it is simple to show that this distance corresponds to a redshift of @xmath260 .
this result relates physically measurable quantities , so is gauge invariant .
the newtonian gauge presents a particularly simple picture for a closed universe ( @xmath261 ) .
for @xmath261 the requirement that @xmath262 is positive means that @xmath263 this places a limit on the speed with which the dust can expand , so the solution describes a finite ball of dust expanding into a vacuum .
this ball expands out to some fixed radius before turning round and contracting back to the origin .
the turning point is achieved where @xmath264 so , for @xmath252 , this occurs when @xmath265 the maximum radius is therefore @xmath266 the time taken to reach the future singularity is given by , since this cosmological model is a special case of spherical collapse in which the density is uniform .
this picture of a model for a closed universe is both simple and appealing .
it allows us to apply newtonian reasoning while ensuring consistency with the full relativistic theory .
the finite ball model for a @xmath261 cosmology is clearly useful when considering experiments with particles carried out near the origin , but globally one must consider the boundary properties of the ball .
a crucial question is whether the particle horizon ( the largest region of the universe with which an observer at the origin is in causal contact ) extends past the edge of the ball or not .
it can be shown that this horizon always lies inside the radius at which @xmath267 becomes imaginary , except at the turnaround point ( the point at which the ball reaches its maximum radius ) , where the two radii coincide .
a suitable choice of cutoff radius is therefore available in either the expanding or contracting phase separately , but what happens at the turnaround point is potentially ambiguous . when discussing field theory , however , the finite ball model is inadequate .
one must instead use a global gauge so , in section [ max - cos ] , we introduce the ` stereographic projection gauge ' .
this provides a global solution which can be shown to be spatially closed .
this solution is used in the study of electromagnetism ( section [ max - cos ] ) and the dirac equation ( section [ de - cos ] ) in a cosmological background .
it is possible to treat the stereographic projection gauge solution in a form of the newtonian gauge , though this possibility is not explored here .
in section [ field ] we derived field equations for the gravitational and dirac fields .
we now turn to the derivation of the maxwell equations in a gravitational background . a number of applications of these equations
are then discussed , including a simple derivation of the characteristic surfaces for both the maxwell and dirac equations .
the basic dynamical variable is the electromagnetic vector potential @xmath24 , for which the coupling to spinor fields was derived in section [ g - dirac ] . under phase rotations of the spinor field , @xmath24 transforms as @xmath268 it follows that , under a displacement , @xmath24 must transform in the same way as @xmath137 , that is , @xmath269 the covariant form of the vector potential is therefore @xmath270 which is the term that appeared in the dirac equation . from @xmath24 , the faraday bivector @xmath59
is defined by @xmath271 this definition implies that , under displacements , @xmath272 is transformed to @xmath273 , where @xmath274 it follows that the covariant analogue of @xmath59 is defined by @xmath275 which is covariant under position and rotation - gauge transformations , and is invariant under phase changes . the same covariant quantity @xmath161 is obtained if one follows the route used for the construction of @xmath276 at .
in particular , the contracted commutator of two covariant derivatives gives ( in the absence of torsion ) @xmath277
\psi & = { { \textstyle \frac{1}{2}}}{{\bar{\mathsf{h}}}}({\partial_b}){{\bar{\mathsf{h}}}}({\partial_a } ) { \mathsf{r}}(a { \ ! \wedge \!}b ) \psi { \nonumber}\\ & = { { \textstyle \frac{1}{2}}}{{\mathcal{r}}}\psi.\end{aligned}\ ] ] the analogous construction for the ` internal ' covariant derivative @xmath278 gives @xmath279 \psi & = e { { \bar{\mathsf{h}}}}({\partial_b } ) { \ !
\wedge \!}{{\bar{\mathsf{h}}}}({\partial_a } ) ( a{\ ! \wedge \!}b ) { \ ! \cdot \!}f \psi { i \hspace{-1pt } { \mbox{\boldmath $ { \mbox{\boldmath $ \sigma$}}$}}_3}{\nonumber}\\ & = 2 e { { \mathcal{f}}}\psi { i \hspace{-1pt } { \mbox{\boldmath $ { \mbox{\boldmath $ \sigma$}}$}}_3 } , \label{em4}\end{aligned}\ ] ] which clearly identifies @xmath161 as a covariant quantity .
unlike @xmath280 , however , @xmath161 is a bivector and equation exhibits a curious interaction between this bivector on the left of @xmath197 , and the fixed bivector @xmath216 on the right .
having defined the covariant bivector @xmath161 , it is clear that the appropriate generalisation of the electromagnetic action to include gravitational interactions is @xmath281 where here @xmath162 is the covariant charge current . unlike the dirac action , the rotation - gauge field @xmath180
does not appear in this action .
it follows that the electromagnetic field does not act as a source of spin .
the action is varied with respect to @xmath24 , with @xmath139 and @xmath162 treated as external fields .
the result of this is the equation @xmath282 where @xmath283 equation combines with the identity @xmath284 to form the full set of maxwell equations in a gravitational background .
we again see that the classical field equations can be expressed in a form from which all reference to the rotation gauge has been removed .
some insight into the equations and is obtained by performing a space - time split ( see section [ o - sta ] ) and writing @xmath285 in terms of these variables maxwell s equations in a gravitational background take the familiar form @xmath286 where @xmath287 is the 3d vector derivative , and the bold cross @xmath288 is the traditional vector cross product : @xmath289 equation shows that the @xmath149-field defines the dielectric properties of the space through which the electromagnetic fields propagate , with @xmath290 determining the generalized permittivity / permeability tensor .
many phenomena , including the bending of light , can be understood easily in terms of the properties of the dielectric defined by the @xmath149-field .
so far we have failed to achieve a manifestly covariant form of the maxwell equations .
we have , furthermore , failed to unite the separate equations into a single equation . in the absence of gravitational effects the equations @xmath291 and @xmath292 combine into the single equation @xmath293 the significance of this equation is that the @xmath88 operator is invertible , whereas the separate @xmath93 and @xmath94 operators are not @xcite .
clearly , we expect that such a unification should remain possible after the gravitational gauge fields have been introduced . to find a covariant equation , we first extend the ` wedge ' equation to include higher - grade terms . to make the derivation general we include the spin term , in which case we find that @xmath294 and , more generally , we can write @xmath295 we can therefore replace equation by @xmath296 we next use the rearrangement @xmath297 to write equation as @xmath298 equations and now combine into the single equation @xmath299 which achieves our objective .
equation is manifestly covariant and the appearance of the @xmath300 term is precisely what one might expect on ` minimal - coupling ' grounds .
the appearance of the spin term is a surprise , however .
gauge arguments alone would not have discovered this term and it is only through the construction of a gauge - invariant action integral that the term is found .
equation should be particularly useful when considering electromagnetic effects in regions of high spin density , such as neutron stars . to complete the description of electromagnetism in a gravitational background
we need a formula for the free - field stress - energy tensor . applying the definition we construct @xmath301 hence , @xmath302 which is the natural covariant extension of the gravitation - free form @xmath303 .
the tensor is symmetric , as one expects for fields with vanishing spin density . in their spacetime algebra form
the maxwell and dirac equations are both first - order differential equations involving the vector derivative @xmath88 . for electromagnetism
, this first - order form of the equations offers many advantages over the equivalent second - order theory @xcite .
we have now seen that gravitational interactions modify both these equations in such a way that the vector derivative @xmath88 is replaced by the position - gauge covariant derivative @xmath143 . as an illustration of the utility of first - order equations , both without and with gravitational effects
, we now give a simple derivation of the properties of characteristic surfaces .
consider , initially , a generic equation of the type @xmath304 where @xmath197 is some arbitrary field , and @xmath305 is some known function .
suppose that an initial set of data is given over some three - dimensional surface in spacetime , and we wish to propagate this information off the surface into the adjoining region .
we pick three vectors , @xmath2 , @xmath3 and @xmath306 , which are tangent to the surface . from our initial data
we can construct @xmath307 , @xmath308 and @xmath309 . now define @xmath310 and use @xmath311 to decompose into @xmath312 the right - hand side of contains the term @xmath313 which is therefore known .
it follows that we know all the terms on the right - hand side of equation and can therefore construct @xmath314 .
this gives us the derivative required to propagate off the surface .
the only situation for which propagation is impossible is when @xmath10 remains in the surface .
this occurs when @xmath315 it follows that the characteristic surfaces for any first - order equation of the type defined by are null surfaces .
these considerations automatically include the maxwell and dirac equations .
it is notable how this result follows from purely algebraic considerations .
the generalisation to a gravitational background is straightforward .
equation is generalised to @xmath316 and we assume that a gauge choice has been made so that all the fields ( apart from @xmath197 ) are known functions of @xmath0 .
again , we assume that the initial data consist of values for @xmath197 over some three - dimensional surface , so we can still determine @xmath307 _ etc . _ since @xmath317 it follows that the vector of interest is now @xmath318 this time we multiply equation by @xmath319 and find that @xmath320 can be constructed entirely from known quantities
. we can therefore propagate in the @xmath321 direction , so we now require that this vector does not lie in the initial surface .
the analogue of is therefore @xmath322 and the characteristic surfaces are now those for which @xmath319 is null .
this is the obvious covariant extension of @xmath10 being a null vector .
the problem of interest here is to find the fields generated by a point source held at rest outside the horizon of a radially - symmetric black hole .
the @xmath149-function in this case can be taken as that of equation .
the solution to this problem can be found by adapting the work of copson @xcite and linet @xcite to the present gauge choices .
assuming , for simplicity , that the charge is placed on the @xmath323-axis a distance @xmath2 from the origin ( @xmath324 ) , the vector potential can be written in terms of a single scalar potential @xmath325 as @xmath326 it follows that @xmath327 and @xmath328 the maxwell equations now reduce to the single partial differential equation @xmath329 where @xmath330 is a @xmath331-function at @xmath332 .
this was the problem originally tackled by copson @xcite who obtained a solution that was valid locally in the vicinity of the charge , but contained an additional pole at the origin .
linet @xcite modified copson s solution by removing the singularity at the origin to produce a potential @xmath325 whose only pole is on the @xmath323-axis at @xmath332 .
linet s solution is @xmath333 where @xmath334 an important feature of this solution is that once is inserted back into the resultant @xmath161 is both finite and continuous at the horizon .
furthermore , since @xmath335 is well - defined at the horizon , both @xmath59 and @xmath336 must also be finite and continuous there .
working in the newtonian gauge has enabled us to construct a global solution , and we can therefore study its global properties .
one simple way to illustrate the global properties of the solution is to plot the streamlines of @xmath337 which , from equation , is divergenceless away from the source .
the streamlines should therefore spread out from the charge and cover all space .
since the distance scale @xmath25 is fixed to the gravitationally - defined distance , the streamlines of @xmath337 convey genuine intrinsic information .
hence the plots are completely unaffected by our choice for the @xmath267 or @xmath338 functions , or indeed our choice of @xmath119-coordinate .
figure [ max - fig1 ] shows streamline plots for charges held at different distances above the horizon .
similar plots were first obtained by hanni & ruffini @xcite although , as they worked with the schwarzschild metric , they were unable to extend their plots through the horizon .
the plots reveal an effective contraction in the radial direction .
it is not hard to show that the contraction is precisely that of a particle moving with the free - fall velocity @xmath339 relative to a fixed observer .
the description presented here of the fields due to a point charge is very different from that advocated by the ` membrane paradigm ' @xcite .
the membrane paradigm was an attempt to develop the theory of black holes in a way that , as far as possible , employed only familiar physical concepts . in this way
, gravitational effects could be incorporated correctly without requiring an understanding of the full theory .
the hope was that astrophysicists would adopt this paradigm when modelling regions where black - hole physics could be significant , such as at the heart of a quasar .
the paradigm works by drawing a veil over the horizon ( the membrane ) and concentrating on the physics outside the horizon as seen by observers remaining at a fixed distance ( fiducial observers ) . our view
, however , is that it is the newtonian gauge which provides the clearest understanding of the physics of black holes while requiring minimal modification to newtonian and special - relativistic ideas .
furthermore , writing the maxwell equations in the form removes any difficulties in applying conventional reasoning to the study electromagnetism in a gravitational background .
there are other ways that our approach offers advantages over the membrane paradigm .
when applying the membrane paradigm one has to work with quantities which are singular at the horizon , and this is hardly a recipe for applying traditional intuition ! as we have seen , once formulated in the newtonian gauge ( or any other gauge admissible in gtg ) all physical quantities are finite .
the membrane paradigm also warns physicists against producing plots such as figure [ max - fig1 ] , because such plots depend on the choice of radial coordinate .
but our intrinsic approach makes it clear that such plots _ are _ meaningful , because @xmath25 is determined uniquely by the riemann tensor .
presenting the plots in the form of figure [ max - fig1 ] enables direct physical information to be read off . in short , the simple physical picture provided by our intrinsic method and newtonian - gauge solution disposes of any need to adopt the artificial ideas advocated by the membrane paradigm .
an interesting feature of the above solution is the existence of a repulsive ` polarisation ' force @xcite , one effect of which is that a smaller force is needed to keep a charged particle at rest outside a black hole than an uncharged one . in their derivation of this force ,
smith & will @xcite employed a complicated energy argument that involved renormalising various divergent integrals . here
we show that the same force can be derived from a simple argument based on the polarisation effects of the dielectric described by a black hole .
first , however , we must be clear how force is defined . in the presence of an electromagnetic field the equation of motion for a point particle
is modified to @xmath340 we therefore expect that any additional force should also be described by a covariant bivector which couples to the velocity the same way that @xmath161 does .
so , if we denote the externally applied force as @xmath341 , the equation of motion for a neutral test particle becomes @xmath342 now , suppose that @xmath341 is chosen so that the particle remains at a fixed distance @xmath2 outside the horizon of a black hole .
the equation for the trajectory is @xmath343 where the constants @xmath344 and @xmath345 specify the angular position of the particle .
the covariant velocity is @xmath346 and the condition that @xmath347 forces @xmath348 so that @xmath349 , and hence @xmath350 , are constant .
equation now reduces to @xmath351 where @xmath352 is the unit outward spatial vector from the source to the particle .
the magnitude of the force is therefore @xmath353 and it is not hard to check that this result is gauge - invariant .
equation reduces to the newtonian formula at large distances and becomes singular as the horizon is approached , where an infinite force is required to remain at rest .
we now want to see how this expression for the force is modified if the particle is charged and feels a force due to its own polarisation field . from the extra term in the force
is simply @xmath161 , and only the radial term contributes . from equation
this is just @xmath354 . since the charge lies on the @xmath323 axis , we need only look at @xmath355 along this axis , for which @xmath356 the singular terms must be due to the particle s own coulomb field , and
so do not generate a polarisation force .
the only term that generates a force is therefore the final one , which is precisely the term that linet added to copson s original formula !
this term produces an outward - directed force on the charge , of magnitude @xmath357 .
the applied force is therefore now @xmath358 so a smaller force is needed to keep the particle at rest outside the horizon .
this result agrees with that found in @xcite , though our derivation avoids the need for infinite mass renormalisation and is considerably simpler .
this result is a good example of the importance of finding global solutions .
the polarisation force is felt outside the horizon , yet the correction term that led to it was motivated by the properties of the field at the origin .
we saw in section [ ss - cos ] that one form of the @xmath149-function for a homogeneous cosmology is defined by @xmath359 where @xmath360 .
when @xmath361 is positive , however , the function is undefined for @xmath362 and so fails to define a global solution .
a globally - valid solution is obtained with the displacement @xmath363 which results in the particularly simple solution @xmath364 this solution is well - defined for all @xmath0 and generates an ` isotropic ' line element .
( the solution can also be viewed as resulting from a stereographic projection of a 3-sphere . )
we want to find the fields due to a point charge in the background defined by .
since the @xmath149-function is diagonal we start with the obvious anstz @xmath365 so that @xmath366 and @xmath367 it follows that the equation we need to solve is simply @xmath368 where the charge @xmath369 is located at @xmath370 .
this equation can be solved using the general technique described by hadamard ( see also copson @xcite for a discussion of a similar problem . ) the solution to turns out to be @xmath371 where @xmath372 as a simple check , @xmath355 reduces to the usual coulomb potential when @xmath251 .
( 300,270 ) ( 0,270 ) ( 147,-10)@xmath0 ( 0,134)@xmath323 fieldline plots of the @xmath337 field defined from are shown in figure [ max - fig3 ] .
the fieldlines follow null geodesics and clearly reveal the existence of an image charge .
the reason for this is can be seen in the denominator of @xmath355 . not only is @xmath355 singular at @xmath373 , it is also singular at @xmath374 so , if the charge lies on the @xmath323 axis , the position of its image is found by solving @xmath375 from which we find that the image charge is located at @xmath376 ( this is the stereographic projection of the opposite point to the charge on a 3-sphere . )
however , if we try to remove this image charge by adding a point source at its position , we find that the fields vanish everywhere , since the new source has its own image which cancels the original charge .
the image charge is therefore an unavoidable feature of @xmath377 cosmologies .
we comment on this further in section [ cos ] .
in this paper we began by introducing the gravitational gauge fields and minimally coupling these to the dirac action . for our final major application
we return to the dirac equation in a gravitational background . in the absence of an @xmath378 field ,
the minimally - coupled equation is @xmath379 here we consider two applications : the case of a black - hole background ; and cosmological models in which the universe is not at critical density .
we have already demonstrated that the newtonian gauge solution dramatically simplifies the study of black - hole physics , so this is the natural gauge in which to study the dirac equation .
we therefore start with an analysis in this gauge , and then consider the gauge - invariance of the predictions made . with the gravitational fields as described in section [ ss - bh ] , the dirac equation becomes @xmath380 if we pre - multiply by @xmath51 and introduce the symbol @xmath381 to represent right - sided multiplication by @xmath216 , so that @xmath382 , then equation becomes @xmath383 where @xmath384 .
one feature that emerges immediately is that the newtonian gauge has recovered a hamiltonian form of the dirac equation ( see @xcite for a discussion of operators , hamiltonians and hermiticity in the spacetime algebra approach to dirac theory ) .
since the newtonian gauge involves the notion of a global time , it might have been expected that this gauge would lend itself naturally to a hamiltonian formulation .
an open question of some interest is whether it is always possible to make such a gauge choice ( see @xcite for an extension to the kerr solution ) .
the hamiltonian contains a subtlety : it is hermitian only away from the origin . to see why , consider the interaction term @xmath385 for this we find that @xmath386_0^\infty , \label{dbh3}\end{aligned}\ ] ] where @xmath387 denotes the projection onto the ` complex ' @xmath388 and @xmath216 terms and @xmath389 ( see appendix [ app - dirac ] and @xcite ) . for
all normalised states the final term in tends to zero as @xmath390 .
but it can be shown from that wavefunctions tend to the origin as @xmath391 , so the lower limit is finite and @xmath392 is therefore not a hermitian operator .
this immediately rules out the existence of normalisable stationary states with constant real energy @xcite .
equation can be used to propagate a spinor defined over some initial spatial surface , and numerical simulations based on this equation give a good picture of the scattering induced by a black hole . here
, however , we wish to focus on an analytical approach .
a result that follows immediately from the hamiltonian form of the dirac equation is that is manifestly separable , so that we can write @xmath393 as usual , the solution of the @xmath119-equation is @xmath394 where @xmath395 is the separation constant .
the non - hermiticity of @xmath392 means that @xmath395 can not be purely real if @xmath197 is normalisable .
the imaginary part of @xmath395 is determined by equation and , for suitably normalised states , we find that @xmath396 this equation shows that the imaginary part of @xmath395 is necessarily negative , so the wavefunction decays with time .
this is consistent with the fact that the streamlines generated by the conserved current @xmath397 are timelike curves and , once inside the horizon , must ultimately terminate on the origin . with the @xmath119-dependence separated out ,
equation reduces to @xmath398 to solve this equation we next separate out the angular dependence .
this is achieved using the spherical monogenics , which are the spacetime algebra equivalent of the @xmath399 pauli spinors .
these are described in detail in @xcite and here we quote the necessary formulae .
the unnormalised monogenic @xmath400 is defined by @xmath401 where @xmath402 , @xmath403 , and @xmath404 are the associated legendre polynomials .
the two properties of the @xmath400 relevant here are @xmath405 now , the operator @xmath406 commutes with the hamiltonian defined by . constructing eigenstates of this operator with eigenvalue @xmath407 , leads to solutions of the form @xmath408 where @xmath407 is a non - zero integer and @xmath409 and @xmath410 are complex functions of @xmath25 ( _ i.e. _ sums of a scalar and an @xmath216 term ) . substituting into equation , and using the properties of the spherical monogenics , we arrive at the coupled radial equations @xmath411 where @xmath412 @xmath413 and @xmath414 are the reduced functions defined by @xmath415 and the primes denote differentiation with respect to @xmath25 .
( we continue to employ the abbreviation @xmath381 for @xmath216 . )
to analyse we first rewrite it in the equivalent form @xmath416 this makes it clear that the equations have regular singular points at the origin and horizon ( @xmath417 ) , as well as an irregular singular point at @xmath418 . to our knowledge , the special function theory required to deal with such equations has not been developed . without it
we either attempt a numerical solution , or look for power series with a limited radius of convergence . here we consider the latter approach , and look for power - series solutions around the horizon . to this end
we introduce the series @xmath419 where @xmath420 .
the index @xmath26 controls the radial dependence of @xmath197 at the horizon , so represents a physical quantity . to find the values
that @xmath26 can take , we substitute equation into and set @xmath421 .
this results in the equation @xmath422 rewriting this in terms of a single matrix and setting its determinant to zero yields the two indicial roots @xmath423 the @xmath373 solution is entirely sensible the power series is analytic , and nothing peculiar happens at the horizon .
the existence of this root agrees with our earlier observation that one can evolve the time - dependent equations without encountering any difficulties at the horizon .
the second root is more problematic , as it leads to solutions that are ill - defined at the horizon . before assessing the physical content of these roots , however , we must first check that they are gauge invariant .
if , instead of working in the newtonian gauge , we keep the gauge unspecified then , after separating out the angular dependence , the equations reduce to @xmath424 we can still assume that the @xmath119-dependence is of the form @xmath425 , so that equation becomes @xmath426 where @xmath427 now , since @xmath428 holds in all gauges , we obtain @xmath429 and substituting in the power series and setting @xmath421 produces the indicial equation @xmath430_{r=2 m } = 0,\ ] ] where @xmath52 is the identity matrix . employing the result that @xmath431 we find that the solutions to the indicial equation are now @xmath432 but in section [ ss - bh ] we established that @xmath433 at the horizon for all solutions with a forward time direction .
this demonstrates that the indices are indeed intrinsic , with the sign of the imaginary term for the singular root picking up information about the time direction implicit in the presence of the horizon .
the fact that @xmath373 is always a solution of the indicial equation means that solutions always exist that are analytic at the horizon .
determining the split between ingoing and outgoing states of these solutions enables one to calculate reflection coefficients and scattering amplitudes @xcite .
the question we wish to consider is whether the second , singular , root can be physically significant . to address this we look at the current .
the covariant current @xmath162 is given by @xmath397 , and satisfies @xmath434 .
the corresponding non - covariant quantity is therefore @xmath435 which satisfies the flatspace conservation equation @xmath436 .
it is the streamlines of @xmath437 that are plotted as functions of @xmath0 and determine the flow of density .
the crucial terms in @xmath437 are the time component and radial component , which ( ignoring the overall exponential decay term ) are given by @xmath438 the @xmath439 and @xmath440 are finite for all admissible solutions so , for the @xmath373 solution , the components of @xmath437 are well - defined at the horizon .
furthermore , it is easily shown that for @xmath373 the radial flux at the horizon always points inwards .
the @xmath373 root therefore describes the case where the flux crosses the horizon and continues onto the singularity . for the singular root we must first decide on a branch for the solution so that @xmath197 is fully specified on both sides of the horizon .
we can then assess whether the discontinuity in @xmath197 , and the discontinuity in the current generated by it , are physically acceptable .
to do this , we first write @xmath441 as @xmath442 we can now write @xmath443 and for the choice of argument we take @xmath444 ( the choice of sign will be discussed further below . ) if we now take the limit @xmath445 from above and below we find that the @xmath51 component of @xmath437 is given by @xmath446 where @xmath447 is a positive - definite , finite term and we have split @xmath395 into real and imaginary parts as @xmath448 equation is valid in all gauges for which @xmath449 at the horizon .
while the density @xmath450 is singular at the horizon , the presence of the positive term @xmath451 ensures that any integral over the horizon is finite and the solution is therefore normalisable .
this link between the properties of @xmath197 at the horizon and at the origin ( where @xmath452 is determined ) provides another example of the importance of finding global solutions to the field equations .
the radial current now turns out to be @xmath453 and is therefore zero at the horizon , and inward - pointing everywhere inside the horizon .
it appears that the existence of the imaginary contribution to @xmath395 does ensure that that the singular solutions have sensible physical properties , and the singularity in @xmath197 at the horizon is no worse than that encountered in the ground state of the hydrogen atom @xcite .
what is less clear , however , is the extent to which the properties of @xmath197 at the horizon are compatible with the original equation . in particular , since @xmath197 is both singular and non - differentiable at the horizon , it does not appear that the singular root can be viewed as defining a solution of over all space .
a number of authors have attempted to give derivations of the hawking temperature and particle flux due to a black hole from an analysis of first - quantised theory , _
i.e. _ from the properties of wave equations alone @xcite .
this work has generated some controversy @xcite , so it is interesting to assess how the ideas stand up in gtg .
these derivations focus on the singular solutions to the wave equation ( either klein gordon or dirac ) , and study the properties of these solutions under the assumption that the energy is real .
if one ignores the problems that @xmath454 introduces for the normalisability of @xmath197 and presses ahead , then from there is now a non - zero current at the horizon and , furthermore , there is a net creation of flux there .
the ratio of the outward flux to the total flux is simply @xmath455 which defines a fermi dirac distribution with temperature @xmath456 remarkably , this is the temperature found by hawking @xcite .
the fact that both the correct fermi dirac statistics and hawking temperature can be derived in this manner is astonishing , since both are thought to be the result of quantum field theory .
but what can we really make of this derivation ?
the first problem is that setting @xmath452 to zero means that the density is no longer normalisable at the horizon any integral of the density over the horizon region diverges logarithmically , which is clearly unphysical .
a further problem relates to the choice of branch .
had the opposite branch been chosen we would not have obtained and , as pointed out in @xcite , there is no _ a priori _ justification for the choice made in . for the above reasons , the ` derivation ' of can not be viewed as being sound .
the remarkable thing is that the same techniques can be used to ` derive ' the correct temperatures for the reissner
nordstrm and rindler cases , as well as the schwinger production rate in a constant electric field . this will be discussed elsewhere .
these further analyses contain another surprise : the temperature at the interior horizon of a reissner
nordstrm black hole is necessarily negative !
however , while these analyses are both interesting and suggestive , it is only through a study of the full quantum field theory in a black hole background that one can be sure about particle production rates .
a final point in this section is that all our analyses have been based on working with the correct time - asymmetric solutions admitted in gtg . on attempting to force through the analysis in the ` schwarzschild ' gauge ( @xmath457 )
, one discovers that the indices are now given by @xmath458 in this case no analytic solution is possible , and even the presence of an exponential damping factor does not produce a normalisable current at the horizon .
this only serves to reinforce the importance of working with global solutions , since there is no doubt that the presence of non - singular , normalisable solutions is an intrinsic feature of horizons . as a second application we consider the dirac equation in a cosmological background
we have a choice of form of @xmath149-function to use , of which the simplest is that defined by , @xmath459 the dirac equation in this background takes the form @xmath460 where the various functions are as defined in section [ ss - cos ] .
our question is this : can we find solutions to such that the observables are homogeneous ? there is clearly no difficulty if @xmath251 since , with @xmath238 , equation is solved by @xmath461 and the observables are fixed vectors which scale as @xmath244 in magnitude @xcite . but what happens when @xmath1 ?
it turns out that the solution must be modified to @xcite @xmath462 for the case of @xmath261 both @xmath139 and @xmath197 are only defined for @xmath463 .
this problem is overcome by using the displacement to transform to the global solution of equation . in this case
@xmath197 is given by @xmath464 which diverges as @xmath390 .
in both the @xmath261 and @xmath465 cases , @xmath197 contains additional @xmath25-dependence and so is not homogeneous .
furthermore , the observables obtained from @xmath197 are also inhomogeneous . in principle
one could therefore determine the origin of this space from measurements of the current density .
this clearly violates the principle of homogeneity , though it is not necessarily inconsistent with experiment .
the implications for cosmology of this fact are discussed in the following section .
( some consequences for self - consistent solutions of the einstein dirac equations are discussed in @xcite and , in the context of general relativity , in @xcite . )
the fact that quantum fields see this ` preferred ' direction in @xmath466 models , whereas classical phenomena do not , reflects the gauge structure of the theory .
dirac spinors are the only fields whose action couples them directly to the @xmath467-function .
all other matter fields couple to the gravitational field through the @xmath149-field only .
dirac spinors therefore probe the structure of the gravitational fields directly at level of the @xmath468-field , which is inhomogeneous for @xmath1 models . on the other hand
, classical fields only interact via the covariant quantities obtained from the gravitational fields , which are homogeneous for all values of @xmath361 .
this conclusion is reinforced by the fact that the klein
gordon equation , for which the action does not contain the @xmath467-field , _ does _ have homogenous solutions in a @xmath1 universe .
in section [ ss - cos ] we discussed some aspects of cosmology as examples of the general treatment of time - varying spherically - symmetric systems .
there we drew attention to the utility of the newtonian gauge as a tool for tackling problems in cosmology .
in addition , in sections [ max - cos ] and [ de - cos ] we studied the maxwell and dirac equations in various cosmological backgrounds . in this section
we draw together some of our conclusions from these sections . specifically , we discuss redshifts , difficulties with @xmath469 models , and the definitions of mass and energy for cosmological models . as a final demonstration of the use of the newtonian gauge , consider a photon following a null path in the @xmath470 plane . in this case
the photon s momentum can be written as @xmath471 and @xmath472 is the frequency measured by observers comoving with the fluid .
we restrict to the pressureless case , so @xmath473 and @xmath474 , but will allow @xmath243 to be @xmath25-dependent .
a simple application of equation produces @xmath475 where @xmath476 .
but , since @xmath474 , we find that @xmath477 , so @xmath478 which holds in any spherically - symmetric pressureless fluid .
for the case of a cosmological background we have @xmath479 , so the angular terms drop out of equation , and we are left with the simple equation @xmath480 which integrates to give the familiar redshift versus density relation @xmath481 other standard cosmological relations , such as the luminosity distance and angular diameter versus redshift formulae , can be easily derived in this gauge ( see also section [ ss - cos ] ) . in @xcite equation
was derived in the gauge of equation , in which all particles comoving with the cosmological fluid are at rest in the background spacetime . in this gauge
the redshift can be attributed to a loss of energy to the gravitational field , although this is a gauge - dependent viewpoint the only physical statement that one can make is embodied in equation .
the explanation of cosmological redshifts in our theory has nothing to do with ` tired light ' , or spacetime playing a dynamic role by expanding , or even anything to do with doppler shifts .
the redshift is simply a consequence of the assumption of homogeneity .
ultimately , all physical predictions are independent of the gauge in which they are made , although certain gauges may have useful computational or heuristic value . at the level of classical ( _ i.e. _ non - quantum ) physics
, there is no doubt that @xmath1 cosmologies are homogeneous .
this is true in both general relativity and gtg .
no prediction derived for classical systems of point particles or for electromagnetic fields can reveal a preferred spatial direction in these models .
however , we saw in section [ de - cos ] that it is impossible to find homogeneous solutions of the dirac equation in a @xmath1 universe .
the consequences of this are , in principle , observable , since the local density gradient will reveal a preferred radial direction .
it has already been pointed out that it is impossible to find a _ self - consistent _ solution of the combined system of einstein
dirac equations for any case other than a spatially flat cosmology @xcite .
we believe that this is the first time that it has been pointed out that even a _ non_-self - consistent dirac field would be observably inhomogeneous .
this is a more damaging result for @xmath1 cosmologies , since it reveals inhomogeneity without assuming that spin - torsion effects have anything to do with the dynamics of the universe .
while the properties of dirac fields pose theoretical difficulties for @xmath1 models , there is no contradiction with present observations .
furthermore , one could question the validity of inferences drawn in extrapolating the dirac equation to cosmological scales
. there is , however , a purely classical effect that does lead one to question the validity of @xmath261 models .
as we have seen , when looking at the properties of fields in a @xmath261 background it is necessary to work in a globally - defined gauge , such as that of equation . in this case
the maxwell equations show that each point charge must have an image charge present in a remote region of the universe .
this is a consequence of a closed universe that we have not seen discussed , although it has doubtless been pointed out before .
the necessity for this image charge raises many problems in attempting to take such a universe seriously .
setting aside the problems with @xmath1 models , a further issue on which our theory sheds some light is discussions of the total matter and energy content of the universe . in section [ colldst ]
we discussed the distinction between the rest - mass energy and the total gravitating energy inside a sphere of radius @xmath25 .
since cosmological models are a special case of the general theory outlined in section [ ss ] , this same distinction should be significant in cosmology .
in section [ colldst ] we identified the total gravitating energy of a sphere of radius @xmath25 with the function @xmath482 .
for all cosmological models , this is given by @xmath483 in strictly homogeneous models the weyl tensor vanishes , and we lose an intrinsically - defined distance scale .
but , if we consider cosmological models as the limiting case of spherically - symmetric systems , then there seems little doubt that is still the correct expression for the gravitating energy within a sphere of radius @xmath25 surrounding the origin .
moreover , attempts to discover the gravitating content of a region rely on perturbations away from ideal uniformity . in these cases
an intrinsic distance scale is well defined , since a weyl tensor is again present .
( determining the gravitating content of a region is important in , for example , determinations of @xmath484 the ratio of the actual density of the universe to the critical density . ) on the other hand , the total rest mass energy within a sphere of radius @xmath25 centred on the origin must still be given by @xmath485 this remains a covariant scalar quantity , and is just @xmath244 multiplied by the covariant volume integral ( this is the volume one would measure locally using light paths or fixed rods ) .
we have now defined two covariant scalar quantities , @xmath482 and @xmath486 , both of which are conserved along fluid streamlines in the absence of pressure .
if the identifications made in the spherically - symmetric case remain valid in the homogeneous case , then the difference between these should give the additional contribution to the total energy beyond the rest - mass energy .
for spatially - flat universes we have @xmath487 so there is no difference .
( in terms of the newtonian - gauge description of section [ colldst ] , the gravitational potential energy cancels the kinetic energy . ) but , for @xmath1 models , there is a difference because @xmath486 is now given by @xmath488 an interesting place to study the difference between @xmath30 and @xmath489 is in a @xmath261 universe at its ` turnaround ' point , as described in section [ ss - cos ] .
there one finds that , to lowest order in @xmath25 , the difference is given by @xmath490 which is precisely the newtonian formula for the self - potential of a uniform sphere of mass @xmath491 .
this is what we would have expected since , at the turnaround point , the kinetic energy vanishes .
the above should only be viewed as suggestive , but one idea that it appears to rule out is the popular suggestion that the total energy density of the universe should be zero @xcite . if the above analysis is correct then there is no possibility of the total energy density @xmath482 ever being zero .
furthermore , for spatially - flat models which we consider the most likely the total energy density resides entirely in the rest - mass energies of the particles in the universe and can not be cancelled by a negative gravitational contribution .
in this paper we developed a theory of gravity consisting of gauge fields defined in a flat background spacetime .
the theory is conceptually simple , and the role of the gauge fields is clearly understood they ensure invariance under arbitrary displacements and rotations .
while it is possible to maintain a classical picture of the rotation gauge group , a full understanding of its role is only achieved once the dirac action is considered .
the result is a theory which offers a radically different interpretation of gravitational interactions from that provided by general relativity . despite this , the two theories agree in their predictions for a wide range of phenomena .
differences only begin to emerge over issues such as the role of topology , our insistence on the use of global solutions , and in the interaction with quantum theory .
furthermore , the separation of the gauge fields into one for displacements and one for local rotations is suggestive of physical effects being separated into an inertia field and a force field .
indeed , there is good reason to believe that mass should enter relativistic multiparticle wave equations in the manner of the @xmath149-field @xcite .
it is possible that , in the development of a multiparticle theory , the @xmath492 and @xmath484 fields will be extended in quite distinct ways .
such possibilities do not appear to be open to a theory such as general relativity , with its reliance on the metric as the ` foundation of all ' @xcite .
probably the closest approach to the theory developed here is the spin-2 field theory discussed by many authors ( see box 17.2 of @xcite and @xcite ) .
this theory is usually viewed as reproducing general relativity exactly , albeit in a somewhat ugly form due to the existence of a background spacetime and the reliance on infinite series of the field variable . by contrast
, we hope to have demonstrated that gtg has an internal attractiveness of its own , as well as simplicity due to its first - order nature . a crucial question to address is whether any experimental tests are likely to distinguish between general relativity and gtg in the immediate future .
the biggest differences between general relativity and gtg to emerge to date lie in the treatment of black hole singularities @xcite , but these are unlikely to be testable for some considerable time ! a more promising area is the link between gravity and quantum spin .
gtg makes a clear prediction for the type and magnitude of this interaction , whereas it is not uniquely picked out in general spin - torsion theories , or in more general poincar gauge theory .
any experiment measuring this interaction would therefore provide a clear test of gtg .
a partial exploration of the effects of spin interactions in gtg is contained in @xcite . the techniques developed here
reveal some remarkable properties of spherically - symmetric systems .
it has been known since the 1930 s @xcite that , in the absence of pressure , the dynamical equations of cosmology can be cast in a newtonian form .
we have now shown that a single , unified , newtonian treatment can be given for all spherically - symmetric pressureless fluids , whether homogeneous or not .
furthermore , effects that have hitherto been viewed as the result of spacetime curvature can now be understood in a simple alternative fashion .
the result is a physical picture in which the background spacetime has no effect on either dynamics or kinematics .
this , we believe , is both new and potentially very useful . for example , simulations of black hole formation , and studies of the behaviour of the universe as a whole , can be carried out in exactly the same framework .
all previous studies have relied on cutting and pasting various metrics together , with the result that no clear , global view of the underlying physics can be achieved .
these advantages are exploited in @xcite to model the growth of a spherically - symmetric perturbation in a homogeneous background cosmology , and to study the effect of the perturbation on the cosmic microwave background .
the intrinsic method described here , and used to study spherically - symmetric systems , is quite general and can be applied to a wide range of problems . in @xcite the method
is applied to a restricted class of cylindrically - symmetric systems , and elsewhere we have presented treatments of axisymmetric systems @xcite . in all cases studied to date , the intrinsic method has brought considerable clarity to what would otherwise be a largely mysterious mess of algebra .
this is achieved by removing the dependence on an arbitrary coordinate system , and instead working directly with physical quantities .
the same technique also looks well suited to the study of cosmological perturbation theory , about which there has been considerable recent debate @xcite .
the interaction between dirac theory and the gauge theory developed here revealed a number of surprises .
the first was that consistency of the action principle with the minimal - coupling procedure restricted us to a theory that is unique up to the possible inclusion of a cosmological constant .
the second was that spatially flat cosmologies are the only ones that are consistent with homogeneity at the level of the single - particle dirac equation .
the final surprise was provided by a study of the dirac equation in a black hole background , which revealed a remarkable link with the hawking temperature and quantum field theory .
much work remains to settle the issues raised by this final point , however . as a final remark
, we also hope to have demonstrated the power of geometric algebra in analysing many physical problems .
many of the derivations performed in this paper would have been far more cumbersome in any other language , and none are capable of the compact expressions provided by geometric algebra for , say , the riemann tensor .
in addition , use of geometric algebra enabled us to remove all reference to coordinate frames from the fundamental equations .
this is a real aid to providing a clear physical understanding of the mathematics involved
. we would encourage anyone interested in studying the consequences of our theory to take time to master the techniques of geometric algebra .
we thank anton garrett for his careful reading of this manuscript and his many suggestions for improvements .
we also thank david hestenes for many enjoyable and thought - provoking discussions , and george ellis for a number of helpful suggestions .
in the dirac pauli representation the @xmath47-matrices are defined as @xmath493 where the @xmath494 are the standard pauli matrices @xcite .
the dirac @xmath495-matrices act on spinors , which are four - component complex column vectors .
a spinor @xmath496 is placed in one - to - one correspondence with an 8-component even element of the spacetime algebra via @xcite @xmath497 the action of the @xmath498 , @xmath381 and @xmath499 operators maps to @xmath500 the dirac equation , @xmath501 now takes the spacetime algebra form @xmath502 recombining to form the vectors @xmath503 and @xmath504 , and postmultiplying by @xmath51 , we arrive at the dirac equation in the form @xmath505 under lorentz transformations the spinor @xmath197 transforms single - sidedly to @xmath506 , hence the presence of the fixed @xmath51 and @xmath198 vectors on the right - hand side of @xmath197 does not break lorentz invariance .
the role of the dirac adjoint is played by the geometric operation of reversion , and the quantum inner product projects out the @xmath507 components from a general multivector .
so , for example , the real part of the inner product @xmath508 is given in the spacetime algebra by @xmath509 and the imaginary part by @xmath510 .
the dirac current @xmath511 is now replaced by the set of components @xmath512 these are simply the components of the vector @xmath397 , decomposed in the @xmath513 frame .
reference to the frame is removed from the vector by defining the current as @xmath514 similarly , the role of the spin current is played by the vector @xmath515 and the spin trivector is simply @xmath516 .
the dirac lagrangian has the equivalent form @xmath517 which is the form used in the main text .
a more detailed discussion of the spacetime algebra formulation of dirac theory is contained in @xcite .
we begin with a set of results for the derivative with respect to the vector @xmath2 in an @xmath10-dimensional space @xcite : @xmath518 the results needed for the multivector derivative in this paper are : @xmath519 where @xmath78 is the projection of @xmath24 onto the grades contained in @xmath72 .
these results are combined using leibniz rule ; for example , @xmath520 for the action principle we also require results for the multivector derivative with respect to the directional derivatives of a field @xmath197 .
the aim is again to refine the calculus so that it becomes possible to work in a frame - free manner .
( the derivations presented here supersede those given previously in @xcite . )
we first introduce a fixed orthonormal frame @xmath521 , with reciprocal @xmath522 , so that @xmath523 .
the partial derivative of @xmath197 with respect to the coordinate @xmath524 is abbreviated to @xmath525 so that @xmath526 we can now define the frame - free derivative @xmath527 the operator @xmath528 is the multivector derivative with respect to the @xmath2-derivative of @xmath197 .
the fundamental property of @xmath528 is that @xmath529 again , more complicated results are built up by applying leibniz rule .
the euler
lagrange equations for the lagrangian density @xmath530 can now be given in the form @xmath531 which is the form applied in the main text of this paper .
we also need a formalism for the derivative with respect to a linear function . given the linear function @xmath532 and the fixed frame @xmath533
, we define the scalar coefficients @xmath534 the individual partial derivatives @xmath535 are assembled into a frame - free derivative by defining @xmath536 the fundamental property of @xmath537 is that @xmath538 which , together with leibniz rule , is sufficient to derive all the required properties of the @xmath537 operator . for example
, if @xmath70 is a fixed bivector , @xmath539 this result extends immediately to give @xmath540 in particular , @xmath541 where the definition of the inverse has been employed .
the derivation of affords a remarkably direct proof of the formula for the derivative of the determinant of a linear function .
the above results hold equally if @xmath492 is replaced throughout by its adjoint @xmath335 , which is the form of the derivative used throughout the main text .
note , however , that @xmath542 thus the derivatives of @xmath543 and @xmath544 give different results , regardless of any symmetry properties of @xmath492 .
this has immediate implications for the symmetry ( or lack of symmetry ) of the functional stress - energy tensors for certain fields .
we finally need some results for derivatives with respect to the bivector - valued linear function @xmath180 .
the extensions are straightforward , and we just give the required results : @xmath545
the reformulation of the gauge theory presented in this paper in terms of conventional tensor calculus proceeds as follows .
a choice of gauge is made and a set of scalar coordinates @xmath546 is introduced .
we define the coordinate frame @xmath208 , @xmath547 with the reciprocal frame denoted @xmath548 . in terms of this the vector derivative is @xmath549 from the frame vectors we construct the vectors @xmath189 the metric is then given by the @xmath550 matrix @xmath190 if the @xmath0-dependence in @xmath551 is replaced by dependence solely on the coordinates @xmath546 then we recover riemann cartan geometry , where all relations are between coordinates and the concept of a point as a vector is lost . the connection is defined by ( following the conventions of @xcite ) @xmath552 where @xmath553 .
we can therefore write @xmath554 since @xmath555 we find that @xmath556 which recovers ` metric compatibility ' of the connection .
this is nothing more than the statement that the @xmath175 operator satisfies leibniz rule .
equation can be inverted to show that the connection contains a component given by the standard christoffel symbol .
the connection can then be written @xmath557 where @xmath558 is the contorsion tensor and is given by @xmath559 here @xmath560 is the torsion tensor , equal to the antisymmetric part of the connection : @xmath561 where @xmath562 is the torsion bivector .
the contorsion is formed from @xmath562 by @xmath563 for our gauge theory of gravity , the torsion is exclusively of the type @xmath564 , where @xmath565 is the spin trivector , in which case @xmath566 if we now consider the covariant derivative of a covariant vector @xmath567 , we find that @xmath568 so that the components of the vector @xmath569 are those expected for tensor calculus .
obviously the fact that @xmath570 implies that @xmath571 , so indices are raised and lowered in the expected manner . for covariant quantities such as the riemann tensor the translation to tensor calculus is straightforward : @xmath572 the general scheme is that any covariant quantity in gtg can be decomposed into tensor components by applying either the @xmath573 or @xmath574 , or a mixture of both , to yield a tensor with the appropriate number of upstairs and downstairs indices .
so , for example , @xmath161 can be decomposed to @xmath575 , @xmath576 or @xmath577 .
tensor calculus is poor at revealing which , if any , of the components represent a physical observable .
such issues are much clearer in gtg , which focuses attention on the single entity @xmath161 .
a vierbein @xmath578 ( essentially an orthonormal tetrad ) is given by @xmath579 where @xmath580 is a fixed orthonormal frame .
any position dependence in the @xmath580 is eliminated with a suitable rotor transformation .
when matrix operators @xmath581 are required these are replaced by the @xmath580 frame vectors using the method described in appendix [ app - dirac ] . in this way frame - free vectors can be assembled .
for example , the dirac operator ( * ? ? ?
* chapter 11 ) @xmath582 has the spacetime algebra equivalent @xmath583 the above relations enable many results from riemann cartan geometry to be carried over into our formalism , though the theory of gravity presented here is restricted in the structures from riemann cartan geometry that it admits ( the torsion is of trivector type , for example ) .
a similar translation scheme is easily set up for the language of differential forms , which is much closer to the spirit of geometric algebra than tensor calculus .
differential forms are scalar - valued functions of an antisymmetrised set of vectors .
they can easily be mapped to an equivalent multivector , and a full translation into geometric algebra is quite straightforward @xcite . here
we note in passing the geometric algebra equivalent of the hodge dual of a differential form , which is @xmath584 where @xmath23 is the multivector equivalent of @xmath585 .
doran , a.n .
lasenby , and s.f .
grassmann mechanics , multivector derivatives and geometric algebra . in z.
oziewicz , b. jancewicz , and a. borowiec , editors , _ spinors , twistors , clifford algebras and quantum deformations _
, page 215 .
kluwer academic , dordrecht , 1993 .
lasenby , c.j.l .
doran , and s.f .
2-spinors , twistors and supersymmetry in the spacetime algebra . in z.
oziewicz , b. jancewicz , and a. borowiec , editors , _ spinors , twistors , clifford algebras and quantum deformations _
, page 233 .
kluwer academic , dordrecht , 1993 .
d. hestenes . a unified language for mathematics and physics .
chisholm and a.k .
common , editors , _ clifford algebras and their applications in mathematical physics ( 1985 ) _ , page 1 .
reidel , dordrecht , 1986 .
doran , a.n .
lasenby , and s.f .
gravity as a gauge theory in the spacetime algebra . in f.
brackx , r. delanghe , and h. serras , editors , _ clifford algebras and their applications in mathematical physics ( 1993 ) _
, page 375 .
kluwer academic , dordrecht , 1993 .
lasenby , c.j.l .
doran , y. dabrowski , and a.d .
rotating astrophysical systems and a gauge theory approach to gravity . in n.
snchez and a. zichichi , editors , _ current topics in astrofundamental physics , erice 1996 _ , page 380 .
world scientific , singapore , 1997 .
lasenby , c.j.l .
doran , and s.f .
astrophysical and cosmological consequences of a gauge theory of gravity . in n.
snchez and a. zichichi , editors , _ advances in astrofundamental physics , erice 1994 _ , page 359 .
world scientific , singapore , 1995 .
lasenby , c.j.l .
doran , and s.f .
cosmological consequences of a flat - space theory of gravity . in f.
brackx , r. delanghe , and h. serras , editors , _ clifford algebras and their applications in mathematical physics ( 1993 ) _
, page 387 .
kluwer academic , dordrecht , 1993 .
the covariant and gauge invariant approach to perturbations in cosmology . in n.
snchez and a. zichichi , editors , _ current topics in astrofundamental physics : the early universe _
, page 1 .
kluwer academic , dordrecht , 1995 . | a new gauge theory of gravity is presented .
the theory is constructed in a flat background spacetime and employs gauge fields to ensure that all relations between physical quantities are independent of the positions and orientations of the matter fields . in this manner
all properties of the background spacetime are removed from physics , and what remains are a set of ` intrinsic ' relations between physical fields . for a wide range of phenomena , including all present experimental tests , the theory reproduces the predictions of general relativity .
differences do emerge , however , through the first - order nature of the equations and the global properties of the gauge fields , and through the relationship with quantum theory .
the properties of the gravitational gauge fields are derived from both classical and quantum viewpoints .
field equations are then derived from an action principle , and consistency with the minimal coupling procedure selects an action that is unique up to the possible inclusion of a cosmological constant .
this in turn singles out a unique form of spin - torsion interaction .
a new method for solving the field equations is outlined and applied to the case of a time - dependent , spherically - symmetric perfect fluid .
a gauge is found which reduces the physics to a set of essentially newtonian equations .
these equations are then applied to the study of cosmology , and to the formation and properties of black holes .
insistence on finding global solutions , together with the first - order nature of the equations , leads to a new understanding of the role played by time reversal .
this alters the physical picture of the properties of a horizon around a black hole .
the existence of global solutions enables one to discuss the properties of field lines inside the horizon due to a point charge held outside it .
the dirac equation is studied in a black hole background and provides a quick derivation of the hawking temperature . some applications to cosmology
are also discussed , and a study of the dirac equation in a cosmological background reveals that the only models consistent with homogeneity are spatially flat .
it is emphasised throughout that the description of gravity in terms of gauge fields , rather than spacetime geometry , leads to many simple and powerful physical insights .
the language of ` geometric algebra ' best expresses the physical and mathematical content of the theory and is employed throughout .
methods for translating the equations into other languages ( tensor and spinor calculus ) are given in appendices .
anthony lasenby , chris doran and stephen gull astrophysics group , cavendish laboratory , madingley road , + cambridge cb3 0he , uk .
this paper first appeared in _ phil .
trans .
r. soc .
lond .
a _ ( 1998 ) * 356 * , 487582 .
the present version has been updated with some corrections and improvements in notation .
some extra citations have been added , but these are not intended to be an exhaustive catalogue of work since the paper was completed in 1996 .
april 2004 part i foundations |
we will work over a ground field @xmath1 of characteristic zero .
a differential graded ( dg ) vector space @xmath2 will mean a complex of @xmath1-vector spaces with a differential of degree one .
the degree of a homogeneous element @xmath3 will be denoted by @xmath4 . in the context of graded algebra
, we will be using the koszul rule of signs when talking about the graded version of notions involving symmetry , including commutators , brackets , symmetric algebras , derivations , _
etc_. , often omitting the modifier _
graded_. for any integer @xmath5 , we define a _ translation _ ( or _ @xmath5-fold desuspension@xmath6 _ @xmath7 $ ] of @xmath2 : @xmath7^p : = v^{n+p}$ ] for each @xmath8 . for two graded vector spaces @xmath2 and @xmath9 , we define grading on the space @xmath10 of @xmath1-linear maps @xmath11 by @xmath12 for @xmath13 .
recall that a _ quantum group _ in the sense of drinfeld and jimbo is an associative , coassociative hopf algebra @xmath14 subject to the condition of being _ quasitriangular _
the latter implies , in particular , the existence of a solution @xmath15 to the _ quantum yang - baxter equation _ @xmath16 set up in @xmath14 .
more conceptually , the quasitriangularity condition provides data needed to put a braided structure on the monoidal category of left @xmath14-modules .
the most basic examples of quantum groups appear as _ quantizations _ or certain types of _ deformations _ ( in the sense of hopf algebras ) of universal enveloping algebras and algebras of functions on groups . in the first case , starting with a lie algebra @xmath17 and a hopf - algebra deformation @xmath18 of its universal enveloping algebra @xmath19 , one passes to the `` _ _
@xmath20semi@xmath6classical limit _
_ '' @xmath21 thus equipping @xmath19 with a _ co - poisson - hopf structure _ with @xmath22 being the _ co - poisson cobracket_. in particular , the restriction @xmath23 becomes a well - defined cobracket on @xmath17 turning it into a lie bialgebra .
the above process can be reversed : as it was shown in @xcite , any finite - dimensional lie bialgebra @xmath24,\delta)$ ] can be _ quantized _ , meaning that one can always come up with a hopf - algebra deformation @xmath18 whose classical limit , in the sense of the above formula , agrees with @xmath25 . while a priori @xmath18 is just a hopf algebra , one would really be interested in having a quasitriangular structure on it . as a special case , it was shown in @xcite that such a structure exists , when @xmath17 is a _ triangular _ lie bialgebra . this class of lie bialgebras is defined as follows : let @xmath17 be a lie algebra and @xmath26 ( `` _ _ a classical @xmath27-matrix _ _ '' ) be a skew - symmetric element satisfying the _ classical yang - baxter equation _ @xmath28+[r_{12},r_{23}]+[r_{23},r_{13 } ] = 0,\ ] ] which can be conveniently restated in the form of the _ maurer - cartan equation _ @xmath29 = 0\ ] ] taking place in the graded lie algebra @xmath30)[1]$ ] with respect to the _ schouten bracket _ for elements @xmath27 of degree one : @xmath31)[1])^1 = ( s(\g[-1]))^2 = s^2(\g[-1])[2 ] = \g \wedge \g$ ] .
such an element @xmath27 , called a _
maurer - cartan element _ , gives rise to a lie cobracket on @xmath17 in the form of the coboundary @xmath32 of @xmath27 taken in the cochain _ chevalley - eilenberg complex _ of @xmath17 with coefficients in @xmath33 ( here , @xmath27 is regarded as a 0-cocycle ) .
the compatibility with the lie - algebra structure on @xmath17 is packed into the relation @xmath34 , thus guaranteeing that @xmath17 with such a cobracket is indeed a lie bialgebra . the _ co - jacobi identity _ , which could be rewritten as @xmath35 = 0,\ ] ] follows from the following statement , which is an odd version of the hamiltonian correspondence in poisson geometry , if one regards @xmath30)[1]$ ] as the shifted gerstenhaber algebra of functions on the odd poisson manifold @xmath36)^*$ ] and @xmath37 , s(\g[-1]))$ ] as the graded lie algebra of vector fields on @xmath36)^*$ ] .
[ hamiltonian ] the linear map @xmath38)[1 ] \to \operatorname{hom}(\g[-1 ] , s(\g[-1]))\ ] ] is a graded lie - algebra morphism .
a _ triangular lie bialgebra _ is a lie algebra @xmath17 provided with a lie cobracket @xmath39 coming out of an @xmath27-matrix @xmath27 . a basic statement concerning this class of lie
bialgebras is that @xmath19 can be quantized to a _
triangular hopf algebra _ @xmath18 .
this condition is stronger than being quasitriangular , and in particular , the category of ( left ) modules over a triangular hopf algebra turns out to be symmetric monoidal , as opposed to just being braided .
the upshot of the above construction is that there is a source of quantum groups coming from the data of a lie algebra @xmath17 and a solution of the maurer - cartan equation in @xmath30)[1]$ ] .
the goal of our project is to promote this construction to the realm of homotopy lie algebras . in particular
, this would generalize the work @xcite done for the case of lie 2-bialgebras .
here is an outline of our program : 1 . develop the notion of a _
triangular @xmath40-bialgebra _ extending the classical one . in analogy with the classical case ,
the input data for this construction consists of an @xmath40-algebra @xmath17 and a solution @xmath27 of a generalized maurer - cartan equation set up in an appropriate algebraic context ; 2 .
show that the universal enveloping algebra @xmath19 of a triangular @xmath40-bialgebra @xmath17 admits a natural _ homotopy co - poisson - hopf _ structure ; 3 .
extend the _ drinfeld twist _
construction @xcite , which equips a cocommutative hopf algebra with a new , triangular coproduct , to the homotopical context .
apply it to the case of universal enveloping algebra @xmath19 of the previous step to obtain a _
quantum@xmath0 group_. the current paper is dedicated to describing the first step of the construction , which we believe might be interesting on its own .
the second step is work in progress .
while the universal enveloping algebra @xmath19 of an @xmath40-algebra @xmath17 is a strongly homotopy associative ( or @xmath41- ) algebra that also turns out to be a cocommutative , coassociative coalgebra object in lada - markl s , see @xcite , symmetric monoidal category of @xmath41-algebras , we would be interested in verifying that @xmath19 is actually a _
hopf@xmath0 algebra _
, that is , possesses an antipodal map satisfying certain compatibility conditions .
we would also need to translate an @xmath40-bialgebra structure on an @xmath40-algebra @xmath17 into a co - poisson@xmath0 structure on the hopf@xmath0 algebra @xmath19 .
this would result in providing @xmath19 with the structure of a cocommutative co - poisson@xmath0 coalgebra object in lada - markl s symmetric monoidal category of @xmath41-algebras .
furthermore , we would like to develop deformation theory of homotopy hopf algebras and use it to quantize homotopy lie bialgebras . a different approach to quantization of homotopy lie bialgebras ( using the framework of props ) was taken in @xcite , in which a different notion of a homotopy hopf algebra ( or rather , homotopy bialgebra ) was used .
that notion depends on the choice of a minimal resolution of the bialgebra properad .
the notion we outline above appears to be more canonical . in the future
we would also be interested in investigating what this program produces for @xmath40-algebras arising in the geometric context , such as generalized poisson geometry , @xmath40-algebroids , or @xmath42-geometry .
recall that the structure of a ( strongly ) homotopy lie algebra ( or _ @xmath40-algebra _ ) on a graded vector space @xmath17 may be given by a _
codifferential _ , _
i.e_. , a degree - one , square - zero coderivation @xmath43 such that @xmath44 , on the graded cocommutative coalgebra @xmath45)$ ] equipped with the shuffle comultiplication . the date given by @xmath43 is equivalent to a collection of `` higher lie brackets '' @xmath46 ) \to \g[1]$ ] , @xmath47 , of degree one obtained by restriction of @xmath43 to the @xmath1th symmetric component of @xmath45)$ ] followed by projection to the cogenerators . the condition @xmath48 is equivalent to the _ higher jacobi identities _ , homotopy versions of the jacobi identity . outside of deformation theory ,
nontrivial examples of homotopy lie algebras are known to arise in the context of multisymplectic geometry @xcite , courant algebroids @xcite , and closed string field theory @xcite . in order to discuss the structure of an @xmath40-bialgebra on a graded vector space @xmath17
, we need to mix the graded lie algebra @xmath49),\g[1])$ ] , used to define the structure of an @xmath40-algebra on @xmath17 , with the graded lie algebra @xmath37 ) , s ( \g[-1]))$ ] , used to define the cobracket on @xmath17 in the classical , lie algebra setting in section [ classical ] . consider the graded vector space @xmath50),s^n(\g[-1]))[2]\ ] ] and
provide it with the structure of a graded lie algebra given by the graded commutator , called the _ big bracket _
, @xmath51 : = f \circ g - ( -1)^{{{\lvertf\rvert}}\cdot { { \lvertg\rvert } } } g \circ f\ ] ] under the _ circle _ , or @xmath52 _ product _ , cf.@xcite and @xcite : @xmath53 where @xmath54 $ ] , @xmath55),s^m(\g[-1]))[2],\\ g \in \prod\limits_{m\geq 0 } \operatorname{hom}(s^{l}(\g[1]),s^m(\g[-1]))[2],\end{gathered}\ ] ] @xmath56 otherwise we set @xmath57 , @xmath58 is the set of @xmath59 _ shuffles _ : permutations @xmath60 of @xmath61 such that @xmath62 and @xmath63 , @xmath64 , @xmath65 is the _ koszul sign _ of the permutation of @xmath66 to @xmath67 in @xmath45)$ ] , and we use sweedler s notation to denote the result @xmath68 of applying to @xmath69)[2]$ ] the ( shifted ) comultiplication @xmath30)[2 ] \to s(\g[-1])[2 ] \otimes s(\g[-1])$ ] followed by the projection @xmath30)[2 ] \to \g[1]$ ] onto the cogenerators in the first tensor factor .
this graded lie algebra @xmath70 , under the assumption that @xmath71 and in a slightly different incarnation , was introduced by y. kosmann - schwarzbach @xcite in relation to lie bialgebras and later used by o. kravchenko @xcite in relation to @xmath40-bialgebras .
the graded lie algebra @xmath70 has the property that its maurer - cartan elements represent @xmath40 brackets and cobrackets on @xmath17 , as well as mixed operations , comprising the structure of an @xmath40-bialgebra on @xmath17 .
here we adopt kravchenko s approach and define an _
@xmath40-bialgebra structure _ on @xmath17 as a maurer - cartan element @xmath72 in the subalgebra @xmath73 ) , s^n(\g[-1]))[2]\ ] ] of the graded lie algebra @xmath70 .
would correspond to more general , _ curved _ @xmath40-bialgebras . ]
this means @xmath74 ) \to s^n(\g[-1])[2 ] \quad \text{of degree 1},\\ \nonumber [ \mu , \mu ] = 0.\end{gathered}\ ] ]
for an @xmath40-algebra @xmath17 , one can generalize the schouten bracket to an @xmath40 structure on @xmath30)[1]$ ] by extending the higher brackets @xmath76 on @xmath17 as graded multiderivations of the graded commutative algebra @xmath30)$ ] .
this @xmath40 structure may also be described via _ higher derived brackets _ ( in the semiclassical limit ) on the @xmath42-algebra @xmath30)$ ] , see ( * ? ? ?
* example 3.4 ) .
the @xmath40 structure can be naturally extended to the completion @xmath77)[1 ] : = \prod\limits_{n\geq 0 } s^n(\g[-1])[1].\ ] ] while investigating the deformation - theoretic meaning of solutions @xmath78)[1][[\lambda]]$ ] , where @xmath79 is the _ deformation parameter _ , that is to say , a ( degree - zero ) formal variable , of the _ generalized maurer - cartan equation _
@xmath80 where @xmath81 refers to multiplication in @xmath82 for @xmath83)[2]$ ] , certain analogies can be drawn with basic constructions of the theory of quantum groups .
note that an @xmath40-algebra structure on @xmath17 endows the graded lie algebras @xmath70 and @xmath84 with the structure of a dg lie algebra .
namely , bracketing with the maurer - cartan element @xmath85 ) , \g[-1])[2]$ ] , representing the @xmath40-algebra structure on @xmath17 , creates a differential @xmath86\ ] ] on @xmath70 and @xmath84 , compatible with the `` big '' bracket .
an @xmath40-_morphism _ @xmath87 from the @xmath40-algebra @xmath88)[1]$ ] to the dg lie algebra @xmath84 , that is to say , a morphism @xmath89)[2])\to s(\b^+[1])$ ] of dg coalgebras mapping 1 to 1 , amounts to defining a series of degree - zero linear maps @xmath90)[2 ] ) \to \b^+[1]$ ] for all @xmath91 satisfying the following compatibility conditions : @xmath92\\ = \sum_{m=1}^n \sum_{\tau \in \operatorname{sh}_{m , n - m}}\!\ ! ( -1)^{{{\lvertx_\tau\rvert } } } \varphi_{n - m+1 } ( l_m(x_{\tau(1 ) } \odot \dots \odot x_{\tau(m ) } ) \odot x_{\tau(m+1 ) } \odot \dots \odot x_{\tau(n)}),\end{gathered}\ ] ] where @xmath93)[2]$ ] and @xmath94 .
there is a _
canonical @xmath40-morphism _
@xmath95)[1 ] \to \b^+$ ] , which may be defined by the maps @xmath96 where @xmath93)[2]$ ] , @xmath97)$ ] , @xmath98)[2])$ ] , @xmath99 , and @xmath100 ) \to s(\s(\g[-1])[2])$ ] is the graded - algebra morphism induced by the obvious linear map @xmath101 \hookrightarrow \s(\g[-1])[2 ] \hookrightarrow s(\s(\g[-1])[2])$ ] .
the following theorem generalizes proposition [ hamiltonian ] to the @xmath40 setting .
the proof of the theorem is a straightforward checkup that reduces the statement to the higher jacobi identities for the @xmath40 brackets on @xmath88)[1]$ ] .
this theorem also generalizes kravchenko s result ( * ? ? ?
* theorem 19 ) , which provides an @xmath40-morphism from an @xmath40-algebra @xmath17 to the graded lie algebra @xmath105 .
@xmath27 is a _
@xmath20generalized@xmath6 maurer - cartan element _
@xmath106 in the @xmath40-algebra @xmath107)[1][[\lambda]]$ ] , _
i.e_. , a degree - one solution of the generalized maurer - cartan equation . sending
an @xmath75-matrix to the subalgebra @xmath84 of the big - bracket dg lie algebra @xmath70 under an @xmath40-morphism @xmath95)[1 ] \to \b^+$ ] would yield a maurer - cartan element @xmath108 : @xmath109 depending on the deformation parameter @xmath79 , in the dg lie algebra @xmath110 $ ] : @xmath111 = 0,\ ] ] or , equivalently , a maurer - cartan element @xmath112 in the graded lie algebra @xmath110 $ ] : @xmath113 = 0,\ ] ] giving rise to an @xmath40-bialgebra structure on @xmath17 , as per section [ bb ] , in analogy with @xmath39 giving rise to an ordinary lie cobracket in the classical , nonhomotopical case , see also example [ example ] below .
we call the @xmath40-bialgebra @xmath114 produced out of an @xmath40-algebra @xmath17 and an @xmath75-matrix @xmath27 by transferring it to a maurer - cartan element @xmath115 in @xmath84 via the canonical @xmath40-morphism @xmath87 , as above , a _
triangular @xmath40-bialgebra_. [ example ] in the case of a classical lie algebra @xmath17 , the graded lie algebra @xmath88)[1]$ ] is just the completed graded lie algebra of ( right-)invariant multivector fields on the corresponding local lie group with the schouten bracket ( or , equivalently , up to degree shift , functions on the formal odd poisson manifold @xmath36)^*$ ] ) , and the canonical @xmath40-morphism @xmath87 is just linear : @xmath116 , equal to the chevalley - eilenberg differential of 0-cochains of the lie algebra @xmath17 with coefficients in the graded @xmath17-module @xmath88)[2]$ ] .
the fact that @xmath87 is an @xmath40-morphism translates into being a dg lie morphism , _
i.e_. , satisfying two compatibility conditions @xmath117 & = \varphi([x , y]),\end{aligned}\ ] ] where @xmath118 is the operator taking the big bracket with the lie structure @xmath119 ) , \g[-1])[2 ] \subset \b^+$ ] : @xmath120.\ ] ] the first condition means that @xmath121 , and the second states that @xmath122 is a lie - algebra morphism , which is the assertion of proposition [ hamiltonian ] .
an @xmath123-matrix is a solution @xmath124)[1])^1 = ( s^2(\g[-1]))^2 = \g \wedge \g$ ] of the generalized maurer - cartan equation , which turns into the classical maurer - cartan equation @xmath125 = 0 $ ] in this case , and thereby @xmath27 is just a classical @xmath27-matrix .
thus , the transfer @xmath126 of an @xmath27-matrix is a lie cobracket @xmath127 on @xmath17 , satisfying the compatibility condition with the lie bracket and the co - jacobi identity , resulting in the structure of a triangular lie bialgebra on @xmath17 .
here we ignored the deformation parameter @xmath79 , because @xmath88)[1]$ ] is just a graded lie algebra and the generalized maurer - cartan equation in @xmath88)[1]$ ] and the morphism @xmath87 have only finitely many terms .
when @xmath17 is a dg lie algebra , the picture is dramatically different from the classical picture of example [ example ] .
now an @xmath123-matrix @xmath27 may have many more components than just one in @xmath128 : @xmath129)[1])^1 = ( \s(\g[-1]))^2 = \prod\limits_{n\geq 0 } ( s^n(\g[-1]))^2 = \prod\limits_{n\geq 1 } ( s^n(\g[-1]))^2.\ ] ] however , @xmath88)[1]$ ] is just a dg lie algebra , and the generalized maurer - cartan equation is still classical , @xmath130 = 0.\ ] ] the @xmath40-morphism @xmath87 is still linear @xmath131 , _
i.e_. , @xmath87 is a dg lie - algebra morphism .
the transfer @xmath132 of @xmath27 via @xmath87 will result in a @xmath40-bialgebra structure on @xmath17 with higher operations @xmath133 , see , which are trivial for all pairs @xmath134 but those with @xmath135 , @xmath91 and @xmath136 , @xmath137 .
thus , even in the case of a dg lie algebra @xmath17 , our construction creates a triangular @xmath40-bialgebra , rather than just a ( dg ) triangular lie algebra .
d. bashkirov and a. a. voronov .
the bv formalism for l@xmath0-algebras .
reprint ipmu14 - 0339 , kavli ipmu , kashiwa , japan , november 2014 . `
arxiv:1410.6432 [ math.qa ] ` , to appear in j. homotopy relat . struct . y. kosmann - schwarzbach .
jacobian quasi - bialgebras and quasi - poisson lie groups . in _ mathematical aspects of classical field theory ( seattle ,
wa , 1991 ) _ , volume 132 of _ contemp .
_ , pages 459489 .
soc . , providence , ri , 1992
. d. roytenberg . on the structure of graded symplectic supermanifolds and courant algebroids . in _
quantization , poisson brackets and beyond ( manchester , 2001 ) _ , volume 315 of _ contemp .
_ , pages 169185 .
soc . , providence , ri , 2002 . | a homotopy analogue of the notion of a triangular lie bialgebra is proposed with a goal of extending basic notions of the theory of quantum groups to the context of homotopy algebras and , in particular , introducing a homotopical generalization of the notion of a quantum group , or quantum@xmath0-group . |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.